Ionization of bromouracil and fluorouracil stimulates base mispairing frequencies with guanine.

To test whether ionized base pairs influence polymerase-catalyzed misinsertion rates, we measured the efficiency of forming 5-bromouracil (B), 5-fluorouracil (F), and thymine base pairs with guanine and adenine as a function of pH using avian myeloblastosis reverse transcriptase. When B, F, and T were present as dNTP substrates, misincorporation efficiencies opposite G, normalized to incorporation of C opposite G, increased by about 20-, 13-, and 7-fold, respectively, as reaction pH increased from 7.0 to 9.5. Incorporation efficiencies to form the correct base pairs, B.A and F.A, normalized to T.A, decreased by 4- and 8-fold, respectively, with increasing pH. The effects of pH on misincorporation efficiencies were about 10-fold greater when B, F, and T were present as template bases. The relative misincorporation efficiencies of G opposite template B, F, and T, normalized to incorporation of A opposite B, F, and T, increased by about 430-, 370-, and 70-fold, respectively, as pH was increased from 6.5 to 9.5, while correct incorporation of A opposite template B and F decreased about 10-fold over the same pH range. Plots depicting incorrect and correct incorporation efficiencies versus pH were fit to a pH titration equation giving the fraction of ionized base as a function of pH. We conclude that avian myeloblastosis reverse transcriptase forms B.G and F.G mispairs in an ionized Watson-Crick conformation in preference to a neutral wobble structure containing favored keto tautomers of B or F. Although participation of disfavored enol tautomers in enzyme-catalyzed base mispair formation cannot be ruled out, the results are inconsistent with the "standard" disfavored tautomer model of mutagenesis. Instead, the data support a model in which ionization of halouracil bases is primarily responsible for B- and F-induced mutagenesis.


From the Department of Biological Sciences, University of Southern California, L o s Angeles, California 90089-1340 and the $Department of Molecular Genetics, Center for Research and Development, Jordi Girona 18-26,08034 Barcelona, Spain
To test whether ionized base pairs influence polymerase-catalyzed misinsertion rates, we measured the efficiency of forming 5-bromouracil (B), 5-fluorouracil (F), and thymine base pairs with guanine and adenine as a function of pH using avian myeloblastosis reverse transcriptase. When B, F, and T were present as dNTP substrates, misincorporation efficiencies opposite G , normalized to incorporation of C opposite G, increased by about 20-, 13-, and v-fold, respectively, as reaction pH increased from 7.0 to 9.5. Incorporation efficiencies to form the correct base pairs, BOA and F*A, normalized to T-A, decreased by 4-and &fold, respectively, with increasing pH. The effects of pH on misincorporation efficiencies were about 10-fold greater when B, F, and T were present as template bases. The relative misincorporation efficiencies of G opposite template B, F, and T, normalized to incorporation of A opposite B, F, and T, increased by about 430-, 370-, and 7O-fold, respectively, as pH was increased from 6.5 to 9.5, while correct incorporation of A opposite template B and F decreased about 10-fold over the same pH range. Plots depicting incorrect and correct incorporation efficiencies versus pH were fit to a pH titration equation giving the fraction of ionized base as a function of pH. We conclude that avian myeloblastosis reverse transcriptase forms BOG and FOG mispairs in an ionized Watson-Crick conformation in preference to a neutral wobble structure containing favored keto tautomers of B or F. Although participation of disfavored enol tautomers in enzyme-catalyzed base mispair formation cannot be ruled out, the results are inconsistent with the "standard" disfavored tautomer model of mutagenesis. Instead, the data support a model in which ionization of halouracil bases is primarily responsible for B-and F-induced mutagenesis.
Watson and Crick's papers describing the structure of DNA also discussed a model to account for transition mutations involving disfavored tautomers of the common bases (1,2). A.C mispairs could occur if either of the bases was present as a favored amino tautomer and the other as a disfavored imino tautomer. Similarly, G. T mispairs could accommodate one of the bases in a favored keto and the other in a disfavored enol conformation. Freese (3) proposed that disfavored tautomers could account for 2-aminopurine and 5-bromodeoxyuridine mutagenesis in Escherichia coli and T4-infected E. coli, assuming 2-aminopurine and B underwent tautomeric conversion more readily than their respective parent compounds, A and T. Topal and Fresco (4) suggested that disfavored base tautomers might also be involved in transversions provided that one of the sugar moieties was present simultaneously as a syn isomer, e.g. GenoI(syn). Aamino( anti). Recently, participation of the disfavored tautomer of 3-deazacytosine has been proposed to explain the incorporation of CTP on a template containing a deaza-C base analogue (5).
NMR and x-ray crystallographic studies have recently focused on determining the structures of a wide variety of base mispairs in duplex DNA. A few examples of structures that have been observed include: G.T wobble (6)(7)(8); A.C protonated wobble (9)(10)(11); G(syn) .A(anti), G(anti).A(syn), and G(anti).A(anti) with both bases as favored tautomers (6,(12)(13)(14). The structure of 5-bromouracil (B) and 5-fluorouracil (F) base pairs with A and G has been determined by NMR. B. G and F.G base pairs exist in a pH-dependent equilibrium between ionized and wobble structures (15, 16), while B . A and F.A are pH-independent . Ionization of B and F allows them to adopt a Watson-Crick structure when forming mispairs with G (15, 16). The observation of 5-bromouracil in an anionic form in a B .G mispair agrees with an earlier proposal by Lawley and Brookes (20) and more generally with the ideas proposed by Ramsay Shaw and co-workers (21) on the involvement of ionized base pairs in mutagenesis.
In this study, we varied pH over a range of 6.5-9.5 and measured the kinetics of forming mispairs involving B, F, and T with G and correct pairs with A. In the lower portion of the pH range the halouracils and T are predominantly electrically neutral, while in the higher range, the halouracils as triphosphates, or inside of the DNA molecule paired with G, are mainly ionized (anionic form) (15,16). We asked whether the efficiencies of mispair formation increased with pH and whether the efficiencies of correct pair formation decreased with pH. The main objective is to determine whether the rate of forming halouracil. G mispairs in the polymerase active cleft is proportional to the concentration of halouracil anions present either on the template strand or as dNTP' substrates.
The abbreviations used are: dNTP, deoxyribonucleoside triphosphate; AMV, avian myeloblastosis virus; RT, reverse transcriptase; FdUTP, fluorodeoxyuridine triphosphate; BrdUTP, bromodeoxyuridine triphosphate; B, 5-bromouracil; F, Ux,uracil derivative where X represents either CH3, Br, or F; when discussing base pairs formed during DNA synthesis, UX.G refers to a mispair formed by incorporating a Ux deoxynucleotide substrate opposite a template G and G.UX refers to a mispair formed by incorporating a G deoxynucleotide substrate opposite a template Ux; the same convention holds for the standard Watson-Crick base pairs; PAC, phenoxyacetyl; DMT, dimethyltrityl. Dependence on p H An increase in halouracil base mispairing frequencies and a concomitant decrease in correct base pairing frequencies with increasing pH would be consistent with ionization models of mutagenesis (20, 21) and inconsistent with models requiring disfavored tautomers to form H-bonded base mispairs (1)(2)(3)(4).

EXPERIMENTAL PROCEDURES
Enzymes-AMV R T (specific activity, 62,700 units/mg) was purchased from U. S. Biochemical Corp. One unit is that amount of enzyme required to incorporate 1 nmol of labeled dTMP into nucleic acid product in 10 min at 37 "C.
Primer-Templates-For assays in which FdUTP and BrdUTP were present as substrates for AMV RT, single-stranded M13 DNA template, isolated from wild type M13 bacteriophage grown in E. coli TAAATCCTTTGCCCG 3') and p14 (5' AAACGGGTAAAATACGT strain JM103, was annealed with primer p5 (5' AT-3') (22). For assays in which B and F were present in the template strand, the 26-mer oligonucleotides (5' CATCACCGXAAACGTCG-TGACTGGGA 3', where X is B, F, or T) were used together with primer p64 (5' TCCCAGTCACGACGT 3'). All synthetic oligonucleotides except for B-and F-containing oligonucleotides were synthesized on an Applied Biosystems DNA synthesizer by Lynn Williams (Comprehensive Cancer Center, University of Southern California, Los Angeles, CA) using standard 2-cyanoethyl phosphoramidites and purified by polyacrylamide gel electrophoresis. Synthesis of Band F-containing Oligonucleotides-An important matter when non-natural base analogues are introduced in DNA templates is to ensure that DNA chemical synthesis does not affect the integrity of the base analogue, When we prepared oligonucleotides containing B and F for NMR studies using phosphate triester methodology (17), a side reaction was observed with 5-bromodeoxyuridine oligonucleotides. The side reaction yielded 10-20% of a modified oligonucleotide in which 5-bromodeoxyuridine was converted to another product that was assigned to 5-amino-2'-deoxyuracil resulting from ammonia displacement of the bromine at the 5-position (23). The modification was prevented by lowering the temperature of the ammonia treatment from the standard overnight treatment at 55 "C to a 3-day treatment at 37 "C (17). Recently, phosphoramidites having more labile protecting groups have been developed (24,25). We have used PAC-protected phosphoramidites for the preparation of modified sequences. Oligonucleotides sequences (A-D) were assembled using commercially available 5'-O-DMT-F and -B and PAC-protected (A and G, phenoxyacetyl; C, isobutyryl) O-(2-cyanoethyl) phosphoramidites (Biogenex, CA) on a 392 DNA synthesizer (Applied Biosysterns): A, 5' CCGBAA 3' (6-mer); B, 5' CCGFAA 3' (6-mer); C, 5' CATCACCGBAAACGTCGTGACTGGGA 3' (26-mer); D, 5' CAT-CACCGFAAACGTCGTGACTGGGA 3' (26-mer).
Hexamers were used to study the deprotection conditions because they could be characterized more easily. Deprotection of hexamers A and B was carried out in two separate aliquots. One half of the oligonucleotide solid support was deprotected by an overnight treatment with concentrated ammonia a t room temperature (recommended conditions for deprotection of PAC-protecting groups) and the other half with the same concentrated ammonia solution but at 55 "C (recommended conditions for deprotection of standard benzoyl and isobutyryl groups). The resulting oligonucleotides were purified by OPC (Applied Biosystems). Aliquots of purified oligonucleotides were analyzed by snake venom phosphodiesterase and alkaline phosphatase digestion (26). 5-Bromodeoxyuridine containing hexamer (A) obtained by the 55 "C ammonia treatment showed the presence of a 15% side product having a retention time similar to dC (retention times using a C,, reverse phase column to separate the products (26) were as follows: side product coming from bromouracil aminolysis, 5.0 min; dC, 5.4 min; dG, 10.2 min; bromodeoxyuridine, 12.4 min; dA, 13.0 min). As before, the side product was likely to be 5-amino-2'deoxyuridine coming from ammonia displacement of bromine at the 5-position (23). Enzymatic digestion of the hexamer obtained by the room temperature ammonia treatment did not show any of this side product (detection limit, 1%). 5'-Fluorouracil containing hexamer (B) did not show any side product either a t room temperature or 55 "C. Based on these results, oligomers C and D were deprotected by ammonia treatment at room temperature and DMT-oligonucleo-tides were purified by oligonucleotide purification cartridge and gel electrophoresis. We recommend the use of more labile phosphoramidites for the preparation of oligonucleotides containing 5-bromouracil in order to avoid the side product obtained at 55 "C.
Reaction Conditions-Procedures of primer 5'-end labeling, primer-template annealing, gel electrophoresis, autoradiography, and densitometry integration were performed as described (22,27). To remove the buffer in primer-template solution, the annealed primertemplate was precipitated by adding 0.1 volume of 5 M ammonium acetate and 2 volumes of cold ethanol, and the DNA pellet was resuspended with HZ0 in the same volume. The primer-template/ enzyme solution was prepared by adding 5 p1 of AMV R T (0.45 units/ PI, 42 nM) to 85 pl of primer-template solution (70 nM). All kinetic measurements were carried out using a "running start" reaction (22,27), where nucleotide incorporation was examined a t a target template site located 2 bases beyond the end of the initial primer 3' terminus. The reactions were started by mixing equal volumes (3 pl) of primertemplate/enzyme solution with dNTP solution (150 mM NaCl, 16 mM MgC12, 1 mM dithiothreitol, and 120 mM Tris-HCI) containing a constant (saturating) concentration of the running start nucleotide and a variable concentration of d N T P for insertion at the target template site. To verify that the reaction pH was not changed significantly at high d N T P concentrations, a t each p H between 6.5 and 9.5, in increments of 0.5 p H unit, we measured pH both in the presence and absence of 5 mM dTTP, dGTP, or UTP at 37 "C; at the highest and lowest pH, the reduction was less than 0.1 pH unit in the presence of 5 mM ribo-or deoxyribonucleoside triphosphate. The reactions were carried out a t 37 "C for 2 min and terminated by adding 12 p1 of 20 mM EDTA in 95% formamide. The reaction conditions were chosen such that product accumulation was linear with reaction time. Labeled primer molecules extended by 0, 1, and 2 or more nucleotides were separated by polyacrylamide gel electrophoresis, and the fraction of each product was determined by integrating band intensities (22,27).
Determination of the Relative Efficiency of Nucleotide Incorporation-The method used to measure the nucleotide insertion kinetics a t a given target site has been described in detail elsewhere (22,27).
Briefly, a running start nucleotide a t constant saturating concentration extended the primer to a site just prior to the target site, and the nucleotide for insertion at the target site was present a t a series of different concentrations to allow the measurement of nucleotide insertion kinetics. The concentrations used for FdUTP were 15 pM, 30 pM, 60 p M , 125 p M , 250 pM, 1 mM, and 5.2 mM; the concentrations for dCTP, BrdUTP, and dTTP are given in the legend to Fig. 3. The relative nucleotide insertion velocity ( u ) at the template target site, given by the ratio of integrated intensities of primers extended to the target site (and beyond) compared with the integrated band intensity of primers extended to a site just prior to the target, was calculated for each [dNTP]. The bands in each lane of a gel autoradiograph were scanned on either a Hoefer GS300 densitometer or a Molecular Dynamics Phosphorimager. The two methods gave equivalent results.

squares fit of u versus [dNTP] fit to the Michaelis-Menten equation
Relative V,,, and K,,, values were obtained by a nonlinear least (22,27). The nucleotide misinsertion efficiency, fins, was determined by the taking ratios of ( VmaX/Km) for incorrect to correct insertions Curve Fitting-The pH rate profiles were fit by assuming that the observed rate, Robs, set equal to either V,,./Km or to the misinsertion efficiency, fins, is given by a population weighted average of the rates for insertion of the protonated and ionized species as shown in Equation 1, where subscripts NH and N-represent protonated and ionized species, respectively, and X represents the mole fraction. The mole fractions of protonated and ionized species can be expressed in terms of [H+], determined by measuring the pH under reaction conditions, and the dissociation constant for the acid (KO) as follows.
Since X N H + XN-= 1, Equation 2 can he rearranged to obtain Equation 2a.
Observed rates as a function of pH were fit by Equation 2a using a nonlinear regression routine contained in Sigma Plot@, Jandel Scientific (Corte Madera, CA) to estimate values for the parameters K,, RNH, and RN-. The data conform reasonably well to the generated titration curves (see Figs. 4, a and b, and 5 ) . However, an accurate estimate for pK. values cannot be obtained because the data exhibit insufficient curvature in the high pH region for incorrect base pairs and low pH regions for correct base pairs. Sufficient curvature is needed to make an accurate extrapolation to minimum and maximum RNH and RNvalues, and to estimate the inflection point (pK. value).
A rough estimate for pK, values was made by fitting the data to a general sigmoidal function, Equation 3, using the nonlinear regression curve-fitting routine in Sigma Plot@, where Robs are the observed values for fin., Rmax, and Rmin are the asymtotic maximum and minimum rate values, respectively, pH is the measured pH, pK. is the calculated pK, (pH value at the inflection point), and 6 is an adjustable slope parameter, where 6 < 0 gives a rising sigmoidal curve. The pK. values obtained were 8.7 ? 0.2, 9.1 & 0.8, and 11 & 8 for insertion of F, B, and T opposite G, respectively (see Fig. 4b, inset), for an average value of b = -11.8 ? 3.8. For the reciprocal measurement having a halouracil base present on the template strand, the pK, values are 8.6 ? 0.4 for insertion of dGMP opposite B and 8.2 ? 0.1 for insertion of dGMP opposite F. The pK, for insertion of dGMP opposite T is 8.6 ? 0.1, indistinguishable from insertion of dGMP opposite B. It should be emphasized that the calculated pK, values use data obtained at lower pH to generate expected data at higher pH and, therefore, must be regarded as no more than rough approximations.

5-
Bromouracil and 5-fluorouracil are predicted to form "stable," multiple H-bonded base mispairs with G ( Fig. 1). Based on NMR spectroscopy of double-stranded oligonucleotides, mispaired species involving ionized and wobble B . G, and F. G were observed in pH-dependent equilibrium (15,16). The triply H-bonded structures predicted to occur when B and F are in unfavored enol tautomeric conformations ( Fig.   1) were not observed. It is possible that disfavored tautomers participate in forming base mispairs during catalysis while a.

7
ionized base pairs, observed by NMR, are present in the final DNA product. Our objective was to investigate whether ionization of B and F influenced deoxynucleotide misinsertion rates. The kinetics of formation of B. G and F. G mispairs and B .A and F.A correct base pairs were measured as a function of pH, with B, F, and T present as dNTP substrates ( Fig. 2a) or as DNA template bases in the reaction (Fig. 26).
Insertion kinetics were carried out in a "running start" reaction (22,27) in which 5'-32P-labeled primers were extended by addition of two nucleotides prior to reaching a template target site G or A (Fig. 2a) or T, B, or F (Fig. 2b).
The running start substrates, dATP or dTTP, were held at constant (saturating) concentrations, and the [dNTP] for insertion at the target site was varied. Primer extension dependence on pH and [dNTP] are shown for correct insertion of C and misinsertion of B, and T opposite template G (Fig. 3). The velocity to insert a nucleotide opposite the target G site is proportional to the ratio of integrated band intensi-

0%. 4)
The nucleotide insertion fidelity is the reciprocal of fins.
Insertion of 5-Bromo-2'-deoxyuridine, and Thymidine Nucleotides Opposite Template G and A as a Function of pH-The relative velocity for insertion of B opposite G is given by the ratio of adjacent integrated band intensities, Z(B . G)/I(A. T) (Fig. 3b). The relative velocity for insertion of B opposite G was considerably more sensitive to pH changes than the insertion of either T opposite G, I ( T . (Fig. 3c), or C opposite G, Z(C .G)/Z(A.T) (Fig.  3a).

G)/I(A.T)
Changes in pH appeared to have a much more pronounced effect on values of relative V,,, than K, (data not shown).

Nucleotide Misinsertion Rate Dependence on pH
Relative V,,, misinsertion values increased 5-50-fold with increasing pH, with the halouracils present either on the template or as dNTP substrates. Apparent K , values for misinsertion of dGMP opposite template halouracil bases showed about a 2-fold decrease between pH 6.5 and 9.5, in a range of 2.5 to 1 mM, from low to high pH. When the halouracils were present as deoxynucleotide substrates, misinsertion K , values varied between 400 p~ and 1.2 mM. For correct insertions opposite template halouracils (dAMP opposite F or B) a 10-to 20-fold decrease in relative V,,, and 2-fold decrease in K,,, (approximately 10 p~ to 5 p~) was observed with increasing pH. With the halouracils as dNTP substrates for correct insertion opposite A, there were no clear systematic changes in K, and Vmax; the range of K,,, values

Effect of p H on Insertion
Efficiencies and Fidelity-V,,./K, and fins values for insertion of C, B, F, and T opposite template G in the pH range 6.5-9.5 are given in Table I, and the values corresponding to insertion of T , B, and F opposite template A are shown in Table 11

FIG. 3. Gel autoradiograms show-
ing the pH-dependent nucleotide incorporation opposite template G by AMV reverse transcriptase. A primer ( P ) is extended, in a running start reaction, by incorporation of two A deoxynucleotides to reach the template target site G (the running start nucleotide, dATP, is present in the assay at saturating concentration, 50 e~, for incorporation a t the template T sites prior to the target G site). Also present in the reaction for incorporation at the target G site are variable concentrations oE a, dCTP; b, BrdUTP; c, dTTP. The reaction pH from left to right is 7.0, 7.5, and 9.0. The unextended "'P-labeled primer is shown in the top left panel, lane 0 I as a function of pH in Fig. 4. The data for the B G, F. G, and T. G mispairs exhibited qualitatively similar "S-shaped" pH profiles predicted by the pH titration equation (Fig. 4a, solid  curues; Equation 2a). Insertion efficiencies, V,,,/K,, increased by about 10-and 14-fold for B .G and F.G mismatches, and about 4-fold for T. G mismatches as the pH was raised Prom 6.5 to 9.5 (Table I).
In contrast to the increase in V,.,/K, for insertion of 5substituted deoxyribouridines as the pH was increased from 6.5 to 9.5, the shapes of V,.,/K,,, curves for correct pairs C. G and T. A (Fig. 4c) appeared as typical enzyme activity profiles, having a maximum in activity a t about pH 7.0, and showed a decrease in activity as the pH was increased from 7 to 9.5.
Although the general shapes of the pH profiles for B . A and F. A appeared qualitatively similar to the C G and T .A curves, the nucleotide analogue insertion efficiencies exhibit a much steeper decrease for pH values above 7. From pH 7.0 to 9.5, B .A and F. A insertion efficiencies decreased by about 9-and 15-fold, respectively, while C . G and T. A efficiencies decreased by only about 2-3-fold (Tables I and 11). The pH profile for insertion of C opposite G should provide a fairly accurate indication of the pH activity profile of the enzyme since C contains no ionizable groups in the pH 7-9 range, and G is roughly 17% ionized at pH 9 (pK, of dGMP = 9.7, Ref. 29). We suggest that ionization contributes to the steep decrease in insertion rates of B and F opposite A because the anionic forms of B and F should not form stable base pairs with A. Also, less efficient binding of the B and F ionized triphosphates to the triphosphate binding site of the polymerase, resulting from the additional negative charge, could also cause a decreased rate of insertion of F and B nucleotides opposite A with increasing pH (Table 11). The relative efficiency, fins, for insertion of F opposite G increased by about 24-fold, from 9.1 X at pH 6.5 to 2.2 X at pH 9.5 (Fig. 4b, Table I)   ' NQ, not quantified autoradiogram band intensities were barely detectable. V,.,/K, < 7 X fins < 3 X  There was about a 20-and 7-fold increase, respectively, in B .
G and T.G fins values at pH 9.5 compared with pH 6.5 (pH 7.0 for the case of B; see Table I). In contrast to the plots of V,,,/K,, which begin to bend over above pH 8.5 (Fig. 4a), the pH profiles for fin, do not exhibit significant saturation even at pH 9.5. The apparent saturation in V,,,,,/K, profiles for the three mispairs observed at high pH is more likely to be caused by a decrease in the polymerase activity, as seen for C . G and T. A (Fig. 4c), than by complete ionization of the bases. The relative insertion efficiencies (Fig. 4b) may correspond more closely than Vm.,/K, values (Fig. 4a) to the degree of ionization of the bases because fins is normalized to enzyme activity, i.e. the efficiency for insertion of C opposite G.
To make a rough estimate of pK, values, a sigmoidal function was used to fit data for pH dependence for insertion of 5-substituted deoxyuracil derivatives opposite G (Fig. 4b,  ). The pK, for insertion of dGMP opposite T is 8.6 f 0.1, indistinguishable from insertion of dGMP opposite B. However, it must be emphasized that the pK, values estimated in this way are, at best, only rough approximations because they rely on the assumption that fins data obtained in a lower pH range can be used to extrapolate to an experimentally inaccessible pH range.

Insertion of G and A Deoxynucleotides Opposite Template
B, F, and T as a Function of pH-Oligonucleotide templates were synthesized containing either B, F, or T at a specific template target site (Fig. 2b), and the effect of pH on the kinetics of G misinsertion or A insertion was measured. The curues describing the kinetics for correct and incorrect base pair formation when B, F, or T was present on the template strand are shown in Fig. 5 , where Vm,,/K, and fins dependence on pH was fit to the pH titration equation (Equation 2a; Fig.  5, a and b, solid curves).
Differences of an order of magnitude or more were observed for the effects of pH on error rates, comparing template (Fig.  5, Table 111) versus substrate halouracils (Fig. 4, Table I). The effect of pH on misinsertion rates was much more pronounced when B and F were located on the template strand. V,,,,,/K, for misinsertion of G opposite B, F, and T increased by about 35-,67-, and 9-fold, respectively, as pH was increased from 6.5 to 9.5 (Fig. 512, Table 111). The corresponding increase in fins was about 400-fold for G . B and G . F mispairs and 70fold for G. T mispairs (Fig. 5b, Table 111).
The effect of pH on the kinetics of forming A . B, A. F, and A ' T correct base pairs is shown in Fig. 5c and Table 111. Unlike the case for the halouracil dNTP substrates (Fig. 4, c  and d ) , there was no indication of a V,,,/K,,, maximum at pH 7.0. Instead, V,,,/K, insertion values appeared to decrease monotonically between pH 6.5 and 9.0, with A inserted opposite B more efficiently than opposite either T (1.7-fold at pH 7.0) or F (4-fold at pH 7.0). The V,,,,,/K, values decreased as a function of pH by about 6-17-fold, while the values for fins, the ratio of V,,,/K, values for A.F and A.B compared with A.T, changed by less than a factor of 2 (Table 111). in text for B . G (V), F. G (0), and T . G (0) mispairs); c, efficiencies of correctly incorporating C opposite template G (V) and T opposite template A (0); d, efficiencies of "correctly" incorporating B (V) and F (0) opposite template A. The symbols represent the experimental data (Tables I and 11). The solid "S-shaped curues in panels a and b were obtained by a non-linear least squares fit of the data to a pH titration equation (see Equation 2a and "Curve Fitting" under "Experimental Procedures"). The solid spline curves in panels c and d were drawn by direct connection of the data points. In the inset, a sigmoidal function was used to fit data for pH dependence for insertion of 5-substituted deoxyuracil derivatives opposite G (see "Curve Fitting" under "Experimental Procedures"). Data were fit using the sigmoidal function in Equation 3 and the nonlinear regression curve-fitting routine in Sigma Plot' by Jandel Scientific. When the data are analyzed in this way pK. values of 8.7 & 0.2, 9.1 -t 0.8, and 11 f 8 are obtained for insertion opposite G of F, B, and T, respectively.

DISCUSSION
According to recent NMR data, a variety of base mispairs in duplex DNA involve anionic (15, 16) and cationic (9)(10)(11)31) structures. The base analogues bromouracil and fluorouracil were observed to form ionized mispairs with G, in equilibrium with neutral wobble structures (15,16). In this paper, we used AMV RT to measure nucleotide misinsertion kinetics as a function of pH to determine whether ionization of F and B may be involved in base mispairing during polymerization.
The transition from keto "* ionized form of the halouracil base increases with pH. Since the equilibrium between keto t , enol forms is independent of pH, an increase in ionized form is accompanied by a concomitant decrease in the neutral enol tautomeric form (Fig. la). Therefore, an observation of increased base mispairing frequencies with p H would be con-sistent with formation of an ionized base mispair during DNA synthesis, while a reduction in mispairing frequencies would be consistent with involvement of the neutral enol tautomer.
A large increase in nucleotide misinsertion efficiencies occurred when p H values were increased from 6.5 to 9.0. Enhanced misinsertion efficiencies were observed in reciprocal experiments with: (i) B, F, and T as dNTP substrates inserted at a defined template position containing G (Fig. 4, a and b; Table I); (ii) dGTP as substrate for insertion opposite either B, F, or T a t a defined template site (Fig. 5, a and b; Table   III). Low misinsertion efficiencies, fin. -5 X 10-~ to 1 X were observed for insertion of F, B, and T opposite G at pH 6.5-7.0. At pH 9.5, the error rates were about 9.2 X for B . G, 2.2 X for F. G, and 5.7 X for T -G, an increase of about 20-, 24-, and 7-fold, respectively (Fig. 4b, Table I).
At pH 6.5, misinsertion efficiencies for T and F deoxynu- cleotides opposite G were 8.5 to 9.1 X (Table I). However, incorporation of B opposite template G (pH 6.5) was beneath the level of detection (Table I) (Fig. 3b). These data suggest the possibility that B . G may form wobble pairs less readily than T. G. It is possible that the relatively efficient incorporation of F opposite template G at pH 6.5 may reflect, in part, a greater degree of ionization of F (pK, 7.6, Ref. 30) compared with B (pK, 8.3;

Ref. 30).
The pH-dependent increase in misinsertion efficiencies was more than 10-fold larger when the halouracil was present as a template rather than dNTP substrate base (Fig. 5b, Table  111). A comparison of Table I (UX.G) with Table I11 for (G. Ux) shows that the effect of pH on V,.,/K, for insertion was about 3-fold greater when B, F, and T were present as template bases compared with dNTP substrates. The more pronounced effect of pH on G. Ux misinsertion efficiencies ( fins ,  Table 111) was caused by a steeper decrease (17-fold for A . B, 6-fold for A.F, and 10-fold for A.T from pH 6.5 to 9.5) in A . Ux correct insertion rates compared with an approximate 2fold decrease in insertion of C deoxynucleotide opposite template G (Table I)  In contrast to the pH dependence for the efficiencies of forming UX.G mispairs, a plot of V,,,/K, for insertion of C opposite G and T opposite A appears to resemble a "typical" enzyme activity uersus pH profile (Fig. 4c). There was at most a 2.5-fold reduction in polymerase activity for insertion of C opposite G between pH 7.0 and 9.5. There was less than a 3fold reduction in activity for T opposite A in the same pH range (Fig. 4c). To explain why it is that T.G shows a pH profile (Fig. 46) but not T. A, it is important to note that when misinsertion efficiencies are very low as for T . G (  Table I), having a small percent of ionized T present (-10%) can greatly increase the misincorporation efficiency. However, in the case of T.A where correct incorporation efficiencies are high (-1, Table 11), the presence of a small amount of ionized T has only a negligible effect on V,,,/K,,, values for incorporating T opposite A.
A(temp1ate) insertion activities with increasing pH (Fig. 4d, Table 11) compared with the relatively shallow %fold decline for T.A(template) (Fig. 4c We suggest that three factors may be contributing to a reduction in the rate of incorporating B and F opposite template A at high pH: (i) ionized halouracils are unable to form two Hbonds with A; (ii) ionized halo-dNTPs might bind with reduced affinity at the triphosphate binding site; (iii) the polymerase is less active at high pH. The reduction in insertion efficiency for F opposite A with increasing pH correlates with a decrease in T,,, with increasing pH for poly(A.F) (30).
When B, F, and T were present on the template strand, insertion of the correct base (A) decreased monotonically with increased p H (Fig. 5c). B was most efficient in directing insertion of A, while T and F were less efficient. Differences in the rates of correct base pair formation, A. B > A. T > A.
F, correlate with T, values (19,33). It is possible that substituents that increase hydrophobicity at the 5-position (bromine > methyl > fluorine) (34) may aid in stabilizing the polymerase. DNA. dNTP complex. Thus, stimulation in nucleotide insertion rates might result from increased base stacking (22).
An important conclusion drawn from this study is that AMV R T catalyzed ionized B.G and F. G Watson-Crick mispairs in preference to neutral wobble structures. This conclusion is based on the qualitative similarity of Ux. G misinsertion rate dependence on pH to a pH titration profile. Note that the concentrations of ionized and keto forms of the bases are in pH-dependent equilibrium with each other (Fig.  la). If AMV RT-catalyzed rates were similar for both ionized and keto forms, then mispair formation would be independent of pH. The observation that Ux. G mispairs rates showed a marked increase with p H is strong evidence that AMV R T favors formation of the ionized Watson-Crick structure. Other enzymes, however, may behave differently. An ability of some enzyme clefts to accommodate both wobble and ionized structures might account for the interesting observation of Driggers and Beattie (32) that changes in pH had little effect on B . G mispairing with E. coli polymerase 111 holoenzyme and T4 polymerase, although excision of mispairs by the potent proof-reading exonucleases of polymerase 111 and T4 polymerase could be responsible for suppressing differences in pH-dependent mispair formation.
We cannot completely eliminate the possibility that ionization of an amino acid side chain in the enzyme active site contributed to the observed fidelity dependence on pH. However, the absence of a titration-like curve for incorporating C opposite G supports the idea that DNA base ionization is the major cause of base mispairs during DNA synthesis. Thus, in light of the recent NMR, x-ray, and enzymatic studies on base mispair structures, it no longer seems reasonable to require involvement of imino and enol tautomers in spontaneous mutagenesis, although a role for disfavored tautomers cannot be formally eliminated.