Deoxynucleotide-polymerizing Enzymes of Calf Thymus Gland

SUMMARY A simplified procedure for purification of terminal deoxy- nucleotidyl transferase is presented. The enzyme is shown to be homogeneous by equilibrium centrifugation and gel electrophoresis. The molecular weight of the enzyme is 32,360, with a V measured as 0.65 cc g-r. The homogeneous enzyme can be dissociated into two subunits, CY and @, by sodium dodecyl sulfate. Subunit molecular weights, esti- mated from gel electrophoresis, are 01 = 8,000 and fl = 26,500. The procedure for partial purification of a terminal deoxy- nucleotidyl transferase from the soluble protein fraction of calf thymus gland reported earlier from this laboratory (1) produces an enzyme that is free of degradative activity and has a useful specific activity. In the earlier work the terminal transferase was fractionated as a contaminant of DNA polymerase and the assay used for terminal transferase did not permit its detection in crude extracts. This report presents some attempts to estimate the amount of terminal trasferanse activity in crude extracts. These experi- ments indicate that our earlier purification procedure, although not specifically oriented toward terminal transferase purification, does provide a major fraction of this activity. We have therefore continued the fractionation and have devised a simple procedure for obtaining a terminal transferase preparation that is homo- geneous by equilibrium centrifugation and gel electrophoresis. Preparations of the homogeneous enzyme have a molecular weight of 32,360 and can be dissociated by sodium dodecyl sulfate into two subunits having molecular weights of 26,500 (/3 chain) and 8,000 (a chain).

From the Department of Biochemistry, University of Kentucky Medical Center, Lexington., Kentucky 40506 SUMMARY A simplified procedure for purification of terminal deoxynucleotidyl transferase is presented. The enzyme is shown to be homogeneous by equilibrium centrifugation and gel electrophoresis.
The molecular weight of the enzyme is 32,360, with a V measured as 0.65 cc g-r. The homogeneous enzyme can be dissociated into two subunits, CY and @, by sodium dodecyl sulfate.
The procedure for partial purification of a terminal deoxynucleotidyl transferase from the soluble protein fraction of calf thymus gland reported earlier from this laboratory (1) produces an enzyme that is free of degradative activity and has a useful specific activity.
In the earlier work the terminal transferase was fractionated as a contaminant of DNA polymerase and the assay used for terminal transferase did not permit its detection in crude extracts.
This report presents some attempts to estimate the amount of terminal trasferanse activity in crude extracts.
These experiments indicate that our earlier purification procedure, although not specifically oriented toward terminal transferase purification, does provide a major fraction of this activity.
We have therefore continued the fractionation and have devised a simple procedure for obtaining a terminal transferase preparation that is homogeneous by equilibrium centrifugation and gel electrophoresis. Preparations of the homogeneous enzyme have a molecular weight of 32,360 and can be dissociated by sodium dodecyl sulfate into two subunits having molecular weights of 26,500 (/3 chain) and 8,000 (a chain). RlETHODS AND MATERIALS Xubstrates-Deoxynucleoside triphosphates were synthesized by the phosphomorpholidate procedure described by Moffatt (2) and purified by chromatography at pH 4.7 on Dowex l-Cl with a linear NaCl gradient followed by rechromatography on Sephadex A-25.HC03 with an ammonium bicarbonate gradient at pH * This research was supported by Grant CA-08487 from the National Cancer Institute.
The previous paper in the series is Reference 5.
7.9. Radioactive dNTPs were purchased from Schwarz Bio-Research and diluted to the desired specific activity with nonradioactive dNTPs. Trideoxythymidylate was prepared by chemical polymerization of dTMP (3) and purified by chromatography on DEAE-cellulose at pH 4.7 (20 mrvr sodium acetate) with a NaCl gradient.
Salmon sperm DNA, pancreatic RNase, bovine serum albumin, and Escherichiu coli alkaline phosphatase were purchased from Worthington.
Bovine heart lactate dehydrogenase (Hq isozyme) was a generous gift from Dr. George Schwert.
All other reagents were commercial reagent grade.
Thymus Fractions-All attempts to estimate activity of terminal transferase in crude extracts were carried out on the soluble fraction prepared by extracting frozen thymus gland with four volumes of a solution that contains 40 mM NaCI, 40 mu potassium phosphate (pH 7.4), and 1 mM 2-mercaptoethanol in a Waring Blendor.
The homogenized sample was then centrifuged for 120 min at 27,000 rpm in the No. 30 rotor of the Spinco centrifuge. The supernatant fraction was used for assays both directly and after concentration by precipitation at 50% (NH&SO4 saturation. The concentrated solution was dialyzed against 0.1 M potassium phosphate, pH 7.5, before assay or gradient fractionation. A4l1 terminal transferase and DNA polymerase precipitated in the 50y0 fraction and no activity could be detected in the 50 to 80% (NH&SO4 fraction.
The 35 to 55% (NH&SO., fraction (Fraction IV) used for further purification was prepared exactly as previously described (I). It should be noted that this (NH&S04 fraction is not identical with the (NH&SO4 concentrate used for assessment of activity in crude extracts since Fraction IV has been treated with phosphocellulose and DEAE-cellulose prior to salt fractionation. The specific activities of the i4C-dNTP for terminal transferase assays were 2 to 4 X lo5 cpm per pmole for partially purified or purified enzyme fractions and 1.2 X lo6 cpm per pmole for assaying crude extracts. Progress of the reaction is followed by measuring the amount of product formed as a func-910 Deoxynucleotide-polymerizing Enzymes. V. Vol. 246,No. 4 tion of time, after work-up on filter paper disks (4). The amount of product formed was determined by scintillation counting of the disks in toluene containing 0.4% 2,5-bis[2-(5.tert-butylbenzoxazolyl)]thiophene.
Average rates, calculated from the time when a fixed amount of substrate is converted to product, are used for specific activity determinations (cf. Reference 5). Reaction rate is expressed in nanomoles of deoxynucleotide incorporated per hour. One unit of enzyme is equal to 1 nmole of dNTP incorporated per hour and specific activity of the enzyme is expressed as enzyme units per mg of protein.
Routine enzyme assays on column effluents or gradient fractions were carried out by addition of 50 ~1 of reaction mixture to 10 ~1 of enzyme solution in Disposo trays (Linbro Chemical Company, model IS-  and incubations were carried out at 35" for 30 min or 1 hour. At the end of the incubation the entire reaction mixture from each well in the tray was adsorbed onto a paper disk and the disk was processed and counted. DNA polymerase activity was determined by incorporation of radioactive dNTP into acid-insoluble material in the presence of denatured DNA as template as reported earlier (6). Gradient Centrifugation-Linear 5 to 20% (w/w) sucrose gradients or 10 to 30y0 (v/v) glycerol gradients were prepared as described by Britten and Roberts (7). The gradients were 0.1 M in potassium phosphate (pH 7.5 or 4.5) and contained 1 mM 2-mercaptoethanol.
The enzyme solutions were loaded onto the gradients with a valve type sample applicator at 6,000 rpm for 30 min in the SW5OL rotor in a Spinco L-2 ultracentrifuge. Centrifugation was carried out for 15 to 16 hours at 40,000 rpm at 5". At the end of centrifugation, 30 equal volume fractions were collected from the top of each gradient tube by displacement from the bottom with 40% sucrose or glycerol solution.
The column was washed with 7 volumes of 50 m&f sodium citrate in 0.1 M potassium phosphate (pH 7.5) followed by equilibration with 0.1 M potassium phosphate (pH 7.5) in 1 InM 2-mercaptoethanol.

Equilibrium
Centrifugation-Equilibrium ultracentrifugation analyses were performed according to the method of Yphantis (8) in a Spinco model E analytical ultracentrifuge equipped with interference optics.
The enzyme was diluted with 0.1 M KH2POI containing 1 mu 2-mercaptoethanol to give final concentrations of 0.045, 0.018, and 0.008%. A 12-mm threechannel Kel-F centerpiece permitting simultaneous observation of the three solutions was used.
The interference pattern was recorded on Kodak metallographic plates and the fringe displacements were measured with a Nikon two-coordinate comparator.
The natural logarithms of the corrected fringe displacement were plotted against the square of the corrected distance from the center of the rotation.
The slopes of the lines were determined by a least squares method and have been defined as half of the effective reduced molecular weight or u/2 (8). We are indebted to Dr. Carole Coffee for her aid in making the equilibrium runs. The determinations of the partial specific volume (2') of purified terminal transferase were carried out according to the method of Edelstein and Schachman (9 was then carried out as described above. To establish that equilibrium was reached, photographs were taken for each run at 16, 20, and 24 hours. Measurements of fringe displacements were made after 20 hours of centrifugation for the 507, D,O run and at 24 hours for the 80% D2180 run. The ratio of slopes in D20 or DslsO to Hz0 and the densities of the solutions are used in the calculation of 0 (8).
Gel Electrophoresis-Electrophoresis was carried out in an apparatus manufactured by Shandon Scientific Company with columns, 5 X 50 mm, of acrylamide gel. Purified enzyme preparations were run on acrylamide gels at pH 4.5 with the discontinuous system described by Reisfeld, Lewis, and Williams (10). Enzyme protein, stored at -20" in 0.1 M KH2POI, was diluted into p-alanine-acetic acid buffer and allowed to stand at 4" overnight.
Approximately 50 pg of enzyme protein in 50 ~1 of /3-alanine-acetic acid buffer were mixed with solid sucrose and applied directly over a 2.5u/h stacking gel. The separating gel was 7.5% acrylamide.
Electrophoresis was carried out for 1 hour at 5" at 5 ma per gel. Similar conditions were used for the urea gels after incubating the enzyme protein in 5 M urea overnight at 5". In the urea system all buffers contained 5 M urea and the concentration of acrylamide in the separating gel was 5%.
Sodium dodecyl sulfate gel electrophoresis was carried out according to the method of Shapiro, Viiiuela, and Maize1 (11). Enzyme protein (30 pg) was diluted into 0.01 M sodium phosphate (pH 7.2) containing 0.1% SDS1 and 1 y0 2-mercaptoethanol and incubated at 35" for 3 hours before being applied directly over the 7.5ci0 gel. Electrophoresis was carried out at 5 ma per gel for 2 hours at room temperature.
All gels were stained for protein with Amido black and destained with 7% acetic acid. Terminal transferase activity was assayed directly on gels by incubating the gel with two changes of a substrate mixture containing 0.2 M potassium cacodylate (pH 7.2), 1 InM dTTP, 1 InM CoC&, 1 mM 2-mercaptoethanol, and 0.01 mu d(pT)3 for a total of 3 hours at 35". The polymer formed on the gel was stained with 100 kg per ml of ethidium bromide in 0.1 M Tris-Cl (pH 8.0), 50 mM NaCl, and 1 mM EDTA (12).
Amino Acid Analysis-The analysis for amino acids was carried out by the general procedures outlined by Moore and Stein (13) on a Technicon automatic amino acid analyzer.
Samples of whole enzyme, or a and fi chains separated on Sephadex G-75 columns, 1 X 100 cm, in 1 y0 SDS, were hydrolyzed with constant boiling HCI for 20 and 70 hours before chromatographic analysis. Cysteine was estimated as cysteine sulfoxide in performic acidoxidized samples subjected to HCI hydrolysis.
Tryptophan was calculated by estimation of the tyrosine-tryptophan ratio from ultraviolet absorption measurements (14). We are indebted to Messrs. John Dorson and Clyde Erley for carrying out the amino acid analysis.  The gradient patterns are shown in Fig. 1, and the summations of activities from the gradients are listed in Columns 2 and 3 of  During this period a turbidity forms and about 50% of the starting protein is precipitated.
All of the DNA4 polgmerase activity is destroyed in this step. The precipitate is removed by centrifugation and the protein in the supernatant fluid (Fraction V,  rpm at 20" in the An-D rotor. Fractions from the Sephadex G-100 column (Fraction VII, Table II) (Fig. 3) The purification scheme is summarized in Table II. No activity is lost upon standing for 4 weeks at 4", but aggregation occurs (and some activity loss) upon freezing and thawing.
Protein aggregation (visible turbidity) also occurs when the pH of the solution is raised above 5, although the enzyme remains completely active over a pH range of 4 to 9. After 7 months of storage at -2O", approximately 60yc of the enzyme activity remains.
Equilibrium Centrifugation-Sedimentation to equilibrium in 0.1 M KHzPOl shows that only one component is present in the purified terminal transferase preparations (Fig. 4). When the natural logarithms of the fringe displacements are plotted against the square of the distance from the center of rotation, the slope of the line is defined as half of the effective reduced molecular weight (8). Disparity in the slopes of such plots for different initial protein concentrations provides a method of detecting heterogeneity if the measured displacements span the same concentration range.
The difference in slopes between lines obtained when the protein concentration is 0.045 and 0.008, and, in other independent runs, is less than 5y;. These small differences in the slopes are within the range of experimental error and thus the enzyme appears to be homogeneous under the conditions used. with a partial specific volume (0) of 0.74 is 43,900. This appears to be too high when compared to the values obtained from gel filtration on Sephadex (18) and sucrose gradient centrifugation and gel electrophoresis presented below. Preliminary amino acid analyses also lead to a calculated B of about 0.73. Since the enzyme may be a metalloprotein (5) and B for a metalloprotein may be significantly different from that estimated from amino acid composition, a preliminary determination of c for terminal transferase was carried out according to the micro-method of Edelstein and Schachman (9). Equilibrium runs in 50% DzO resulted in a ratio of slope D20 to slope Hzo equal to 0.94, indicating a rather low fi for terminal transferase. More accurate measurement of B in the low range is obtained from runs carried out in 80% Dz180. The ratio of slope &I80 to slope H20 equal to 0.711 gives a corresponding ti of 0.65 (Fig. 5). The molecular weight of terminal transferase calculated from equilibrium sedimentation with fi equal to 0.65 and a/2 = 3.19 (Fig. 4) is 32,360. Sedimentation Constant of Terminal Transferuse at pH 4.5 or pH 7.5-The molecular weight determination of terminal transferase by equilibrium centrifugation is carried out at pH 4.5. Since terminal transferase reactions are usually carried out at neutral pH, it is important to know whether the same molecular species of the enzyme exists at both acidic and neutral PH. The sedimentation constant of terminal transferase has been determined on sucrose gradients at pH 4.5 and 7.5 in 0.1 M potassium phosphate with bovine heart LDH-(HJ, pancreatic RNase, bovine serum albumin, and E. coli alkaline phosphatase as markers. Fig. 6 shows the results from sucrose gradient centrifuga- tion. The ~20,~ value is 3.65 at pH 7.5 and 3.70 at pH 4.5, indicating no gross difference in sedimentation rate of the enzyme activity at these two pH values. Gel Electrophoresk-at pH 4.5 the enzyme behaves as a single cationic protein band on 7.5% acrylamide gels as shown in Fig.  7A, and this band also corresponds to terminal transferase activity assayed directly on the gel (Fig. 7B). As indicated previously, there are some aggregation problems with this protein that we do not understand completely at this time, but enzyme samples stored at -20" in KH~POI, diluted into @-alanine-acetic acid buffer, and allowed to stand at 4" overnight before electrophoresis show one principal band of protein and corresponding activity.
Electrophoresis in 0.1 y0 SDS 7.5% acrylamide gels (11) shows two major bands (Fig. 7C). The rapidly migrating band is called cx and the slower one 6. By running appropriate molecular weight standards the subunits are estimated to have approximate molecular weights of 8,000 and 26,500, respectively. No activity can be detected on such gels, and 0.01 y0 SDS is sufficient to cause complete inhibition of terminal transferase activity in solution.
Neutral detergents, such as 1% Triton X-100, do not affect activity.
Gels run at pH 4.5 in 5 M urea show a diffuse band of protein at the origin of the separating gel and a fast moving band (Fig. 70).
We suspect that this is the result of subunit separation, with aggregation of the higher molecular weight peptide, but have not obtained direct proof of this yet. No enzyme activity can be shown in enzyme treated with 5 M urea or on 5 M urea gels.

915
Amino Acid Composition--Amino acid analysis was carried out on samples of homogeneous enzyme, and cy and /? subunit chains were isolated by gel filtration in l'ih SDS on Sephadex G-75 columns.
The limited supply of material did not permit extensive or high precision analysis, so the results presented in Table III must be considered as very preliminary, or perhaps even qualitative.
We note, however, that all amino acids are detected and the calculated ti is as expected for an "average" protein.
The measured fi (see above) is quite different. The most striking result of the amino acid analysis is the preponderance of acidic (66) over basic amino acids (47, including histidine).
Since terminal transferase behaves as a basic protein on electrophoresis and chromatography (p1 = 8.6), we believe that a large fraction of the acidic amino acids must be present as amides.

DISCUSSION
This paper should be considered an extension of the first paper in the series (1) since in that investigation the activity levels and purification of terminal transferase were treated as a side issue in the purification of DNA polymerase. In this investigation we ignore DNA polymerase, for the most part, and direct our attention to activity levels and purification of terminal transferase.
The level of terminal transferase activity in crude extracts of thymus gland cannot be accurately estimated with the usual assay with short oligonucleotide initiators.
The reason for this is not completely clear, but there are several possible interpretations of this fact. Since we are measuring the production of acid-insoluble material, endonucleases or exonucleases could materially reduce the apparent activity.
The other possibility concerns the form of terminal tra,nsferase in crude extracts, that is, whether it exists as a part of an enzyme complex or as an inhibitor-enzyme complex. For purposes of fractionation it is desirable to have some means of assessing total activity, masked and unmasked, in the crude extract,. We believe that the simple "fractionation assays" that have been carried out now provide us with this information.
The best current estimate is that there are approximately 200 units of terminal transferase per ml of crude extract (17.4 units per mg), and our guess is that the absolute level does not exceed twice that value because none of the fractionation steps in the new procedure leads to major increases in activity.
As mentioned in Footnote 3, another approach to the problem of accurate assessment of activity level in crude extracts uses modified assay conditions. This approach, with polymer initiators to form exonucleaseresistant product, provides a rough confirmation of the data presented here, but more detailed proof will be required to ensure that the activities measured are one and the same.
Proceeding on the assumption that the preliminary steps of our previous procedure (Steps I to IV in Reference 1) do provide a major fraction of the terminal transferase activity, we have now used this material (Fraction IV, the 35 to 55% (NH&SO4 fraction) to develop a procedure for obtaining a pure preparation of the enzyme. The pure enzyme is about 6 times as active as the material obtained from the earlier fractionation.
In the earlier work (I) sedimentation velocity runs on the terminal transferase fraction indicated that it was largely composed of protein having s20,W = 5.2. It was later found that the t.erminal transferase activity has sZo ,W = 3.6, and thus contamination by higher molecular weight proteins or protein aggregates must have contributed the major weight fraction to the partially purified terminal transferase preparations.
The starting point of the new purification was an examination of the pH stability curve for terminal transferase.
In carrying out this study it was found that terminal transferase was completely stable at 4" at pH 4.5 in 0.1 hf cacodylic acid, whereas DNA polymerase activity Ivas destroyed and a large fraction of the total protein (largely high molecular weight and aggregating material) \vas precipitated. After the acid treatment terminal transferase is precipitated at 70% (NH&SO4 saturation as compared to 55% in earlier fractions, indicating that it was probably fractionating as an aggregate in the earlier steps. Aggregation reactions of basic proteins are well known but rather poorly understood.
At pH 4.5 the enzyme behaves as a single molecular species in the ultracentrifuge.
After removal of the major aggregating contaminants, molecular sieving on Sephadex G-100 removes a small amount of higher and lower molecular weight material.
Chromatography on hydroxylapatite removes any trace contamination with exonuclease and provides a homogeneous enzyme. In our first hydroxylapatite column runs the column fractions lost activity rather rapidly upon standing at 4". This may have been the result of metal ion contamination since the enzyme is susceptible to inactivation by a number of divalent ions. JVe now find that a 50 m&r sodium citrate wash of t,he hydroxylapatite column greatly enhances the stability of the column fractions.
Finally, the enzyme is rather unstable in dilute (less than 100 pg of protein per ml) solution.
This means that column fractions should be concentrated as soon as possible. Column effluents (from citrate-washed hydroxylapatite columns) lose activity at a rate of about 15% per 24 hours. This loss may be prevented by addition of 400 pg per ml of bovine serum albumin.
Only limited amounts of the pure terminal transferase have been prepared up to the present time, and these are sufficient for the preliminary characterization. The enzyme is homogeneous by equilibrium centrifugation and gel electrophoresis. Perhaps the most unusual findings are the dissociation into two subunits and the measured fi of 0.65. The preliminary ammo acid composition is not unusual, but should be helpful in designing degradations for sequence determination.
The availability of a procedure for obtaining homogeneous enzyme provides the means for further study of mechanism and structure of the terminal deoxynucleotidyl transferase from calf thymus gland.