ABSTRACT
By replacing a stretch of five A-U base pairs in the acceptor stem with G-C pairs, mitochondrial tRNASerGCU lacking a D arm could be expressed in Escherichia coli cells in considerable amounts. The expressed tRNA with no modified nucleoside was serylated in vitro with the mitochondrial enzyme. The tRNASerGCU derivatives carrying identity elements for alanine tRNA and the related anticodons were expressed. However, this expression event did not affect cell growth, probably because the expression started from the late log phase, which suggests that these mitochondrial tRNA derivatives are not involved in E.coli gene expression systems. Although there are some restrictions in the secondary structure of tRNAs that can be expressed by this method, it could prove useful for preparing large amounts of heterologous tRNAs in vivo.
A number of animal mitochondrial (mt) tRNAs exhibit unique structural features: a lack of invariant or semi-invariant sequences such as GG in the D loop or T[Psi]CG in the T loop; the presence of many more A-U pairs than standard tRNAs in the stem regions (1 ,2 ), and in certain cases, the absence (or truncation) of a D arm or T arm, which appear to be important for maintaining the L-shaped tertiary structure in canonical tRNAs (3 -5 ). Elucidation of the structure-function relationships existing in these non-canonical tRNAs should be valuable, therefore, in improving our understanding of the structural basis of tRNA functions.
However, this work is seriously hampered because only the scanty amounts of mt tRNAs can be prepared from animal mitochondria and it is often difficult to procure suitable organs in sufficient numbers. We therefore decided to attempt the overproduction of mt tRNAs in Escherichia coli cells as a means of obtaining larger quantities. The wild-type tRNASerGCU has been shown not to be expressed in E.coli (Taupin et al., 15th International tRNA Workshop, Cap d'Agde, May 30-June 4, 1993), apparently because of the conformational instability of the tRNA. We therefore conducted expression experiments with tRNA derivatives possessing a more stable secondary structure. As we previously demonstrated that replacing five A-U pairs in the acceptor stem of mt tRNASerGCU by G-C pairs did not affect its aminoacylation by homologous seryl-tRNA synthetase (SerRS) (6 ), we decided to employ this pair replacement strategy as a means of stabilizing the tRNA.
The tRNA derivatives thus constructed were found to be expressed in vivo due to the G-C richness in the acceptor stem, which probably causes both correct 3'-processing and stabilization of the tRNA molecules. The strategy employed could prove useful for preparing large amounts of heterologous tRNAs in vivo.
Plasmid vector pUC19 with the mt tRNASerGCU gene insert (pT7AU5) was prepared as reported previously (6 ,7 ). The plasmid of the derivative (pT7GC5) (for derivative abbreviations, see the legend to Fig. 1 ) was constructed in the same manner as pT7AU5. The terminator of T7 RNA polymerase was obtained from pET15b (Novagen) and inserted into the BamHI-HindIII sites of each plasmid using T4 DNA ligase (Takara Shuzo). The resulting two plasmids were named pT7AU5T and pT7GC5T, respectively. The plasmids for the other tRNA derivatives described were all constructed and transformed in the same manner as pT7GC5T.
Production of the wild-type and derivative tRNAs was carried out by growing a 100 ml culture of an expressing clone at 28oC in LB medium supplemented with 100 mg/ml ampicillin. When the culture reached 0.5 A600, expression of the cloned tRNA gene was induced by adding IPTG to a final concentration of 1 mM followed by incubation at 37oC. After 3 h incubation, E.coli cells were harvested by centrifugation and resuspended in 10 ml of 4 M guanidium thiocyanate, 25 mM sodium cianate (pH 7.0), 0.5% sodium lauril sarkocinate and 0.1 M [beta]-mercaptoethanol. The RNA fraction was extracted by phenol treatment followed by ethanol precipitation according to the usual method. tRNASerGCU was isolated by the selective hybridization method using a solid-phase DNA probe (8 ). The oligodeoxyribonucleotides (22mer) used for the probes (purchased from Biologica) were 5'-AGCAGTTCTTGCATACTTTTTC-3' and 5'-AGCAGTTCTTGCATACCCCCCC-3', which were complementary to the region extending from the 5'-acceptor stem to the 5'-anticodon stem of the native tRNASerGCU and tRNASer(GC5) genes, respectively.
The nucleotide sequence of the expressed tRNA was determined by Donis-Keller's method (9 ). Nucleoside analysis was carried out using a photodiode-array detector (SPD-M10A, Shimadzu) as described previously (10 ).
The amino acid acceptor activity of the expressed tRNAs with the E.coli enzyme (11 ) or mt SerRS (6 ) was assayed as described previously, using a [14C]amino acid mixture (1.9 GBq/mgAtom), l-[U-14C]alanine (5.6 GBq/mmol) or [14C]serine (5.8 GBq/mmol), all of which were purchased from Amersham.
RNA fractions extracted from BL21(DE3) cells harboring plasmid pT7GC5T or pT7GC5AAT at various times after induction were analyzed by electrophoresis on 8% polyacrylamide gel with 7.5 M urea and 20% formamide. After the gel was stained by toluidine blue, each band strength on the gel was monitored by 660 nm light using a densitometer (Beckmann DU, 640).
All the transcripts used in this study were prepared as described previously (7 ). tRNA was separated by electrophoresis on 8% polyacrylamide-7 M urea-30% formamide gel in TBE (0.09 M Tris-borate and 2 mM EDTA, pH 7.5) buffer.
pT7AU5 and pT7GC5 linearized by HindIII were used as templates for the in vitro transcription reactions with T7 RNA polymerase. The transcribed tRNA precursors were purified by electrophoresis on 10% polyacrylamide gel containing 7 M urea and 20% formamide. Each of the precursors contained 35 extra nucleotides at the 3'-end. The condition of the processing assay was as described by Cudny et al. (12 ). The reaction mixture was run on 8% polyacrylamide gel containing 7.5 M urea and 30% formamide.
The sequences of bovine mt tRNASerGCU and its derivatives tRNASerGCU(GC1) to tRNASerGCU(GC5) [hereafter abbreviated to tRNASer, and tRNASer (GC1) to tRNASer(GC5), respectively] described in this study are shown in Figure 1 . Synthetic genes for these mt tRNAs with the IPTG inducible T7 promoter at their 5'-end were constructed (6 ), and the terminator for the T7 RNA polymerase obtained from pET15b was attached to each at the 3'-end for expression of the tRNA genes in vivo. The plasmids were named pT7AU5T for the wild-type tRNASer, and pT7GC1T to pT7GC5T for the respective derivatives tRNASer(GC1) to tRNASer(GC5).
With plasmids pT7AU5T and pT7GC5T, 4.0 mg tRNA mixture was recovered from 1.2 g E.coli cells harboring each plasmid. Although there was no expression of the wild-type mt tRNASer (Fig. 2 A), a large amount of tRNASer(GC5) with seven G-C pairs in the acceptor stem was expressed. The band position detected by hybridization was the same as that of the in vitro transcript of the tRNA derivative, indicating that the product was processed to the actual tRNA size (Fig. 2 B). About 0.4 mg of tRNASer(GC5) expressed in the cells was isolated by the selective hybridization method (8 ) and its nucleotide sequence was confirmed by Donis-Keller's method (9 ). HPLC analysis showed no modified nucleoside in the tRNA derivative (data not shown).
To investigate the effects of plasmids pT7AU5T and pT7GC5T on the growth of BL21(DE3) cells, the relevant genes were induced by addition of IPTG when the density of the cultures reached 0.5 A600. Unlike in other tRNA expression systems, induced cells carrying either plasmid continued to grow at the same rate as cells carrying no plasmid (Fig. 3 A, curves 1-3).
To determine whether the expressed tRNA possessed amino acid acceptor activity, the Km values for serylation were measured for tRNASer(GC5), as well as for the wild-type tRNASer as a reference. As shown in Table 1 , the tRNASer(GC5) expressed in vivo could be serylated with mt SerRS in a similar way to tRNASer(GC5) that was transcribed in vitro. The Km values for both of these tRNAs were several times higher than that of the wild-type tRNASer (6 ,14 ). In contrast, none of these tRNAs was aminoacylated with E.coli cell extracts (S100) (15 ). It was thus evident that the mt tRNASer derivative was processed correctly, but not serylated in E.coli cells.
Table 1
To elucidate the relationship between the structure and the processing capability of mt tRNAs, mt tRNASer precursors for the wild-type and tRNASer(GC5), both of which contained 35 extra nucleotides in their 3' termini, were incubated with E.coli S100 (Fig. 4 ). A major product band corresponding to the mature tRNA was observed only in the case of the tRNA derivative. On the other hand, the wild-type tRNA precursor was hardly processed to the mature tRNA, and was degraded within 60 min. This observation strongly suggests that 3'-processing enzymes require a substrate tRNA possessing a G-C-rich acceptor stem.
To determine the minimum number of G-C pairs necessary for the 3'-processing of tRNASer derivatives, plasmids pT7GC1T, pT7GC2T, pT7GC3T and pT7GC4T were constructed in which the A-U pairs in the acceptor stem were replaced by one to four G-C pairs (Fig. 1 ) and introduced into E.coli BL21(DE3) cells in which the expression of the tRNA derivatives was examined (Fig. 5 ). All the tRNASer derivatives except for tRNASer(GC1) were found to be expressed in the cells. Since the amount of expressed tRNA increased exponentially with the number of G-C pairs, it is very likely that at least three G-C pairs at the top of the acceptor stem are necessary for the 3'-processing of tRNASer derivatives in E.coli cells.
Since the identity elements of E.coli tRNAAla are known to be only the G3-U70 base pair and A73 (17 ,18 ), we considered it possible that the introduction of alanine identity elements into the derivatives of mt tRNAs expressed in E.coli cells would enable the tRNAs to be charged with alanine even in E.coli cells. Experiments were therefore planned to determine whether cell growth was influenced by plasmids containing the mt tRNASer(GC5) gene in which an alanine identity element (G3-U70 in the standard numbering) was introduced and the anticodon was replaced by one for alanine or amber suppressor tRNA.
tRNASer(GC5A) is a tRNASer(GC5) derivative in which the base pair A3-U57 (U70 in the standard numbering) and the discriminator base G60 (G73 in the standard numbering) were replaced by G3-U57 and A60, respectively, and the serine anticodon GCU was replaced by the suppresser anticodon CUA. Another tRNASer(GC5) derivative in which the anticodon of tRNASer(GC5A) was changed to the alanine anticodon GGC was named tRNASer(GC5AA) (Fig. 1 ).
No inhibition of cell growth was observed in E.coli cells harboring either of the plasmids carrying genes for the derivatives tRNASer(GC5A) and tRNASer(GC5AA), even 8 h after induction (Fig. 3 A), although staining with toluidine blue (Fig. 3 B) and Northern hybridization analysis (data not shown) clearly demonstrated that these tRNAs were actually produced in the E.coli cells. The cell growth and amounts of the tRNAs expressed were approximately the same as those for tRNASer(GC5), which possessed no alanine identity element.
There have been several reports on the expression in E.coli cells of homologous (20 -24 ) and heterologous tRNAs (25 ), as well as one on a minihelix derived from E.coli tRNAGly (26 ). However, these tRNAs had a normal structure or formed part of the E.coli tRNA. In the present work, we found that derivatives of bovine mitochondrial tRNASerGCU lacking a D arm could be produced in E.coli cells by introducing a stretch of G-C base pairs into the acceptor stem of the tRNA gene loaded onto the expression vector plasmid, and that the expressed tRNAs were chargeable with mt seryl-tRNA synthetase.
Four conditions need to be met for a heterologous tRNA gene to be expressed in E.coli BL21(DE3) cells: (i) a plasmid containing the tRNA gene should be maintained in the cells, (ii) the tRNA gene should be efficiently transcribed in vivo by T7 RNA polymerase, (iii) the expressed precursor should be recognized by processing enzymes in order to be converted to the mature tRNA, and (iv) the tRNA should be stable enough to remain in the cells. Both the plasmid carrying the wild-type tRNA gene (tRNASer) and that carrying the derivative [tRNASer(GC5)] were present in the cells in similar amounts as judged from the band density of each plasmid on gel electrophoresis (data not shown). Also there appeared to be no appreciable difference in the efficiencies of the in vitro transcription reactions of mt tRNASer and tRNASer(GC5), despite a report that T7 RNA polymerase prefers a G- or C-rich sequence downstream of the promoter in the in vitro transcription reaction (27 ). With respect to the third condition, it was demonstrated that even mt tRNA precursors having unusual secondary structures could be processed normally. The in vitro 3' processing assay also showed that while deletion of the D arm did not influence the processing, some G-C pairs in the acceptor stem were indispensable. It was also found that the more G-C pairs there are in the acceptor stem, the more efficiently the tRNA is expressed (Fig. 5 ), which is probably related to the stability of the tRNA.
The expressed tRNASer(GC5) was found to be chargeable in vitro with mt SerRS, but not with E.coli S100, which is consistent with the findings of our previous work (15 ). However, once the identity determinant for E.coli AlaRS was introduced into tRNASer(GC5), the tRNA became chargeable with alanine by E.coli S100. These results clearly indicate that mt tRNAs expressed in E.coli cells are folded into the active form in their tertiary structures, as is the case with canonical tRNAs.
It is intriguing that cell growth was never influenced by the expression of the mt tRNAs, irrespective of their charging activity with E.coli S100. This behavior differs from that of expressed minihelices derived from E.coli tRNAGly, which compete with tRNAGly for glycine charging and cause inhibition of cell growth (26 ). The reason may lie in the fact that both mt tRNASer(GC5) and tRNASer(GC5AA) start to be expressed in E.coli cells only from the very late log phase (almost the stationary phase) (Fig. 3 A), which suggests that there may be some unknown mechanism that delays the expression of heterologous tRNA genes until the cells have nearly reached the stationary phase so that the expressed tRNAs do not hinder the E.coli gene expression systems, including the translation system. Further studies are needed to reveal this postulated mechanism.
It has already been determined that although seryl-tRNASerGCU of bovine mitochondria is able to bind to bacterial EF-Tu to form a ternary complex with GTP (28 ), it cannnot be transferred to the ribosomal A site (29 ; Yokogawa et al., unpublished data). Therefore, it is evident that even when alanyl-tRNASer(GC5A) or alanyl-tRNASer(GC5AA) is present in E.coli cells (which it has been shown to be), the tRNA would be non-functional in the E.coli translation system.
The system for the heterologous expression of tRNAs in E.coli cells described here may be applicable to various fields of study. As the expressed tRNAs are not modified, they may be good substrates for tRNA-modification enzymes. Furthermore, since even non-canonical tRNAs lacking a D arm could be expressed in vivo, the system may be a useful means of providing various kinds of tRNA substrates for tRNA-processing enzymes, especially 3'-processing enzymes, which have so far been little studied in comparison with 5'-processing enzymes (30 ,31 ). It could also provide a powerful method for preparing larger amounts of RNA samples, especially for use in structural analysis such as by X-ray crystallography or NMR spectroscopy. Though the system has some limitations in terms of the RNA structures that can be obtained, a sufficient amount of RNA (~0.3 mg tRNA per 1 g cells) can be recovered. It should be also noted that since the system works in minimal medium, it could be a useful method for preparing many kinds of labeled RNA molecules.
We thank Drs T. Nojima and T. Yokogawa for providing the E.coli S100 fraction and mt SerRS, respectively, and for helpful discussions, and Prof. S. Yokoyama for allowing us to use facilities in his laboratory at the Institute of Physical and Chemical Research (RIKEN) for some of the experiments. This work was supported by Grants-in-aid for Scientific Research on Priority Areas from the Ministry of Education, Science, Sports and Culture of Japan and by the Human Frontier Science Program Organization.
*To whom correspondence should be addressed. Tel/Fax: +81 3 5800 6950; Email: kw@kwl.t.u-tokyo.ac.jp
tRNA derivative
Km ([mu]M) for serylation with mt SerRS
Bovine mt tRNASerGCU isolated from mitochondria
0.178a
Bovine mt tRNASerGCU transcribed in vitro
0.189
tRNASer(GC5) transcribed in vitro
1.04
tRNASer(GC5) expressed in vivo
1.07
tRNA derivative
Km ([mu]M) for serylation with E.coli S 100
tRNASer(GC5A)
6.1
tRNASer(GC5AA)
6.2
Escherichia coli tRNAAla
2.1b
Escherichia coli tRNAAla minihelix
7.5a
REFERENCES