Highly conserved modified nucleosides influence Mg²⁺-dependent tRNA folding

Abstract

Transfer RNA structure involves complex folding interactions of the TΨC domain with the D domain. However, the role of the highly conserved nucleoside modifications in the TΨC domain, rT₅₄, Ψ₅₅ and m⁵C₄₉, in tertiary folding is not understood. To determine whether these modified nucleosides have a role in tRNA folding, the association of variously modified yeast tRNA^Phe T-half molecules (nucleosides 40–72) with the corresponding unmodified D-half molecule (nucleosides 1–30) was detected and quantified using a native polyacrylamide gel mobility shift assay. Mg²⁺ was required for formation and maintenance of all complexes. The modified T-half folding interactions with the D-half resulted in K_ds (rT₅₄ = 6 ± 2, m⁵C₄₉ = 11 ± 2, Ψ₅₅ = 14 ± 5, and rT₅₄,Ψ₅₅ = 11 ± 3 µM) significantly lower than that of the unmodified T-half (40 ± 10 µM). However, the global folds of the unmodified and modified complexes were comparable to each other and to that of an unmodified yeast tRNA^Phe and native yeast tRNA^Phe, as determined by lead cleavage patterns at U₁₇ and nucleoside substitutions disrupting the Levitt base pair. Thus, conserved modifications of tRNA’s TΨC domain enhanced the affinity between the two half-molecules without altering the global conformation indicating an enhanced stability to the complex and/or an altered folding pathway.

INTRODUCTION

All species of cytoplasmic tRNA appear to have the same basic L-shaped tertiary structure. The anticodon stem and loop and the helical aminoacyl stem of the cloverleaf secondary structure remain independent of the global tertiary fold. However, the loop containing dihydrouridine (D-loop), and that containing ribothymidine and pseudouridine (TΨC-loop) form the junction between these two physically separate domains (Fig. 1). Though tRNA nucleotide sequences are not well conserved across kingdoms, some of tRNA’s post-transcriptional modifications are highly conserved throughout all cytoplasmic tRNA species (1). Modifications in tRNA’s TΨC domain represent some of the most highly conserved RNA modifications. Pseudouridine at position 55, Ψ₅₅, occurs at a frequency of 90%. Ribothymidine, or 5-methyluridine (rT₅₄), occurs at position 54 in 60% of all sequenced tRNAs, and m⁵C at positions 48, 49 or both in 22%, 11% and 6%, respectively (1). Conservation of the nucleoside modifications in the TΨC domain suggests that they may play an important role in a function common to all tRNAs or in the folding of tRNA into its conserved tertiary structure for proper function.

Figure 1.

(A) Cloverleaf secondary structure of yeast tRNA^Phe color-coded by domain. Red is the aminoacyl stem; blue is the T-stem–loop; orange is the variable loop; green is the anticodon stem–loop; and black is the D-stem–loop. (B) Crystal structure of yeast tRNA^Phe (4) showing tertiary interactions between the T-loop and D-loop folding tRNA into its native L structure. Color coding corresponds to the secondary domains indicated in (A). The placement of Mg²⁺ ions was determined by X-ray diffraction as reported previously and is indicated by pink spheres. (C) The unmodified sequence of chemically synthesized constructs that mimic the T domain and D domain in yeast tRNA^Phe. Positions of incorporated modifications are boxed; m⁵C₄₉, m⁵U₅₄ (rT₅₄) and Ψ₅₅ were singly or doubly incorporated into the T-half molecule. The Levitt base pair in yeast tRNA^Phe occurs between G₁₅ and C₄₈ and is indicated by a dashed line. Constructs altering the potential for a Levitt base pair contained a U at position 15 in the D-half molecule (D-half G₁₅U), or a U at position 48 in the T-half molecule (T-half C₄₈U).

tRNA’s complex folding, the result of long-range, tertiary structure hydrogen bonding and metal ion coordination between the D-, TΨC- and variable-loops, occurs soon after transcription and before processing of the tRNA is complete (2,3). The folding interactions of the conserved L-shaped structure of cytoplasmic tRNAs are characterized by hydrogen bonding of G₁₉ in the D domain with C₅₆ in the TΨC domain and by a G₁₈ base pair to Ψ₅₅ (4). The tRNA molecule does not need to be in its native, fully processed, mature size for modification enzymes to produce rT₅₄ and Ψ₅₅, yet some structure of the RNA molecule is required (5–7). While neither rT₅₄ nor Ψ₅₅ is required for tRNA aminoacylation, elongation factor recognition or codon recognition on the ribosome in vitro (8), Escherichia coli lacking the enzyme for Ψ₅₅ synthesis are at a disadvantage in competing with wild type cells (9). In addition, E.coli lacking the gene for rT₅₄ synthesis are not viable (10). However, the rT₅₄ synthesis activity does not appear to be the cause, since cells with inactive enzymes are viable (11).

tRNA’s tertiary folding interactions in vitro are observed with transcripts completely devoid of modifications (12,13). Yet, thermal denaturation of the fully modified, native tRNA occurs at a higher T_m than the corresponding unmodified transcript (12,14,15). Incorporation of rT₅₄ increased the T_m of a 17mer analog of the yeast tRNA^Phe TΨC stem and loop domain with a corresponding decrease in ΔG, whereas Ψ₅₅ had no measurable effect on the melting temperature or free energy (7). Though the base-stacking contributions of Ψ compared with uridine were significant in a base-paired region of tRNA (16–19), hydrogen bonding through pseudouridine’s N1 may be favored in tertiary interactions between loops. Therefore, rT₅₄ increases the thermal stability of tRNA at both the secondary (7,20) and tertiary level (21) and Ψ₅₅ provides base-stacking and/or hydrogen bonding contributions to the tertiary interaction with the D domain. However, the contributions of the individual modified nucleosides for the folding process, and not the resulting functional structure, are unclear. Perhaps the modifications affect the kinetics of tRNA’s folding interactions and/or the number of folding pathways.

Empirical data on the contributions of individual modified nucleosides to the folding pathways for tRNA have been limited because experimental systems either utilize completely modified, native tRNA or completely unmodified transcripts (12,14,15). A structural study of the isolated stem and loop of the TΨC domain suggested a modification-dependent folding intermediate (20). The 17mer with the rT₅₄ modification adopted a structure similar to that of the same domain within the native tRNA’s crystal structure, but distinctly different from the completely unmodified 17mer RNA (22,23). This result, in addition to the folding interaction of Ψ₅₅ with the D domain, suggests that modifications in the TΨC domain may enhance the natural folding process of tRNAs.

The folding of RNA’s highly negative phosphate backbone requires the presence of counterions, typically Mg²⁺. Similarly, tRNA folding depends on the presence of counterions (12), but any contributions of rT₅₄ and Ψ₅₅ to this counterion dependence have yet to be determined. The counterion-dependent folding of native yeast tRNA^Phe and the unmodified tRNA^Phe transcript suggests that several folding pathways exist in which distinct, stable intermediate conformations can be observed (15,24). However, the site-specific coordination of Mg²⁺ ions in tRNA has been challenged (25). Although Mg²⁺ is coordinated between the D- and TΨC-loops within the crystal structure of yeast tRNA^Phe (4,26,27), recent results with monovalent ions have suggested that electrostatic shielding of the phosphate backbone alone localizes Mg²⁺ to certain regions of the tRNA without coordination to specific ligands (25). Modified nucleosides have been implicated in the specific coordination of Mg²⁺ by native tRNAs (28,29) and in lowering the concentration of Mg²⁺ required for tertiary structure interactions (28). In contrast to native tRNA, tertiary structure interactions of unmodified transcripts have an absolute requirement for either high millimolar concentrations of Mg²⁺ (12–14) or molar concentrations of monovalent ion (15). However, the specific effect of rT₅₄ and Ψ₅₅ on Mg²⁺ coordination in the tertiary structure has not been determined (13,30).

To better understand the role of modified nucleosides in the ion-dependent folding interactions of tRNA, we have developed an experimental bimolecular model system. The model system allows us to site-specifically incorporate the naturally occurring, highly conserved modified nucleosides and monitor a bimolecular interaction of two RNA fragments mimicking the TΨC domain and the D domain of yeast tRNA^Phe. The interaction of chemically synthesized half molecules, corresponding to nucleosides 1–30 (D-half) with nucleosides 40–72 (T-half) of yeast tRNA^Phe, was monitored in a quantifiable polyacrylamide gel mobility shift assay. Here, we report that formation of the complex by the two half molecules required Mg²⁺ and introduction of the highly conserved modifications rT₅₄ and Ψ₅₅ and the commonly occurring m⁵C₄₉ enhanced the affinity of the unmodified D-half molecule for the T-half molecule, but did not alter the global fold of the complex, which was similar to native yeast tRNA^Phe.

MATERIALS AND METHODS

RNA synthesis, purification and analysis

tRNA half molecules were chemically synthesized with the nucleotide sequence of yeast tRNA^Phe (Fig. 1C) using 2′-bis(acetoxyethoxy)-methyl ether (ACE; Dharmacon, Inc., Boulder, CO) (31), 2′-O-triisopropylsilyl-oxy-methyl (TOM; Xeragon, AG, Zurich, Switzerland) (32) and standard phosphoramidite chemistries (NCSU Nucleic Acids Facility, Raleigh, NC) (33,34). The unmodified D-half molecule RNAs corresponded to the 30mer sequence of nucleotides 1–30 in yeast tRNA^Phe and the variously modified T-half RNAs corresponded to the 33mer sequence of nucleotides 40–72 in yeast tRNA^Phe (Fig. 1). The 1–72 nt unmodified sequence of yeast tRNA^Phe was chemically synthesized (Xeragon, AG) to mimic the completely unmodified yeast tRNA^Phe transcript. RNAs were 3′-end labeled with [5′-³²P]Cp using RNA ligase (35) and then diluted with known quantities of unlabeled half molecule. Radiolabeled [5′-³²P]Cp was produced from [γ-³²P] ATP (Amersham Biosciences, Piscataway, NJ) and CMP using polynucleotide kinase (35).

Gel mobility-shift assay

Gel mobility-shift assays were conducted with 15% native polyacrylamide gels composed of a Tris–borate buffer (TB; 89 mM Tris–HCl, 89 mM boric acid, pH 8.1). Mg²⁺, present only in the gel-forming reaction and in TB running buffer, was at a concentration of 3 mM, unless noted. The electrophoresis apparatus was surrounded by ice to maintain a temperature <25°C. All 3′-end-labeled T-half molecules were purified by a 7 M urea denaturing polyacrylamide gel electrophoresis (PAGE) (15%) prior to reaction with the D-half molecule in the native mobility-shift assay. Purified T-half molecules, eluted from gels using denaturing elution buffer (0.5 M ammonium acetate, 1% SDS, 0.1 mM ethylenediamine-tetraacetic acid), were reacted with the unmodified D-half in 15 µl of native loading buffer (30% glycerol in 89 mM Tris–HCl, 89 mM boric acid, pH 8.1). A fixed concentration of the T-half molecule was titrated with increasing amounts of the D-half molecule in individual reaction mixtures as described in the figure legends. The reaction mixtures were heated at 65°C for 10 min (in the absence of Mg²⁺), then allowed to cool to room temperature prior to loading on the gel. After electrophoresis, the gel was dried and visualized by phosphorimaging (36). T-half molecules free and within the T-half/D-half complex were quantified by analyzing the volume of the phosphorimage using ImageQuant software (Molecular Dynamics, Amersham Biosciences). Only the two prominent bands, free and bound, were analyzed. The material between the two bands was not included in the analysis of free or bound material (see Fig. 5). A lane containing no RNA was used for a background subtraction from all lanes with bound and free RNA. The percent complex formation was determined from the volumes and then normalized to 100% for relative K_d determination. The data were fitted to a one-site binding model using Prism software and errors were based on the deviations from the least squares fit (GraphPad Software, San Diego, CA).

Figure 5.

(A) Mobility-shift assays of ³²P-end-labeled, unmodified and modified, T-half molecules with increasing amounts of unmodified D-half molecule. In the native PAGE (15%, 3.0 mM Mg²⁺) at the top of the figure, complex formation is detected when the unmodified, ³²P-labeled T-half molecule interacts with the unmodified D-half molecule. Below, a complex between the Ψ₅₅-containing T-half molecule (T-half-Ψ₅₅) with the unmodified D-half molecule is detected. All lanes contained 84.2 µM T-half molecule. Lanes 2–7 of each gel contained increasing amounts of unmodified D-half molecule as follows: 8.42, 25.3, 42.1, 63.2, 84.2 and 253 µM, respectively. (B) Percent complex formed versus concentration of the D-half molecule for the unmodified D-half molecule with unmodified T-half molecule (closed square) or with the Ψ₅₅-containing T-half molecule (closed triangle). Data were normalized to 100% complex formation and curves were fitted to the data with a non-linear regression assuming one-site binding. The percent complex was determined by quantifying the phosphorimages from the PAGE in (A) using ImageQuant software (Molecular Dynamics) and errors were determined as deviations from the least squares fit.

Ribonuclease T1 and lead cleavage reactions

The 3′-end-labeled RNAs and RNA complexes (D- and T-half molecule complexes) were purified by PAGE (15%). The RNAs were identified on the gel by ethidium bromide staining. Individual RNAs were eluted from the gel in denaturing elution buffer (0.5 M ammonium acetate, 1% SDS, 0.1 mM EDTA), whereas complexes of the D- and T-half molecules were eluted with native elution buffer (20 mM HEPES, 250 mM KCl and 5 mM MgCl₂). The eluted RNAs were ethanol precipitated, dried and resuspended to a concentration of 20 µM RNA in water. RNAs that were not in complexes were denatured and renatured by heating to 70°C for 5 min and slowly cooled to room temperature prior to use.

Lead cleavage reactions (10 µl) contained 4 µM RNA (40 mM NaCl, 10 mM Tris–HCl, pH 7.2, 10 mM MgCl₂) and 0.5 mM Pb(II) acetate. Reaction mixtures were stored on ice, initiated by the addition of Pb(II) acetate and incubated at 20°C for 15 min. The lead cleavage sites were mapped by generating an alkaline hydrolysis ladder. Alkaline hydrolysis was conducted by heating reaction mixtures (10 µl) containing 2 µM RNA and formamide (1:5 by volume in 50 mM NaHCO₃/Na₂CO₃ pH 9.1) at 100°C for 10 min. In addition, the three Gs (G₁₈, G₁₉ and G₂₀) in the D-loop that are 3′ to the expected lead cleavage site at U₁₇, were mapped with RNase T1 digestion (Promega, Madison, WI). RNase T1 partial digestions were performed in reaction mixtures (10 µl) containing 2 µM RNA (50 mM sodium citrate, 4.5 M urea) with 5 U RNase T1 at 55°C for 20 min. Cleavage reactions were stopped by addition of 10 µl formamide loading dye (100% deionized formamide, 10 mM EDTA, 0.25% bromophenol blue and 0.25% xylene cyanol FF). Radiolabeled products of the cleavage reactions were analyzed by 7 M urea–PAGE (15%) and visualized by phosphorimaging.

RESULTS

Design of a bimolecular system to mimic tRNA folding

A bimolecular experimental model was designed to determine the contributions of modified nucleosides to the folding interaction of tRNA’s TΨC domain with the D domain. The experimental system consisted of a 33 nt T-half molecule and a 30 nt D-half molecule (Fig. 1C). The T-half molecule was synthesized unmodified or with the modified nucleosides rT₅₄, Ψ₅₅, m⁵C₄₉ or with both rT₅₄ and Ψ₅₅ (Fig. 2). While a previous study had used a similar bimolecular model to determine the minimal fragments of yeast tRNA^Phe required for leadzyme activity (37), we extended the lengths of our half molecules to include the aminoacyl stem and anticodon stem base-paired regions (Fig. 1C). Though a 26 nt D-half formed a complex with a 29 nt T-half molecule, an extension to slightly larger RNAs was necessary to achieve complete T- and D-half molecule complex formation as a gel mobility shift under the native PAGE conditions used in this study (Fig. 3). Mobilities of RNAs in native gels are apparently sensitive to molecular conformation and Mg²⁺ concentrations (38). In the presence of 3 mM Mg²⁺ and increasing concentrations of the D-half molecule, the intensity of the faster moving band of individual half molecules decreased with a concomitant increase in the intensity of the slower moving complex (Fig. 3A, lanes 2–5). An increase in the intensity of the faster moving band was observed with addition of excess D-half molecule (Fig. 3A, lanes 6–9). Complex formation required the presence of both half molecules. No gel shift was evident with either half molecule alone, even at 10-fold higher concentrations indicating that the individual half molecules did not aggregate (Fig. 3B).

Figure 2.

The chemical structure and hydrogen bonding of the conserved modifications, rT₅₄, Ψ₅₅ and m⁵C₄₉. Modified T-half molecule constructs were chemically synthesized with the modified nucleosides rT₅₄, Ψ₅₅ and m⁵C₄₉ individually, or with both rT₅₄ and Ψ₅₅. The hydrogen bonding of each of these modified nucleosides as found in the X-ray crystallographic tertiary structure yeast tRNA^Phe is shown in the figure. Ribothymidine or 5-methyluridine (rT₅₄) forms a tertiary structure base pair with m¹A₅₈. However, our T-half molecules are constructed with the unmodified A₅₈. Methyl-5-cytidine forms a canonical Watson–Crick base pair with G₆₅.

Figure 3.

Complex formation between the unmodified D- and T-half molecules was detected in a native PAGE (15%, 3.0 mM Mg²⁺) with ethidium bromide staining. Free D- and T-half molecules migrated similarly, and stained less well than the slower migrating complex. (A) Lanes 1–9 contained 5.6 µM unmodified T-half molecule. Lanes 2–9 contained increasing amounts of unmodified D-half molecule with the following µM concentrations: 0.56, 1.68, 2.81, 4.21, 5.61, 16.8, 56.1 and 112. (B) Increasing amounts of unmodified T-half molecule alone in a native PAGE (15%, 3.0 mM Mg²⁺) demonstrated that the half molecule does not aggregate. Lanes 1–6 contained unmodified T-half with the following µM concentrations: 1.6, 3.0, 4.2, 5.6, 7.0 and 8.4.

Complex formation also required Mg²⁺ in the gel and running buffer. Mg²⁺ in the sample alone was not sufficient to demonstrate complex formation by a gel shift (Fig. 4A). At least 0.75 mM Mg²⁺ was required for a complex to form whether the T-half was unmodified or had Ψ₅₅ (Fig. 4B). A 3 mM concentration of Mg²⁺ was determined to be at saturation for complex formation (Fig. 4B). Since Ψ₅₅, the most highly conserved of the three T-half modifications, did not lower the requirement for Mg²⁺ (Fig. 4B), experiments were routinely conducted with 3 mM Mg²⁺ to ensure that the effects of modifications, and not of Mg²⁺, were being measured.

Figure 4.

(A) The absence of Mg²⁺ in both the running buffer and gel precluded complex formation. A native polyacrylamide gel (15%) was polymerized in the absense of Mg²⁺. The running buffer (TB) also lacked Mg²⁺. RNA was detected with ethidium bromide staining. Samples for lanes 1 and 2 of the gel contained 84.2 µM unmodified T-half and D-half, respectively, but the D-half did not stain well with ethidium bromide, due to lack of secondary structure (John Stuart and Paul F. Agris, personal communication). Lane 3 contained 84.2 µM unmodified T-half and 84.2 µM D-half. (B) A Mg²⁺ concentration >0.75 mM is required for complex formation. Complex formation was analyzed under different Mg²⁺ concentrations from 0 to 5 mM. Samples of 84.2 µM unmodified (lanes 1, 3, 5, 7, 9 and 11), or Ψ-containing T-half (lanes 2, 4, 6, 8, 10 and 12) and 84.2 µM D-half molecules were subjected to native PAGE in which the gel, sample and running buffers contained the same concentration of Mg²⁺. The composite of gels depicts examples of the results used to produce the graph below (unmodified T-half, open circles, solid line; Ψ-containing T-half, cross, dotted line). (C) The left-hand native gel demonstrates a lack of complex formation for the unmodified T-half molecule with increasing concentrations of the D-half molecule in the presence of only 0.5 mM Mg²⁺. Lane 1 contained 84.2 µM unmodified T-half molecule. Lanes 2–9 contained 84.2 µM unmodified T-half with the following µM concentrations of D-half: 8.42, 25.3, 42.1, 63.2, 84.2, 253, 842 and 1684 µM. The right-hand native gel demonstrates a lack of complex formation for the Ψ₅₅-containing T-half molecule in the presence of 0.5 mM Mg²⁺. Lane 1 contains 84.2 µM Ψ₅₅ T-half alone. Lanes 2–9 contain 84.2 µM Ψ₅₅ T-half with the following µM concentrations of D-half: 8.42, 25.3, 42.1, 63.2, 84.2, 253, 842 and 1684 µM.

A gel mobility shift, indicating complex formation of the T- and D-half molecules, was observable with ethidium bromide staining (3 mM Mg²⁺, pH 8.1, 25°C). The individual T- and D-half molecules, though of different lengths and sequences, exhibited very similar migrations under native PAGE conditions. However, labeling either half molecule with ³²P allowed us to observe that half molecule as free or as participating in the complex (Fig. 5). The complex was detected independently of which half molecule was labeled, but the T-half molecule was routinely labeled for all quantified experiments. The integrity of the complex could be maintained during extraction from the gel in the presence of Mg²⁺. Upon removal of Mg²⁺, the complex dissociated into the individual D- and T-half molecules (data not shown).

TΨC domain modifications enhanced affinity of the D-half molecule for the T-half molecule

The binding affinities of the unmodified and variously modified T-half molecules for the completely unmodified D-half molecule were compared. Binding affinities for the interacting half molecules were determined by titrating the 3′-³²P-labeled T-half molecules with increasing concentrations of the unlabeled D-half molecule (Fig. 5). Free and complexed T-half molecules were separated by native PAGE and quantified by phosphorimaging, as described in Materials and Methods. Binding affinity, calculated as dissociation constants (K_d), varied with the modification present in the T-half molecule (Table 1). All K_d values were determined under saturating conditions although, for clarity, binding isotherms are not shown to saturation (Fig. 5). While all of the naturally occurring TΨC domain modifications that we investigated, m⁵C₄₉, T₅₄ and Ψ₅₅, enhanced the affinity of the half molecules, the rT₅₄-containing T-half molecule exhibited the highest affinity for the unmodified D-half molecule (Table 1). The T-half molecule with rT site-specifically incorporated at its naturally occurring position, 54, exhibited an affinity for the D-half molecule, 7-fold higher (K_d = 6 ± 2 µM) than that of the unmodified T-half molecule (K_d = 40 ± 10 µM). Pseudouridine at position 55 increased the binding affinity of the T-half molecule ∼3-fold (K_d = 14 ± 5 µM), compared with the unmodified T-half molecule. The m⁵C₄₉ modification increased the affinity of the T-half almost 4-fold (K_d = 11 ± 2 µM) compared with that of the unmodified T-half molecule. A doubly modified T-half molecule containing both rT₅₄ and Ψ₅₅ exhibited an affinity (K_d = 11 ± 3 µM) comparable with that of the m⁵C₄₉-containing T-half molecule and ∼4-fold higher than that of the completely unmodified T-half molecule. The affinity of the rT₅₄ and Ψ₅₅ modified T-half molecule for the D-half was intermediate to that of the rT₅₄ or Ψ₅₅ modification alone.

Table 1. Binding affinities and free energy for complex formation between the unmodified D-half and variously modified T-half molecules.

T-half moleculea	Kd (µM)b	ΔG37c
Unmodified	40 ± 10	–6.2 ± 1.6
RT54	6 ± 2	–7.4 ± 2.0
m5C49	11 ± 2	–7.0 ± 1.3
Ψ55	14 ± 5	–6.9 ± 2.5
RT54;Ψ55	11 ± 3	–7.0 ± 1.9

^aT-half molecule constructs containing the site-specific incorporations of the indicated modified nucleosides.

^bBinding constants were determined by quantifying mobility shift assays (Materials and Methods), plotting % complex formation versus concen tration of D-half molecule, and fitting data with non-linear regression.

^cΔG₃₇ (kcal/mol at 37°C) was calculated from the association constants (ΔG = –RT ln K).

Modifications do not lower the Mg²⁺ requirement for complex formation

Binding affinities were determined by mobility-shift assays conducted under excess Mg²⁺ concentrations to minimize any contributions the modified nucleosides may have had in coordinating Mg²⁺. In order to determine if the most highly conserved modification in tRNA (1), Ψ₅₅, altered the Mg²⁺ requirement, native PAGE mobility-shift assays were conducted at various concentrations of Mg²⁺ (0–5 mM). The unmodified T-half and Ψ₅₅-containing T-half molecules were titrated with the unmodified D-half molecule. The minimum Mg²⁺ concentration required for the observable interaction between the two unmodified half molecules was ∼0.75 mM (Fig. 4B). In the presence of 0.5 mM Mg²⁺, even a 20-fold concentration of D-half to T-half molecule failed to generate an observable complex formation (Fig. 4C). Incorporation of Ψ₅₅ into the T-half molecule did not lower the Mg²⁺ concentration required to observe the mobility-shift (Fig. 4B and C). Additions of very small amounts of Mg²⁺ readily alter the conformation of tRNA^Phe, as revealed by hydroxy radical cleavage patterns (39). Therefore, while the conserved modifications enhanced the affinity of the T-half molecule for the D-half molecule, they did not lower the Mg²⁺ requirement for the interaction of the two half molecules under the conditions of the assay.

Comparison of global folds of the D and TΨC complex and native yeast tRNA^Phe

Yeast tRNA^Phe has a G₁₅:C₄₈ tertiary structure base pair in common with some 65% of all tRNAs (1,40). This base pair, known as the Levitt pair, occurs between G₁₅, in the D-loop, and C₄₈, in the variable loop, in yeast tRNA^Phe (Fig. 1C). To ensure that complex formation observed under native PAGE conditions was a result of known tertiary structure interactions between D- and T-half molecules, and not simply a result of canonical base-pairing between the aminoacyl and anticodon stems, RNAs were synthesized to eliminate the possibility of the G₁₅:C₄₈ Levitt tertiary structure base pair. D- and T-half molecules with U substitutions in both positions 15 (D-half G₁₅U) and 48 (T-half C₄₈U) would not be expected to form a complex, and none was observed (Fig. 6). No complex was observed when the completely unmodified, native sequence T-half molecule was combined with the D-half G₁₅U, since a U₁₅:C₄₈ base pair would have to form (Fig. 6). However, when the unmodified, but native sequence D-half molecule was incubated with the T-half C₄₈U molecule, a small amount of RNA with the mobility of the native sequence D-half/T-half complex and a faster moving, diffuse band of RNA in the gel were observed. The diffuse band of RNA in the gel could have resulted from a G₁₅:U₄₈ tertiary interaction plus tertiary structure mis-pairings that yielded complexes with many conformations (Fig. 6). Observation of a mobility-shift that required the two half-molecule constructs, the Levitt base pair and Mg²⁺ was indicative of the complex having a global fold similar to that of native yeast tRNA^Phe. To confirm that the overall structure of the complex composed of the T- and D-half molecules was indeed similar to that of native tRNA^Phe, we took advantage of the tRNA’s ability to self-cleave with Pb²⁺ only when in the native fold (41). A T-half/D-half complex with a fold similar to that of native tRNA^Phe should exhibit the well-characterized self-cleavage reaction catalyzed by lead (37). When correctly folded native tRNA^Phe acts as a leadzyme, a specific and strong, single site cleavage of the D-loop between D₁₇ (or U₁₇) and G₁₈ results in identifiable fragments (Fig. 7A). RNase T1 digestions, under denaturing conditions, and alkaline hydrolysis ladders confirmed the location of the lead cleavage (Fig. 7A). The location of the lead cleavage produced by an unmodified, chemically synthesized and slightly truncated tRNA^Phe (nucleotides 1–72) was identical to that of the native tRNA (Fig. 7A), as reported previously for an unmodified transcript (37). A specific self-cleavage by lead between U₁₇ and G₁₈ was achieved with the complex composed of the unmodified half molecules indicating that the global fold of the complex was similar to that of native tRNA^Phe (Fig. 7B). Lead cleavage of the D-half molecule alone revealed random cleavages quite similar to that of an alkaline hydrolysis ladder, and dramatically different to the site-specific cleavage when complexed with the T-half molecule (Fig. 7B). Since the TΨC domain modifications enhanced the interaction of the T-half molecule for the unmodified D-half molecule, the modifications may also alter the structure of the resulting complex. Therefore, we performed the lead cleavage analysis to determine if modifications influenced the overall tertiary conformation of the complex. The strong, site-specific lead cleavage of a complex composed of the unmodified D-half and a rT₅₄,Ψ₅₅-containing T-half molecule was similar to that of the unmodified D-half and T-half complex (Fig. 7B). Thus, the lead cleavage patterns of the complexes, native yeast tRNA^Phe, and the unmodified analog of yeast tRNA^Phe, were all similar.

Figure 6.

Absence of complex formation by T- and D-half molecules without the potential for Levitt base-pair. Lane 1 contained 84.2 µM of unlabeled T-half C₄₈U and 84.2 µM D-half G₁₅U. The single RNA band, visualized with ethidium bromide staining, is that of the free T-half C₄₈U and D-half G₁₅U molecules that migrate similarly in the gel. A mobility shift indicating complex formation was not detectable. Lane 2 contained the positive control of unmodified T-half molecule (84.2 µM) with unmodified D-half (84.2 µM). The major band was the mobility-shifted, ³²P-end-labeled T-half molecule in complex with D-half molecule. The RNA was visualized by phosphorimaging. Some free T-half molecule is also observed. Lanes 3–5 contained ³²P-end-labeled unmodified T-half molecule (84.2 µM) with an increasing concentration of the D-half G₁₅U, 8.42, 84.2 and 253 µM, respectively. A Levitt base pair interaction by the half molecules in lanes 3–5 would require a U₁₅C₄₈ base pair to occur. Lane 6 contained ³²P-end- labeled, unmodified T-half molecule (84.2 µM) only. Lanes 7–9 contained the T-half C₄₈U molecule (84.2 µM) with increasing amounts of unmodified D-half molecule, 8.42, 84.2 and 253 µM, respectively. A Levitt base pair interaction by these half molecules would require formation of a G₁₅U₄₈ base pair. Lane 10 contained 84.2 µM ³²P-end-labeled C₄₈U T-half molecule, alone. Mobility of the free T-half molecule varied slightly across the gel.

Figure 7.

Lead, Pb(II), cleavage of native tRNA^Phe and the D-half molecule in the presence of the unmodified and the rT₅₄ and Ψ₅₅ modified T-half molecules produced a distinct cleavage pattern not observed in the D-half molecule alone. (A) Native tRNA^Phe and unmodified tRNA^Phe 72mer, and (B) free D-half molecule, the complex from unmodified D- and T-half molecules and that from the unmodified D-half with the rT₅₄ and Ψ₅₅ modified T-half molecule were ³²P-3′-end-labeled and subjected to: (–) no treatment, (OH) alkaline hydrolysis, (T1) T1 nuclease and (Pb) 0.5 mM Pb²⁺. Alkaline hydrolysis and T1 nuclease treatments were used to confirm cleavage sites. All reactions are described in Materials and Methods. Reaction products were separated by 10% PAGE–7 M urea and visualized by phosphorimaging. The positions of the nucleotides sensitive to T1 cleavage (G18–20) and Pb²⁺ cleavage (between U16 and U17) are indicated in the figure. The heavy bands at the top of the lanes in (A) and (B) are the full-length native tRNA^Phe and the unmodified, 72mer analog.

DISCUSSION

The contributions of some highly conserved modifications to the folding of tRNA have not been well documented, although the Mg²⁺-dependent folding of both unmodified and fully modified tRNAs has been studied (12,13,24). With tRNA containing the largest and most variant number of modified nucleosides, there may be a role for modifications in the folding process of tRNA by either stabilizing intermediates or altering local conformations for the overall folding of the tRNA into its native tertiary structure. In particular, site-specifically conserved modifications, such as rT₅₄ and Ψ₅₅, may be important for tRNA to efficiently and effectively achieve functional conformation under physiological conditions. To assess the roles of highly conserved, modified nucleosides and Mg²⁺ in the folding interactions of tRNA’s domains, we developed an experimental bimolecular model. Tertiary interactions of the TΨC domain with the D domain in the unimolecular yeast tRNA^Phe were mimicked by the interactions of half-molecules. Incorporation of the highly conserved modifications rT₅₄ and Ψ₅₅, or the commonly occurring m⁵C₄₉, enhanced the affinity of the T-half molecule with the unmodified D-half. However, Ψ₅₅ did not lower the Mg²⁺ requirement for this interaction. Modifications in the T-half molecule did not affect the global conformation of the resulting complexes. Taken together, the enhanced affinity of the modified T-half molecule for the unmodified D-half and the early events of rT₅₄ and Ψ₅₅ syntheses in pre-tRNA (3) indicate that these conserved modifications contribute to the native TΨC and D domain folding interactions.

The TΨC domain folding interaction with the D domain required both the Levitt tertiary structure base pair present in the native yeast tRNA^Phe and the presence of Mg²⁺. No complex was observed in our gel mobility shift analysis when base substitutions completely negated the Levitt pair. Disruption of the Levitt base pair, as well as all D- to T-loop tertiary structure interactions, in an effort to define a functional, minimal sequence tRNA^Leu analog, completely negated aminoacylation by the E.coli leucyl-tRNA synthetase (42). Thus, formation of the Levitt pair may define one of the lowest energy conformations in the RNA’s folding landscape (43) on the path to the native global fold. In tRNA’s three-dimensional structure (4,26), the highly charged phosphate backbone folds sharply at the TΨC and D domain interface. The requirement for Mg²⁺ could be explained by neutralizing electrostatic repulsion (44,45), in particular the large local negative charge at this interdomain interface (24,25). Though high concentrations of Mg²⁺ were required, the unmodified yeast tRNA^Phe transcript exhibited, in NMR spectra, the tertiary and secondary structure hydrogen bonding similar to that of the native tRNA and expected of a folding interaction between the TΨC and D domains (13). Results of Levitt base pair substitutions and site-specific lead cleavages reported here are consistent with the NMR investigation. Cleavage of native tRNA^Phe, an unmodified analog of tRNA^Phe, the complex formed from the unmodified D- and T-half molecules and the complex formed from the unmodified D-half molecule with variously modified T-half molecules all produced similar fragmentation patterns (Fig. 7). A Mg²⁺ requirement for the bimolecular folding interaction in our model system is also consistent with electrostatic interactions dominating the free energy of RNA folding (46). The result is that stable tertiary structures of RNAs are observed only in the presence of divalent metal ions or with very high concentrations of monovalent ions (46).

Our bimolecular model system allows investigation of the contributions of individual modifications to interdomain contacts and folding interactions within tRNA. As with other RNAs, the folding process of tRNA involves a series of early and late folding events (46,47), secondary and tertiary interactions (48), as it is being transcribed in the cell and is then processed as pre-tRNA to the modified and sized, functioning molecule. Initiation of the early, secondary interactions in the tRNA folding process is likely to occur in a time scale (10–100 µs) approximately 1000 times more rapid than the late-occurring, tertiary interactions (49,50). Support that our model system is reporting the late occurring, specific tertiary interactions, is that the ΔG values observed in our model system (Table 1) are of a similar magnitude for interdomain contacts in other RNAs (1.7 ± 2 to 7.6 ± 1.2 kcal/mol) (51). In addition, the specific lead cleavage pattern and the presence of the Levitt base-pair in the completely unmodified D- and T-half molecule suggests that the contributions of modifications are to stabilize, rather than to form, the tertiary contacts.

Modifications in the TΨC domain possibly enhance the correct tertiary structure contacts for tRNA folding interactions. Thus, unproductive pathways and stable, but mis-folded, conformations that have been observed in unmodified RNAs (52) are thwarted. The m¹A₉-dependent folding of a mitochondrial tRNA that has a small D domain loop (53) may be an example of a modified nucleoside required for a late folding event to the native structure. However, one of the most common modifications, rT₅₄, probably contributes to an early secondary structure alteration of the TΨC domain conformation such that the native folding interaction with the D domain would be affected. rT₅₄’s ability to stabilize and affect the local conformation of the T stem and loop (7,20), and the tRNA^fMet species from E.coli and Thermus thermophilus (21), indicates that enhanced secondary structure interactions could be responsible for our observation of the higher affinity of the modified T-half molecule than the unmodified for the D-half. The fact that catalytically inactive yeast m⁵U₅₄-tRNA methyltransferase is required for stabilizing at least one of the cell’s tRNAs, tRNA^Ser_CGA (11), indicates that the modification enzyme may play a role in tRNA folding prior to catalyzing formation of rT₅₄. After methylation of U₅₄, the resulting folded structure would be retained through the stabilizing contributions of the modification.

In contrast to rT₅₄, Ψ₅₅ may directly favor tertiary interactions between the TΨC domain with the D domain without contributing secondary interactions to the TΨC domain alone. The presence of Ψ₅₅ in the otherwise unmodified 17mer TΨC domain did not further stabilize the RNA (7), but the presence of Ψ in base-paired regions provides stability to RNA (16–19,54). Ψ₅₅ may not contribute to local stability of the TΨC domain, as does T₅₄. However, Ψ’s additional hydrogen bonding and/or base stacking capabilities (16–19) would understandably favor tertiary interactions between the TΨC domain and the D domain. Compared with the unmodified T-half molecule, the T-half molecule with Ψ₅₅ had a higher affinity for the D-half molecule. Introduction of Ψ₅₅ into the T-half molecule resulted in an additional –0.7 kcal/mol of free energy for the association of the two half molecules. Similarly, the introduction of the conserved Ψ in snRNA stabilized that RNA’s interaction with an intron branch point by –0.7 kcal/mol (55). These results, together with our finding that the unmodified D-half molecule’s affinity for the doubly modified T-half rT₅₄;Ψ₅₅ (K_d = 11 ± 3 µM) was less than that for the singly modified T-half (rT₅₄, K_d = 6 ± 2 µM), yet greater than that of the Ψ₅₅-modified T-half (K_d = 14 ± 5 µM) and the unmodified T-half (K_d = 40 ± 10 µM), suggest that rT₅₄ and Ψ₅₅ could play different roles in stabilizing the tertiary interactions in tRNA. The contributions of these individual modifications were not additive in our experimental model. When the other, adjacent, modification was present, each could influence the contributions of the other to the RNAs’ folding interactions.

Recent theoretical and empirical approaches to folding describe the folding interactions of RNA molecules with Mg²⁺ as one of loose and diffusely bound hydrated Mg²⁺ ions participating in a rapid, non-specific collapse of the RNA into an intermediate that then finds, within this restricted conformational space, its stable native conformation (56,57). The enhanced folding interaction observed when conserved modifications are present in the TΨC domain may be indicative of modifications reducing conformational space and thus affecting the folding pathway and/or kinetics of reaching the native structure. We have observed by NMR that additions of a modification to an already modified yeast tRNA^Phe anticodon stem and loop domain restricted its conformational space accessible in the direction of the native structure (J.W.Stuart and P.F.Agris, unpublished results). Although we have as yet to identify a stable folding intermediate by native PAGE under conditions of low Mg²⁺ concentration, the application of urea titrations (14,15) and circular dichroism studies to our bimolecular approach may yield information about modified nucleoside contributions to the formation of specific intermediates in the tertiary folding of tRNA. Because tRNA processing enzymes and the translation machinery of the cell are sensitive to the tertiary fold of tRNA, effects of modification on the pathway and rate of tRNA folding and unfolding could determine an organism’s ability to survive (58,59).