THE JOURNAL OF BIOLOQICAL CHBYIBTBY Vol. 247, No. 1, Issue of January 10, pp. 241-248, 1972 Printed in U.S.A. Enzymatic Synthesis of Deoxyribonucleic Acid XXXVI. A PROOFREADING FUNCTION FOR THE 3' + 5' EXONUCLEASE ACTIVITY IN DEOXYRIBO- NUCLEIC ACID POLYMERASES* (Received for publication, July 16, 1971) DOUGLAS BRUTLAGS AND ARTHUR KORNBERG From the Department of Biochemistry, Stanford University School of Medicine, Stanford, California 94305 SUMMARY The Escherichia coli and T4 DNA polymerases do not ex- tend chains in which the 3'4erminal nucleotide (primer terminus) is not paired with the template. By using syn- thetic double-stranded polynucleotides, the 3' -+ 5` exonu- clease function of these polymerases was shown to be di- rected specifically against a mispaired or unpaired primer terminus. Chain extension of such termini begins only after all mispaired nucleotides have been removed and a base- paired terminus is reached. The latter is completely con- served while polymerization is maintained. These results suggest that the function of the 3' --* 5` exonuclease activity of DNA polymerases is to remove mispaired nucleotides which have been incorrectly incorporated, thereby increasing the fidelity of template copying. The function of other E. coli exonucleases suggested by their specificities on poly- nucleotide substrates are trimming loose ends of DNA for exonuclease I and enlarging nicks and gaps within helical regions for exonuclease III. An exonuclease activity which degrades DNA chains in the 3' to 5' direction is a component of DNA polymerase isolated from Escherichia coli (1). The 3` -+ 5' exonuclease is also closely associated with and appears to be part of DNA polymerases induced in E. coli by infections with bacteriophages T2, T4, and T5 (2-4). The specificity and possible functions of this nuclease have been obscured by the presence in E. coli DNA polymerase of another and even more active exonuclease activity that de- grades chains from 5' to 3`. Recently it has become possible to separate and physically isolate these two nuclease activities by proteolytic cleavage of DNA polymerase (5, 6). From the parent polypeptide chain (mol wt 109,000), two active fragments were isolated: the small fragment (mol wt 36,000) contains the 5' -+ 3' exonuclease activity; the large fragment (mol wt 75,000) contains the 3' + 5' exonuclease activity as well as the polym- erizing functions (5, 6). * This study was supported in part by grants from the National Institutes of Health (United States Public Health Service) and the National Science Foundation. The previous paper in this series is Reference 33. # Predoctoral Fellow of the National Science Foundation. The studies reported here on the specificity of the 3' 4 5' exonuclease support earlier speculations (3,7,8) that this activity may be designed to remove mispaired nucleotides which have been incorrectly incorporated at the 3'-hydroxyl end of the double-stranded DNA and thereby serves a corrective function.' By the use of specifically labeled homopolymers and block copolymers as substrates we show in this report that under polymerising conditions a mispaired primer terminus is quan- titatively removed by the DNA polymerase whereas a correctly base-paired primer terminus is completely conserved. It thus appears that the 3' -+ 5` exonuclease may serve a proofreading role in removing mispaired nucleotides at the growing end of a chain and that the polymerase will not extend a chain until the mispaired 3` terminus is removed. The availability of these polynucleotide substrates has enabled us also to examine the specificities of the E. coli exonucleases I and 111. These nucleases appear to perform complementary roles in trimming single-stranded ends of DNA chains and in enlarging nicks and gaps within helical regions of double-stranded DNA. MATERIALS Nucbotides-Unlabeled deoxyribonucleotides were purchased from P-L Biochemicals except for dUTP which was purchased from Calbiochem. dITP was a gift from Dr. M. Fikus and was prepared by deamination of dATP as described by Inman and Baldwin (9). [aH]dTTP and taH]dATP were purchased from Schwarz BioResearch. iaH]dUTP was purchased from Amersham-Searle and was purified before use by paper chro- matography (Schleicher and Schuell, No. 589 orange ribbon) using an isopropyl alcohol-concentrated NHIOH-water (7 : 1 :2) solvent. [CY-~*P]~TTP was purchased from ICN and was checked for radiochemical purity by paper electrophoresis (20 mu sodium citrate buffer, pH 3.5). ["CIdTTP was purchased from New England Nuclear. All specific activities were determined by spectral measurement and radioactivity measured on Whatman GF/C glass filters in a Nuclear Chicago scintillation counter. 2',3'-Dideoxythymidine triphosphate and the 6`-deoxy-6'- homothymidine pyrophosphoryl phosphonate were gifts from Dr. J. G. Moffatt. 8,y-dTTP methylene diphosphonate was purchased from Miles Laboratories. Enzymes--Micrococcal nuclease and spleen phosphodiesterase were purchased from Worthington Biochemicals. Alkaline L. E. Orgel and F. H. C. Crick, personal communication. 241 242 Enzymatic Synthesis of DNA. XXXVI Vol. 247, No. 1 phosphatase was isolated and purified according to the method of Malamy and Horecker (10). DNA polymerases induced by T4 umN82 and T4 umN82 tsL56, the gifts of Dr. W. M. Huang, had been purified as described (3, 11). E. coli DNA polymerase was purified as described (12). The large fragment obtained by proteolytic cleavage of this enzyme appeared in the purification procedure of the native DNA polymerase and was also purified to homogeneity (6). All DNA polymerases used in this study yielded a single protein band on sodium dodecyl sulfate-polya- crylamide gel electrophoresis. E. coli exonuclease I was purified through the hydroxylapatite stage (specific activity of 15,000 units per mg) (13). E. coli exonuclease I11 had a specific activity of 130,000 units per mg (12). Terminal deoxynucleotidyltrans- ferase (terminal transferase) purified from calf thymus according to Kat0 et al. (14) was a gift from Dr. F. N. Hayes (Los Alamos Scientific Laboratory, Ins Alamos, New Mexico). METHODS Polynucleotides-The polynucleotides d(T)260 and d(I)~s were prepared with terminal transferase as described (15) except that 20 rnM potassium phosphate buffer (pH 7.0), 20 mM potassium cacodylate buffer (pH 7.0), and 8 mM MgClz were used for the polymerization of dITP. d(A)mo was prepared according to Riley, Maling, and Chamberlin (16) and d(C)looo according to Chamberlin and Patterson (17). Terminally Labeled Polynucleotides-Polynucleotides, radio- activity labeled at their 3'4ermini with various nucleotides d(T)z60-[SH]d(U) 1.8, and d(1)~,-[~H]d(T) were synthesized by incubating labeled triphosphates with unlabeled polynucleotides and terminal transferase. To add a purine nucleotide, dATP for example, to d(T)Z60, 0.3 mu d(T)26Oa was incubated with 5 /.LN (3HIdATP (17.7 Ci per mmole), and 0.15 mg per ml of terminal transferase in 20 mN potassium phosphate buffer (pH 7.0), 20 mM potassium cacodylate buffer (pH 7.0), 8 mM MgC12, and 1 mM P-mercaptoethanol at 37' for an amount of time (predeter- mined by a small scale reaction) sufficient to add about 1 mole of nucleotide per mole of polynucleotide chain. Incorporation was monitored by adsorption to DEAE-paper as described below. To add a pyrimidine nucleotide, 0.1 M potassium phosphate buffer (pH 7.0), 0.1 M potassium cacodylate buffer (pH 7.0), and 0.25 mM CoC12 were used. The reaction was terminated by cooling and adding 50% KOH to a final concentration of 0.3 M. After 10 min at O', the mixture was neutralized with 85% Hap04 and dialyzed extensively against 1 M NaCl, 10 mM Tris butrer (pH S.O), and then uerm 10 mM Tris buffer (pH 8.0) to remove the salt. The final average number of residues added was determined from the specific activity of the added nucleotide, the length of the polynucleotide, and the radioactivity present in the labeled polymer per mole of total nucleotide as determined by ultraviolet absorption. Determination of Distribution of Labeled Residues-Hayes et al. (18) have shown that limited addition of nucleotides to short oligonucleotide primers by terminal transferase gives a Poisson * Polynucleotides labeled at the 3' terminus are abbreviated as d(T)260-[aH]d(C)l,~, where 260 gives the average number of residues in the polymer of unlabeled thymidylate to which has been added on the average 1.1 moles of aH-labeled deoxycytidylate per mole of polymer. 8 Polynucleotide concentrations are expressed in terms of moles of total nucleotide. (d(T)260-[~H]d(T)o.s, d(T)tso-['Hld(C)~~J ~(T)~BO-[*HI~(A)O.Q, distribution of added residues. This conclusion was found to apply also to the longer polynucleotide primers used in this study. This determination was made from an independent measure of the distribution of nucleotides as well CM the average number of residues added per chain. Degradation of the labeled polymers by micrococcal nuclease and spleen phosphodiesterase converts all the internal nucleotides to 3'-nucleotides and the 3'-terminal residue to a nucleoside. The ratio of label converted to nu- cleoside to the total label found in the nucleoside and 3'- nucleotide, combined, gives the fraction of the labeled residues at the 3`-terminal position (FJ. For this analysis, 3 nmoles of each of the terminally labeled polynucleotides were degraded with micrococcal nuclease and spleen phosphodiesterase and the products were separated by chromatography as described by Wu and Kaiser (19). The hydrolysis converted >98% of the total label to nucleoside and nucleotide. Table I summarizes the average number of labeled residues and the fraction of these residues which was terminal. The observed fraction of terminal residues was in good agreement with that predicted by a Poisson distribution. It can be demonstrated that this analysis is formally equivalent to showing that the fraction of chains to which no labeled residues were added conforms to the Poisson distribution. Determination of Polynucleotide Length-The number-average length of a polynucleotide was determined by end group labeling analysis as described by Weiss, Live, and Richardson (20). Nuclease Assays-Nuclease assays were performed in 0.2 rnl containing 2 nmoles of [aH]d(T)~~O, with or without 5 nmoles of unlabeled d(A) 4000, 0.05 M N-hydroxyethylpiperazine-Wethane- sulfonate buffer (pH 7.4), and 5 m~ MgC12. The mixture was was heated to the appropriate temperature and a 20.~1 sample was taken and applied to a 1.5-cm square of Whatman DE-81 paper. Then 10 J of enzyme diluted in bovine serum albumin (1 mg per ml), 0.05 M HEPES4 buffer (pH 7.4), and containing less than 1 pmole of enzyme were added to the reaction; 20.~1 samples were removed at appropriate intervals and adsorbed onto 1.5 ern squares of DE-81 paper. The squares were washed three times by gentle agitation for 5 min in 100 ml of ammonium formate, 0.3 M (adjusted to pH 7.8), and dehydrated by two washes in 95% ethanol and one in anhydrous ether. The squares were air dried and the amount of radioactive label re- maining in polynucleotide determined in a scintillation counter. (Polymerization was also measured with this technique by using labeled deoxyribonucleoside triphosphates.) Nuclease rates were determined from a least squares fit to initial linear points of nucleotide hydrolyzed plotted against time. Rates were determined from only those assays in which >20% of the polynucleotide had been degraded and in which the rate was linear for at least 3 time points. RESULTS S' + 6` Exonuclease Rates of DNA Polymerases: Influence of Secondary Structure and Temperature-The nuclease rate of the large fragment (a proteolytic fragment of the E. coli enzyme containing only polymerase and 3` -+ 5' exonuclease activities) on d(T)300 was 4fold greater with the polymer in a 4 The abbreviations used are: HEPES, N-2-hydroxyethyl- piperazine-N'-2-ethanesulfonic acid; ddTl'P, 2',3'-dideoxythy- midine triphosphate; dTMPPCP, B,-pdTTP methylene diphos- phonate. Issue of January 10, 1972 D. Brutlag and A. Kornberg 243 I I I a NodTTP TABLE I Distribution of terminal residues on synthetic polynucleotides I I I b No dTTP 1 Polynucleotide composition d(T)zso-['H]d(T) o .86 d(T) zao-['HId(C)1.11 d(T)~aa-[~Hld(A)0.8T d(T) zao-['HId (U)I .81 d(I)zes-['€IId(T)~ .6t 6 The fraction of labeled residues which are at the 3'-OH termi- nal position of a chain (F,) was determined by nuclease degrada- tion. Specific activities of the labeled nucleotides were from 7 to 20 Ci per mole. b Ft = (1 - e-r)/r, for a Poisson distribution of labeled residues where r is the average number of labeled residues per chain. Fraction of labeled residues at the 3' terminus Observeda Poissonb ______._ 0.54 0.67 0.56 0.60 0.50 0.46 0.47 1 0.49 0.42 0.67 TABLE I1 Injuence of temperature on 9' -+ 6' exonuclease rates of DNA polymerase acting on single- and double-stranded polynucleotides 1 Substrate I 19 1100 ' 850 790 1 I 4.8 350 27 11 DNA polymerase Double- stranded Temperature E. coli large fragment T4 wild typeo T4 wild type" T4 L56" 37" 37 30 30 ap-Mercaptoethanol (10 mM) was included in assays of the T4 enzyme. single-stranded as compared to a double-stranded conformation (d(T) $O@:d(A) 4000) (Table 11). A similar preference for single- stranded d(T)sao was observed at 37" with the T4 enzyme which has a much higher nuclease activity than the E. coli enzyme. When the nuclease rates of the T4 polymerase on single- stranded d(T)3oo at 30" are compared with the rates at 37" only a slight reduction in rate (23%) was found. However, there was a 13-fold reduction in rate on the double-stranded substrate (Table 11). A similar marked temperature dependence of the nuclease rate of the E. coli large fragment on double-stranded substrates has also been observed.6 This temperature depend- ence indicates a high energy of activation for some rate-limiting step on double-stranded substrates, possibly local denaturation at the 3' terminus. Thus the 3' 3 5' exonuclease might require that the terminal nucleotides not be base-paired (i.e. frayed). Nucbase Action of DNA Polymerases on Primer Terminus- Polynucleotides containing on the average about one radioactive nucleotide residue per chain at the 3' terminus were synthesized employing terminal transferase (Table I). The terminal residue was either the same as the rest of the polymer or distinctive from it. The distribution of labeled residues added to the polynucle- otide chains approximates a Poisson distribution. For simplic- ity, these 3' terminally labeled polymers are referred to with a 6 P. Setlow and A. Kornberg, unpublished results. FIG. 1. Specificity of the 3'- 5' exonuclease action of the large fragment of the Escherichia coli DNA polymerase on terminally labeled polynucleotides. An excess of large fragment (5.4 pmoles) was incubated at 37" under standard assay conditions with the indicated polymers and triphosphates. Samples were taken at indicated times. a, 0.6 nmoles of d(T)~,o-['H]d(T)l with and without 5.4 nmoles of d(A)roao; b, 0.66 nmole of ~(T)~so-[~H]~(C)I with and without 5.4 nmoles of d(A),000; c, same as a with d(A)raoo and with 50 p~ [dZP]dTTP(18,000 cpm per nmole) ; d, same b with d(A)tooo and with 50 pd [o(-32P]dTTP. subscript "1" even though the actual average number of labeled residues only approximates unity. The rate of removal of the terminal label from ~(T)PBo-[~H]~(T)~ by the large fragment was too great to permit accurate rate measurements (Fig. la). However, the decrease innuclease rate with the polynucleotide in a double-stranded conformation was large enough to be apparent. Thus the rate of hydrolysis of these terminal residues reflects the preference for single-stranded substrate initially observed with uniformly labeled polynu- cleotides. The terminal residue of d(T)2eo-[WJd(C) I, which does not base pair with d(A)4ooa, was removed at a rapid rate both in the single and the double-stranded conformation (Fig. lb). Although the relative rates of hydrolysis of dCMP in these two cases cannot be determined, the rate of removal of dCMP from d(T)260-[3H]d(C)1:d(A)rooo was far more rapid than that of dTMP from d(T)zso-[8Hld(T)1:d(A),rm (compare Fig. 1, a and b). Thus the preference of the nuclease for nucleotides which are not able to base pair with the template is readily apparent. Removal of Terminal Nucleotides Prior to Polymerization- Polymerization had a dramatic effect on the 3' ---f 5' exonuclease activity. When the large fragment and the polymer d(T)PBO- ['H]d(T), :d(A)dOOO were incubated with dTTP so that polymeri- zation could proceed, there was no detectable loss of terminal la- Vol. 247, No. 1 Enzymatic Synthesis of DNA. XXXVI FIQ. 2. 3' + 5' exonuclease action of the large fragment on the terminal residues of d (T)lao-[PzP]d(T)s-[aH]d(C) 16' d(A)4000. The large fragment (5.4 pmoles) was incubated under standard as- say conditions at 37" with the following; a, 0.75 nmole of d(T)2so- [a2P]d(T)I-[aH]d(C)la and 2.0 nmoles of d(A)mo; b, same &s a but with 50 p~["c]d'M'P (10,400 cpm per nmole). bel during the incorporation of d"TP (Fig. IC). The incorpora- tion of dTTP completely protected the terminal nucleotides from exonuclease action. In a similar experiment in which d(T)ZBO- [aH]d(T)l was present in a single-stranded conformation, there was no detectable incorporation of dTTP nor did the dTTP protect the terminal nucleotides from nuclease. Thus both the template and the triphosphate which participate in po- lymerization are required for protection of the terminal residues. A third requirement for the protection of a terminal residue from exonuclease action is that it be base-paired to the template. When d(T)~ao-[~H]d(C)~:d(A)400o was incubated with polym- erase and dTTP, the terminal dCMP residue was removed rapidly after which polymerization proceeded normally with the primer that remained (Fig. Id). After extensive polymerization little or no [aH]dCMP was detectable in the product, indicating that the DNA polymerase does not polymerize on a mispaired 3' terminus; instead the enzyme first removes the non-base- paired nucleotides by 3' + 5' exonuclease action. In order to investigate further how strict is the base pairing specificity of this exonuclease during polymerization, a copolymer was synthesized containiing a block of 5 [c+P]dTMP residues subterminal to a block of 16 PHIdCMP residues (d(T)~o-[~P]d- Qs-[aH]d(C)~~:d(A),o~o). The fate of the terminal block of 16 non-base-paired dC residues and the subterminal block of base- paired dT residues were both observed. In the absence of dTTP, the enzyme (large fragment) removed the terminal dCMP resi- dues at a maximum rate of 20 nucleotides per min per chain, whereas the dTMP residues were removed only after a significant lag and then only at a maximum rate of 4 residues per min per chain (Fig. 2u). The rate difference is not immediately appar- ent from Fig. 2a because the plot of the percentage of residues remaining does not reflect that there are initially three times as many dCMP residues as dTMP residues. When dTTP was included so that polymerization could proceed, the terminal dCMP residues were still removed as rapidly but the subterminal, base-paired dTMP residues were retained (Fig. 2b); loss of dTMP was less than 0.5 residue per chain. As expected, the polynucleotide supported synthesis only after a lag, during which the dCMP residues were removed. This experiment indicates TABLE I11 Inhibition oj 3' -+ 5' exonuclease is specific for deozyribonucleoside triphosphates which are incorporated The deoxyribonucleoside triphosphate, 50 pra, was incubated with 0.6 nmole of d(T),,o-[*Hld(T)~, 2.0 nmoles of d(A)tooo, and 5.4 pmoles of large fragment at 37" under standard assay condi- tions for 4 min. Incorporation was followed using 14C- or **P- labeled triphosphate (10,OOO cpm per nmole) except where noted. I I TriphosphFte Labeled primer terminus incorporationD remainin2 Nucleoside triphosphate added dATP dGTP dCTP Homophosphonate' dTMPPCP dTTP dUTP ddTTP None None None Noned Noned Normal Normal One % o/ initial <5 <5 <5 <5 <5 100 95 97 6 "None" indicates no incorporation detectable (less than resi- due per chain); "nor nal" indicates rapid chain extension; "one" indicates a single residue incorporated per chain (22). Values of <5% indicate that both the rate and the extent of hydrolysis were the same as a control without any deoxyribonu- cleoside triphosphate added. c 6'-Deoxy-6'-homothymidine pyrophosphoryl phosphonate. d The inability of DNA polymerase to incorporate these ana- logues was shown by their inability to replace dTTP in a poly- [d(A-T)] primed reaction (Brutlag and Kornberg, unpublished results). nucleotide, but also the inability of the polymerase to utilize such a nucleotide as a primer terminus for polymerization. Inhibitioa of S' -+ 6' Exonuclease by Deoxyribonucleoside Triphosphates-The previous expcrimcnts demonstrated that dTTP in an exonuclease assay with the base-paired template primer d(T)~~~-[~wd(T) :d(A)tooo completely inhibited the 3' -+ 5' exonuclease activity of the DNA polymerase on terminal dTMP residues. Incorporation of nucleotides on a terminally labeled polynucleotide covers the labeled residue and protects it from exonuclease action. In order to determine whether incor- poration of the triphosphate is required for protection of a ter- minal nucleotide, other deoxyribonucleoside triphosphates were tested for their ability to inhibit exonuclease (Table 111). Nu- cleotides not complementary to the template d(A)cooo (i.e. dATP, dGTP, and dCTP) were not incorporated and also failed to in- hibit exonuclease. The phosphonates, dTMPPCP and 6'- deoxy-6'-homothymidine pyrophosphoryl phosphonate, are analogua of dTTP, which bind to the triphosphate-binding site on the enzyme (21),6 and have the same base pairing specificity as dTTP. Neither analogue is incorporated by DNA polym- erase nor does either analogue inhibit exonuclease. Three nu- cleotides which were incorporated (dTTP, dUTP, and ddTTP) all inhibited exonuclease (Table 111). Since ddTTP lacks a 3'-hydroxyl group only 1 residue is incorporated per chain (22). Nevertheless, this limited incorporation was sdcient to protect essentially all of the terminal nucleotides from exonuclease. Secondly, ddTTP is incorporated at a rate The behavior of ddTTP is of interest in two respects. not only the specscity of the 3' -+ 5' exonuclease for amispaired 8 D. Brutlag and A. Kornberg, unpublished results. Issue of January 10, 1972 D. Brutlag and A. Kornberg 245 E. coli large fragment.. .. T4 wild type. ........... T4 W6.. ................ TABLE I\' Yispairing of 3'-terminal nucleotide as basis of nuclease action by large fragment and TQ DNA polymerase An excess of DNA polymerase (4 to 5 pmoles) was incubated with 0.6 nmole of primer polynucleotide d(T)m (first three col- umns) or d(I)%Ba (fourth column) labeled at the 3' terminus with the designated (*) nucleotide and annealed to 5 pmoles of a tem- plate (d(A)tooo or d(C)lm). dTTP or dGTP (50 p~) was present, as indicated. Incubation was 4 min under standard assay condi- tions except that 10 mM p-mercaptoethanol was included in assays of the T4 polymerase. The temperature was 37" except as noted. yo of indial labeled primer terminus remaining 98 3 2 3 97 100" --&OH --TT