Multispecific Aspartate and Aromatic Amino Acid Aminotransferases in

Two aminotransferases from Escherichia coli were purified to homogeneity by the criterion of gel electrophoresis. The first (enzyme A) is active on L-aspartic acid, L-tyrosine, L-phenylalanine, and L-tryptophan; the second (enzyme B) is active on the aromatic amiono acids. Enzyme A is identical in substrate specificity with transaminase A and is mainly an aspartate aminotransferase; enzyme B has never been described before and is an aromatic amino acid aminotransferase. The two enzymes are different in the Vmax and Km values with their common substrates and pyridoxal phosphate, in heat stability (enzyme A being heat-stable and enzyme B being heat-labile at 55 degrees) and in pH optima with the amino acid substrates. They are similar in their amino acid composition, each enzyme appears to consist of two subunits, and enzyme B may be converted to enzyme A by controlled proteolysis with subtilsin. The conversion was detected by the generation of new aspartate aminotransferase activity from enzyme B and was further verified by identification by acrylamide gel electrophoresis of the newly formed enzyme A. The two enzymes appear to be products of two genes different in a small, probably terminal, nucleotide sequence.

phenylalanine, and L-tryptophan; the second (enzyme B) is active on the aromatic amino acids. Enzyme A is identical in substrate specificity with transaminase A and is mainly an aspartate aminotransferase; enzyme B has never been described before and is an aromatic amino acid aminotransferase. The two enzymes are different in the V,,, and K, values with their common substrates and pyridoxal phosphate, in heat stability (enzyme A being heat-stable and enzyme B being heat-labile at 55") and in pH optima with the amino acid substrates.
They are similar in their amino acid composition, each enzyme appears to consist of two subunits, and enzyme B may be converted to enzyme A by controlled proteolysis with subtilisin. The conversion was detected by the generation of new aspartate aminotransferase activity from enzyme B and was further verified by identification by acrylamide gel electrophoresis of the newly formed enzyme A. The two enzymes appear to be products of two genes different in a small, probably terminal, nucleotide sequence.
We have reported recently the extensive purification of transaminase A of Escherichia coli by isoelectric focusing, gel filtration in Sephadex, and acrylamide gel electrophoresis into two forms designated IA and IB with apparent isoelectric points of about 4.55 and 4.60 and molecular weights of 82,000 and 88,000, respectively (I). The two forms catalyzed the transamination of L-tyrosine, n-phenylalanine, L-tryptophan, and, marginally, L-methionine with either ol-ketoglutarate or oxalacetate as cosubstrates. Form IA was not affected, but form IB was repressed by growth of the organisms in the presence of tyrosine. An aspartate:a-ketoglutarate aminotransferase activity coincided with form IA in electrofocusing columns; however, we were not prepared to claim that this activity resided in the same protein * This investigation was supported by a grant from the Medical Research Council of Canada.
with the aromatic activities, especially in view of conflicting reports claiming the nonidentity (2) and identity (3) of the aspartate and aromatic amino acid components. In the present study, forms IA and IS were recognized to represent different enzymes rather than forms of one enzyme and are abbreviated for convenience to enzyme A and enzyme B, respectively. Enzyme Assays-Tyrosine:2-oxoglutarate aminotransferase was assayed by the method of Diamondstone (6). The reaction mixture contained in 3.2 ml the following: 0.2 M potassium phos phate buffer (pH 7.3), G.0 mM L-tyrosine, 9.4 mM a-ketoglutarate, 38 PM pyridoxal phosphate, and enzyme. The mixture was preincubated at 37" for 10 min and the reaction was stopped with 0.2 ml of 10 N NaOH. After 30 min at room temperature the optical density was read against a control to which NaOH had been added prior to a-ketoglutarate.
A molar extinction coefficient of 19,900 M-' cm-i was used.
Phenylalanine aminotransferase was assayed as the tyrosine aminotransferase except that tyrosine was replaced by 6 mM phenylalanine and the optical density was read at 315 nm. A molar extintion coefficient of 17,500 M-l cm-l was used.
Tryptophan aminotransferase was assayed also aa the tyrosine aminotransferase except that tyrosine was replaced by 6 mM Ltryptophan and the optical density was read at 335 nm. A molar extinction coefficient of 10,000 M-' cm-i was used.
Aspartate:2-oxoglutarate aminotransferase (EC 2.6.1.1) was assayed by the method of Karmen (7) by coupling with malate dehydrogenase at 25". The reaction mixture contained in 3.0 ml the following: 0.1 M potassium phosphate buffer (pH 7.6), 178 mM n-aspartate, 6.4 mM ru-ketoglutarate, 38 PM pyridoxal phosphate, 0.24 mM NADH, 10 units of malate dehydrogenase, and enzyme. The reaction was started by adding the enzyme and was followed bv the fall of ontical densitv at 340 nm. A molar extinction coefficient of 6.2 X i03 M-I cm-i was used.
One enzyme unit is defined as that amount of enzyme which catalyzes the conversion of 1 pmol of substrate to product per min at the temperature of the assay.
Acrylamide Gel Electrophoresis-The method of Davis (S) was used with an acrylamide concentration of 10% in a Polyanalyst electrophoresis apparatus (Buchler Instruments) at 1-Z". The gels were prepared in glass tubes (75 or 100 X 5 mm) at 3 ma per gel, with bromphenol blue as the tracking dye. Proteins were visualized with Coomassie brilliant blue R. Aminotransferase activity was detected by incubating the gels in the mixture used by Ryan et al. (9) for 20 to 30 min. Purple rings were formed at the position of transaminase activity. Staining for aspartate aminotransferase was poor compared with the excellent staining obtained for the activities toward the aromatic amino acids and we prolonged the incubation with aspartate to 45 to 60 min which practice also resulted in much darker backgrounds.
Inferior staining in the presence of aspartate has been observed also by others (2). The molecular weight of subunits was estimated by sodium dodecyl sulfate gel electrophoresis as described by Weber and Osborn (10).
Heat Sensitivity Tests-The purified enzyme solutions were diluted IO-fold with 0.2 M potassium phosphate buffer, p1-I 7.3, to a final volume of 4.0 ml and serum bovine albumin was added to a final concentration of 5 mg/ml. The solutions were placed in thin walled glass tubes (20 X 100 mm) supplied with small magnetic stirring bars and were immersed into a water bath at 100" with constant stirring until the temperature was raised to about 52". This required about 20 s. The tubes then were transferred to a water bath at 55" and kept for 10 min with constant stirring. Aliquots were withdrawn at specified time intervals, quickly chilled in ice, and subsequently assayed. An aliquot was kept in ice throughout and served as the zero time control. Activities were expressed as percentages of the control.
Amino Acid Analysis-Samples of enzyme A (0.2 ml) and enzyme B (0.6 ml) were hydrolyzed with 6 N HCI at 110' for 24 hours in tubes sealed after evacuation. After hydrolysis the tubes were opened and HCl was removed in a vacuum desiccator and the sample was dissolved in 0.3 M sodium citrate buffer, pH 2.0. Analysis of the hydrolysate was carried out on a TSM amino acid autoanalyzer.
Apoenzyme Formation-The purified enzyme solutions were allowed to form the pyridoxamine form, which is believed to be more dissociable than the pyridoxal form (II), in the presence of L-tyrosine. Purified enzymes A and B were diluted 20-fold and 40fold, respectively, with 0.1 M potassium phosphate buffer, pH 7.3, supplemented with 1 mM EDTA, 1 mM dithiothreitol, and 5 mM n-tyrosine to a volume of 8.0 ml and were allowed to stand at 4" for 4 hours. They then were dialyzed for 19 hours against two changes of 2 liters each of the above buffer minus L-tyrosine. Full reactivation of both enzymes was obtained upon saturation with phate, enzyme A still possessed 56% of the activity obtained upon full activation with the cofactor, the corresponding figure being 9% for enzyme B. The experiment was repeated replacing Ltyrosine with L-phenylalanine with no more success. Therefore, in kinetic experiments designed to calculate the K, and V,,,, of pyridoxal phosphate, the plots of the velocity (vertical axis) versus the apparent (added) pyridoxal phosphate concentration were extrapolated to zero velocity. The intercept of the curve with the horizontal concentration axis (to the left of the velocity axis) was set as zero pyridoxal phosphate concentration and the new real cofactor concentrations were written on the horizontal axis. Double reciprocal plots then were constructed to calculate the approximate K, and-V,,,, values for pyridoxal phosphate.
Kinetic Experiments-Double recinrocal nlots according to Lineweaver and Burk (12) were used for the calculation of kinetic constants. Since in transamination reactions more than one substrates is involved, the K, and V,,,,, values given in Table I are apparent values at pH 7.3 and were calculated for each variable substrate, while pyridoxal phosphate and a-ketoglutarate were at saturating and L-tyrosine at 6.0 mM concentrations.

PurQicafion
and Resolution oj Two Enzymes on Preparafive Isoelectric Focusing Column-Purification and resolution by focusing in the small (110 ml) column resulted in partially overlapping bands so that it was possible to obtain only small amount,s of enzyme from either band without cross-contamination.
In the large (440 ml) preparative column, which could be loaded with 5 to 6 times more enzyme units, superior resolution was obtained and sufficient amounts of enzymes were recovered for kinetic and other experiments. Fig. 1 shows the separation obtained in the large column, and this profile was reproducible.
The intermediate low activity fractions were rejected and only the fractions of the highest activity from each peak were pooled, dialyzed, and concentrated by ultrafiltration and tested for protein and enzyme activity bands by gel electrophoresis.
Each of the two enzymes gave a single protein band and a single enzyme activity band when stained in the presence of L-tyrosine, L-phenylalanine, and L-tryptophan and single protein bands with acrylamide concentrations varying from 5 to 11% (not shown). Therefore, by the criterion of gel electrophoresis, the two preparations, A and U, ing-Isoelectric focusing, not shown here, on a small (110 ml) column, resolved an enzyme preparation from DEAE-cellulose column chromatography into two peaks, A and 1~ (enzymes A and B), when the fractions were assayed for transaminase activity with L-tyrosine, L-phenylalanine, and L-tryptophan. Assaying for aspartate aminotransferase resulted in a single peak completely coincident with Peak A. The coincidence of aspartate and aromatic amino acid activities in enzyme A was further confirmed by gel electrophoresis of enzyme A. Staining the gels for protein, aspartate aminotransferase, and a mixture of aspartate and tyrosme aminotransferases resulted in a single band in each case (Fig. 2). Kinetic Experiments- Table  I lists the K, and V,,, values and their B:A ratios for enzymes A and B. The V,,, values clearly indicate that the major activity toward the aromatic amino acids resides in enzyme 13, the B:A ratio varying from about 14 to 25. In the presence of equimolar amounts of the two enzymes, a maximum of only about 7% of the total V,,, toward phenylalanine and only a few per cent toward tryptophan and tyrosine would be contributed by enzyme A. On the other hand, the major activity of enzyme A is toward aspartate as attested by the high I',,,,, value for this amino acid compared to the V,,,,, values for the aromatic amino acids (Table I). Since aspartate aminotransferase was assayed at 25", whereas the aromatic aminotransferases were assayed at 37", the relative magnitude of V Inax for aspartate is actually underestimated in Table I by a factor of about 2 (if about 2.0 be considered the value of temperature coefficient Q10 in this instance). The V,,, values for the two keto acids and pyridoxal phosphate (in the presence of tyrosine as the amino acid substrate) were also higher with enzyme B than with enzyme A as indicated by their high B:A

Prot
Asp Tyr +Asp - Polyacrylamide gel electrophoresis (in tubes, 100 X 5 mm) after isoelectric focusing, dialysis, and concentration of enzyme A. The gels were prerun for 1 hour and then were overlayered with 0.48 Kg (left gel) and 1.92 pg (middle and right gels) of enzyme A and run for 4.5 hours at 3 ma per gel. The left gel was stained for protein (Prot.1, the middle gel for aspartate aminotransferase, and the right gel for both tyrosine and aspartate aminotransferases. ratios. The K, values for a-ketoglutarate and pyridoxal phosphate were 35-and 40.fold higher with enzyme B than with enzyme A, respectively, indicating significantly higher affinities of the keto acid and t,he cofactor for enzyme A than for enzyme 13 (Table I). The lower K, value of the cofactor with enzyme A is in agreement with its tighter physical binding during dialysis in our experiments for apoenzyme formation. The same relationship, qualitatively, was found in the K, values of the other keto acid substrate oxalacetate (Table I).
Subunit Structure-Each enzyme on acrylamide gel electrophoresis in the presence of sodium dodecyl sulfate yielded a single protein band which corresponded to a molecular weight of 42,000 to 45,000 (range of five experiments). Since the molecular weights of enzyme A and 1~ were estimated to be about 82,000 and 88,000, respectively (l), each appears to consist of two subunits of equal size. In two experiments, one of which is shown in Fig. 3, we were able to calculate a slightly higher subunit molecular weight for enzyme 13 (43,000) than for enzyme A (42,000), in The nanomoles of each enzyme in the purified preparations were estimated from the known molecular weights and protein concentrations. The latter were calculated from the amino acid analyses. keeping with the higher molecular weight of enzyme B. The failure to calculate consistently a higher subunit molecular weight for enzyme B than for enzyme A is due to the short distance traveled by proteins of so similar size in the sodium dodecyl sulfate gels (about 1.7 cm in 5 hours at 8 ma per gel).
plf Activity Curves- Fig.  4 illustrates the dependence of activity of the two enzymes on pH. The pH optima for enzyme A with its four substrates vary from 8.0 (toward aspartate) t.o 9.0 (toward tyrosine) with 8.5 being the pH optimum toward phenylalanine and tryptophan (Fig. 4A). For enzyme 1'1 the pH optimum toward tyrosine drops to 8.0 with 7.5 and 8.0 being the optima toward phenylalanine and tryptophan (Fig. 4B). In view of the physical evidence of homogeneity, the differences in the pH optima of each enzyme with its substrates reflect the differences in the pK values of the substrates rather than the existence of different enzymes.
Heat Sensitivity Tests-The time courses of act,ivity loss of the two enzymes toward their substrates (tryptophan was not included) at 55" are shown in Fig. 5. Enzyme B rapidly lost activity with tyrosine and phenylalanine as substrates so that, after 10 min, less than 10% of the original activity was left. Enzyme A was assayed with aspartate (Asp), phenylalanine (Phe), and tyrosine (Tyr) and enzyme B was assayed with phenylalanine and tyrosine.

4131
Enzyme A, however, was relatively heat stable, with 52 and 647" of the original activity remaining after 10 min with tyrosine and phenylalanine, respectively, as substrates. For each enzyme the initial rate of inactivation as well as the final activity loss (especially for enzyme A) was more pronounced with tyrosine than with phenylalanine as substrate and this was verified in another experiment (not shown). More striking was the almost complete absence of inactivation of enzyme A with aspartate as substrate seen against its partial loss of activity with the two aromatic amino acids (Fig. 5). Kinetic evidence (to be reported in detail elsewhere) indicates that of the two ionizing groups at the active site of enzyme A, only one is affected by the binding of aspartate, but both are affected by the binding of tyrosine. The less stringent requirement for aspartate binding most probably accounts for the heat stability of the enzyme versus this substrate. A heat sensitivity test such as ours applied to a crude or semipurified preparation containing the two enzymes would yield a heatstable aspartate aminotransferase, a more labile phenylalanine aminotransferase, and an even more labile tyrosine aminotransferase and it would be misinterpreted to suggest the existence of one enzyme specific for aspartate and one enzyme (and possibly two) specific for the aromatic amino acids. Such a plot, derived from an experiment with a crude extract, was published recently (2) in support of the nonidentity of the two types of activity.
Amino Acid Composition-The amino acid composition of the two enzymes was looked into with the purpose of obtaining some insight into the structure of the two enzymes. Their similarity in size and especially in charge suggested that they might be similar proteins. The data of Table II do indeed indicate that their amino acid compositions are very similar. The significant differences in the number of amino acid residues appear to be limited to glutamic acid, glycine, alanine, valine, and possibly to lysine and phenylalanine.
This suggested to us that extensive homology might exist between the two proteins and we were successful in converting enzyme B to enzyme A by controlled proteolysis.
Conversion of Enzyme B to Enzyme A by Subtilisin Treatment-Incubation for 30 min of enzyme R at 37" in the presence of TyrA, tyrosine aminotransferase.
subtilisin resulted in a nearly 6-fold increase in its very low basal aspartate aminotransferase activity in the first 10 min and the activity declined only slightly thereafter (Fig. 6). The tyrosine aminotransferase activity declined rapidly in the first 10 min to about 50% of its zero time value and remained practically unchanged thereafter.
Incubation in the absence of subtilisin (control) resulted in some loss of tyrosine aminotransferase activity, whereas the aspartate aminotransferase activity did not change in agreement with the relative heat stabilities of these two activities (Fig. 5). Incubations with subtilisin concentrations double and quadruple that used in Fig. 6 resulted in the generation of much lower net aspartate aminotransferase (not shown). These results suggested that enzyme A with its typical aspartate aminotransferase activity was the product of the controlled proteolysis of enzyme B by subtilisin. This was demonstrated by gel electrophoresis of subtilisin-treated enzyme B. Fig. 7 (Gel 2) shows clearly an enzyme activity band with the mobility of enzyme A. This band is absent from the zero time control (Gel 1) and the subtilisin control  5) ; (d) optima pH for the same substrates (Fig. 4). On the other hand, the two enzymes are similar in amino acid composition, size, and net charge, similarities which readily explain the failure of Rudman and Meister to distinguish between them in their classic study on transamination in E. coli (13) with the then available techniques. In addition, there appears to exist extensive homology in the primary structures of the two en zymes.
Enzyme A, formerly reported as form IA (l), was first described by Rudman and Meister and given the name transaminase A (13). Enzyme B, formerly reported as form IB (l), has not been previously recognized in E. coli as a distinct enzyme. According to the recommendations of the Commission of Biochemical Nomenclature the term "multiple forms" should be used for proteins possessing the same enzyme activity (14). Accordingly, the two proteins of Fig. 1 do not represent multiple forms of an aromatic aminotransferase but, rather, two different enzymes, one of which (enzyme A) is active mainly with aspartate and the other (enzyme B) with the aromatic amino acids (Table  I). Our previous reluctance to identify aspart'ate aminotransferase with enzyme A (1) led us to believe that the aromatic activities of the two proteins represented multiple forms of an essentially aromatic aminotransferase, with aspartate aminotransferase being perhaps an additional distinct protein. The physical evidence for the identity of aspartate aminotransferase with enzyme A rests with two powerful analytical methods, namely, isoelectric focusing and acrylamide gel electrophoresis, failing to resolve the activity toward aspartate from the activity toward the aromatic amino acids in enzyme A. We are thus in disagreement with a recent report (2) claiming the nonidentity of aspartate and aromatic amino acid activities in enzyme A. We propose multispecific aspartate aminotransferase as the recommended name (14) to replace the name transaminase A.
Enzyme B, an aromatic amino acid aminotransferase, has never been described before in E. coli and its "dissection" from enzyme A has some important consequences regarding the actual aminotransferase involved in the regulation of tyrosine and phenylalanine biosyntheses. It has been reported (15) that transaminase A of E. coli is specifically repressed by L-tyrosine and so is enzyme B of the present study (1). The enzyme, assayed only as tyrosine and phenylalanine aminotransferases, was found in crude extracts to be heat-labile and to account for most of the activity toward the two aromatic amino acids (15). The heat lability and substrate specificity conform with the properties of our enzyme B (Table I, Fig. 5). It is now apparent that, by assaying only for aromatic aminotransferase activities, the above investigators (15) were observing in the crude extracts the repression not of transaminase A but of t,he new enzyme B. They were led to their erroneous conclusion because of the omission to assay for aspartate aminotransferase.
Accordingly, references in the literature on the repression by tyrosine of transaminase A assayed as tyrosine aminotransferase (which is the usual practice) reflect the repression of enzyme B, not of transaminase A which is insensitive to repression by tyrosine.
In the same study (15) evidence was presented purporting to show that there exists in E. coli a specific phenylalanine aminotransferase distinct from transaminase A with the following properties studied in crude extracts: (a) relatively heat-stable at 60" (25 to 30% loss of activity after 10 min) (Fig. 1 (1) ; (c) contributes little to the total activity toward phenylalanine (Table I); and (d) was separated with some difficulty from the strictly aromatic activity of enzyme B, with techniques not used or unavailable to the previous investigators (15). It must be emphasized that we are not claiming that a distinct phenylalanine aminotransferase does not exist in E. coli, only that the earlier studies (15), rather than providing evidence for such an enzyme, merely described properties of transaminase A itself against the background of the unsuspected enzyme B. In the absence of rigorous enzymological evidence and in the light of the present data its occurrence in these organisms must remain in doubt. Consequently, and until a specific phenylalanine aminotransferase is unequivocally established, it is enzyme B that appears to be involved in the terminal step of phenylalanine, as well as of tyrosine, biosynthesis.
The conversion of enzyme B to enzyme A by controlled proteolysis in the presence of subtilisin is of interest. The conversion experiments were suggested by, and are in agreement with, the similarity in amino acid composition of the two enzymes. It appears likely that cleavage of a terminal polypeptide tract from each subunit of enzyme B is responsible for the generation of enzyme A under our conditions of limited proteolysis.
It is unlikely that the process is of physiological significance. Since enzyme B is selectively repressed by L-tyrosine whereas enzyme A is unaffected (1) the two enzymes are products of two genes different only in a small, probably terminal, nucleotide sequence. The enzymological implications of the conversion are of interest also. It seems clear that cleavage of polypeptide(s) from enzyme B is sufficient to generate a new activity, aspartate aminotransferase, typical of enzyme A. In this respect, the conversion is reminiscent of the long known zymogen-enzyme relationships. The very low aspartate aminotransferase activity of enzyme B noted in Fig. 6 in the absence of subtilisin may mean that masking of the activity in enzyme B is not 100% complete or that enzyme B was slightly contaminated with enzyme A during isolation from the isolectric focusing column. To the best of our knowledge this is the first instance of an enzyme being converted to another distinct, physiologically occurring enzyme by controlled proteolysis.