ABSTRACT
The alaS gene encoding the alanyl-tRNA synthetase (AlaRS) from Thermus thermophilus HB8 was cloned and sequenced. The gene comprises 2646 bp, corresponding to 882 amino acids, 45% of which are identical to the enzyme from Escherichia coli. The T.thermophilus AlaRS was overproduced in E.coli, purified and characterized. It has high thermal stability up to ~65oC, with a temperature optimum of aminoacylation activity at ~60oC, and will be valuable for crystallization. The purified enzyme appears as a dimer with a specific activity of 220 U/mg and kcat/KM values of 118 000/s/M for alanine and 114 000/s/M for ATP. By genetic engineering a 53 kDa fragment of AlaRS comprising the N-terminal 470 amino acids (AlaN470) was also overproduced and purified. It is as stable as entire AlaRS and sufficient for specific aminoacylation of intact tRNAAla, as well as acceptor stem microhelices with a G3-U70, but not U3-A70, I3-U70 or C3-U70, base pair. The reduced binding strength of such microhelices to AlaN470 enabled, due to the resulting fast exchange of the microhelices between free and complexed states, preliminary NMR analyses of the binding mode and intermolecular recognition.
Alanyl-tRNA synthetase (AlaRS), which belongs to class II of the aminoacyl-tRNA synthetases (aaRSs; 1 ), is one of the few aaRSs whose three-dimensional structure has not been solved to date. Since the known aaRSs are composed of several different domain modules in various combinations (2 ), a knowledge of each individual aaRS contributes to our understanding of the evolution of the genetic code. Also, the modules used for tRNA binding and recognition and their modes of interaction are surprisingly varied. In this respect AlaRS is particularly interesting, since the organization of this enzyme in domains and their functions are not yet fully understood.
Several experimental approaches have been used to identify common structural features of AlaRSs from different organisms and to determine the functions of the various enzyme regions (see 3 for a review). The N-terminal 368 amino acids of Escherichia coli AlaRS comprise the active site domain, where the three sequence motifs characteristic of class II aaRSs are found. This module is sufficient for alanylation (4 ). It is believed to interact with the G1-C72 base pair of tRNAAla by approaching the acceptor stem from the major groove side (5 ). A retroviral-like zinc binding motif situated between motifs 2 and 3 contributes to tRNA recognition by this module (6 ,7 ). An additional module (amino acids 385-461), however, is needed for specific interaction of this enzyme with tRNAAla by minor groove interactions at its conserved acceptor helix base pair G3-U70 (8 ). This base pair is the major determinant for the identity of tRNAAla. Amino acids 699-808 resemble a module which is believed to be responsible for oligomerization of E.coli AlaRS. This domain is different in the monomeric eukaryotic enzymes. At least two other modules with unknown function are found in the known AlaRSs.
Truncated tRNA model molecules, consisting of the acceptor stem of tRNAAla from E.coli, are recognized by AlaRS as substrates and efficiently aminoacylated with alanine, provided the proper base pair is present at the critical 3-70 position (9 ). It has become a crucial question whether a helical distortion produced by the G3-U70 base pair (10 ) or a functional group on it (11 ) is required for recognition of the RNA by AlaRS. In addition to elucidation of the three-dimensional structure of the AlaRS-tRNAAla complex by X-ray crystallographic analysis, NMR studies in particular are promising to clarify the mode of interaction with this acceptor helix base pair.
In this work we report cloning, sequence determination and overexpression of the alaS gene from Thermus thermophilus and purification of AlaRS and its biologically active monomeric fragment. The sequence information and the availability of large amounts of this thermostable enzyme are valuable for further structural investigations, since crystallization of AlaRS from mesophilic organisms has not been successful to date, for unknown reasons. In preliminary NMR experiments the N-terminal 53 kDa fragment of AlaRS (AlaN470) has proved particularly useful for studying its interactions with the acceptor stem of tRNAAla microhelices.
Thermus thermophilus HB8 (ATCC 27634) was grown as previously reported (12 ). Escherichia coli strains DH10BTM (Gibco BRL, Bethesda, MD) and BL21(DE3)pLys (Novagen, Germany; 13 ) were grown in Luria Bertani (LB) nutrient medium (14 ) or Terrific broth (TB; 15 ). For cloning and sequencing plasmid pUC18 (16 ) and for expression plasmid pET-28c (Novagen, Germany; 13 ) were used.
The first purification steps of AlaRS from T.thermophilus involving anion exchange chromatography on Q-Sepharose-FF and gel permeation chromatography on Sephacryl S200 HR were performed essentially as previously described for purification of overproduced elongation factor Tu from T.thermophilus (17 ). Fractions with AlaRS activity were pooled and the protein precipitated by addition of solid ammonium sulfate to 70% saturation. The precipitate was dissolved in a minimal volume of buffer P10 (10 mM potassium phosphate, pH 7.0). The solution was dialyzed against 2* 2 l buffer P10, centrifuged and the particle-free supernatant loaded on a hydroxyapatite column (20 * 1 cm, particle size 15 [mu]m, flow rate 1 ml/min; Merck, Germany) equilibrated with buffer P10. After the flow-through was eluted, the column was developed with a linear gradient of 10-200 mM potassium phosphate in 120 ml buffer P10. The fractions containing AlaRS were pooled, precipitated as above, dissolved in a minimal volume of buffer A [50 mM Tris-HCl, pH 7.5, 10 mM MgCl2, 1 mM 2-mercaptoethanol, 20 [mu]M PMSF and 5% (v/v) glycerol] and dialyzed against 2* 1 l buffer A. The supernatant after centrifugation was loaded on a heparin-Sepharose CL-6B column (10 * 1 cm, flow rate 1 ml/min; Pharmacia Biotech, Freiburg, Germany). The column was developed with a linear gradient of 0-125 mM KCl in 100 ml buffer A. The fractions containing AlaRS were pooled and the protein precipitated as above. For use, the AlaRS precipitate was dissolved in an appropriate buffer and dialyzed. As assessed by SDS-PAGE, AlaRS was ~95% pure.
The overproducing bacteria, E.coli BL21(DE3)pLys with one of the T7 promoter plasmids, were grown at 30oC in LB or TB in the presence of 40 [mu]g/ml kanamycin and 25 [mu]g/ml chloramphenicol. Fermentation of bacteria and the first purifications steps of AlaRS or AlaN470, involving heat treatment at 65oC and anion exchange chromatography on Q-Sepharose-FF, were performed essentially as previously described for purification of overproduced phenylalanyl-tRNA synthetase from T.thermophilus (18 ). AlaRS and AlaN470 was selectively precipitated with 45 and 50% saturated (NH4)2SO4, respectively, and loaded on a hydroxyapatatite column as described. Collected fractions with AlaRS activity were dissolved in buffer A. Imidazole was added to a final concentration of 5 mM. A Ni-HiTrap® Chelating affinity column (volume 5 ml, flow rate 1 ml/min; Pharmacia Biotech, Freiburg, Germany) was equilibrated with buffer AI (buffer A + 5 mM imidazole) and saturated by addition of 2.5 ml 100 mM NiSO4. The protein solution was loaded on this column. After elution of the flow-through, the column was developed with a linear gradient of 5-300 mM imidazole in 50 ml buffer AI. The fractions containing apparently homogeneous AlaRS or AlaN470 were pooled, dialyzed against 2* 1 l buffer A and finally against 100 ml buffer A + 50% glycerol and stored at -20oC.
For cyanogen bromide cleavage 80 [mu]g protein were dialyzed against water and dried in a vacuum centrifuge. An estimated 100-fold molar excess of cyanogen bromide dissolved in 70% (v/v) trifluoroacetic acid over AlaRS methionines was added and incubated at room temperature for 12 h. After adding 0.5 ml water the peptide solution was lyophilized; this was repeated twice. The peptides were separated by SDS-PAGE and blotted onto polyvinylidene difluoride (PVDF) membrane (Schleicher & Schüll GmbH, Dassel, Germany). The N-terminal amino acid sequences were determined with an ABI 473A protein sequenator as previously described (19 ).
Standard recombinant DNA methods were performed according to the protocols of Ausubel et al. (20 ) and Sambrook et al. (15 ). PCR was carried out in 2 mM MgCl2, 200 [mu]M dNTPs,100 [mu]M primers, 2 [mu]g T.thermophilus genomic DNA partially digested with SauIIIA as template, 2.5 U Taq DNA polymerase (Fermentas Ltd, Vilnius, Lithuania) in 10 mM Tris-HCl, pH 8.8, 50 mM KCl under the following conditions (in a volume of 100 [mu]l): 5 min at 95oC; 30 cycles of 1 min at 95oC, 50 s at 40oC and 1 min at 72oC; 7 min at 72oC.
The determined amino acid sequences (data not shown) were used to design degenerate oligodeoxyribonucleotides taking into account the peculiar codon usage of T.thermophilus. Oligodeoxyribonucleotides Ala 1 [ATGAAGCG(CG)AC(CG)GC(CG)GAGATCCG(CG)GAGAAGTT] and Ala 2 [CCCTTCTTGAAGA- TCTC(CG)CGCTCGAA(CG)GCCAT] were successfully used in PCR to amplify a product of 1.3 kb. This fragment was used as a probe to identify a 9 kbp SphI fragment in a partial genomic DNA library constructed in pUC18. The corresponding plasmids were designated pUCAlaRS Sph9-1 and pUCAlaRS Sph 9-3 (insert in opposite orientation). The alaS gene was further subcloned, resulting in plasmids pUCAlaRS 4H1 (pUC18 with a 4 kbp SphI-HindIII fragment comprising the 3'-portion of the alaS gene) and pUCAlaRS 3.4H1 (pUC18 with a 3.4 kbp HindIII fragment comprising the 5'-portion of the alaS gene), which were used to determine the nucleotide sequence of the alaS gene region.
For overexpression of AlaRS and AlaN470 the vector pET-28c, which allows purification of N-terminally histidine-tagged fusion proteins (20 additional amino acids at the N-terminus MGSSHHHHHHSSGLVPRGSH), was used. A NdeI restriction site overlapping the start codon of alaS was introduced by PCR into a 415 bp DNA fragment comprising nucleotides 1-404 of the alaS gene using primers AlaRSNde (TAATCAAACATATGCGCACGGCGGAGATCC) and Ala P2 (GCCTCGTCGTCGTCCTCAAAGACCGTGACC). The amplification product was hydrolyzed with NdeI and HindIII (internal restriction site) and ligated with expression vector pET-28c. A 4 kb HindIII fragment comprising the 3'-portion of the alaS gene was linked using the internal HindIII site. Escherichia coli BL21(DE3)pLys was transformed with the resulting expression plasmid (pETAlaRS1). To obtain an expression plasmid for AlaN470, a stop codon (at AlaRS amino acid position 471) and an adjacent EcoRI restriction site were simultaneously introduced into the alaS gene by site-directed mutagenesis (21 ) using oligodeoxyribonucleotide Ala Eco/am (GCCTCGAGGGCGTTCTAGCCCAAGAATTCCGTGGCCCCCC). The EcoRI site was used to identify the mutated clones. The changed nucleotide sequence in the resulting plasmid (pETAlaRS471am), as confirmed by sequencing, reads 5'-GCCACGGAATTCTTGGGCTAGAAC-3' (EcoRI site and stop codon in bold). This plasmid expresses the 470 N-terminal amino acids of T.thermophilus AlaRS (AlaN470).
The specific activities of AlaRS and AlaN470 were determined by a filter binding assay in 100 mM sodium cacodylate, pH 7.6, 10 mM KCl, 8 mM MgSO4, 2 mM rATP, 5 mM 2-mercaptoethanol, 20 [mu]M [14C]alanine (120 Ci/mol; Amersham Buchler, Braunschweig, Germany), 70 [mu]M T.thermophilus tRNAbulk (isolated according to the protocol of Zubay; 22 ) at 60oC in a volume of 60 [mu]l. The protein concentration was determined by the BioRad (Hercules, CA) assay. One unit (U) of AlaRS activity was defined as the amount of enzyme needed to attach 1 nmol [14C]alanine to the tRNA in 1 min at 60oC. For determination of the temperature dependence of the specific enzyme activity the sample was equilibrated at the respective temperature for 5 min before the reaction was started by addition of the enzyme. Prior to use the tRNA was preincubated at 60oC for 5 min and then added to the assay. To determine the aminoacylation of microhelices derived from the tRNAAla acceptor stem, the chemically synthesized oligoribonucleotides (see below) were annealed by heating to 80oC for 3 min and slow cooling to room temperature. The resulting duplices at a concentration of 30 [mu]M were assayed at 40oC with 30 nM enzyme in a total volume of 70 [mu]l.
The molecular masses of proteins were determined by size exclusion chromatography on Superdextm 200 HR 10/30 (Pharmacia Biotech, Freiburg, Germany) by comparison with standard proteins (BioRad, Hercules, CA): thyroglobulin, 670 kDa; bovine [gamma]-globulin, 158 kDa; chicken ovalbumin, 44 kDa; equine myoglobulin, 17 kDa; cyanocobalamin, 1.35 kDa. Aliquots of 100 [mu]g purified protein were analyzed by developing the column with 50 mM Tris-HCl, pH 7.5, 10 mM MgCl2, 150 mM KCl.
The reaction mixture (300 [mu]l) contained 100 mM Tris-HCl, pH 7.5, 10 mM MgCl2, 2 mM KF, 2 mM [32P]pyrophosphate (0.5 mCi/ mmol; DuPont NEN, Bad Homburg, Germany). ATP and alanine concentrations were in the range 40-600 and 100-800 [mu]M respectively. After preincubation for 3 min at 50oC the reaction was started by addition of 10 nM enzyme. Aliquots of 40 [mu]l were withdrawn at 2, 4, 6, 8 and 12 min and the reaction was stopped by addition of 70 [mu]l 10% perchloric acid containing 300 mM sodium pyrophosphate. Aliquots of 100 [mu]l of this solution were filtered through activated carbon paper (Schleicher & Schüll, Dassel, Germany). After washing with 5 ml water and 5 ml ethanol the filter was dried and the radioactivity of the absorbed ATP was counted. Kinetic data were analyzed using the Eadie- Hofstee and Lineweaver-Burk plots (23 ).
The protein solutions (1.5 and 3.7 [mu]M for AlaRS and AlaN470, respectively) in 50 mM HEPES, pH 7.5, 10 mM MgCl2, 50 mM KCl were heated in a spectrophotometer (Hewlett Packard model 8452A) in 1oC steps with a linear temperature gradient from 20 to 90oC. The UV spectra in the range 240-350 nm were recorded after equilibration of the probe at each temperature for 3 min. Protein unfolding was determined by following the change in the absorbance difference at two wavelengths ([Delta]A274-286).
The oligoribonucleotides (7mer and 11mer strands) were chemically synthesized on a Gene Assembler Plus (Pharmacia Biotech, Freiburg, Germany) using the H-phosphonate method as described previously (24 ). RNA synthons were produced as specified (25 ). The oligoribonucleotides were purified by HPLC on a Nucleogen- DEAE 500-7 column (Macherey-Nagel, Germany).
For the NMR measurements the oligoribonucleotides (~0.25 mM strand concentration for one-dimensional studies, ~1.3 mM for the NOESY experiments) were dissolved in 0.5 ml H2O/D2O (9:1 v/v) with 100 mM NaCl, 4 mM MgCl2, 10 mM sodium phosphate, pH 7. The strands were annealed by heating to 80oC for 3 min and slow cooling to room temperature. Approximately 0.2 mM 2,2-dimethyl-2-silapentane-5-sulfonate (DSS) was added as an internal chemical shift reference.
All NMR spectra were acquired on a DRX 500 spectrometer (Bruker), equipped with an Aspect Station computer, at a proton resonance of 500 MHz. For the one-dimensional measurements the 1-3-3-1 pulse sequence (26 ) was used to suppress the strong water signal. The NOESY spectra were recorded using a 1-1 `jump-return' sequence (27 ) for the last (read) pulse with the excitation maximum in the imino region (~12.8 p.p.m.). Eight thousand data points in t2 and 512 t1 increments were used in the phase-sensitive mode, giving 4000 * 1000 two-dimensional NMR spectra after Fourier transformation. NOESY spectra were apodized by [pi]/2-shifted squared sine functions in both dimensions. Data processing and spectra analysis were performed on Hewlett-Packard 9755 or INDIGO2 (Silicon Graphics) workstations using the NDEE program package developed by F.Herrmann (Symbiose Software, Bayreuth, Germany). The spectral width was 10 kHz, the mixing time 300 ms.
To obtain sufficient amounts of AlaRS for crystallization and NMR studies via overexpression, as well as to enable mutagenesis, we first cloned the gene encoding AlaRS (alaS) from T.thermophilus. About 300 [mu]g AlaRS were purified from T.thermophilus cells using standard chromatographic procedures. The enzyme was identified by its specific alanylation activity on tRNAbulk from T.thermophilus. To gain sequence information, the N-terminal amino acid sequences of the protein and of cyanogen bromide fragments were determined (data not shown). This was used to design gene probes, which were successfully used to clone the alaS gene on a 9 kb SphI fragment. After subcloning the relevant region, the nucleotide sequence of 4019 bp covering the alaS gene was determined on both strands and deposited in the EMBL database (accession no. Y08363). The alaS gene comprises 2646 bp, corresponding to 882 amino acids of the encoded AlaRS.
The protein is similar to the other known AlaRSs, e.g. 45% of the amino acids are identical in the corresponding enzyme of E.coli (Fig. 1 ). Sequence similarities are particularly pronounced not only around the three signature motifs of class II aaRSs but also around amino acids 400-411 (with respect to T.thermophilus AlaRS), as well as in a domain (T.thermophilus amino acids 500-716) of unknown function. The zinc binding motif reported for E.coli AlaRS (E.coli amino acids 179-192) was not found at a homologous position of T.thermophilus AlaRS (6 ). Cys666 (numbered including the N-terminal methionine) in the C-terminal region, which is important for aminoacylation activity of E.coli AlaRS (28 ), is also present in the T.thermophilus enzyme.
For overproduction of AlaRS from T.thermophilus and AlaN470 in E.coli we used pET-28c as the expression vector. This appended a histidine tag and a thrombin cleavage site at the N-terminus of the protein, thus facilitating purification. A NdeI restriction site overlapping the start codon of alaS was introduced by means of PCR, the alaS gene was put together from two neighboring restriction fragments and inserted into the vector (resulting in pETAlaRS1). For production of AlaN470, comprising the first 470 amino acids of AlaRS, codon 471 of the alaS gene was replaced by an amber stop codon by site-directed mutagenesis (pETAlaRS471am). Both the entire AlaRS and AlaN470 were overproduced in E.coli and purified by heat treatment of the crude cellular extract and chromatographies on Q-Sepharose, hydroxyapatite and immobilized nickel (Table 1 ). The proteins, apparently homogeneous on Coomassie Blue stained SDS gels (data not shown), were used for subsequent investigations.
Though AlaRS is one of the most interesting and biochemically extensively characterized aaRSs, structural data have not been available to date. The reason is probably the low stability of this enzyme from mesophilic organisms, which renders difficult both crystallization and production of the homogeneous, highly concentrated solutions that are needed for NMR studies. Numerous examples of stable, well crystallizing proteins from T.thermophilus prompted us to clone its gene and overproduce AlaRS from this extreme thermophilic organism. Indeed, the protein proved rather stable and could be purified without major complications. Fortunately, the genetically engineered N-terminal 53 kDa fragment of AlaRS (AlaN470), which is sufficient for specific aminoacylation of cognate tRNAAla, proved as stable as the entire enzyme and could also be purified in greater amounts. This protein is, due to its reduced size, particularly valuable for NMR investigations.
Primary sequence alignment of the T.thermophilus AlaRS with other known AlaRSs revealed, as expected, a high degree of similarity around the three characteristic motifs of class II aaRSs. However, a retroviral-like zinc binding motif (C-X2-C-X6-H- X2-H) located between motifs 2 and 3 in the E.coli AlaRS (29 ) was not found at the homologous position of the T.thermophilus AlaRS (amino acids 173-186), a property shared with the known eukaryotic AlaRSs. Only one amino acid, namely the second cysteine (Cys176), of this zinc binding motif is present in the T.thermophilus AlaRS, while the first cysteine is replaced by a serine. The two histidines completing the motif in the E.coli enzyme are not found in the T.thermophilus counterpart, whose sequence is rather different in that region. It is, therefore, unlikely that the T.thermophilus AlaRS binds a metal atom in this region. For the T.thermophilus AlaRS as well as for the eukaryotic AlaRSs a protein-bound metal may not be a requirement for tRNA recognition, in contrast to the corresponding E.coli enzyme, as demonstrated by mutagenesis and binding studies (6 ,7 ). However, a cysteine (Cys666) that is not involved in zinc binding in the E.coli AlaRS (28 ) is present at the homologous position in the T.thermophilus enzyme (amino acid 682). Though located outside the N-terminal fragment (461N), which is sufficient for specific aminoacylation, this residue was shown to contribute to the catalytic efficiency of E.coli AlaRS. The whole region (T.thermophilus AlaRS amino acids 500-716) where this cysteine is located shows a high degree of sequence similarity with other known AlaRSs, indicating its, albeit unknown, functional importance.
A previous report demonstrated specific DNA binding by the E.coli AlaRS (30 ). It remained unclear which domain was responsible for this property. The T.thermophilus phenylalanyl-tRNA synthetase, which belongs to the same subclass (IIc) of aaRSs, also specifically binds DNA fragments located upstream of its own genes (our unpublished results). A sequence comparison of the assumed DNA binding domain 5 of the [beta]-subunit of this enzyme (31 ) with regions of unknown function of the AlaRS, however, revealed no obvious similarities. The structural basis of this rather unusual and poorly understood common property of both aaRSs has to be established.
A high degree of sequence similarity with other known AlaRSs was also found at amino acids 400-411, which is in the region that probably recognizes the critical G3-U70 base pair of tRNAAla. This region is supposed to form a module that folds back to the catalytic center to ensure proper tRNA recognition by making contacts with its acceptor stem (8 ). The accumulated conserved aromatic amino acids in this region may interact with the acceptor stem nucleotides.
The AlaRS from T.thermophilus was found to be a dimer. Recently it was shown that this also applies to the E.coli AlaRS (32 ). In an earlier publication it appeared as a tetramer (4 ). Association of E.coli AlaRS dimers to form larger species seems to occur only at lower temperatures. For the thermostable T.thermophilus AlaRS, however, even at room temperature, which is ~40oC below its optimum, no such association was found. Thus, the thermophile AlaRS resembles the homodimeric structure of most of the other prokaryotic class II aaRSs.
The observed non-aminoacylation by AlaN470 of the C3-A70 acceptor stem duplex in vitro is at variance with results obtained with the E.coli AlaRS (10 ), wherealanylation of mutant suppressor tRNAAla containing a C3-A70 mismatch pair was found in vivo. The differing behavior might be associated with the strongly reduced binding strength of the acceptor helix as compared with intact tRNAAla. The weakened interaction might prevent induced adaptation (`fine adjustment') of the AlaRS contact region to the recognition region of the RNA (mismatch pair), for which tight binding of the RNA could be an essential prerequisite, in particular if the recognition element has been modified with respect to the wild-type sequence.
In our preliminary NMR investigations presented here the acceptor helix of tRNAAla from E.coli, comprising 7 bp and the single-stranded ACCA 3'-terminus (inset of Fig. 4 ), was used to study the interaction with AlaN470. Interaction between the tRNA microhelix and AlaN470 is distinctly weaker as compared with the native, original molecule, since both interaction partners have been truncated. For NMR analysis this was advantageous, since the expected reduced association constant gave rise to a fast exchange of the RNA between the free and complexed forms, thus resulting in well-resolved NMR spectra of the RNA. In the case of a tightly bound complex, however, the effective molecular weight of ~60 kDa would have rendered a detailed NMR analysis impracticable, due to the greatly increased linewidths for the RNA moiety and the correspondingly poor spectral resolution.
From the preliminary NMR studies a weak, but specific, interaction between the acceptor stem of tRNAAla and AlaN470 can be derived. The most pronounced effect in the imino and amino resonance regions of the proton NMR spectra is found for the first base pair (G1-C72), indicating a specific contact of this site with the protein. Thereby, the exchange rate of the corresponding labile imino and amino protons is enhanced in comparison with the non-complexed state. This might be due to a widening or even disruption of the first base pair or specific contact with a side chain of the protein which is able to catalyze proton exchange upon interaction of the acceptor stem with AlaN470 in the major groove. Only slight effects upon the exchangeable proton resonances of the identity element base pair G3-U70 could be observed in the presence of AlaN470. In particular, the G3 amino-G3 imino crosspeak is not influenced by addition of AlaN470 to the acceptor helix solution. The presence of this crosspeak is all the more noteworthy as the G3 amino group is not involved in Watson-Crick base pairing and protrudes into the minor groove. Although it should thus be even more easily accessible to solvent molecules than in the case where the NH2 group is involved in hydrogen bonding, the opposite is found. In the present case a drastic reduction in amino proton exchange due to particular protection of this group is detected, maybe by a tightly bound water molecule, as found in a crystal structure analysis of a G-U pair-containing mismatch duplex (33 ). The lack of major changes in the exchangeable proton resonances cannot be taken as evidence for a lack of any specific interaction between the involved nucleotides and the protein and, in particular, between the G3-NH2 group and functional groups of the enzyme (9 ), since here only the exchangeable proton resonances were analyzed. Nevertheless, a change in the hydrogen bonding pattern of the G3 amino group seems improbable in view of the reported results. The linewidth (and, accordingly, linewidth changes) of imino resonances are mainly determined by the exchange rates of the corresponding imino protons with the solvent (H2O). The exchange rate can be affected by the accessibility to exchange-catalyzing buffer ions (34 ), by the presence of proper amino acid side chains of the protein or by alterations of the hydrogen bond (base pair) geometry. This can be induced by interaction with the protein and in turn gives rise to altered solvent accessibility. Consequently, from these analyses no assertions about possible changes in the helical geometry, in particular in the vicinity of the G3-U70 identity element, upon interaction with AlaN470 can be deduced. It seems possible that specific interactions between certain functional groups of the G3-U70 base pair and the enzyme occur, which, however, do not cause significant variations in the imino proton exchange rates of the exchangeable protons of the identity element base pair. More detailed structural information can be expected from an analysis of the non-exchangeable proton NMR spectra (in particular NOESY), which is presently underway.
We thank Prof. Dr M.Sprinzl for encouraging the present work and providing facilities, N.Grillenbeck for skillful technical assistance, Dr H.Faulhammer for peptide sequencing, Drs L.Arnold and G.Ott for RNA syntheses and M.Vogtherr and H.Schübel for support with the preparation of NMR samples.
*To whom correspondence should be addressed. Tel: +49 921 552433; Fax: +49 921 552432; Email: roland.kreutzer@uni-bayreuth.de
REFERENCES