Purification, Properties, and Identity of Liver Mitochondrial Tyrosine Aminotransferase*

Abstract Mitochondrial tyrosine aminotransferase (l-tyrosine + 2-oxoglutarate = p-hydroxyphenylpyruvate + l-glutamate, EC 2.6.1.5) has been purified about 900-fold to homogeneity from rat liver and was crystallized. The enzyme is cationic and has a molecular weight of approximately 90,000, determined by amino acid composition and molecular sieve chromatography and of 114,000 by sedimentation analysis. It catalyzes the conversion of tyrosine to p-hydroxyphenylpyruvate and the reverse reaction. From the reverse direction with p-hydroxyphenylpyruvate, the reaction has a 3-fold greater rate with l-glutamate as cosubstrate than with aspartate. In the forward direction with l-tyrosine and α-ketoglutarate or oxaloacetate as cosubstrate, l-aspartate, α-methyl-l-aspartate, and β-methyl-l-aspartate are potent inhibitors. With α-ketoglutarate, amino acids were effective as cosubstrates in the following order of activity: l-aspartate g l-phenylalanine g l-cysteine g l-tyrosine g monoiodo-l-tyrosine = dihydroxy-l-phenylalanine g l-tryptophan g l-methionine = l-asparagine = p-chloro-l-phenylalanine. Evidence based on substrate specificity, inhibition by common substrates, inhibition by specific antibodies upon enzyme activity, and physical chemical properties indicate that the mitochondrial tyrosine aminotransferase is identical with mitochondrial aspartate aminotransferase (laspartate: 2-oxoglutarate aminotransferase, EC 2.6.1.1). This activity relationship would then distinguish this enzyme from the tyrosine aminotransferase and aspartate aminotransferase found in cytosol on the basis of properties and function.

The enzyme is cationic and has a molecular weight of approximately 90,000; determined by amino acid composition and molecular sieve chromatography and of 114,000 by sedimentation analysis. It catalyzes the conversion of tyrosine to p-hydroxyphenylpyruvate and the reverse reaction.
From the reverse direction with p-hydroxyphenylpyruvate, the reaction has a S-fold greater rate with L-glutamate as cosubstrate than with aspartate.
In the forward direction with L-tyrosine and cY-ketoglutarate or oxaloacetate as cosubstrate, L-aspartate, cY-methyl-L-aspartate, and /3-methyl-L-aspartate are potent inhibitors.
Evidence based on substrate specificity, inhibition by common substrates, inhibition by specific antibodies upon enzyme activity, and physical chemical properties indicate that the mitochondrial tyrosine aminotransferase is identical with mitochondrial aspartate aminotransferase (Laspartate: 2 -oxoglutarate aminotransferase, EC 2.6.1.1). This activity relationship would then distinguish this enzyme from the tyrosine aminotransferase and aspartate aminotransferase found in cytosol on the basis of properties and function.
Hepatic tyrosine aminotransferase activity is found in cytosol and in mitochondria (l-6), but this activity is the result of the action of different enzymes (6). The mitochondrial form of * This research was supported by Research Grants AM-08350 from the National Institute of Arthritis and Metabolic Diseases and by CA-10439 from the National Cancer Institute.
i Recinient of Postdoctoral Training Grant CA-05197 from the N$tional Cancer Institute.
Present aYddress, Division of Biological Research, G. D. Searle and Company, Chicago, Illinois 60680.
5 To whom correspondence regarding this paper should be addressed.
the enzyme seems to have a broader tissue distribution than the cytosol form (5, 7).
The finding of very low levels of the hepatic cytosol enzyme in a patient who had increased levels of urinary p-hydroxyphenylpyruvate suggested that, at least under some physiological conditions, the liver mitochondrinl enzyme can function metabolically (5). This study was uudertaken to isolate and investigate the properties of the mitochondrial tyrosine aminotransferase from rat liver.
Evidence is presented to suggest that the mitochondrial tyrosirre aminotransferase reaction is catalyzed by the same protein that catalyzes aspartate transamination.
Tyrosine transnmirrntion, perhaps, is also a property of a anine aminotransferasc (L-alanine : 2-oxoglutarate aminotranlsferase, EC 2.6. 1.2) and the action of all of these enzymes can result in the utilizatiou or formation of tyrosine.
The mitochondrial suspension was homogenized, adjusted to a final concentration of 5 volumes of buffer per g of original tissue, and frozen for at least 48 hours. The preparation was allowed to thaw at 37" in a shaking water bath and was removed just as the last ice was disappearing.
The preparation was homogenized again gently with a Potter-Elvehjem homogenizer and centrifuged at 15,000 rpm in a Servall centrifuge for 30 min to sediment the membranous material and other particulate material which might be present.
Purification of Enzyme from .lfitochondrial Extracts-'Tyrosine aminot,ransferase activity was purified from the crude extract by addition of solid ammonium sulfate to final concentrations of 3Oc/;,, 50$& 6055,, and 907; of saturation.
When possible, the 50 to 6040 fraction was colle&d by allowing the preparation to stand overnight to improve the yield of enzyme activity utilizing pyrurate as amino group acceptor.
If the yield of enzyme was incomplete in the 60 to 9O(j', fraction, more ammonium sulfate was added to bring the concentration to 95% saturation. The active fraction, 60 to 90y0, was suspended in 0.01 M potassium phosphate, pH 7.2, containing 0.01 mM pyridoxal-5'.P and 0.1 rnM cu-ketoglutarate and dialyzed overnight against 200 volumes of the same buffer. Occasionally, protein precipitation occurred during dialysis and this was not, preventled by adding 1 mM EDTA alld 1 rnnr dithiothreitol to the buffer. This could be overcome partially by increasing the buffer concentration to 0.05 M or 0.1 M and then diluting the preparation to 0.01 M just prior to addition of the enzyme to the column for the next step of the purification. The preparation was added to a column of carboxymethyl-Sephadex C-50 (3 x 28 cm) which had been equilibrated with 0.01 RI Medium A (0.01 M potassium phosphate, pH 7.2, containing 0.01 rnM pyridoxal-5'-P and 0.1 mM oc-ketoglutarate). After the initial protein peak was collected, the concentration of the I\ledium 1 was increased to 0.02 M for the equivalent of 1.5 col-urn11 rolumes and then was increased to 0.05 M for collection of the enzyme. Initial studies with 0.03,0.04, 0.05 M buffer demonstratcd that the enzyme came off over a broad range from about 0.03 to 0.05 M depending somewhat on the $1 of the buffer system. The buffer concentration was then raised to 0.1. M to remove most of the remaining protein and to determine that all of the enzyme had been recovered in the previous fraction.
Only traces of enzyme could be found at hisher molarities.
Elevation of the buffer concentration to 0.3 M yielded no more enzyme and only small amounts of material with absorbance at either 260 or 280 nm. The active fractions were pooled and concentrated using an Amicon ultrafiltration unit consisting of Dia-Flo model 50 ultrafiltration cell and a l)ia-Flo ultrafiltration membrane U&l-10.
Pressure was maintained by connection to an external nitrogen source with a filter and drying agent in the system. Ultrafiltration was carried out at approximately 4". Recovery was quantitative with no activity in the filtrate solution. The concentrated enzyme preparation was subjected to isoelectrofocusing using a 110.ml column, a pI1 gradient of 3 to 10, 300 volts, and a running time of approximately 30 hours. No detectable difference was seen when a time range of 19 to 65 hours were compared except that the peak may be more diffuse at 19 hours. The active fractions around pH 9.3 were pooled and applied to a column of Sephadex G-100 for rapid removal of ampholines.
The enzyme was concentrated by ultrafiltration as described above and stored frozen in 0.1 M Medium A in a concentration range of 5 to 10 mg of protein per ml.
Ultracentrifugafion Studies--Bnalytical ultracentrifugation was performed with a Beckman model E instrument. Sedimentation coefficients were determined by standard procedures using a double sector cell at 60,000 rpm at 20". Photographs were taken at 8-or 16.min intervals.
Diffusion constants were determined using an artificial boundary cell at 22,000 rpm and 20" by determining the slope of a plot of a2 in square centimeters versus 2t in seconds (8,9). Photographs were taken at 8-or 16.min intervals.
Measurements of distances moved by the maximum ordinate or of c were made with a Nikon two-dimensional microcomparator. Molecular weight values were calculated from the s2c,w and the D20,W values at 10 mg protein per ml, determined experimentally, according to the Svedberg equation For this equation the partial specific volume (8) was calculated from the amino acid composition to be 0.75 according to the method of Cohn and Edsall (10).
Enzyme Assays-Tyrosine aminotransferase was measured by the modified Briggs assay (12, 13), the Diamondstone method (14), and the method of Gabay and George (15). A unit of enzymatic activity is defined as 1 nmole of product formed per min at 37". The specific activity is given as nanomoles of p-hydroxyphenylpyruvic acid forrned per min per mg of protein.
The aspartate aminotransferase reaction was followed by the procedure described by Banks et al. (16).
Protein Jrleasurements-Protein was measured by the biuret method (17), the method of Lowry et al. (18), the ultraviolet method of Warburg and Christian (19), and qualitatively by measuring absorbance at 280 nm.
Interaction of Tyrosine Aminotransfernse with Antibodies to Rat Liver Anionic and Cationic Aspartate Aminotransferase Isozymes-Antibodies to soluble (anionic) and mitochondrial (cationic) aspartate aminotransferase (kindly provided by Dr. H. Wada) were incubated for 60 min at 37" in various ratios with soluble aspartate aminotransferase or to our enzyme preparation from 1: 1 to 2000 : 1 and then the enzymes were assayed for the respective activities.
The preparations were allowed to stand for 24 hours at, 4" and assayed again.

RESULTS
Release of Yyrosine Aminotransferase jrom Rat Liver Mitochondria-In order to purify and characterize this mitochondrial enzyme, it had to be released from the mitochondria in a reproducible and quantitative but simple fashion wit,h minimal loss of activity.
Detergents such as sodium dodecyl sulfate and sodium deoxycholate were tried but did not improve the yield and in some cases resulted in lower activity.
Sonic treatment of fresh mitochondria resulted in about 70% release, but recovery varied considerably (40 to 80 "/c), with most of the remainder of the activity being recoverable in the pellet after two I-min treatments at  maximum output using a Bronson Sonifier with a probe attachment. Samples were simultaneously cooled by immersion in an ice bath with a l-min interval for cooling the probe and the sample separately.
Sonic treatment of mitochondria &ich had been frozen in the appropriate salt medium at a given level of dilution improved the yield if 6 ml per g but not 2 ml per g were used. Yields were lower in some cases presumably due to heat denaturation.
To test this, mitochondria were suspended in 0.25 M sucrose and aliquots maintained at various temperatures for 5 min with shaking.
Heating to 60" resulted in a loss of 80% of the activity, and after treatment at 70" no activity remained when assayed. The effects of salt media and dilution on the release of tyrosine aminotransferase from crude mitochondrial preparations are shown in Table I. Almost complete release was obtained when 6 volumes of medium were used per g of original liver.
Samples should be frozen at least 48 hours to obtain maximum results.
In water only 3374 of the activity was released while 66% was released by 0.05 M Tris-HCl, pH 8.4. Sucrose in the range tested yielded only about 30': release. Samples were checked with oxaloacetate to monitor ratio changes and the results paralleled those with ol-ketoglutarate.

PuriJication of Liver Mitochondrial
Tyrosine Aminotrcrnsferase-A general purification scheme is presented in Table II. The six  fractionation steps produced approximately a 900-fold purification with approximately a 250/, yield. When liver was homogenized and the homogenate assayed for activity and compared with recovery from subcellular fractions, 95% of the total activity was recovered in the fractions with about llyO being found in the nuclear fraction.
The activity found in the nuclear fraction could be almost completely "washed out" and had activity toward oxaloacetate similar to that of the mitochondrial enzyme. Thus the activity in the nuclear fraction is mostly due to the mitochondrial enzyme and approximately 10 to 15% of the enzyme was lost to this fraction during mass preparation of the enzyme. Normally little of the enzyme is lost to the cytosol fraction (less than 5s;lu). One third of the liver activity is mitochondrial. Initially there is enzyme activity with pyruvate. By using ratios of activity with the various keto acids it was determined that about 30% of the activity observed in mitochondria may be due to the enzyme which utilized pyruvate and may be a mito- 1. Sedimentation velocity of purified mitochondrial tyrosine aminotransferase.
A double sector filled epon centerpiece pyridoxal-P. The experiment was conducted at 20" and photowas used. The protein concentration was 12.4 mg per ml. The graphs were made at 16-min intervals after reaching a speed of 60,000 rpm. solvent was 0.1 M phosphate buffer, pH 7.5, containing 0.1 M The direction of sedimentation is from left to right and the photographs shown were taken at 0, 16,32,48 and 64 min. chondrial alanine aminotransferase.
Each of these factors was considered when the final yield was calculated.
The final specific activity was 0.18 (0.145 to 0.180) pmoles of p-hydroxyphenylpyruvate formed per min per mg of protein at 37" in 0.2 M potassium phosphate buffer, pH 8.1, with 9 mM tyrosine as amino donor.
Activity with pyruvate was reduced to less than 0.05% of activity with oc-ketoglutarate. Most of the pyruvate activity could be recovered between 50 and 60% saturation and most of the oxaloacetate activity between 60 and 90% saturation.
These results are consistent with the existence of two mitochondrial enzymes utilizing tyrosine as a substrate.
An additional study was made on the crude extract from mitochondria using the three keto acids as given in Table III. At 9 rnM tyrosine, negligible inhibition was observed with oxaloacetate at 50 mM final concentration. When pyruvate was incubated in the presence of one of the other two keto acids, an additive effect was seen when both keto acids were at saturating levels. Similar results were obtained with monoiodotyrosine instead of tyrosine. These data also suggest that perhaps 30% of the mitochondrial activity could be the result of an alanine aminotransferase. Properties of Puri$ed Enzyme-In the analytical ultracentrifuge the enzyme moved as a single component as can be seen in Fig. 1. Dependence of the sedimentation coefficient upon protein concentration is shown in Fig. 2 where it can be seen that the value of 4.5 x lo-l3 see-l is invariable over a substantial concentration range of the enzyme.
The diffusion coefficient has been measured at a protein concentration of 10 mg per ml as shown in Fig. 3. Using these values in the molecular weight equation (see "Experimental Procedure"), a molecular weight of 114,000 was calculated.
Sephadex G-100 chromatography yielded a symmetrical curve with respect to protein and activity and removed ampholines present after electrofocusing.
The presence of ampholines could be qualitively determined by adding calorimetric reagents used in the Briggs assay (12, 13) and observing the white precipitate that formed. The molecular weight for mitochondrial tyrosine aminotransferase as determined on Sephadex G-100 was approximately 90,000. The purified enzyme was subjected to isoelectric focusing on both pH 7 to 10 and pH 8 to 10 gradients and one peak occurred at pH 9.3 (Fig. 4). Similar results were observed at various steps in the purification process. The elution volume for the enzyme before and after isoelectrofocusing from Sephadex G-100 was essentially unchanged, suggesting that no major alteration in molecular weight had resulted from this step. Disc gel electrophoresis (11) in sodium dodecyl sulfate disc l-l-l-l-l-l-l-1-0 2000 4000 6000 8000 2 t (seconds) FIG. 3. Measurement of the diffusion coefficient of purified mitochondrial tyrosine aminotransferase at 10 mg per ml. The determination was made with a double sector artificial boundary centerpiece at 20" and 22,000 rpm giving a value of 3.7 X 1OV cm2 set-l.
gel-urea yielded two major bands (RF = 0.18 and 0.22) and one minor band which was barely visible (0.26). Tiselius free boundary electrophoresis at pH 7.6 in 0.1 M potassium phosphate yielded two closely traveling peaks. These data suggest that there may be binding of the enzyme to some substance which alters its mobility, that there is still some minor protein contaminant with a molecular weight very similar and a pHr of around 9.3 on isoelectrofocusing, or that fractionationof the enzyme into its subunit occurred.
The enzyme was found to elute from CaP04 gel over a wide range of buffer concentration.
In an attempt to find several enzymes the ratio of activity in fractions eluted with 0.05 M, 0.1 M, and 0.3 M were compared for cr-ketoglutarate and oxaloacetate.
No ratio difference occurred suggesting that the same proteinwas eluting from the gel. Repeated  washing of the CaP04 gel with 0.05 RI buffer would not remove all of the enzyme. The apparent Michaelis constants for the enzyme are as follows: tyrosine, 12 mM; a-ketoglutarate, 0.4 mM; and oxaloacetate, 2 InM.
When the purified enzyme was concentrated to about 10 mg of protein per ml or more, spontaneous crystallization occurred when dilute buffers were employed (0.01 M).
The crystals were usually shaped in the form of a triangle.
This "crystahization" may represent the condition where the protein is insoluble with low salt concentration.
This may explain why protein precipitated under certain conditions during dialysis. Precipitation during dialysis and crystallization may be identical processes and prolonged standing out of solution results in loss of activity such that it is difficult to resolubilize the protein.
Amino Acid Analysis of Mitochondrial Tyrosine Aminotransferase-Samples of the final preparation were analyzed for amino acid content.
The results are presented in Table IV. Substrate Specificity-The relative initial velocity of mitochondrial tyrosine aminotransferase with various amino acids was determined using 14C-ol-ketoglutaric acid as the amino group acceptor.
Activity of the enzyme was followed measuring 14Cglutamate formation.
The results are presented in Table V showing a comparison of each amino acid with tyrosine which has been adjusted arbitrarily equal to 1. Several amino acids, other than tyrosine, are good amino group donors.
The ratio data indicate that the optimum couple would be aspartate-glutamate or the reaction catalyzed by mitochondrial aspartate aminotransferase. Further evidence that mitochondrial aspartate aminotransferase might be catalyzing the reaction for mitochondrial tyrosine transamination was obtained with the addition of aspartate to the complete system for the tyrosine aminotransferase reaction since, as can be seen in Table VI, aspartate effectively blocks the conversion of tyrosinc to p-hydroxyphenylpyruvate. In contrast, aspartate has no effect under these conditions on the activity of the cytosol tyrosine aminotransferase.
EJect of Antibodies to Mitochondrial Aspartate Aminotransferase and to Soluble Aspartate Anainotransferase on Mitochondrial Tyrosine Aminotransjerase-Whereas preparations of mitochondrial tyrosine aminotransferase at various stages of purity would catalyze the conversion of tyrosine and aspartate, the cytosol tyrosine aminotransferase would not convert aspartate nor would the cytosol aspartate aminotransferase convert tyrosine to p-hydroxyphenylpyruvate.
It was of interest to know what effect antibodies to the aspartate aminotransferases would have on our preparations of mitochondrial tyrosine aminotransferase. These results arc presented in Table VII.
Activity of Mitochondrial Tyrosine Aminotransjerase with p-Hydroxyphenylpyruvate and Glutamate-The purified enzyme was tested with 0.1 mM p-hydroxyphenylpyruvate and 10 mM and 20 mM glutamate and aspartate.
The results obtained with glutamate are presented in Table VIII.
Aspartate was found to be about 33% as effective as glutamate under the same conditions. At nearly 0.01 the level of substrate (0.1 m&l p-hydroxyphenylpyruvate versus 9 mM tyrosine) the reaction rate u~as 18oj, of the rate with tyrosine when p-hgdroxyphenylpyruvate was ut,ilizctl.  Vol. 246,No. 10 to have a second enzyme which is not subject to large changes in activity (5, 6) and has an unfavorable apparent Michaelis-Menten constant for tyrosine of 12 ~RI. In this paper we have presented evidence suggesting that mitochondrial tyrosine aminotransferase activity is identical wit#h mitochondrial aspartate aminotransferase and also with alanine aminotransferase. These proteins could thus use various amino acids, including tyrosine, to regulate the important keto acids, oxaloacetate and a-ketoglutarate, for oxidative metabolism.
This could explain the role of these enzymes within liver mitochondria.
The existence of the mitochondrial form in other tissues such as brain (7), heart, and kidney (26) could aid in the conservation of tyrosine at the expense again of aspartate and glutamate which would then be available to the Krebs cycle. Recent studies with p-hydroxyphenylpyruvate i n viva support the concept that the level of blood tyrosine could, in part, be controlled in such a manner at least under certain conditions.' The association of mitochondrial tyrosine aminotransferase activity with mitochondrial aspartate aminotransferase suggests the need to investigate the role of mitochondrial aspartate aminotransferase in the metabolic control of these amino acids and keto acids.