LINKAGE OF POLYNUCLEOTIDES THROUGH PHOSPHODIESTER BONDS BY AN FROM ESCHERICHIA COLI* BY BALDOMERO M\I. OLIVERAt AND I. R. LEHMAN

DEPARTMENT OF BIOCHEMISTRY, STANFORD UNIVERSITY SCHOOL OF MEDICINE, PALO ALTO, CALIFORNIA Communicated by Arthur Kornberg, March 20, 1967 There is now convincing evidence that recombinants are formed in genetic crosses by the breakage and joining of DNA molecules.' It is also likely that a similar breakage and rejoining occurs during "dark" repair of UV-irradiated DNA.2 have been described which are, in principle, capable of catalyzing the specific breakage of DNA molecules.3 Recently, Gellert4 described an activity in extracts of E. coli which converted "hydrogen-bonded circles" of phage X DNA5 to a "covalently circular" form,6' I a reaction which may be analogous to the joining of DNA molecules. In this paper we describe the purification and some of the properties of an enzyme from E. coli that joins preformed polynucleotides in phosphodiester linkage. It does so only under the specific condition that the polynucleotides to be joined be part of a double-stranded structure. The purified enzyme requires for its action a diva- lent cation (1\Ig++ or Ca++) and a "," as yet unidentified. The cofactor is heat-stable; its activity is lost after treatment with Norit, and it cannot be re- placed by adenosine 5'-triphosphate (ATP) or any other known nucleoside tri- phosphate. The purified enzyme will also catalyze the formation of covalent circles from hydrogen-bonded circles of X DNA and would therefore appear to be similar, if not identical, to the activity described by Gellert. The substrate used in most of the studies to be described is a double-stranded homopolymer pair consisting of multiple dT8 units of 150-200 residues, hydrogen- bonded to a dA chain of approximately 3000 residues. The dT chains are labeled with p32 5'-phosphoryl termini so that the reaction is conveniently measured by the conversion of the labeled monoester phosphate to a form which is not susceptible to alkaline . The reaction may be formulated as shown in Figure 1. Although the relationship of this enzyme, which we shall for convenience desig- nate as "polynucleotide-joining enzyme," to the act of genetic recombination is presently unknown, the availability, in reasonably purified form, of an enzyme that can link polynucleotides by a standard 3'5'-phosphodiester bond now makes it possible to explore recombinational and related phenomena at the enzymatic level. Experimental Procedure.-Materials: Unlabeled nucleotides and salmon sperm DNA were purchased from CalBiochem. HI-labeled phage X DNA (linear form) was isolated according to the method of Doherty and Hogness.9 It was converted to hydrogen-bonded circles as described by Wang and Davidson.10 3'-Deoxythymidine monophosphate (3'-dTMP) was synthesized according to Turner and Khorana.1 _yP32 ATP was prepared by the method of Glynn and Chappei1P2 as modified by Wu and Kaiser."3 Polynoucleotide kiniase was isolated as described by Richardson ;14 venom 5'- was purified by the method of Lehman, Roussos, and Pratt;'5 DNA polymerase from E. coli was plrepared according to Richardson et al.16 and micrococcal nulclease'7 was the gift of Dr. C. A. 1)ekker. The T4 phage-induced 3'-nucleotidase was isolated according to Becker and Ilurwitz.8 B. subtilis was prepared as described by Kerr, Chien, atid Lehman,19 and E. coli exo- nruclease I was purified by the method of Lehimami aiid Nimssbaum.20 E. coli alkaline phosl)hattse" was purchased fromi Worthington Biochemical Co. 1426 Downloaded by guest on September 29, 2021 VOL. 57, 1967 BIOCHEMISTRIYI: OLIVERA I ND LEHAJLAN 1427 5'

P {P{P{'P{'P{PNV1~|Joinin{gAenz{{m' A A A A A A A A Cofcactor A A A A A A A A TT T T T T M ++ T T T T T T

FIG. 1.-Synthesis of dT product b)y polylnucleotide-joining enzyme.

Alethods: Polynucleotide concentrations are expressed as equivalents of nucleotide phosphoruis unless otherwise indicated; they were determined by measuring the absorbance at 260 mA and applying the extinction coefficients given by Riley, Mlaling, and Chamberlin.22 (a) Preparation of homopolymer substrates: The homopolymers dA and dT (mol wt, approxi- mately 106) were prepared as described by Riley et al.;22 the dT was degraded to smaller units by treatment with . The progress of by the nuclease was measured by treating small aliquots with an excess of polynucleotide kinase and yp32 ATP to determine the number of 5'-hydroxyl dT termini formed. The reaction was terminated when the average chain length was approximately 250 nucleotide residues. The 3'-phosphoryl termini of the dT were removed by treatment with bacterial (10 units, 30 min at 370 for 0.3 /Lmole of dT). The phosphatase was inactivated by adjusting the pH to 10.5 with NaOH and heating in a boiling water bath for 30 min. The reaction mixture was brought to a pH of approxi- mately 8.5 by the addition of 1 M Tris-HCl, pH 7.5, then dialyzed against 0.01 M Tris-HCl, pH 7.5. The 5'-termini were labeled with p32 in a reaction with polynucleotide kinase as described by Richardson.14 The reaction mixture (5 ml) containing approximately 1 X 10-4 M dT (5 X 10-7 M in termini) was incubated at 370; 10-unit portions of polynucleotide kinase were added at 15-min intervals and the reaction was followed by determining acid-precipitable p32 until a limit was reached at between 60 and 90 min.23 The reaction mixture was dialyzed repeatedly against 1-liter changes of 1 M NaCl, containing 0.1 M Tris-HCl, pH 8.0, and 0.001 M ethylene- diaminetetraacetate (EDTA) until the dialysate contained fewer than 300 cpm per ml; this was followed by dialysis against two changes of 0.01 M Tris-HCl buffer, pH 8.0. The a' p32 dT was then centrifuged in an alkaline CsCl gradient, as described by Riley et al.,22 to separate it from the used in its preparation. After centrifugation, the dT solution was adjusted to pH 8.0-8.5 with 1 M Tris-HCl buffer, pH 7.5, and dialyzed against three 1-liter changes of 1 Al NaCl, containing 0.1 M Tris-HCl, pH 8.0, and 0.001 M EDTA, and three changes of 0.001 M Tris, pH 8.0. The 51_p32 dT prepared in this way had an average chain length of approximately 150 residues and a specific activity of 3-5 X 103 cpm per MSmole of 5'.p32 dT termini. (b) Assay of polynucleotide-joining enzyme: The standard assay for the polynucleotide-joining enzyme measures the conversion of the p32 5'-phosphoryl group of dT to a form insusceptible to the action of bacterial alkaline phosphatase. The incubation mixture (0.10 ml) contained 0.01 M Tris-HCl, pH 8.0, 0.002 M MgCl2, 0.001 M EDTA, 1.6 X 10 6 M dA and 1.6 X 10 6 M 5'- p32 dT (10-8 M in termini), 0.003-0.015 unit of enzyme and 2 ,A of boiled extract (20 mg protein/ ml; A260 = 389, prepared by heating a crude extract of E. coli (see below) for 5 min at 1000 then quickly chilling). After incubation of 30 min at 300, the reaction was terminated by boiling for 2 min. Five units of bacterial alkaline phosphatase24 were added and after thorough mixing, incubated for 15 min at 85°.25 The reaction mixture was chilled in ice and 0.2 ml of 1 N HCl, containing 0.1 AI Pi; 0.2 ml of 1 N HCl, containing 0.1 M PPi, 0.1 ml of bovine plasma albumin (1 mg/ml), and 0.2 ml of an acid-washed Norit suspension (20% packed volume) were added. The mixture was shaken, then filtered on a GF/C, 2.4-cm Whatman glass filter. The Norit was washed three times with 8 ml of 1 N HCl containing 0.1 A! Pi, and five times with 8 ml of 1 N IICI. The filter was dried and counted using a Nuclear-Chicago model 186 gas-flow counter equipped with a Micromil window. In the absence of enzyme, less than 0.5%7, of the input P32 remained Norit-adsorbable after phosphatase treatment. One unit of enzyme is defined as the amount catalyzing the transforma- tion of 1 MAmole of p32 5'-phosphoryl terminus of dT to a form insusceptible to bacterial alkaline phosphatase in 1 mill. The reaction rate was pioportional to the amount of enzyme added iii the range between 3 X 10-3 amd 15 X 1()-3 ,,,it. Downloaded by guest on September 29, 2021 1428 BIOCHEMISTRY: OLIVERA AND LEHMANV PROC. N. A. S.

(c) Other methods: Protein concentrations were determined by the method of Lowry et al.26 In a few instances where concentrations were below 10 jg/ml, estimates were made by reading the absorbance at 280 mwu.2 Nucleotides were separated by paper electrophoresis at 40 in 0.02 Al sodium citrate buffer, pH 3.5, at a potential of 5000 v for 30 min. After electrophoresis, the paper was dried and the nucleotides were visualized under ultraviolet light. The paper was then cut into strips and the radioactivity determined in the gas-flow counter. When necessary, the nucleotides were eluted from the strips by soaking them in distilled water overnight at room temperature. H3-labeled samples were counted with a toluene-base scintillator using a Nuclear- Chicago model 724 liquid scintillation counter. (d) Purification of polynucleotide-joining enzyme: The source of enzyme was a strain (1100) of E. coli deficient in I'5 which was very generously provided by Professor H. Hoffman- Berling. Because of the extreme susceptibility of the dA:dT substrate to endonuclease I, the polynucleotide-joining enzyme was barely detectable in extracts of wild-type E. coli. All operations were performed at 0-4°. Centrifugations were for 20 min at 12,000 X g in the Servall model 1332 centrifuge. Unless otherwise specified, buffers contained 0.001 M EDTA and 0.001 M glutathione. (i) Preparation of crude extract: A 60-liter culture of E. coli strain 1100 was grown in yeast extract-phosphate-glucose medium16 to an absorbance at 595 mAi of 0.7-1.0 (approximately 5 X 108 cells/ml). The cells were collected by centrifugation and stored at - 200 for use when needed. Frozen cells (40 gm) were suspended in 100 ml of 0.1 M glycylglycine, pH 7.0, and disrupted by sonic irradiation (Branson Sonifier model S75). Cell debris was removed by centrifugation. The supernatant fluid (94 ml) was collected and brought to a concentration of 19 mg protein per ml by the additioh of 0.1 M glycylglycine, pH 7.0 (fraction I, 172 ml). (ii) Streptomycin precipitation: To 171 ml of fraction I were added, with stirring, 34 ml of a freshly prepared 5% streptomycin sulfate solution (Merck and Co.). After 20 min at 00, the suspension was centrifuged and the precipitate discarded. The supernatant fluid (188 ml) was diluted threefold with 376 ml of distilled water and an additional 188 ml of 5%7 streptomycin sulfate were added with stirring. The mixture was again allowed to stand for 20 min at 00, then centrifuged. The supernatant fluid (fraction II, 735 ml) was immediately subjected to ammonium sulfate fractionation. (iii) Ammonium sulfate fractionation: To 735 ml of fraction II were added 212 gm of ammonium sulfate slowly with stirring. After 30 min at 0°, the suspension was centrifuged and the precipi- tate discarded. An additional 87.8 gm of ammonium sulfate were added, and after 30 min the precipitate was collected and dissolved in 50 ml of 0.1 M glycylglycine, pH 7.0 (fraction III, 52 ml). (iv) DEAE-cellulose chromatography: A column of DEAE-cellulose (2.5 X 21 cm) was pre- pared and equilibrated with 1 liter of 0.02 M potassium phosphate buffer, pH 7.5. Fraction III was dialyzed for 4 hr against two 2-liter changes of the same buffer, then applied immediately to the column at a rate of approximately 40 ml per hour. The adsorbent was washed with 100 ml of 0.05 M potassium phosphate buffer, pH 7.5, then a linear gradient (1 liter total vol) from 0.05 M to 0.3 M potassium phosphate, pH 7.5, was applied. The flow rate was 86 ml per hour and 10-ml fractions were collected. The enzyme was eluted between 0.15 and 0.18 M buffer (fraction IV, 38 ml). (v) Phosphocellulose chromatography: A phosphocellulose column (2.5 X 11 cm) was equili- brated with 1 liter of 0.02 M potassium phosphate buffer, pIT 6.5. Fraction IV (38 ml) was dia- lyzed for 4 hr (3 changes) against 0.02 M potassium phosphate, pH 6.5, and immediately applied to the column. The adsorbent was then washed (flow rate, approximately 80 ml per hr) with 120 ml of 0.02 M potassium phosphate buffer, pH 6.5, followed by 100 ml of 0.05, Al potassium phos- phate buffer, pH 6.5. The enzyme was eluted with 0.10 M potassium phosphate buffer, pH 6.5. Fractions of 5 ml were collected and assayed within a few hours after elution from the column. The peak fractions were combined (fraction V, 35 ml) and concentrated 20- to 40-fold by dialysis against solid sucrose. Approximately 75%0 of the activity of fraction V was recovered after concentration. A summary of the purification is given in Table 1. Fraction V was approximately 650-fold purified over the crude extract and represented 12% of the activity initially present; it was used for all the studies to be described. Downloaded by guest on September 29, 2021 VOL. 57, 1967 BIOCHEMISTRY: OLIVERA AND LEHMAN 1429 TABLE 1 PURIFICATION OF POLYNUCLEOTIDE-JOINING ENZYME FROM E. coli Total activity Specific activity Fraction (units) (units/mg) I. Extract 8680 2.4 II. Streptomycin* (3620) (2.6) III. Ammonium sulfate 5040 8.1 IV. DEAE-cellulose 2800 75'.0 V. Phosphocellulose 1022 1575.0 * Residual streptomycin inhibits the reaction, giving falsely low values for this fraction. When stored at 0°, fraction V showed no noticeable loss of activity (<5%) on repeated assays over a period of several weeks. The following fractions were unstable (>30% loss of activity per week): the dialyzed ammonium sulfate fraction, the dialyzed DEAE-cellulose fraction, and the phosphocellulose fractions before concentration. In addition to concentration against sucrose, the phosphocellulose fractions could be stabilized by dilution 1:1 with glycerol and storage at - 200. The crude extract and ammonium sulfate fraction showed no significant inactivation (<20% loss per week) for a period of several weeks. Fraction V was relatively free of and DNA polymerase activity. Under standard reaction conditions, using 0.06 unit of enzyme (an amount which converted more than 60% of added 51p32 dT to a phosphatase-insensitive form in 30 min), there was no (<1%) detectable degradation of the 5'_p32 dT to an acid-soluble form either in the presence or absence of dA. A similar level of enzyme showed no detectable DNA polymerase activity under condi- tions optimal for either the polynucleotide-joining enzyme or the E. coli polymerase.16 Results.-Properties of the purified enzyme: Conversion of the p32 5'-phosphoryl group of dT to a phosphatase-insensitive form required a divalent cation, dA, and a cofactor present in boiled extracts of E. coli. Thus, in a reaction in which there was a conversion of 0.80 ,.iumole of 5'_p32 dT to a phosphatase-insensitive form, less than 0.01 jymole was converted in the absence of any one of these com- ponents. Mg++ was the most effective of the cations tested; the optimal level was 1-3 X 10-3 M. At a concentration of 3 X 10-3 M, Ca++ was 60 per cent and 1In++ only 13 per cent as effective as Mg++. The rate of reaction was proportional to the concentration of boiled extract in the range 0.01-0.2 Mul, until a saturating level was reached (>0.2 JAI). Cofactor activity was lost after treatment of the boiled extract with Norit, but it was not affected by levels of bacterial alkaline phosphatase that converted >95 per cent of added y_-P32-labeled ATP (0.1 MM) to a Norit nonadsorbable form. It was also unaffected by treatment with pancreatic RNase and DNase. The boiled extract could not be replaced by ATP and deoxythymidine 5'-triphosphate (dTTP) (1 MAM). Attempts to purify and characterize the cofactor are currently in prog- ress.2S With an excess of dT over dA (on' a nucleotide basis), the extent of conversion was proportional to the dA concentration. The effect of varying the level of dA is shown in Figure 2. When the concentrations of dA and dT were equivalent, ap- proximately 90 per cent of the 5 _p32 termini became phosphatase-insensitive ill 30 minutes. When the concentration of dA was only 0.3 that of the dT, only 30 per cent conversion was observed. Further addition of 5'_p32 dT at 20 minutes had i1o effect. In contrast, increasing the concentration of dA to a level 1.7 that of the dT resulted in a rapid conversion of the dT, at about one-half the rate29 and nearly to the same extent observed when the concentrations of dA and dT were initially equal. Downloaded by guest on September 29, 2021 1430 BIOCHEMIST7RY: OLIVERA AND LEHMAN Pnoc. N. A. S. FIG. 2.-Effect of dA concentration ______on rate and extent of reaction. The 0 reactions were performed using stand- ard assay conditions (Methods) except d A = dT that the concentration of dA was varied / as indicated. The reaction mixture / contained 1.6 X 10' M 5'-p32 dT - M a 0.7 (10-8 in termini) and 0.28 unit of 0 / enzyme per ml. The points shown 0 / -- represent assays on 0.10-ml aliquots, 0 0.6 / dA= 1.XdT at 20 containing 1 iujumole of dT chains. 0 I The volume for the reaction in which / o'dA = dT was 1.0 ml; for the reaction o i 'ls0. in which dA = 0.3 dT, it was 2.0 ml. /* In the latter, two 0.3-ml aliquots loJ* were taken at 20 min. To one was L 0.4 _ I et _ added 0.024 ml of 5'.p32 dT (2.1 X f I d10-6 M, 1.27 X io-M in termini); c/c dT doubled at 20' this increased the dT concentration to 0.IFo2 A 2 X 1(-8 M (termini). To the other dA=0.3dT aliquot was added 0.015 ml of dA (4.6 0 1/ X 10-5 M). Aliquots (0.10 ml) were 0 0.2 withdrawn at the times indicated and

£E assayed as described under Methods. ll The number of jurumoles made insus- - ceptible to alkaline phosphatase was calculated by correcting for the dilu- introduced by the further addition nodA tion 0 r___ _ _ of dA and dT. The reactions in the 02l 0 30d0 absence of dA were carried out in two 0 1 0 20 30o 40 5 0 60 separate tubes (0.1 ml each), one Time in minutes incubated for 30 min and the other for 60 min.

IzFIG.I I3.-Electrophoresis| T I I II I of B. sub- dTMP 3,5-dTDP tilis nuclease digest of dT and 400 m F1 _ the product of polynucleotide-joining5'_P32 enzyme. Two reaction mixtures (0.365 1200 - ml each) were prepared containing 0.01 M Tris-JICi buffer, pH 8.0, 0.001 M EDTA, 0.004 M MgCl2, 0.68 o000 I m,.moles 5' p32 dT (4.2 juvmoles of 0oecenznma added I termini), 2.8 mjsmoles of dA, and 15 jul S... ControI of boiled extract. Polynucleotide-join- *0oo I ing enzyme (0.3 unit) was added to IIl I' one of the reaction mixtures; the other I I lserved as a control. Both mixtures CC600-boo _ I \ : | _ were incubated at 30P for 30 min, then 0 l l l I heated at 1000 for 2 min. One-tenth \ I ml of 0.01 M 0.04 ml of 1 M ,+00 Tris-HCl, pH CaCl2,8.0, and 520 units of I\ g z ~~~B. ubtili8 were added and 20020| \ tI A. -- the mixture was incubated at 370 for 1 30 min. At this point, the p32 was >99% acid-soluble. The reaction mix- 0 - - - tures were then chilled and treated I 1 with, kNoriti xx asx Xdescribed under Methods. 0 2 4 b 8 10 12 14 lb 18 20 22 24 26 28 30 The Norit was washed twice with 1-ml Cm. from the origin portions of cold water, and then eluted twice (00) with 1-ml portions of 50% ethanol containing 0.02 M NH4OH. Approximately 50% of the adsorbed p32 was eluted from the Norit by this procedure. The combined eluates were evaporated to dryness, the residue re- dissolved in 25,l of H20, and spotted on Schleicher and Schuell 5890 orange ribbon paper. 3'-dTMP and 3',5'dTI)P (0.1 jmole each) were added as carrier. Electrophoresis was then per- formed as described in Methods. After drying the paper, strips (1.5 cm long and 2.5 cm wide were cut out and counted directly in the gas-flow counter. Downloaded by guest on September 29, 2021 VOL. 57, 1967 BIOCHEMISTRY: OLIVERA AND LEHMAN 1431

Characterization of the product of the reaction: (1) Evidence that a phosphodiester bond is formed: The product synthesized in a standard assay was quantitatively converted to mononucleotides by the action of E. coli exonuclease O20 (100 units) and the B. subtilis nuclease19 (100 units), suggesting that the 5'-phosphoryl group of the dT was incorporated into a phosphodiester linkage. M\lore definitive evi- dence was provided by the following experiment. The product of the reaction was degraded to 3'-mononucleotides by the B. subtilis nuclease and then subjected to paper electrophoresis. As shown in Figure 3, only two radioactive peaks appeared, one coincident with authentic 3'5'-deoxy- thymidine diphosphate (3'5'-dTDP) (15% of the total) and the second showing the same mobility as dTMIP (85% of the total). The latter would be expected to be the 3' isomer based on the known specificity of the B. subtilis enzyme."9 The identity of the dTMP was further substantiated by its differential susceptibility to the specific 3'- and 5'-. Thus, treatment of the monophosphate with 3'-nucleotidase resulted in 93 per cent dephosphorylation, whereas only 5 per cent of the phosphate was removed by the action of 5'-nucleotidase. In the control experiment, treatment of unreacted P32 5'-phosphoryl dT with B. subtilis nuclease resulted in conversion of over 70 per cent of the radioactivity to 3'5'-dTDP (the remainder appeared as inorganic phosphate). Electrophoresis of the Norit-adsorbable products of B. subtilis nuclease digestion of the unreacted P32 5'-phosphoryl dT yielded a single peak of radioactivity at the position of 3'5'-dTDP (Fig. 3). There was no detectable P32 dTMP (either 3' or 5'). The identity of the 3'5'-dTDP was further verified by the finding that the P32 was essentially insusceptible to the action of either 3' or 5'-nucleotidase (6% converted to P132); on the other hand, it was almost quantitatively (97%0) removed upon sequential treatment with 3'-nucleotidase followed by 5'-nucleotidase. (2) Evidence that dT increases in molecular weight: The product of the reaction catalyzed by the polynucleotide-joining enzyme when the ratio of dA/dT was approximately 1 was examined by sedimentation in an alkaline sucrose gradients (Fig. 4). The initially short dT chains (150 nucleotides) were increased consider- ably in length as a result of the action of the enzyme. Sixty-five per cent of the

30 FIG. 4.-Alkaline sucrose density gradient centrifugation of products of polynucleotide-joining enzyme. Re- action mixtures (0.2 ml) in which dA E = dT and dA = 0.3 dT were prepared ° 20 - control as described in the legend to Fig. 3. a 30-min incubation (300), 25 - _ sl of 0.2 M EDTA and 25 ,ulof 2 M /. .dA-d~s/@|NaOH were added and the mixtures k& a were layered onto sucrose gradients cl0 - (5-20%) containing 0.3 M NaOH //,Ad \.8 0.7 M NaCl, and 0.002 M EDTA. The control sample is the dT sub- \ strate that had not been treated with g/,aSv o___f enzyme. Sedimentation was performed I e a SW39 rotor O !_*ge-0 4 0 F8OO~o0 12 14 lb 8 20 22 modelusing L centrifuge for in22 hrtheat Spinco37,000 rpm at 100. At the conclusion of the run, the bottoms of the tubes were punctured and 17-drop fractions were collected directly onto glass filters. The fractions were then counted in the gas-flow counter. The total number of cpm per tube was between 1000 and 1500. The direction of sedimentation is from rightI'Afterto left. Downloaded by guest on September 29, 2021 1432 BIOCHEMJSTRY: OLIVERA AND LEHMAN Pitoc. N. A. S. dT attained a sedimentation coefficient at least twice as great as the unreacted dT, implying that 65 per cent of the product was at least four times longer than the original dT. By the same criterion, approximately 40 per cent of the dT chains increased in length by at least tenfold and 5 per cent by 20-fold. Also shown in Figure 4 is the product of a reaction in which the dA/dT ratio was 0.3. There is a suggestion of two peaks of radioactivity, one of which (representing 30% of the total) had a median sedimentation coefficient threefold greater than the remainder of the material, which sedimented at the same rate as unreacted dT. (3) Conversion of "hydrogen-bonded circles" of X DNA to "covalent circles": The polynucleotide-joining enzyme catalyzed the conversion of hydrogen-bonded circles of X DNA to the covalently closed form. The latter is characterized by its very rapid sedimentation rate at alkaline pH.7 Treatment of hydrogen-bonded circles (10 mumoles of nucleotide/ml) with 1.3 units of enzyme for 30 minutes re- sulted in the conversion of 30 per cent of the DNA to a form whose sedimentation coefficient at alkaline pH was fourfold greater than that of the untreated circles. Incubation of the hydrogen-bonded circles with 0.2 unit of enzyme for only 5 minutes produced the same (30%) extent of conversion. The value of 30 per cent may be a minimal one since the preparation of hydrogen-bonded circles used may well have contained a significant fraction of linear molecules or circles lacking a terminal 5'- phosphoryl group. These findings implythat the covalent closure of X circles occurs by a reaction involving the 3'-hydroxyl and 5'-phosphoryl termini of the X DNA molecule, resulting in the formation of two new phosphodiester bonds. Discussion.-The polynucleotide-joining enzyme catalyzes the linkage of dT chains through phosphodiester bonds. The stoichiometry of the reaction at various dA/dT ratios indicates that only those dT chains that are hydrogen-bonded to dA may be linked. Under the conditions used, there would appear to be no mechanism operative permitting more than a stoichiometric number of dT mole- cules to be linked on a dA template. The mechanism of the joining reaction is unknown and must remain so until the cofactor has been purified and identified. The cofactor (or one of its components) presumably provides the energy needed to form the phosphodiester bond. Its insensitivity to bacterial alkaline phosphatase would tend to eliminate a nucleoside polyphosphate. If this finding is in fact substantiated upon chemical identification in the cofactor, the reaction catalyzed by the polynucleotide-joining enzyme would be unusual in the participation of an energy source other than a nucleoside di- or triphosphate in phosphodiester bond formation. The reaction is also novel in that it leads to the formation of a phosphodiester bond at the 5'-terminus of a polynucleotide. Sumnmary.-An enzyme, purified approximately 600-fold from extracts of Escher- ichia coli, catalyzes the condensation of short (150 residues) polydeoxythmidylate chains to form deoxythymidylate polymers whose length has increased by as much as 20-fold. It does so through the formation of typical 3'5' phosphodiester link- ages. For the reaction to occur, the enzyme requires (1) a divalent cation (Mlg++ or Ca++); (2) a factor or factors present in boiled extracts of E. coli; and (3) fixation of the polydeoxythymidylate chains by hydrogen bonds to a long (3000 residues) deoxyadenylate polymer. The enzyme also catalyzes the formation of "covalent circles" from "hiydrogen-bonded circles" of phage X DNA. Downloaded by guest on September 29, 2021 VOL. 57, 1967 BIOCHEMISTRY: OLIVERA AND LEHMAN 1433

Note added in proof: Recent studies indicate that the "cofactor" can be replaced by DPN. We wish to acknowledge the expert help of Mrs. Janiice It. Chien. * Supported in part by a grant from the National Istitutes of Health, USPHS. f Postdoctoral fellow of the 1amon Runyon Memorial Fund for Cancer Research. ' Meselson, M., J. Mol. Biol., 9, 734 (1964). 2 Howard-Flanders, P., Radiation Res., suppl. 6 (1967), in press. I Lehman, I. R., Ann. Rev. Biochem. (1967), in press. 4Gellert, M., these PROCEEDINGS, 57, 148 (1967). 5 Hershey, A. D., E. Burgi, and L. Ingraham, these PROCEEDINGS, 49, 748 (1963). 6 Young, E. T., and R. L. Sinsheimer, J. Mol. Biol., 10, 562 (1964). 7 Bode, V. C., and A. D. Kaiser, J. Mol. Biol., 14, 399 (1965). 8 The designations for a polynucleotide chain and the abbreviations used are those recom- mended by J. Biol. Chem., 242, 1 (1967). The following additional abbreviations are used: dT, a homopolymer composed of deoxythymidylate residues; dA, a homopolymer composed of de- oxyadenylate residues; dA: dT, a double-stranded polymer composed of dA and dT chains hydrogen-bonded together; DEAE-cellulose, diethylaminoethyl cellulose; EDTA, ethylene di- amine tetraacetate; Pi, inorganic orthophosphate; 3'5'-dTDP, 3'.5' deoxythymidine diphosphate. 9 Doherty, R., and D. S. Hogness, to be published. °0 Wang, J. C., and N. 1)avidson, J. Mol. Biol., 15, 111 (1966). "Turner, A. F., and H. G. Khorana, J. Am. Chem. Soc., 81, 4651 (1959). 12 Glynn, I. M., and J. B. Chappell, Biochem. J., 90, 147 (1964). 13 Wu, R., and A. D. Kaiser, these PROCEEDINGS, 57, 170 (1967). 14 Richardson, C. C., these PROCEEDINGS, 54, 158 (1965). 15 Lehman, I. R., G. G. Roussos, and E. A. Pratt, J. Biol. Chem., 237, 819 (1962). 16 Richardson, C. C., C. L. Schildkraut, H. V. Aposhian, and A. Kornberg, J. Biol. Chem., 239, 222 (1964). 17 Cunningham, L., B. W. Catlin, and M. Privat De Garilhe, J. Am. Chem. Soc., 78, 4642 (1956). 18 Becker, A., and J. Hurwitz, J. Biol. Chem., 242, 936 (1967). We are grateful to Dr. Hurwitz for permitting us to see this manuscript in advance of its publication. 19 Kerr, I. M., J. R. Chien, and I. R. Lehman, J. Biol. Chem. (1967), in press. 20 Lehman, I. R., and A. L. Nussbaum, J. Biol. Chem., 239, 2628 (1964). 21 Garen, A., and C. Levinthal, Biochim. Biophys. Adta, 38, 470 (1960). 22 Riley, M., B. Maling, and M. Chamberlin, J. Mol. Biol., 20, 359 (1966). 23 The concentration of _yP32 ATP used (3 X 106 M, 6 X 109 cpm/gmole) was considerably below the Km value of 14.3 X 10-6 M (Novogrodsky, A., M. Tat, A. Traub, and J. Hurwitz, J. Biol. Chem., 241, 2933 (1966)); a large excess of kinase was therefore required to achieve quanti- tative phosphorylation. 24 Malamy, M. H., and B. L. Horecker, Biochemistry, 3, 1893 (1964). 25 It was necessary to carry out incubations with the phosphatase at 850 because the 5'-phos- phoryl termini of dT are resistant to the enzyme when the dT is hydrogen-bonded to dA. 26 Lowry, 0. H., N. L. Rosebrough, A. L. Farr, and R. J. Randall, J. Biol. Chem., 193, 265 (1951). 27 Warburg, O., and W. Christian, Biochem. Z., 310, 384 (1942). 28 Dr. Martin Gellert has informed us that he has observed the requirement for a heat-stable cofactor, different from ATP, for a polynucleotide linkage activity currently under study in his laboratory. 29 The decrease in rate is the result of some inactivation of the enzyme during the 20-min incuba- tion. 30 Martin, R. G., and B. Ames, J. Biol. Chem., 236, 1372 (1961), Downloaded by guest on September 29, 2021