Magnesium Ion-mediated Binding to tRNA by an Amino-terminal Peptide of a Class II tRNA Synthetase*

Aspartyl-tRNA synthetase is a class II tRNA synthetase and occurs in a multisynthetase complex in mammalian cells. Human Asp-tRNA synthetase contains a short 32-residue amino-terminal extension that can control the release of charged tRNA and its direct transfer to elongation factor 1α; however, whether the extension binds to tRNA directly or interacts with the synthetase active site is not known. Full-length human AspRS, but not amino-terminal 32 residue-deleted, fully active AspRS, was found to bind to noncognate tRNAfMet in the presence of Mg2+. Synthetic amino-terminal peptides bound similarly to tRNAfMet, whereas little or no binding of polynucleotides, poly(dA-dT), or polyphosphate to the peptides was found. The apparent binding constants to tRNA by the peptide increased with increasing concentrations of Mg2+, suggesting Mg2+mediates the binding as a new mode of RNA·peptide interactions. The binding of tRNAfMet to amino-terminal peptides was also observed using fluorescence-labeled tRNAs and circular dichroism. These results suggest that a small peptide can bind to tRNA selectively and that evolution of class II tRNA synthetases may involve structural changes of amino-terminal extensions for enhanced selective binding of tRNA.

Aminoacyl-tRNA synthetases catalyze the covalent attachments of amino acids to cognate tRNAs in the first step of protein biosynthesis. Extensive studies of the structure and function of this family of enzymes have provided excellent understanding of fundamental principles of RNA-protein interactions and structures of synthetases. Almost all synthetases contain a core catalytic domain carrying out adenylation of an amino acid and an anti-codon binding domain essential for aminoacylation of cognate tRNA. The core catalytic domains in class I synthetases resemble dinucleotide binding folds and reside in the amino termini, whereas a seven-stranded antiparallel sheet with three ␣-helices characterizes active sites in class II synthetases and locates in the carboxyl termini of synthetases.
Beyond the basic amino-and carboxyl-terminal catalytic domains, eukaryotic and mammalian synthetases (1-3) have evolved idiosyncratic extensions dispensable for aminoacylation of tRNA. Additionally, extensive association of synthetases occurs in high eukaryotic organisms; thus, 9 of the 20 synthetases associate as a multienzyme complex, ProRS and GluRS form a fused GluProRS (4,5), and ValRS associates with the EF1H as a separate complex (6,7). The function of the extensions in synthetases has attracted more attention recently. Controlled proteolysis or hydrophobic interaction chromatography dissociates several synthetases from the synthetase complex without significantly affecting enzymatic activities (8,9); thus, the extensions in mammalian synthetases likely play pivotal roles in the structural organization of the synthetase complex. Extensions in synthetases can be involved in RNA binding (10,11) or function as cytokines (12)(13)(14). It appears that extensions in synthetases are multifunctional in at least some cases. Multienzyme complexes of aminoacyl-tRNA synthetases also provide a model for studying organization of cellular protein biosynthetic machinery (15,16). The primary and quaternary structures of the protein components and the gross structure of the synthetase complex are well established, but the structural and functional organization of the synthetase complex is not well understood.
AspRS 1 is a class II synthetase, its primary structures from more than 50 sources have been determined, and the threedimensional structures from bacteria and yeast have been resolved (17,18). Mammalian AspRS occurs as the smallest synthetase in the multisynthetase complex and has been used as a model for better understanding of the multisynthetase complex and the extensions in synthetases. Human AspRS (hDRS) has a very short, 32-residue, near neutral extension (19,20) as opposed to the 71-residue, highly basic extension in the corresponding yeast AspRS (21). All other mammalian synthetases have longer extensions than AspRS (1); in particular, IRS has an extension of 180 residues. Deletion of the amino-terminal extension in mammalian AspRS affects the stability, the Michaelis-Menten parameters (22), the dimerization (12), and cellular localization (23,24), but deletion mutants remain fully active. Interestingly, the amino-terminal extension reduces the rate of release of charged tRNA from hDRS (11), and elongation factor 1␣ with GTP stimulates the charged tRNA release; thus, the extension in hDRS directly or indirectly mediates the direct transfer of charged tRNA from RS to elongation factor 1␣ (25). The full-length hDRS has appreciably lower K m to tRNA than that of the truncated hDRS (22). Altogether, the results of the steady-state and single-turnover kinetic analyses suggest that the hDRS extension likely interacts with tRNA or the active site in a manner that may provide additional understanding of the process of synthetase⅐RNA interactions. Furthermore, in contrast to the lysine-rich extensions described in several recent reports (10,26,27), the 32-residue extension in hDRS is short and is not highly basic. In this paper, we focus on the tRNA binding by the amino-terminal extension in hDRS using fluorescence spectroscopy and circular dichroism (CD). We report that the hDRS extension indeed binds to tRNA and mag-nesium ion plays an essential role in this case of a neutral nonlysine-rich extension. Preliminary reports of this study have appeared previously (28).

EXPERIMENTAL PROCEDURES
Unfractionated Escherichia coli tRNA, yeast tRNA, and E. coli tRNA fMet were obtained from Roche Molecular Biochemicals. Rabbit reticulocytes were purchased from Green Hectare. The synthetase complex was purified from rabbit reticulocytes as described previously (29). Ethidium bromide, trifluoroethanol, octadecaphosphate, soluble calf thymus DNA, and ribonuclease were obtained from Sigma. Poly(A), poly(U), poly(dA), and poly(dA-dT) were purchased from Amersham Pharmacia Biotech. The full-length (hDRS) and the amino-terminal 32-residue-deleted (hDRS⌬32) forms of human AspRS were expressed in bacteria as glutathione S-transferase fusion proteins, subsequently thrombin-digested, and purified as free synthetase as described previously (22). The sequence of the 21-mer synthetic peptide, hDRS (Thr 5 -Lys 25 ), was AcTQRKSQEKPREIMDAAEDYAK-amide, and the sequence of the hexadecamer peptide, hDRS (Asp 12 -Arg 27 ), was AcDPREIMDAAEDYAKER-amide. Both peptides were synthesized and purified using RPC-C18 high pressure liquid chromatography to be more than 95% pure by Peptide Technology (Bethesda, MD). The molecular weights of hDRS (Thr 5 -Lys 25 ) and hDRS (Asp 12 -Arg 27 ) were 2535 (expected 2535) and 1950 (expected 1950), respectively, as determined by mass spectrometry.
Steady-state fluorescence measurements were carried out using a Hitachi PerkinElmer Life Sciences MPF2A spectrofluorimeter and 0.4ϫ 0.4-cm 2 Hellma quartz microcuvettes at room temperature as described previously (30). Fluorescence titrations were performed by adding 1-to 5-l aliquots to an initial volume of 250 l in a standard buffer containing 10 mM Tris-HCl (pH 7.5) and 4 mM MgCl 2 unless specified otherwise. Samples for fluorescence measurements had absorbance Ͻ 0.05 absorbance unit at the excitation wavelength. The extents of decrease of fluorescence intensity due to inner filter effect were determined by carrying out parallel titrations with tyrosine or tryptophan at the same molar concentration as that of the peptide or protein. The net extents of quenching were used for calculating binding constants.
CD measurements were carried out using a Jasco700 CD spectrophotometer under nitrogen at 10 p.s.i. as described previously (11). The wavelengths were calibrated using camphor sulfate. The spectra were recorded using 1-cm cells and scanned at least three times from 190 to 300 nm. CD titrations were carried out by adding 1-l aliquots to an initial volume of 300 l in a standard buffer of 10 mM potassium phosphate (pH 7.5) and 4 mM MgCl 2 unless specified otherwise. CD spectra were deconvoluted using the convex constraint analysis (31) to resolve the five fundamental component CD spectra for various secondary structures of a given sample and 25 proteins with known secondary structures using software supplied by Jasco.
Binding of tRNA to peptides was analyzed by fluorescence quenching at varying concentration of peptides (⌬F) and saturating concentration of peptides (⌬F max ). Combining these values with the equation of conservation of mass and the Scatchard equation, Eq. 1 can be derived: where R 0 , P 0 , n, and K d are the total concentration of tRNA, the total concentration of peptide, the binding stoichiometry, and the dissociation constant, respectively. Eq. 1 was used instead of the simple hyperbolic equation to include correction of the total concentration of peptide to the concentration of unbound peptide. Nonlinear least square fit was carried out using Sigma Plot. CD titration was similarly analyzed except the net increases in ellipticity above the sum of tRNA and peptide ellipticities were used by subtraction of ellipticities of peptide and tRNA alone from those of mixtures of peptide and tRNA.
Binding of Mg 2ϩ to tRNA was analyzed by fluorescence quenching of tRNA-bound ethidium by Mg 2ϩ at varying concentrations of Mg 2ϩ (⌬F) and at saturating concentration of Mg 2ϩ (⌬F max ). The Hill equation (Eq. 2) was used to obtain Hill coefficients: where [Mg 2ϩ ], h, n, and K d are the total concentration of MgCl 2 , the Hill coefficient, the stoichiometry, and the dissociation constant, respectively.

RESULTS
Binding of tRNA fMet to hDRS and hDRS⌬32-The aminoterminal extension in hDRS dramatically reduces the rate of release of charged tRNA from hDRS (11), suggesting that tRNA interacts directly or indirectly with the amino-terminal extension in hDRS. Noncognate tRNA fMet from E. coli was used to examine tRNA binding capability of the amino-terminal extension in hDRS. Noncognate tRNA was used, because the relatively small amino-terminal extension was not expected to exhibit tRNA identity and noncognate tRNA should not bind to the cognate tRNA binding site. As shown in Fig. 1, the fulllength hDRS bound to tRNA fMet as judged by the quenching of the intrinsic fluorescence of hDRS, and an apparent dissociation constant of tRNA fMet to hDRS of 0.27 M was obtained based on fluorescence titration. No fluorescence quenching of hDRS by tRNA fMet was observed in the absence of MgCl 2 suggesting Mg 2ϩ is essential in the noncognate tRNA binding. The linear decrease of the fluorescence in hDRS⌬32 was due to the UV absorption by tRNA, and the negligible fluorescence quenching of hDRS⌬32 by tRNA fMet precluded determination of a binding constant. The large difference in the tRNA fMet affinity between hDRS and hDRS⌬32 suggests that the aminoterminal extension is needed for hDRS to bind to noncognate tRNA.
Binding of the 21-mer Peptide hDRS (Thr 5 -Lys 25 ) to tRNA fMet -The binding of tRNA fMet to hDRS could result from interaction of the amino-terminal extension with the active site or attribute to a direct binding of tRNA by the extension. The interaction of tRNA fMet with the amino-terminal extension in hDRS was next investigated using synthetic peptides. A 21residue peptide hDRS (Thr 5 -Lys 25 ) that contains a putative amphiphilic helix in the amino-terminal extension in hDRS was synthesized. The fluorescence intensity of ethidium enhances over 20-fold upon binding to a single strong binding site in tRNA (32,33), and the fluorescence of tRNA-bound ethidium was monitored to assess binding of tRNA to amino-terminal peptides. As shown in Fig. 2, the fluorescence of tRNA-bound ethidium was quenched 25% by hDRS (Thr 5 -Lys 25 ) at 4 mM MgCl 2 . In the absence of MgCl 2 , on the contrary, the fluorescence of tRNA-bound ethidium was not at all affected by the peptide. Free ethidium fluorescence in the absence of tRNA was not affected by the peptide under the same conditions either. Fluorescence titration of tRNA fMet and hDRS (Thr 5 - Lys 25 ) gave an apparent dissociation constant of 1.7 M. The number of peptide binding sites in tRNA was determined by Job analysis (34,35), in which the fluorescence intensities were determined at varying molar ratios of the peptide to tRNA without changing their total molar concentration. As shown in Fig. 2, the maximal change was obtained when tRNA and the peptide were present at a 1:1 molar ratio; thus, there is one binding site per tRNA molecule.
When the Tyr fluorescence of the peptide was monitored, a moderate quenching by tRNA fMet , 11%, was observed after correction for the inner filter effect. In the absence of MgCl 2 , the Tyr fluorescence of the peptide was not quenched by tRNA fMet .
When the CD spectrum of the peptide was compared with that of the peptide in the presence of tRNA after subtraction of CD of tRNA alone, significant changes of the peptide CD between 200 nm and 250 nm were observed (Fig. 3). Comparison of the CD spectra of tRNA in the absence and in the presence of the peptide (after subtraction of CD of peptide) indicated that no detectable changes in the spectrum between 250 nm and 300 nm where tRNA absorbs. Again, no CD changes of the peptide or tRNA were found in the absence of MgCl 2 . An apparent dissociation constant of 0.9 M was obtained by monitoring ellipticity at 208 nm in the binding of tRNA fMet to the peptide.
The binding of the peptide to tRNA fMet appears to be selective to tRNA among polyanions and polynucleotides. As shown in Table I, the fluorescence of poly(dA-dT)-bound ethidium was not affected by the peptide hDRS (Thr 5 -Lys 25 ). The Tyr fluorescence of hDRS (Thr 5 -Lys 25 ) was also monitored for binding to nucleic acids other than tRNA. As shown in Table I, tyrosine fluorescence quenching was small for poly(A), poly(U), poly(dA-dT), or an oligodeoxynucleotide. The tyrosine fluorescence of the hDRS (Thr 5 -Lys 25 ) was not at all affected by polyphosphate (Table I). Finally, the ellipticity of the peptide was monitored to detect the binding of the peptide to polyphosphate and oligodeoxynucleotides. No change in the CD spectrum of the peptide was observed upon addition of polyphosphate. When the CD spectra of the peptide were compared in the absence and presence of the oligodeoxynucleotide after correction of the oligodeoxynucleotide CD, no changes in the CD spectrum of the peptide were found.
The binding of hDRS (Thr 5 -Lys 25 ) to tRNA fMet was analyzed at varying concentrations of ethidium to determine whether hDRS (Thr 5 -Lys 25 ) and ethidium bind competitively to tRNA fMet . The binding of the hDRS (Thr 5 -Lys 25 ) to tRNA fMet was determined at varying concentrations of ethidium in 10  Roles of MgCl 2 on tRNA Binding-Mg 2ϩ was needed for the binding of the amino-terminal peptide to tRNA fMet as demonstrated by fluorescence quenching and CD changes. Divalent cations such as Ca 2ϩ and Zn 2ϩ or the monovalent cation Na ϩ could not substitute for Mg 2ϩ in the binding of the peptide to tRNA fMet as judged by the extent of quenching of the fluorescence of tRNA-bound ethidium by the peptide. The roles of Mg 2ϩ ion in the interaction of tRNA and the peptide could be maintaining the proper conformation of tRNA or mediating the formation of the complex.
To assess the role of Mg 2ϩ in the binding of the peptide to tRNA, fluorescence titration of the tRNA-bound ethidium with the peptide was carried out at varying concentrations of MgCl 2 . The effect of Mg 2ϩ may be represented by the mass action of the process: When K a is defined in terms of the equilibrium concentrations of tRNA, peptide and the complex, then it is readily shown that where ⌬n is the difference of the average numbers of Mg 2ϩ bound in free tRNA and peptide-bound tRNA. The apparent binding constants of the peptide to tRNA, K a , increased with increasing concentrations of MgCl 2 . The linear relationship of the plot of log K a versus log [Mg 2ϩ ] gave a slope of 1.33 (Fig. 5).
A slope of greater than one suggests at least one additional Mg 2ϩ ion bound in the tRNA⅐peptide complex upon the formation of the complex as compared with free tRNA. The association of Mg 2ϩ with tRNA was next analyzed by monitoring the fluorescence changes of tRNA fMet -bound ethidium by Mg 2ϩ . As shown in Fig. 6, the binding of Mg 2ϩ is highly cooperative and the binding isotherm gave a Hill coefficient of 3.4. This is in good agreement with earlier studies (36). When the association of Mg 2ϩ with tRNA fMet was simi-larly analyzed in the presence of a saturating amount of the 21-mer peptide, a Hill coefficient of 4.5 was obtained.
When the effect of NaCl on the binding of the peptide to tRNA in the presence of MgCl 2 was analyzed, in contrast to the stimulatory effect of Mg 2ϩ , the extent of fluorescence quenching decreased at increasing concentrations of NaCl. The apparent binding constants of the peptides to tRNA fMet also decreased at increasing concentrations of NaCl and gave a linear relationship of log K a versus log[NaCl] with a slope of Ϫ0.2.
Binding of the 16-mer Peptide hDRS (Asp 12 -Arg 27 ) to tRNA fMet -A shorter 16-mer peptide in the amino-terminal extension of hDRS was also examined. The peptide hDRS (Asp 12 -Arg 27 ), which has an additional acetylated Asp as a helix cap and an additional basic Arg residue at the carboxyl terminus, has significantly higher propensity of forming helical secondary structure than hDRS (T5-25) (11). The binding of tRNA fMet by the peptide hDRS (Asp 12 -Arg 27 ) was similarly analyzed. As shown in Fig. 7, the fluorescence of ethidium noncovalently bound to tRNA fMet was similarly quenched by hDRS (Asp 12 -Arg 27 ), showing an apparent dissociation constant of 8.9 M. Again, in the absence of Mg 2ϩ , no change of the ethidium fluorescence was observed.
Molecular Model of the Peptide⅐tRNA Complex-Although the three-dimensional structures of yeast and bacteria AspRS have been determined, the amino-terminal extension was mobile and its structure so far remains unresolved. The conformation of the peptide was first modeled based on its 48% sequence identity to Salmonella typhimurium glutamine synthetase (97-117) in the Rutgers Protein Data Base. Coordinates of the peptide backbone were assigned to the peptide according to the coordinates of the reference glutamine synthetase (PDB 2LGS). The energy of the structure was minimized using the DISCOVER Homology module from Biosym by running 1000 dynamics steps followed by 100 equilibration steps at 300°K and 1-ps time step. The dynamics process was followed by 3000 energy minimization steps. The lowest energy was Ϫ118 kcal/mol, and the predicted secondary structure had 11 hydrogen bonds. The peptide was then docked, using the DIS-COVER Affinity module, onto tRNA Phe in the Rutgers Protein Data Base with all possible orientations and 100 iterations while the intermolecular energy between the peptide and tRNA was monitored. The lowest energy among docked structures was Ϫ99 kcal/mol. Binding of the peptide to tRNA involved conformational changes in the peptide and resulted in a slightly higher disordered secondary structure than that in the free peptide. A stereo view of the backbone model of the complex of hDRS (Thr 5 -Lys 25 ) with tRNA Phe is shown in Fig. 8.

DISCUSSION
Nonspecific tRNA Binding by the Amino-terminal Extension of a Class II Synthetase-The present study shows that the amino-terminal peptide in hDRS binds to tRNA based on fluorescence and CD changes in tRNA fMet or the peptide. Ethidium was used as an extrinsic fluorescence probe in analyzing the binding of tRNA by the peptide, and increasing the concentration of ethidium was found to reduce the affinity of tRNA to the peptide. A tighter binding to tRNA by the peptide can be expected in the absence of ethidium under the same conditions. Indeed, a slightly higher apparent binding constant based on the ellipticity at 208 nm was observed as compared with those obtained using fluorescence-labeled tRNA or ethidium-bound tRNA. Both fluorescence and CD results demonstrated the binding of tRNA by the peptide. The binding appears to be selective to tRNA, because other polynucleotides and polyphosphate did not bind to the synthetic peptide. These results are in notable contrast to the amino-terminal peptide in yeast DRS. Polyphosphate and DNA bind indiscriminately to the highly basic extension in yeast DRS (37). It appears that the amino-terminal extension in DRS evolved from a highly basic extension in yeast to a nearly neutral extension in mammals and resulted in its selective binding to tRNA. To our knowledge, this is the first time an amino-terminal peptide was found to bind tRNA selectively. Several nonspecific tRNA binding domains have been found recently in GlnRS (26), AspRS (10), and AlaRS (27) and are characterized by clusters of basic residues in the tRNA binding domains. The nonspecific RNA binding domains evidently facilitate the formation of a specific and essential tRNA-protein complex and optimize recognition of tRNA. In addition, a zinc binding domain distal to the tRNAbinding site in IRS is essential in tRNA binding (38). The mechanisms of tRNA recognition through these RNA binding domains in synthetases are yet to be elucidated. to be explored. Na ϩ cannot replace Mg 2ϩ in facilitating the binding of the peptide to tRNA, and Na ϩ actually reduces tRNA affinity to the peptide. The slope of log K a versus log [Na ϩ ] was small and negative in comparison to those such as drug⅐DNA interactions dominated by electrostatic interaction. The apparent dissociation constants of hDRS (Thr 5 -Lys 25 ) and the 16-mer to tRNA fMet were found to be 1.7 and 8.9 M, respectively. Three basic residues in hDRS (Thr 5 -Lys 25 ) were absent in the 16-mer; thus, additional basic residues in hDRS (Thr 5 -Lys 25 ) moderately enhanced affinity to tRNA. It appears that electrostatic interactions between tRNA and the basic residues in the extension may complement the roles of Mg 2ϩ in tRNA binding.

Roles of Magnesium Ion and Electrostatic Interaction-Mg
Conformation of the Peptide⅐tRNA Complex-The molecular model of the peptide, which showed a disordered amino half of the peptide before Pro and a partial helical carboxyl half after Pro, is in general consistent with previously reported CD analyses. CD of a 22-mer, hDRS (T5-E26), and a 16-mer, hDRS (Asp 12 -Arg 27 ), under various conditions indicated hDRS (Asp 12 -Arg 27 ) has a high propensity of forming helical secondary structure and hDRS (T5-E26) has a low propensity (11). In the presence of trifluoroethanol, hDRS (Asp 12 -Arg 27 ) exhibited up to 61% helical content. However, the model must be considered tentative. Molecular modeling of the tRNA⅐peptide complex suggests that the peptide is preferentially located at the inside surface of the Lshaped tRNA with proximity to the acceptor and anti-codon stems. Such a model is consistent with the experimental data but must be considered tentative. It is intriguing that the binding of the peptide to tRNA is inhibited by ethidium, considering that the ethidium binding site is near the base of the acceptor stem as localized by singlet-singlet energy transfer experiments (39). Because binding of ethidium to tRNA could be nonintercalative and nonfluorogenic, the nature of the competition of the peptide with ethidium remains to be clarified; nonetheless, the quenching of fluorescence of ethidium in tRNA fMet is consistent with the suggestion that the peptide binds near the base of the acceptor stem. The quenching of the fluorescence of tRNA (8 -13) and unchanged fluorescence in tRNA labeled at the 3Ј-end and D20 by hDRS (Thr 5 -Lys 25 ) are also consistent with the model of the peptide and tRNA.
The Multisynthetase Complex and the Synthetase Extensions-The present results support the likelihood that the evolution of extensions in mammalian synthetases may facilitate transferring tRNA between the periphery of the enzyme and the active site of the enzyme. The ability of tRNA binding by the amino-terminal extension in hDRS is in accord with its control of the release of charged tRNA (25). The amino-terminal extension in hDRS also reduces the tRNA Michaelis-Menten constant as compared with hDRS⌬32 (22). Similarly, the nonspecific tRNA binding extension in yeast AspRS decreases the tRNA Michaelis-Menten constant (10). Clearly, fusion of a nonspecific tRNA binding domain to either class I or class II synthetase significantly enhances synthetase⅐tRNA interactions in terms of affinity and specificity (10,26,27).
The RNA binding capability of extensions in synthetases can conceivably facilitate not only the intramolecular transfer of RNA but also intermolecular transfers of RNA from one protein to another. The amino-terminal extension in hDRS mediates the direct transfer of charged tRNA from hDRS to the elongation factor 1␣ (25). Such intermolecular transfer is likely to play roles in the observed effects of yeast Arc1p on MetRS (40,41) or a human homologue, p43, on ArgRS (42). The detailed mechanisms of action and the tRNA binding motifs in these cases are yet to be determined.
Highly charged and potentially amphiphilic helices are prevalent in mammalian synthetases but were absent in bacterial synthetases (2). The highly basic extensions in yeast syntheta-ses endowed the yeast synthetases with the ability to bind polyanions (37). Some mammalian synthetases such as Valand LysRSs also contain such highly basic extensions (43,44).
The present results open the possibility that the potentially amphiphilic helices in other mammalian synthetases may be involved in the RNA⅐synthetase interactions as well.
Better understanding of the RNA⅐peptide interactions provides models of RNA⅐protein interactions and should be useful in the elucidation of the mechanisms in the transport of RNA from one protein to another and in the intracellular transport of RNA (45) as well as in protein biosynthesis (46).