Two distinct domains of the β subunit of Aquifex aeolicus leucyl-tRNA synthetase are involved in tRNA binding as revealed by a three-hybrid selection

Abstract

The Aquifex aeolicus αβ-LeuRS is the only known heterodimeric class Ia aminoacyl-tRNA synthetase. In this study, we investigated the function of the β subunit which is believed to bind tRNA^Leu. A yeast three-hybrid system was constructed on the basis of the interaction of the β subunit with its cognate tRNA^Leu. Then, seven mutated β subunits exhibiting impaired tRNA binding capacities were selected out from a randomly mutated library. Two mutations were identified in the class Ia-helix-bundle-domain, which might interact with the D-hairpin of the tRNA analogous to other class Ia tRNA:synthetases complexes. The five other mutations were found in the LeuRS-specific C-terminal domain of which the folding is still unknown. tRNA affinity measurements and kinetic analyses performed on the isolated β subunits and on the co-expressed αβ-heterodimers showed for all the mutants an effect in tRNA affinity in the ground state. In addition, an effect on the transition state of the aminoacylation reaction was observed for a 21-residues deletion mutant of the C-terminal end. These results show that the genetic approach of the three hybrid system is widely applicable and is a powerful tool for the investigation of tRNA:synthetase interactions.

INTRODUCTION

The aminoacyl-tRNA synthetases (aaRSs) catalyze the activation of their cognate amino acids and transfer them to the relevant tRNA molecules. The precise recognition of cognate tRNAs and amino acids by aaRSs guarantees the high fidelity of protein synthesis (1). The specific recognition of tRNAs by their cognate aaRSs depends on a set of identity determinants that are mostly located at the two distal extremities: the anticodon loop and the amino acid accepting stem. Most determinants are in directed contact with cognate synthetases (2), nevertheless, the aminoacylation fidelity is controlled by kinetic differences more than by binding differences (3).

The twenty aaRSs can be classified into two families of ten members each on the basis of conserved sequences and characteristic structural motifs (4). Leucyl-tRNA synthetase (LeuRS), together with arginyl-tRNA synthetase (ArgRS), cysteinyl-tRNA synthetase (CysRS), isoleucyl-tRNA (IleRS), methionyl-tRNA (MetRS) and valyl-tRNA synthetase (ValRS), comprise the class Ia aaRSs. The resolution of at least one crystal structure for each class Ia synthetase shows that the overall structure consists of the typical Rossmann fold and of a specific class Ia α-helical domain called ‘helix-bundle-domain’ (5–15). The helix-bundle-domain binds the tRNA anticodon and can be considered as the signature of the class Ia enzymes. Critical residues involved in anticodon recognition have been identified in ArgRS (16,17), IleRS (18), MetRS (19–25) and ValRS (26). It has also been shown that communication signals between the active site and the anticodon-binding domain are essential for the catalysis (17,27–29). Several idiosyncratic domains are found appended to this class Ia-scaffold of which several are remarkable. For example, a specific N-terminal domain is found in all ArgRSs that binds the D-loop of tRNA^Arg (5,8). Large enzymes (IleRS, LeuRS and ValRS) share CP1 domains involved in tRNA editing (9,13,30). Some bacterial MetRSs exhibit an OB-fold homologous to Trbp111, EMAPII, p43 and Arc1p. Moreover, IleRS, LeuRS and ValRS contain C-terminal domains which recognize the anticodon loop and arm [IleRS, (13)], or the D-loop [ValRS, (26)]. The function and the three-dimensional (3D) structure of the C-terminal domain of LeuRS has still to be determined (7).

The crystallographic structure of Thermus thermophilus LeuRS has been solved in the presence of leucine, leucyl-adenylate analog (7) and several analogs of the distinct pre- and post-tranfer editing substrates (30). The overall architecture is similar to that of IleRS and ValRS except that the editing domain is inserted at a different position of the primary structure and differs by the presence of an additional domain called ‘leucyl-specific’ domain which is located in the catalytic core just before the KMSKS sequence. It has been shown that this domain is essential for the aminoacylation reaction (31). Until now, no structure of the tRNA^Leu: LeuRS complex is available, and data on tRNA^Leu recognition by LeuRS mainly result from mutational studies on the tRNA identity elements. Unlike in other class Ia aaRSs systems, the anticodon triplet is mostly not recognized in LeuRS system. The main identity elements of the tRNA^Leu, which is a class II tRNA with an extra-long variable region, consist of the discriminator base A73 and tertiary interactions between bases from the D- and T-loops (32–39).

In the hyperthermophilic bacterium Aquifex aeolicus, leucyl-tRNA synthetase (αβ-LeuRS) is found as a heterodimer αβ resulting from the spilt of the canonical monomeric LeuRS into two polypeptide chains. In analogy with the T.thermophilus LeuRS of which the 3D structure is known, the split occurs in the leucyl-specific domain (40,41). We previously showed that the fusion of the α and β subunits resulted in a product with an activity comparable to the native αβ-LeuRS (31). Whereas the α subunit cannot be correctly folded when expressed alone, the isolated β subunit composed of 289 amino acid residues (Mr = 33.5 kDa) is stable in vivo and in vitro. Although the β subunit is catalytically inactive, it can bind and recognize tRNA^Leu (40,41). In analogy to the T.thermophilus LeuRS, the β subunit includes half of the leucyl-specific domain, the second half of the Rossmann fold, the helix-bundle-domain and the C-terminal domain (41). In theory, these domains, or parts of them, should be involved in tRNA^Leu recognition. In this study, we examine the contribution of these domains of the β subunit to the tRNA^Leu binding in the ground state, and to the stabilization of the transition state of the aminoacylation reaction. For this we used the yeast three-hybrid system (3HS) developed for the in vivo study of RNA-protein interactions (42,43). A functional 3HS system that uses A.aeolicus β subunit of LeuRS and was set up. A selection of loss-of-binding mutants was subsequently used to identify the β subunit residues involved in tRNA^Leu binding. Mutants were further analyzed after reconstitution of the whole αβ-LeuRS enzymes. This work is the first report of a functional 3HS implicating a tRNA and an aminoacyl-tRNA synthetase. It also shows that the 3HS can be used as a new tool for studying tRNA recognition in vivo.

MATERIALS AND METHODS

Materials

l-Leucine, dithiothreitol (DTT), ATP, CTP, UTP, GTP, 5′-GMP, inorganic pyrophosphatase and 3-amino-1, 2, 4-triazole (3-AT) were purchased from Sigma (USA). [α-³²P]GTP, [¹⁴C]-leucine (300 Ci/mol) and ³²P-labeled tetrasodium pyrophosphate were obtained from Amersham Biosciences (England). GF/C filters were from Whatman Company (Germany). T4 DNA ligase, and restriction endonucleases were obtained from Sangon Company (Canada, Shanghai Branch). T7 RNA polymerase was purified from an overproducing strain in our laboratory (44).

Strains and plasmids

Plasmids were isolated in Escherichia coli strain DH10B. Plasmid pSBETb (45) and pET-15b (Novagen) both with the T7 promoter were used for protein expression in BL21 (DE3). The 3HS procedures were carried out in the Saccharomyces cerevisiae strain L40_coat (46). The shuttle plasmids pACTII with a LEU2 marker, and pIIIA-MS2-1, pIIIA-MS2-2 with a URA3 marker were used to express the hybrid protein and hybrid RNA molecules, respectively, in yeast (46). pIIIA-MS2-1 and pIIIA-MS2-2 differ only in the reverse orientations of the SmaI/XmaI site and in the location of the MS2-binding sites. Hybrid RNAs were transcribed from the yeast RNase P RNA (RPR1) promoter.

Construction of the yeast 3HS

By PCR, using the two primers 5′-CAGGAAACAGACCCCGGGAATT-3′ and 5′-CAAAACAGCCCGGGTTGCATGCCT-3′ (SmaI sites indicated in italic), the genes for E.coli , , and A.aeolicus were amplified from the plasmids carrying the corresponding tRNA genes (42,43). Then, the SmaI digested DNA fragments were inserted into the same site of the plasmid pIIIA-MS2-1 and pIIIA-MS2-2. The recombinant plasmids derived from pIIIA-MS2-1 are named pIIIA-MS2-, pIIIA-MS2- and pIIIA-MS2-. Plasmids derived from pIIIA-MS2-2 are named as pIIIA--MS2, pIIIA--MS2 and pIIIA--MS2. The two series of vectors differ in the position of the tRNA gene relative to the MS2 RNA, for instance in pIIIA-MS2- the MS2 RNA is located at the 5′ end of the tRNA^Leu. The series of pIIIA-MS2-derived plasmids were used to express the hybrid RNAs in yeast 3HS. On the other hand, the gene encoding the LeuRS β-subunit was inserted between the NcoI and BamHI sites of plasmid pACTII; the resulting plasmid was named pACT-LRSB. Then, the yeast strain L40_coat was co-transformed with the pIIIA-MS2-derived plasmids and plasmid pACT-LRSB. Transformants were grown at 30°C on synthetic medium (YNB) lacking uracil, histidine and leucine for selection of the URA3, HIS3 and LEU2 marker genes, respectively. Testing of reporter gene activation was carried out as described (47). Colonies that appeared 5–6 days later were further analyzed for lacZ expression by plating on medium lacking uracil and leucine and supplemented with 80 μg/mL 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal). To test for HIS3 activation, drops (5 μL) of yeast double transformants grown to an OD₆₀₀ = 0.1 were applied to YNB plates deprived of uracil, leucine, histidine and supplemented with different concentrations (0–100 mM) of the competitive His3p inhibitor 3-amino-1, 2, 4-triazole (3-AT). To determine levels of lacZ expression, β-galactosidase activity was measured in extracts from L40_coat cells expressing both the β subunit and the hybrid RNA as described (47).

Expression and purification of A.aeolicus LeuRS and its β subunit

The genes of the α and the β subunit of A.aeolicus LeuRS were inserted into the plasmid pSBET and pET-15b, respectively. The resulting plasmids pSBET-LRSA and pET-LRSB were used to overexpress the α and β subunit. To purify the β subunit, E.coli BL21(DE3) cells were transformed with plasmid pET-LRSB. To purify the heterodimeric form αβ-LeuRS, the two plasmids pSBET-LRSA and pET-LRSB were co-introduced into the E.coli BL21(DE3) cells. Transformants harboring the two recombinant plasmids were grown in 1 l of LB with 100 μg/ml of ampicillin and 50 μg/ml of kanamycin at 37°C. When the cell density reached A₆₀₀ = 0.6, the expression was induced by adding 1 mM isopropyl-1-thio-β-d-galactopyranoside (IPTG) and cells were grown for an additional 4 h. Since all the mutated proteins were thermostable, a heating step at 75°C for an hour was employed to precipitate most of the bacterial proteins. Then, the mutated β subunits and the αβ-LeuRSs were purified according to the procedure used for the wild-type protein (41).

Binding assay of the A.aeolicus LeuRS and its β subunit with tRNA

Labeled A.aeolicus transcripts were prepared as follows: T7 transcripts were generated in a 50 μl reaction mixture, containing 40 mM Tris–HCl (pH 8.0), 5 mM DTT, 10 mM MgCl₂, 1 mM of each ATP, UTP and CTP, 10 μM [α-³²P] GTP (3000 Ci/mmol), 10 mM 5′-GMP, 1 μg of BstNI-digested template DNA, 1000 U/ml RNasin (Takara, Japan), 1 U/ml of inorganic pyrophosphatase and 2 mg/ml of pure T7 RNA polymerase, at 37°C for 4 h. The transcripts were purified by 15% denaturating polyacrylamide gel electrophoresis and folded by heating at 75°C for 5 min in the presence of 5 mM MgCl₂ and slowly cooled down to 30°C. Protein-tRNA interactions were examined by a gel-shift assay. Wild-type LeuRS or mutated β subunits were mixed with the labeled tRNA^Leu in a 5 μl volume containing 100 mM Tris–HCl buffer (pH 6.8), 30 mM KCl, 12 mM MgCl₂ and 0.1 mM EDTA at 0°C for 15 min. A 20% sucrose solution (3 μl) containing tracer dyes was added immediately before loading. The final mixture was then loaded on a 1 mm thick 6% native polyacrylamide gel in Tris–glycine buffer (50 mM Tris–base, 50 mM glycine) and was run for 1 h at 100 V at 8°C in the same buffer. After electrophoresis the gel was dried and radioactive bands were analyzed by phosphorimaging.

Kinetic assays

ATP-PPi exchange and aminoacylation activities of LeuRS were measured at 37°C as described (41). The aminoacylation activity was determined at 37°C in the reaction mixture containing 100 mM Tris–HCl (pH 7.8), 30 mM KCl, 12 mM MgCl₂, 4 mM ATP, 0.5 mM DTT, 10 μM tRNA^Leu, appropriate amounts of [¹⁴C]leucine and enzymes. The kinetic constants of enzymes were determined using various concentrations of the relevant substrates.

RESULTS

Construction of the yeast 3HS based on the interaction of the β subunit of A.aeolicus LeuRS and its cognate tRNA

The first step to develop a functional yeast 3HS is to find a couple of protein and RNA partners able to interact as hybrids with the activation domain of HIS3-GAL and with the MS2-RNA respectively (Figure 1A). The β subunit of A.aeolicus LeuRS was chosen for this assay since it binds its cognate tRNA in vitro (41,48). It was also shown that this complex could be displaced by the E.coli (41). In consequence, we tested both A.aeolicus and E.coli , and E.coli which is chargeable by the A.aeolicus αβ-LeuRS up to one-fourth of the k_cat for the charging of the native tRNA (41). Six recombinant pIIIA-MS2-derived plasmids producing the different hybrid RNAs (see Materials and Methods) were introduced into the L40_coat cells containing plasmid pACT-LRSB. Transformants were then replicated on plates supplemented with X-gal. After 3 days, we found that only the L40_coat cells transformed with pACT-LRSB and pIIIA-tRNA_Aqui^Leu-MS2 exhibited a blue coloration, which reveals an interaction between the RNA and protein partners. The five other pIIIA-derivatives failed to give a productive interaction in the 3HS, suggesting that the complexes between the β subunit and the E.coli tRNA^Arg and tRNA^Leu cannot occur in vivo (Figure 1B). The authenticity of the observed interaction was verified in order to exclude unspecific interactions between the activation domain and the A.aeolicus or between the MS2 RNA and the β subunit (for this we used the empty pIIIA-MS2 and pACTII). Then, the expression level of the reporter gene HIS3 was assayed by growth on plates lacking histidine and supplemented with the inhibitor 3-AT. The highest permissive concentration of 3-AT was obtained for the β subunit and its cognate tRNA^Leu (50 mM) that is higher than the 25 mM measured for the positive control iron responsive element (IRE) and iron regulatory protein (IRP) (Figure 1B, lane 6) (46). Thus, we concluded that the β subunit interacts with its cognate in the yeast 3HS. The interaction is very specific since active substrates like E.coli and cannot form the hybrid, and the interaction can only occur when the tRNA is located at the 5′ end of the hybrid RNA molecule (pIIIA--MS2). Previous studies have reported functional three-hybrid interactions between an aminoacyl-tRNA synthetase and an intron (49) and between a tRNA and intranucleolar component of the nuclear tRNA export machinery (50), but to our knowledge, this is the first report of a functional 3HS based on the interaction of a tRNA and an aminoacyl-tRNA synthetase.

Figure 1.

Three-hybrid formation with the A.aeolicus β subunit and tRNA^Leu. (A) Schematic presentation of the yeast 3HS adapted to a tRNA bait. The two hybrid proteins are the LexA-MS2 coat protein and the LeuRS β subunit in fusion with the GAL4 activation domain. The hybrid RNA consists of the MS2 RNA and the A.aeolicus tRNA^Leu. Three-hybrid formation leads to expression of the two reporter genes HIS3 and lacZ. (B) Assay of the expression level of HIS3 by drop test. The L40_coat yeast cells containing pACT-LRSB and pIIIA--MS2 were grown on 3-AT-containing medium. The highest permissive concentration of 3-AT (50 mM) was measured for the β subunit with its cognate tRNA (lane 3). The positive control is IRP with IRE (25 mM, lane 6). HIS3 activation was not detectable when the GAL4 activation domain alone was expressed with the A.aeolicus tRNA^Leu bait (lane 1), or when the β subunit was expressed with the MS2 RNA alone (lane 2), or when the β subunit was expressed with the E.coli tRNA^Arg (lane 4), or when the β subunit was proposed to the E.coli tRNA^Leu (lane 5).

Selection of β subunits with impaired tRNA binding capacity by the 3HS

We used the three-hybrid system to identify mutants of the β subunit presenting impaired tRNA binding capacity. The loss-of-binding mutants were identified by reduced expression levels of the reporter genes HIS3 and lacZ. Mutants were created by random mutagenesis and site-directed mutagenesis.

First, a library of randomly mutated β subunit genes was screened in order to identify residues that lost tRNA binding when mutated. The randomly mutated pACT-LRSB library was produced by hydroxylamine treatment (47). L40_coat cells containing plasmid pIIIA--MS2 were transformed with the pACT-LRSB library and plated on selective medium lacking uracil and leucine. As before, the plates were replicated onto plates containing X-gal and checked for the coloration. The rare white colonies that appeared among a majority of blue ones were further analyzed. We screened ∼100 000 transformants and isolated 90 white colonies in the first round. A second screening was carried out in order to exclude the false positive clones that resulted from mutations located in the yeast replicon, in the promoter, or in the activation domain of the fusion gene. For this, the mutagenized pACT-LRSB plasmids from the 90 white colonies were isolated and the genes encoding the β subunits were re-cloned in plasmid pACTII. These new pACT-LRSB plasmids were introduced into L40_coat cells containing the plasmid pIIIA--MS2 and were checked again for the blue/white coloration. After this second round of selection only 35 colonies were white from the 90 original colonies. The 35 genes encoding the β subunits from these white clones were inserted into the plasmid pET-15b and expressed in E.coli BL21(DE3). Only 14 genes were successfully expressed and their DNAs were sequenced. Most of genes that did not give any expression product contained premature stop codons at various positions. This agrees well with previous findings, which show that the easiest way to inactivate a protein is to cause truncated forms by means of premature stop codons (16). Among the expressed proteins, two contained nonsense mutations at residues Q269 (Q269stop) and V286 (V286stop) leading to deletions of 21 and 4 residues, respectively, at their C-terminal ends (the whole β subunit contains 289 residues). The other 11 expressed mutants contained the following single mutations: A156V, K160N, Q234H, G237D, and L283F that was found seven times. These mutants are located in the characteristic class Ia α-helix-bundle-domain or in the C-terminal domain which was disordered in the crystal of T.thermophilus LeuRS (7).

Test of site-directed mutants of the β subunit by the 3HS

To complete the study of the tRNA binding site on the β subunit, we identified additional tRNA binding residues by means of site-directed mutagenesis combined with the 3HS selection. Based on multiple sequence alignment of LeuRSs, on the X-ray structure of LeuRS (7), and on the first results from the 3HS selection, we selected a set of six residues that we mutated in Ala in order to remove the contribution of their side chains to a hypothetical contact with the tRNA molecule (Figure 2). Three mutants (N152A, M159A, N163A) were constructed in the third long helix of the helix-bundle-domain (residues 740–761 in T.thermophilus LeuRS), the region where the previously isolated loss-of-binding mutations A156V and K160N are located. These three residues are highly conserved or invariant in the LeuRS sequences. In addition, the quasi-invariant residue R94 located in the close vicinity, but in the adjacent helix, was mutated in Ala. We also mutated two residues of the C-terminal domain, N236 and K238, both highly conserved and located close to the loss-of-binding mutants Q234H, G237D.

Figure 2.

Schematic diagram of the domain structure of T.thermophilus and A.aeolicus LeuRSs and partial sequence alignment of the four β subunit peptides investigated. Amino acid sequences were aligned in the four peptidic regions here studied. Aa, A.aeolicus; Ec, E.coli; Bs, Bacillus subtilis; MSc, mitochondrial S.cerevisiae; Mt, Mycobacterium tuberculosis; Dr, Deinococcus radiodurans; Uu, Ureaplasma urealyticum; Cp, Chlamydia pneumoniae AR39; Cm, Chlamydia muridarum; Hp, Helicobacter pylori; Hi, Haemophilus influenzae; Mp, Mycoplasma pneumoniae; Sc, Streptomyces coelicolor; Nm, Neisseria meningitidis; Tt, T.thermophilus. Amino acids that are completely conserved in all sequences are boxed in black. Homologous residues are boxed in gray. The residues studied in this work with the yeast 3HS are marked with an arrow. The indicated secondary structure elements are from the T.thermophilus LeuRS 3D structure (7) and from the prediction of the GCG SeqWeb 2.0 software for the C-terminal domain.

All the six mutants were tested in 3HS. Coloration defects on X-Gal-containing medium, and measurement of β-galactosidase activity as well as drop tests on 3-AT-containing medium were determined in order to assess the effects of the mutations on the tRNA–protein interaction (Figure 3). Only mutants R94A, M159A and K238A induced phenotypes of loss-of-binding with white phenotypes, and drops of 3-AT tolerance. Mutants N152A and N163A only reduced the interaction slightly, with a light-blue phenotype, whereas N236A exhibited no effect at all.

Figure 3.

Measurement of the tRNA:β subunits interactions by the yeast 3HS. Growth in the presence of the 3-AT inhibitor was tested by drop test and the β-galactosidase activity was assayed on crude extracts from yeast (see Materials and Methods). The vertical axis shows the 3-AT concentrations and the percentage of the measured β-galactosidase activity compared to the activity measured with the wild-type β subunit.

tRNA affinity of the β subunits and αβ-heterodimers for tRNA^Leu

In order to correlate the in vivo results with the 3HS with changes in affinity of the β subunit of LeuRS for tRNA^Leu, we performed band-shift assays using transcript of . This excludes putative effects of the fusions of the protein and RNAs in the hybrids. Eight mutants (R94A, A156V, K160N, Q234H, G237D, L283F, V286stop and Q269stop), that displayed the most severe perturbation in 3HS were assayed by gel mobility shift (Figure 4). The mutated β subunits were overexpressed and purified to homogeneity. Remarkably, the thermostability of these eight proteins was not changed by the heating step of one hour at 75°C, suggesting that the tertiary structure of the mutants was preserved. The tRNA affinity values measured by gel-shift are shown in Figure 4B. They correlate perfectly with the β-galactosidase and 3-AT data obtained in vivo by the 3HS (Figure 3). The six single mutants showed a 2-fold decrease of the tRNA affinity whereas the two deletion mutants exhibited a significantly higher decrease (12-fold).

Figure 4.

Measurement of the dissociation constant of LeuRS proteins for tRNA^Leu by gel-shift assay. (A) In vitro transcribed ³²P-labelled A.aeolicus was incubated with increasing concentrations of the β subunits or whole αβ-LeuRS (see Materials and Methods). Apparent dissociation constants (K_d) of the proteins were estimated from quantification of the free and retarded tRNA. Band shifts are displayed along the linear schematic representation of the β subunit (NH₂- to the COOH-end). For clarity the protein concentrations were omitted from each picture. The large increases of K_d were shown by a shift of the gel picture towards the right side, which symbolizes the high values. (B) K_d values and binding energy variations for the β subunits and αβ-LeuRS proteins.

Then, the eight mutants R94A, A156V, K160N, Q234H, G237D, L283F, V286stop and Q269stop, were co-expressed with the wild-type α-subunit in E.coli and the mutated heterodimers were purified to homogeneity. The αβ-LeuRSs were thermostable at 75°C for 1 h, like the β subunits displaying the same mutations. The K_d values of αβ-LeuRSs for were determined by gel-shift assay. Under these experimental conditions, the K_d of the wild-type αβ-LeuRS for its tRNA is 0.8 μM. K_d values from the mutated αβ-LeuRSs range from 1.2 to 30 μM (Figure 4). The resulting loss of tRNA binding energy is basically the same as for the β subunits alone (Figure 4), except for the deletion mutants V286stop and Q269stop which display a significant higher loss of binding energy when in the heterodimeric complex with the α subunit.

Kinetic analysis of the αβ-LeuRS proteins

Four additional αβ-heterodimers (N152A, M159A, N163A and K238A) were purified. The kinetic parameters of the 12 mutated αβ-LeuRSs for A.aeolicus tRNA^Leu were measured at the steady-state level. Both ATP-PPi exchange and aminoacylation activities were measured (Table 1). The ATP-PPi exchange activities were unaffected by the substitutions and deletions, excluding the participation of the C-terminal end of the β subunit to the leucine activation. Most of the mutated proteins that exhibited binding losses in the 3HS and gel-shift assays, showed moderate variations of their aminoacylation rates, except for N152A and Q269stop which showed a 5- and 100-fold reduction. K_m values for tRNA^Leu were only increased for the deletion mutants V286stop and Q269stop. In non-saturating conditions the tRNA aminoacylation of the Q269stop mutant is almost undetectable. Thus, removal of 21 residues at the C-terminal end induces an almost complete decoupling of the two enzyme's activities, and renders the enzyme only capable to catalyze the leucyl-adenylate formation.

Table 1. Kinetic constants for A.aeolicus αβ-LeuRSs at 37°C.

Shaded boxes mark significant changes of the catalytic parameters.

DISCUSSION

The yeast 3HS has been developed for the research and for the study of interactions between RNA and protein (42,43). Here we report the first successful 3HS based on the interaction of tRNA and an aminoacyl-tRNA synthetase, more precisely between the A.aeolicus tRNA^Leu and the β subunit of its homologous LeuRS. Remarkably, this interaction occurs in vivo despite the fact that the dissociation constant measured in vitro for these two molecules is rather low (∼5.8 μM) when compared to typical K_d values measured for whole aminoacyl-tRNA synthetases with their cognate tRNA (∼1 μM and less). Despite this high dissociation constant, the β subunit and tRNA^Leu, in fusion with the HIS3-lacZ activation domain and in fusion with MS2 RNA respectively, interact in vivo in the 3HS and the binding strength was even higher than the positive control (IRE). This interaction was specific for the A.aeolicus tRNA, since E.coli tRNA^Leu that is recognized and charged by the whole αβ-LeuRS (41), failed to produce a functional 3HS. Thus, the E.coli tRNA^Leu can only give a productive complex in the presence of the α subunit that may provide additional binding energy. We previously demonstrated that in vitro, E.coli tRNA^Arg is able to compete and to displace a complex formed by the A.aeolicus β subunit and A.aeolicus tRNA^Leu (41). Nevertheless, we show here that tRNA^Arg does not form a complex with the β subunit of LeuRS, since no functional 3HS was generated in vivo, suggesting that the cellular environment required for three-hybrid formation is more stringent than the in vitro conditions used for complex formation.

Two special considerations about the hybrid RNA should be taken into account when the yeast 3HS is applied to a tRNA–protein system. The first is that tRNA processing enzymes present in the yeast cells can process the hybrid tRNA molecules. We only detected the expression of the reporter genes when the prokaryotic A.aeolicus tRNA^Leu was located at the 5′ end and not at the 3′ end of the hybrid RNA molecule. This difference might be attributed to the destruction of the tRNA-like molecule by the tRNA processing enzymes. A second consideration might be that the appending of the MS2 RNA to the acceptor stem of tRNA interferes with the tRNA–enzyme interaction. These complications suggest assaying the tRNA bait on both 5′ and 3′ ends of the MS2 RNA, but also assaying tRNA mutants deficient in processing as well as adding RNA spacers to prevent the steric hindrance.

As a direct application of the efficient 3HS, we selected mutations with loss-of-binding phenotypes and demonstrated that the yeast 3HS is sensitive enough to map the tRNA binding site on the β subunit of A.aeolicus LeuRS. Five point mutations and two C-terminal deletions that exhibited drastic decreases in the binding strength were selected from a randomly mutated library. On the other hand, based on other class Ia tRNA-aaRS complexes, and based on sequences homologies with LeuRSs from different organisms, six residues located in the potential tRNA^Leu binding site were altered by site directed mutagenesis. Severe effects were observed for three of them, and moderate or no effects for the three others. All together, the loss-of-binding effects observed for these mutants underline the strength of the 3HS to map and identify specific molecular interactions that are important to tRNA–enzyme complex formation.

We found that the mutated β subunits exhibited changes in the K_d range of 5.6 up to 70 μM. In this range, the yeast colony phenotype varied from the deep-blue (K_d = 5.6 μM) to the white (K_d = 70 μM). The variation in binding energies resulting from these variations in affinities ranges between −0.21 to −0.54 kcal/mol for the single substitution mutants [ΔΔG_b β_mut/β_wt, Figure 4B, (51,52)]. These energy losses might correspond to the binding energy of a weak hydrogen bond. For the two deletion mutants, a −1.49 kcal/mol loss in binding energy was calculated, which suggests that stronger interactions have been lost. In the αβ heterodimers, the variations in binding energy are basically unchanged (ΔΔG_b αβ_mut/αβ_wt, Figure 4B) except for the V286stop and Q269stop mutants which showed a significant decrease of the binding energy (−2.1 kcal/mol). This suggests that the assembly of the deleted β subunits with the native α subunit induces an additional negative impact on the tRNA binding, which might be related to some structural alteration of the β subunit itself or of the α subunit by a distal effect.

According to the moderate loss of binding energy that was observed for the single substitution mutants, the mutations cause only slightly altered k_cat and K_m values for the aminoacylation using the αβ constructions. With the two C-terminal deletion mutants, remarkable effects on the aminoacylation catalysis were observed. Both exhibited severe increases in the K_m for tRNA^Leu, and whereas the shortest deletion, V286stop (4 residues), exhibited no change in its catalytic rate, a 100-fold decrease in the k_cat for the Q269stop mutant (21 residues deleted) was measured.

LeuRS belongs to the class Ia synthetases that comprise ArgRS, CysRS, IleRS, LeuRS, MetRS and ValRS. LeuRS is more closely related to IleRS and ValRS, a subgroup characterized by the presence of large editing domain (CP domain), and C-terminal extensions appended to the idiosyncratic class Ia helix-bundle-domain. So far, three class Ia synthetases structures have been solved in complex with their tRNA, the Staphylococcus aureus IleRS (13), the T.thermophilus ValRS (9,26) and the S.cerevisiae ArgRS (8). The tRNAs found in these complexes interact by their D-loop side and approach the active site by the minor groove of their acceptor stem, according to the class I rules.

The helix-bundle-domain shared by the six class Ia synthetases is the major constituent of the tRNA anticodon recognition. The first and third α-helices of the bundle harbor-specific residues that are involved in tRNA–protein interactions (8,13,26). A 3D model of the β subunit interacting with tRNA has been previously proposed according to the 3D structure of T.thermophilus LeuRS and sequences homologies with the A.aeolicus LeuRS (Figure 5) (31). Unfortunately, the 63-residues long C-terminal domain could not be modeled since it is disordered in the crystal structure probably due to some flexibility or mobility property (7). If we consider the seven mutants that we have selected by the 3HS selection, only two of them can be localized on the 3D structure of the β subunit (A156V and K160N). Their homologs in the T.thermophilus enzyme are A746 and E750, and are located in the third long helix of the bundle. If we superimpose the helix-bundle-domain of LeuRS onto the equivalent domains of the Ile-, Val- and ArgRS complexed with their tRNA, it appears that A746 and E750 might interact with the D arm of tRNA^Leu at the level of the base pairs 13:22, or 14:21. In this study, we showed that the mutation of these two residues in the A.aeolicus enzyme produces exclusive decreases in the stability of the complex in its ground state, without any effect on the transition state of the aminoacylation reaction.

Figure 5.

Partial representations of the LeuRS structure from T.thermophilus. (A) View of the leucine-specific domain (shown in magenta) and helix-bundle-domain (shown in yellow) of T.thermophilus LeuRS (7) corresponding to the β subunit of A.aeolicus LeuRS (31). T.thermophilus tRNA^Val from the ValRS:tRNA^Val complex (26) was docked after superimposition of the helix-bundle-domains of the T.thermophilus LeuRS and ValRS. Base pairs 11:24, 12:23, 13:22 and 14:21 of the D-arm are shown. The 63 last amino acid residues of the protein are missing in the structure (7). (B) View focused on the helix-bundle-domain of T.thermophilus LeuRS. Mutated residues are labeled according to the β subunit numbering. These residues are conserved in the two enzymes, except K160 of the A.aeolicus subunit that is E750 in T.thermophilus.

The five other mutants selected by 3HS are found in the unsolved C-terminal domain. The single substituted residues showed a moderate negative effect on the tRNA binding in its ground state. In addition, the 21-residues deletion mutant was impaired in the stabilization of the transition state of the aminoacylation reaction. This effect is similar to those observed for mutants of the C-terminal coiled-coil domain of ValRS, which interacts electrostatically with A20 and hydrophobically with the G19:C56 tertiary base pair, providing binding energy and tRNA structure stabilization (26). ArgRS, which is not really a ‘large class Ia synthetase’ in the sense it has no editing domain, uses instead of a C-terminal domain a specific N-terminal domain that recognizes the D-loop side of tRNA^Arg (8). Thus, at least for ValRS and ArgRS, the binding of the single stranded tRNA region called ‘variable pocket’ formed by the D- and T-loops is crucial (53,54). This binding may promote recognition but also structural stabilization of the tRNA L-shape during the aminoacylation reaction (26,55). Collectively, these data highlight the recruitment of new domains by these enzymes in order to recognize, stabilize or bind tRNA atoms located on its D-loop face. Concerning the tRNA^Leu binding on the A.aeolicus LeuRS no experimental data are presently available. In analogy to the E.coli tRNA^Leu, the discriminator base and a specific tertiary network of interactions located in the elbow region should be essential for the leucine identity (32–35,39). No identity elements have been found in the bacterial anticodon loop, in contrast to the S.cerevisiae tRNA^Leu (36). Since the tertiary structure of the tRNA^Leu elbow region is crucial for its identity, it is tempting to hypothesize that the C-terminal domain of LeuRS occupies the spatial location of the coiled-coil domain of ValRS that stabilizes the elbow structure of tRNA^Val. At this location, it would check the tertiary structure features of the tRNA that control the leucine identity. We cannot rule out the possibility that the C-terminal domain is folded in some way that allows also the checking of the variable stem of the tRNA. Indeed, LeuRS from the archaea Haloferax volcanii measures the length of the variable stem and recognizes the nucleotides at position 47C, 47D, 47H (38). Previous studies on other synthetases that recognize the class II tRNA have shown that both SerRS and TyrRS recognize the shape of the variable stem of their cognate tRNA by an N- and C-terminal domain, respectively (56,57). Definitive answer to these questions would need the resolution of a complex between the A.aeolicus αβ-LeuRS with its tRNA.