Pleiotropic effects of intron removal on base modification pattern of yeast tRNA^Phe: an in vitro study

Abstract

Cell-free yeast extract has been successfully used to catalyze the enzymatic formation of 11 out of the 14 naturally occurring modified nucleotides in yeast tRNA^Phe (anticodon GAA). They are m²G₁₀, D₁₇, m²₂G₂₆, Cm₃₂, Gm₃₄, Ψ₃₉, m⁵C₄₀, m⁷G₄₆, m⁵C₄₉, T₅₄ and Ψ₅₅. Only D₁₆, Y₃₇ and m¹A58 were not formed under in vitro conditions. However, m¹G₃₇ was quantitatively produced instead of Y₃₇. The naturally occurring intron was absolutely required for m⁵C₄₀ formation while it hindered completely the enzymatic formation of Cm₃₂, Gm₃₄ and m¹G₃₇. Enzymatic formation of m²₂G²⁶, Ψ₃₉, m⁷G₄₆, m⁵C₄₉, T₅₄ and Ψ₅₅ were not or only slightly affected by the presence of the intron. These results allow us to classify the different tRNA modification enzymes into three groups: intron insensitive, intron dependent, and those requiring the absence of the intron. The fact that truncated tRNA^Phe consisting of the anticodon stem and loop prolonged with the 19 nucleotide long intron is a substrate for tRNA: cytosine-40 methylase demonstrates that the enzyme is not only strictly intron dependent, but also does not require fully structured tRNA.

Introduction

Formation of modified nucleotides in transfer RNA is an important part of the complex multistep maturation process which starts immediately after transcription of the corresponding gene by RNA polymerase III. In eukaryotic cells, the modification of nucleotides in tRNA begins in the nucleus and continues further in the cytoplasm. For those tRNAs that are transported into the cellular organelles, additional nucleotide modifications may also occur within these compartments (for review see 1,2). For those tRNA precursors that contain an intron, an additional maturation step consists in intron removal by a specific tRNA splicing machinery located in the inner part of the nuclear membrane, just before the tRNA is transported to the cytoplasm through the nuclear pore (for review see 3). As a result, most nucleotide modifications in these intron-containing tRNAs occur before intron removal (4–6). Since the presence of an intron may change significantly the folding and the 3D-architecture of the tRNA molecule, at least in the anticodon stem-loop region, one might expect that the formation of modified nucleotides, normally present in the anticodon stem-loop of fully matured tRNA molecules, may be influenced positively or negatively by the presence of an intron. Indeed, formation of several modified nucleotides, for example Ψ₃₅ in eukaryotic tRNA^Tyr (6–11), Ψ₃₄ and Ψ₃₆ in yeast tRNA^Ile (12) and m⁵C₃₄ in yeast tRNA^Leu (13) have been reported to be strictly intron dependent. On the other hand, enzymatic formation of several modified nucleotides, such as Gm₁₈, Cm₃₂, Ψ₃₂, Gm₃₄, Q₃₄, galQ₃₄, m¹G₃₇, Y₃₇, i⁶A₃₇, Um₄₄, was shown to be hindered by the presence of an intron and to take place only after intron excision (5,7,13–18), reviewed in (19). Interestingly, the introduction of m2 2G at position 26 (20) or m¹A at position 9 (Dr C.Florentz, personal communication) in certain tRNA molecules, seems to influence the folding pathway towards the correct cloverleaf tRNA structure. Thus, the observed stepwise addition of modified nucleotides may result in part from sequential structural reorganization of the 2D/3D conformation of the tRNA molecules during the complex maturation process (discussed in 21). [Symbols and common abbreviations of modified nucleotides are those of (22). Universal numbering system for tRNA positions corresponds to that used in (23).]

Fully matured yeast tRNA^Phe(GAA) contains 14 modified nucleosides out of a total of 76 nucleotides (nt) (18.5%), respectively m²G₁₀, D₁₆, D₁₇, m²₂G₂₆, Cm₃₂, Gm₃₄, Y(Wybutosine)₃₇, Ψ₃₉, m⁵C₄₀, m⁷G₄₆, m⁵C₄₉, T₅₄, Ψ₅₅ and m¹A₅₈ (23) (Fig. 1A). The primary transcript of yeast tRNA^Phe gene contains a 19 nt intron in the anticodon loop in addition to its 5′- and 3′-nucleotide extensions (24) (Fig. 1B). The overall folding of precursor tRNA^Phe in solution was studied by chemical and enzymatic probing (25–27) and by NMR (28).

In the present work we used yeast tRNA^Phe(anticodon GAA) as a model for studying, in detail, the relation between the presence of an intron and the sequential addition of modified nucleotides. To avoid the problems related to transcription of tRNA genes in vivo and to focus specifically on the tRNA maturation processes that involve splicing and nucleotide modifications, we used the yeast extract and T7 transcripts of synthetic tRNA^Phe genes as substrates. Our data demonstrate that some modification reactions are negatively affected by the presence of the intron, the synthesis of many other modified nucleotides is not influenced by the intron, while the enzymatic formation of m⁵C at position 40 in the anticodon stem is strictly dependent on the presence of the intron. Enzymatic formation of m⁵C₄₀ is independent of the whole architecture of the tRNA molecule and takes place on a minisubstrate composed of the anticodon stem and loop of yeast tRNA^Phe prolonged by its natural 19 nt intron. These results demonstrate that the intron bears at least part of the recognition elements for methylation of cytosine-40.

Figure 1

Cloverleaf structures of naturally modified S.cerevisiae tRNA^Phe(GAA) (A) and its intron-containing precursor (B). The numbering of positions in tRNA molecule as in (23), abbreviations of modified bases as in (22). Arrows (B) show the sites of excision of intron (shown in lower case letters). Thick bar in intron-containing tRNA (B) designates the anticodon (GAA).

Materials and Methods

Chemicals and enzymes

α-³²P-radiolabelled nucleotide triphosphates (400 Ci/mmol) were from Amersham (UK). Tris, DTE and DTT, nucleoside triphosphates, Penicillium citricum nuclease P1, Aspergillus oryzae RNAse T2, phenylmethylsulfonylfluoride (PMSF) and spermidine were from Sigma (USA). Diisopropylfluorophosphate (DFP) and S-adenosyl-L-methionine (SAM) were from Boehringer- Mannheim (Germany). Bacteriophage T7 RNA polymerase and restriction enzymes were from MBI Fermentas (Vilnius, Lithuania), RNasin was from Promega (USA). Thin layer cellulose plates were from Schleicher & Schuell (Germany). Ultra pure ammonium sulphate was from BRL (USA). All other chemicals were from Merck Biochemicals (Germany).

Plasmids

The plasmids designated p67YF0 and pUC13Phe carrying respectively the synthetic gene of yeast tRNA^Phe(anticodon GAA) (gift of Dr O.Uhlenbeck, Boulder, CO, USA) and of its intron-containing precursor (pre-tRNA^Phe, gift of Dr J.Abelson, Pasadena, CA, USA) were both described in detail elsewhere (29–31). The DNA template for T7 transcription of the introncontaining minisubstrate (anticodon stem-loop prolonged with the 19 nt long natural intron) was prepared by PCR amplification of the corresponding sequence present in the plasmid pUC13Phe using two oligonucleotides, respectively: TAATACGACTCACTATAgccagactgaagaaaaaac and TGGccagataacttgaccg (capital letters correspond to nucleotides of the T7 promoter in the first oligonucleotide and the half of the MvaI restriction site in the second oligonucleotide, while lower case letters in both oligonucleotides correspond respectively to sense 5′ end and antisense 3′ end nucleotides of the intron). The minisubstrate composed of the anticodon stem-loop of tRNA^Phe was prepared by T7 transcription using synthetic double strand DNA template. PCR amplification and DNA fragment recovery were performed using standard techniques as described in (32).

In vitro transcription of tRNA genes

Transcription was done using ∼1 µg of the MvaI linearized plasmid (2 pmol in the case of the DNA fragment corresponding to the PCR product of the stem-loop and intron) with 50–100 µCi (100 µM final) of one of the [α-³²P]NTP in 10 µl of reaction mixture containing 40 mM Tris-HCl, pH 7.9, 6 mM MgCl₂, 10 mM DTE, 2 mM spermidine, 1 mM each of the other NTPs. The reaction was initiated by the addition of 20 U T7 RNA polymerase, followed by 2–3 h incubation at 37°C. The resulting runoff transcription product was purified by electrophoresis on a 6% denaturing PAAG and recovered as described elsewhere (18).

Yeast S100 extract

The Saccharomyces cerevisiae strain PP1001 (αpep4-3, ade2, leu2; a gift from Dr K.Straby, University of Umea, Sweden) was grown overnight at 28°C in YPD liquid medium supplemented by 0.004% w/v adenine and 0.01% l-leucine. All of the following operations were carried out at 4°C. The cells were harvested by low speed centrifugation and resuspended in twice their weight of lysis buffer [50 mM Tris-HCl, pH 7.5, 100 mM KCl, 0.1 mM EDTA, 10 mM MgCl₂, 10% glycerol, 10 mM β-mercaptoethanol, 1 mM phenylmethylsulfonylfluoride (PMSF), 1 mM diisopropylfluorophosphate (DFP)]. The frozen cell suspension was passed through a French Press at ∼4000−5000 p.s.i. and the resulting homogenate was centrifuged at 12 000 g for 20 min to remove cell debris (S10 extract). The supernatant was centrifuged for 20 min at 30 000 g (S30 extract) followed by 1 h at 100 000 g in a Beckman TL-100 ultracentrifuge. The resulting supernatant (S100 extract at ∼10−12 mg total protein/ml) was aliquoted and quickly frozen in liquid nitrogen. Aliquots were kept at −80°C until used.

Enzymatic assays

The standard reaction mixture for assaying the formation of modified nucleotides in tRNA transcripts was composed of 100 mM Tris-HCl, pH 8.0, 100 mM ammonium acetate, 5 mM MgCl₂, 2 mM DTT, 0.1 mM EDTA, 20 µM SAM, 50–100 fmol (5–10 nM) of ³²P-labelled tRNA substrate in a final volume of 10 µl. Reaction was initiated by the addition of S100 yeast extract to a final concentration of ∼0.5–1.0 mg of total protein/ml. After incubation at 30°C, the reaction mixture was mixed with denaturing loading buffer, heated 2 min at 65°C and directly applied on 6% denaturing PAAG. The radioactive tRNA bands, as revealed by autoradiography of the gel, were eluted overnight, precipitated by ethanol and redissolved in appropriate buffer for further hydrolysis with nuclease P₁ (0.02 µg/µl) or RNAse T2 (0.01 U/µl) as described earlier (33).

Analysis of modified nucleotides

Identification of the ³²P-labelled nucleotides in tRNA hydrolysates was performed by two-dimensional chromatography on thin layer cellulose plates. Two chromatographic systems were used: in both cases, the first dimension was developed with isobutyric acid:ammonia 25%:water (66:1:33/v:v:v), the while second dimension (in system ‘N/R’) was developed with 0.1 M sodium phosphate pH 6.8:solid ammonium sulphate:n-propanol or (system ‘N/N’) with 2-propanol:HCl 37%:water (68:17.6:14.4/v:v:v). Radioactive spots were quantified using a PhosphoImager instrument (Molecular Dynamics, USA) with integrated software ImageQuant. Calculations of molar amount of modified nucleotide per mol of tRNA substrate were done taking into account the nucleotide composition of tRNA transcript. The experimental error was estimated to be ∼0.05 mol/mol of tRNA. These errors were determined using the identically treated tRNA transcripts analyzed in triplicate.

Results

Intron-containing yeast tRNA^Phe is processed in yeast extract

The endonucleolytic degradation of ³²P-labelled T7 transcripts corresponding to intronless tRNA^Phe (76 nt) and to end-mature, intron-containing pre-tRNA^Phe (95 nt, see Fig. 1B) were assayed in vitro, using undialyzed cell-free yeast extract (S10 or S100, see Materials and Methods). After various incubation periods, the reaction mixture was analyzed by gel electrophoresis. The results of Figure 2 indicate that the T7 transcript of intronless tRNA^Phe (left part of figure) is stable, while under the same conditions, the intron-containing tRNA^Phe (right part of figure) clearly undergoes a rapid endonucleolytic cleavage (estimated half-life 20 min). The sizes of the resulting cleavage products correspond to a mixture of two half molecules (37 and 39 nt) and to the intron (IVS of 19 nt). During the incubation, the two RNA halves become ligated again to produce the full length intronless tRNA^Phe molecule (76 nt). This religation reaction is an ATP-dependent process (6,34,35) and, as expected, the yield of the religated tRNA molecule can be enhanced by the addition of extra ATP to the reaction mixture (data not shown). However, with the low amount of endogeneous ATP normally present in the non-dialyzed cell-free extract, the accumulation of the endonucleolytic 3′- and 5′-halves of the precursor tRNA^Phe (splicing intermediates) and of the newly synthesized full-length tRNA^Phe allowed to analyze easily the modification pattern of isolated oligonucleotides at the pre- and post-splicing steps of the in vitro tRNA^Phe maturation (see below).

Not all modified nucleotides can be formed in intronless yeast tRNA^Phe transcript

When the intronless tRNA^Phe transcript, separately radiolabeled with each of the four NTPs, was incubated with S100 yeast extract in the presence of SAM as donor of methyl groups, several nucleotides were enzymatically modified. Figure 3A-D shows typical autoradiograms of 2D tlc analysis of radiolabeled modified nucleotides present in hydrolysates of tRNA transcript. Figure 4A and B shows the time course of formation of each of modified nucleotides, detected by such analysis. Nuclease P1 allowed to identify 5′-³²P-modified nucleotides, while RNase T2 allowed to identify those 3′-³²P-modified nucleotides that are 5′-adjacent to the 5′-[³²P]NTP used for transcription. Therefore, analysis of the P1 and T2 data for each series of four tRNA samples, separately radiolabelled with one of the four 5′-[³²P]NTP, allowed to check twice each position of the tRNA molecule for the presence of a given modified nucleotide.

Figure 2

Splicing of intron-containing tRNA^Phe in yeast extract (right half). The figure shows the autoradiography of a 12% PAAG with radiolabelled RNA recovered from the incubation mixture with S100 yeast extract. Intronless tRNA^Phe is shown (as control) on the left. Incubation time at 37°C is indicated on top. The sizes of splicing products are given on the right. Intron-containing tRNA^Phe precursor (95 nt), intronless tRNA^Phe (76 nt), mixture of two half molecules (37 and 39 nt), intron (19 nt).

The results indicate that nine out of 11 different kinds of modified nucleotides normally present in fully matured yeast tRNA^Phe (Fig. 1A) were present in the hydrolysates of intronless tRNA^Phe: they were m²GMP (arising from position 10), m²₂GMP (from position 26), CmMP (from position 32), GmMP (from position 34), ΨMP (from both positions 39 and 55, see below), m⁷GMP (from position 46), m⁵CMP (from position 49 only, see below) and TMP (from position 54). The presence of m¹GMP at position 37 (instead of Y₃₇) fits well with the earlier observation that the enzymatic formation of m¹G₃₇ corresponds to the first step in the biosynthesis of the more complex hypermodified wybutosine-37 (Y₃₇) (36). The lack or the instability of the necessary cofactors for Y₃₇ formation from m¹G₃₇ in the S100 yeast extract may explain the absence of Y₃₇ in tRNA^Phe. Alternatively, the Y₃₇-forming enzymes might be unstable in vitro or depleted upon high speed centrifugation because of tight binding to the ribosomal fraction or to other high molecular mass entities. However, the later hypothesis can be ruled out since the use of freshly prepared S10, instead of S100 extract in the incubation mixture, did not catalyze Y₃₇ formation even in trace amounts (data not shown). Unexpectedly, the formation of m¹A₅₈ was not detected in the tRNA transcripts incubated with S100 yeast extract. However, this modified nucleotide can be readily formed in vitro in several yeast tRNA transcripts (including intronless tRNA^Phe) upon incubation in a rat liver extract (data not shown) or after microinjection into the cytoplasm of Xenopus laevis oocytes (18,37 and unpublished results). Therefore, the absence of m1A58 in both intronless tRNA^Phe and its precursor (see below) must be due to some as yet undetermined properties of the corresponding yeast methylase and/or to inappropriate experimental conditions in vitro. As in the case of Y₃₇ formation, the absence of tight binding of tRNA:A₅₈ methylase to a high molecular mass entity was verified experimentally by testing of S10 yeast extract. Finally, only low amounts of the modified nucleotide D₁₇, was detected (Fig. 3A). Due to its migration on the tlc plate very close to the much more intense [³²P]UMP spot in both of the chromatographic systems used, this modified nucleotide is always difficult to detect and to quantify. Apparently no D₁₆ was formed since no DMP was detected in the T2-hydrolysate of UTP-labelled transcript. Essentially the same results as described above for the intronless tRNA^Phe transcript (76 nt) were obtained for the religated tRNA^Phe transcript (76 nt) after intron splicing and recovery.

Figure 3

Autoradiograms of selected 2D chromatographies on thin-layer plates of modified nucleoside monophosphates after P1 or T2 digestion of incubated transcript. (A-D) Correspond to intronless tRNA^Phe labelled with GTP or UTP, digested by RNAse T2 or nuclease P1 as indicated on the figure. (E-H) Correspond to the intron-containing precursor (tRNA^Phe + IVS) similarly labelled and treated. Spots corresponding to modified nucleoside monophosphates labelled with either 5′- or 3′-³²P as well as unmodified radiolabelled nucleotides (AMP, UMP, CMP and GMP) are indicated.

Figure 4

Time course of the formation of modified nucleotides in intronless tRNA^Phe (A and B) and IVS-containing tRNA^Phe (C and D) incubated with S100 yeast extract. (A and C): ○, m²₂G₂₆, ●, m¹G₃₇, ◆, Gm₃₄, □, m⁷G₄₆, ▪, Cm₃₂, Δ, D₁₇; (B and D): ▪, Ψ₃₉+Ψ₅₅, ▴ , sum of m⁵C₄₀ and m⁵C₄₉ (dashed line), ●, T54, □, m¹A₅₈, Δ, D₁₆.

Figure 4B shows that almost 2 mol of pseudouridine per mol intronless tRNA^Phe were formed after 10 min incubation. Identical results were obtained from the P1 hydrolysate of [³²P]UTP-labelled and from T2 hydrolysate of [³²P]CTP-labelled transcripts. This amount corresponds to the sum of Ψ39 and Ψ₅₅ formation because to identical 3′-neighbouring cytosine (Fig. 1A) one cannot follow their formation separately. Enzymatic formation of Cm at position 32, m¹G at position 37 and T at position 54 also occurred with high yields (Fig. 4A and B), while formation of m²₂G at position 26, Gm at position 34 and m⁷G at position 46 (Fig. 4A) was less efficient but nevertheless easily detectable. The clear lag phase in the time courses of Gm and m⁷G formation (Fig. 4A) was reproducibly observed in several independent experiments and may indicate that some modified nucleotides should arise first at other sites before nucleotides at positions 34 and 46 can be enzymatically modified. Formation of D at position 17 (detected after T2 digestion of GTP-labelled tRNA) and m⁵C at position 49 (see below) occur only with low yields. As stated above, formation of Y₃₇, m¹A₅₈ and also D₁₆ (from T2 analysis of UTP labelled tRNA) were undetectable, even after prolonged incubation.

Figure 5

The presence (or absence) of negatively intron-dependent tRNA modifications in tRNA precursor (tRNA^Phe + IVS) (A and B), in the mixture of two tRNA halves (C and D), in religated tRNA molecule (E and F) and in intron-containing tRNA^Phe minisubstrate (G and H). The tRNA^Phe precursor was labelled with CTP (upper row) or GTP (bottom row), while the minisubstrate was labelled with CTP and UTP, followed by P1 or T2 hydrolysis as indicated. Experimental conditions are described in Materials and Methods.

Modification patterns of intron containing tRNA^Phe and intronless tRNA^Phe transcripts are different

As shown in Figures 3 and 4, several modified nucleotides present in hydrolysates of intronless tRNA^Phe were not found in hydrolysates of intron-containing precursor tRNA^Phe (Fig. 3, compare A-D to E-H; Fig. 4, compare A and B to C and D). These are CmMP derived from position 32, GmMP from position 34 and m¹GMP from position 37, all located in the anticodon loop. The other modified nucleotides (m²GMP, m²₂GMP, ΨMP, m⁷GMP, m⁵CMP and TMP), which occur in regions of the tRNA other than the anticodon loop, were present in both transcripts. Except for m⁵CMP, the time course of these ‘intron-independent’ nucleotide modifications are more or less the same (Fig. 4, compare A and B to C and D). The case of m⁵CMP is noteworthy. Indeed, from T2 analysis of UTP-labelled transcripts, only 0.2 mol m⁵CMP per mol of intronless tRNA^Phe was detected, while in the intron-containing tRNA^Phe, the formation of 1.2 mol/mol m⁵CMP was observed. Because yeast tRNA contains m⁵C at positions 40 and 49 with a 3′-adjacent uridine in both cases, it was impossible to differentiate them using the nearest-neighbour method of analysis (Fig. 1A). Taking into account that m⁵C₄₉ is frequently found in many yeast tRNAs, including those for which the corresponding genes do not contain an intron, the observed difference in m⁵C level is most probably related to the absence of C₄₀ methylation in intronless tRNA. A control test with tRNAAsp, which contains only m⁵C₄₉, showed a low yield of m⁵C formation in vitro (data not shown), as found in the case of intronless tRNA^Phe.

Direct evidence for intron-dependent formation of m⁵C₄₀ comes from experiments with truncated precursor tRNA^Phe. Using the PCR technique, a DNA fragment corresponding to the anticodon stem-loop, elongated with the natural intron and flanked on its 5′-end with a T7 promoter, was prepared. The corresponding runoff transcript (40 nt) was then transcribed in vitro, purified on PAAG and subsequently used as a ‘minisubstrate’ for the cytosine methylase present in S100 yeast extract. As shown in Figure 5G, this 40 nt minisubstrate was quantitatively modified to m⁵C; the rate of methylation of cytosine-i34 (which corresponds to C₄₀ in intronless tRNAs) was very similar to that observed for C₄₀ modification in intron-containing pre-tRNA^Phe. Under identical experimental conditions, the control 22 nt RNA fragment, corresponding to the intronless anticodon stem and loop, remained completely unmodified (results not shown). Evidently, part of the identity elements required for tRNA recognition by the corresponding cytosine methylase are located within the 19 nt intron. Beside m⁵C₄₀ formation, no other modified nucleotides were detectable in the intron-containing minisubstrate, including Ψ₃₉ 5′-adjacent to the methylated C₄₀ (Fig. 5H).

Table 1

Positive or negative intron-dependence and intron insensitivity of modified nucleotide formation in yeast tRNA^Phe as tested in homologous yeast extract

Modifications within the anticodon loop of tRNA^Phe occur only when the intronless 5′- and 3′-halves are ligated

Accumulation of the two halves of tRNA molecules during incubation of intron containing tRNA^Phe with yeast extract, as shown in Figure 2 (right half), allowed us to analyze the modification pattern after the endonucleolytic reaction but before the ligation reaction. The results indicate that all except three modified nucleotides were present in the 3′- and 5′-halves, as compared to the results with religated 76 nt tRNA^Phe (Fig. 5E and F; see also Fig. 3A-D). Only CmMP derived from position 32, GmMP from position 34 and m¹GMP from position 37 in the anticodon loop were absent (Fig. 5C and D), a situation which corresponds exactly to that described above for intron-containing tRNA^Phe (Fig. 5, compare C and D to A and B; see also Fig. 3E-H). However, these missing modified bases were formed after religation of the two half molecules (37 + 39 nt) leading to the 76 nt yeast tRNA^Phe (Fig. 5E and F). These results demonstrate that the target nucleotides have to be in a loop structure in order to undergo the modification by the corresponding enzymes. On the other hand, none of these three modified nucleotides appeared when the 22 nt minisubstrate consisting of just the anticodon stem-loop was incubated in yeast extract (data not shown), which suggest that not only the anticodon loop but also the tRNA architecture is required for enzymatic 24-O-methylation of the riboses at positions 32 and 34 and of the guanosine at position 37.

Discussion

The choice of yeast tRNA^Phe(GAA) to study the effect of intron removal on nucleotide modification pattern was made for several reasons. First, naturally occurring yeast tRNA^Phe is rather extensively modified (14 out of 76 nt, 18.6%). Moreover among the modified nucleotides, D₁₆, D₁₇, T₅₄, Ψ₅₅ are common to almost all tRNAs, m²G₁₀, m²₂G₂₆, m¹G₃₇, Ψ₃₉, Cm₃₂, m⁷G₄₆, m⁵C₄₉, m¹A₅₈ are frequently found in many tRNA species, while Gm₃₄, Y₃₇ and m⁵C₄₀ are unique features of eukaryotic tRNA^Phe. Thus, this molecule allows to explore several different types of tRNA modification enzymes. Second, the gene of yeast tRNA^Phe contains a 19 nt intron and the intron-containing or intronless tRNA^Phe can be easily produced in vitro by transcription of the corresponding synthetic gene with T7 polymerase (29,30). Third, the 3D-structure of naturally occurring yeast tRNA^Phe(GAA) harbouring all the modified nucleotides, and also the 3D-architecture of the corresponding T7 transcripts (with or without intron, but lacking all the modified nucleotides) was extensively studied by chemical and enzymatic probing (25,26,38–41), NMR (28,42–44) and, in the case of naturally occurring tRNA^Phe, by X-ray crystallography at high resolution (45). One aspect of yeast tRNA^Phe that has not yet been explored in detail, is the problem of stepwise nucleotide modification.

Figure 6 summarizes the results obtained for enzymatic formation of modified nucleotides when intronless tRNA^Phe transcript (Fig. 6A), intron-containing tRNA^Phe transcript (Fig. 6B), intron-containing minisubstrate (Fig. 6C) and anticodon stem-loop fragment (Fig. 6D) are incubated in a yeast extract. These results allow to divide the corresponding tRNA modification enzymes into three distinct groups (Table 1).

The first group of modification enzymes includes those that catalyze the modification reaction with equal efficiency in both intronless and intron-containing tRNA^Phe precursor (m²G₁₀, D₁₇, m²₂G₂₆, Ψ₃₉, m⁷G₄₆, m⁵C₄₉, T₅₄ and Ψ₅₅) (Table 1). These modified nucleotides are found in many yeast tRNAs independently of the presence of an intron in the corresponding tRNA genes. As one can expect, most of these modifications (except Ψ₃₉) are located outside the anticodon stem-loop. The fact that U₃₉ located at a proximal position to the anticodon loop can be converted to Ψ39 in both intron-containing tRNA^Phe and intronless tRNA^Phe, but not in the minisubstrates, implies that the corresponding pseudouridine synthase recognizes at least part of the tRNA architecture that is beyond the anticodon loop.

The second group of modification enzymes comprizes those which act only after intron removal (Cm₃₂, Gm₃₄ and m¹G₃₇). As expected, these modifications (middle column in Table 1) are present in the anticodon loop of tRNA. The fact that these modifications appear only after ligation of the two halves molecules implies that the corresponding enzymes recognize the sevenmembered anticodon loop of tRNA as identity element.

Figure 6

Summary of modification patterns of intronless (A) and intron-containing (B) tRNA^Phe(GAA), and of two minisubstrates (with and without intron) (C and D), upon incubation in S100 yeast extract. Modified nucleotides that are formed are shown by arrows. Shadowed boxed positions correspond to the nucleotides that, in contrast to those present in natural tRNA^Phe, are not modified in vitro. Intron sequence is shown in lower case letters.

The third group of modification enzymes corresponds to those that are strictly intron-dependent. In the case of tRNA^Phe only one enzyme of this type (tRNA methylase specific to m⁵C₄₀) participates in the maturation process at the precursor level. This enzyme also works on the minisubstrate composed of the tRNA^Phe anticodon stem-loop elongated by the intron, thus demonstrating that only a limited structural domain, located within the prolongated anticodon stem-loop of the pre tRNA is necessary for recognition. This type of positive intron-dependence has already been demonstrated for formation of Ψ₃₅ in tRNA^Tyr of different origins (6–9,11), Ψ₃₄ and Ψ₃₆ in minor yeast tRNA^Ile (12,46) and m⁵C₃₄ in yeast tRNA^Leu (13). In all of these cases, the intron provides the transient RNA structure which bears the necessary identity elements for enzyme recognition (19). This phenomenon may be similar to intron-dependent types of editing in GluB-receptor mRNA in rat brain (47,48) and trans-guiding of 2′-O-methylations of eukaryotic rRNA by complementary sno- RNAs (49,50). The case of m⁵C₃₄ formation in yeast tRNA^Leu (13) is of special interest with regard to the present study. Indeed, as for the enzymatic formation of m⁵C₄₀ in yeast tRNA^Phe, the target cytosine-34 in yeast tRNA^Leu is found in a helical double-stranded RNA region transiently stabilized by the intron (27). One might ask the question, therefore, whether the same enzyme in yeast catalyzes the methylation of cytosine at both position 34 in tRNA^Leu and position 40 in tRNA^Phe.

It is noteworthy that modified nucleotide m⁵C₄₀ was found only in tRNA specific for phenylalanine from fungi (S.cerevisiae and N.crassa) for which the corresponding genes contain an intron. All other eukaryotic tRNA^Phe sequenced so far (including X.laevis tRNA^Phe (51) have an unmodified cytosine or a pseudouridine at position 40; moreover the corresponding tRNA^Phe genes have no intron (2). Therefore, these eukaryotic cells may not contain the kind of RNA:cytosine methylase as we found in yeast. Indeed, when the intron-containing minisubstrate of yeast tRNA^Phe was incubated in X.laevis S100 extract, no m⁵C formation was detected (data not shown). One exception seems to be tRNA^Phe from the fungi S.pombe which was reported to contain m⁵C₄₀ (52) while the corresponding gene has no intron. However, reevaluation of the details of the sequence data published by these authors (52), supports the view that in fact C40 is probably not modified to m⁵C₄₀ in tRNA^Phe from S.pombe. The possibility therefore exists that S.pombe also lacks the intron-dependent cytosine-40 methylase, as is the case for X.laevis oocytes.

Our results on the modification profile of pre-tRNA^Phe in yeast are in good agreement with the earlier observations made in vivo for expressed synthetic genes of supressor tRNA^Phe (19). The absence of Cm₃₂ in the tRNA^Phe precursor and the intron dependency of m⁵C formation was already suggested in that study, but in the latter case no distinction could be made between m⁵C at positions 40 and 49.

Last, but not least, the simple posttranscriptional methylation of cytosine-40 in anticodon arm of yeast tRNA^Phe has been shown to facilitate a tight site-specific Mg²⁺ binding which enables the anticodon stem-loop to form more than one conformation through a process regulated by Mg²⁺ concentration (53,54, reviewed in 55). This m⁵C-dependent, Mg²⁺-induced conformational changes of the anticodon stem-loop, in conjunction with the presence of other modified nucleotides in tRNA (m¹G₃₇, Ψ₃₉), are important parameters for the biological function of the tRNA molecule, such as its recognition by aminoacyl-tRNA synthetases and codon reading on the ribosome during translation.

Acknowledgements

This work was supported in France by research grants to H.G. from the ‘Centre National de la Recherche Scientifique’, ‘Actions Concertées Coordonnées des Sciences du Vivant’ (ACC-SV-5) and the ‘Association pour la Recherche sur le Cancer’ (ARC). H.Q.J. and Y.X.J. were each recipients of a 3 month fellowship from the French-Chinese Association for Scientific and Technical Research (PRA 94-4). Y.M. was supported by a Long-Term FEBS Fellowship (1995)1996). Plasmids, containing the cloned yeast tRNA^Phe or its precursor were kindly provided respectively by Dr O.Uhlenbeck (Boulder, CO, USA) and Dr J.Abelson (Pasadena, CA, USA). We thank Dr J.-P.Waller (CNRS, Gif-sur- Yvette) and Dr R.Tewari (NCL, Pune, India) for comments on the manuscript.

Author notes

The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors