THEJOURNAL OF BIOLOGICAL CHEMISTRY VOl. 266, No. 12, Issue of April 25, pp. 7W7892,1991 0 1991 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S. A.

Specificity and Enzymatic Mechanism ofthe Editing Exonuclease of Escherichia coli DNA Polymerase 111"

(Received for publication, December 3, 1990)

Stephan BrenowitzS, Sunye KwackS, Myron F. Goodman$,Mike O'Donnellli, and Harrison EcholsS From the $Division of Biochemistry and Molecular Biology, University of California, Berkeley, California 94720, the §Department ofBiological Sciences, University of Southern California,Los Angeles, California 90089, and the VDepartment of Microbiology, Cornell University Medical College, New York, New York10021

Exonucleolytic editing is a major contributor to the Earlier studieswith a single mismatch indicated that both pol fidelity of DNA replication by the multisubunit DNA 111 core and isolated t have a preference for a mispaired 3' polymerase (pol) I11 holoenzyme. To investigate the terminus (8,12). In this work we have examined in more source of editing specificity, we have studied the iso- detail the specificity and the mechanism of exonucleolytic lated exonuclease subunit, t, and the pol I11 core subas- proofreading. sembly, which carries the 6, 0, and a (polymerase) sub- First, we wanted to determine if editing specificity is a units. Using oligonucleotides with specific terminal property of the exonuclease subunit alone. Based on the mismatches, we have found that both t and pol 111 core crystal structure, the editing specificity of DNA polymerase I preferentially excise a mispaired 3' terminus and therefore have intrinsic editing specificity. For both appears not to be intrinsic to the exonuclease domain, but and pol I11 core, exonuclease activity is much more depends on the transferof single-strand DNA from the poly- effectivewith single-strand DNA; with adouble- merase domain tothe exonuclease domain (13, 14). The strand DNA, the exonuclease is strongly temperature- exonuclease active site must function solely as a single-strand dependent. We conclude that the e subunit of pol I11 exonuclease, because there is no room for duplex DNA (13, holoenzyme is itself a specific editing exonuclease and 15); therefore the exonuclease site makes no direct selection that the source of specificity is the greater melting for a mispaired base. Kinetic studies havesuggested that capacity of a mispaired 3' terminus. editing by pol I is achieved mainly by a delay in elongation from a mismatched primer terminus (16). Thus editing does not result from the intrinsic specificity of the exonuclease, but ratherdepends on a kinetic delay in polymerization, which DNA replication is carried out with extremely high accu- allows more time for transfer of single-strand DNA into the racy. Error frequencies during duplication of the Escherichia exonuclease site. If the editing mechanism suggested for pol I coli genome are 10-9-10"0 per base replicated (1).To achieve was also used by pol 111, the isolated t subunit would not this fidelity, a DNA polymerase must have an exceptional distinguish a correctly paired from a mispaired 3' terminus. ability to discriminate against incorrect base pairs, which may In this work, we have examined the editing specificity of the exhibit only slight structural and energetic differences from t subunit and pol I11 core with a series of correctly paired or the correct base pairs. Fidelity is achieved by a polymerase in mispaired oligonucleotides annealed to bacteriophage M13 a two-step process: (1) base selection, correct selection of the DNA. These DNA substrates provided all 16 possible combi- complementary dNTP during 5' --$ 3' incorporation; (2) ed- nations of correct and incorrect base pairs at the primer 3' iting, 3' + 5' exonucleolytic excision of a noncomplementary terminus. We have found that both pol I11 core and t specifi- deoxynucleotide misinserted at the3' end of a growing DNA cally excise incorrectly paired 3' termini more rapidly than chain.With the additionalcontribution of postreplicative correctly paired termini. mismatch repair, the high fidelity of genome duplication is We also wanted to determine the feature of DNA structure achieved (2-4). that allows the exonuclease to recognize mispairs. There are DNA pol' I11 holoenzyme is the enzyme primarily respon- two plausible general mechanisms for selectivity in editing: sible for chromosomal replication in E. coli and therefore is recognition of departures from the equivalent geometry of the probably the major determinant of the fidelity of genome Watson-Crick base pairs,and melting capacity of the 3' duplication. The pol I11 holoenzyme contains 10 distinct terminus. Geometric recognition has been implicated as the polypeptide subunits: CY,t, 6, T, y, 6, 6', J., x, and p (5,6). The critical determinantfor the specificity of base selection in the t subunit, the dnaQ gene product, is the 3' + 5' proofreading polymerase reaction (4, 17, 18). In the melting model, the exonuclease (7,8). The CY subunit, the dnuE gene product, is exonuclease is intrinsically a single-strandnuclease that pref- the 5' + 3' polymerase (9, 10). The CY, 6, and t subunits erentially removes a misinserted base because the mismatch compose pol I11 core, the smallest subassembly of pol I11 at the 3' end is more often in a single-strand configuration prepared from the holoenzyme (11).By the use of isolated t (19, 20). In the case of pol I, Brutlag and Kornberg (19) subunit and pol I11 core, we can study the mechanism of demonstrated thatthe exonuclease works best on single- editing in the presence and absence of the polymerase subunit. strand DNA and that thereis a marked increase in exonucle- ase activity with duplex DNA as temperature is increased. * This work was supported by National Institutes of Health Grants These data suggest that melting of the 3' terminus is the CA 41655,GM 21422, and GM 38831. The costs of publication of this primary determinant for editing. As noted above, structural article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accord- studies with pol I have also supported a melting model for ance with 18 U.S.C. Section 1734 solely to indicate this fact. editing by pol I. In this study, we have examined the effect of The abbreviations used are: pol, polymerase; N, nucleoside. temperature andstrand specificity on the exonuclease activity

7888 Editing Specificity of E. coli DNA Pol 111 7889 of pol I11 core and e; our data support the melting model for of a correctly paired base. Autoradiographic data are shown editing by pol 111. in Fig. 1 for the exonuclease activity of t on correct and Finally, we have carried out a steady-state analysis of the incorrect base pairs opposite template C. Removal of a 3'- exonuclease activity of pol I11 core to determine the kinetic terminal G from a G. C pair proceeded more slowly than for parameters that provide editing specificity. We have found any of the othermispaired bases opposite C. After 8 min, less that VmmXdiscrimination is the primary kinetic property that than 10% of the correctly paired G was removed, whereas characterizes preferential excision of a misinserted base at more than 90% of the three incorrectly paired bases were the 3' terminus. excised. The autoradiographs were scanned by a densitometer, and thepeaks corresponding to primer bands were integrated. EXPERIMENTAL PROCEDURES The ratio of the intensity of the primer band to the sum of the intensities of all bands present was plotted against time Materials-Purified f subunit was prepared as described previously (8). Purified core subassembly was prepared as described previously using a semi-log scale. By fitting the data to a first order (11). T4 polynucleotide kinase was purchased from New England exponential curve, kinetic rate constants for the exonuclease Biolabs. Oligonucleotide substrates and the 35-mer template were reaction were determined. Examples of the graphed data are synthesized by conventional solid phase methods. Single-strand M13 shown in Fig. 2 for correct and incorrect base pairs opposite DNA was purified from E. coli strain JM103 using published proce- template C (Fig. 2, A and B) and T (Fig. 2, C and data dures (21). [-pR'P]ATP (>5000 Ci/mmol, 10 mCi/ml) was purchased D); from Amersham Corp. are presented for e (Fig. 2, A and C)and for pol I11 core (Fig. Preparation of DNA Substrates-Sixteen oligonucleotide sub- 2, B and D).The rate constants are presented in Table I for strates were annealed to M13 DNA to generate the 16 possible correct all 16 combinations for e and pol I11 core. and incorrect base pairs. Oligonucleotides (20-mers) were synthesized Both e and core preferentially attack a mispaired 3' termi- complementary to four unique M13 sites with a variable 3'-terminal nus, although the relative specificities vary. We observed the base (G,A, T, C). The M13 DNA coordinates were: N .C (3576-3595); greatest specificity for mispairs opposite template C and G, N .T (2223-2242); N. A (5386-5405); N. G (4637-4656). Oligonucleo- tide substrates were labeled at the 5' end with :'lP in a reaction presumably because the correctly paired 3' terminus is so mixture containing 50 mM Tris-HC1 (pH 7.5), 10 mM MgC12, 13 mM resistant to exonucleolytic attack. Conversely, we found the dithiothreitol, 1 p~ [-p3'P]ATP, 20 units of T4 polynucleotide kinase, least specificity for mispairs opposite template A and T, for and varying amounts of oligonucleotide substrate. The reaction was which the correctly paired 3' terminus was relatively sensitive incubated at 37 "C for 80 min and was terminated by heating to to degradation. The major conclusion of our data on exonu- 100 "C for 5 min. Oligonucleotide substrates were annealed to tem- clease specificity is that t has intrinsic discrimination for a plate DNA either at a 1.5:l molar ratio (M13 template in excess) or at a 1:1 molar ratio (oligonucleotide template). Annealing was carried mispaired 3' terminus. Moreover, the properties of the e out by adding NaCl to a final concentration of 50 mM, then heating exonuclease are qualitatively very similar to those of the pol the oligonucleotide template mixture to 85 "C for 10 min, followed by I11 core. slow cooling to room temperature over 2-4 h. Evidence for a Melting Mechanism for t and pol 111 Core- ExonucleaseAssays-For standard exonuclease assays, reaction As noted in the introduction, there are two general mecha- mixtures (10 pl for epsilon and 15 pl for core) contained 40 mM Tris- nisms for exonucleolytic specificity toward a mispaired 3' HCI (pH 7.5), 10 mM MgC12, 5 mM dithiothreitol, 50 pg/ml acetylated bovine serum albumin, and labeled duplex DNA substrate at appro- terminus: geometric recognition and melting capacity. For the priate concentrations (40 nM for c assays and 10 nM for core assays). melting model, there are three clear predictions: (i) and A. T Reactions were initiated by adding enzyme to a final concentration correct pair should be more sensitive to exonuclease than a of 150 nM for or 4 nM for core. Assays were carried out at 37 "c G C pair; (ii) the exonuclease activity on duplex DNA should (except temperature-effect assays which were carried out at 26, 30, be strongly temperature-dependent; (iii) the exonuclease ac- and 37 "C). A new preparation of pol 111 core with a higher specific tivity should be most effective with a single-strand DNA activity was used for the temperature-effect assays. At appropriate time points,aliquots were removed and quenched in formamide/ 5 3' EDTA. Samples were heated to 100 "C for min, and oligonucleotides 5' 32PJ /N differing in length were separated by electrophoresis in 8%polyacryl- amide gels. Densitometry was carried out essentially as described C (18).For measurements of kinetic parametersof exonuclease activity, G opposite C A opposite C reactions (10 pl) contained 40 mM Tris-HC1 (pH 7.5), 10 mM MgC12, 5 mM dithiothreitol, 50 pg/ml acetylated bovine serum albumin, 2 nM Time (min) core, and varying amounts of substrate ranging from 25 to 850 nM. Primer - a I RESULTS Exonuclease Specificity of e and pol Core-To determine 111 'I' opposite C C opposite C the specificity of pol I11 core and the isolated e subunit, we have used a gel electrophoresis assay to measure the rate of Time (min) n I 2 4 8 0 1 2 4 8 exonuclease action. DNA substrates were oligonucleotides, Primer - labeled with ''lP at the 5' end, with a 3' terminus that was either mispaired or correctly paired to theM13 template. The velocity of exonuclease action was measured by electropho- resis of the oligonucleotide on a sequencing gel that resolves FIG. 1. Exonuclease activity off on mispairs opposite tem- plate C. The substrate DNA used for the assay is shown at the top as discrete bands single nucleotide differences in length. As- of figure. The 20-mer oligonucleotides, each containing a different says were carried out with oligonucleotide substrates that 3"terminal base, were labeled with :'2P at the5' end andannealed to provided all 16 possiblecombinations of correct and incorrect single-strand M13 DNA. Exonuclease assays were performed with terminal pairs. The experiments were done under conditions isolated c as described under "Experimental Procedures." Aliquots in which the relative velocity of the exonuclease reactions from the reaction were removed at theindicated times and subjected to gel electrophoresis. The top band in each lane represents the intact should be proportional to the V,,,.,/K,,, specificity parameter oligonucleotide. The lower bands represent oligonucleotide substrates (substrate concentration very much less than Km). that have been degraded by one and two nucleotides, respectively. In every case, exonucleolytic attack on anincorrectly paired Data are shown for exonuclease assays carried out using substrates base at the 3' terminus proceeded more rapidly than removal containing 3"terminal G.C, A. C, T. C, and C.C pairs. 7890 Editing Specificity of E. coli DNA Pol 111 A B

FIG. 2. Specificity of c and core exonuclease for mispaired terminal bases. A and B, activity of e and pol I11 core, respectively, on substrates contain- ing mispairs opposite template C. 3'- 0 2 4 6 8 0 2 4 6 8 terminal bases are: G.C (-, O), A.C Time (min) Time (min) (---, B), T.C (--, 0), andC.C (- -, 0).C and D,activity of e and pol C I11 core, respectively, on 3'-terminal D mispairs opposite template A. 3"termi- nal bases are: A.T (-, O), G .T (-- -, B), T.T (- -, 01, andC.T (- -, 0).

0 2 4 6 8 0 2 4 6 8 Time (min) Time (min)

TABLEI effects, we carried out exonuclease assays at 26,30, and 37 "C. Rate constants for exonuclease of pol III core and isolated t The exonuclease activity with duplex DNA was compared First order rate constants were obtained from an exponential least with that found forthe unpaired single-strand oligonucleotide. squares fit to a plot of substrate remaining versus time, using the Examples of the data are presented in Fig. 3 for assays with equation [SI, = (SI,* e"'eXd) (see F'g.1 2). The kn/kco,,column gives the kexonumbers normalized to thecorrect Watson-Crick pair. pol I11 core with A opposite template T (Fig. 3A) and C opposite template G (Fig. 3B). There is a notable increase in exonuclease activity with increased temperature. A more com- plete set of data is collected in Table 11. There isa very large S" S" thermal effect on exonuclease for both c and pol 111 core for c 5.3 1.0 3.4 1.0 correctly paired G. C and A.T and for the G .T mispair. There G 3633 6.2 11 N*G A is very high exonucleaseactivity with the single-strand DNA 98 19 21 6.1 substrate butonly a small thermal effect. T 76 8.414 29 The data presented in Table I1 demonstrate that theprin- T 34 1.0 15 1 .o ciple predictions of the melting mechanism are fulfilled. G 53 1.6 31 2.1 N-A A Therefore, we conclude that the melting capacity of the 3' 49 1.4 130 8.8 terminus is the primary recognition determinant for the ed- C 53 1.6 130 8.3 iting exonuclease. A 20 1.0 4.2 1.0 Steady-state Kinetic Parameters of Exonuclease Activity: 2.9 14 3.5 G 1457 2.9 V,, discrimination-To assess the relative contribution of N*T 140 7.4 16 3.9 K,,, and V,,, to editing specificity, we carried out a steady- C 2065 3.3 4.7 state kinetic analysis of pol I11 core. A similar analysis was not possible for E alone because the K, value was too high, as G 2.3 1.0 1.9 1 .o noted previously (12). A 35-mer oligonucleotidetemplate was N*C A 49 21 57 30 T 49 20 61 31 used to obtain substrate concentrations higher than those C 46 18 23 11 obtainable with M13. The 35-mer was annealed to oligonucle- otides generating a correct 3"terminal C.G pair and an substrate. For a geometric model, A. T and G .C should be incorrect T. G pair. Conditions were determined that satisfied equivalent, and an especially strong temperature effect would Michaelis-Menten kinetics: reaction velocity wasdirectly pro- not be anticipated. portional to enzyme concentration, and reaction velocity was The datapresented in Table I demonstrate a preference for constant over the time periodmeasured (4 minfor C .G 3' termini in A.T pairs over G.C; this preference is especially substrate and2 min for T. G substrate). pronounced whenc is actingby itself.To examine temperature Reaction velocities were measured at substrate concentra- Editing Specificity of E. coli DNA Pol III 789 1 A

l5 t

0 200 400 tsl (nu) 0 2 4 6 8 FIG. 4. Exonuclease velocity as a function of substrate con- Time (mln) centration. Velocities were measured at various substrate concen- trations using substrates with a 3"terminal correct C. G pair (0)or B a 3'-terminal incorrect T.G pair (0).

TABLEI11 Kinetic parameters of pol III core exonuclease The exonuclease velocity wasmeasured at substrate concentrations ranging from 25 to 850 nM on substrates with either C .G or T.G terminal pairs. Vmaxand K,,, were obtained from a linear least squares fit to an Eadie-Hofstee plot (V/[S] uersus l/[S]). Primer terminus Vm*X K, v,../K, X 10-4

fml/s nM S" 0 e C.G 2.8 k 1.3 420 -+ 80 6.7 T.G 17 4.5 460 290 37 0 2 4 6 8 Time (mln) of a Lineweaver-Burke plot; these values were closely similar to V,,JKm obtained from the Eadie-Hofstee plot (data not FIG. 3. Effect of temperature on pol 111 core exonuclease. shown). The values obtained for K,,, are nearly identical for Exonuclease assays were carried out at 26 "C (D), 30 "C (m), and 3'-terminal C.G and T.G pairs. A similar K, equivalence 37 "C (0).A, substrate containing 3'-terminal G.C pair. B, substrate containing 3"terminal A.T pair. was observed previously by Maki and Kornberg (12) with a different substrate. The VmaXvalue for degradation of a T - G TABLEI1 mispair is about 6-fold higher than VmaXfor degradation of a Temperature effect on exonuclease ofpol III core and c correct C .G pair. This V,,, is the primary kinetic parameter A. Temperature effect on rate constantsfor exonuclease of pol defining editing specificity. 111 core and c" The increased VmaXfor removal of the mismatched T. G supports the concept that the source of editing specificity is k.. X for core kx,,X for t Primer 3' 10' 10' the greater melting capacity of a mispaired 3' terminus. If the terminus 26°C30°C 37 "C 26°C30°C37°C t subunit inpol I11 core binds with similar affinity to a S" S" matched and mismatched terminus, it can be readily shown G.C 1.4 3.6 10 1.2 2.1 3.0 (and is intuitively clear) that V,,, will be proportional to the A.T 2.5 5.0 9.2 40 11 27 fraction of melted 3' ends: G.T 14 33 65 16 31 54 Single-stranded 340280 410 220210 250 vmax = - B. Relative kexodata normalized to sinfle-strand velocity (k.? k,) kx E at each temperature where 12, and 12, are the rate constantsfor formation of melted Primer 3' Pol I11 core e subunit and annealed 3' ends, respectively, 12, is the rate of single- terminus 37°C30°C26°C37°C30°C26°C strand exonuclease activity, and E is the enzyme concentra- G.C 1 2.1 4.8 1 1.6 2.0 tion. Since the fraction of melted 3' termini is greater for A*T 1 2.9 11 1 2.1 4.5 mismatched bases, V,,, values should be higher for mispairs. G*T 1 1.9 3.1 1 1.7 2.7 If the source of editing specificity was direct binding recog-

a Rate constants at26, 30, and 37 "C were obtained as for Table I. nition of a mispaired 3' terminus, discernible K, differences Rate constants were normalized to values at 26 "C for each wouldbe expected. Thus our kinetic dataare completely substrate, then normalized to the velocity of single-strand DNA consistent with a melting mechanism. degradation at each temperature. DISCUSSION tions ranging from 25 to 850 nM. Fig. 4 shows a typical V uersus [SI curve for exonuclease activity on a correctly or Specificity of Exonucleolytic Editing-Recent structural incorrectly paired 3' terminus. Table I11 presents the values work has demonstrated that exonucleolytic editing is a com- for V,,, and K, obtained from Eadie-Hofstee plots (average plex process requiring a communication between different of threeexperiments). Values for the specificity constant domains or subunits of the DNA polymerase. For pol I, the V,,,/K, were also obtained by taking theinverse of the slope exonuclease domain is not only separated from the polymerase 7892 Editing Specificity of E. coli DNA Pol 111 domain by some 30 A, but the exonuclease site can accept The probable explanation for the quantitativediscrepancy only single-strand DNA (13,15,22). Thus polfor I, the editing between editing in vivo and exonuclease in vitro is that “decision” appears to depend entirely on eventsat the polym- intrinsic exonucleasespecificity mustbe augmented by a erase site. Indeed, kinetic studies have indicated strongly thatcontribution from the polymerase site, most likely a kinetic editing specificity in DNA replication is achieved mainly be amplification provided by inefficient polymerization from a delayed DNA chain elongation froma mispaired 3‘ terminus mismatched 3’ terminus. Thisdelayed chain elongationwould (16). This lag in polymerase action presumably allows more allow the exonucleasemore time to act at amispaired 3’ time for a mismatched terminal base to melt and slide into terminus before the next base insertion event (16, 28). As the exonuclease site (14). noted above for polI, this feature is themajor determinant of For pol 111, the exonuclease active site is also presumably editing specificity(16). A similarkinetic amplification of spatially separated from the polymerase site, because the two editing has been noted recentlyfor phage T7 DNA polymerase biochemical activities reside on distinct subunits (8, 10, 12). (29). A modestkinetic amplificationfrom the polymerase step To examine the questionof intrinsic editingspecificity by the would boost the specificity of pol 111 into the range expected exonuclease subunit, we have carried out a detailed compari- from in vivo work. Thus pol 111 probably partitions its speci- son of the exonuclease activity of pol I11 core and t subunit. ficity determinants for editing between the intrinsic discrim- We havefound that t hasintrinsic specificity forany 3‘ ination of the exonuclease site and the kinetic contribution terminus with a mispaired base. The specificity of E is quali- of the polymerase site. tatively similar to thatof pol 111 core. However, the addition of the polymerase subunit markedly increases the activityof REFERENCES the exonuclease for all duplex substrates; thiseffect probably 1. Drake, J. W., Allen, E. F., Forsberg, S. A., Preparata, R., and derives from the contribution of the a subunit to effective Greening, E. 0.(1969) Nature 221, 1128-1131 recognition of the 3’ terminus (12). 2. Kornberg, A. (1980) DNA Replication, W. H. Freeman and Co., San Francisco Mechanism of Exonucleolytic Editing-The initial experi- 3. Loeb, L. A., and Kunkel, T. A. (1982) Annu. Reu. Biochem. 51, ments with polI and T4 polymerase strongly indicated a 429-457 melting mechanismfor exonucleaseaction (19,20).The struc- 4. Echols, H., and Goodman, M. F. (1991) Annu. Reu. Biochem., in tural analysisof pol I defined a surprisingly extreme formof press melting mechanism; the exonuclease site could accept only 5. Kornberg, A. (1988) J. Biol. Chem. 263, 1-4 single-strand DNA (13, 22). Our work with the editing exo- 6. McHenry, C. S. (1988) Annu. Rev. Biochem. 57,519-550 7. Scheuermann, R., Tam, S., Burgers, P. M. J., Lu, C., and Echols, nuclease of pol 111 strongly supportsa melting mechanismfor H. (1983) Proc. Natl. Acad. Sci. U. S. A. 80, 7085-7089 both pol I11 core and the isolated t subunit. Thus the major 8. Scheuermann, R. H., and Echols, H. (1984) Proc. Natl. Acad. Sci. difference between pol I and pol I11 appears tobe the intrinsic U. S. A. 81, 7747-7751 editing specificity of t itself. As noted below, this intrinsic 9. Welch, M. W., and McHenry, C. S.(1982) J. Bacteriol. 152,351- specificity is unlikely to be sufficient to explain the complete 356 contribution of editing to thefidelity of replication by pol 111. 10. Maki, H., and Kornberg, A. (1985) J. Biol. Chem. 260, 12987- 12992 However, this feature of the exonuclease probably allows pol 11. McHenry, C. S.,and Crow, W. (1979) J. Bid. Chem. 254,1748- 111 to achieve the very high editing precision needed for an 1753 enzyme dedicated to chromosomal duplication. Althoughwe 12. Maki, H., and Kornberg, A. (1987) Proc. Natl. Acad. Sci. U. S. A. do notknow whether a melting mechanismwill be a universal 84,4389-4392 property of editing exonucleases, the idea is attractive, be- 13. Steitz, T. A., Beese, L., Freemont, P. S., Friedman, J. M., and

Sanderson. M. R. (1987).. Cold SDrinp” Harbor Symp.~. Quant.~ Bid. cause a melting mechanismallows a ready evolutionary tran- 52,465-472 sition from a free single-strand nuclease to an editing subunit 14. Joyce, C. M., Friedman, J. M., Beese, L., Freemont, P. s., and or domain designed to excise misinserted nucleotides. Steitz. T. A. (1988) in DNA Replication and Mutagenesis Contribution of Editing to Replication Fidelity-An impor- (Moses, R. E., and Summers, W. C.,-eds)pp. 220-226, American tant biological question is whether the intrinsicspecificity of Society for Microbiology, Washington, D. C. the exonuclease is sufficient to explain the contribution of 15. Derbyshire, V., Freemont, P. S., Sanderson, M.R., Beese, L., Freedman, J. M., Joyce, C. M., and Steitz, T. A. (1988) Science editing to the fidelity of DNA replication. An accurate esti- 240, 199-201 mate of the editing concentration in vivo is complicated by 16. Kuchta, R. D., Benkovic, P., and Benkovic, S.J. (1988) Biochem- the interplay of fidelity systems (4). The mostdefective mu- istry 27,6716-6725 tations in the dnaQ gene codingfor t confera very large 17. Echols, H. (1982) Biochimie 64,571-575 increase in mutation rate, up to 104-105-fold (23, 24). How- 18. Sloane, D. L., Goodman, M. F., and Echols, H. (1988) Nucleic ever, this number is an overestimate, because the mismatch Acids Res. 16, 6465-6475 19. Brutlag, D., and Kornberg, A. (1972) J. Biol. Chem. 247, 241- repair system becomes ineffective at high mutation rates (25, 248 26). The best currentguess forthe contributionof exonucleo- 20. Bessman, M. J., and Reha-Krantz,L. J. (1977) J. Mol. Biol. 116, lytic editing in vivo isin the 102-103 range (4, 25). This 115-123 number is in the same rangeas a rough estimate obtained for 21. Messing, J. (1983) Methods Enzymol. 101, 20-78 pol 111 in vitro by comparing theoverall fidelity of pol I11 with 22. Ollis, D. L., Brick, P., Hamlin, R., Xuong, N. G., and Steitz, T. base insertion data for pol I (27). A. (1985) Nature 313,762-766 23. Degnen, G. E., and Cox, E. C. (1974) J. Bacteriol. 117,477-487 Our work on theexonuclease specificity of pol I11 core and 24. Cox, E. C., and Horner, D. L. (1982) Genetics 100,7-18 t indicates a much lower intrinsicdiscrimination for the 25. Schaaper, R. M. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 8126- editing exonuclease than the editing contribution to DNA 8130 replication inferredin vivo. Our datain Table I were obtained 26. Schaaper, R. M., and Radman, M. (1989) EMBOJ. 8,3511-3516 under conditions in which the relative rates of exonuclease 27. Fersht, A. R., Knill-Jones, J. W., and Tsui, W.-C. (1982) J. Mol. Biol. 156, 37-51 activity (kex,,)should represent the relative specificities for 28. Mendelman, L. V., Petruska, J., and Goodman, M. F. (1990) J. terminal nucleotides ( VmaX/Km).This conclusion is supported Biol. Chem. 265,2338-2346 by the detailed kinetic dataof Table 111. For pol I11 core, these 29. Donlin, M. J., Patel, S. S., and Johnson, K. A. (1991) Biochem- specificity ratios range from 2- to 30-fold. istry, 30,538-546