Transition State Stabilization by the N-terminal Anticodon-binding Domain of Lysyl-tRNA Synthetase* 210

Lysyl-tRNA synthetase from Bacillus stearothermophilus (B.s. LysRS) (EC 6.1.1.6) catalyzes aminoacylation of tRNALys withl-lysine, in which l-lysine was first activated with ATP to yield an enzyme (lysyladenylate complex), and then the lysine molecule was transferred from the complex to tRNALys. B.s. LysRS is a homodimeric enzyme with a subunit that consists of two domains, an N-terminal tRNA anticodon-binding domain (TAB-ND: Ser1-Pro144) and a C-terminal Class II-specific catalytic domain (CAT-CD: Lys151-Lys493). CAT-CD alone retained catalytic activity, although at a low level; TAB-ND alone showed no activity. Size exclusion chromatography revealed that CAT-CD exists as a dimer, whereas TAB-ND was a monomer. The formation of a complex consisting of these domains was detected with the guidance of surface plasmon resonance. In accordance with this, the addition of TAB-ND to CAT-CD significantly enhanced both the l-lysine activation and the tRNA aminoacylation reactions. Kinetic analysis showed that deletion of TAB-ND resulted in a significant destabilization of the transition state of CAT-CD in the l-lysine activation reaction but had little effect on the ground state of substrate binding. A significant role of a cross-subunit interaction in the enzyme between TAB-ND and CAT-CD was proposed for the stabilization of the transition state in the l-lysine activation reaction.

where AA denotes the amino acid; E, aaRS; PPi, inorganic pyrophosphate; and E⅐AAϳAMP, an aaRS⅐aminoacyladenylate complex. Because the tRNA aminoacylation reaction is critical for the fidelity of translation of genetic information into the structure of a protein, aaRSs must have gained a high degree of substrate specificity for each heterogeneous substrate during evolution. Recent progress in x-ray crystallographic analysis has revealed that aaRSs can be classified into two groups according to their active site topology. Class I aaRSs possess a catalytic core composed of a Rossmann fold, whereas Class II aaRSs possess a catalytic core consisting of antiparallel ␤-sheets (2). AaRSs are considered to have emerged at an early stage in the development of the contemporary system of protein synthesis (3). In addition, it was reported that several isolated catalytic domains of aaRS retained full or part of the amino acid activation activity (4 -8). Together, these results have led to the proposal that the modern aaRS evolved from a primordial ancestor that had either of the two types of the classdefining domains, by getting the idiosyncratic region of each aaRS (3,9,10). In fact, the typical insertion in Class I aaRSs of the variable connective polypeptides (11) is responsible for the stabilization of the transition state of methionine activation as well as methionine transfer to the 3Ј end of tRNA in Escherichia coli MetRS (12), and also for the hydrolysis of misacylated Val-tRNA Ile in Thermus thermophilus IleRS (13). On the other hand, the crystallographic structures of the aaRS⅐tRNA complex (8,9,14,15) have revealed that an aaRS prepares a non-catalytic domain for the interaction with the anticodon stem/loop of the cognate tRNA and that each anticodon-binding domain shows considerable variation in its structure. Thus far, SerRS has been an exceptional case in which the anticodon stem/loop of the tRNA was not recognized by aaRS. In this system, the N-terminal non-catalytic domain interacts with an abnormally long variable arm of tRNA Ser (16).
Recently, it has been suggested that the interface between the anticodon-binding domain and the catalytic domain plays an important role in the tRNA-dependent conformational change in the active site, not only in GlnRS, which requires the cognate tRNA for the amino acid activation reaction (17), but also in MetRS, which does not require it (3,18). Furthermore, it has been reported that an accurate anticodon-aaRS interaction was essential for enhancement of the catalytic constant of aminoacylation (10, 19 -21). These results suggest that in some aaRSs, the signal generated by the appropriate interaction between the anticodon of tRNA and the anticodon-binding domain was transmitted to the active site through either the aaRS or the tRNA molecule (10,17,18). In contrast to the tRNAdependent domain-domain interaction, the role of the interaction between the anticodon-binding domain and the catalytic domain has never been reported in the absence of tRNA. Thus far, only for dimeric HisRS (Class IIa), both the catalytic domain and the non-catalytic domain (this domain has not yet been proven to be the anticodon-binding domain) have been purified separately (7). In this system, it was shown that in both the tRNA-independent amino acid activation and the aminoacylation reactions, the Cterminal non-catalytic domain contributed significantly to the stabilization of the transition state, but not to the ground state of substrate binding. In this case, however, because the catalytic domain of HisRS was purified as monomer, it was not clear whether the observed destabilization of the transition state upon deletion of the non-catalytic domain was because of the dissociation of the catalytic domain into a monomer or because of the loss of the interaction between the catalytic domain and the non-catalytic domain.
In the course of the studies on B.s. LysRS (L-lysine:tRNA Lys ligase (AMP forming); EC 6.1.1.6), a Class II enzyme, from Bacillus stearothermophilus (22)(23)(24)(25), we have dissected the enzyme into N-terminal and C-terminal domains. By analogy with the crystallographic structure of Escherichia coli Lys-RS(U) (26), homodimeric B.s. LysRS (molecular mass of the subunit is 57,273 Da (25)) was considered to have a simple modular organization in which the C-terminal catalytic domain (CAT-CD) was joined through a short flexible linker to the N-terminal anticodon-binding domain (TAB-ND) (Fig. 1). In this study, we have succeeded in purifying CAT-CD as a dimer. The aim of this study was to elucidate the role of the interaction between the catalytic domain and the anticodon-binding domain in the tRNA-independent L-lysine activation reaction, and to elucidate the mechanism of the transition state stabilization in the enzymatic reaction.

EXPERIMENTAL PROCEDURES
Proteins-The recombinant form of wild-type B.s LysRS was purified from E. coli BL21(DE3) cells by the method reported previously (25). The enzyme concentration was determined spectrophotometrically with ⑀ 280 of 70,600 M Ϫ1 cm Ϫ1 at pH 8.0, 25°C (25). Streptomyces subtilisin inhibitor was a gift from Dr. B. Tonomura. Superoxide dismutase from bovine erythrocytes was the product of Wako Pure Chemical. Gel filtration molecular weight markers were purchased from Sigma.
Construction of Plasmids-Based on the amino acid sequence alignment between E. coli LysRS(U) and B.s. LysRS (25), the anticodonbinding domain (Ser 1 -Pro 144 ) and the catalytic domain (Lys 151 -Lys 493 ) of B.s. LysRS were designed to be expressed in the form with an Nterminal T7 Tag sequence as TAB-ND and CAT-CD, respectively (Table  I). Two sense N-terminal primers with BamHI restriction sites, 5Ј-TA-CCACGGATCCAAAGATATCGAGCAG-3Ј (N1-primer) and 5Ј-AGGTG-GATCCGGTATGAGCCATG-3Ј (N2-primer), and two antisense C-terminal primers with BamHI restriction sites, 5ЈCGCAACAGGGGAGG-ATCCTGGTTA-3Ј (C1-primer) and 5Ј-CGTGGGATCCTTACGGCAGC-GGA-3Ј (C2-primer), were synthesized. By PCR reaction with pBLX45 containing the B.s. LysRS gene (25) and Pfu DNA polymerase, the genes corresponding to TAB-ND and CAT-CD were amplified with N2-and C2-primers and N1-and C1-primers, respectively. The amplified DNA fragments were digested with BamHI and inserted into the corresponding site of the pET11a expression vector. After transformation of E. coli BL21(DE3) cells, the DNA sequences of the TAB-ND and CAT-CD genes were determined with an ABI PRISM 377 Sequencer (Applied Biosystems).
Protein Expression and Purification-TAB-ND and CAT-CD were induced at 37°C with 1 mM isopropyl-1-thio-␤-D-galactoside in E. coli BL21(DE3) cells grown in Luria-Bertani medium containing 100 g/ml ampicillin. TAB-ND and CAT-CD were purified in the same manner as the intact enzyme until the ammonium sulfate precipitation (60% sat-uration), except for omitting the heat treatment at 55°C (25). The solution was loaded to a DEAE-Toyopearl 650M column and eluted with a linear gradient of 0 -1 M NaCl. The CAT-CD fraction was loaded to a Gigapite column and eluted with a linear gradient of 10 -400 mM phosphate buffer (pH 6.8). On the other hand, the TAB-ND fraction was loaded to a Red Toyopearl column and eluted with a linear gradient of 0 -1.0 M NaCl. Each dialyzed fraction was loaded to a Q-Sepharose HP column and eluted with a linear gradient of 0 -1.0 M NaCl. After dialysis against 100 mM Tris-HCl (pH 8.0), each fraction was applied to a TSK gel phenyl-5PW column in the high performance liquid chromatography system, and proteins were eluted with a linear gradient of ammonium sulfate from 1.2 to 0 M. The TAB-ND and CAT-CD fractions were further applied to a TSK gel G3000 SW XL column equipped with TSK guard column SW XL . All operations were done at 4°C except the high performance liquid chromatography operations that were carried out at room temperature.
Protein Concentration-Protein concentration was measured by the method of Lowry et al. (27) with crystalline bovine serum albumin as the standard. At pH 8.0 and 25°C, the estimated ⑀ 280 values of the TAB-ND monomer (17,965 Da) and CAT-CD dimer (82,571 Da) were 5,200 and 45,000 M Ϫ1 cm Ϫ1 , respectively. The concentrations of the TAB-ND monomer and CAT-CD dimer were determined spectrophotometrically using these values.
N-terminal Amino Acid Sequence Analysis-The amino acid sequences of the N termini of TAB-ND and CAT-CD were determined using a gas-phase protein sequencer (Applied Biosystems model 477A) with an on-line PTH analyzer (Applied Biosystems model 120A).

SDS-PAGE and Size Exclusion
Chromatography-200 ng of purified TAB-ND and CAT-CD were analyzed by SDS-PAGE on 10 -20% gradient gels at a constant current of 40 mA for 90 min, and visualized by Coomassie staining. The quaternary structures of TAB-ND and CAT-CD were investigated by size exclusion chromatography on a TSK gel G3000 SW XL column with a flow-rate of 0.8 ml/min. The running buffer was 100 mM Tris-HCl buffer (pH 8.0) containing 10 mM MgCl 2 .
UV Absorption, Fluorescence, and CD Spectra-All measurements were done in 100 mM Tris-HCl buffer (pH 8.0) containing 10 mM MgCl 2 . UV absorption spectra and far-UV CD spectra were measured at 25°C with a Shimadzu spectrophotometer UV-2200 and a Shimadzu-Aviv circular dichroism spectrophotometer model 202, respectively. Fluorescence spectra were measured at 30°C with a Hitachi fluorescence spectrophotometer 850.
SPR Analysis-Real-time interaction between TAB-ND and CAT-CD was detected with the guidance of surface plasmon resonance (SPR) at 25°C using a BIAcore 2000. Activation of the carboxylmethylated dextran in the sensor chip was carried out by mixing equal volumes of 100 mM N-hydroxysuccinimide in water and 400 mM N-ethyl-NЈ-(3-dimethylaminopropyl)carbodiimide hydrochloride in water, and injecting the mixture at 20 l/min for 7 min. 50 g/ml TAB-ND and 2,000 g/ml CAT-CD dissolved in 10 mM acetic acid buffer (pH 5.0) were separately injected over the activated surface of the sensor chip for 7 min at a flow rate of 20 l/min. The unreacted sites of the sensor chip were masked by the injection for 7 min of 1 M ethanolamine hydrochloride adjusted to pH 8.5 using NaOH. The molecules that non-specifically bound to the sensor chip surface were removed by washing with HBS buffer until the value of RU became nearly constant. The signals obtained were ϳ1,000 RU, which were in the permitted range for determining kinetic constants. In the present analysis, no regeneration procedure was applied because of a significant loss in the binding capacity of each domain of B.s. LysRS. Binding experiments were carried out by injecting CAT-CD in HBS buffer at 40 l/min onto the sensor surface on which TAB-ND was immobilized, and vice versa. k a , k d , and K d were estimated by assuming a simple bimolecular binding equilibrium (see Supplemental Materials). In the apparent association-reaction, the formation of the complex was represented as follows. On the other hand, in the apparent dissociation reaction, the disappearance of the complex was represented as follows, where R n was R at time t n , and R 0 is R at an arbitrary starting time, t 0 . The k d value was obtained as the linear slope of the ln(R 0 /R n ) versus (t n -t 0 ) plot. ATP-PPi Exchange Reaction-The L-lysine activation reaction was measured by using the L-lysine-dependent ATP-PPi exchange reaction at pH 8.0, 37°C, as reported previously (22). The standard reaction mixture contained in 0.5 ml: 100 mM Tris-HCl buffer (pH 8.0), 10 mM MgCl 2 , ATP, L-lysine, and PPi (9.1 mCi/mmol). Kinetic parameters were calculated by the nonlinear least-squares method with KaleidaGraph (Synergy software). K m,Lys was estimated over a range of L-lysine concentrations from 10 to 200 M, whereas K m,ATP was estimated over a range of ATP concentrations from 20 to 500 M. K m,PPi was estimated over a range of PPi concentrations from 30 to 500 M. In all cases, the initial concentration of the other two substrates was 1 mM. 1 mM might not be the saturating concentration for ATP (see Table II), but this concentration was chosen after considering the ratio of ATP to Mg 2ϩ . Enzyme concentrations were 5 nM B.s. LysRS, 500 nM CAT-CD, and 75 nM TAB-ND plus 5 nM CAT-CD.
Aminoacylation Reaction-The tRNA aminoacylation reaction was measured at pH 8.0, 37°C, as reported previously (22). The standard reaction mixture contained in 0. Area of Domain-Domain Interactions in E. coli LysRS(U)-B.s LysRS consists of 493 amino acid residues and the amino acid sequence has relatively high homology (53%) to E. coli LysRS(U) (504 residues) (25). Therefore, to obtain information on the domain-domain interactions of B.s. LysRS, each contact area between the domains of E. coli LysRS(U) was estimated based on the 2-5 Å resolution structure of the complex with the substrate L-lysine (26) (Fig. 1). A dimer of E. coli LysRS(U) was chosen, and hydrogens were added to the groups ionized at pH 8. After removing the L-lysine substrate and water molecules, the accessible surface area of the free dimer enzyme was estimated with Insight II/Homology (Biosym Technologies, San Diego, CA) by rolling a sphere of 1-4 Å radius corresponding to a water molecule around the enzyme (27). In the same way, the accessible surface areas of N A , N B , C A , C B , where the N-terminal domain (Phe 14 -Pro 153 ) and C-terminal domain (Asp 161 -Pro 502 ) in subunit A of the E. coli LysRS(U) dimer were denoted as N A and C A , respectively, and in subunit B as N B and C B .

Purification and Structures of TAB-ND and CAT-CD-Both
TAB-ND and CAT-CD were more highly expressed in E. coli BL21(DE3) cells compared with the intact enzyme, and were purified to homogeneity as judged by SDS-PAGE (Fig. 2). Their migrations on SDS-PAGE were consistent with the molecular mass of TAB-ND, 17,965 Da, and that of CAT-CD, 41,286 Da (Table I). N-terminal amino acid sequences of TAB-ND and CAT-CD were determined as ASMTGGQQMGRGSGMSHEEL and ASMTGGQQMGRGSKDIEQRY, respectively, in complete accordance with those expected from their DNA sequences (Table I). The underlined regions show the sequence from Ser 1 to Leu 5 of TAB-ND and from Lys 151 to Tyr 157 of CAT-CD. Size exclusion chromatography (Supplemental Materials) indicated that TAB-ND existed as a monomer but CAT-CD was a dimer under the conditions used, because the molecular masses of TAB-ND and CAT-CD were estimated to be 13,500 and 88,300 Da, respectively (Supplemental Materials). The emission spectra of TAB-ND at an excitation wavelength ( ex ) of 280 nm has a max value of 304 nm, corresponding to 306 nm of N-acetyl-L-tyrosine ethyl ester, whereas no appreciable fluorescence was observed at ex ϭ 295 nm. These data were in accordance with the fact that TAB-ND contains Tyr residues but no Trp residues, as deduced from the primary structure of B.s. LysRS (Table I). On the other hand, the emission spectra of CAT-CD at ex ϭ 295 nm indicates the presence of Trp residues and that the max value of 336 nm was very close to the corresponding 335 nm of the intact B.s. LysRS (26). TAB-ND and CAT-CD (515 residues) was larger than that of B.s. LysRS (493 residues) because of the inclusion of the Nterminal T7 tag sequence (Table I), it may be reasonable that the ellipticity of the synthesized spectra of the complex of TAB-ND and CAT-CD was somewhat negatively larger than those of B.s. LysRS. On the other hand, despite having the same number of amino acid residues, the ellipticity of the observed spectra of TAB-ND plus CAT-CD was negatively larger than the synthesized one, suggesting that the interaction between TAB-ND and CAT-CD affects the secondary structures of either or both domains.  Fig. 3B. Estimated k a and k d are 48,800 Ϯ 2,300 M Ϫ1 s Ϫ1 and 0.0008 Ϯ 0.0009 s Ϫ1 . When k d was estimated by linear fitting in the plot of ln(R 0 /R n ) versus t n Ϫ t 0 according to Equation 2, curved lines which indicate a slowdown of the dissociation rate were observed (Supplemental Materials). Therefore, k d values were estimated from the slope of the tangent to the initial straight parts (Ϲ20 s). The averaged k d value is 0.00120 Ϯ 0.00009 s Ϫ1 , which agrees with the value estimated above in the association phase. With k a ϭ 48,800 M Ϫ1 s Ϫ1 and k d ϭ 0.0008 s Ϫ1 , the K d (ϭk d /k a ) value was calculated to be 16.4 Ϯ 1.2 nM (Table II). When the interaction between free CAT-CD and immobilized TAB-ND was also investigated in a similar way, binding of CAT-CD to immobilized TAB-ND was observed, but CAT-CD did not dissociate from the immobilized TAB-ND (data not shown). This may be because of the bivalent property of dimeric CAT-CD (see Supplemental Materials) in binding to TAB-ND.

UV Absorption Spectra and Fluorescence Emission
Enzymatic Activities of TAB-ND and CAT-CD-Neither the L-lysine-dependent ATP-PPi exchange activity nor the tRNA aminoacylation activity could be measured with TAB-ND, whereas CAT-CD retained both activities at a very low level. The kinetic constants of CAT-CD in the ATP-PPi exchange reaction were estimated with 500 nM CAT-CD (Table II). The k cat value of CAT-CD was 470-fold smaller than that of the intact B.s. LysRS, whereas K m values of CAT-CD (K m,Lys , K m,ATP , and K m,PPi ) were comparable with those of the intact enzyme. In the presence of 20 A 260 units of E. coli tRNA, the kinetic constants of CAT-CD in the aminoacylation reaction were estimated with 300 nM CAT-CD (Table II). The k cat value of CAT-CD was 29,000-fold smaller than that of the intact enzyme, whereas K m values of CAT-CD (K m,Lys and K m,ATP ) were comparable with those of the intact enzyme. Because it is known that T. thermophilus LysRS interacts with the anticodon region of the tRNA transcripts through the N-terminal anticodon-binding domain (15), we consider that the affinity of CAT-CD for tRNA Lys must be weakened significantly. In the case of AspRS, which belongs to the same subclass as LysRS (Class IIb), the deletion of the anticodon-binding domain re-sulted in a 100-fold increase in the K m for tRNA Asp but k cat remained unchanged (29). Although the estimated k cat and K m of CAT-CD in the aminoacylation reaction were apparent values because of the possible nonsaturating conditions of tRNA, it should be noted that CAT-CD alone can bind to E. coli tRNA and catalyze aminoacylation without the N-terminal anticodon-binding domain.
Enhancement of the Enzymatic Activity of CAT-CD Induced by TAB-ND-The addition of TAB-ND to CAT-CD enhanced both the L-lysine-dependent ATP-PPi exchange activity and the tRNA aminoacylation activity as compared with CAT-CD alone (Table II, Fig. 4). The kinetic parameters of the TAB-ND⅐CAT-CD complex in the ATP-PPi exchange reaction were estimated with 75 nM TAB-ND plus 5 nM CAT-CD (Table II). Under these conditions, most of the CAT-CD exists in a complex with TAB-ND (Fig. 3). The estimated k cat value of the complex was 170-fold larger than that of CAT-CD alone, whereas the K m values of the complex were comparable with those of CAT-CD but even closer to those of the intact B.s. LysRS. Similarly, the kinetic parameters of the complex in the aminoacylation reaction were estimated with 150 nM TAB-ND plus 10 nM CAT-CD. The estimated k cat of the complex was 390-fold larger than that of CAT-CD, whereas the K m values of  the complex were comparable with those of CAT-CD but closer to those of the intact B.s. LysRS. It is noted that the k cat of the TAB-ND⅐CAT-CD complex corresponds to 37% of the k cat of B.s. LysRS in the ATP-PPi exchange reaction, but only to 1.4% of that in the aminoacylation reaction. This significant difference in the restoration of k cat may indicate that the K m for tRNA Lys of the TAB-ND⅐CAT-CD complex was considerably larger than that of the intact B.s. LysRS. If so, the estimated k cat of the complex in the aminoacylation reaction would have been underestimated because of the low concentration of tRNA used. Based on the relationship between the ATP-PPi exchange activity and the initial concentration of TAB-ND added (Fig. 4), we can estimate the K d value for the complex of TAB-ND and CAT-CD assuming a simple bimolecular binding equilibrium in that two TAB-ND monomers bind to a CAT-CD dimer independently, where TAB-ND, CAT-CD m , TAB-ND⅐CAT-CD m are the TAB-ND monomer, the CAT-CD monomer, and the complex of TAB-ND monomer and CAT-CD monomer, respectively. Assuming that the binding of TAB-ND to one CAT-CD m in a CAT-CD dimer does not affect the activity of the other CAT-CD m , the enzymatic reaction velocity observed was represented as follows under saturating substrate conditions (see Supplemental Materials).  (Table II). DISCUSSION The general structural organization of E. coli LysRS(U) (26) was illustrated in Fig. 1. The subunit consists of three  domains: (a) the tRNA anticodon-binding domain in the N terminus; (b) the catalytic site domain in the C terminus; and (c) a short region connecting them. The dimerization of E. coli LysRS(U) was sustained by three domain-domain interactions: (i) between the catalytic domains; (ii) between the anticodon-binding domain and the catalytic domain of the other subunit; (iii) between the anticodon-binding domain and the catalytic domain of the same subunit. These contact areas were calculated to be: 5,000 Å 2 (i), 2,100 Å 2 (ii), and 500 Å 2 (iii), respectively. Size exclusion chromatography revealed that CAT-CD exists as a dimer, whereas TAB-ND was a monomer (Supplemental Materials). These results agree with the x-ray crystallographic structure of E. coli LysRS(U) (26), in which there was no interaction between the two N-terminal domains in the dimer, whereas the two C-terminal domains have significant interactions with each other (contact area 5,000 Å 2 ). Prokaryotic MetRS (Class I) (4), AlaRS (Class II) (5), HisRS (Class II) (7), and SerRS (Class II) (6), which are all dimeric enzymes, were converted into monomeric catalytic domains upon deletion of the non-catalytic domain, in contrast to the present case, in which CAT-CD retains a dimeric structure. This indicates that the interaction between the catalytic domains of B.s. LysRS was considerably stronger than that of the other aaRSs, which has enabled us to investigate the role of the interaction between the anticodon-binding domain and the catalytic domain in the cata-lytic reaction of B.s. LysRS.
The result of SPR measurements (Fig. 3) and the increase in the CAT-CD catalytic activity by the addition of TAB-ND (Fig.  4) indicates that the interaction between the anticodon-binding domain and the catalytic domain was achieved in the TAB-ND⅐CAT-CD complex as it was in the intact B.s. LysRS. The K d value of the complex, 30.7 Ϯ 4.0 nM (37°C), which was estimated by the increase in the ATP-PPi exchange reaction activity of CAT-CD in the complex (Fig. 4), was comparable with 16.4 Ϯ 1.2 nM (25°C), which was estimated from the SPR measurements (Fig. 3, Table II). The values of the free energy of binding between CAT-CD and TAB-ND were calculated with these K d values to be Ϫ10.7 kcal mol Ϫ1 (37°C) and Ϫ10.6 kcal mol Ϫ1 (25°C).
If we assume that the differences in the kinetic parameters between CAT-CD and the TAB-ND⅐CAT-CD complex reflect the interactions between the anticodon-binding domain and the catalytic domain, and that the differences between the TAB-ND⅐CAT-CD complex and the intact B.s. LysRS reflect the covalent link between the two domains in a subunit, then their energetic contributions to the L-lysine-dependent ATP-PPi exchange reaction can be estimated according to Scheme 2, where E denotes B.s. LysRS, CAT-CD, or the TAB-ND⅐CAT-CD complex, and k cat and K m are the kinetic parameters in the ATP-PPi exchange reaction.
Estimated activation free energies (⌬G ‡ ) and binding free energies (⌬G 0 ) are listed in Table III Table  III assuming that the binding processes for L-lysine, ATP, and PPi were in rapid equilibrium and that the forward reaction toward the transition state, E⅐Lys⅐ATP 3 [E⅐Lys⅐ATP] ‡ , was the rate-determining step.   (Table III). These results evidently indicate that the direct link between these domains does not play a significant role in the ground state of B.s. LysRS, but contributes to some extent to the stabilization of the transition state in the ATP-PPi exchange reaction.
Thus, the present study was the first case for an aaRS in which the binding of the non-catalytic domain restores the reduced activity of the catalytic domain both in the ATP-PPi exchange reaction and the aminoacylation reaction. It was suggested that the conformational change induced by the deletion of the anticodon-binding domain was related to the reduction in the k cat of the dimeric catalytic domain of B.s. LysRS in the ATP-PPi exchange reaction. Thus far, only in the case of dimeric HisRS (Class IIa), both the non-catalytic domain and the catalytic domain have been separately purified (7). However, in that case, the catalytic domain of HisRS was purified as a monomer. Although the deletion of the noncatalytic domain of HisRS resulted in a preferential decrease in k cat both in the ATP-PPi exchange and the aminoacylation reactions, the addition of the non-catalytic domain to the catalytic domain did not restore the activity of the catalytic domain in both reactions.
Following the above discussions, we would like to tentatively propose that the cross-subunit interaction of 2,100 Å 2 between the anticodon-binding domain and the catalytic domain contributes significantly to the stabilization of the transition state.
The following lines of evidence support this proposition. First, the amino acid residues at the cross-subunit interface of E. coli LysRS(U) are well conserved in B.s. LysRS (Fig. 6). Of 17 amino acid residues, 15 residues were conserved in B.s. LysRS, and only Asn 37 and Arg 40 , which interact with each other, were converted to Lys and Glu residues, respectively. This strict conservation may support the significance of the cross-subunit interface. Second, six residues (Glu 252 , Tyr 283 , Tyr 466 , Gly 467 , Pro 469 , and Pro 470 ), which belong to either motif 2 or 3, were located in the cross-subunit interface. Because motifs 2 and 3 constitute most of the active site of Class II aaRS, it seems likely that the signal from the cross-subunit interface was transmitted through these residues to other residues that participate in the stabilization of the transition state. Although no residues that specifically stabilize the transition state have as yet been identified in LysRS, it has been revealed that a motif 2 loop in E. coli LysRS(U) ( 264 EGISVRHN 271 ), which was close to the active site and disordered in the complex with L-lysine, became ordered in the presence of adenine molecules (30). Because the loop was strictly conserved in B.s. LysRS except that Val 268 was substituted by a Thr residue (25), such a conformation of the loop may be held in the complex of B.s. LysRS with an adenine molecule. The unfavorable changes in ⌬G ‡ and ⌬G ATP 0 for CAT-CD in the energy diagrams (Fig. 5) may be because of a failure in the formation of the ordered structure of the motif 2 loop because of the loss of the crosssubunit interaction between the catalytic domain and the anticodon-binding domain. Amino acid residues that contribute to the stabilization of the transition state but not the ground state have been reported for MetRS (31,32) and TyrRS (33,34). The energy diagrams (Fig. 5) also indicate that CAT-CD suffers from a disturbance in ⌬G 0 Lys as well as in ⌬G ‡ and ⌬G 0 ATP . The loss of the cross-subunit interaction may also have disturbed the conformational changes that may be induced by the addition of L-lysine to B.s. LysRS as was observed with E. coli LysRS(S) (35). Overall, it was considered that the cross-subunit interaction between the catalytic domain and the anticodonbinding domain in B.s. LysRS contributes more significantly to the activation free energy than to the binding free energy, suggesting a fine mechanism for the transition state stabilization. In the process of biological evolution where the anticodonbinding domain was appended to the catalytic domain and the cross-subunit interaction between them has been finely tuned, the original molecular mechanism of the transition state stabilization in the primordial LysRS may have been lost and changed to require a cross-subunit interaction in the modern B.s. LysRS.
It has been reported that the binding of eukaryotic aaRSs to tRNA was reinforced through an additional RNA-binding helper domain which functions as either a trans-acting (36,37) or cis-acting cofactor (38 -40), as compared with their prokaryotic counterparts. Recently, it was revealed that an N-terminal polypeptide chain extension of about 60 amino acid residues in the hamster LysRS (which was also conserved in human LysRS) was a new type of RNA-binding domain that acts as a cis-acting cofactor for tRNA binding (40). In the report (40), the hamster LysRS aminoacylated a microhelix that mimics the amino acid acceptor minihelix at a very low level as compared with an intact tRNA (40-fold increase in K m for tRNA and 65,000-fold decrease in k cat ). These data indicate that the interaction between the tRNA anticodon and the anticodon-binding domain of the enzyme was very important in the transition state stabilization of the aminoacylation reaction. Is the communication between the region participating in the recognition of anticodon and the catalytic site achieved by a direct interaction between the anticodon-binding domain and the catalytic domain or an indirect interaction that was mediated through the "elongated" N-terminal polypeptide chain extension? The finding that the effect of the deletion of the N-terminal polypeptide chain extension on the aminoacylation of intact tRNA was limited to a change in the K m for tRNA and no observed change in k cat (40) may support the importance of a direct interaction in the transition state stabilization. In mammalian LysRS, it remains to be uncovered whether the direct interaction between the anticodon-binding domain and the catalytic domain was a cross-subunit interaction or not.