DNA replication fidelity: kinetics and thermodynamics

Mutation Research, 200 (1988) 11-20

11

Elsevier MTR 02322

DNA replication fidelity: kinetics and thermodynamics M.F. Goodman Molecular Biology Section, Department of Biological Sciences, University of Southern California, Los Angeles, CA 90089-1481 (U.S.A.) (Accepted 3 March 1988)

Keywords: DNA synthesis fidelity; DNA polymerase kinetics; Base-pairing thermodynamics

Summary Mechanisms that control the fidelity of DNA replication are discussed. Data are reviewed for 3 steps in a fidelity pathway: nucleotide insertion, exonucleolytic proofreading, and extension from matched and mismatched 3'-primer termini. Fidelity mechanisms that involve predominately K m discrimination, Vmax discrimination, or a combination of the two are analyzed in the context of a simple model for fidelity. Each fidelity step is divided into 2 components, thermodynamic and kinetic. The thermodynamic component, which relates to free-energy differences between right and wrong base pairs, is associated with a K m discrimination mechanism for polymerase. The kinetic component, which represents the enzyme's ability to select bases for insertion and excision to achieve fidelity greater than that available from base pairing free-energy differences, is associated with a Vmax discrimination mechanism for polymerase. Currently available fidelity data for nucleotide insertion and primer extension in the absence of proofreading appears to have relatively large K m and small Vmax components. An important complication can arise when analyzing data from polymerases containing an associated 3'-exonuclease activity. In the presence of proofreading, a Vmax discrimination mechanism is likely to occur, but this may be the result of two K m discrimination mechanisms acting serially, one for nucleotide insertion and the other for excision. Possible relationships between base pairing free energy differences measured in aqueous solution and those defined within the polymerase active cleft are considered in the context of the enzyme's ability to exclude water, at least partially, from the vicinity of its active site.

Mutations in DNA are the source of a wide variety of biological effects having both positive and negative consequences. Mutations play a positive role as a driving force behind evolution. The origin of mammalian defense against disease involves the synthesis of antibodies, an extraordinarily diverse group of molecules. Antibody synthe-

Correspondence: Dr. M.F. Goodman, Molecular Biology Section, Department of Biological Sciences, University of Southern California, Los Angeles, CA 90089-1481 (U.S.A.).

sis requires large-scale DNA sequence rearrangements and is accompanied by hypermutation, characterized by an abnormally high frequency of base-substitution mutations in the variable region of the immunoglobulin genes (Malipiero et al., 1987). Negative consequences of mutations include single nucleotide substitutions in DNA responsible for oncogene activation (Marshall et al., 1984) and for debilitating metabolic diseases such as cycle-cell anemia and Lesch-Nyhan syndrome (Kelley and Wyngaarden, 1972). Considering the central role that mutations play in the complex

0027-5107/88/$03.50 © 1988 Elsevier Science Publishers B.V. (Biomedical Division)

12 patterns of evolution and survival, an understanding of mutagenesis at the molecular level is vital to understanding cell behavior. Base-substitution mutations may arise during DNA replication and repair as a result of mistakes made by DNA polymerase in copying a D N A template strand. An important question is to try to understand how DNA polymerase selects a nucleotide for insertion into DNA. What portion of nucleotide insertion fidelity can be attributed primarily to the polymerase and what portion to differences in hydrogen bonding interactions for complementary compared to non-complementary base pairs? Nucleotide misinsertion frequencies, based on in vitro data using purified DNA polymerases, are typically in the range 10- 3_ 10- 5 (Loeb and Kunkel, 1982). This 100-fold range of variability may result from some polymerases being generally more accurate than others and from the possibility that different enzymes might exhibit unique base mispairing specificities. For many DNA polymerases, particularly those obtained from procaryotes, nucleotide insertion accuracies can be markedly improved by the presence of an associated proofreading exonuclease (Brutlag and Kornberg, 1972; Muzyczka et al., 1972). This exonuclease is capable of removing mismatched bases before the polymerase can continue adding nucleotides following the error. The same questions posed concerning the mechanisms of insertion fidelity are also valid for the case of exonuclease excision fidelity; how much of the specificity of error correction might be accounted for by differences in stability between correct and incorrect base pairs, and how much by enzyme selective proofreading? Is there an additional component of recognition residing within the proofreading exonuclease enabling it to remove melted out mispairs with greater specificity than melted out base pairs? A polyacrylamide gel assay has recently been developed to measure the fidelity and kinetics of nucleotide insertion and mismatch extension at any site along a D N A template strand (Randall et al., 1987; Boosalis et al., 1987). Data from the assay can be used to address the important effects of neighboring bases on fidelity. Questions of enzyme mechanisms of insertion and excision fidelity will be discussed in the context of a simple

model for D N A synthesis (Galas and Branscomb, 1978; Clayton et al., 1979; Goodman and Branscomb, 1986) which treats nucleotide insertion and excision as competing reactions acting on annealed and melted our primer termini respectively. Nucleotide insertion is assumed to require the presence of annealed 3'-primer termini while nucleotide excision takes place from melted-out primer termini. The model is useful in being able to delineate fidelity components attributable primarily to differences in stability between correct and incorrect base pairs from those involving polymerase and exonuclease base selection.

Replication fidelity model A simple model for D N A synthesis is shown in Fig. 1. It contains two branches; nucleotide insertion occurs in the lower branch and exonucleolytic proofreading in the upper. The insertion step is assumed to proceed exclusively from the state A where the 3'-primer terminus is fully annealed. The excision step takes place exclusively from the state M where the 3'-primer terminus is melted out. At each stage of the polymerization process, direct competition is occurring between insertion of the next nucleotide or removal of the previous nucleotide. After a nucleotide is inserted or excised, repositioning of the enzyme is assumed to occur only after states A and M reach equilibrium. If a correct nucleotide is inserted, the enzyme is more likely to cycle to state A than to state M in accordance with the equilibrium constant K a = k a / k m. Conversely, if an incorrect nucleotide is inserted, the enzyme has a higher probability to be repositioned in the melted state. The fidelity of D N A synthesis depends on the kinetics of the competing insertion and excision reactions. The activity of the proofreading exonuclease is defined by the rate constant k x. If k x is large, then the exonuclease may succeed in removing a terminal melted nucleotide from state M before reannealing occurs. If reannealing has occurred and a d N T P substrate is available, then the polymerase can insert a nucleotide from state P. A summary of 3 important properties of the model follows:

13

kx

l

Excision

\

\

/

" ~

/

\

,,:,'ilk.

/

/

/ //"

"-..~

/

-~

t~

//i/ // I/

•L i

km

/

\-Z/ /

/

(a)

~

~

(A)

k.1

k~

°° kcat

Insertion [ _ _ o . L ~ ~

(p)

Fig. 1. States and transitions of the simple kinetic model. In state A, the d N T P binding site is empty and the terminal base is in its base-pairing position. A transition out of state A occurs by either the association of a d N T P (transition to state P, rate constant kl) or dissociation of the terminal base from its base pairing position (transition to state M, rate constant km). A transition from state M occurs by either the excision of the 3'-terminal base (rate constant kx) or reassociation of the terminal base to its pairing configuration (transition to state A, rate constant ka). A transition from state P occurs by either the formation of phosphodiester bond (rate constant kcat) or the dissociation of the d N T P (transition to state A, rate constant k _ l ) . Following either excision or insertion, a shift occurs one base backward or forward to allow the cycle to repeat. When cycling occurs, the terminal base is assumed to be in an equilibrium distribution between states A and M.

(1) All base pairs are created equal No intrinsic distinction need be made between right and wrong d N T P substrates or matched and mismatched 3'-primer termini. Both properly and improperly paired nucleotides can either be melted out or annealed. The sole difference between right and wrong primer termini is the magnitudes of

their respective equilibrium constant, K a. A previously inserted correct nucleotide is most likely to be annealed, so that K a >> 1. Following insertion of a correct nucleotide, the enzyme will be more likely to cycle to state A to add another nucleotide rather than cycle to state M to remove the correct, but transiently melted out, nucleotide. Similarly, a previously inserted incorrect nucleotide is more likely to appear melted out, K m << 1. Following insertion of an incorrect nucleotide, the enzyme will favor removal of the mismatch by cycling to state M rather than A.

(2) Nuclease/polymerase ratio is an important fidelity component In accordance with the idea originally put forth by M.J. Bessman (Muzyczka et al., 1972; Bessman et al., 1974), in the simplest form of the model, replication fidelity is determined primarily by the ratio of exonuclease/polymerase activity. Here, it is important to distinguish between the terms "activity" and "specificity". The proofreading exonuclease exhibits specificity only for single stranded primer termini, not for the identity of the nucleotide at the primer terminus. In the context of the model, if the enzyme cannot distinguish between a melted primer terminus containing correct or incorrect base pairing partners, then the rate constant to remove any melted out nucleotide is equal to k x. If the enzyme were found to distinguish incorrect from correct melted termini, this additional specificity could be dealt with by assigning two different exonuclease rate constants, one to remove wrong melted termini, kx(W ) and one to remove right melted termini, kx(R ). Nucleotide insertion by the polymerase can be treated in a similar way. The simplest assumption is that the enzyme does not distinguish between right and wrong candidate dNTP's so that kcat (Fig. 1) is the same to insert right and wrong nucleotides. In this picture, the fidelity of nucleotide insertion (equal to the reciprocal of the misinsertion frequency) is determined primarily by the relative Km'S, Km(W)/Km(R), for competing wrong versus fight dNTP's. Again, the simplest model assumes that the enzyme cannot adjust its catalytic rate to favor right over wrong nucleotides but instead inserts,

14 with equal probability, any bound dNTP. In this case, a correct dNTP will be inserted much more often than an incorrect one because Km(W ) >> Km(R ). In other words, the polymerase-primer template-dNTP substrate complex is likely to be much less stable for mispaired compared to correctly paired substrates. The Km'S approximate equilibrium binding constants for right and wrong dNTP substrates, K m = ( K _ 1 + k c a t ) / k 1 ~ k _ l / k 1. If the polymerase is able to adjust its phosphodiester bond catalytic rate, kcat, to accelerate the rate of inserting correctly matched over mismatched nucleotides, then two different values of kcat, one for right and one for wrong dNTP's will be required. (3) D N A polymerase is able to sample an equi#brium distribution of melted~annealed primer termini

Following nucleotide insertion or excision, equilibrium is established between melted and annealed primer termini states. In other words, the rate constants for the melting and annealing of primer termini, k m and k a (Fig. 1), are assumed to be large compared to rate constants for polymerization (kca t), excision (k x), and enzyme translocation. Following each insertion and excision step, the enzyme is assumed to recycle to an equilibrated primer terminus, at each template site, until completion of the DNA chain. Cycling of the enzyme to an equilibrium distribution of melted and annealed primer termini is an important feature of the model which insures that the proofreading exonuclease can never be completely inhibited, even at saturating concentrations of correct dNTP. Thus, there will always by a possibility that the enzyme will bind to a melted out terminus, and remove it, even when the terminal base pair is correct, in compliance with experimental data showing turnover of a substantial fraction of correctly inserted nucleotides (Clayton et al., 1979; Fersht et al., 1982). Kinetic mechanisms of fidelity The model (Fig. 1) is an Occam's razor attempt to treat the fidelity of DNA replication by assuming the minimum number of adjustable parameters necessary to describe the kinetics (Galas and Branscomb, 1978; Clayton et al., 1979; Goodman

and Branscomb, 1986). In this model, a deliberate attempt is made to separate replication fidelity into individual components: the first component contains the enzymes, the polymerase and proofreading exonuclease, and the the second component is determined by differences in stabilities between correct and incorrect base pairs. The simplest assumption concerning the properties of the polymerase is that it is constrained to insert all candidate nucleotides resident on the polymerase-primer-template complex. It is not able to adjust its catalytic rate constant, kca t, to favor correct or disfavor incorrect base pairs. Similarily, the simplest assumption regarding the proofreading exonuclease is that it shows equal specificity toward the removal of all melted-out primer termini, both correctly matched and mismatched. The fidelity is determined by a combination of enzyme-specific and base-pairing components. The enzyme-specific component is given by the ratio of exonuclease/polymerase activity. The base-pairing component for insertion fidelity is determined by the difference in stability between polymerase-primer-template complexes containing right versus wrong dNTP substrates; the base-pairing component for excision fidelity is determined by the relative probabilities that wrong and right 3'-primer termini are melted-out. There are several objections that can be raised to this simple model. First, enzymes typically exhibit highly selective properties for the right substrates. Second, differences between matched and mismatched base-pairing free energies, as deduced from thermodynamic melting temperature measurements in aqueous solution, are much too small to account for nucleotide insertion or excision fidelity (Loeb and Kunkel, 1982). However, these are precisely the types of objections which should be addressed within the context of a simple model. If experimental data are shown to be inconsistent with predictions based on the simple model, then either added layers of complexity are required to restore consistency, or failing that, the model must be discarded. Kinetics and fidelity of nucleotide insertion

A general expression for the nucleotide misinsertion frequency, ( f ) , in terms of Km, Vma×,

15

and dNTP nucleotide pools is dGTP

I(W) [dWTP] Km(R ) Vmax(W) f = I(R-----~- [dRTP'---~~K m ( W Vm~x(R)

5,_32p primer ACGAATJl (1)

The misinsertion ratio is directly proportional to the ratio of wrong to right dNTP's competing for insertion at a given template site. The mass action effect of competing nucleotide pools has been verified in vitro (Clayton et at., 1979; Fersht, 1979). Nucleotide pool biases have been used to drive mutations to detectible levels in systems where purified polymerases are used to copy a defined DNA template in vitro, followed by transfection of the copied primer-template DNA into a suitable recipient cell in vivo (see e.g. Loeb and Kunkel, 1982). Data from the transfection studies also show that the mutation rate is proportional to [dWTP]/[dRTP]. In the simplest form of the model, the catalytic rate of phosphodiester bond formation is assumed to be the same for right and wrong nucleotides. In this extreme case, and assuming the presence of equal concentrations of right and wrong dNTP's, the misinsertion ratio should be given just by the ratio of right to wrong Km'S, and one should also observe that Vm~,(W) = Vmax(R). The use of K m discrimination to control nucleotide insertion fidelity has been observed for the case of misinsertions involving the mutagenic base analogue 2aminopurine, A P - T versus A - T (Clayton et al., 1979) and A P - C versus AP. T (Watanabe and Goodman, 1982). More generally, one might expect to observe a mixture of both K m and Vma. discrimination. The ratio of Vma~/K m is a direct measurement of nucleotide insertion efficiency (Fersht, 1977). It can be seen from Eqn. (1) that the reciprocal of the misinsertion frequency is polymerase insertion fidelity ( l / f ) . Nucleotide insertion fidelity at one specific site

We have recently developed an assay to measure polymerase kinetics and fidelity at arbitrary sites along a DNA template strand (Randall et al., 1987; Boosalis et al., 1987). The assay takes advantage of the fact that single nucleotide additions to a 32p-labelled primer strand can be resolved by

dNTP

/

... templateTGCTTACCTAGG... b dTTP

5'_32p _primer _ ACGAATGGN / template TGC T T ACCTAGG ""

5'-ACGAATGGN 3'-TGCT TACCT Fig. 2. DNA configurations used to measure DNA polymerase insertion and extension kinetics and primer-template thermal stability. (a) Primer-template used to measure enzymatic rates of insertion of matched and mismatched dNTP substrates opposite base T in template. (b) Primer-template used to measure enzymatic rates of extending matched and mismatched 3'-primer termini by addition of T opposite A. (c) Synthetic duplex 9 mer used to measure melting temperature for matched and mismatched 3'-primer termini. In (a) and (b), the synthetic primer strands are 23 and 20 nucleotides long respectively, labeled at the 5'-end with 32p and annealed to their complementary sections of circular M13 DNA template.

electrophoresis of labelled primer molecules on a polyacrylamide gel. Following electrophoresis, primer molecules of all sizes can be individually quantitated by autoradiography. Using this assay, we measured insertion kinetics for A, G, C and T opposite a template T site (Fig. 2a). The values of K m and Vmax for insertion of fight (A) and wrong nucleotides (G, C, T) were deduced by measuring integrated gel band intensities as a function of dNTP concentration (Boosalis et al., 1987). The data were obtained using Drosophila DNA polymerase a intact holoenzyme complex having no detectible 3'-exonuclease proofreading activity (Kaguni et al., 1983; Cotterill et al., 1987). Base selection by polymerase a appears to be governed primarily by differences in K m for right versus wrong insertions. Compared to the correct insertion of A opposite T, the value of K m is

16 ll00-fold greater for misinsertion of G opposite T and 2600-fold greater for misinsertion of C and T opposite T (Boosalis et al., 1987). In contrast to these large differences in Km, there is only a 4-fold reduction in Vma~ for misinserting G opposite T compared to A opposite T and an 8-fold reduction for misinserting C or T opposite T.

Kinetics and fidelity of nucleotide extension Suppose that a wrong nucleotide had just been inserted, resulting in a mismatched templateprimer terminus. How well does such a mismatch support further DNA synthesis by Drosophila DNA polymerase a? We have addressed this question by comparing D N A synthesis from a preannealed primer molecule containing either correctly matched or mismatched 3'-primer termini (Fig. 2b). The same M13 D N A sequence used to measure mismatch insertion (Fig. 2a) was also used to measure extension from mismatched primer termini (Petruska et al., in preparation). Data comparing insertion of the next correct nucleotide (T opposite A) from matched and mismatched 3'-primer termini (J. Petruska et al., in preparation) show that the efficiency to add the correct nucleotide, T opposite A, onto mismatched primer termini is greatest for G . T and much less for C . T and T . T (Fig. 2b). A comparison of the kinetics of elongation from matched and mismatched primer termini shows that most of the discrimination against mismatch elongation is attributable to increases in Kin, not decreases in Vmax. The apparent Km's to insert T opposite A next to G . T, C- T, and T . T mispairs are 56-fold, 400-fold, and 300-fold larger than to insert T opposite A next to an A - T base pair. The apparent Vmax values are lower by only 3-6-fold to extend mismatched primer termini compared to the correctly matched A • T terminus. Thus, for the M13 template sites shown in Fig. 2, polymerase appears to use K m discrimination to guard against inserting a wrong nucleotide and to reduce the efficiency of elongating mismatched primer termini. The relative frequency of misinsertion (2.1 x 10-4-4.9 x 10 -5) is roughly 10-20-fold lower than for mismatch extension (5.2 x 10 3-5.2 X 10-4). Although it appears to be about an order of magnitude more efficient to add a nucleotide onto a mismatch than it is to form a

mismatch, elongating a mismatch is nevertheless highly inefficient. A simple way to visualize the efficiency of elongating a mismatch is to note that at dTTP concentrations on the order of 1-10/~M, roughly 200 A . T base pairs can be extended in the time it takes to extend a single G . T mispair, and roughly 2000 A . T base pairs can be extended per extension of a single T . T mispair. Inefficient elongation of mismatched primer termini may play an important role in determining the fidelity of D N A synthesis in vivo. For example, delayed elongation could allow for additional opportunity to proofread. It is also possible that D N A polymerase has a higher probability to dissociate from the D N A following misinsertion and a lower probability to reassociate to a melted out terminus. Hence, partially replicated DNA molecules containing aberrant primer termini may have low probabilities to complete replication even if proofreading does not take place.

Kinetics and fidelity of nucleotide excision In our investigations on the kinetics of proofreading, we have utilized 2-aminopurine as a convenient biochemical marker to quantitate the frequency of 2-aminopurine excisions per incorporation opposite template thymine (Clayton et al., 1979; G o o d m a n and Branscomb, 1986). 2Aminopurine. thymine base pairs form two hydrogen bonds but are less stable than A . T base pairs in D N A (Eritja et al., 1986). Bacteriophage T4 wild-type DNA polymerase inserts 2-aminopurine opposite T at a frequency of 14% compared to insertion of A (Bessman et al., 1974; Clayton et al., 1979). The K m to insert the analogue is 6-fold higher than that for A, while Vmax is indistinguishable for the two substrates. Following insertion of 2-aminopurine, opposite T, the active 3'-exonuclease associated with bacteriophage T4 wild-type polymerase removes about 40% of the misinserted analogue compared to a removal frequency of 20% for A (Bessman et al., 1974; Clayton et al., 1979). By comparison, T4 L141 antimutator polymerase containing a very active proofreading exonuclease removes about 80% of the 2-aminopurine and 40% of the adenine. However, the specificity of removal, defined by the ratio of 2-aminopurine/adenine excision, is the same for both exonucleases. In comparison to

17 the wild-type enzyme, the more active exonuclease activity of L141 antimutator allows it to excise a greater proportion of both incorrect and correct (transiently melted out) primer termini from 3tate M (Fig. 1) before reannealing can occur to allow subsequent nucleotide addition from state A. The data are consistent with the idea that the exonuclease does not distinguish between 2aminopurine and adenine per se. Instead, the proofreading exonuclease appears to respond to the conformation of the 3'-primer-template terminus. In other words, 2-aminopurine is removed in preference to adenine because it has a higher probability of being melted out, not because it is "recognized" as being aberrant by the enzyme. Further insight into the relationship between DNA stability and proofreading can be obtained by considering the effect of nucleotide sequence on the distribution of 2-aminopurine in DNA. In an experiment using L141 antimutator DNA polymerase to copy q~X174 DNA in vitro, 2aminopurine and adenine deoxynucleotides were allowed to compete at equimolar concentrations for incorporation opposite T (Pless and Bessman, 1983). The relative incorporation of 2-aminopurine/adenine opposite 57 separate template sites was determined by DNA sequencing. The ratio of 2-AP/A incorporation was observed to vary significantly for each of the four possible 5'-nearest neighbor stacking partners. The A P / A ratio varied between 0 and 20% for incorporation next to a primer T, 0 and 14% next to C, 0 and 7% next to A and 0 and 6% next to G. The data appear to be distributed uniformly within each range. Since the 2-aminopurine misincorporation frequency did not correlate simply with the identity of the nearest-neighbor on the primer strand, we decided to test the idea proposed by Bessman and Reha-Krantz (1977), that polymerase errors occurring within relatively stable regions of DNA might be proofread with greater difficulty than those located in less stable regions. In accordance with this idea, we found that the ratio of 2-AP/A incorporation correlated well with the ratio of G - C / A . T base pairs surrounding the site of 2-aminopurine misincorporation (Petruska and Goodman, 1985). The correlation between 2-AP/A incorporation and DNA stability was greatest

when the stability calculation included the surrounding 4-5 base pairs, located symmetrically on both sides of the misincorporation site. The stability of base pairs present at the "upstream" 5'-side of a misinsertion site can affect proofreading by increasing the probability that polymerase cycles to a melted out configuration (state M, Fig. 1) following misinsertion. In other words, a mispaired nucleotide at the primer terminus is less likely to be transiently annealed (state A, Fig. 1) when 5'-neighboring base pairs are A - T rich. The stability of base pairs present at the "downstream" 3'-side of a misinsertion site can affect proofreading by modulating the ability of the exonuclease to "peelback" correct base pairs to confront an error that escaped initial proofreading (Goodman et al., 1974). Recently, it has been shown that wild-type T4 polymerase is able to excise (peelback) as many as 4 normal base pairs downstream from an error before editing can no longer be detected (Sinha, 1987). The identity of the 5'-nearest neighbor appears to have a profound effect on the relative insertion of 2-AP/A. It can be deduced that insertion of 2-aminopurine competes much more favorably with insertion of adenine when the 5'-primer base is a pyrimidine. On the surface, this observation appears to contradict the widely held notion that purines stack better next to purines than next to pyrimidines. However, it is important to remember that 2-aminopurine and adenine are competing for insertion opposite T. Therefore, it is the difference in stacking energies of the two purine bases that determines their relative stabilities (gm'S) on the polymerase-primer-template complex. We believe that both 2-aminopurine and adenine stack better next to purines on the primer strand, but that their differences in base stacking energies are less when stacked next to pyrimidines than purines (Petruska and Goodman, 1985). Hence, A P / A insertion ratios will be greater next to pyrimidines.

Relating base-pairing thermodynamics and polymerase kinetics In describing a complex sequence of reactions, it is often necessary to include intricate kinetic details associated with the reaction pathway. In our simple model describing DNA synthesis fidel-

18 ity (Fig. 1), several kinetic pathways that might have an effect on fidelity were omitted deliberately. For example, not included is the possibility that polymerase might dissociate more frequently following insertion of a mispair than a correct base pair. Enzyme dissociation at various stages of the reaction pathway, in general, can affect apparent g m and Vmax values. It is welldocumented that different polymerases exhibit different processivities that are influenced by the presence of auxiliary proteins (Kornberg, 1980). Our approach to investigate fidelity is to divide the problem into two distinct parts, one involving g m and the other Vmax (Goodman and Branscomb, 1986). K m contains the thermodynamic contributions to fidelity attributable to free energy differences, AAG °, between correct and incorrect base pairs in the environment of the enzyme's active site. Vmax reflects fidelity components specified primarily by the polymerase and proofreading exonuclease. Since free-energy differences between the states of a system must be independent of pathway, then when confronting the thermodynamic part of the problem, it may be possible to avoid having to define detailed kinetic models or to invoke conformational properties of enzymes. As shown earlier, fidelity of nucleotide insertion and extension at the M13 target site in Fig. 2a, b is determined primarily by K m discrimination. In the context of the simple model shown in Fig. 1, the K m component of discrimination is identified with the relative stability of the polym e r a s e - p r i m e r - t e m p l a t e - d N T P complex. The ratio of Km'S for insertion of right and wrong nucleotides, can be expressed in terms of the free-energy difference for dissociation of nucleotides from template in the polymerase active site

KIn(W) - e aaG°/RT)

(2)

Km(R) For discrimination at the nucleotide insertion step, ratios are in the range of 1100 to 2600, corresponding to AAG ° values of 4.3-4.9 kcal/mole. On the other hand, for extension of mismatched versus matched primer termini, K m K m

ratios are between 56 and 400, about an order of magnitude less than for insertion. AAG° values for extension are between 2.4-3.6 kcal/mole. An immediate difficulty arises when one attempts to relate free-energy differences based on fidelity measurements with free energy differences obtained from D N A melting studies in aqueous solution. The problem is that corresponding AAG ° values in solution are much smaller. Whereas fidelity AAG ° values, given by Eqn. (2), are typically in the 3-5 k c a l / m o l e range, melting temperature measurements comparing matched and mismatched base pairs generally give AAG ° values less than half as large (Loeb and Kunkel, 1982).

Effects of water exclusion on A A G °

DNA bases can form hydrogen bonds with each other as well as with water molecules. When double-stranded D N A melts to form two single-stranded molecules, the loss of base pairing H-bonds is compensated for by the bases H-bonding to water. Hence, the decrease in free energy accompanying the transition from random coil to double helix is attributable to base stacking interactions, not hydrogen bonding (Petruska et al., 1986). It seems reasonable to assume that H-bonding interactions play a central role in determining specificities of nucleotide insertion, excision and primer extension by D N A polymerase. Since Hbonding contributions to solution AAG °'s appear negligible, it is not surprising that the magnitudes of AAG°'s from solution melting data are not directly comparable with those based o n K m differences from Eqn. (2). We have investigated the possibility that exclusion of water in the active cleft of the enzyme may amplify H-bonding free energy differences between right and wrong base pairs (Petruska et al., 1986). Perhaps a reduction in the concentration of water molecules in the vicinity of the active site can decrease the probability that a dNTP substrate in the cleft forms H-bonds with water. If so, a mechanism involving water exclusion could allow for an H-bonding contribution to AAG ° for right versus wrong base pairs at the active site of polymerase which would be absent in aqueous solution.Free-energy differences contain contribu-

19 tions from differences in enthalpy and entropy, A A G ° = 3 A H ° -- T A A S

°

(3)

In collaboration with C. Cheong and I. Tinoco at University of California, Berkeley, we are investigating the thermodynamics of melting D N A containing the same local sequence (Fig. lc) used to measure nucleotide insertion and extension fidelity (Petruska, et al., in preparation). Analysis of these data, as well as earlier data from Tinoco's group (Aboul-ela et al., 1985) and from Breslauer's group (Breslauser et al., 1986), shows that a strong correlation exists between A A H ° and AAS °. The A A H ° values are on the order of 1-3 k c a l / m o l e for terminal base pairs and 3-5 k c a l / m o l e for internal base pairs. However, these differences in enthalpy are accompanied by correspondingly large differences in entropy resulting in relatively small AAG °'s by Eqn. (3). One can speculate that a polymerase might exploit free-energy differences between right and wrong base pairs in two ways. The first could involve the imposition of fairly rigid geometrical constraints to reduce the AAS ° and thereby allow AAG ° values to approach A A H ° (Petruska et al., in preparation). The second, could include the amplification of A A H ° by partial exclusion of water from the polymerase active site as described above. The architecture of the enzyme active site would have to be involved in the possible reduction of the number of degrees of freedom for bound d N T P ' s and primer termini and in the partial exclusion of water. However, the mechanism of fidelity for nucleotide insertion, excision, and extension would still be dependent primarily on K m discrimination. Any further enhancement required for D N A synthesis fidelity could be accommodated by enzyme selective mechanisms involving Vmax discrimination.

Concluding remarks The observation that a major portion of insertion and extension fidelity at a single site appears to be explained by differences in g m rather than Vmax suggests that it would be profitable to carry out kinetic measurements at additional template

sites using a variety of polymerases to see if Vmax discrimination can play an important role. However, a cautionary point should be noted concerning interpretation of kinetic data for polymerase accompanied by proofreading exonucleases. If most of the incorrect d N T P is turned over by the exonuclease while most of the correct d N T P is incorporated, then Vmax differences will be large and K m differences small. However, an observation of Vmax discrimination for incorporation does not imply a mechanism of base selection by the enzyme during insertion or excision. Indeed, two K m discrimination mechanisms acting in series, one for insertion and the other for excision, can give rise to Vmax discrimination for incorporation. It is only by analysis of the individual discrimination steps during insertion, extension, and excision that one can distinguish between kinetic mechanisms involving polymerase base selection as opposed to thermodynamic mechanisms involving base-pairing free-energy differences.

Acknowledgement I want to thank Dr. John Petruska and Lynn Mendelman for reading the manuscript and making numerous insightful suggestions. This research was supported by National Institutes of Health Grants GM21422 and GM33863.

References Aboul-ela, F., D. Koh, I. Tinoco Jr. and F.H. Martin (1985) Base-base mismatches, Thermodynamics of double helix formation for dCA3XA3G+dCT3YT3G(X, Y= A, C, G, T), Nucleic Acids Res., 13, 4811-4824. Bessman, M.J., and L.J. Reha-Krantz (1977) Studies on the biochemical basis of spontaneous mutation, V. Effects of temperature on mutation frequency, J. Mol. Biol., 116, 115-123. Bessman, M.J., N. Muzyczka,M.F. Goodman and R.L. Schnaar (1974) Studies on the biochemical basis of spontaneous mutation, II. The incorporation of a base and its analogue into DNA by wild-type, mutator, and antimutator DNA polymerases, J. Mol. Biol., 88, 409-421. Boosalis, M.S., J. Petruska and M.F. Goodman (1987) DNA polymerase insertion fidelity: gel assay for site-specific kinetics, J. Biol. Chem., 262, 14689-14696. Breslauer, K.J., R. Frank, H. Blocker and L.A. Marky (1986) Predicting DNA duplex stability from the base sequence, Proc. Natl. Acad. Sci. (U.S.A.), 83, 3746-3750. Brutlag, D., and A. Kornberg (1972) Enzymatic synthesis of

20 deoxyribonucleic acid, XXXVI. A proofreading function for the 3 ' ~ 5' exonuclease activity in deoxyribonucleic acid polymerases, J. Biol. Chem., 247, 241-248. Clayton, L.K., M.F. Goodman, E.W. Branscomb and D.J. Galas (1979) Error induction and correction by mutant and wild type T4 DNA polymerases: kinetic error discrimination mechanisms, J. Biol. Chem., 254, 1902-1912. Cotterill, S.M., M.E. Reyland, L.A. Loeb and I.R. Lehman (1987) A cryptic proofreading 3 ' ~ 5' exonuclease associated with the polymerase subunit of the D N A polymerase-primase from Drosophila melanogaster, Proc. Natl. Acad. Sci. (U.S.A.), 84, 5635-5639. Eritja, R., B.E. Kaplan, D. Mhaskar, L.C. Sowers, J. Petruska and M.F. Goodman (1986) Synthesis and properties of defined DNA oligomers containing base mispairs involving 2-aminopurine, Nucleic Acids Res., 14, 5869-5884. Fersht, A. (1977) Enzyme Structure and Mechanism, Freeman, San Francisco, p. 91 Fersht, A.R. (1979) Fidelity of replication of phage ~174 DNA by DNA polymerase III holoenzyme; spontaneous mutation by misincorporation, Proc. Natl. Acad. Sci. (U.S.A.), 76, 4946-4950. Fersht, A.R., J.W. Knill-Jones and W.-C. Tsui (1982) Kinetic basis of spontaneous mutation, J. Mol. Biol., 156, 37-51. Galas, D.J., and E.W. and Branscomb (1978) Enzymatic determinants of DNA polymerase accuracy: theory of coliphage T4 polymerase mechanisms, J. Mol. Biol., 124, 653-687. Goodman, M.F., and E.W. Branscomb (1986) DNA replication fidelity and base mispairing mutagenesis, in: T.B.L. Kirkwood, R.F. Rosenberger and D.J. Galas (Eds.), Accuracy in Molecular Processes: Its Control and Relevance to Living Systems, Chapman and Hall, New York, pp. 191-232. Goodman, M.F., W.C. Gore, N. Muzyczka and M.J. Bessman (1974) Studies on the biochemical basis of spontaneous mutation, III. Rate model for DNA polymerase-effected nucleotide misincorporation, J. Mol. Biol., 88, 423-435. Kaguni, L.S., J.M. Rossignol, R.C. Conway and I.R. Lehman (1983) Isolation of and intact DNA polymerase-primase from embryos of Drosophilamelanogaster, Proc. Natl. Acad. Sci. (U.S.A.), 80, 2221-2225. Kelley, W.N., and J.B. Wyngaarden (1972) The Lesch-Nyhan

syndrome, in: J.B. Stanbury, J.B. Wyngaarden and D.S. Fredrickson (Eds.), The Metabolic Basis of Inherited Diseases, 3rd edn., McGraw-Hill, New York, pp. 969-991. Kornberg, A. (1980) DNA Replication, Freeman, San Francisco, Chapter 5. Loeb, L.A., and T.A. Kunkel (1982) Fidelity of DNA synthesis, Annu. Rev. Biochem., 52, 429-457. Malipiero, U.V., N.S. Levy and P.J. Gearhart (1987) Somatic mutation in anti-phosphorylcholine antibodies, Immunol. Rev., 96, 59-73. Marshall, C.J., K.H. Vousden and D.H. Phillips (1984) Activation of c-Ha-ras-1 protooncogene by in vitro modification with a chemical carcinogen, benzo[a]pyrene diol-epoxide, Nature (London), 310, 586-589. Muzyczka, N., R,L. Poland and M.J. Bessman (1972) Studies on the biochemical basis of spontaneous mutation, I. A comparison of the deoxyribonucleic acid polymerase of mutator, antimutator, and wild type strains of bacteriophage T4, J. Biol. Chem., 247, 7116-7122. Pless, R.C., and M.J. Bessman (1983) Influence of local nucleotide sequence on substitution of 2-aminopurine for adenine during deoxyribonucleic acid synthesis in vitro, Biochemistry, 22, 4905-4915. Petruska, J., and M.F. Goodman (1985) Influence of neighboring bases of DNA polymerase insertion and proofreading fidelity, J. Biol. Chem., 260, 7533-7539. Petruska, J., L.C. Sowers and M.F. Goodman (1986) Comparison of nucleotide interactions in water, proteins, and vacuum: model for DNA polymerase fidelity, Proc. Natl. Acad. Sci. (U.S.A.), 83, 1559-1562. Randall, S.K., R. Eritja, B.E. Kaplan, J. Petruska and M.F. Goodman (1987) Nucleotide insertion kinetics opposite abasic lesions in DNA, J. Biol. Chem., 262, 6854-6870. Sinha, N.K. (1987) Specificity and efficiency of editing of mismatches involved in the formation of base-substitution mutations by the 3' ~ 5' exonuclease activity of phage T4 DNA polymerase, Proc. Natl. Acad. Sci. (U.S.A.), 84, 915-919. Watanabe, S.M., and M.F. Goodman (1982) Kinetic measurement of 2-aminopurine-cytosine and 2-aminopurine-thymine base pairs as a test of DNA polymerase fidelity mechanisms, Proc. Natl. Acad. Sci. (U.S.A.), 79, 6429-6433.