Protein Engineering of Homodimeric Tyrosyl-tRNA Synthetase to Produce Active Heterodimers*

Heterodimers of tyrosyl-tRNA synthetase from Ba-cillus stearothermophilus have been produced by mu- tagenesis at the subunit interface. Oppositely charged groups have been engineered into the subunits so that they can form a complementary pair. Wild-type tyro-syl-tRNA synthetase is a symmetrical dimer in which the side chains of the 2 Phe-164 residues interact at the subunit interface. Phe-164 was mutated to Asp in tyrosyl-tRNA synthetase and to Lys in a truncated enzyme (des-(321-419)tyrosyl-tRNA synthetase) which lacks the two tRNA-binding sites, but which can catalyze pyrophosphate exchange. The size difference allows subunit association to be studied by gel filtration chromatography. These changes induce reversible dissociation from active dimers into inactive monomers at pH values which favor ionization at position 164. A mixture of the two mutants near neutral pH is appar- ently fully active in pyrophosphate exchange and con-sists of a heterodimer of [A~p'~~]tyrosyl-tRNA synthe- tase and [Ly~'~~]des-(321-419)tyrosyl-tRNA synthetase. Despite having only one binding site for tRNA, heterodimer has full aminoacylation activity at high concentrations of tyrosine. We have therefore produced a family of dimers that differ in stability near neutral pH. This novel approach using protein engi-

Heterodimers of tyrosyl-tRNA synthetase from Bacillus stearothermophilus have been produced by mutagenesis at the subunit interface. Oppositely charged groups have been engineered into the subunits so that they can form a complementary pair. Wild-type tyrosyl-tRNA synthetase is a symmetrical dimer in which the side chains of the 2 Phe-164 residues interact at the subunit interface. Phe-164 was mutated to Asp in tyrosyl-tRNA synthetase and to Lys in a truncated enzyme (des-(321-419)tyrosyl-tRNA synthetase) which lacks the two tRNA-binding sites, but which can catalyze pyrophosphate exchange. The size difference allows subunit association to be studied by gel filtration chromatography. These changes induce reversible dissociation from active dimers into inactive monomers at pH values which favor ionization at position 164. A mixture of the two mutants near neutral pH is apparently fully active in pyrophosphate exchange and consists of a heterodimer of [A~p'~~]tyrosyl-tRNA synthetase and [Ly~'~~]des-(321-419)tyrosyl-tRNA synthetase. Despite having only one binding site for tRNA, heterodimer has full aminoacylation activity at high concentrations of tyrosine. We have therefore produced a family of dimers that differ in stability near neutral pH. This novel approach using protein engineering allows specific dimerization of subunits of the same size that have different defined mutations, each subunit being tagged by the charge. Such hybrid proteins can be used to study subunit interaction.
Wild-type tyrosyl-tRNA synthetase is a homodimer (Mr = 2 X 47,300) (1) with each subunit having one active site (2). In common with several other enzymes, only one active site appears functional per dimer (3). Neither the mechanism nor the role of this half-of-the-sites activity is known. The enzyme ( E ) catalyzes the aminoacylation of tRNA as a two-step reaction (Equations 1 and 2).  (2). The side chains of the 2 Phe-164 residues interact at the subunit interface and lie on the axis of symmetry (2). Mutation of Phe-164 to Asp in tyrosyl-tRNA synthetase introduces potential negative charges at the hydrophobic interface, and induces pH-dependent reversible dissociation from active dimer into inactive monomers (4). Deletion of the carboxyl-terminal domains of wild-type tyrosyl-tRNA synthetase (producing des-(321-419)tyrosyl-tRNA synthetase,' M , = 2 x 36,300) abolishes tRNA-binding (Reaction 2), but the formation of tyrosyl adenylate is unaffected (Reaction 1) (5). In this study Phe-164 is mutated to Lys-164 in des-(321-419)tyrosyl-tRNA synthetase to introduce potential positive charge into the hydrophobic subunit interface and so induce activity and dissociation analogous to that of [A~p'~~]tyrosyl-tRNA synthetase but with the opposite pH dependency. We mixed monomeric  Fig. 1). We show that such heterodimers are much more stable a t neutral pH than are the parent homodimers, and that they are apparently fully active.

RESULTS
Previous work (4) has shown that the monomeric form of the mutant [A~p'~~]tyrosyl-tRNA synthetase is inactive and has only weak, if any, affinities for tyrosine and ATP. The dimer is active and has higher affinities for these ligands. Association of subunits is therefore favored by high concentrations of the substrates. [A~p'~~]Tyrosyl-tRNA synthetase has the same value of kcat in pyrophosphate exchange at pH 7.8 as does wild-type enzyme, but the value of K, for tyrosine is greatly increased (4). [Asp'64]Tyr~~yl-tRNA synthetase is predominantly monomeric above pH 6 and so forms enzymebound tyrosyl adenylate only slowly and with a low stoichiometry which falls with increasing pH as the Asp-164 ionizes (4) (Fig. 2). [Lys"j4]des-(321-4l9)Tyrosyl-tRNA synthetase behaves in an analogous manner, having a reversible change of activity with pH except that activity decreases at low pH as Lys-164 becomes protonated. But, on mixing the two mutants at pH 7.8, enzyme-bound tyrosyl adenylate is formed with a stoichiometry approaching 1 mol/mol of dimeric enzyme, as found with the wild type (Fig. 2 ) . As the mixture of the two mutants is more active than the constituents, there must be a functional interaction between the two types of subunit.
The composition of the active heterodimer was confirmed by determining its apparent M , by gel filtration using FPLC. Both mutants at position 164 are predominantly monomers in the absence of tyrosine (Table I) synthetase is predominantly dimeric at high pH when the tamino groups are more readily deprotonated. At pH 7.8, both mutants are mainly monomeric even in the presence of tyrosine (Table I). A mixture of the two mutants at p H 7.8 in the absence of tyrosine elutes as a single broad peak of apparent M , = 57,000, indicative of weak dimerization. In the presence of tyrosine, the mixture elutes as a single sharper peak of apparent M , = 77,000 which is much greater than that for either of the constituents under the same conditions and is close to that expected for the heterodimer ( In the presence of low concentrations of tyrosine at pH 7.8, the individual mutants have low activity in pyrophosphate exchange (Reaction 1) and aminoacylation of tRNA (Reactions 1 and 2 ) ( Table II), showing that inactive monomers predominate under these conditions. On mixing the two mutants they have identical activity to wild-type enzyme in pyrophosphate exchange and low activity in tRNA charging (Table 11). Adding tRNA before tyrosine and ATP, compared with adding tRNA after the other substrates, gives lower rates of charging or pyrophosphate exchange by [A~p'~~]tyrosyl-tRNA synthetase or by heterodimer. This is because tRNA binds to catalytically inactive monomeric [A~p'~~]tyrosyl-tRNA synthetase and apparently inhibits formation of active dimer The mutation was verified by dideoxy DNA sequencing (7). All enzymes were expressed and purified as described by Lowe et al. (8). Each preparation was homogeneous on sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Active site titrations were performed by measuring accumulation of the stable enzyme-tyrosyl adenylate complex in the presence of pyrophosphatase (Reaction 1) (4). Enzyme (60 pl) was added to 40 p1 of assay mixture and incubated for 5 min. mutations is not limited in this way. The difference in sizes was used only to facilitate analysis of production of heterodimers. Production of heterodimers using urea resulted in a significant irreversible loss of activity (lo), which did not occur in the current work when polar mutations were used. Employing polar mutations rather than reversible denaturation, heterodimers can be prepared in much larger quantities. We have shown that heterodimers of polar mutants of position 164 predominate over parent monomers or dimers so that a mixture of full-length [A~p'~~]tyrosyl-tRNA synthetase and full-length [Lys"j4]tyrosy1-tRNA synthetase will give heterodimers. This will allow future experiments in which individual amino acid residues in defined subunits may be varied where both monomers have tRNA binding domains. Conservation of the overall size symmetry should facilitate x-ray crystallography studies. Such enzymes will also be valuable for studying half-of-the-sites activity. Although the mutations introduced   in this study do weaken the association of subunits, they do not appear to change the functional interactions across the subunit interface as heterodimers are apparently fully active and show properties similar to wild type rather than parent mutant homodimer. Heterodimer and wild type have similar activities in pyrophosphate exchange at low substrate concentrations and both show half-of-the-sites reactivity. Detailed analysis of these mutant enzymes will indicate if changes a t position 164 alter only the association of subunits or whether they also affect the activity of the dimers after they are formed. If the absolute properties of the dimers can be measured, this system can be used to investigate the basis of subunit association. The kinetic behavior of the dissociating enzymes would provide a means of assaying functional consequences of structural changes. By making two mutations at position 164 of tyrosyl-tRNA synthetase we have produced a family of dimers which decrease in stability near neutral pH in the order: wild type > heterodimer >> [A~p'~]tyrosyl-tRNA synthetase = [Lys"j4] des-(321-419)tyrosyl-tRNA synthetase. Heterodimers produced both by urea denaturation (9, 10) and by salt bridging (Table 11) have charging activity despite having only one tRNA-binding site. Thus, the model for aminoacyl-tRNA synthetase and tRNA interaction where pseudosymmetrical regions of tRNA are recognized by symmetrical regions on two tRNA-binding domains of the enzyme (11) does not apply in this case.
Formation of hybrid oligomeric proteins by the classical technique of reversible denaturation and reassociation has been valuable in the study of subunit interactions, especially where the subunits are of differing composition (12). Many enzymes cannot regain significant activity after denaturation. Our novel application of protein engineering to change quaternary structure allows hybrids to be formed where subunits are of the same size and, possibly, in cases where enzymes may not be reversibly denatured.