Proliferating cell nuclear antigen promotes misincorporation catalyzed by calf thymus DNA polymerase delta.

A proliferating cell nuclear antigen (PCNA)-dependent complex, detectable after nondenaturing polyacrylamide gel electrophoresis, is formed between calf thymus DNA polymerase δ (pol δ) and synthetic oligonucleotide template-primers containing a mispaired nucleotide at the 3′-terminal position of the primer. This complex is indistinguishable in composition from that formed with a fully base paired template-primer. Extension of a mispaired primer terminus is a component of DNA polymerase fidelity. The fidelity of pol δ on synthetic oligonucleotide template-primers was compared with and without its specific processivity factor, PCNA. In the absence of PCNA, pol δ misincorporates less than one nucleotide for every 100,000 nucleotides incorporated correctly. Addition of PCNA to reactions reduces fidelity by at least 27-fold. PCNA also confers upon pol δ, the ability to incorporate (and/or not excise) the dTTP analog, 2′-deoxythymidine-5′-O-(α-phosphonomethyl)-β,γ-diphosphate. A model is proposed whereby the increased stability (decreased off-rate) of the pol δtemplate-primer complex in the presence of PCNA facilitates unfavorable events catalyzed by pol δ. This model suggests an explicit mechanistic requirement for the intrinsic 3′-5′-exonuclease of pol δ.

DNA polymerase ␦ (pol ␦) 1 is thought to be responsible for most DNA synthesis during mammalian replication (1). Together with its processivity factor, proliferating cell nuclear antigen (PCNA), pol ␦ is responsible for the bulk of leading strand synthesis as well as significant replication of the lagging stand. It is thought that PCNA enhances the processivity of pol ␦ by binding to both DNA and pol ␦ and by acting as a "sliding clamp" that stabilizes the interaction between pol ␦ and template-primer (2,3). On model template-primers, PCNA stabilizes by nearly 2000-fold the pol ␦⅐template-primer interaction (4,5) and apparently increases the rate of single nucleotide incorporation by the polymerase (6).
The dramatically increased stability of the pol ␦⅐templateprimer interaction in the presence of PCNA has profound functional implications. For example, it was demonstrated by O'Day et al. (7) that PCNA could facilitate synthesis by yeast pol ␦ past a thymine dimer present in the template strand. Both cis-syn and trans-syn-L-thymine dimers could be bypassed. Similarly, the ␤-subunit of Escherichia coli pol IIIholoenzyme, thought to be the functional prokaryotic homolog of eukaryotic PCNA (see Refs. 2,8), also affects replicative bypass of lesions, although in a complex way (see e.g. 9, 10). Finally, thioredoxin, the processivity subunit of T7 DNA polymerase and thus analogous to PCNA, also has effects on polymerase fidelity (11). In light of these many observations, we reasoned that addition to pol ␦ incubations of homologous PCNA might lead indirectly to an increased incidence of any relatively unlikely event (e.g. incorporation versus misincorporation).
In the present study we analyzed the impact of PCNA on two different sorts of unlikely events catalyzed by pol ␦, 1) misincorporation, i.e. incorporation of a normal dNMP opposite a noncomplementary template nucleotide, and extension without excision of a mismatched primer nucleotide; and 2) incorporation of an abnormal nucleotide, i.e. a nucleotide analog. For both, PCNA enhanced the activity of pol ␦, essentially as was expected based on enhanced stability of the pol ␦⅐templateprimer complex. These observations suggest compelling evolutionary selection pressure to account for the active 3Ј-5Ј-exonuclease associated with pol ␦.

EXPERIMENTAL PROCEDURES
Materials-A modified analog of dTTP, 2Ј-deoxythymidine-5Ј-O-(␣phosphonomethyl)-␤,␥-diphosphate (dTCH 2 PP 2 ; for structure, see Fig.  4A), was synthesized as described (12). Unlabeled deoxyribonucleoside triphosphates (dNTPs) were high performance liquid chromatographypurified and were from Pharmacia Biotech Inc. 32 P-Labeled ATP and 32 P-labeled dNTPs were from Amersham Corp. Synthetic oligonucleotides of defined sequence were prepared by conventional phosphoramidite chemistry and purified by PAGE in the presence of 7 M urea. (dT) 4 was also made synthetically by conventional phosphoramidite chemistry. PCNA and pol ␦ were both purified from calf thymus according to published protocols (6,13,14). Micrococcal nuclease was from Sigma. All other materials were obtained commercially; chemicals were of reagent grade and were used without further purification.
DNA Polymerase Assays-DNA polymerase assays were performed essentially as described previously (6,15,16). The standard reaction mixture for pol ␦ contained 40 mM bis-Tris, pH 6.7, 6 mM MgCl 2 , 1 mM dithiothreitol, 10% glycerol, and 40 g/ml bovine serum albumin. Additional details are provided in the figure legends. Fidelity assays were performed according to Creighton and Goodman (19). For kinetic measurements, 5% or less of the total primer present before incubation with enzyme was extended. Results of primer extension were quantified after PAGE of reaction products in the presence of 7 M urea using a Molecular Dynamics 445 SI PhosphorImager. Unless indicated other-* These studies were supported by Research Grants ES-04068 (to P. A. F.) and DK-26206 (to K. M. D.) from the National Institutes of Health and N96-04-48277 (to A. A. K.) from the Russian Foundation for Basic Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed. Tel.: 516-444-3067; Fax: 516-444-6229; E-mail: paul@pharm.sunysb.edu. 1 The abbreviations used are: pol ␦, DNA polymerase ␦; PCNA, proliferating cell nuclear antigen; PAGE, polyacrylamide gel electrophoresis; dTCH 2 PP 2 , 2Ј-deoxythymidine-5Ј-O-(␣-phosphonomethyl)-␤,␥diphosphate; dNTPs, deoxyribonucleoside triphosphates; bis-Tris, wise, all kinetic constants were calculated as described previously (18). Inhibition reactions were performed within the linear region of product formed versus time. With poly(dA)-(dT) 16 as template-primer, inhibition was quantified by placing some of each reaction mixture on 1 ϫ 1-cm squares of Whatman DE-81 filter paper and washing filter paper squares in a 5% solution of Na 2 HPO 4 ⅐12H 2 O to remove unincorporated dNTPs. Inhibition with other template-primers was quantified by phosphorimager analysis after denaturing PAGE. Kinetic constants were determined by least-squares nonlinear regression to a rectangular hyperbola. The ratio of V max /K m was determined either from direct calculation (see above) or from the slope in the linear region of the rectangular hyperbola (19).
Nondenaturing PAGE Band Mobility Shift Assays-Nondenaturing PAGE band mobility shift assays were performed essentially as described previously (4) but without MgCl 2 and otherwise as detailed in figure legends. EDTA was included in each incubation and in the gel electrophoresis buffer at 3 mM.
Nearest Neighbor Analyses-Nearest neighbor analyses were used to prove incorporation of dTCH 2 PP 2 into oligonucleotide primer and were performed essentially as described (20,21). 21-mer primer was annealed to template such that the next three nucleotides to be incorporated were dTMP, dCMP, and dTMP (Fig. 4B). This primer was extended with [␣-32 P]dCTP and either unlabeled dTTP or unlabeled dTCH 2 PP 2 such that 32 P-labeled 24-mers were synthesized. Resulting 24-mers were PAGE-purified in the presence of 7 M urea and digested to completion with micrococcal nuclease, and products were analyzed either by denaturing PAGE or silica gel thin layer chromatography run in isopropyl alcohol:NH 4 OH (25% in H 2 O):H 2 O (7:1:2).
In Situ Determination of DNA Polymerase Activity-Determination of DNA polymerase activity in polyacrylamide gels in situ was performed as described previously (5) after standard nondenaturing PAGE-band mobility shift assays. Nucleic acid template-primers were unlabeled. After electrophoresis, entire gels were incubated for 30 min at 37°C in a standard pol ␦-reaction mixture (4, 6) containing 6 mM MgCl 2 , 40 M dCTP, and 60 Ci of [␣-32 P]dTTP. After incubation, gels were washed extensively (24 -48 h) in cold 5% trichloroacetic acid to remove unincorporated dTTP after which they were subjected to autoradiography.

RESULTS
Formation of a PCNA-dependent Complex Between Calf Thymus Pol ␦ and Mismatched Template-Primers-We previously reported that PCNA stabilized the pol ␦⅐template-primer interaction nearly 2000-fold (5). To study the interaction of pol ␦ with mismatched template-primers, PAGE-band mobility shift assays were initially performed with a single 30-mer template annealed to four different 21-mer primers. Each primer contained a different dideoxynucleoside monophosphate at the 3Ј-terminal position. The structures of the four template-primers used are shown (Fig. 1A). PAGE-stable complex formation was detected by autoradiography after 5Ј-32 P labeling of nucleic acid. Although complex formation was strongest with the matched template-primer (I) (Fig. 1B), PCNA-dependent complex formation was detected with each of the three mismatched template-primers (II-IV) ( Fig. 1, C-E). The experiment shown was performed in the continuous presence of EDTA. Identical results to those shown ( Fig. 1) were obtained when primers containing 3Ј-OH groups were used instead of 2Ј,3Ј-H groups (not shown, but see Fig. 2).
Before extensive characterization of the complex formed among pol ␦, PCNA, and mismatched template-primers, it was necessary to prove that the complex detected after nondenaturing PAGE still contained unmodified (i.e. mismatched) nucleic acid. Accordingly, 32 P-labeled primer was recovered from the nondenaturing PAGE-purified complex formed among pol ␦, PCNA, and mismatched template-primer and was subjected to denaturing PAGE to analyze primer size. Detection of fulllength (21-mer) primer (not shown) was considered proof that the mismatched primer was not modified during incubation with pol ␦ and that a pol ␦⅐PCNA⅐mismatched template-primer complex was indeed forming.
Ferguson plot (25) analysis was used to show that the complex formed by pol ␦ and PCNA with a matched templateprimer was indistinguishable from that formed with a mismatched template-primer. Gels of five different polyacrylamide concentrations were used. At all polyacrylamide concentrations and with all three mismatched template-primers, the pol ␦⅐PCNA⅐matched template-primer complex comigrated with the pol ␦⅐PCNA⅐mismatched template-primer complexes (not shown). This suggests that the complex formed among pol ␦, PCNA, and a matched template-primer contains the same molecules in the same ratio as that formed among pol ␦, PCNA, and a mismatched template-primer.
In situ DNA polymerase activity determination revealed that the complex formed among pol ␦, PCNA, and mismatched template-primer could be distinguished from the complex formed with a matched template-primer. The template-primers used for in situ DNA polymerase assays are shown ( Fig.  2A). When a 3Ј-OH-terminated primer is used and the pol ␦⅐PCNA⅐template-primer complex is formed in EDTA, the only additional ingredients required to visualize DNA synthesis are MgCl 2 and radiolabeled dNTP (5). Accordingly, complex was formed with unlabeled matched or mismatched template-primer; to monitor formation and mobility, complex was also formed with 32 P-labeled template-primer. Both labeled and unlabeled complexes were subjected to nondenaturing PAGE, and the portion of the gel containing complex formed with unlabeled template-primer was incubated in [␣-32 P]dTTP, dCTP, and 6 mM MgCl 2 . Nucleotide incorporation was readily detected coincident in PAGE mobility with the complex formed among pol ␦, PCNA, and labeled template-primer (Fig. 2, B and C, compare lanes a; see also Fig. 2C, lane M). In contrast, mismatched template-primers, although supporting substantial complex formation when pol ␦ and PCNA were added (Fig. 2B, lanes  b-d), supported much less nucleotide incorporation in the in situ DNA polymerase assay (Fig. 2C, lanes b-d). Curiously, primers with A⅐C and A⅐G mismatches at the 3Ј-position apparently supported more incorporation (mismatch extension) than a primer with an A⅐A mismatch at the 3Ј-position (Fig. 2C,  lanes b-d). This observation is of uncertain significance.
PCNA Reduces the Fidelity with Which Calf Thymus Pol ␦ Replicates a Synthetic Oligonucleotide Template-It was established that extension of mispaired primer termini contributes significantly to errors made during DNA replication (for a review, see Ref. 17). Formation of a PCNA-dependent complex between pol ␦ and template-primers containing 3Ј-terminally mispaired nucleotides led us to determine, directly, the effect of PCNA on the fidelity of pol ␦. The strategy of Creighton and Goodman (19) was employed. For this, we used the templateprimers shown in Fig. 3A. Initially, this consisted of a 5Ј-32 Plabeled 17-nucleotide primer (Fig. 3A, underlined) which was annealed to an unlabeled 64-mer template such that the next nucleotides to be incorporated were (in order) dGMP, dCMP, dTMP, and dAMP. All incubations contained 5 ng of pol ␦ and dGTP such that the first nucleotide specified could be incorporated. In addition, incubations contained either no PCNA (see Fig. 3B) or 70 ng of PCNA (Fig. 3C), dTTP at 10 M (Fig. 3, B and C, lanes 2-6), or as indicated, and dCTP as indicated. Data from these experiments were quantified by phosphorimager. To determine relative velocities for the correct nucleotide, results shown in Fig. 3, B and C, lanes 1-6, were quantified; radioactivity in the band resulting from dTMP incorporation plus radioactivity in the band resulting from dCMP incorporation was divided by radioactivity in the band resulting from dGMP incorporation. The results from Fig. 3C are plotted versus dCTP concentration (Fig. 3D). To determine relative velocity for the incorrect nucleotide, results shown in Fig. 3, C, lanes 1 and 7-9, were quantified; radioactivity in the band resulting from dTMP incorporation was divided by radioactivity in the band resulting from dGMP incorporation. The results are plotted versus dTTP concentration (Fig. 3E). There was no misincorporation of dTMP detected for pol ␦ alone (Fig. 3B, lanes 7-9) or in the presence of dGTP only (Fig. 3, B and C, lanes 1).
To determine fidelity (efficiency of incorporation (misincorporation), and extension (26)), V max /K m values were calculated as described previously (18) from the data shown (Fig. 3). As noted, with pol ␦ alone, there was no detectable misincorporation. We estimate that pol ␦ misincorporates less than once for every 10 5 nucleotides incorporated correctly. In the presence of PCNA, pol ␦ catalyzed one misincorporation event (dTMP instead of dCMP) for every 3700 nucleotides incorporated correctly. This amounts to at least a 27-fold decrease in pol ␦ fidelity brought about by PCNA.
To investigate the effect of primer length on PCNA-induced misincorporation catalyzed by pol ␦, each of the primers shown in Fig. 3A was 5Ј-32 P-labeled and annealed to the template shown in Fig. 3A. Incubations were formulated as indicated; only results with 21-mer are shown (Fig. 3F). Identical results were obtained with 13-, 17-, 25-, 29-, 36-, 41-, and 46-mer (not shown). Only in the presence of both dTTP and PCNA was a second dTMP residue incorporated (Fig. 5, downpointing arrows), indicative of both misincorporation of dTMP in place of dCMP and extension without excision of the mismatched primer terminus.
PCNA Promotes the Stable Incorporation of Nucleotide Analog, dTCH 2 P, by Calf Thymus Pol ␦-The structure of dTCH 2 PP 2 is shown in Fig. 4A. dTCH 2 PP 2 has been used diagnostically to distinguish different DNA polymerases (16). Incorporation of this analog in place of dTMP was evaluated using the synthetic template-primer shown (Fig. 4B). Extension of this primer with either pol ␦ alone (Fig. 5, lanes 2-6) or pol ␦ plus PCNA (Fig. 5, lanes 7-11) revealed that PCNA FIG. 2. Difference between pol ␦⅐PCNA⅐matched templateprimer complex and pol ␦⅐PCNA⅐mismatched template-primer complex as revealed by in situ DNA polymerase assay. A, to allow incorporation, experiment was performed with 3Ј-OH-terminated primers; a single 21-mer was annealed to each of four 30-mers as shown to generate one matched template-primer (I) and three terminally mismatched template-primers (II-IV). B, each of the template-primers shown in A was 5Ј-32 P-labeled, mixed with pol ␦ and PCNA, and subjected to nondenaturing PAGE. Arrow to the right of the panel designates the migration position of the pol ␦⅐PCNA⅐nucleic acid complex. Lanes a-d, complex formed with 32 P-labeled template-primers (I-IV), respectively. C, results of in situ DNA polymerase assay. Lane M, complex formed with 32 P-labeled matched template-primer (I) included as a mobility marker; lanes a-d, complex formed with unlabeled template-primers (I-IV), respectively (see A). Arrow to the right of the panel designates the migration position of the pol ␦⅐PCNA⅐nucleic acid complex.
increased dramatically the ability of pol ␦ to apparently incorporate (and/or not excise) dTCH 2 P. Also apparent from inspection of Fig. 5, lane 8, is PCNA-induced misincorporation and extension of dTMP opposite template dGMP. In separate experiments, we showed that in the presence of PCNA, pol ␦ could extend a 3Ј-dTCH 2 P terminated primer generated after incorporation from a dTCH 2 PP 2 precursor (not shown).
Modified nearest neighbor analyses (20,21) were used to prove that the dTMP analog dTCH 2 P was indeed being incorporated (and/or not excised) by pol ␦ in the presence of PCNA. Two separate incubations contained pol ␦, PCNA, synthetic template-primer (Fig. 4B), [␣-32 P]dCTP, and either dTTP or dTCH 2 PP 2 ; radiolabeled 24-mer was synthesized in both cases and PAGE-purified in 7 M urea. The two purified 24-mers (Fig.  6, lanes 1 and 2) were treated identically with micrococcal nuclease, and digests were analyzed by PAGE in the presence of 7 M urea. Results demonstrated that after digestion of the dTMP-containing 24-mer (Fig. 6, lane 1), only 32 P-labeled dTMP was recovered (Fig. 6, lane 3); identical digestion of the putatively dTCH 2 P-containing 24-mer (Fig. 6, lane 2) resulted in a 32 P-labeled species of dramatically different mobility (Fig.  6, lane 4). This species was presumably 5Ј-dAMP-dTCH 2 P-dCM[ 32 P]dTCH 2 OH-3Ј. The dramatically different products recovered after micrococcal nuclease treatment thus confirmed incorporation of dTCH 2 P by pol ␦ in the presence of PCNA. Identical conclusions were reached after considering results from silica gel thin layer chromatography (not shown).
Quantitative comparison of dTMP and dTCH 2 P incorporation (Fig. 7) revealed that dTTP was much more efficiently utilized than dTCH 2 PP 2 . (Efficiency is defined as V max /K m , see Ref. 26.) V max and K m were calculated as described previously (18). The V max and K m for dTTP were 4.6 and 0.6 M, respectively. The V max and K m for dTCH 2 PP 2 were 1.6 and 18.2 M, respectively. Hence, in the presence of PCNA, dTTP is utilized by pol ␦ 87-fold more efficiently than dTCH 2 PP 2 . In the absence  (lanes 1-6) or T⅐G (lanes 7-9); and 20-mer, T⅐A). Incubation was for 45 s at 37°C during which approximately 5% of the primer molecules were extended; 95% remained unmodified. C, exactly the same as B except that 70 ng of PCNA was included in all incubations. D, data in C, lanes 1-6, were quantified and plotted. Relative velocities were calculated from the ratio, (19-mer (C⅐G) radioactivity ϩ 20-mer (T⅐A) radioactivity)/18-mer (G⅐C) radioactivity. Efficiency of dCTP (correct) incorporation was determined from experimentally measured V max and K m to be 1.3 M Ϫ1 . E, data in C, lanes 1 and 7-9, were quantified and plotted. Relative velocities were calculated as in D. Efficiency of dTTP (incorrect) incorporation was determined from experimentally measured V max and K m to be 3.5 ϫ 10 Ϫ4 M Ϫ1 . Identical experiments and calculations were performed for pol ␦ without PCNA (not shown; see B for primary data). F, standard reaction mixtures contained 5Ј-32 P-labeled 21-mer primer as indicated (see Fig. 3A for sequence) annealed to a 64-mer template (see Fig. 3A for sequence), and additions as indicated. When dTTP was present, it was at 1 mM. When dGTP was present, it was at 30 M. When pol ␦ was present, 10 ng was added. When PCNA was present, 70 ng was added. Incubations were for 10 min at 37°C after which products were analyzed by PAGE performed in the presence of 7 M urea followed by autoradiography. Downpointing arrow designates the band resulting from stable misincorporation and extension of dTMP into primers opposite a template dGMP residue.
FIG. 4. dTCH 2 PP 2 and the template-primer used to monitor dTCH 2 P incorporation. A, the structure of dTTP analog, dTCH 2 PP 2 . B, the template-primer used to monitor dTCH 2 P incorporation by pol ␦.
of PCNA, pol ␦ apparently does not use dTCH 2 PP 2 as a substrate.
The Effect of PCNA on Utilization of dTCH 2 PP 2 by Pol ␦ Occurs through an Indirect Mechanism-The ability of PCNA to promote utilization of dTCH 2 PP 2 by pol ␦ could be either direct, through increased affinity for dTCH 2 PP 2 or rate of dTCH 2 P incorporation, or indirect, through increased stability of the pol ␦⅐PCNA⅐template-primer complex. If direct (e.g. through increased affinity for dTCH 2 PP 2 ), we would expect PCNA to alter significantly inhibition of incorporation by dTCH 2 PP 2 ; indirect action (e.g. through increased affinity for template-primer) would not change the inhibitory potency of dTCH 2 PP 2 . To distinguish direct from indirect effects, we examined the ability of dTCH 2 PP 2 to compete for dNTP utilization by pol ␦. As a preliminary, it was noted that dTCH 2 PP 2 only inhibited incorporation of dTMP and did not affect utilization of dGTP on a template-primer which directed dGMP incorporation (Fig. 8A). With a similar template-primer but which directed dTMP incorporation, dTCH 2 PP 2 was a potent inhibitor of pol ␦ activity; inhibitory potency was not affected by PCNA (Fig. 8B). On a different template-primer that directed dTMP incorporation, poly(dA)-(dT) 16 , dTCH 2 PP 2 was also a potent inhibitor of dTMP incorporation, and inhibition was not affected by PCNA (Fig. 8C). PCNA is without significant effect on the affinity of pol ␦ for dTTP (not shown).

DISCUSSION
Under static conditions, PCNA decreases dissociation of pol ␦ from the pol ␦⅐template-primer complex nearly 2000-fold (4,5). This is reflected under dynamic conditions, i.e. during template-directed dNMP incorporation, in decreased K m for appropriate template-primers (6), and dramatically increased processivity of DNA synthesis (see e.g. Ref. 14). Although the PCNA-induced increase in the stability of the pol ␦⅐templateprimer complex is necessary for efficient DNA replication (see e.g. Ref. 27), our results demonstrate that it also causes substantial decreases in the fidelity with which pol ␦ catalyzes replication as well as increases incorporation (and/or inhibits excision) of a nucleotide analog, dTCH 2 P, relative to unmodified dTMP.
We propose that increased efficiency of both unlikely events results from prolonged residence of pol ␦ at the templateprimer junction induced specifically by PCNA and that this is intrinsic to the mechanism of "clamp-type" processivity factors. Similar proposals have previously been advanced for prokaryotes (see e.g. Refs. 9 -11). We demonstrated directly that PCNA stabilizes the interaction of pol ␦ with a template-primer containing a 3Ј-terminally mismatched primer (Fig. 1) in a complex that is indistinguishable from that formed with a fully base-paired template-primer (Fig. 2). Moreover, the terminally mispaired primer can apparently be extended, at least under some circumstances (Fig. 2). These observations correlate with at least a 27-fold decrease in the fidelity of pol ␦ brought about by the addition of PCNA (Fig. 3).
Formally, the apparent increase in misincorporation of dTMP as measured in our assay could result if the dTTP used was contaminated with dCTP and PCNA acted to reduce the K m of pol ␦ for dCTP. Given the vendor-stated purity of the dTTP used, we considered this highly unlikely. Nevertheless, to evaluate this possibility directly, we determined the K m of pol ␦ for dCTP both with and without PCNA. K m values of 7.5 and 12 M, respectively, were determined, 2 thus ruling out K m effects as a mechanism to account for the apparent misincorporation of dTMP. We also considered the possibility of template contamination to account for dTMP misincorporation. However, if this was the case, the ratio of dTMP incorporation versus dCMP incorporation would not be expected to change dependent on the presence or absence of PCNA.
We think it noteworthy that in the assay used to evaluate pol ␦ fidelity, we did not distinguish among nucleotide misincorporation, mismatch excision, and extension of a 3Ј-terminally mispaired primer. It is clear from the data shown that PCNA promotes complex formation with a mispaired primer terminus ( Fig. 1) and, at least under some conditions, can allow such a primer to be extended (Fig. 2). In several contexts, we were unable to demonstrate any effect of PCNA on mismatch excision. 3 There is currently no information pertinent to effects of PCNA on misincorporation of normal nucleotides (as distinct from mispaired primer extension) by pol ␦. However, PCNA clearly promotes the misincorporation (and/or inhibits excision) certainly an area for future research that will have considerable bearing on understanding the effects of PCNA on replication by pol ␦. We would also like to note that the conditions used to delineate PCNA-induced infidelity by pol ␦ differ significantly from those found in vivo. In our assays, only a single dNTP, dTTP, is provided, potentially biasing our system toward misincorporation. In vivo, all four dNTPs are present. It remains to be determined if PCNA would promote misincorporation by pol ␦ during in vivo replication.
The other unfavorable event catalyzed by pol ␦ that PCNA promotes is incorporation of the dTMP analog, dTCH 2 P (Figs. 4 -8). This analog was used diagnostically to distinguish DNA polymerase ␣ from DNA polymerase ⑀ (16). In light of our current observations, such results must be interpreted with caution since, in vivo, it is possible that hypothetical and as yet undiscovered accessory factors could influence dramatically analog utilization by DNA polymerases. Inhibition data (Fig. 8) demonstrate that PCNA enhancement of dTCH 2 P incorporation by pol ␦ occurs through an indirect mechanism rather than a direct effect of PCNA on the interaction of pol ␦ with dTCH 2 PP 2 . The ability of dTCH 2 PP 2 to compete with dTTP is unaffected by PCNA (Fig. 8). Accordingly, it is our conclusion that PCNA can promote stable nucleotide misincorporation (as opposed to extension of mismatched primer termini) indirectly through stabilization of interaction of pol ␦ with the primed template.
Ultimately, it is, in our estimation, the effect of PCNA on dissociation of pol ␦ from the template-primer that provides the evolutionary selection pressure necessary to maintain the 3Ј-5Ј-exonuclease intrinsic to pol ␦. Any DNA polymerase that functions processively by virtue of decreased dissociation from the primed template (enhanced template-primer binding) is inherently more likely to misincorporate, immortalize a misincorporated nucleotide, or incorporate a nucleotide analog in place of a normal nucleotide. In all instances, the 3Ј-5Ј-exonuclease is ideally suited to deal efficiently with any errors committed by pol ␦. In contrast, DNA polymerases ␣ and ␤, which are not highly processive, lack intrinsic 3Ј-5Ј-exonuclease activities. However, we would also like to note that in our in vitro assays, the 3Ј-5Ј-exonuclease of pol ␦ is demonstrably active (see Ref. 6), yet both effects of PCNA, enhanced misincorporation/mismatched primer extension and greater incorporation of  Fig. 4B for structure and sequence), and varying concentrations of either dTTP (E) or dTCH 2 PP 2 (Ç) as indicated. Incubations were for 30 s at 37°C during which less than 5% of the primer was extended. After incubation, nucleic acids were subjected to PAGE in the presence of 7 M urea, and products were quantified. B, Lineweaver-Burk double-reciprocal plots of the data shown in A.
FIG. 8. Effect of PCNA on dTCH 2 PP 2 inhibition of dTTP utilization by pol ␦. Standard reaction mixtures contained 10 ng of pol ␦, either with 70 ng of PCNA (Ⅺ) or without PCNA (E) and dTCH 2 PP 2 as indicated on the abscissas. A, 0.4 M 64-mer template was annealed to an equal amount of 5Ј-32 P-labeled 17-mer primer (see Fig. 3A for structure and sequence); this template-primer directed dGMP incorporation. 10 M dGTP was included in all incubations. B, 0.4 M 40-mer template was annealed to an equal amount of 5Ј-32 P-labeled 21-mer primer (see Fig. 4B for structure and sequence); this template-primer directed dTMP incorporation. 10 M dTTP was included in all incubations. C, 0.4 M poly(dA) template was annealed to unlabeled (dT) 16 primer (200:1 ratio of nucleotide:nucleotide); this template-primer directed dTMP incorporation. 10 M [␣-32 P]dTTP was included in all incubations. dTCH 2 P, were easily observed. In cells, there are presumably other factors that contribute to the accuracy of pol ␦. The assays described here may be useful to identify such factors in vitro