Terminal Deoxynucleotidyltransferase of 60,000 Daltons from Mouse, Rat, and Calf Thymus PURIFICATION BY IMMUNOADSORBENT CHROMATOGRAPHY AND COMPARISON OF PEPTIDE STRUCTURES*

For rapid, simple purification of terminal deoxynu- cleotidyltransferase from limited amounts of tissues, we have developed an immunoadsorbent column chro- matographic method using antiterminal transferase antibody-conjugated Sepharose 4B. The column specif- ically adsorbed all mammalian terminal deoxynucleotidyltransferase (terminal transferases) tested and, in all cases, nearly homogeneous enzymes were recovered at extremely high yields of activity and protein. By this method, we first succeeded in purifying rodent enzymes from rat or mouse thymus, which enzymes were comprised of a single polypeptide chain (Mr = 60,000). The enzyme purified from calf thymus by the same procedure showed the two well known subunits (a: M, = 10,000 and 8: M, = 32,000). However, the calf preparation purified in the presence of protease inhibitors ex- hibited several polypeptides with molecular weights ranging from M, = 42,000 to M, = 60,000, but did not contain the two-subunit form. From peptide mapping analyses, it was evident that each of the high molecular weight polypeptides contained sequences of both of the two low molecular weight subunits. These results indicate

For rapid, simple purification of terminal deoxynucleotidyltransferase from limited amounts of tissues, we have developed an immunoadsorbent column chromatographic method using antiterminal transferase antibody-conjugated Sepharose 4B. The column specifically adsorbed all mammalian terminal deoxynucleotidyltransferase (terminal transferases) tested and, in all cases, nearly homogeneous enzymes were recovered at extremely high yields of activity and protein. By this method, we first succeeded in purifying rodent enzymes from rat or mouse thymus, which enzymes were comprised of a single polypeptide chain (Mr = 60,000). The enzyme purified from calf thymus by the same procedure showed the two well known subunits (a: M, = 10,000 and 8: M, = 32,000). However, the calf preparation purified in the presence of protease inhibitors exhibited several polypeptides with molecular weights ranging from M, = 42,000 to M, = 60,000, but did not contain the two-subunit form. From peptide mapping analyses, it was evident that each of the high molecular weight polypeptides contained sequences of both of the two low molecular weight subunits. These results indicate that the two subunits (a and p) of the calf thymus enzyme reported earlier may be proteolytic products derived from a single polypeptide of M, = 60,000, which may be the native form. It was noted that an extensive homology existed in primary structure of the enzymes from three species of mammals.
A unique DNA-polymerizing enzyme, terminal deoxynucleotidyltransferase (EC 2.7.7.31), is generally localized in mammalian thymus and bone marrow cells and abnormally in the circulating lymphocytes of leukemic patients (1-4). The homogeneous enzyme purified from calf thymus is well known to have two subunits, one of M , = 8,000-10,000 (the a subunit) and one of M, = 24,000-30,500 (the fi subunit) (5-lo), and a similar subunit structure has been observed in a purified enzyme from human leukemic cells (11). Recently, however, it has been reported that the human enzymes purified from lymphoblasts of leukemic patients are comprised of a single polypeptide chain with a molecular weight of 62,000 (8,12), while the enzyme from human thymus shows the subunit structure similar to that from calf thymus (8,9). By immu-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. noprecipitation assays of the enzyme labeled metabolically with radioactive amino acids, a similar single polypeptide with high molecular weight has been detected in both human and murine cultured lymphoblastoid cells and also in murine thymus and bone marrow lymphocytes (13, 14). However, it has not been established whether these differences in enzyme structures, as well as in their molecular weights, reflect species or tissue specificities of the enzyme or are due simply to a proteolytic degradation during the purification process.
In the present study, in order to clarify this issue, we developed a new, rapid purification method for terminal transferase' using highly specific immunoadsorbent chromatography. Since the anti-calf terminal transferase antibody used here cross-reacted extensively with all mammalian enzymes tested, the method was applicable to the purification of terminal transferase from various mammals, even though limited amounts of tissues, such as rodent thymuses, were available. Using this method, we demonstrated for the fiist time the existence of high molecular weight forms of terminal transferase from the thymus of calf, rat, and mouse. Furthermore, we have proved by tryptic peptide mapping analyses that the subunit structure of calf thymus enzyme reported in earlier studies is derived from a single polypeptide chain of M , = 60,000 by proteolysis during the purification process and also that the enzymes from the three mammalian species show a marked homology in the primary structure.

8745
were obtained from Miles-Yeda and P-L Biochemicals, respectively. DNA-cellulose (native form) and crystallized bovine serum albumin were from Wako Pure Chemicals. Silica gel 60-coated thin layer plates were from E. Merck. DNA polymerase a(6.5 S form, 4,000 units/mg of protein) and p(lO,000 units/mg of protein) were purified from calf thymus as described previously (15)(16)(17).
Enzyme Assays-The reaction mixture for determining terminal transferase activity contained, in a final volume of 125 pl, 0.2 M potassium cacodylate (pH 7.5), 1 mM 2-mercaptoethanol, 80 pg of BSA, 2.5 pg of oligo(dA)12-ls, 0.2 m~ [3H]dGTP (5,000 or 20,000 cpm/ nmol), 0.8 mM MnCL, and 5-30 pl of enzyme fraction. For routine assays, activated and heat-denatured calf thymus DNA (20 pg/assay) was also used as a primer (18). Enzyme assays for the DNA polymerase a and B and E. coli DNA polymerase I were performed with the activated calf thymus DNA as described previously (18). After incubation at 35 "C for 15 or 30 min, the acid-insoluble radioactivity was measured by the method of fdter paper discs (Whatman No. 3 " ) as described previously (19). One unit of enzyme activity is defined as the amount which catalyzes the incorporation of 1 nmol of total deoxynucleotides/h under the assay conditions described above.
Extraction and Prepurification of Terminal Transferase-Normal thymuses were obtained from 8-to 12-month-old calves, 2-monthold rats (Wistar-King A), and 2-month-old mice (C57BL/6J). A human thymus was from a patient with thymoma. Tissues were suspended in a minimum volume of PBS (20 m~ sodium phosphate, pH 7.2, and 130 m~ NaCl) and stored at -80 "C until use.
All subsequent operations were carried out at 4-6 "C. The frozen tissues (usually 100 g) were thawed, minced, and homogenized with a Waring Blendor in 4 volumes of Buffer A (20 mM potassium phosphate, pH 7.2,1 m~ 2-mercaptoethanol, and 1 mM EDTA). After centrifugation at 10,000 X g for 20 min, the supernatant (extract I) was collected through four layers of gauzes to remove fats. The nuclear pellets were homogenized in 4 volumes of Buffer A containing 250 mM KCl. After the centrifugation, the supernatant (extract 11) was saved and the residual pellets were homogenized again in 3 volumes of the same buffer. The supernatant (extract 111) was obtained after the centrifugation. By this procedure, terminal transferase was completely extracted (more than 98%). The extracts I, 11, and I11 were mixed and passed through a DEAE-cellulose column (2.5 X 30 cm, DE23 equilibrated with Buffer A containing 160 m~ KCl) to remove nucleic acids. The flow-through fractions were collected and dialyzed against 100 m~ KC1 in Buffer A. After clarifying by centrifugation at 24,000 X g for 30 min, the supernatant was adsorbed onto a phosphocellulose column (P-11,2.5 X 30 cm) previously equilibrated with Buffer A containing 100 m~ KCl. After washing with 120 mM KC1 in Buffer A, the column was eluted with 600 mM KC1 in Buffer A. Recovery of the terminal transferase activity from the column was more than 90% (see Table 11). All active fractions were pooled, dialyzed against 150 m~ KC1 in Buffer A, and the resulting protein aggregates were removed by centrifugation. These phosphocellulose fractions were used as enzyme sources for enzyme neutralization tests and immunoadsorbent chromatographies.
In the cases indicated, protease inhibitors (PMSF, bemamidine/ HC1, and ovomucoid) were added to all buffers for both extraction and chromatography, except that ovomucoid was omitted from the dialysis buffers (20). PMSF was dissolved in isopropanol at 100 mM and added to the buffers just prior to use (21).
Purification of Calf Thymus Terminal Transferase-The terminal transferase-rich fraction, corresponding to the extract I1 in Table 11, was prepared from 6 kg of calf thymus as mentioned above and initially passed through a DEAE-cellulose column (15 X 30 cm, DE11 equilibrated with Buffer A containing 200 mM KC1) to remove nucleic acids. The flow-through fractions were collected, diluted to 100 mM KC1 with Buffer A, and then adsorbed onto phosphocellulose (6 liters of P-11 previously equilibrated with Buffer A containing 100 mM KC11 by a batchwise procedure according to Bollum et al. (19). After extensive washing with the equilibration buffer, the resin was packed into a column (15 X 40 cm) and further washed with Buffer A containing 120 mM KC1. The enzyme was eluted with 500 m~ KC1 in Buffer A and dialyzed against 120 m~ KC1 in Buffer A. After clarifying by centrifugation, the supernatant was loaded onto a second phosphocellulose column (5 X 70 cm, P-11 equilibrated with the dialysis buffer) and eluted with a linear gradient of 120-800 mM KC1 in Buffer A. The highly active fractions were collected and dialyzed extensively against 40 mM KC1 in Buffer B (20 mM Tris-HC1, pH 8.1, 1 mM 2mercaptoethanol, and 0.2 m~ EDTA). The clarified fraction was adsorbed onto a DEAE-Sephadex A-50 column (5 X 50 cm) equilibrated previously with Buffer B containing 30 mM KC1 and developed with a linear gradient of 50-400 m~ KC1 in Buffer B. The peak fractions were collected as a 30-65% ammonium sulfate precipitable fraction and dialyzed against Buffer C (5 mM potassium phosphate, pH 7.2, 1 m~ 2-mercaptoethanol, 10% (v/v) glycerol, and 300 mM KC1). After the centrifugation, the supernatant (50 m l ) was loaded onto a Sephadex G-100 column (5 X 90 cm) equilibrated with Buffer C and eluted with Buffer C. The terminal transferase fractions were adsorbed immediately onto a hydroxylapatite column (2.5 X 20 cm) equilibrated with Buffer C and developed with a linear gradient of 5-300 mM potassium phosphate in Buffer C.
The enzyme fractions eluted at around 50 m~ potassium phosphate were dialyzed against Buffer B containing 10 m~ KC1 and 20% (v/v) glycerol and applied to a native DNA-cellulose column (0.9 X 20 cm) equilibrated with the dialysis buffer. The column was developed with a linear gradient of 10-250 mM KC1 in Buffer B containing 20% (v/v) glycerol. The active fractions appeared at around 70 mM KC1, were dialyzed against Buffer B containing 40 m~ KC1 and 20% (v/v) glycerol, and then adsorbed onto a column of DEAE-Sephadex A-50 (0.9 X 7 cm) equilibrated with the dialysis buffer. The terminal transferase, which was eluted as a single peak at 130 m~ KC1 using a linear gradient of 40-300 mM KC1 in Buffer B containing 20% (v/v) glycerol, was dialyzed against 20 mM potassium phosphate (pH 7.2), 100 m~ KCl, 1 mM 2-mercaptoethanol, 0.02 m~ EDTA, and 50% (v/v) glycerol and stored at -20 "C.
Thus, 17.3 mg of homogeneous terminal transferase was obtained from 6 kg of calf thymus. A sodium dodecyl sulfate-polyacrylamide gel electrophoretic pattern of this preparation is shown in lune A of Fig. 3. The specific activity of this preparation was 104,800 units/mg of protein with oligo(dA)la-ls.
Preparation of Antibody against Calf Thymus Terminal Transferase-Antiserum directed against the homogeneous terminal transferase was prepared in female Japanese white rabbits according to Bollum's method (221, except that cross-linking of the antigen was omitted. The antiserum (100 m l ) obtained from one rabbit was precipitated with ammonium sulfate at 45% saturation and the precipitate was dialyzed against 20 mM Tris/HsP04 (pH 8.0). After clarifying by centrifugation, the sample was passed through a DEAEcellulose (DE23, 2.5 X 30 cm) column equilibrated with the same buffer and the recovered fraction (IgG fraction) was dialyzed against PBS. An aliquot of the IgG fraction was adsorbed at a flow rate of 3 ml/h onto an affinity column (0.9 X 8 cm) of Sepharose 4B which had been conjugated with 3 mg of the homogeneous terminal transferase by the cyanogen bromide method (23) according to the Pharmacia manual. After extensive washing with PBS, the highly specific antibody was eluted from the column with 50 mM glycine/HCl (pH 2.8) containing 200 m~ NaC1, and the eluate was immediately neutralized with 1 M Tris-HC1 (pH 8.0). A total of 20 mg of the affinity-purified antibody was pooled from eight experiments. antibody (7 mg of IgG) was immobilized on Sepharose 4B beads (6 Preparation of Zmmunoadsorbent Column-The affinity-purified ml) by the cyanogen bromide method described above. The resin was packed into a double-ended column (Whatman IEC column, 1 X 8 cm). A column (1 x 9 cm) of normal rabbit IgG (40 mg)-conjugated Sepharose 4B was also prepared and used as a precolumn to remove nonspecific binding proteins and protein aggregates.
SDS-Polyucrylumide Gel Electrophoresis-SDS slab gel electrophoresis was carried out according to Laemmli (24) using 12.5% or 15% running gel (2 mm in thickness) with 4% stacking gel. Gels were stained with Coomassie brilliant blue R-250 and destained according to the original method (24). The molecular weights of the polypeptides observed were estimated from protein standards of known molecular Peptide Mupping-Radioiodination, digestion with trypsin or chymotrypsin, and separation of the peptides on thin layer plates were carried out according to the method described by Elder et ul. (25) with some modifications described previously (26,27). In brief, individual stained bands were cut out from a SDS-polyacrylamide slab gel and radioiodinated by the chloramine-T method. The gel slices were washed, dried, and then digested with 75 p g / d of trypsin or 75 pg/ml of chymotrypsin in 0.5 ml of 50 m~ NH~HCOR (pH 8.4) for 20 h at 37 "C. The resulting supernatant in each tube was lyophilized, dissolved in 20 pl of electrophoresis buffer (acetic acid/formic acid/ water, 15:5:80), and a 4-to 6-pl portion of this solution was spotted on a silica gel-coated thin layer plate (20 X 20 cm). The digested peptide was resolved by electrophoresis in the f i t dimension and ascending chromatography in the second dimension (27). The plates were dried and then exposed to Kodak X-Omat R film.
by the dye-binding assay devised by Bradford (28) using a Bio-Rad Protein Determinution-Protein concentrations were determined protein assay kit with bovine y-globulin as a standard.

RESULTS
Specificity of Antiterminal Transferase Antibody-The affinity-purified antibody described under "Experimental Procedures'' inhibited more than 90% of the activity of both homogeneous terminal transferase and partially purified enzyme from calf thymus (Table I). It also strongly inhibited a human enzyme, but inhibited rat and mouse enzymes only to lesser extents. However, the addition of the secondary antibody (anti-rabbit LgG sheep IgG) to the systems precipitated almost a l l of the remaining activities ( Table I). The results of double antibody experiments indicate that the antibody against a calf enzyme has a capacity to bind all of the enzymes from other mammalian species tested, whereas its efficiency t o neutralize the activity varies depending on the sources of enzymes.
In contrast, the antibody did not cross-react significantly with calf DNA polymerases a, b, and E. coli DNA polymerase I, tested by either single or double antibody assays ( Table I).
Purification of Terminal Transferase by Immunoadsorbent Chromatography-The phosphocellulose fractions described under "Experimental Procedures" were used as enzyme sources for the immunoadsorbent chromatography. All procedures were carried out at 4-6 "C (Fig. 1).
The enzyme fraction was applied to the precolumn of normal rabbit IgG-conjugated Sepharose 4B which had been directly connected with the bottom inlet of an immunoadsorbent column of Sepharose 4B conjugated with antiterminal transferase antibody ("Experimental Procedures"). The passthrough effluent from the precolumn flowed upward into the immunoadsorbent column at a flow rate of 3 ml/h. After washing the precolumn with 20 ml of Buffer I (20 m~ potassium phosphate, pH 7.2, 150 mM KC1, 1 mM 2-mercaptoethanol, 1 mM EDTA, 2 mM PMSF, 5 mM benzamidine/HCl, and 0.2 mg/ml of ovomucoid), two columns were separated and the immunoadsorbent column was washed by ascending flow with 30-50 ml of the same buffer at the same flow rate. The column was washed further with 30-50 ml each of Buffer I1 (20 mM potassium phosphate, pH 8.0) and Buffer 111 (20 mM potassium phosphate, pH 6.4). Both buffers contained 400 mM KC1, 1 mM 2-mercaptoethanol, 0.5 mM EDTA, 5% (v/v) glycerol, and 0.5 mM benzamidine/HCl. The termind transferase was eluted from the column with 15 ml of a pH gradient buffer (pH 4.0-3.2) containing 50 mM glycine/HCl, 200 mM KC1, 1 mM 2-mercaptoethanol, 0.5 mM EDTA, 0.5 mM benzamidine/ HC1, and 50% (v/v) glycerol (Buffer 1%'). The elution was carried out using descending flow at a flow rate of 6 ml/h, and 2-ml fractions were collected in test tubes containing 0.3 ml of neutralizing buffer (0.5 M potassium phosphate, pH 7.2, 1 mM 2-mercaptoethanol, 20% (v/v) glycerol, and 2 mM PMSF). The residual enzyme was eluted completely with Buffer V (pH 2.9) and then with Buffer VI (pH 2.3). Buffers V and VI had the same composition as Buffer IV except for the pH values. The column was neutralized immediately with Buffer I for recycling. The precolumn was also washed successively with Buffers 11, 111, and VI and then neutralized with Buffer I. An elution profie of calf thymus preparation from the immunoadsorbent chromatography is shown in Fig. 1. The total purification steps are summarized in Table 11.  (11). Then, 20 pl incubated for another 2 h. Then the mixture was centrifuged at 20,000 of BSA (1 mg/ml) was added to each tube and an aliquot (20 pl) of X g for 15 min and 20 pl of the supernatant was assayed. Activities the mixture was subjected to enzyme assays with activated DNA as were expressed as per cent of the control (I  " Enzyme activity was determined by polymerization of Mn*+:dGTP onto a synthetic primer, oligo(dA),?.,s.
The precolumn was effective in removing nonspecific binding proteins and protein aggregates but did not bind any terminal transferase. The maximum adsorption of the enzyme to the immunoadsorbent column was obtained at a flow rate less than 4 ml/h. At 3 ml/h, more than 95% of the enzyme was adsorbed and did not leak out during the washing under the above conditions ( Fig.  1). Using a homogeneous calf terminal transferase (600 pg), the conditions to elute the enzyme from the column were examined. At pH 2.9 (Buffer V), approximately 95% of the enzyme protein was eluted, but the recovery of enzyme activity was only 32% of the applied activity. As shown in Fig. 2, the calf enzyme was fairly stable in buffers of pH above 3.5, but was gradually inactivated at lower pH values. Therefore, the reduction in enzyme activity may be due to the partial inactivation of the enzyme by temporary exposure to the acidic conditions (below pH 3.5) during the elution. The maximum recovery of enzyme activity was achieved by using a pH gradient elution as described above. As shown in Table 11, the fraction eluted with a pH gradient from pH 4.0-pH 3.2 contained 64% of total eluted proteins and 83% of total recovered activity. This fraction showed the highest specific activity (approximately 72,700 units/mg), which was comparable to that (104,800 units/mg) of a homogeneous calf enzyme purified by the conventional method ("Experimental Procedures"). The pH 2.9 fraction contained 30% of the total protein but only 15% of the activity and showed relatively low specific activity. The pH 2.3 fraction contained negligible amounts of protein and activity. From five experiments, the overall yield of the enzyme activity ranged from 63-74% of the total activity in the initial extract. Usually, 1.35-1.70 mg of purified enzyme protein was obtained from 100 g of calf thymus. Rodent terminal transferase from 27 g of rat thymus and from 10 g of mouse thymus was purified by one-step elution with Buffer V (pH 2.9) to avoid loss of the enzyme protein.
Analysis of Terminal Transferase by SDS-Polyacrylamide Gel Electrophoresis-The homogeneous calf enzyme used as antigen showed two major polypeptide bands ( M , = 32,000 and 10,000) on a SDS-polyacrylamide gel (Fig. 3 , lane A). These two bands may correspond to the / 3 and a subunits (5), respectively. The molar ratio of p to a was estimated to be approximately 1:l by scanning densitometry of the stained gel at 595 nm. Furthermore, three minor bands of M,-= 28,000, M , = 42,000, and M , = 57,000 were also detected.
The immunoadsorbent-purified calf enzyme, prepared without protease inhibitors during the prepurification process, showed the same subunit bands (Fig. 3, lane R ) . A calf enzyme purified with low concentrations of protease inhibitors showed A fixed amount (10 units in 5 pl) of homogeneous calf terminal transferase was added to 85 p1 of various buffers used for the immunoadsorbent chromatography as described under "Results." All buffers were previously supplemented with BSA (0.5 mg/ml) to prevent the enzyme from being inactivated by the dilution. After preincubation at the pH values indicated (at 6 "C for 2 h), the enzyme solution was neutralized with 10 pl of 1 M potassium phosphate (pH 7.2) and then a 2 0 4 aliquot was subjected to the enzyme assay. The inset shows the enzyme stability in the pH 2.9 buffer for the times indicated. a distinctive band of M , = 42,000 (Fig. 3, lane C ) in addition to the a and p subunit bands. When the prepurification was done in the presence of high concentrations of inhibitors (Fig.   3, lane D), several new bands of higher molecular weights appeared, while the two bands corresponding to the a and p subunits were barely detectable. The major three bands were On the other hand, rat enzyme prepared with low concentrations of protease inhibitors was mainly composed of a single polypeptide corresponding to M , = 60,000 (Fig. 3, lane E). A mouse preparation purified in a similar way showed a distinctive band of M , = 60,000 and a minor one of M , = 42,000 (Fig.  3, lane F ).
Peptide Mapping Analysis-In order to determine possible relationships between these several polypeptides obtained by the immunoadsorbent column chromatography, each polypeptide was analyzed by tryptic peptide mapping.
At first, homogeneous terminal transferase (Fig. 3, lane A) from calf thymus purified by a modification of the method of Bollum et al. (19) ("Experimental Procedures") was analyzed. As shown in Fig. 4, A and B  The representative spots derived from the a subunit are indicated by arrows.
that these two subunits do not contain common sequences. In contrast, the mapping pattern of the M, = 42,000 polypeptide consisted of two groups of spots; one corresponds to those from the a subunit and another from the /3 subunit (Fig. 4 0 . To confirm this homology, the radioiodinated peptides of the a and /3 subunits were mixed after digestion and were mapped by the same procedure (Fig. 40). This map is indeed indistin-guishable from that of the M, = 42,000 protein (Fig. 4 0 ; additional or missing spots could not be detected. These results indicated clearly that the a and fl subunits of the calf enzyme had been derived from a single polypeptide of M , = 42,000. A faint band of M , = 28,000 was shown to be a partially degraded form of the /? subunit because its mapping pattern was identical with that of the /3 subunit with the exception that it lacked at least three spots. Furthermore, the polypeptide of M , = 57,000 was shown to be still another form of terminal transferase by this technique. T h e polypeptides of M , = 10,000, M, = 32,000, and M, = 42,000 in the immunoadsorbent-pured preparations (Fig. 3,  lanes B and C) showed mapping patterns which were identical with those of polypeptides of the corresponding molecular weights shown in Fig. 4, A-C. In the calf enzyme prepared using large amounts of protease inhibitors, the bands at M , = 60,000, M , = 57,000 and M , = 42,000 were prominent (Fig.  3, lane D). The (Fig. 3,  lanes E and F). The spots indicated by arrows ( E and F) are common to the calf enzyme (C).
shown.) It is puzzling that only one additional spot was present since the molecular weights of the former two are much higher (by at least 15,000) than that of the latter, based on the SDS gel. To clarify this point further, another digestion was performed with chymotrypsin. The mapping patterns differed completely from the tryptic peptide maps. However, the maps of the M , = 60,000, M , = 57,000, and M , = 42,000 proteins were quite similar and the difference among the three polypeptides was again only one spot, as shown in Fig. 5, B and D (the map of M, = 57,000 protein is not shown). The other minor components (Fig. 3, lane D)  rodent enzyme were common to those of calf enzyme (Fig. 5,   C, E , and F), while a few spots may be specific for each species (not marked).
Sephadex G-100 Column Chromatography-Gel filtration chromatography was performed in order to examine whether the polypeptide of M , = 60,000 observed on the SDS-polyacrylamide gel is enzymatically active. In the presence of high concentrations of protease inhibitors, the phosphocellulose fraction and the immunoadsorbent-purified enzyme were prepared and subjected to Sephadex G-100 column chromatography (Fig. 6A). Both showed similar elution profiles. The activity was eluted as heterogeneous peaks ranging from M , = 67,000 to M, = 48,000. In contrast, no activity was observed in the high molecular weight range with purified calf enzyme having the two-subunit structure (Fig. 6A) or with the crude enzyme prepared in the absence of protease inhibitors (data not shown). Furthermore, a rat enzyme prepared with the low concentrations of protease inhibitors showed essentially a single peak with a mean molecular weight of 67,000, while a mouse enzyme prepared in a similar way was eluted apparently as two peaks, a major peak with M, = 67,000 and a minor one with M, = 44,000 (Fig. 6B).
It should be noted that the molecular weights estimated from the gel filtration were somewhat larger than those expected, based on the results on the SDS gel. For example, purified calf thymus terminal transferase was eluted as a single peak at the position of M, = 48,000 (Fig. 6A) which is larger than the value (MI = 42,000) calculated from the molecular weights of two subunits (Fig. 3, lane A ) . These results indicate that the high molecular weight forms (M, = 60,000 or 57,000), as well as the smaller enzyme, have the terminal transferase activity and proteolytic cleavage of the large forms occurs rapidly during the prepurifkation.

DISCUSSION
The immunoadsorbent chromatography developed here has many advantages to conventional methods for purifying an enzyme from s m d amounts of tissue (about 10 g), such as rodent thymus or human clinical specimens. The method also has advantages in purifying total enzyme exhibiting apparent multiple forms (29,30). Furthermore, pure enzyme can be easily obtained in extremely high yield (Table 11) by this simple procedure.
By this method, we f i s t succeeded in purifying nearly homogeneous terminal transferase from rodent thymus, which enzyme is mainly comprised of a single polypeptide of M , = 60,000 (Fig. 3, lunes E and F). This value is in good agreement with that predicted by immunoprecipitation of metabolically labeled enzyme from cultured cells (13, 14). In contrast, calf thymus enzyme showed two distinctive subunits, the a and p polypeptides, although the purification was performed under similar conditions (Fig. 3, lane C). However, if high concentrations of protease inhibitors were used during the purification, the enzyme exhibited several polypeptides having higher molecular weights, including the highest one of M , = 60,000, while the two polypeptides corresponding to the a and /3 subunits almost disappeared ( Fig. 3, lane D). Peptide mapping analyses clearly indicate that the high molecular weight forms contain both sequences of the a and p subunits (Figs. 4 and 5). In all preparations (Fig. 3), we could not detect any trace of the M, = 79,000 protein reported previously (31).
These results suggest that the single polypeptide of M, = 60,000 must be the original form of bovine terminal transferase as well as rodent enzyme, and the two-subunit structure ( a and p subunits) of calf enzyme reported previously (5-11) may be a proteolytic product derived from the high molecular forms, via an intermediated form of M , = 42,000, during the purification process (see Figs. 4 and 5).
The degradation of the enzyme may occur rapidly during the prepurification, especially in the case of calf thymus extract (Fig. 6). Bovine hemoglobin, which was reported to be effective as a protease-binding protein in purifying RNA polymerase from Bacillus subtilis (32), was not effective in this case (data not shown). In view of these results, we cannot exclude the possibility of the existence of intermediate forms  (e.g. M , = 42,000 or 57,000) in calf thymus tissue.
Although large differences in molecular weights were observed among the polypeptides (M, = 42,000, M , = 57,000, and M, = 60,000) of calf enzyme (Fig. 3, lane D ) , only one additional spot was clearly detected on both tryptic and chymotryptic peptide maps (Fig. 5). There are at least two explanations for this discrepancy. The fragments of the larger polypeptides may include only a small number of amino acid residues capable of being labeled with ' ' ' I by the chloramine-T method and/or there are only few digestible sites for trypsin or chymotrypsin (25).
By the peptide mapping analysis, it was found that terminal transferase from bovine, rat, and mouse show extensive homology in their primary protein structures (Fig. 5). The excellent cross-reactivity of the antibody (Table I and  The number of terminal transferase molecules per calf thymocyte can be estimated as follows. One-hundred g of calf thymus should contain 1.69 mg of enzyme because the final yield of 1.49 mg corresponds to 88% of the total extracted enzyme as summarized in Table 11; the tissue consists of approximately 4 X 10" cells as estimated by the DNA content; approximately 65% of the cells have the enzyme (33); and the native molecular weight is 42,000 as shown in lane C of Fig. 4. A thymocyte, therefore, contains approximately 9.3 X lo4 molecules of the enzyme, a value which is close to that estimated by radioimmunoassay (6).
Terminal transferase has been speculated to be a somatic mutator in the generation of immunological diversity in the early stages of lymphocyte differentiation (1,34). We have recently proposed a model in which terminal transferase can help DNA polymerase a bypass certain base damages on the template DNA strand in vitro, resulting in insertion of mismatched bases (35). It may be interesting to examine whether the high molecular weight enzyme has a similar function in vitro and in vivo.