THE JOURNAL OF BIOLOQICAL CHEMISTRY Vol. 239, No. 1. January 1964 Printed in U.S.A. A Deoxyribonucleic Acid Phosphatase-Exonuclease from Escherichia coli 11. CHARACTERIZATION OF THE EXONUCLEASE ACTIVITY * CHARLES c. RICHARDSON,t I. R. LEHMAN, AND ARTHUR KORNBERG From the Department of Biochemistry, Stanford University School of Medicine, Palo Alto, California (Received for publication, July 18, 1963) An enzyme detected initially by its capacity to increase the priming capacity of deoxyribonucleic acid was identified upon its purification from Eschmichia coli extracts as a phosphatase acting specifically on 3'-phosphoryl-terminated deosyribonucleic acid chains (1, 2). Concurrent with the purification of this DNA phosphatase, there was an enrichment for an esonucleolytic activity which could be distinguished from the two known exo- nucleases of E. coli. Evidence for the identity of the phosphatase and exonuclease activities with a single enzyme is the subject of this report and the basis for naming this enzyme a DNA phosphataseexonucle. The specificity of the enzyme in attacking native double stranded DNA and its inability to act on small polynucleotides make it a unique and useful reagent in studies of DNA structure and metabolism. EXPERIMENTAL PROCEDURE Materials The 32P- and SH-labeled DNAs and RNA, the mised polymer of ribo- and deoxyribonucleotides, the synthetic oligonucleotides, and the 3'- and 5'-phosphoryl-terminated DNA substrates were prepared as described in the previous paper (1). Heat-denatured DNA was prepared by heating 3LP-labeled E. coli DNA (0.3 pmole per ml) in a boiling water bath for 10 minutes in 0.05 M KCl and then quickly cooling it in an ice bath. Unlabeled deoxyribonucleoside triphosphates were purchased from the California Corporation for Biochemical Refearch. Deosythy- midme-2-1% was purchased from the New England Nuclear Corporation; it was phosphorylated to 5'-deoxythymidylate by E. coli deoxythymidine kinase, and subsequently to 5'-deosy- thymidine t,riphosphate as previously described (3). The synthetic dAT copolymer' was prepared by synthesis de novo (4) with 14C-labeled dTTP, dATP, and purified E. coli DNA polymerase. The synthetic rAU copolymer was prepared with E. coli RNA polymerase in a dAT-primed synthesis utilizing IC-ATP and UTP as substrates (5). Bacillus subtilis DNA labeled with W-deoxythymidylate at its 3'-hydrosyl end was * This work was supported in part by Grants GM 07581 and 5T1 Gm 196 from the National Institutes of Health, United States Public Health Service. t Postdoctoral Fellow, United States Public Health Service (Grant GF 13,634). Present addrese, Department of Biological Chemistry, Harvard Medical School, Boston 15, Massachusetts. * The abbreviations used are: dAT copolymer, copolymer of deoxyadenylate and deoxyt,hymidylate; rAU copolymer, copoly- mer of adenylate and uridylate; the prefix "r" denotes "ribo." prepared by treating B. dtilia DNA with IC-deoxythymidylate in the presence of purified E. coli DNA polymerase (6). Hybrid DNA with one strand 16N-2H-labeled and the other strand normal, isolated from B. sztbtilis, was a gift of Dr. A. T. Ganesan. The latter DNA was labeled in both strands with 'H (specific activity, 2 x IOs c.p.m. per pmole). The deoxythymidine 3'- and B'-p-nitrophenyl esters were prepared by Dr. A. Nussbaum. The fractions of E. coli DNA phosphataseexonuclease were prepared and assayed as described in the previous paper (1). Fraction VI, the phosphocellulose pervaporate, was used as the purified enzyme source in all the esperiments to be described. E. coli phosphodiesterase, specific for single stranded DNA, and specific 5'-nucleotidase from Crotalus adamanteus venom were purified and assayed as previously described (7, 8). E. coli alkaline phosphatase was obtained from the Worthington Bio- chemical Corporation and characterixed as before (1). Methods Assay of Ewnuckase Activity of E. wli DNA Phosphdase- ezonuckuse-This assay measures the conversion of a 52P-labeled native DNA to acid-soluble fragments. The incubation misture (0.3 ml) contained 20 pmoles of Tris-HC1 buffer, pH 8.0, 0.2 pmole of MgCl,, 0.3 pmole of 2-mercappethano1, 50 mpmoles of 32P-l&ehd native E. coli DNA, and 0.1 to 2.0 units of enzyme? The reaction mixture was incubated for 30 minutes at 37"; 0.2 ml of a cold solution of calf thymus DNA (2.5 mg per ml) and 0.5 ml of cold 7% perchloric acid were added. After 5 minutes at O", the resulting precipitate was removed by centrifugation at 10,OOO x g for 5 minutes; 0.2 ml of the supernatant fluid was pipetted into a planchet, and 1 drop of 1 N KOH was added to the aliquot. The solution was taken to dryness, and the radio- activity was measured. The supernatant fluids obtained from control incubations with enzyme omitted contained 0.3 to 0.5% of the added radioactivity. A unit of exonuclease activity is defined as the amount causing the production of 1.0 mpmole of acid-soluble 32P in 30 minutes. The radioactivity produced w.1~ proportional to the enzyme concentration at levels of 0.1 to 2.0 units of enzyme. Thus, with the addition of 0.005, 0.01, 0.02, and 0.03 ml of a 1:lW dilution of Fraction VI, specific activities of 6,300, 6,250, 6,400, and 6,200, respectively, were obtained. Unless otherwise noted, all enzyme units in this paper refer to those obtained with the exonuclease assay. 2 Enzyme dilutions were made as previously described for the DNA phosphatase assay (1). 251 252 DNA Phosphatase-Exonuclease from E. coli. II Vol. 239, No. 1 TABLE I Identification of acid-soluble product of DNA phosphatase-exonu- clease action on native DNA as 6'-mononucleotides The reaction mixtures (0.3 ml) contained 20 pmolea of Tris- HCl buffer, pH 8.0, 0.2 pmole of MgCl,, 0.3 pmole of 2-mercapto- ethanol, 25 mpmoles of a*P-labeled E. cola DNA, and 30 units of DNA phosphatase-exonuclease. After incubation for 30 minutes, a 0.05-ml aliquot was assayed in the routine manner for the pro- duction of acid-soluble SP. Radioactivity susceptible to E. coli alkaline phosphatwe (20 pg) in a 0.05-ml aliquot was deter- mined in a Norit assay as described for the DNA phosphatase activity (1) ; for susceptibility to 5'-nucleotidase, a 0.05-ml aliquot was assayed in a reaction mixture (0.3 ml) containing 30 pmoles of glycine buffer, pH 9.6, pmoles of MgCIz, and 5 units of venom 5'-nucleotidase. I Acid-soluble, Norit-nonadsorbable s*P I Exueriment I Acid-soluble "P 1 I ".. I Venom 1. cog1 Control ,&%FW 5'-nucleotidad I nrpmotcr I mpmdw 1 7.8 he formation of phosphate esters susceptible to E. wli alkaline phosphatase or 5'-nucleotidase. Dmily Gradient Centrifugation-The technique described by Meselson, Stahl, and Vinograd (10) was followed. Solid CsCl (Harshaw Chemical Company, optical grade) was added to the DNA solution to raise the density to values between 1.725 and 1.750 g per cm3, depending on the sample being examined. The esact density of the solution was determined by means of the linear relation between refractive index and density (11). Ap- proximately 0.75 ml of the final CsCl solution, containing 0.05 M Tris at pH 8.0, was placed in a cell (plastic Kel-F centerpiece) and centrifuged in a Spinco model E analytical ultracentrifuge at 44,770 r.p.m. at 25". After 20 hours of centrifugation, ultra- violet absorption photographs were taken on Kodak commercial film. Tracings were then made with a Joyce-Loebl double beam recording microdensitometer with an effective slit width of 30 ~.r in the film dimension. Densities were calculated by using the position of Tetrahymena pyriformh (buoyant density of 1.684 g per cma) as a reference (12). Other Methods-Protein, inorganic orthophosphate, and de- oxyribose were determined as described in the previous paper (1). '2P waa counted in a windowless gas flow counter; 14C and 5H were measured in the Packard Tri-Carb liquid scintillation counter. REsms Ezonucleolytic Action of Enzyme Identi$cation of Acid-soluble Product of Reaction-When native 82P-labeled E. co2i DNA was incubated with the enzyme, 31% of the radioactivity was rendered acid-soluble. Greater than 95% of the acid-soluble label was susceptible to E. coli alkaline phosphatase as measured by the formation of Norit-nonadsorb- able 3zP. Treatment of the product with the specific venom 5'-nucleotidase also rendered 91 % of the radioactivity Norit- nonadsorbable, thus identifying the acid-soluble product as 5'-mononucleotides (Table I). Emnucbase Action on W"Te7minally Labeled DNA-When DNA terminally labeled at its 3'-hydroxyl end with 14Cdeoxy- thymidylate was treated with the E. coli DNA phosphatase- esonuclease, 90% of the radioactive deoxynucleotides of the molecules were made acid-soluble in 15 minutes. At this time, less than 1% of the unlabeled nucleotides had been released as judged by the appearance of acid-soluble ultraviolet-absorbing material (Fig. 1). This result is similar to that observed with venom diesterase (6, 7), an enzyme which attacks DNA or polynucleotides stepwise from the end bearing a free 3'-hydrosyl group (13). Digestion of a terminally labeled DNA with pancreatic DNase, an endonuclease which attacks DNA in a random manner (14), has been previously shown to release unlabeled nucleotides at a rate similar to the release of radio- activity (7). It therefore appears that the DNA phosphatase- exonuclease carries out a stepwise attack on DNA, starting from the 3'-hydroxyl end and producing 5'-mononucleotides in a manner analogous to venom diesterase. 80 - 0 60 c CL t 0 0, @ 40 2 a 20 0 I I I I Fro. 1. DNA phosphatase-exonuclease action on DNA termi- nally labeled with 'Cdeoxythymidylate. The incubation mix- ture (2.4 ml) contained 0.64 pmole of DNA phosphate terminally labeled with W-deoxythymidylate (2250 c.p.m. per pmole of DNA phosphate), 160 pmoles of Tris-HC1, pH 8.0, 1.6 pmoles of MgClt, 2.4 pmoles of 2-mercaptoethanol, and 16 units of the purified enzyme (Fraction VI). Incubation was at 37". At the times indicated, 0.3-ml aliquots were removed; 0.2 ml of calf thymus "carrier" DNA (2.5 mg per ml) and 0.5 ml of cold 7% perchloric acid were added. After 5 minutes at O", the suspensions were centrifuged and the optical density of the supernatant fluids was determined at 260 mp. The radioactivity in the supernatant fluid was determined by pipetting 0.2 ml into an aqueous scin- tillator solution and is expressed as a percentage of the total radioactivity in the DNA added at zero time. Measurement of the radioactivity in the precipitate aa previously described (7) gave values for hydrolysis that agreed within 8%. ptnuary 1964 C. C. Richardson, I. R. Lehman, and A. Kornberg 253 Sequential Release of Pi and Mononuchtides from $-Phs- I-terminated DNA-Unlike venom diesterase (13) or E. osphodiesterase,8 the DNA phosphatase-exonuclease a 3'-phosphoryl-terminated DNA. The enzyme first moves the 3'-phosphoryl group, as described in the previous (I), and then carries out a stepwise attack on the molecule, from the newly formed 3'-hydroxyl end. When 32P- 3'-phosphoryl-terminated DNA was incubated with the there was an immediate release of Pi, which reached a in approximately 30 minutes (Fig. 2). The release of -adsorbable radioactivity (5'-mononucleotides) occurred ter a lag of approximately 5 minutes. However, the rate of mononucleotides rapidly increased, and reached a e which corresponded to that initially seen with the . It appears that the enzyme attacks the 3'-hydroxyl of the molecule only after having fist removed the 3'- end group. The enzyme apparently has the same this monoesterified group and cleaves it at about as the phosphodiester linkage. NA Structure on Rate of Reaction-The DNA hataseexonuclease hydrolyzed B. subtilis DNA, E. coli and the synthetic dAT copolymer at similar rates (Table . J2P-Labeled E. coli DNA, partially digested with either coccal nuclease or E. coli endonuclease (see "Methods"), ed on by the enzyme at the same rate as the untreated DNA. at-denatured E. coli DNA is hydrolyzed at approximately rth the rate seen with the native DNA. The addition of natured DNA (30 mMmoles) to the standard reaction produced no inhibition. S*P-labeled E. coli ribosomal RNA was added to the mixture, acid-soluble 32P was released at B rate 2% of seen with native DNA. However, the hydrolysis of the was inhibited by Mg++, and the products of the hydrolysis found to be oligonucleotides as judged by the limited ceptibility to E. wli alkaline phosphatase. No 5'-mOnO- otides were produced as tested by the formation of venom idase-sensitive phosphate. These results suggest that phosphatase-exonuclease does not attack ribosomal that the slow degradation of RNA to oligonucleotides e to a contaminant of E. coli ribonuclease (15). Since y structure of the DNA substrate clearly influenced extent of hydrolysis, the hydrogen-bonded, double U copolymer (5) was incubated with the purified thetic RNA was not hydrolyzed than 0.1% the rate on native DNA) as measured by the radioactivity. The synthetic oligo- is not a substrate for the enzyme, nor dine 3'- and Ei'-p-nitrophenyl esters f pTpTpTpTpT to the standard detectable inhibition. ture m Extent of Reaction-The exonuclease coli or B. subtdis DNA to only 35 to 45% le 11). The addition of more enzyme or result in further hydrolysis of the tive DNA. However, if an additional 50 mpmoles of Sip- coli DNA were added, there was a further release of ity at a rate similar to that initially seen with the native icating that the enzyme had not been inactivated and e products were not inhibitory. The dAT copolymer, which, because of its alternating, re- 8 I. R. Lehman, personal communication. I I I I 'A 0 /norgan;c orthophosphatc? o 5 'Monon ucleo fides I 'I Minutes F~G. 2. Sequential release of Pi and mononucleotides from 3'-phosphoryLterminated DNA. The reaction mixture (2.4 ml) contained 400 mpmoles of "P-labeled 3'-phosphoryl-terminated DNA, 160pmoles of potassium phosphate buffer, pH 7.0,24 pmoles of MgC12, 2.4 pmoles of 2-mercaptoethanol, and 1.2 units of DNA phosphatase-exonuclease (DNA phsophatase units). Incubation was at 37". At the times indicated, 0.3-ml aliquots were removed; 0.2 ml of calf thymus "carrier" DNA (2.5 mg per ml) and 0.5 ml of cold 10% trichloroacetic acid were added. After 5 minutes at 0", the suspensions were centrifuged and 0.2 ml of the supernatant fluid was pipetted into a planchet for determination of the acid- soluble **P. A similar aliquot of 0.5 ml of the supernatant fluid was treated in a manner identical with the routine DNA phos- phatase assay (1) to obtain the acid-soluble, Norit-nonadsorbable SIP (Pi). The difference between the **Pi released and the acid- soluble J*P represents the release of mononucleotides. peating sequence, remains double stranded even after extensive hydrolysis (16), was hydrolyzed almost to completion. With heat-denatured E. coli DNA, the initial rate of hydrolysis was slow and decreased progressively, but could be carried to 18% by the addition of large amounts of enzyme (150 units) and prolonged incubation (2 hours). Effect of Ineubdion at 46" on Limit Reached Wiu, Nalwe and Heat-denatured DNA---Additional evidence that the secondary structure of the DNA plays a significant role in the catalytic property of the enzyme is shown by experiments in which incuba- tion was carried out at 45O, a temperature which decreases the nonspecific hydrogen bonding in heatdenatured DNA (17). A preparation of the heat-denatured DNA is degraded to a limit of only 3% at 45O, as compared to 18% at 37" (Table 111). When the native DNA is incubated at 45", a limit of 32% is obtained as compared to one of 38% at 37". 254 DNA Phosphatase-Exonuclease from E. coli. 11 Vol. 239, No. 1 TABLE I1 Eflect of DNA structure on exonuclease activity The rate of reaction was measured in the standard assay with replacement of the usual native E. coli DNA by the compounds listed. The preparation and characterization of these substrates were described in the previous paper (1). Native E. coli DNA, native B. subtilis DNA, dAT copolymer, heat-denatured E. coli DNA, ribosomal RNA, and the endonuclease-treated DNAs (40 mpmoles of each) were incubated in the standard reaction mixture (0.3 ml) containing 20 pmoles of Tris-HC1 buffer, pH 8.0, 0.2 pmole of MgC12, 0.3 pmole of 2-mercaptoethanol, and 0.5 to 2.0 units of enzyme. Because of a small contamination of the enzyme preparation wiyh E. coli RNase as described in the text, the release of 5'-mononucleotides from the RNA substraOes was assayed with venom 5`-nucleotidase (see Table I). The synthetic substrates were tested at several concentrations ranging from 40 to 200 mpmoles per reaction mixture and at several pH values and MgCll concentrations with 10 to 100 units of enzyme. The natural DNAs, treated and untreated, aa well as the dAT co- polymer, were assayed by determination of acid-soluble radio- activity. Hydrolysis of pTpTpTpTpT was assayed by the forma- tion of phosphate esters susceptible to 5`-nucleotidase as described in Table I. Hydrolysis of the thymidine 3`- and 5`-p-nitrophenyl esters was determined by spectrophotometric assay (see "Met.hods"). For determination of the eztent of reac- tion, t,he reaction mixtures contained these compounds and an excess of enzyme (50 to 100 units) ; the total release of acid-soluble radioactivity was determined after 60 minutes. In each case, additional enzyme (50 units) and incubation for 30 minutes resulted in no further release of radioactivity. The amount of radioactivity made acid-soluble is recorded as the percentage of the total radioactivity present in the reaction mixture. Compound dAT copolymer. .......................... Native E. coli DNA. ...................... Native B. subtiEis DNA.. .................. 3`-Hydroxyl-terminated DNA*. ............ 3`-Phosphoryl-terminated DNA*. .......... Heat-denatured E. coli DNA.. ............. Ribosomal RNA. .......................... rAU copolymer. ........................... pTpTpTpTpT ............................. Thymidine a'-p-nitrophenyl ester. ......... Thymidine 5'-p-nitrophenyl ester. ......... mwwh/ 1370 1130 2100 1460 1391 350t <1 <1 <6# < 13 < 13 nidmg % total 92 38 39 44 50 18 precipitable Product of Reaction-The acid-insoluble product which remains as a result of the extensive action of the DNA phosphatase-exonuclease is largely single stranded, as judged by its susceptibility to E. coli exonuclease I, an enzyme specific for single stranded DNA (7). The susceptibility of the acid- insoluble a*P to this enzyme increases with the progressive action of the DN.4 phosphatase-exonuclease on native DNA (Fig. 3). When 10% of the DNA has been converted to acid-soluble mononucleotides, only 2% of the residual DNA is susceptible to exonuclease I. -4s the DNA phosphatase-exonuclease ap- proaches its limit of degradation (approximately 35% of the total radioactivity), there is a rapid increase in the susceptibility of the acid-precipitable DNA to exonuclease I. Since exo- nuclease I requires a single stranded region terminated by a free 3'-hydrosyl group, these findings indicate that the 3'-hydroxyl termini are protected in double stranded regions until the limit of action of the DNA phosphatase-exonuclease is approached. Buoyant Density Distribuiion in CsCl of Acid-precipitable Product of Extensive DNA Phosphatase-Exonuclease Action- Further evidence for the formation of single stranded DNA by the extensive action of the DNA phosphatase-exonuclease was obtained by following the buoyant density distribution in CsCl of a hybrid B. subtilis DNA, one strand labeled with 15N-2H and one with W-lH. As shown in Fig. 4, such a density-labeled hybrid bands in CsCl at a buoyant density of 1.728 g per cmJ, and on heating and fast cooling the strands separate and band at 1.770 and 1.721 g per cm3, the expected densities of single stranded heavy (`5N-SH-labeled) and single stranded light B. subtilis DNA (18). After extensive digestion with the DNA phosphatase-exonuclease (42 % acid-soluble nucleotides), two new bands appeared, one at a density (1.770 g per cm3) corre- sponding to that of single stranded heavy DNA, and the other at a density (1.721 g per cm3) corresponding to that of single stranded light DNA. Furthermore, the original hybrid band at a density of 1.728 g per cms had increased in density, with a major portion banding in the region (1.735 g per cms) expected of hybrid molecules which were predominately single stranded but unable to completely separate and band separately. Action on Mixed Ribo-Dmxydonucleotide Polymer-DNA chains which contain interspersed J*P-rCMP residues served as substrates for the enzyme. The initial rates were from 3 to 6% of those observed with native E. coli DNA, but the extent of hydrolysis ranged from 77 to 82% (Fig.5); of the labeled nucleo- * The acid-soluble radioactivity obtained with the partially digested substrates reflects not only mononucleotides released but also those oligonucleotides whose size decreased to the "acid- soluble" range. t Proportionality to enzyme concentration with the heat-de- natured DNA as substrate was obtained only up to the release of 0.40 mpmole of nP. Addition of 20 mpmoles of heat-denatured DNA to the standard assay produced no inhibition. $Addition of 100 mpmoles of pTpTpTpTpT to the standard assay produced no inhibition. TABLE I11 Extent of hydrolysis of native and heat-denatured DNA at S7' and 46' The extent of hydrolysis of native and heat-denatured "P- labeled E. coli DNA was determined under the standard assay conditions at 37" and 45". In each reaction mixture were present 40 mpmoles of DNA phosphate. At both temperatures and with each substrate, SO units of DNA phosphatase-exonuclease were added at 15-minute intervals until no further release of acid- soluble radioactivity was obtained. The limit is expressed as the percentage of total srP present in the reaction mixture. The DNA phosphatase-exonuclease appears to require double stranded regions in the molecule in order to hydrolyze the DNA substrate. Heat-denatured DNA apparently contains enough such regions at 37" to enable the enzyme to act on it at a reduced rate and to a limited extent. At 46, where these regions are eliminated, the heat-denatured DNA is hardly degraded at all. Action of E. coli Eronuclease I (Phosphodiesterase) on Acid- DNA Extent I I 370 I 45- % Native .............................. Heat-denatured. ..................... January 1964 C. C. Richardson, I. R. Lehman, and A. Komberg 255 ' 50 - 2s 2? 522 40- ZS 2P 30- a* 2O- 2 2 Extent OC hydrolysis OF DNA by DNA cr: phosphatase-exonuclease, % acid soluble FIG. 3. Action of E. coli exonuclease I ("phosphodiesterase") on the acid-precipitable product of DNA phosphatase-exonuclease reaction. The reaction mixture (2.2 ml) contained 0.27 pmole of S*P-labeled E. coti native DNA, 140 pmoles of Tris-HC1 buffer, pH 8.0, 1.4 pmoles of MgClz, 2.1 pmoles of 2-mercaptoethanol, and 80 units of E. coli DNA phosphatase-exonuclease. Incubation was at 37'. At intervals, 0.1-ml aliquots were removed and added to 0.2 ml of water. The acid-soluble S*P was determined as in the standard exonuclease assay. Similar aliquots (0.1 ml) were also removed and added to reaction mixtures (0.3 ml) containing 20 pmoles of glycine buffer, pH 9.2, 2 pmoles of MgCls, 0.02 pmole of ZnClz (to inhibit DNA phosphatase-exonuclease), and 150 units of E. coli exonuclease I. After incubation at 37' for 30 minutes, the reaction mixture was assayed for acid-soluble arP as above. The percentage of acid-insoluble made acid-soluble by the E. coli exonuclease I is plotted against the percentage of total DNA made acid-soluble by the DNA phosphatase-exonuclease. tides rendered acid-soluble, 95% were attacked by 5'-nucleotidase to Pi (nonadsorbable to Norit), indicating that 5'-rCMP was released by the enzyme. The rather estensive digestion of the mised polymers may be explained by the assumption that these mised polymers are synthesized by addition to chains of native DNA primer as a repair process (6, 19) rather than as new strands. Therefore, even extensive removal of the mised poly- mers would not be expected to alter the essentially native charac- ter of the DNA to which they were initially attached; the DNA exonuclease is thus binding a native DNA rather than single stranded substrate. Evidence that DNA Phosphatase and Emnucbase are Part of a Single Enzyme Constant Ratio of Activities during Purification-During the course of a 1300-fold purification, the ratio of DNA phosphatase to exonuclease activity remained essentially constant (Table IV);' the ratio was also constant throughout the DEAE-cellulose and phosphocellulose peaks. Similar Rcactwn Rates-When the enzyme acts on 82P-labeled 3'-phosphoryl-terminated DNA under the conditions found optimal for the DNA phosphatase activity, it is possible to distinguish the rate of release of orthophosphate and of mono- nucleotides (Fig. 2). The linear rate of release of Pi from the 3'-phosphoryl end groups (1.26 mpmoles in 30 minutes) was 4 The exonuclease assay could be used to follow the purification of the DNA phosphatase-exonuclease provided that soluble RNA and the antiserum to E. coli endonuclease were present inthe assay mixture (see Table IV). followed by an identical rate of release of mononucleotides (1.30 mpmoles in 30 minutes). Dependence on DiYabnt Cations and pH-Optimal conditions for each activity depend on complicated relationships between pH, buffer, divalent cation concentration, and the nature and amount of the DNA substrate. For example, the optimal pH for exonuclease activity at 7 X lo-' M MgCL is 7.7 to 8.4, but with a 10-fold increase in MgCL the pH optimum is 7.0 to 7.4. The latter conditions are just those which were found to be optimal for DNA phosphatase activity (1). In the absence of added Mg*, esonuclease activity (standard assay) is only 6% of the optimal value, and with 0.01 M EDTA, it is undetectable. J Na-tivs DNA treated with DNA phosphatase axonuclaasa I -------- 1.770 1.728 1.721 - Density FIG. 4. Buoyant density distribution in CsCl of the product of extensive digestion of lsN-zH-hybrid B. 8ubtili8 DNA by DNA phosphatase-exonuclease. When a solution of native I'N-EH-hy- hrid DNA labeled in both strands with *H (band profile shown in top tracing) was heated for 5 minutes at a concentration of 30 mpmoles per ml at 100" in 0.05 M KCl and quickly cooled, the strands separated and banded at the expected density for single stranded heavy and single stranded light B. subtilis DNA (second tracing). Treatment of 60 wmoles of thenative hybrid DNA with 16 pg of the DNA phosphatase-exonuclease (Fraction VI) under standard assay conditions resulted in a limit of 42% of the radioactivity becoming acid-soluble. After dialysis first against 1.0 M KCl and then 0.05 M KCI, the product banded as shown in the third tracing. 256 DNA Phosphatase-Emmuckuse from E. coli. 11 Vol. 239, No. 1 I. Extract. .................... 111. Acetone.. ................... IV. DEAE-cellulose ............. VI. Phosphocellulose pervapo- rate.. ....................... , -99 unilJ/ml 780 270 1,040 400 393 123 19,500 6,400 a Hour FIG. 5. Hydrolysis of a mixed ribo-deoxyribonucleotide poly- mer compared to DNA. The mixed polymer contained 32P-rCMP interspersed among the four deoxyribonucleotides with a fre- quency of about 1 per 10 nucleotides (1). The incubation mix- tures (0.3 ml) contained 0.07 M potassium phosphate buffer, pH 7.0,O.Ol M MgClz, 0.001 M 2-mercaptoethanol, mixed polymer con- taining 0.98 mpmole of rCMP (l), and 8 pg of Fraction VI. For DNA hydrolysis, the mixture contained 31 mpmoles of S*P-native E. coli DNA (1.3 X lo5 c.p.m. per fimole) and 4 pg of Fraction VI. Acid-soluble '*P was determined as in the standard exonuclease assay. TABLE IV Ratio of DNA phosphatase and exonuclease activities during purijication The standard DNA phosphatase assay, utilizing the 3'-phos- phoryl-terminated DNA as substrate, was used to determine the DNA phosphatase activity. The standard exonuclease assay, with native DNA as substrate, was used to determine the exo- nuclease activity. In assaying the exonuclease activity in the crude extract and acetone fractions, 10 mr moles of soluble RNA and 0.005 ml of a 1:5 dilution of rabbit antiserum prepared against E. coli endonuclease were added. Fraction and step Ratio of bosphatase to exo- nuclease 2.9 2.6 3.2 3.0 MnCL was about equally effective as MgCL when tested in the range of 2 to 7 x 10-4 M. InhiMwn by ZnC12 and p-Chloromc~rihwai?e--ZnC1~ at 3.3 X lo-& M inhibited DNA phosphatase by 90% (1) and the exonuclease activity by 88%. CaCI2 (3 x lo-* M) inhibited neither activity. The DNA phosphatase activity requires pro- tective sulfhydryl compounds for prolonged incubation (see footnote 5 in (1)); both the DNA phosphatase and exonuclease activities are sensitive to p-chloromercuribenzote. At 1 X lo-' Y in the standard assays (2-mercaptoethanol omitted), p-chloro- mercuribenzoate inhibited the DNA phosphatase activity by 50% and the exonuclease activity by 90%. Heat Inactivation-The two activities follow an identical heat inactivation curve (a. 6). After 30 minutes at 37" in the absence of Mg" or substrate, only 7% of each activity remained. DISCUSSION The enzyme described here is an exonuclease attacking step- wise from the 3'-hydroxyl end of a DNA molecule. This is established by (a) the formation of 5'-mononucleotides as the only acid-soluble products, (b) the preferential removal of labeled nucleotides from a DNA molecule labeled at the 3'-hydroxyl end, and (c) the sequence of reaction on a DNA terminated by a 3'-phosphoryl end; liberation of a 5'-nucleotide occurs only after the 3'-phosphoryl group is removed. The exonuclease attacks denatured DNA at a slow rate and to a very liited extent. The requirement for a double stranded helical structure is suggested by the almost total lack of hydrol- ysis (less than 5%) of heat-denatured DNA at an elevated tem- perature. The inability of the enzyme to hydrolyze short oligo- nucleotides or model phosphodiester substrates and its capacity IOC 90 BC 7c 6C > t > 50 T 0 - ' 40 0 t S 0 t f 0, .- .- 30 ? 2 2c 10 0 5 10 15 Minutes at 37'c FIG. 6. Heat inactivation of DNA phosphatase-exonuclease. Fraction VI (1.6 pg) was diluted to 1.5 ml with the standard diluent and incubated at 37". At the times indicated, 0.005-ml aliquots were added to the standard DNA phosphatase assay and 0.05-ml aliquots to the standard exonuclease assay. The percentage de- crease in activity is relative to the zero time aliquots set at 100. January 1964 C. C. Richardson, I. R. Lehman, and A. Kmberg 257 5` \ <. 4 3` 5' ` 5' $ FIQ. 7. Postulated mechanism of action of the exonucleolytic activity of the DNA phosphataae-exonuclease. to digest the dAT copolymer virtually to completion also dem- onstrate the importance of secondary structure of the substrate in reactions catalyzed by this enzyme. The inability to attack RNA, even when double stranded, must therefore be due to a specificity for the DNA structure. Consistent with the requirement for a double stranded DNA is the fact that its degradation stops when 35 to 45% has been digested. Assuming that the enzyme initiates its stepwise attack from both 3'-hydroxyl ends of the double stranded molecule, then at or near 50% degradation, all of the residual acid-insoluble DNA should be single stranded (Fig. 7) and resistant to further attack. Supporting this is the observation that the residual DNA is largely susceptible to the E. coli phosphodiesterase, an enzyme specific for single stranded DNA. Further evidence for this model is provided by the examination of the buoyant density in CsCi of the acid-insoluble product formed as a result of extensive digestion of a density-labeled hybrid DNA (one strand containing 16N and 2H atoms). Band- ing of the product in CsCl clearly reveals that single strands of both light and heavy buoyant density are formed. The finding that a certain proportion of the hybrid DNA increases in density but does not give rise to strand separation suggests that hydrogen bonding still exists to some extent in these molecules. It is also significant that the dAT copolymer, which, because of its alter- nating structure, remains double stranded even after extensive degradation, is digested to 94% of completion. A most interesting and unusual property of the enzyme is its ability to cleave 3'-phosphoryl groups from the terminus of a DNA chain, a feature which distinguishes it from known exo- nucleases, which are either blocked or inhibited by such end groups. These two activities appear to reside on the same pro- tein molecule and reflect the ability of this enzyme to cleave a 3`-phosphoryl linkage whether it is involved in a mono- or diester linkage. TABLE V Comparison of E. coli exonucleases * Exonuclease I is the E. coli phosphodiesterase, specific for single stranded DNA; exonuclease I1 is the E. coli nuclease purified with DNA polymerase; and exonuclease I11 is the E. coli DNA phosphatase-exonuclease described in this paper. For a comprehen- sive review of the nucleases of E. coli, see (20). Required end group on DNA: Required DNA structure Extent of action Type of attack Products from DNA Action on synthetic oligonucleo- tides : PTPTPTPTPT TPTPTP Exonuclease I Active Inactive Single stranded Up to terminal dinuc-dotide Stepwise beginning at 3'-hy- droxyl end of chain Mono- and dinucleotides with 5`-phosphodiester group Produces 6'-mononucleotides Inactive and pTpT Exonuclease I1 Active Inactive Single or double stranded Complete Stepwise beginning at 3'-hy- droxyl end of chain 5'-Mononucleotides Produces 5'-mononucleotides Inactive Exonuclease III Active Active; because initial attack removes Pi terminus Double stranded Up to 40% degradation; re- sidual single stranded chains are resistant Stepwise beginning at 3'-by- droxyi or phosphoryl end of chain Pi, 5`-mononucleotides, and large molecular weight single stranded oligonucleotides Inactive Inactive 258 DNA Phosphatase-Exonuclease from E. coli. II Vol. 239, No. 1 In addition to the endonuclease in E. coli, there are now three zharacterized exonucleases. Their common and distinguishing properties are summarized in Table V. The DNA phosphatase-exonuclease has properties which may be exploited in several ways. Unlike venom diesterase, spleen diesterase, and H. coli exonuclease I, the DNA phosphatase- exonuclease prefers and may therefore be used for stepwise deg- radation of native, double stranded DNA. Moreover, regardless of the presence of 3'-hydroxyl or 3'-phosphoryl end groups, the enzyme is able to initiate its exonucleolytic action and therefore attack all the DNA molecules. These two properties make it possible to dissect away a bio- logically active DNA by observing the retention as well as loss of selected markers. A preliminary study with B. subtilis DNA showed that when 5% of the DNA had been converted to mono- nucleotides, 50% of the control transforming activity remained, as measured by transformation of the Tryit' (indole) marker.6 Since kinetics of this inactivation was consistent with an exo- slucleolytic attack on the DNA, the absence of endonuclease activity in the purified enzyme was confirmed. Combined with the use of another phosphatase such as the E. coli alkaline phosphatase, which acts on both 5'- and 3'-phos- phoryl termini, it should be possible to determine the occurrence of (and factors producing) such termini in DNA. SUMMARY 1. A deoxyribonucleic acid phosphatase purified extensively from extracts of Escherichia coli (described in the preceding paper) is also an exonuclease carrying out a stepwise attack from the 3'-hydroxyl end of the deoxyribonucleic acid chain, releasing 5'-mononucleotides. 2. The enzyme preferentially attacks native, double stranded DNA, degrading it to an extent of 35 to 45% mononucleotides; the residual, predominately single stranded DNA is resistant to further attack. 3. A rnixed polymer of ribo- and deoxyribonucleotides is cleaved by the enzyme, although at a reduced rate. The enzyme can therefore liberate a 5'-ribonucleotide when it terminates a DNA chain. 4. The enzyme is unable to hydrolyze single stranded short 6 We are indebted to Dr. Walter Bodmer of the Department of Genetics, Stanford University, for carrying out the B. subtilis- transformation assays. oligonucleotides, ribosomal ribonucleic acid, or the adenylate- uridylate copolymer. 5. Identitmy of the phosphatase and esonuclease activities is supported by the constant ratio of the activities throughout purification, similar reaction rates, inhibition by ZnClz and p-chloromercuribenzoate, and the rates of heat inactivation. REFERENCES 1. RICHARDSON, C. C., AND KORNBERG, A., J. Biol. Chem., 239. 242 (1964). KORNBEBG, A., BODMER, W., AND LEDERBERG, J., in H. VOGEL (Editor), Symposium on informational macromole- cules, Academic Press, Inc., New York, 1963, p. 13. 3. OKAZAKI, R., AND KORNBERG, A., J. Biol. Chem., 239, 269 (1964). 4. SCHACHMAN, H. K., ADLER, J., RADDING, C. M., LEHYAN, I. R., AND KORNBERQ, A., J. Biol. Chem., 236, 3242 (1960). 5. CHAMBERLIN, M., BALDWIN, R. L., AND BERG, P., J. Molecular Biol., in press. 6. ADLER, J., LEHMAN, I. R., BESSMAN, M. J., SIMMS, E. s., AND KORNBERG, A., Proc. Nall. Acad. Sci. U. S., 44, 641 (1958). 7. LEHMAN, I. R., J. Bio2. Chem., 136. 1479 (1960). 8. LEHMAN, I. R., Roussos, G. G., AND PRATT, E. A., J. BioZ. 9. RAZZELL. W. E.. AND KHORANA. H. G.. J. Biol. Chem., 236, 2. RICHARDSON, c. c., SCHILDKRAUT, c. L., APOSHIAN, H. v., Chem., 157. 819 (1962). 1144 (1961). ' Acad. Sci. U. S., 43, 581 (1957). 10. MESELSON, M., STAHL, F. W., AND VINOGRAD, J., Proc. Natl. 11. IFFT. J.. VOET. P. H.. AND VINOGRAD. J.. J. Phus. Chem.. 66. ,, 1138 (iwi). ' 12. SUEOKA. N.. J. Molecular Biol.. 3. 31 (1961). 13. RAZZELL, W. E., AND KHORAN~, H. G.., J.'Biol. Chem., 134, 14. DEKKER, C. A., AND SCHACHMAN, H. K., Proc. Natl. Acad. Sci. 15. SPAHR, P. F., AND HOLLINGWORTH, B. R., J. Biol. Chem., 236. 16. INMAN, R. B., AND BALDWIN, R. L., J. Molecular Biol., 6, 185 17. STOLLAR, D., AND GROSSMAN, L., J. Molecular Biol., 4, 31 (1962). 18. SCHILDKRAUT, C. L., MARYUR, J., AND DOTY, P., J. Molecular Biol., 3. 595 (1961). 19. RICHARDSON, C. C., SCHILDKRAUT, C. L., AND KORNBERG, A. Cold Spring Harbor spaposia on quantitative biology, Vol. 98, Long Island Biological Association, Cold Spring Harbor, New York, 1963, in press. 20. LEHMAN, I. R., in W. COHN AND J. N. DAVIDSON (Editors), Progress in nucleic acid research, Vol. II, Academic Press, Inc., New York, in press. 2114 (1959). U. S., 40, 894 (1954). 823 (1961). (1962).