Bacteriophage T7 DNA replication in vitro. Electron micrographic analysis of T7 DNA synthesized with purified proteins.

Extensive replication of duplex T7 DNA is catalyzed in reactions contining T7 DNA polymerase, T7 gene 4 protein, and T7 RNA polymerase. When the product of this reaction is analyzed in the electron microscope, many eye form and Y form replication intermediates are observed. Replication in vitro is not initiated at a single region of the T7 genome. However, we tentatively conclude that initiation does occur preferentially at a few specific sites along the DNA, and that these sites may be near promoters at which the T7 RNA polymerase initiates transcription.


1) Recipient of Public Health Service Career Program Award GM-
. The hyperphase contained 8 pl of 2 mM Tris, pH 8.5, 60 111 of twice-distilled H20, 2 to 3 pl of DNA sample, 8 pl of cytochrome c ( 1 mg/ml), and 65 pl of formamide. The hypophase contained 10 mM Tris, 1 m~ EDTA, pH 8.5, and 15% formamide and was used immediately after preparation. When samples were mounted under partial denaturation conditions (8), the hyperphase contained 6.5 p1 of H20, 2.5 pl of the T7 DNA sample, 2 pl of a mixture of +X174 viral and R F DNA (stock 5 pg/ml of each), 6 pI of cytochrome c (0.5 mg/ml in 0.5 M Tris, 0.05 M EDTA, pH 8.5), and 83 p1 of forrnamide (added just before spreading). The hypophase contained I mM Tris, 0.1 mM EDTA, pH 8.5, and 50% formamide.
DNA was picked up from the hyperphase on copper grids which had been coated with collodion within the previous 24 h. The grids were stained with uranyl acetate, then rotary-shadowed with platinum/palladium (80:20). Random fields or selected molecules were photographed with an AEI EM 801 or Philips 200 electron microscope a t magnifications of 4,000 to lO,oOO, then enlarged and measured with an electronic planimeter (Numonics Graphics Calculator, model 250-117). 4x174 viral DNA and relaxed circular +X RF DNA were measured as internal length standards.

Frequency of Different Types of Replication Intermediates
Formed in Vitro-Samples from the T7 DNA synthesis reactions were prepared for electron microscopy without any enrichment for replicating molecules. Since the large quantity of RNA formed in the presence of T 7 RNA polymerase obscures the analysis, samples were routinely treated with RNase. Three classes of replicative structures were observed on the T7 DNA molecules in these samples. The first class, branched DNA was full length T7 DNA molecules containing a duplex DNA arm which was shorter than either of the other two arms. These forms presumably arose from single strand breaks in the template. They could have been formed when synthesis was initiated a t a break or when one replication fork of an eye form intermediate ran into a break. The second class, eye forms, was replication bubbles which appeared similar to the structure formed in vivo during bidirect.  The frequency with which each of these structures was observed in reactions containing several combinations of enzymes is summarized in Table I. As described in the preceding paper (5) very little synthesis was observed in reactions con-taining only the T7 DNA polymerase or the T7 DNA polymerase plus gene 4 protein. Essentially no replication intermediates were observed in these reactions. Approximately 10% of the DNA molecules in reactions containing T7 DNA polymerase and T7 RNA polymerase had branches or eyes. However, consistent with the fact that very little DNA was synthesized in these reactions, all of these branches and eye forms were small (56%-unit length) (Fig. l a ) . In reactions containing T7 RNA polymerase alone, or T7 RNA polymerase plus gene 4 protein, where no incorporation of [''HIdTTP was detected, only linear DNA molecules were observed after the RNase treatment.
In contrast to these results, reactions containing T7 DNA polymerase, gene 4 protein, and T7 RNA polymerase produced a high frequency of replication intermediates. More than half of the DNA molecules contained some type of

Electron Micrographic
Analysis of Ti' DNA Produced in Vitro replication fork after 10 min of synthesis, and 17% of the molecules contained more than one such structure. Furthermore, in this reaction large regions of T7 DNA were replicated; several of the eye and Y forms extended along more than 50% of the genome. This result was consistent with the measurement of [3H]TTP incorporation, which indicated that about 0.3 eq of DNA was synthesized in this reaction. We have also determined the frequency of each type of replication intermediate in samples taken at several different times during the reaction (Table 11). In general the extent of replication observed in the electron microscope correlates well with the amount of r3H]DNA synthesized in each reaction.

Replication in Vitro May Start at Specific Sites on the T7
Genome-In uiuo, T7 DNA replication is initiated predominantly at a site about 17% from the left end of the genome. The replication eye forms produced in vitro by the T7 DNA polymerase in the presence of gene 4 protein and T7 RNA polymerase were not initiated at a unique site on the T7 DNA. However, we noted that DNA synthesis in vitro appeared to be preferentially initiated in certain regions of the DNA. For example, more than half of the small (55% unit length) eye forms were clustered in the middle of the DNA molecules between 40 and 60%. To more precisely determine the sites of initiation we used partial denaturation mapping (2, 8), a technique which relies on an asymmetrical distribution of ATrich regions along the DNA to orient replicating molecules. The partial denaturation maps constructed for T7 DNA by Dressler et al. (2) and by Gomez and Lang (8) have several features which are useful for orienting DNA molecules. First, there are major denaturation sites located at about 16,21, and 26% from the genetic left end of the DNA. Thus, the region from 15 to 30% is preferred for denaturation while the corresponding region at the left end (70 to 85%) is one of the last regions to exhibit strand separation. This feature has been used to orient most of our partially denatured DNA molecules, and 45 of the 48 DNA molecules shown in Fig. 2A are denatured in the region from 15 to 30%. In contrast, only 5 molecules are denatured in the region from 70 to 85%, and 4 of these molecules are highly denatured. Other features of the T7 DNA partial denaturation map were also used to orient the molecules. For example, denaturation of the regions at 60 to 70% and around 90% is preferred relative to the corresponding regions at 30 to 40% and around 10%. However, denaturation sites are not located at precise positions on the T7 DNA molecule. The number and position of these sites fluctuate considerably even among molecules taken from a single preparation. Furthermore, to maintain the integrity of small replication eye forms on the DNA, it was necessary for us to keep the overall denaturation low (ll%), which produced an average of 4 denatured sites/DNA molecule. Gomez and Lang (8) have found that an average of 5 to 15 denatured sites/T7 DNA molecule seems to optimize pattern recogni-

FIG. 2. Replication intermediates after partial denaturation.
A, line drawings of replication intermediates oriented after partial denaturation. Synthesis was carried out for 20 min at 30°C as described in Table 11. Samples from the reaction were analyzed in the electron microscope after partial denaturation as described under "Methods." The overall denaturation of the DNA is 11%, and the average number of denatured sites per DNA molecule is 4.

The denatured regions are shown by the horizontal black bars. The dashed lines represent alternative orientations for the branches. B, denaturation histogram constructed from the data in A.
tion. Thus, the orientation of some of the DNA molecules shown in Fig. 2A is ambiguous. However, a histogram (Fig.  2B) constructed from the data shown in Fig. 2A is in good agreement with the results of Dressler et al. (2) and Gomez and Lang (8), suggesting that at least most of the molecules have been oriented correctly.
Most of the replicated regions observed on the oriented T7 DNA molecules (Fig. 2 A ) were located on the right half of the genome, and molecules with more than one eye or Y were frequent. In this particular experiment, more than half of the small eyes were located between 43 and 58%. Representative molecules are shown in the photographs in Fig. 3.
To determine the preferred sites for the initiation of these eye forms, we have constructed the histograms shown in Fig.  4. Since it is likely that replication does not always proceed at  the same rate in both directions from an initiation site, the origins of replication can be most accurately mapped using molecules with small eye forms (1 to 5%). The histogram constructed from these data (Fig. 4a) shows four preferred regions of initiation on the T7 genome. These regions are located at approximately 35, 45 to 50, 55, and 75% from the genetic left end of the DNA. An inspection of the individual molecules ( Fig. 2 A ) used to construct this histogram suggests that the region at 45 to 50% is actually at least two sites. The individual molecules were grouped into sets of overlapping eye forms and the average center was determined for each set. The major initiation sites as determined from this average were at 34, 46, 50, 55, and 73%. Less prominent sites were observed in the regions 17 to 22%, 61 to 65%, and 95 to 100%.
We have also analyzed eye forms from micrographs of DNA spread under nondenaturing conditions. We have oriented these molecules (by inspection and also with the aid of a computer) to provide the best match with the initiation sites observed in the partial denaturation analysis (Fig. 5, a and b). A histogram, in which lengths were measured to the closest end, was also constructed from the data using only molecules with eye forms of length 510% (Fig. 5c). These data support the conclusion that DNA synthesis in vitro is not initiated at random sites along the T7 genome. A statistical analysis (chi square with a single classification and equal expectations) of the data in Fig. 5c indicates that the probability that these eye forms are located at random sites along the DNA is extremely small ( x 2 with 49 degrees of freedom = 185.6, p << 0.005). Half of the small eye forms (20/40) observed on these molecules are located in the region between 40 and 60%. The distribution of small eye forms on these nonoriented DNA molecules is in general agreement with the distribution of initiation sites mapped by partial denaturation. In uiuo, T7 DNA replication is initiated predominantly at a site about 17% from the genetic left end, at least during early stages of infection (2). However, since phage which have deleted this region of the chromosome are viable (16), other sites for the origin of replication must exist. Indeed, when cells were infected with T 7 phage a t a high multiplicity, Dressler et al. (2) found numerous eye forms in which the eye was located in the center or right portion of the T7 DNA molecule. Panayotatos and Wells (17) have constructed a plasmid containing a T7 DNA segment from 12.05 to 16%. This DNA is an active template for the T7 RNA polymerase, and also stimulates DNA synthesis in a cell-free system prepared from Alternatively, additional proteins may be required to direct the initiation of replication to the site at 17%.
Richardson et al. (10) have purified a bacterial protein which stimulates DNA synthesis on intact duplex T7 DNA. Although the purified protein contains no detectable nuclease activity, in the presence of T 7 DNA polymerase and gene 4 protein the "initiation protein" apparently acts by breaking the r-strand of the template predominately at 18% and less frequently at 31% and 86% from the left end. The resulting nick is then used to initiate DNA synthesis, and a branched DNA molecule is produced in which the newly synthesized DNA remains covalently attached to the template. If this protein.is used for initiation in viuo, then additional factors must be required to release the product from the template and generate eye-shaped replication intermediates. It is unclear how the T7 RNA polymerase might function in this process.
Our experiments with purified proteins suggest that the T7 RNA polymerase can function in the initiation of T7 DNA replication. Other experiments, using a T7 gene 1 mutant which produces a temperature-sensitive RNA polymerase, suggest that the T7 RNA polymerase may also play a role in phage DNA replication in uiuo (19). Until a system is developed in which DNA replication with purified proteins exactly mimics T7 DNA replication in uivo, the molecular mechanism for initiation of replication remains uncertain. However, the mechanism by which T 7 DNA replication is initiated in our in vitro system may closely resemble the mechanism of initiation in uiuo.