Dissociation of the Protein Primer and DNA Polymerase after Initiation of Adenovirus DNA Replication*

Initiation of adenovirus DNA replication occurs by a jumping back mechanism in which the precursor terminal priming protein (pTP) forms a pTP·trinucleotide complex (pTP·CAT) catalyzed by the viral DNA polymerase (pol). This covalent complex subsequently jumps back 3 bases to permit the start of chain elongation. Before initiation, pTP and pol form a tight heterodimer. We investigated the fate of this pTP·pol complex during the various steps in replication. Employing in vitro initiation and elongation on both natural viral templates and synthetic oligonucleotides followed by glycerol gradient separation of the reaction products, we established that pTP and pol are separated during elongation. Whereas pTP·C and pTP·CA were still bound to the polymerase, after the formation of pTP·CAT 60% of the pTP·pol complex had dissociated. Dissociation coincides with a change in sensitivity to inhibitors and inK m for dNTPs, suggesting a conformational change in the polymerase, both in the active site and in the pTP interaction domain. In agreement with this, the polymerase becomes a more efficient enzyme after release of the pTP primer. We also investigated whether the synthesis of a pTP initiation intermediate is confined to three nucleotides. Employing synthetic oligonucleotide templates with a sequence repeat of two nucleotides (GAGAGAGA … instead of the natural GTAGTA … ) we show that G5 rather than G3 is used to start, leading to a pTP·tetranucleotide (CTCT) intermediate that subsequently jumps back. This indicates flexibility in the use of the start site with a preference for the synthesis of three or four nucleotides during initiation rather than two.

Initiation of adenovirus DNA replication occurs by a jumping back mechanism in which the precursor terminal priming protein (pTP) forms a pTP⅐trinucleotide complex (pTP⅐CAT) catalyzed by the viral DNA polymerase (pol). This covalent complex subsequently jumps back 3 bases to permit the start of chain elongation. Before initiation, pTP and pol form a tight heterodimer. We investigated the fate of this pTP⅐pol complex during the various steps in replication. Employing in vitro initiation and elongation on both natural viral templates and synthetic oligonucleotides followed by glycerol gradient separation of the reaction products, we established that pTP and pol are separated during elongation. Whereas pTP⅐C and pTP⅐CA were still bound to the polymerase, after the formation of pTP⅐CAT 60% of the pTP⅐pol complex had dissociated. Dissociation coincides with a change in sensitivity to inhibitors and in K m for dNTPs, suggesting a conformational change in the polymerase, both in the active site and in the pTP interaction domain. In agreement with this, the polymerase becomes a more efficient enzyme after release of the pTP primer. We also investigated whether the synthesis of a pTP initiation intermediate is confined to three nucleotides. Employing synthetic oligonucleotide templates with a sequence repeat of two nucleotides (GAGAGAGA . . . instead of the natural GTAGTA . . . ) we show that G5 rather than G3 is used to start, leading to a pTP⅐tetranucleotide (CTCT) intermediate that subsequently jumps back. This indicates flexibility in the use of the start site with a preference for the synthesis of three or four nucleotides during initiation rather than two.
The replication of the linear 36-kilobase pair adenovirus (Ad) 1 DNA initiates at two origins located at either molecular end employing a protein-priming mechanism. Initiation requires at least two viral proteins, the DNA polymerase (pol) and the precursor of the terminal protein (pTP), which acts as the primer and is linked covalently to the first nucleotide, a dCMP molecule. Initiation can be stimulated by the cellular transcription factors NFI and Oct-1 which guide the pTP⅐pol complex to the core origin employing their DNA binding domains. This leads to stabilization of the preinitiation complex and increases the number of initiation events by enhancing the V max of the reaction. In addition, the virus-coded singlestranded DNA-binding protein stimulates initiation by lowering the K m for dCTP (1).
The virus uses a jumping back mechanism for initiation. At position 4 from the 3Ј-end of the template a pTP⅐CAT intermediate is synthesized which jumps back to be paired to the template residues 1-3 before elongation starts (2). This jumping or sliding-back mechanism was first described for 29 (3) and appears to be universal for protein-priming replication systems (4 -6), enabling errors made during initiation and small deletions to be restored. Subsequent elongation concomitant with the displacement of the non-template strand further requires the Ad DNA-binding protein in addition to the polymerase (for review, see Refs. 7 and 8).
The multiprotein preinitiation complex assembled at the origin consists of five proteins that interact with DNA and with each other with different affinities. Several of the proteinprotein interactions have been studied individually. The polymerase interacts with the NFI DNA binding domain (9 -11) and possibly also with the DNA-binding protein. pTP contacts the Oct-1 POU homeodomain (12,13). These interactions are mostly weak, and dissociation occurs already at limited (200 mM) salt concentrations. In contrast, pTP and pol form a stable heterodimer that can only be dissociated in vitro under strong denaturing conditions (1.7 M urea) (14) or by binding of specific antibodies (15,16).
We are interested in the dynamic aspects of association and dissociation events occurring during initiation as well as during the transition to chain elongation. Previous studies employing an immobilized DNA replication system have shown that the interaction between NFI and pTP⅐pol is disrupted already early in initiation (17), in agreement with its role as temporarily acting recruitment factor. Recently dissociation of Oct-1 from pTP⅐pol complex was shown to take place upon elongation (13). No information exists about dissociation of the pTP⅐pol complex. A priori these two proteins could stay together during replication, but this would require high flexibility of the end of the nascent strand anchored to the pTP which would have to be present in the moving fork.
We have studied the dissociation of pTP and polymerase employing separation of initiation and elongation products by glycerol gradient centrifugation and find that the heterodimer is disrupted early after the synthesis of the pTP⅐CAT initiation intermediate. This is accompanied by changes in the catalytic properties of the polymerase. Moreover, we provide evidence for flexibility in the choice of the position in the origin at which the formation of initiation intermediates starts.

EXPERIMENTAL PROCEDURES
Nucleotides and DNA Templates-Ad serotype 5 TP-DNA was isolated as described previously (18 cleotide templates were obtained from Pharmacia Biotech Inc. The sequence of the wild-type 30-mer containing the template strand of the Ad5 origin is 3Ј-GTAGTAGTTATTATATGGAATAAAACCTAA-5Ј. The template strand of the GA repeat template is 3Ј-GAGAGAGATAT-TATATGGAATAAAACCTAA-5Ј. Oligonucleotides SP1 (5Ј-GATCA-CAGTGAGTAC) and SP1Cϩ6 (5Ј-TCTATTGTACTCACTGTGATC) were purified by electrophoresis on 8 M urea, 20% polyacrylamide gel electrophoresis. Oligonucleotide SP1 was 5Ј-end labeled with [␥-32 P]ATP and T4 polynucleotide kinase and purified further by polyacrylamide gel electrophoresis. Partially double-stranded primer-template structures were made by hybridizing labeled SP1 to the nonlabeled SP1Cϩ6 oligonucleotide in the presence of 0.2 M NaCl and 60 mM Tris-HCl, pH 7.5. The mixture was heated at 70°C and was allowed to cool down slowly to room temperature.
The pTP⅐pol complex and the free Ad DNA polymerase were expressed in Sf-9 cells employing recombinant baculoviruses that were constructed as described Ref. 19. Purification of both the pTP⅐pol complex and the free Ad DNA polymerase to apparent homogeneity was described previously (20). Polyclonal antibodies were raised against the pTP⅐pol complex in rabbits as described before (12).
Initiation and Partial Elongation-The standard incubation mixture (25 l) for initiation contained 0.4 g of pTP⅐pol and 0.7 g of singlestranded oligonucleotide as template in a buffer containing 20 mM Hepes, pH 7.5, 1 mM dithiothreitol, 1 mM MgCl 2 , 1 g of bovine serum albumin, and NaCl to a final concentration of 55 mM. The initiation reaction was allowed to proceed for 1 h unless otherwise indicated at 30°C in the presence of a 0.75 M concentration of one of the [␣-32 P]dNTPs (5 Ci). The reaction was stopped by the addition of 10 l of 0.25 M sodium pyrophosphate and 2 l of 0.2 M EDTA. Reaction products were precipitated with 5 l of trichloroacetic acid (20% final concentration) for 20 -30 min at 0°C. The precipitate was spun down by centrifugation for 15 min at 12,000 rpm in an Eppendorf centrifuge, and the pellet was dissolved in Laemmli buffer (2% SDS, 10% glycerol, 5% ␤-mercaptoethanol, 100 mM Tris, pH 9.0, 0.02% bromphenol blue). Samples were heated for 5 min at 100°C and separated by electrophoresis in 7.5% polyacrylamide-SDS gels and detected by autoradiography.
Initiation coupled to partial elongation was performed under conditions similar to those for initiation, except for the added dNTPs. One of the four dNTPs was labeled with ␣-32 P (5 Ci, 0.75 M unless otherwise indicated). The other dNTPs were unlabeled and present at 40 M; dGTP was replaced by 20 M dideoxy-GTP. This results in an elongation block after position 26, the first cytidine residue in the template. For the kinetic studies the concentrations of the dNTPs were variable as indicated in the legends. Samples were treated as described above for the initiation reaction.
Glycerol Gradient Sedimentation-The standard incubation mixture (200 l) for glycerol gradient analysis contained 1.6 g of pTP⅐pol and 5 g of single-stranded oligonucleotide (wild-type or mutant template ⌬3G7C), 4 g of Ad5 TP-DNA, or 5 g of double-stranded oligonucleotide in a buffer containing 20 mM Hepes, pH 7.5, 1 mM dithiothreitol, 1 mM MgCl 2 , 8 g of bovine serum albumin, and NaCl to a final concentration of 55 mM. Replication was allowed to proceed for 60 min at 30°C in the presence of 0.75 M [␣-32 P]dCTP (2 Ci), unlabeled ddGTP (40 M), and dATP and dTTP (40 M). After incubation a 190-l sample was loaded on a glycerol gradient as described below, and a 10-l sample was withdrawn for control.
Samples were layered on top of a 5-ml linear 18 -30% (v/v) glycerol gradient containing 50 mM Hepes, pH 7.5, 1 mM dithiothreitol, 1 mM EDTA, 150 mM NaCl, and 14 g/ml heparin. Gradients were centrifuged for 24 h at 50,000 rpm in an SW50 rotor at 4°C. Fractions of 120 l were collected and analyzed by electrophoresis in 7.5% polyacrylamide-SDS gels. 32 P-Labeled replication products were visualized by autoradiography or PhosphorImager. The pTP⅐pol complex was visualized by immunoblotting using an anti-pTP⅐pol antiserum as described before (12).
Control gradients with glucose-6-phosphate dehydrogenase (104 kDa, 5 g) and luciferase (62 kDa, 2,500 microunits) as molecular mass markers were run under similar conditions. Fractions of 120 l were collected and analyzed. For analysis of the glucose-6-phosphate dehydrogenase activity 5 l of the fraction was mixed with 12.5 l of 6.67 mM glucose 6-phosphate, 25 l of 5 mM NADP ϩ , in the presence of 50 mM Tris, pH 7.6 and 100 mM MgCl 2 ; after an incubation of 10 min the increase in NADPH that is formed in this reaction was measured by the absorbance at ϭ 340 nm. To determine the position of luciferase in the gradient, 10 l of the fractions was mixed with luciferin. Luciferase levels were determined using a Lumac/3 M Biocounter M 2010A. Luciferin was purchased from Sigma.
DNA-primed Polymerization-The hybrid molecule between SP1 with SP1Cϩ6 containing a 5Ј-protruding end of six nucleotides was used as a primer-template for DNA polymerization. Standard incubation mixtures of 12.5 l contained 0.12 ng of the hybrid molecule, 12.5 ng of the Ad DNA polymerase in a buffer containing 50 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, 4% glycerol, 0.1 mg/ml bovine serum albumin, and 40 a M concentration of the four dNTPs. After a preincubation for 15 min at 20°C, reactions were started by the addition of MgCl 2 to a final concentration of 1 mM. After incubation at 37°C for the indicated time, the reactions were stopped by the addition of EDTA to 10 mM. Samples were analyzed by 20% polyacrylamide gel electrophoresis in 8 M urea and autoradiography.

RESULTS
Dissociation of the pTP⅐pol Complex during Ad DNA Replication-We incubated a 30-nucleotide-long template containing the single-stranded Ad5 origin with pTP⅐pol and a low concentration (0.75 M) of [␣-32 P]dCTP, 40 M dATP, 40 M dTTP, and 20 M ddGTP. Under these conditions partial elongation until position 26, the first C in the template, can take place, whereas also part of the pTP⅐CAT product accumulates. At higher dCTP concentrations all pTP⅐CAT products can be extended, indicating that it is an intermediate in replication (1). By using low dCTP concentrations sufficient amounts of the intermediate product accumulate to enable the study of both products in one reaction. The reaction products were subsequently separated by glycerol gradient centrifugation. To disrupt the interaction between proteins and DNA we included heparin at 14 g/ml in the gradient. This concentration was chosen carefully such that the pTP⅐pol heterodimer remained intact as much as possible during centrifugation. When higher heparin levels or high salt were used, we observed spontaneous dissociation of the pTP⅐pol complex, most likely induced by centrifugal forces.
Fractions were collected and analyzed by immunoblotting with a polyclonal antiserum raised against the pTP⅐pol complex, which preferentially recognizes epitopes on pTP and is less reactive against pol. Fig. 1A shows that most of the pTP coelutes with the polymerase near the bottom of the gradient, which corresponds to the position of the intact pTP⅐pol heterodimer, present in excess. The 32 P-labeled pTP⅐26 replication product sedimented around the position of the 104-kDa marker protein (Fig. 1, B and C), corresponding with its molecular mass (97 kDa, assuming that the DNA is double-stranded because of renaturation). This shows that this replication product has dissociated from the polymerase. In contrast, the pTP⅐CAT intermediate was found distributed over two peaks (Fig. 1C), one cosedimenting with pTP⅐26 and one at the pTP⅐pol position, indicating that approximately 40% of the pTP⅐CAT is still bound to the polymerase. Employing a mutant single-stranded origin that was mutated at position 7 from G to C (G7C), we also analyzed the sedimentation behavior of a pTP⅐7N product. This product sedimented similarly to the pTP⅐26 product (data not shown).
Replication products formed on the natural TP-DNA template and analyzed in the same way showed a similar dissociation pattern of the replication products, with pTP⅐26 dissociated and the pTP⅐CAT intermediate distributed over two peaks (Fig. 2, A and D). The same results were obtained at various pTP⅐pol or template concentrations or when NFI was included in the reaction (not shown). To determine the moment of dissociation in more detail we performed the reaction in the presence of one (dCTP) or two (dCTP and dATP) nucleotides allowing only the formation of pTP⅐C or pTP⅐CA products. In contrast to pTP⅐CAT, both products cosedimented almost exclusively with the pTP⅐pol complex (Fig. 2, B-D). This shows that dissociation only starts after the synthesis of pTP⅐CAT.
After pTP⅐CAT formation the intermediate jumps back. One possibility is that jumping back and dissociation are somehow coupled. To test this we used an oligonucleotide template lacking the first three nucleotides and mutated at position 7 (⌬3G7C, 3ЈG 4 TACTTA . . . ). On this template jumping back is not possible, and the presence of a C at position 7 forces the use of the first G in the mutant template for pTP⅐CAT formation. As expected, replication on this template in the presence of dCTP, dATP, and dTTP resulted in the synthesis of pTP⅐CAT only without pTP⅐26 formation. Subsequent analysis in a glycerol gradient (Fig. 3) showed a distribution of pTP⅐CAT similar to that with wild-type, despite the absence of a jump. These results suggest that dissociation occurs independently of the jumping back step.
Dissociation from pTP Enhances the Polymerase Activity-Dissociation of the pTP⅐pol complex implies that the free polymerase performs elongation. To compare the polymerizing efficiencies of the free and pTP⅐bound enzyme we prepared highly purified Ad DNA polymerase and pTP⅐pol complex from recombinant baculovirus-infected insect cells and measured the rate of DNA synthesis on a DNA-primed template. Equimolar amounts of the free and the complexed polymerase were preincubated with all four nucleotides and a 5Ј-end-labeled 15-mer primer base paired to a 21-mer template (see "Experi-mental Procedures"), in the absence of Mg 2ϩ . After initiation of the reaction by the addition of 1 mM MgCl 2 , extension of the primer was allowed for the indicated periods, and the products were analyzed on a sequence gel. As shown in Fig. 4, almost of all primers were elongated within 30 s by uncomplexed DNA polymerase, whereas this occurred only after 32 min for the pTP⅐pol complex. By quantitation of the products we calculated a 10-fold higher initial rate of elongation for the free polymerase compared with the pTP⅐pol complex, indicating that dissociation from pTP after initiation renders the DNA polymerase more efficient for elongation. Formally we cannot exclude that the properties of free polymerase and dissociated DNA polymerase are not identical, but it is technically very difficult to study the properties of the dissociated polymerase since only a small percentage (less than 0.1%) of pTP⅐pol participates in the reaction.
Dissociation Is Accompanied with Changes in the Catalytic Properties and Inhibitor Sensitivity of the Polymerase-Previously we observed that the optimal conditions for formation of pTP⅐dCMP differ considerably from those required during elongation. The initiation reaction is insensitive to ddNTPs, aphidi-colin, and (S)-g-(3-hydroxy-2-phosphonylmethoxypropyl)adenine diphosphoryl, all strong inhibitors of elongation (21)(22)(23)(24)(25). Moreover, the K m for dCTP is 3.2-fold lower during initiation (1). Similar results were obtained when we compared the K m values for dATP and dTTP (Fig. 5). For the formation of the pTP⅐CAT intermediate, values of 3.8 and 1.0 M were found, respectively, whereas for pTP⅐26 this was 5.1 and 5.5 M. The various parameters are summarized in Table I. The most likely explanation for these results is that changes occur in the polymerase active site upon transition from initiation to elongation, concomitant with dissociation from pTP and jumping back.
Ad5 DNA Polymerase Can Use Different Initiation Sites Depending on the Template-With the Ad5 template (3Ј-GTAG-TAGTTA . . . ) the second GTA triplet is used for the synthesis of the pTP⅐CAT intermediate. To investigate whether it is always the second repeat that is used by the Ad5 DNA polymerase, we employed a template with a terminal repeat consisting of four times GA (3Ј-GAGAGAGATA . . . ), which occurs naturally in Ad4. As on the Ad5 template (2), only dCMP could be coupled when any of the 32 P-labeled dNTPs was added (Fig.  6A, lanes 1-4). Elongation until position 26 was possible on the GA repeat as with wild-type (Fig. 6A, lanes 5 and 6). To examine which of the 4 G residues was used for initiation we mutated the A residues at the positions 2, 4, 6, and 8 one by one to a C and studied incorporation of ␣-32 P-labeled nucleotides in the intermediate using partial elongation conditions. As shown in Fig. 6B 1-4 and 5-8), indicating that the synthesis of the intermediate had not started from template residues 1 or 3. With A6C and A8C, however, the intermediate could also be labeled with [␣-32 P]dGTP (lanes 9 -12 and 13-16). These results are consistent with a start of synthesis from template residue G5 from which a pTP⅐tetranucleotide is synthesized which jumps back to enable elongation. For the A6C and A8C mutants, these intermediates would be pTP⅐CGCT and pTP⅐CTCG, respectively. Alternatively, the results could be explained by the synthesis of a mixture of pTP⅐CG and pTP⅐CT originating from the use of both G5 and G7, which would also result in labeling with C, G, and T. We consider the latter possibility less likely since replacing dGTP by 20 M ddGTP completely blocked elongation on the A8C mutant (not shown). If a pTP⅐dinucleotide (pTP⅐CT) would have been formed starting at G5 this would have been able to jump back and continue elongation. However, in the case of synthesis of pTP⅐CTCG the presence of the dideoxy group at the 3Ј-end would prevent elongation, and this is exactly what we observed.
In conclusion, for Ad5 DNA polymerase the choice of the G residue seems not to be determined by the most proximal repeat but rather by the position of this residue. A preference exists for position 4 or 5 rather than 3, provided that a repeat is still present and starts with a G residue. This indicates a certain amount of flexibility in the choice of the starting nucleotide.

Dissociation of the pTP⅐Pol Complex Occurs at or after
pTP⅐CAT Formation-By studying the sedimentation properties of four well defined replication products (pTP⅐C, pTP⅐CA, pTP⅐CAT, and pTP⅐26) we have obtained strong evidence for dissociation of the pTP primer and the DNA polymerase early during the initiation process. In these experiments it was essential to remove the proteins from the template while keeping the pTP⅐pol interaction intact. We observed that, although the interaction between pTP and pol is reported to be very stable (14), the application of centrifugal forces in combination with FIG. 3. Dissociation is not coupled to jumping back. An oligonucleotide lacking the first 3 bases and containing a mutation at position 7 was used as template, thereby preventing jumping back. Panel A, the resulting pTP⅐CAT products synthesized in the presence of dCTP, dATP, and dTTP were analyzed as in Fig. 1. Panel B,  high salt or high heparin concentrations disrupted the complex. We finally found conditions (low concentrations of heparin) under which the pTP⅐pol complex remained intact while the contacts with DNA were disrupted.
We observed dissociation of the pTP⅐pol complex after the synthesis of the pTP⅐CAT intermediate. Approximately 60% of the pTP⅐CAT product becomes dissociated. That dissociation is not yet complete might indicate that the events triggering dissociation take place gradually. One of the events occurring after the pTP⅐CAT formation is the jumping back step. However, a link between dissociation and jumping back could not be established since the replication product formed by employing the ⌬3G7C template (on which no jumping back of pTP⅐CAT is possible) showed the same distribution pattern as observed using the wild-type template. After elongation of the pTP⅐CAT intermediate additional dissociation occurs and is (almost) complete when seven nucleotides have been synthesized.
We have strong indications that other events, besides jumping back, are taking place after pTP⅐CAT synthesis. Before and after pTP⅐CAT formation the catalytic properties of the DNA polymerase differ considerably as revealed by changes in the K m for dCTP, dATP, and dTTP and by changes in inhibitor sensitivities (summarized in Table I). This indicates a change in the active site of Ad DNA polymerase after pTP⅐CAT formation. Such a conformational change in the polymerase active site might also influence the pTP interaction domain of the polymerase and cause dissociation of the pTP⅐pol complex. Similarly in DNA polymerase III holoenzyme the dissociation of two ␤ subunits was accompanied by a conformational change (26).
Previously a role was proposed for TP or pTP in elongation, based upon studies employing isolated nuclei and monoclonal antibodies against the pTP. However, these antibodies were not well characterized, and the inhibition occurred also in uninfected cells (27). Although we cannot completely rule out a role for pTP in vivo, for instance related to its property to attach to the nuclear matrix (28), our experiments make a role in vitro very unlikely. Assuming that a maximal rate of DNA replication gives the virus a selective advantage, dissociation of pTP from the polymerase is a logical event since the polymerase is rendered more efficient after the release of pTP (this paper and 29). Moreover, dissociation of pTP activates the exonuclease activity thus enabling proofreading (20).
Flexibility in the Site of Initiation-We have previously mapped the precise initiation site used in Ad5 (with the template sequence 3Ј-GTAGTAGTT . . . ) at position G4, from which a pTP⅐trinucleotide intermediate is formed. On templates containing a different sequence consisting of a 4-fold repetition of two nucleotides (3Ј-GAGAGAGA . . . ) and mutants thereof, position G5 is used, and a pTP⅐tetranucleotide intermediate is formed. The selection of the initiation site is apparently determined by the template sequence rather than by the distance from the molecular end or the conserved pTP⅐pol binding site. Assuming that the binding of pTP⅐pol to the core origin is independent of the sequences near the molecular end, this suggests a certain amount of flexibility in locating a G residue of the template in the active site. Recent structural comparisons of a proofreading complex of the Klenow fragment of Escherichia coli DNA polymerase I and the "polymerizing" complex of the Taq DNA polymerase show that DNA polymerases can make a sliding motion along the template, indicating considerable flexibility in protein-DNA contacts (30).
Using a variety of templates, the first residue coupled to pTP is always a dCMP, suggesting that this is an intrinsic property of Ad5 pol. Similar restrictions for a specific starting nucleotide are not found in other protein-priming DNA replication systems that use a similar initiation mechanism (3-5). Bacteriophages 29 and Cp-1 seem to have a built-in property to start replication from the second and third template residues, respectively, irrespective of the nucleotide in this position (3,5), indicating that the position rather than the nucleotide determines the start site. In the case of adenovirus, a limited amount of flexibility in the position seems to be permitted, probably dictated by the need for a G in the template. Employing the GA repeat template, the Ad5 DNA polymerase prefers to start from position G5 rather than G3. Given the fact that on the natural template the start is at G4, why would G5 be favored over G3? Most terminal sequences present in all adenovirus serotypes (31) allow the formation of a pTP⅐trinucleotide or pTP⅐tetranucleotide. The benefit of an initiation intermediate of three or four nucleotides long, compared with two, could be to enable proofreading because this requires the synthesis of at least three or four nucleotides to initiate, at least in E. coli pol I (32).   1-4) or with ␣-32 P-labeled dCTP, dATP, dTTP, and ddGTP (lane 6) allowing partial elongation. Lane 5 contains the wildtype origin. Panel B, partial elongation reactions on the indicated mutant templates, using one ␣-32 P-labeled dNTP as indicated as well as the other unlabeled dNTPs at 40 M except dGTP, which was added at 0.7 M.