Mammalian Deoxyribonucleic Acid Ligase

An ATP-dependent DNA ligase, partly purified from calf thymus, forms a covalent enzyme-adenylate complex on incubation with [14C]ATP. The isolated complex joins DNA single strand breaks with 3’-hydroxy and 5’-phosphate termini, in the absence of ATP, and the adenylate residue is released during this reaction. On incubation of the ligaseadenylate complex with pyrophosphate, the radioactive moiety is released as ATP. These data indicate that the complex is a reaction intermediate, and that DNA ligase from mammalian cells acts by the same mechanism as microbial DNA ligases.

DNA ligase, partly purified from calf thymus, forms a covalent enzyme-adenylate complex on incubation with [14C]ATP.
The isolated complex joins DNA single strand breaks with 3'-hydroxy and 5'-phosphate termini, in the absence of ATP, and the adenylate residue is released during this reaction.
On incubation of the ligaseadenylate complex with pyrophosphate, the radioactive moiety is released as ATP.
These data indicate that the complex is a reaction intermediate, and that DNA ligase from mammalian cells acts by the same mechanism as microbial DNA ligases.
The XTXD-dependent DNA ligase of Escherichia coli and the ,kTP-dependent DNA ligase induced by phage T4 in E. coli have been studied extensively, as these enzymes appear to be involved in the replication and repair of DXA (1). Both enzymes act by first forming a covalent enzyme-adenylate complex that, subsequently joins single strand breaks in DNA (2)(3)(4)(5), with the int.ermittent formation of a DNALadenylate complex (3). An ATP-dependent DNA ligase has also been found in extracts of cells from higher organisms (6-12).
While the ligase from mammalian cells has been purified in several laboratories (6, 10,12), active reaction intermediates have not been isolated. Attempts to do this have been hampered by the lability of the ligase preparations from some sources of mammalian cells (6), and by the small amounts of ligase activity found in higher organisms in comparison with the microbial systems. Thus, the ATPdependent DNA ligase activity in a cell extract from "1'4.infected E. coli is 500 to 1000 times higher than that in the most active ext.racts from mammalian tissues (6) or plant cells (8). The reaction mechanism of the mammalian DNA ligase has therefore remained somewhat unclear, in spite of the previous observation of an ATP This enzyme preparation (20 mg of protein in 2 ml) was considerably more stable than the DNA ligase preparations previously made from rabbit tissues (6, 13), and retained 50% of its activity after 3 weeks at 0". The enzyme could be preserved for long time periods (6 months) without loss of activity by rapidly freezing the protein pellet obt,ained after the final ammonium sulfate precipitation, followed by storage at -70". The frozen enzyme was thawed, dissolved iu cold buffer, and dialyzed as above before use.
Nucleic Acids-Nollradioactivc a~ltl [l'C]thymidine labeled l)NA (25,000 cpm l)er pg) from phage Ti were prepared by the method of Richardson (14). Single atralld breaks with 5'-phosphate termini w\'crc introduced into the 1 )NA1 to produce 0 to 4% acid-soluble material by incub&tion of the l)NA (200 pg per ml) for 10 min at 25" with pancreatic DNase I (0.05 to 1 pg per ml) in 0.05 M Tris-HC1, 0.02 M MgCl2, pH '7.4. The reaction was stopped by addition of EDTA to 0.05 M, heating at 55" for 5 min, and dialysis at 2" against 0.05 M Tris-HCl, lOA M EDTA, pH 7.4. I)enatured nicked DNA was obtained by heating this mat'erial at 100" for 10 min, followed by rapid cooling. DNA strand breaks with 5'-OH termini were introduced by treatment with micrococcal nuclease (0.05 pg per ml) for 20 min at 25' in 0.02 M N-2-hydroxyethylpiperazine-N'Qethanesulfonic acid-KOH, 0.006 M CaCl*, pH 8.5, to produce 4y0 acid-soluble material. The react.ion was stopped by addition of EDT-4 and dialysis as above. Pancreatic DNase (1 x tryst.), micrococcal nuclease, and alkaline phosphatase were obtained from Worthington, and crystalline hexokinase from Boehringer.
Assays-DNA ligase was measured by the method of Weiss et al. (15) as previously described (13), except, that incubations mere for 75 min at, 20". With an excess of enzyme, a a-fold higher amount of Dhe single strand breaks in the extensively nicked ligase substrate could be repaired at 20" t,han at 37". One enzyme unit converts 1 nmole of 32P to an alkaline phosphataseresistant form under the standard assay condibions.
ATP, ADP, AMP, and adenosine and adenine were separated by paper chromatography on Whatman No. 3?vIM paper in iso-but& acid, HzO, concentrated K&OH, 0.2 RI El)TA (66:33: For determinations of acid-soluble material in nucleuse-treated, nonradioactive DNA preparations, aliquots of the DNA solutions were incubated with an equal volume of 0.8 Y perchloric acid for 5 min at O", followed by centrifugation at 10,000 X g for 10 min. The supernatant solutions were recovered, and the concentration of acid-soluble oligonucleotides determined spectrophotometritally. Radioactive DNA solutions were precipitated in the same fashion together with 0.1 mg per ml of denatured calf thymus DNA as carrier, and after centrifugation the radioactivity of an aliquot of the supernatant solution was determined in a scintillation counter in the presence of 5 ml of "Aquasol" (NEN Chemicals) . In experiments with the radioactive enzyme-adenylate complex acid-solubility was measured in the same fashion, except that 0.5 mg per ml of bovine serum albumin was employed as carrier instead of DNA.

Isolation of a Protein-Adenylate
Complex-The partly purified calf thymus DNA ligase was incubated with [14C]ATP, and then chromatographed at Sephadex G-100 (Fig. 1). One percent of the applied radioactivity appeared in the protein fraction. This fraction was concentrated by the addition of (NH&S04 to 80% saturation, followed by centrifugation and freezing of the protein pellet in several aliquots.
The yield was 4 mg of protein, COKItaining 45,000 cpm '"C and 10 ligase units, as measured in the standard assay. In this frozen form, the complex could be preserved with undiminished biological activity for >6 months at -70". For further experiments the material was thawed, dissolved in 0.3 M NaCl, 0.05 M Tris-HCl, IO-3 M EDTA, 10e3 M dithiothreitol, pH 7.4, and dialyzed against the same solvent for 5 to 10 hours before use. Ninety-two per cent of the 14C was present in an acid-insoluble form.
The experiment described in Fig. 1 was also performed with the addition of both ['aC]ATP (5 X lo6 cpm) and [y-32P]ATP (5 X 106 cpm) to the reaction mixture.
In the protein fraction, 1.0% of the 1% and 0.2yc of the 32P were then recovered.
The 32P-labeled material was acid-insoluble (85 to 90y0), but was released into an acid-soluble form by treatment with alkaline phosphatase (15 pg per ml) for 10 min at 65". In contrast to the '4C-labeled material, no detectable amount of 32P ( <5oj,) was released by exposure to nicked DNA (see below, Fig. 2). The nature of the asp-labeled material was therefore not further studied; presumably it was due to contamination of the DNA ligase by protein kinase activity.
The maximum amount of free [Y-~~P]ATP that could have been present in the protein fraction directly after gel filtration was estimated to be lo-* M from these data. The K, for ATP in the standard ligase assay was sepa-rately determined to be 8 x lo-' M for the calf thymus DNA ligase.
Incubation of an aliquot of the 14C-labeled protein in 0.1 M HCI for 4 min at 100" converted the radioactive moiety (>857J to an acid-soluble form. This materia,l was identified as AMP (80 % yield). A minor proportion (20 %) of the material was also recovered as adenine due to acid-catalyzed depurinat,ion. Heating of free [l (C]ATP under identical conditions caused <1570 conversion to AMP. Xtability of Compler-The [14C]AMP residue remained acidinsoluble (.<lO% solubilization) after incubation for 10 min in 0.1 M HCl or 60 min in 0.1 M NaOH at 25". This stability of the complex indicates strongly that the adenylate residue was covalently bound to a macromolecule. Incubation of the complex for 10 min at 100" in 0.1 M HCl or 1 M NaOH caused t,he l*C residue to be released (>857J in an acid-soluble form, while treatment with 0.1 M HCl for 30 min at 25' or with 0.05 M Tris-HCl, 10Va M EDTA, pH 7.4 for 10 min at 100" caused 30 to 35% conversion to acid-solubility. Further, incubation with 3.1 M hydroxylamine at pH 4.75 as described by Gumport and IJehman (16) caused 67y0 and 85a/, of the label to be released in 15 and 30 min, respectively, while incubation with 0.2 M hydroxylamine at pH 7 (16) caused <20% release. These results a,re similar to data previously reported for the enzyme-adenylate complexes of microbial ligases (2)(3)(4)16), and suggest that the adenylat,e residue is bound to protein by a phosphoamide bond (16).
Enzymatic ActivityL of Complex-The protein-adenylate complex, which contained 9270 acid-insoluble radioactive material, was incubated at 0" with DNA-containing single strand breaks (Fig. 2). Most of the radioactivity (65%) was rapidly released in an acid-soluble form when 22 pg per ml of DN-4, pretreated with pancreatic DNase to produce single strand breaks with 5'phosphate termini, was added to the incubation mixture. In contrast, DNA without single strand breaks, or with single strand breaks containing 5'-hydroxy tern&i, showed little or no ability (<5% of that of nicked DNA with S/-phosphate termini) to promote the release of r4Clabel from the complex. Further, if the DNA-containing nicks with 5'-phosphate termini was heat-denatured before use, or if MgClz was not included in the reaction mixture, no release of 1% was detected. These data demonstrate that at least 6570 of the isolated protein-AMP complex is accounted for as an enzyme that interacts with single strand breaks in bihelical DNA, and the data below suggest that this activity is due to DNA ligase.
The acid-soluble material released by DNA ligase in the reaction with nicked DNA was identified as free AMP (>90%). When the enzyme-adenylate complex was incubated under the same conditions with lop4 M pyrophosphate instead of nicked DNA, the 14C label was released to >90% in an acid-soluble form, and this material was identified as ATP (95y0 yield; <3oj, of either ADP, AMP, adenosine, or adenine). These data show that the formation of an enzyme-adenylate complex that occurs on incubation of mammalian DNA ligase with ATP is a reversible reaction, as expected from our previous results on ligase-catalyzed pyrophosphate exchange (6).
When the ligase-AMP complex was studied in the standard ligase assay, performed in the absence of added ATP, conversion of the 5'-a2P residues to an alkaline phosphatase-resistant form could be demonstrated to the extent of 17 pmoles per mg of protein. DNA ligase, not preincubated with [14C]ATP, caused no detectable repair under these conditions (see below, Fig. 3). For each 32P residue that was converted to a phosphatase-resistant form, the release of 0.94 [14C]AMP residues from the com-  (Table I). It, thus, appears that 1 AMP residue was released per repairable single strand interruption. Five picomoles of 32P in the substrate were repaired, when the maximum amount of ATP that could have been present as a, contaminant in the preparation of the complex was 1 pmole. Thus, at least 80% of the DNA joining catalyzed by the enzyme-adenylate complex must have occurred without ATP consumption. On incubation of the enzyme-adenylate complex (100 fig per ml) with nicked [14C]T7 DNA (2 pg per ml, < 1 70 acid-soluble) for 10 min at 0" under the solvent conditions given in Fig. 2, followed by phenol extraction and dialysis, it was observed by alkaline sucrose gradient centrifugation that an increase in the sedimentation coefficient of the DNA strands from 24.5 S ( <5oj, of the DNA sedimenting as intact strands of T7 DNA) to 33 S (50% of the DNA sedimenting as intact strands of T7 DNA) had been obtained.

ATP-independent DNA Ligase
Activity-When the calf thymus DNA ligase was prepared as described, except that the incubation with pyrophosphate before the final gel filtration step was omitted, 1.6% of the ligase activity in the standard assay was independent of ATP (Fig. 3). However, after incubation with pyrophosphate, no such ATP-independent activity could be detected (<0.2%).
In order to control if the residual activity could be due to ATP contamination, the ligase substrate and the enzyme were separately preincubated in the assay buffer with crystalline hexokinase (1 pg per ml) and glucose (0.1 M) for 10 min at 209 before mixing.
This preincubation did not reduce the amount of ATP-independent ligase activity. It is therefore concluded that the DNA ligase preparation contained a small proportion of the enzyme in the form of a ligase-AMP complex.

DISCUSSION
The reported properties of mammalian DNA ligase demonstrate the great similarity between this enzyme and the wellcharacterized DNA ligases from E. coli and phage-infected E.
coli. These enzymes form a covalent complex of similar chemical stability with AMP in the presence of cofactor, and the adenylate moiety of the complex is discharged in a Mg+f requiring reaction during the repair of double stranded DNA with single strand breaks having 5'-phosphate and 3'.hydroxy termini. Consequently, it seems fairly certain that all three enzymes act by the same mechanism.
In agreement with this notion, a DNA-adenylate complex has been isolated in low yield (8%) after incubation of the mammalian DNA ligase-adenylate complex with nicked DNA at pH 6.5 for 30 s at O", and the adenylate residue of this DNA-AMP complex is released on further incubation with DNA 1igase.l A requirement in the assay of mammalian DNA ligase for protein factor(s) present in boiled crude cell extracts has been reported by Spadari et al. (10) for a ligase partly purified from human cells. With t.he calf thymus enzyme preparations employed here, we have been unable to detect any dependence of 1 S. 66derhSil1, in preparation. this type.2 The ligases from these two sources have been purified to approximately the same extent, but not by identical techniques. At present, there is thus, no indication of an involvement of a second protein in the reaction between calf thymus DNA ligase and nicked DNA.
In DNA ligase preparations from E. coli (a), T4-infected E. coli (4) and calf thymus, a minor proportion of cofactor-independent activity has been detected.
Moreover, cofactor-independent DNA ligase activity has been found in disrupted virions of Rous sarcoma virus (17). In the case of the T4 and mammalian enzyme, that require ATP, the cofactor-independent activity is abolished by preincubation of the enzyme with pyrophosphate.
In the present work, EDTA was included in the buffers used during the initial purification of the enzyme, in an attempt to prevent complex formation in vitro.
However, it cannot be decided from these data if a ligase-adenylate complex was in fact present in viva, or if a small amount of complex was synthesized during the disruption and extraction of the cells.