In some Candida species, the universal CUG leucine codon is translated as serine. However, in most cases, the serine tRNAs responsible for this non‐universal decoding (tRNASerCAG) accept in vitro not only serine, but also, to some extent, leucine. Nucleotide replacement experiments indicated that m1G37 is critical for leucylation activity. This finding was supported by the fact that the tRNASerCAGs possessing the leucylation activity always have m1G37, whereas that of Candida cylindracea, which possesses no leucylation activity, has A37. Quantification of defined aminoacetylated tRNAs in cells demonstrated that 3% of the tRNASerCAGs possessing m1G37 were, in fact, charged with leucine in vivo. A genetic approach using an auxotroph mutant of C.maltosa possessing this type of tRNASerCAG also suggested that the URA3 gene inactivated due to the translation of CUG as serine was rescued by a slight incorporation of leucine into the polypeptide, which demonstrated that the tRNA charged with multiple amino acids could participate in the translation. These findings provide the first evidence that two distinct amino acids are assigned by a single codon, which occurs naturally in the translation process of certain Candida species. We term this novel type of codon a ‘polysemous codon’.
The universality of the genetic code was once considered to be one of the essential characteristics of life, which led to the conception of the ‘frozen accident theory’. This theory proposes that all extant living organisms use the universal genetic code, which was born by accident and ‘frozen’, and that they originate from a single, closely interbreeding population (Crick, 1968). However, in recent years a number of non‐universal genetic codes have been reported in various non‐plant mitochondrial systems, as well as in several nuclear systems (reviewed in Osawa et al., 1992; Osawa, 1995), which contradict the frozen accident theory.
Among these deviations from the universal codes, Kawaguchi et al. (1989) demonstrated that CUG, a universal leucine codon, is translated as serine in an asporogenic yeast, Candida cylindracea. We identified the serine tRNA having the anticodon CAG, which is responsible for the assignment of codon CUG as serine (termed tRNASerCAG), and revealed its decoding mechanism by means of an in vitro translational assay system (Yokogawa et al., 1992; Suzuki et al., 1994). Furthermore, when we investigated the distribution of this non‐universal genetic code in fungi, as well as C.cylindracea, eight other Candida species—C.albicans, C.zeylanoides, C.lusitaniae, C.tropicalis, C.melbiosica, C.parapsilosis, C.guilliermondii and C.rugosa—were found to utilize the codon CUG for serine instead of leucine, all having tRNASerCAG as the mediator in the unusual decoding (Ohama et al., 1993; Ueda et al., 1994). Several other investigators have also shown that the codon CUG is actually translated as serine in vivo in C.albicans and C.maltosa (Santos and Tuite, 1995a; Sugiyama et al., 1995; Zimmer and Schunck, 1995).
One of the most remarkable structural features observed in most of these tRNASerCAGs is that the nucleotide 5′‐adjacent to the anticodon (position 33) is occupied not by the conserved U residue (U33) but by a G residue (G33). It has been speculated that U33 is necessary for forming the U‐turn structure of the anticodon loop in all tRNAs reported so far (Quigley and Rich, 1976; Sprinzl et al., 1996). Moreover, the nucleotide at position 37, 3′‐adjacent to the anticodon CAG, is 1‐methyl guanosine (m1G) in almost all tRNASerCAGs except for that of C.cylindracea (A37), while all the serine tRNAs in fungi corresponding to the universal serine codons UCN (N: A, C, U or G) and AGY (Y: U or C) have modified adenosine at this position without exception (Sprinzl et al., 1996). The question then arises as to why these tRNASerCAGs specific for this non‐universal codon possess such unique features in the sequence surrounding the anticodon.
In the past 10 years, the mechanism by which aminoacyl‐tRNA synthetases recognize their cognate tRNAs has been extensively investigated both in vitro and in vivo (Schimmel, 1987; Schulman and Abelson, 1988; Yarus, 1988; Normanly and Abelson, 1989; Shimizu et al., 1992; McClain, 1993; Schimmel et al., 1993). This line of study began with the artificial conversion of leucine tRNA of Escherichia coli to serine tRNA by Abelson's group 10 years ago (Normanly et al., 1986). Recently, tRNA identity elements of Saccharomyces cerevisiae leucine tRNA were elucidated using unmodified variants synthesized by T7 RNA polymerase (Soma et al., 1996), indicating that in addition to the discriminator base, A73, the second letter of the anticodon, A35, and the nucleotide 3′‐adjacent to the anticodon, m1G37, are important for recognition by leucyl‐tRNA synthetase (LeuRS). The majority of Candida tRNASerCAGs have A35 and m1G37, while the discriminator is occupied by a nucleoside other than adenosine (mostly G73). In this respect, tRNASerCAG seems to be a potentially chimeric tRNA molecule capable of being recognized not only by seryl‐ but also by leucyl‐tRNA synthetases.
Previously, we showed that these tRNASerCAGs would have originated from the serine tRNA corresponding to codon UCG (Ueda et al., 1994). This suggests an evolutionary pathway in which conversion from A to m1G would have taken place at position 37 just after the emergence of tRNASerCAG had brought about a change in the universal code. Since such a mutation at position 37 might potentially result in the leucylation of tRNASerCAG, we attempted to elucidate the charging properties of these tRNASerCAGs both in vitro and in vivo. Based on the results of in vitro aminoacylation reactions using tRNA variants constructed by the microsurgery method, the direct analysis of aminoacylated tRNAs in cells and a genetic approach, we demonstrate here that these serine tRNAs are actually leucylated both in vitro and in vivo. Furthermore, m1G at position 37 was found to be indispensable for the leucylation of tRNASerCAGs. In fact, the tRNASerCAG of C.cylindracea, which has A at position 37, exhibits no leucylation activity. C.cylindracea has a high G+C content (63%) and utilizes CUG as a major serine codon. However, the other Candida species have no such high G+C content and utilize the CUG as a minor serine codon (Kawaguchi et al., 1989; Lloyd and Sharp, 1992; our unpublished observation). Considering the relationship between the usage of the codon CUG as serine and the leucylation properties of tRNASerCAG, it seems that only Candida species with a genome in which the incidence of the CUG serine codon is very low possess serine tRNASerCAG that can be leucylated. Furthermore, such tRNASerCAGs charged with heterogeneous amino acids should be utilized equally in the translation process. This is the first demonstration that a single tRNA species is assigned to two different amino acids in the cell. We propose designating this type of codon having multiple amino acid assignment as a ‘polysemous codon’. The correlation between the dual‐assignment state and the pathway of genetic code diversification is also discussed.
Candida zeylanoides tRNASerCAG is leucylated in vitro
First the leucylation of tRNASerCAGs from C.zeylanoides and C.cylindracea was examined using LeuRS partially purified from C.zeylanoides, since it is known that leucine tRNAs of yeast have one of their identity determinants at position 37 (Soma et al., 1996) and tRNASerCAGs of C.zeylanoides and C.cylindracea have different nucleotides at this position (m1G and A, respectively) (Figure 1A). Both tRNAs showed almost full serylation activity (∼1200–1500 pmol/A260 unit), as shown in Figure 1B. The tRNASerCAG of C.zeylanoides was evidently leucylated (the kinetic parameters are given in the uppermost row of Table I), though the charging activity was lower than that for serylation. This low acceptance of leucine of tRNASerCAG may be due to the partial purification of LeuRS and high Km value of LeuRS towards tRNASerCAG. On the contrary, tRNASerCAG of C.cylindracea was not leucylated at all, as was the case when another species of serine tRNA specific for codon AGY (Y: U or C) (tRNASerGCU) was employed as a control substrate (Figure 1B, right‐hand graph). The Km value of C.zeylanoides LeuRS towards tRNASerCAG (5.0 μM) is only one order of magnitude larger than that of the serylation of this tRNA (0.22 μM) as well as that of leucylation toward the cognate leucine tRNAs of S.cerevisae (0.34 μM; Soma et al., 1996).
In order to verify that the leucylation activity observed for the tRNASerCAG of C.zeylanoides actually came from the tRNASerCAG itself, and not from a trace amount of leucine tRNA contaminating the tRNA sample, the leucylated 3′‐terminal RNA fragment derived from leucyl‐tRNASerCAG was analyzed in the following manner. 14C‐leucylated tRNASerCAG from C.zeylanoides was first acetylated with acetic anhydride to prevent deacylation, and then digested with RNase T1. The resulting 3′‐terminal fragment with 14C‐labeled acetylleucine was analyzed by cellulose TLC. The results are shown in Figure 1C. If leucylated tRNASerCAG were digested with RNase T1, 14C‐labeled acetylleucyl‐CCA should be released as a labeled fragment (Figure 1C, lane 3), because G is located at position 73 of the tRNASerCAG (Figure 1A, left‐hand structure). Any contaminated leucine tRNAs, if they exist, will give some 14C‐labeled fragments larger than the tetramer (Figure 1C, lane 4), because all the leucine tRNAs of yeasts so far analyzed (Sprinzl et al., 1996) including those of C.zeylanoides (T.Suzuki, unpublished result) are known to have A73 at their 3′‐ends, which are resistant to RNase T1. The mobility of the acetylleucyl‐oligonucleotide derived from tRNASerCAG from C.zeylanoides (Figure 1C, lane 1) was identical to that of acelylleucyl‐CCA prepared from the RNase U2 digests of leucyl‐tRNALeus from C.zeylanoides (lane 3).
This observation clearly demonstrates that leucine is definitely attached to the tRNA possessing G73; the tRNA therefore must be tRNASerCAG and not tRNALeu. Thus, it is concluded that the tRNA which incorporated leucine in vitro is in fact tRNASerCAG. This deduction is supported by the results of an additional experiment: incorporation of [14C]leucine into the tRNASerCAG sample with LeuRS was reduced by the addition of SerRS and non‐labeled serine to the reaction mixture (data not shown), which clearly indicates that the same tRNA molecule is competitively aminoacylated by these two enzymes.
To conclude that tRNASerCAG is aminoacylated with leucine, we carried out a further experiment. The tRNASerCAG was charged with serine and serylated tRNASerCAG was separated from non‐aminoacylated tRNASerCAG by gel‐electrophoresis under acidic conditions. After deacylation, the leucylation activity of the tRNASerCAG was unequivocally detected. This experiment clearly indicates that a tRNASerCAG molecule with serylation activity simultaneously possesses leucine‐accepting activity.
Leucyl‐tRNA synthetase from C.zeylanoides also leucylated tRNASerCAGs from C.albicans, C.lusitaniae and C.tropicalis (data not shown), but the tRNASerCAG of C.cylindracea was not leucylated at all. This charging property was not due to the heterologous combination of the synthetase and tRNA, since similar results were observed with LeuRSs from both C.cylindracea and S.cerevisiae (data not shown).
m1G37 is responsible for recognition by leucyl‐tRNA synthetase
Among the tRNASerCAGs of several Candida species, that of C.cylindracea is unique because it alone possesses no leucylation capacity. A sequence comparison of these tRNAs (Figure 1A) prompts us to speculate that the nucleotide at position 37 is strongly associated with leucylation, because all tRNASerCAGs possessing leucylation activity have m1G in common, while only the tRNASerCAG of C.cylindracea, which possesses no leucylation activity, has A at this position.
To examine the validity of this speculation, a series of tRNASerCAG variants was constructed by the in vitro transcription method using T7 RNA polymerase, as well as by the microsurgery method, and the leucylation activity of each variant was measured. When the tRNASerCAG of C.zeylanoides synthesized by in vitro transcription was employed as a substrate, no leucylation activity was detected, not even for the tRNA transcript having G37 (Figure 3A). On the other hand, as shown in Figure 3A, serylation activity exceeded 1000 pmol/A260unit. These results strongly suggested that some nucleoside modification is necessary in tRNASerCAG for recognition by LeuRS. We thus attempted to replace the m1G37 of C.zeylanoides tRNASerCAG with G (the variant is symbolized as m1G37G) or A (m1G37A), by the microsurgery method (Figure 2A and B; for details, see Materials and methods) to examine the contribution of m1G37 to leucylation and the contribution of A37 of C.cylindracea tRNASerCAG to the prevention of leucylation.
When aminoacylation of m1G37A and m1G37G was examined (Figure 3A), the results indicated that both substitutions lead to complete loss of leucylation (Figure 3A, right‐hand graph), although no apparent influence was observed on serylation (Figure 3A, left‐hand graph). These findings strongly indicate that the methyl group of m1G37 plays a crucial role in enhancing the leucylation activity of tRNASerCAG.
The slight reduction in leucylation activity observed in the control variant z‐G33G (Figure 2A) compared with native tRNA (Figure 3A, right‐hand graph) was found to have resulted from the partial deacetylation of 4‐acetyl cytidine (ac4C) due to acid treatment of the 5′‐half fragment of tRNASerCAG (see Materials and methods). This is considered further in the Discussion.
G33 acts as a modulator of leucylation
In addition to m1G37, another unique feature of the serine tRNASerCAGs in these Candida species is the presence of G at position 33, where a pyrimidine (mostly U) is completely conserved in usual tRNAs (Sprinzl et al., 1996). Since we considered it is possible that this notable feature may be in some way related to the unusual aminoacylation characteristics described above and/or to the translation of non‐universal genetic code, we examined the effect of residue 33 on the aminoacylation and translation activities of mutated tRNASerCAG by introducing a point mutation at this position in the tRNAs of C.zeylanoides and C.cylindracea using the microsurgery method (see Figures 2A and C, and 7B and C).
The effect of mutation at position 33 in these two tRNAs was found to be quite different. In the case of the C.cylindracea tRNA, none of the mutations at position 33 caused leucylation of the tRNA, as was observed with the native tRNASerCAG, and there was no reduction in serylation activity (Figure 3C). In contrast, the replacement of G33 by pyrimidines in C.zeylanoides tRNASerCAG considerably enhanced the leucylation activity (Figure 3B, right‐hand graph), while no significant difference was observed in the serylation activity (Figure 3B, left‐hand graph). The kinetic parameters of leucylation for the variants of C.zeylanoides tRNA are shown in Table I. It is notable that the Km values of the two pyrimidine mutants, z‐G33U (1.4 μM) and z‐G33C (1.3 μM), are clearly lower than those of the two purine mutants, z‐G33A (6.7 μM) and z‐G33G (5.6 μM). The Vmax value of z‐G33U (1.2 pmol/min) is 39% of that of z‐G33C (3.1 pmol/min), which could explain why z‐G33U shows lower leucylation activity than z‐G33C despite having nearly the same Km value (Figure 3B, right‐hand graph). Judging from the sequence analysis (data not shown), the slight reduction in the leucylation of z‐G33G (5.6 μM) compared with that of the native tRNASerCAG (5.0 μM) is probably due to the partial deacetylation of ac4C at position 12, as mentioned above. This was confirmed by the observation of a slight reduction in leucylation activity also in acid‐treated native tRNASerCAG (data not shown). It is thus concluded that replacement of a pyrimidine by a purine at position 33 has a repressive effect on leucylation of the tRNASerCAG of C.zeylanoides.
The translation efficiencies of the variants with a mutation at position 33 were also examined in a cell‐free translation system of C.cylindracea (Yokogawa et al., 1992; Suzuki et al., 1994), to evaluate the effect of G33. A change from G to U at position 33 apparently enhanced the translation activity 2.5‐fold, although their decoding properties did not change at all (data not shown). We thus consider that G33 serves as a modulator of leucylation of tRNASerCAG, despite a slight disadvantage in translation activity.
Evidence for leucylation of C.zeylanoides tRNASerCAG in vivo
At this point, we had established that the tRNASerCAG of C.zeylanoides is actually able to accept leucine in vitro. However, considering the facts that SerRS and LeuRS coexist in cells and, judging from their Km values, that the affinity of tRNASerCAG toward SerRS is one order of magnitude higher than that toward LeuRS, we needed to ascertain whether the tRNASerCAG of C.zeylanoides is in fact leucylated in vivo. For this purpose, we adopted a newly developed method for quantifying an individual aminoacyl‐tRNA in cells (Suzuki et al., 1996).
Aminoacyl‐tRNAs separately prepared from cells of C.zeylanoides and C.cylindracea were immediately subjected to acetylation using [1‐14C]acetic anhydride to label the amino acids as well as to stabilize the aminoacylated tRNAs. From each of the acetylated aminoacyl‐tRNA mixtures, tRNASerCAGs from C.zeylanoides and C.cylindracea were fished out by a solid‐phase‐attached DNA probe as described previously (Tsurui et al., 1994; Wakita et al., 1994). A single band for each of the aminoacyl‐tRNAs was detected by staining (Figure 4A) with which the radioactivity coincided in each case (Figure 4B).
Acetylated amino acids attached to these tRNAs were deacylated by alkaline treatment and analyzed by TLC. As shown in Figure 4C, acetylserine was observed as a major amino acid derivative in both tRNASerCAGs, but acetylleucine was detected only in the C.zeylanoides tRNASerCAG; the acetylserine and acetylleucine spots were identified as described previously (Suzuki et al., 1996). The radioactivities remaining on the origins probably came from the direct acetylation of some nucleotides in the tRNAs, as discussed previously (Suzuki et al., 1996). From comparison with the radioactivity of acetylserine, it was calculated that ∼3% of the tRNASerCAG was attached with acetylleucine. These results were reproducible.
Digestion of purified acetyl‐aminoacyl tRNASerCAG with RNase T1 also gave only a 14C‐labeled CCA fragment, as shown in Figure 1C. When the acetylated amino acid released from the fragment purified from the corresponding spot on TLC was analyzed by TLC, the ratio of acetylleucine to acetylserine was also found to be 3% (Figure 4D), indicating that acetylleucine is covalently attached to the tRNASerCAG fragment with G73. It thus became clear that the tRNASerCAG of C.zeylanoides was in fact charged with leucine by 3% of the amount of serylation of the same tRNASerCAG in C.zeylanoides cells.
Incorporation of leucine is dependent on the CUG codon in C.maltosa
Aminoacylation has generally been considered to be the final stage determining translational accuracy (reviewed in Parker, 1989; Kurland, 1992; Farabaugh, 1993). However, in the case of tRNAGln charged with glutamate in the chloroplast, Glu‐tRNAGln is rejected by an elongation factor so that the chloroplast translation machinery does not employ the mischarged aminoacyl‐tRNA (Stanzel et al., 1994). It is likely that this is an exceptional case due to the lack of glutamyl‐tRNA synthetase in the chloroplast.
In order to prove that leucylated tRNASerCAGs actually participate in the translation process in Candida cells without such a rejection mechanism, we utilized a URA3 gene expression system derived from S.cerevisiae in C.maltosa, which was developed by Sugiyama et al. (1995). Candida maltosa utilizes the codon CUG as serine and possesses the relevant tRNASerCAG gene (Sugiyama et al., 1995; Zimmer and Schunck, 1995). Since the tRNASerCAG gene has G at position 37, G37 should be modified to m1G in tRNA, and tRNASerCAG may hence become chargeable with leucine in addition to serine in C.maltosa cells.
In the URA3 gene of S.cerevisiae, only one CTG codon appears, at the 45th position, and this is translated as leucine according to the universal genetic code, which is essential for the activity of orotidine 5′‐monophosphate decarboxylase (ODCase) (Rose et al., 1984; Sugiyama et al., 1995). In the present study, this URA3 gene, with the CTG codon replaced by various leucine or serine codons, was utilized as a marker gene (Figure 5A). First, a plasmid in which the S.cerevisiae URA3 gene was inserted downstream of a C.maltosa‐specific promoter (C‐p) was constructed and designated as pCSU–CTG (Sugiyama, 1995). As controls, mutant plasmids of pCSU–CTG, in which the codon CTG was replaced by either the serine codon TCT or the leucine codon CTC, were constructed and named pCSU–TCT and pCSU–CTC, respectively. In addition, a plasmid (pCCU) consisting of the URA3 gene of C.maltosa having a CTT leucine codon at the corresponding site, combined with the C.maltosa‐specific promoter, was also used as a positive control. These variant plasmids were introduced into a URA3‐defective C.maltosa strain CHU1 (his5, ade1, ura3::C‐ADE1/ura3::C‐ADE1) (Ohkuma et al., 1993), the growth of which was monitored on minimal medium SD plates in the presence and absence of uracil.
When uracil was supplied to the SD plate for the positive control experiments, all the transformants grew normally (Figure 5B, middle row). However, in the absence of uracil, cells harboring pCCU and pCSU–CTC showed normal growth, whereas no growth was observed in those harboring pCSU–TCT and pUTH18 that contained no URA3 gene insertion. Cells harboring pCSU–CTG showed weak but significant growth (Figure 5B, uppermost row). These results demonstrate that if the codon at position 45 is translated as leucine, active ODCase will be produced and the cells will be able to grow, but translation of the codon with serine will produce inactive ODCase and the cells will be unable to grow. The result with cells harboring pCSU–CTG clearly demonstrates that the URA3 mutation on the C.maltosa chromosome was in some way complemented by the introduced pCSU–CTG plasmid, suggesting that the CTG codon was read at least partially as leucine in C.maltosa cells possessing tRNASerCAG.
In order to quantify the growth rate of the cells harboring pCSU–CTG, the viability of the cells was examined in liquid medium without uracil. As shown in Figure 5C, whereas translation of the CTG codon as serine completely blocked cell growth in the case of pCSU–TCT, and full complementation was observed in the case of pCSU–CTC in which the CTC codon was read as leucine, intermediate cell growth was observed in the case of pCSU–CTG, indicating that ODCase was expressed in an active form, albeit at a low level, when there was a slight incorporation of leucine at the CTG codon. The slow growth of the cells harboring pCSU–CTG was not due to the spontaneous reversion of the CTG codon to another leucine codon or due to any other mutation, because the cells harvested from the colony on the SD‐plate show the same growth phenotype. These results are unlikely to reflect the different expression levels of the URA3 gene variants because the URA3 mRNA level is not altered by mutations at position 45 (Ohkuma, 1993). Furthermore, the possibility that the URA3 gene with CTG at position 45 is translated more efficiently than the gene with TCT at the same site due to codon preference (Ikemura, 1982) is excluded by the fact that the TCT codon is the most preferred of all the serine codons, including the CUG codon, in C.maltosa (Sugiyama et al., 1995).
ODCase activity resulting from the translation of the URA3 gene was examined in the presence of a pyrimidine analog, 5‐fluoroorotic acid (5FOA), an inhibitor in pyrimidine biosynthesis. Incorporation of 5FOA with ODCase results in the formation of 5‐fluorouridylate, which is harmful to cell propagation (Boeke et al., 1984). Thus, URA3‐defective strains grow normally on a medium containing 5FOA, whereas cells possessing the active URA3 gene are unable to grow on this medium. Cells harboring the respective plasmids were cultivated in the presence of 5FOA in addition to uracil.
As shown in the bottom row of Figure 5B, cells harboring pCSU–CTG exhibited similar growth on the agar plate to those with pCSU‐TCT and pUTH18, although the transformants with pCSU–CTC and pCCU were unable to grow. These results indicate that the CTG codon at position 45 was mainly translated as serine in C.maltosa, so as to produce the inactive ODCase. However, when the liquid medium was supplied with 5FOA, a slight reduction in the growth rate was observed in the case of pCSU–CTG, compared with pCSU–TCT (Figure 5C), while very slow growth was observed in the case of pCSU–CTC used as a control. In order to detect a low level of ODCase activity arising from a slight incorporation of leucine at the CUG codon in the 45th position, we adjusted the ratio of 5FOA and uracil as shown in Materials and methods. This growth rate reduction clearly suggests that the slow growth observed in the SD medium was due to low expression of active ODCase. Thus it is concluded that the CUG codon is partially translated as leucine in C.maltosa cells.
We believe that tRNASerCAG is the only molecule responsible for the leucine insertion corresponding to codon CUG in C.maltosa cells, based on the following obervations. We have purified and sequenced a number of leucine and serine tRNAs from Candida species, in which codon CUG is translated as serine, and failed in finding tRNA with the anticodon sequence potentially complementary to codon CUG other than tRNASerCAG (Yokogawa et al., 1992; Ohama et al., 1993; Suzuki et al., 1994; Ueda et al., 1994; our unpublished observation). Futhermore, tRNA genes for serine and leucine from these Candida species were sequenced following the amplification by cloning and/or PCR methods, and we found that only tRNASerCAG is able to translate codon CUG (Yokogawa et al., 1992; Ohama et al., 1993; Suzuki et al., 1994; Ueda et al., 1994; our unpublished observation). Thus, it could be concluded that only the tRNASerCAG species inserts leucine into polypeptide corresponding to codon CUG.
The observations presented here clearly demonstrate that, in certain living organisms, a single codon can be simultaneously assigned to two distinct amino acids. Most codons in the genetic code degenerate, but our findings show that some amino acids are also able to degenerate with respect to a particular codon. Such codon ambiguity is governed by a tRNA acceptable to two amino acids simultaneously, as described above. We propose to designate a codon corresponding to multiple amino acids a ‘polysemous codon’.
A high degree of accuracy in tRNA aminoacylation has been considered crucial for preserving fidelity in protein synthesis. It has been established that aminoacyl‐tRNA synthetase is able to discriminate precisely its cognate amino acid from other structurally related amino acids at the adenylation reaction step, and its cognate tRNAs from non‐cognate ones (reviewed in Parker, 1989; Kurland, 1992). The misacylation error in this process has been estimated to range between 10−4 and 10−5 (Lin et al., 1984; Okamoto et al., 1984). Discrimination of cognate tRNA from non‐cognate tRNAs is mediated by positive and negative identity determinants localized on the tRNA molecule (Yarus, 1988; Normanly and Abelson, 1989). The only exception reported so far is that tRNAGln is aminoacylated with glutamate in Gram‐positive bacteria and in some organelles (Lapointe et al., 1986; Schön et al., 1988). However, this differs from misaminoacylation in that this process is indispensable to compensate for the lack of glutamyl‐tRNA synthetase in these organisms. In general, high fidelity in the aminoacylation process is considered to be indispensable for translating genes into functionally active proteins with a high degree of accuracy.
The discovery of a polysemous codon in a Candida species contradicts the established notion of aminoacylation with high fidelity. We have shown that a single tRNA is acceptable to two different amino acids, and that it can therefore transfer two different amino acids corresponding to a particular codon. The expression experiment using the ODCase‐encoding URA3 gene containing codon CUG at the site essential for its activity (see also Sugiyama et al., 1995) suggested that leucine could be incorporated into the gene product corresponding to codon CUG in C.maltosa, as judged from the complementation tests with the URA3 mutation. Although the amount of leucine incorporated per CUG codon was not quantitatively determined, it is clear that the incorporation was mediated by the leucyl‐tRNASerCAG. We thus concluded that codon CUG was simultaneously assigned to serine and leucine in the normal translation process in C.maltosa. A quantitative analysis of the amino acids attached to the tRNA indicated that 3% of tRNASerCAG is leucylated in C.zeylanoides cells. Such a high level of leucylation is far beyond conventional misacylation, whose rate is estimated to be less than 10−4. Unless a proofreading mechanism exists on the ribosome, incorporation of leucine at CUG codon sites may reflect the relative ratio of tRNASerCAG leucylation, which is two orders of magnitude higher than that of conventional mistranslation.
To date, artificial manipulations of molecules participating in the translation process, such as the overproduction of aminoacyl‐tRNA synthetase (Swanson et al., 1988), mutations of tRNAs etc. and/or control of growth conditions, such as deprivation of amino acids in the medium (Edelmann and Gallant, 1977; O'Farrell, 1978; Parker and Precup, 1986), have been found to increase the error rate in translation (reviewed in Parker, 1989). However, our observation is based on experiments using wild‐type cells grown in a rich medium suitable for high viability. In these respects, the polysemous codon is a phenomenon completely different from these artificial translational errors. It is known that many examples exist for alternative decoding of universal codons—initiation codons other than AUG (Gold, 1988; Kozak, 1983), leaky stop codons caused by nonsense suppresser or native tRNAs (Murgola, 1985), the UGA codon used for incorporation of selenocysteine (Leinfelder et al., 1988) and so on. However, because of strong dependence on the context effects or possible secondary structures of mRNAs, these recoding events are those which are programed in the mRNAs (Gesteland et al., 1992). We have sequenced several genes in Candida genomes, but we could not find any secondary structure around the codon CUG in these genes. Considering that the polysemous codon is mediated by a single tRNA, it is unlikely that a polysemous codon occurs under the influences of the neighboring regions in mRNAs. Alternative decoding of a polysemous codon CUG is possible, assuming that LeuRS is overexpressed under a certain physiological condition. Depending on the increased amount of the LeuRS in cells, incorporation of leucine corresponding to codon CUG may occur frequently, which causes the production of polypeptides with new functions. This possibility should be examined in further experiments.
The idea of a polysemous codon also differs from the ‘near‐cognate’ concept proposed by Schultz and Yarus (1994). They claimed that ambiguous decoding may occur as a consequence of an irregular codon–anticodon interaction induced by the 27–43 base pair at the anticodon stem of the tRNA, resulting in a genetic code change transition state. The polysemous codon found in our study is caused by the tRNA aminoacylation process of tRNA with codon–anticodon interaction proceeding precisely in the conventional manner (Suzuki et al., 1994). Furthermore, since the hypothesis of Schultz and Yarus is based on experiments using an artificial mutation, and it does not reflect experimental observation in an extant living organism.
On the basis of peptide sequences, several research groups have reported that codon CUG corresponds only to serine in C.maltosa (Sugiyama et al., 1995) and C.albicans (Santos and Tuite, 1995a; White et al., 1995). No leucine‐inserted peptide was detected in these studies. However, we consider that any peptide with a leucine which was inserted for the codon CUG might have been missed during purification or was undetectable in the peptide sequencing, because the amount of leucine‐inserted peptide (∼3%) would have been too low to be positively identified in sequencing experiments.
We have shown that tRNASerCAG in Candida species is a chimera of tRNASerCAG and tRNALeuCAG in so far as it is the substrate for both SerRS and LeuRS. The Km value for LeuRS is 5.0 μM, which is only one order of magnitude larger than that for SerRS (0.22 μM). In an in vitro aminoacylation experiment >30% of tRNASerCAG subjected to the reaction could be converted to leucyl‐tRNASerCAG using an increased amount of LeuRS and a longer incubation time (data not shown). We observed that while the presence of SerRS and non‐radioactive serine reduced leucylation, complete loss of leucylation could not be achieved (data not shown), indicating that the affinity of LeuRS toward tRNASerCAG is relatively high.
In proliferating cells of C.zeylanoides, the leucyl‐tRNASerCAG in the cells was estimated to be 3% of the seryl‐tRNASerCAG, which is much lower than that obtained in the in vitro experiments. We consider that such a reduction in leucylation is due to the competition for the tRNASerCAG between SerRS and LeuRS in the cells. Despite this competition, the distinct detection of leucylated tRNASerCAG in vivo supports the existence of an ambiguous aminoacylation reaction toward the single tRNASerCAG species.
The polysemous codon results from the coexistence of tRNA identity determinants for serine and leucine in a single tRNA molecule. Construction of tRNASerCAG variants by the microsurgery method led to the finding that a single methyl moiety of m1G at position 37 is involved in the leucylation process. In contrast, the tRNASerCAG of C.cylindracea, which has A at the same position, is deprived of such leucine‐accepting activity. Himeno and his co‐workers noted that three nucleotides of leucine tRNAs were strongly recognized by S.cerevisiae LeuRS using unmodified variants transcribed by T7 RNA polymerase (Soma et al., 1996). Although the discriminator base, A73, is the strongest recognition site among them, A35 and G37 in the anticodon loop also play roles as determinants in tRNA. They were able to compare the activities of variants mutated at position 37 with A or G using the variants with A at the discriminator position which effectively elevates leucylation activity. In our work, we utilized serine tRNA with a modified nucleoside and with G at the discriminator position as a substrate for LeuRS, because the T7 transcript of tRNASerCAG showed no activity for leucylation. Our experiments using microsurgery methods indicated that m1G is of great importance in leucylation, despite the fact that the presence of G at the discriminator position is unsuitable for the recognition of LeuRS. Some modified nucleotides in tRNA are known to be involved in recoginition of some synthetases (Muramatsu et al., 1988). Pütz et al. (1994) showed that m1G at position 37 of yeast tRNAAsp is one of the negative determinants for arginyl‐tRNA synthetase.
We have also demonstrated that the nucleotide at position 33, where only tRNASerCAG uniquely possesses G, modulates the leucine‐accepting activity. G33 may prevent tRNASerCAG from excessive leucylation, which might be deleterious to the cells. Leuker and Ernst (1994) failed to express S.cerevisiae tRNALeu genes having anticodons UAG or CAG on a low‐copy (two or three copies) plasmid in C.albicans, indicating that such a high level of leucine residue incorporation into the CUG codon causes a lethal situation. In contrast, Santos and Tuite (1995b) reported that the C.albicans tRNASerCAG gene existing in a single copy plasmid could be expressed in S.cerevisiae cells, but that the viability of the cells decreased substantially. This finding suggests the polysemous state may be tolerated only when the ambiguous translation is under a strict constraint. We consider that G33 functions as a negative modulator in the leucylation of tRNASerCAG, thereby controlling the relative seryl‐ to leucyl‐tRNASerCAG ratio.
Several lines of experiment have suggested that U33 is involved in the tRNA function on ribosomes, such as in rigid codon–anticodon interaction, proper GTP hydrolysis of the ternary complex and the efficient translation of termination codons (Bare et al., 1983; Dix et al., 1986). Indeed, the replacement of G33 by U in C.cylindracea tRNASerCAG increased the efficiency of in vitro translation by 2‐ to 3‐fold (data not shown). The negative effect of G33 on translation may indicate involvement in some mechanism for decoding the polysemous codon. This possibility needs to be clarified by further study. Nevertheless, we have shown here that one of the roles of G33 is the suppression of leucylation, and we consider that the nucleotide at position 33 is not directly involved in recognition by LeuRS. On the basis of our observation that no leucylation was detectable in the C.cylindracea tRNASerCAG variants in which G33 was replaced by a pyrimidine base (c‐G33U and c‐G33C), we speculate that G33 influences the location and/or conformation of m1G37, accompanied by the alteration of the anticodon loop structure, decreasing the affinity of LeuRS toward tRNASerCAG.
It has been generally considered that reconstructed tRNA does not lose its activity during the several reaction steps needed in the microsurgery method, such as cleavage of the tRNA strand and ligation of tRNA fragments (Ohyama et al., 1985). However, a slight reduction of leucylation activity was observed in the control variant, z‐G33G, compared with that of the native tRNA (Figure 3A, right‐hand graph), which turned out to result from the partial deacetylation of 4‐acetyl cytidine (ac4C) due to acid treatment of the 5′‐half fragment (see Materials and methods). Nevertheless, it is reasonable to deduce the effect of base replacement on the aminoacylation activity by comparing the activities of these reconstructed tRNAs, because the same 5′‐half fragments were used for all the manipulated tRNA molecules of C.zeylanoides.
A plausible mechanism by which LeuRS could recognize cognate leucine and serine tRNAs specific for codon CUG is illustrated in Figure 6. LeuRS contacts and recognizes its cognate leucine tRNA from the 3′‐side of the anticodon loop, which is afforded by the uridine‐turn structure due to U33 (Figure 6A). The methyl moiety of m1G37 is directly recognized by LeuRS. In the case of C.zeylanoides, the anticodon loop distorted by G33 decreases the affinity toward LeuRS, judging from the observation that G33 increased the Km value for leucylation approximately 4‐ to 5‐fold in comparison with that with prymidine bases at the position (Figure 6B). In C.cylindracea, m1G is replaced by A, which means that the tRNA has lost the two major determinants for LeuRS, m1G and the discriminator base (Figure 6C). Consequently, LeuRS is unable to recognize tRNASerCAG at all, and G33 concomitantly loses its function as a modulator. LeuRS is, of course, unable to recognize other serine isoacceptor tRNAs corresponding to universal codons, because they have modified A at position 37.
How did this interaction between LeuRS and tRNASerCAG evolve? Candida species utilizing CUG as serine can be classified into two distinct groups: group 1 contains the species that have tRNASerCAG with leucylation activity, and includes C.zeylanoides, C.maltosa and others (Figure 6B); group 2, which is represented solely by C.cylindracea, contains species that have tRNASerCAG without leucylation activity (Figure 6C). A plausible evolutionary process is that group 1 would have arisen prior to group 2 after the genetic code change, which is speculated on the basis of the following observations. First, the homology between tRNASerCAGs in group 1 and its isoacceptor tRNAs for codon UCG is higher than that between the tRNASerCAG from C.cylindracea and its isoacceptor (Ueda et al., 1994). Second, C.cylindracea (group 2) possesses high copy numbers of the tRNASer CAG genes (∼20 copies) on the diploid genome (Suzuki et al., 1994), while low copy numbers (two or four copies) are observed for group 1 tRNASerCAG genes (Santos et al., 1993; Sugiyama et al., 1995; T.Suzuki, personal observations). Third, the codon CUG is utilized as a major serine codon on several genes in C.cylindracea, such as lipase (Kawaguchi et al., 1989) and chitin synthase (unpublished results), while CUG appears infrequently on the genomes of other species belonging to group 1 (Lloyd and Sharp, 1992; Sugiyama et al., 1995; T.Suzuki, personal observations). During the course of the change in the genetic code, the genome should pass through a state where the codon is under restrictive usage. Species with a high frequency of usage of the serine codon CUG are likely to be posterior to those with a lower frequency of usage. Furthermore, group 1 has a comparatively A+T‐rich genome which would be a disadvantage for the usage of codon CUG, whereas the opposite situation occurs in group 2 (Ohama et al., 1993). Fourth, the phylogenetic tree of these species and relatives constructed by using several genes also supports this evolutionary pathway (manuscript in preparation). Group 1, with a polysemous codon, thus seems to be closer to the point of genetic code change than group 2.
On the basis of their sequence similarity, Pesole et al. and ourselves have proposed that tRNASerCAG could have originated from the serine tRNA isoacceptor responsible for codon UCG (Ueda et al., 1994; Pesole et al., 1995). Thus, the nucleotide at position 37 seems likely to have mutated in the direction modified A→m1G(group 1)→unmodified A (group 2). This hypothesis raises two questions: (i) why did the mutation changing modified A to m1G, which give rise to a polysemous codon disadvantageous for the fidelity of protein synthesis, occur; and (ii) why was there a subsequent reversion in group 2? These questions can only be answered by clarifying the evolutionary significance of the polysemous codon.
Alternative splicing generates a multiple protein sequence from a single gene at the mRNA level. In contrast, a polysemous codon potentially results in diversification of the amino acid sequence of the protein derived from a single gene in the translation process. We have shown that the polysemous state of a codon is tolerated when the codon appears infrequently, as observed in group 1 species. Furthermore, our analyses of the nucleotide sequences of several genes of groups 1 and 2 suggest that codon CUG tends to appear in non‐housekeeping genes with high frequency (data not shown). Whereas a house‐keeping gene requires a high degree of fidelity in the translation process to maintain its function, which is essential for the fundamental mechanism of the cell, non‐housekeeping genes could allow the ambiguity of protein synthesis caused by a polysemous codon. We speculate that such ambiguity could have given rise to proteins with multiple amino acid sequences in non‐house‐keeping genes, which may have conferred multifunctionality on the proteins. Since the C.cylindracea strain was developed industrially for the production of lipase, such multifunctionality, caused by a polysemous codon, would not have been essential to cell growth, but might have been advantageous for the natural evolution of the cell. The advantage of a polysemous codon in evolution is of particular interest to us, and is a question which we are now addressing.
Materials and methods
[14C]acetic anhydride (185 MBq/mmol) was purchased from American Radiolabeled Chemicals. Uniformly labeled [14C]serine (5.99 MBq/mmol) and leucine (11.5 MBq/mmol) were from Amersham. 5‐fluoroorotic acid monohydrate (5FOA) was from PCR inc. 3′‐Biotinylated DNA probes were synthesized by Sci. Media, Japan. Synthetic RNA oligomers and a chimeric oligonucleotide composed of DNA and 2′‐O‐methyl RNA were synthesized by Genset Co. Ltd. Most of the enzymes used for the microsurgery were from Takara Shuzo (Tokyo, Japan). Other chemicals were obtained from Wako Chemical Industries.
Strains and media
Candida cylindracea (ATCC14830) was obtained from the American Type Culture Collection. Candida zeylanoides (JCM1627) was from the Japan Collection of Microorganisms (JCM, RIKEN, Wako, Saitama, Japan). Candida maltosa strain CHU1 (his5, ade1, ura3::C‐ADE1/ura3::C‐ADE1) was kindly provided by Drs M.Takagi and R.Ohtomo of the Department of Agricultural Chemistry, The University of Tokyo. Complete medium YPD (1% yeast extract, 2% Bacto‐peptone and 2% dextrose) and minimal medium SD [0.67% yeast nitrogen base without amino acids (Difco) and 2% dextrose] supplied with 24 mg/ml uracil were used for the cultivation of yeast cells.
SD‐plates with or without uracil were prepared by adding agar at a final concentration of 2% to each medium. 5FOA medium (1.4% yeast nitrogen base without amino acids, 4% dextrose, 20 mg/l uracil and 4 mg/ml 5‐fluoroorotic acid) was adjusted to evaluate the different viabilities of transformants with pCSU–TCT and pCSU–CTG in the presence of 5FOA. 5FOA plates consisted of 0.7% yeast nitrogen base without amino acids (Difco), 2% dextrose, 50 mg/l uracil, 1 mg/ml 5FOA and 2% agar, prepared according to the literature (Boeke et al., 1984).
Plasmid construction and transformation
In order to introduce mutation at the 45th codon in the reading frame of the S.cerevisiae URA3 gene, pUC‐CSU (Sugiyama et al., 1995) was first cut by EcoT14I so as to remove a fragment consisting of 80 bps, including the 45th CTG codon. A totally synthesized DNA fragment with a mutation of CTG to TCT at the 45th position was inserted into the vector in place of the removed fragment to construct pUC–CSU–TCT. The HindIII fragment of pUC–CSU–TCT consisting of the C.maltosa promoter and the S.cerevisiae URA3 coding region where the 45th position was mutated from CTG to TCT was inserted into the SmaI site of pUTH18 (Ohkuma et al., 1993) to construct pCSU–TCT. Plasmid pUTH18 containing an autonomously replicating sequence of C.maltosa (Takagi et al., 1986) and C‐HIS5 (Hikiji et al., 1989) were used as expression vectors developed from pUC18 by Ohkuma et al. (1993).
Other plasmids—pCSU–CTG, denoted as pCSU in the literature (Sugiyama et al., 1995), consisting of the C.maltosa promoter and S.cerevisiae URA3 coding region having one CTG codon at the 45th position, pCSU–CTC (Sugiyama et al., 1995) consisting of the C.maltosa promoter and S.cerevisiae URA3 coding region where the 45th position was mutated from CTG to CTC, and pCCU (Sugiyama et al., 1995) having the native URA3 of C.maltosa—were kindly provided by Drs M.Takagi and R.Ohtomo of the Department of Agricultural Chemistry, University of Tokyo. Transformation of C.maltosa was carried out by electroporation as follows. An overnight culture of C.maltosa CHU1 in YPD medium was harvested by centrifugation at 0°C. The cells were washed with 5 ml ice‐cold 1 M sorbitol solution and collected by centrifugation. The pellet was resuspended with 200 μl chilled 1 M sorbitol. A 50 μl aliquot of the cells mixed with an appropriate amount of the plasmid DNA was pulsed by an electroporator (Electroporator II, Invitrogen Corp.) under the conditions described in the manufacturer's instruction manual. The electrified cells were spread on a SD‐plate containing uracil and incubated at 30°C.
In vitro aminoacylation assay
Seryl‐ or leucyl‐tRNA synthetases were partially purified from C.zeylanoides cells as described previously (Suzuki et al., 1994), both of the activities having been completely separated by column chromatographies. The aminoacylation reaction was carried out at 37°C in 100 μl of reaction mixture containing 100 mM Tris–HCl (pH 7.5), 10 mM KCl, 50 mM NaCl, 15 mM MgCl2, 10 mM DTT, 2 mM ATP, 15 μM [14C]serine (5.99 GBq/mmol) or 15 μM [14C]leucine (11.5 GBq/mmol), 0.1–2.0 μM purified tRNA, and appropriate units of partially purified SerRS or LeuRS from C.zeylanoides. The reaction was started by addition of the enzyme to the pre‐incubated reaction mixture. After appropriate incubation periods, 25 μl of the reaction mixture was spotted onto a filtration paper (Wattman 3MM), which was washed with 5% TCA to remove the free [14C]amino acids. The radioactivity remaining on the filter was then measured by a liquid scintillation counter.
Large‐scale purification of tRNASerCAGs from C.zeylanoides and C.cylindracea
Candida cylindracea cells (3.1 kg) were treated with phenol, from which 150 000 A260 units of unfractionated tRNA were extracted. Eighty thousand A260 units of tRNA mixture were obtained by DEAE–cellulose chromatography with stepwise elution, which was then applied onto a DEAE–Sephadex A‐50 column (6×100 cm). Elution was performed with a linear gradient of NaCl from 0.375 to 0.525 M in a buffer consisting of 20 mM Tris–HCl (pH 7.5) and 8 mM MgCl2. The fraction rich in tRNASer was applied onto a RPC‐5 column (1×80 cm) and eluted with a linear gradient of NaCl from 0.4 to 1 M NaCl in a buffer consisting of 10 mM Tris–HCl (pH 7.5) and 10 mM Mg(OAc)2. As a result of these chromatographies, 300 A260 units of purified tRNASerCAG were finally obtained.
One hundred and fifty thousand A260 units of tRNA from C.zeylanoides cells (3.7 kg) were fractionated on DEAE–Sepharose fast‐flow column (3.5×130 cm) with a linear gradient of NaCl from 0.25 to 0.4 M in a buffer consisting of 20 mM Tris–HCl (pH 7.5) and 8 mM MgCl2. About 300 A260 units of C.zeylanoides tRNASerCAG were finally obtained by further column chromatography with Sepharose 4B in a reverse gradient of ammonium sulfate from 1.7 to 0 M with a buffer consisting of 10 mM NaOAc (pH 4.5), 10 mM MgCl2, 6 mM β‐mercaptoethanol and 1 mM EDTA.
Construction of tRNA variants with mutation at position 33
The microsurgery procedures were basically carried out according to the literature (Ohyama et al., 1985, 1986). Limited digestion of 4 mg purified tRNASerCAG from C.zeylanoides with RNase T1 was performed at 0°C for 30 min in a reaction mixture containing 50 mM Tris–HCl (pH 7.5), 100 mM MgCl2, 0.5 mg/ml of the tRNA and 25 000 units/ml RNase T1 (Sigma). After phenol extraction, the resulting fragments were treated with 0.1 N HCl at 0°C for 12 h in order to cleave the 2′, 3′ cyclic phosphate of the 3′‐end of the fragments formed in the limited digestion, and then the 5′‐ and 3′‐half fragments were separated by 10% PAGE containing 7 M urea (10×10 cm). Four hundred and thirty micrograms of the 5′‐half and 520 μg of the 3′‐half fragments were recovered from the gel. The purified 5′‐half fragment was dephosphorylated with bacterial alkaline phosphatase (Takara Shuzo), and G33 at the 3′‐end of the 5′‐half fragment was removed by oxidation with sodium periodate as described in the literature (Keith and Gilham, 1974). After dephosphorylation of the 3′‐end of this fragment, the truncated 5′‐half fragment was ligated with each of four synthetic tetramers—pUCAGp, pCCAGp, pACAGp or pGCAGp—with T4 RNA ligase (Takara Shuzo) at 10°C for 16 h in 200 μl of a reaction mixture containing 50 mM Tris–HCl (pH 7.5), 15 mM MgCl2, 3.5 mM DTT, 15 μg/ml BSA, 5% PEG6000, 480 μM ATP, 4.0 nmol of the 5′‐half fragment, 20 nmol of the tetramer and 200 units T4 RNA ligase. Ligation efficiency was >80%. Ligated fragments were separated from unligated ones by MonoQ column chromatography with a linear gradient of NaCl from 0.38 to 0.5 M in a buffer consisting of 20 mM Tris–HCl (pH 7.5) and 1 mM EDTA. The fragments were precipitated with isopropanol and dephosphorylated. The ligated 5′‐half and the phosphorylated 3′‐half were incubated at 65°C for 10 min and annealed at room temperature. The ligation was performed in 300 μl of a reaction mixture consisting of 58 mM Tris–HCl (pH 7.5), 17.5 mM MgCl2, 3.5 mM DTT, 10 μg/ml BSA, 50 μM ATP, 50 μg annealed tRNA and 200 units T4 RNA ligase. The 5′‐end of the ligated tRNA was phosphorylated and the 3′‐end was dephosphorylated simultaneously with T4 polynucleotide kinase (Toyobo) at pH 6.9, and then the 3′‐end was repaired at 37°C for 1 h with CCA enzyme partially purified from C.zeylanoides in 400 μl of a reaction mixture consisting of 50 mM Tris–HCl (pH 8.7), 10 mM MgCl2, 100 mM KCl, 1.75 mM ATP, 0.5 mM CTP, 2.5 mM DTT and an appropriate amount of the CCA enzyme. The reconstituted tRNAs were purified by 12% PAGE containing 7 M urea (0.5×20×40 cm). About 50 μg of each of the four variants mutated at G33 from C.zeylanoides tRNASerCAG were finally obtained; these were denoted as z‐G33U, z‐G33C, z‐G33A and z‐G33G, respectively. Transfer RNA variants mutated at position 33 from C.cylindracea tRNASerCAG were constructed. Limited digestion of 10 mg purified tRNASerCAG from C.cylindracea using RNase A was performed to obtain the 5′‐half and 3′‐half fragments. The dephosphorylated 5′‐half fragment was ligated with each of four kinds of dimer—pUCp, pCCp, pGCp or pACp—for mutation at position 33. Each of the ligated 5′‐half fragments was annealed with the phosphorylated 3′‐half fragment and ligated at the anticodon loop. Then the 3′‐end of each variant was repaired with T4 polynucleotide kinase and the CCA enzyme as described above. About 65 μg of each of the purified variants of C.cylindracea tRNASerCAG mutated at position 33 was obtained; these were denoted as c‐G33U, c‐G33C, c‐G33A and c‐G33G, respectively.
Construction of tRNA variants mutated at position 37
To obtain the mutation of m1G37 of C.zeylanoides tRNASerCAG, we selected the restrictive digestion technique with RNase H using a chimera oligonucleotide splint composed of DNA and 2′‐O‐methyl RNA as developed by Inoue et al. (1987). We designed a splint oligonucleotide according to the literature (Hayase et al., 1990) which was designated as CZE‐37 (5′GmCmCmCmAmAmUmGmGmAmAmdCdCdTdG–CmAmUmCmCmAmUm3′), possessing a cleavage site between positions 37 and 38 of C.zeylanoides tRNASerCAG. Two hundred micrograms of purified tRNASerCAG from C.zeylanoides was incubated at 65°C for 10 min with 14.4 nmol CZE‐37 in a buffer consisting of 40 mM Tris–HCl (pH 7.7), 0.5 mM NaCl, 0.1 mM DTT, 0.0003% BSA and 0.4% glycerol (500 μl), and then annealed at room temperature. Magnesium chloride was added to the mixture up to a final concentration of 4 mM and the reaction was carried out at 30°C for 2 h by the addition of 600 units of RNase H (Takara Shuzo). About 60 μg of the cleaved 3′‐half fragment was obtained by purification using 10% PAGE containing 7 M urea. Either of two synthetic oligo‐RNAs, pCAGAp or pCAGGp, was ligated with the same 5′‐half fragment digested by RNase T1 as the variants mutated at position 33 under the conditions described above. The ligated and dephosphorylated 5′‐half fragments were annealed and ligated with the 3′‐half fragment digested by RNase H. About 50 μg of each of the two variants from C.zeylanoides mutated at position 37—m1G37A and m1G37G—was obtained by the phosphorylation of the 5′‐end and purification by 12% PAGE containing 7 M urea.
Identification of amino acids attached to tRNASerCAGs in the cells
Identification of aminoacyl‐tRNASerCAG from Candida cells was carried out by a new method developed recently by us (Suzuki et al., 1996). The experimental conditions were the same as those reported. To fish out the aminoacyl‐tRNAs, we designed two 3′‐biotinylated DNA probes: 5′AGCAAGCTCAATGGATTCTGCGTCC3′ for C.cylindracea tRNASerCAG and 5′GAAGCCCAATGGAACCTGCATCC3′ for C.zeylanoides tRNASerCAG. These probes were immobilized with streptavidin agarose (Gibco BRL) as reported previously (Wakita et al., 1994).
The authors are grateful to Drs M.Takagi and R.Ohtomo of the Department of Agricultural Chemistry, the University of Tokyo, for providing the C.maltosa strain and plasmids, Dr H.Sugiyama of the Brewing Research Laboratories, Sapporo Breweries, for valuable discussions, and Dr M.Hasegawa of the Institute of Statistical Mathematics, for encouragement and valuable comments. Special thanks are also due to Dr S.Osawa of the Biohistory Research Institute and Dr Thomas H.Jukes of University of California, Berkeley for critical reading of the manuscript and valuable suggestions, Dr H.Himeno of Hirosaki University and Drs M.A.S.Santos and M.F.Tuite of the University of Kent for communicating their experimental results to us before publication and for useful suggestions, and Drs N.Nishikawa and T.Yokogawa of Gifu University and S.Asakawa of Keio University for their technical advice. This work was supported by Grant‐in Aid for Scientific Research on Priority Areas from Ministry of Education, Science, Sports, and Culture, Japan and by a JSPS Fellowship for Japanese Junior Scientists to T.Suzuki.
- Copyright © 1997 European Molecular Biology Organization