Analysis of Sequential Steps of Nucleotide Excision Repair in Escherichia coli Using Synthetic Substrates Containing Single Psoralen Adducts*

Escherichia coli ABC excinuclease initiates the re- moval of dodecanucleotides from damaged DNA in an ATP-dependent reaction. Using a synthetic DNA frag- ment containing a psoralen adduct at a defined position we have investigated the interaction of the components of the enzyme with substrate by DNase I footprinting. We find that the UvrA subunit binds to DNA specifically in the absence of cofactors and that the binding affinity is stimulated about 4-fold by ATP and only marginally inhibited by ADP. The UvrA-DNA complexes formed in the absence of co-factors or in the presence of either ATP or ADP are remarkably similar. In contrast, adenosine 5’-0-(thiotriphosphate) increases nonspecific binding and completely abolishes the UvrA footprint. The UvrB subunit can associate with the UvrA subunit on DNA in the absence of ATP, but this ternary UvrA-UvrB-DNA complex is qualitatively different from that formed in the presence of ATP. The UvrC subunit elicits no additional change in the UvrA-UvrB footprint. Helicase I1 (UvrD protein) does not alter the UvrA-UvrB footprint but does ap-pear to interact at the 6‘-incision site of the postinci- sion complex. DNA polymerase I fills in the excision gap in the presence or absence of helicase I1 and ap-parently

( 1 Present address: Microprobe Corp., 1725 220th St. 104, Bothell, WA 98021. ** To whom reprint requests should be sent. (Sancar and Rupp, 1983). This enzyme is composed of the UvrA, UvrB, and UvrC proteins, which assemble on DNA at the site of the altered nucleotides (Van Houten et al., 1987). The UvrA subunit, which has a DNA-independent ATPase activity, is the damage recognition subunit with a 1000-fold higher affinity for DNA containing a damaged nucleotide compared to nonadducted DNA (Van Houten et al., 1987). Binding of UvrA to DNA is stimulated by ATP hydrolysis Yeung et al., 1983) and by addition of the UvrB subunit (Van Houten et al., 1987). The UvrB subunit interacts with the UvrA subunit to form a stable complex with damaged DNA (Kacinski and Rupp, 1982;Yeung et al., 1983;Thomas et al., 1985). This ternary complex exhibits a %fold increase in the rate of ATP hydrolysis Oh and Grossman, 1986) compared to UvrA alone. Addition of the UvrC subunit to this complex results in the hydrolysis of the 8th phosphodiester bond 5' and the 5th phosphodiester bond 3' to the altered nucleotides (Sancar and Rupp, 1983;Sancar and Sancar, 1988). The exact role of ATP binding and hydrolysis in the formation of the active complex and/or incising DNA is not known although hydrolysis of ATP appears to be essential for the nicking activity (Seeberg, 1978).
The postincision ABC excinuclease complex does not dissociate but requires both DNA polymerase I and helicase I1 for release of the three Uvr proteins as well as the excised dodecamer carrying the modified nucleotide (Caron et al., 1985;Husain et al., 1985). Gap filling by DNA Pol I' occurs concomitantly with the excision step, and DNA ligase completes the repair process by sealing the repair patch (Husain et al., 1985;Caron et al., 1985). I n vivo and in vitro estimates of the number of nucleotides inserted per repair event give a range of 13-25 (Ley and Setlow, 1977;Ben-Ishai and Sharon, 1978;Carlson and Smith, 1981;Matson and Bambara, 1981;Kuemmerle et al., 1981Kuemmerle et al., , 1982Caron et al., 1985). Genetic and biochemical experiments give conflicting results regarding the importance of helicase I1 in the gap-filling step mediated by DNA Pol I. Kumura et al. (1985) have suggested that helicase I1 acts by allowing the turnover of the ABC excinuclease complex but is not necessary for repair synthesis by Pol I while Caron et al. (1985) have reported that helicase I1 is absolutely required for repair synthesis by Pol I. In vivo studies using several different alleles of helicase I1 mutants provided evidence that normal tracts of repair synthesis occur at low fluences of UV light (Carlson and Smith, 1981;Kuemmerle et al., 1982). Since helicase I1 needs a single strand gap of at least 12 nucleotides

16553
This is an Open Access article under the CC BY license. to move in a 3' to 5' direction (on the single strand) and unwind the two DNA strands, the enzyme may not participate in turnover of ABC excinuclease by DNA unwinding but may act by direct protein-protein interactions (Matson, 1986).
We have recently reported the use of DNase I footprinting to follow the formation of the active ABC excinuclease complex on a DNA fragment containing a site-specific psoralen adduct (Van Houten et al., 1987). In the present study we use this technique to investigate the role of ATP in nucleotide excision as well as the interactions of the six proteins (UvrA, -B, and -C, helicase 11, Pol I, DNA ligase), which are thought to be necessary and sufficient for nucleotide excision repair. In addition, using a double-stranded M13 DNA substrate carrying a single psoralen adduct at a defined position, we have measured the size of the repair patch made by Pol I in t h e presence or absence of helicase 11.

MATERIALS AND METHODS
Chemicals and Enzymes-DNase I, restriction endonucleases, T4 polynucleotide kinase, E. coli DNA ligase, and T4 DNA ligase were purchased from Bethesda Research Laboratories. ATP, ATP+, and ADP (less than 0.1% ATP) were purchased from Boehringer Mannheim, and NAD and NMN were from Sigma. Radioisotopes were obtained from Du Pont-New England Nuclear or ICN Radiochemicals.
The UvrA, UvrB, and UvrC subunits of ABC excinuclease were prepared and stored as described previously . Helicase I1 was a kind gift of S. Matson (University of North Carolina).
DNA Substrates-A 137-bp-long DNA fragment containing a psoralen adduct was constructed as described elsewhere (Van Houten et al., 1987). Briefly, a 12-mer containing a 4'-hydroxymethyl-4,5',8trimethylpsoralen (HMT)-furan side monoadducted thymine was ligated to five other properly kinased oligomers of overlapping complementarities to obtain the desired duplex. To ensure full-length duplex DNA without internal nicks, the ligation products must be purified using successive denaturing and nondenaturing 8% polyacrylamide gels. M13mp19 replicative form containing a HMT-furan side monoadducted thymine at the KpnI site of the polylinker was synthesized by the method of Kodadek and Gamper (1988).
ABC Excinuclease Assay-The reaction buffer contained 50 mM Tris-HC1,50 mM KCl, 10 mM MgC12, 2 mM ATP, 2 mM dithiothreitol, and 100 pg/ml bovine serum albumin. The UvrA, UvrB, and UvrC subunits were incubated individually or in combinations in this buffer at 37 "C for 5 min prior to the addition of DNA. We have found this preincubation necessary to yield maximum enzymatic activity and to give more reproducible results, probably by promoting the formation of AzBl complex (Orren and . DNase Z Footprinting-DNase I footprinting with the ABC excinuclease alone was performed as described previously (Van Houten et al., 1987). The following modifications were made for footprinting experiments which were done using ABC excinuclease under turnover conditions. The Uvr subunits (5-40 nM) were incubated with the HMT-modified DNA (0.5-2.0 nM) for 20 min at 37 "C, and then all 4 dNTPs (40 p~ each), helicase I1 (5-20 nM), and Pol I (0.5 units) were added, and the reaction mixture was incubated for an additional 10 min at 22 "C. In reactions containing ligase the mixtures were first incubated with E. coli DNA ligase (0.5 units) and NAD at 2 mM, and then NMN was added to 10 mM before adding CaC12 (2 mM) and DNase I (0.5 ng). This order of addition was to ensure the sealing of the repair gaps but not the nicks of DNase I and thus enabled us to obtain a footprint in the presence of ligase. Following DNase I treatment the samples were made 15 mM in EDTA, frozen in dry iceethanol, and processed as described previously (Van Houten et al., 1987).
Repair Patch Assay-The size of the nucleotide excision repair patches was determined as follows. The UvrA (0.5 pmol), UvrB (1.3 pmol), and UvrC (1.5 pmol) subunits were added to 50 p1 of reaction buffer, and after 5 min at 37 "C, modified or unmodified M13mp19 (8.3 fmol) was added to the enzyme. The mixture was incubated at 37 "C for 15 min, and then dATP, dGTP, and TTP (40 pM each) plus 30 pCi of [ c Y -~~P I~C T P (6000 Ci/mmol), Pol I (2 units), helicase I1 (0.43 pmol), and T4 DNA ligase (2 units) were added, and the mixture was incubated for an additional 10 min at 37 "C. The reaction was terminated by the addition of EDTA to 15 mM followed by extraction with an equal volume of phenol. The aqueous phase was extracted with ether and the DNA precipitated with ethanol. The DNA was then resuspended in buffer and digested with PuuII and one of the enzymes that cuts in the polylinker region, thus liberating two labeled DNA fragments. The fragments were separated on an 8% polyacrylamide gel. After autoradiography the bands corresponding to each fragment were cut out and the radioactivity measured by Cerenkov counting. From the distribution of label in the two restriction fragments the patch size was estimated. The amount of radioactivity incorporated 3' to the 3"incision site of ABC excinuclease was taken to be a measure of nick translation.

RESULTS
The Role of ATP in Formation of UvrA-DNA or UvrA. UvrB. DNA Complexes-ABC excinuclease is an ATP-dependent nuclease, but the exact role of ATP in the nuclease function is not known. Filter binding studies showed that ATP stimulated specific binding of UvrA, the ATPase subunit of the enzyme, to UV-irradiated DNA while ADP inhibited specific binding, and ATPyS abolished specific binding by increasing nonspecific binding of the protein . To gain an insight into the role of ATP hydrolysis in the formation of specific complexes we conducted DNase I footprinting with various combinations of the psoralen-adducted region even in the absence of ATP. In the presence of ATP, the affinity is increased about 4-fold without any change in the quality of the footprint. This suggests that the UvrA-DNA complex does not undergo a significant conformational change during ATP hydrolysis. As reported previously (Van Houten et al., 1987) when UvrB is added to UvrA in the presence of ATP, two characteristic changes are seen: the footprint shrinks from 33 to 19 bp and a DNase I-hypersensitive site appears at the 11th phosphodiester bond 5' to the HMT-modified thymine (indicated by an asterisk). In the absence of ATP, the UvrA-UvrB footprint was quantitatively as well as qualitatively different. The UvrA-UvrB footprint is smaller than the UvrA footprint, but this shrinkage is not as prominent as in the presence of ATP. Perhaps more significantly, the hypersensitive site produced in the presence of ATP is never observed in the absence of ATP, even at very high UvrB concentrations (compare lanes 17 and 21, and data not shown). As expected, addition of UvrC to the UvrA.UvrB .DNA ternary complex results in cleavage of the DNA only when ATP is present (compare lanes 18 and 22). This subunit had no discernible effect on the UvrA-UvrB footprint in the absence of ATP (compare lanes 6 and 7). It thus appears that UvrC does not associate with the UvrA. UvrB .DNA complex in the absence of ATP.
To discern whether the binding of ATP, its hydrolysis, or the binding of ADP is responsible for formation of the productive UvrA. UvrB .DNA complex (defined as the complex which upon addition of the UvrC subunit results in bi-incision) we performed footprinting experiments in the presence of ADP and the nonhydrolyzable ATP analog, ATPyS. The results of these experiments are shown in Fig. 2. UvrA binds specifically to the HMT-modified DNA in the presence of ADP with about 2-fold less affinity compared to its binding in the absence of any cofactor (compare lane 4 with lane 7).

BOTTOM TOP
Lanes I 2 3 4 5 6 7 8 9 IO " I lll_- 3. Effects of helicase I1 on the formation of the ABC excinuclease complex. Helicase I1 (20 nM) was added to the 137bp HMT-modified DNA (0.5 nM) in the absence or presence of UvrA (20 nM), UvrB (26 nM), or UvrC (30 nM), and the DNA was subjected to limited digestion with DNase I. The arrow indicates the new band which appears when helicase I1 is added to the postincision ABC excinuclease complex. A, B, C, and H refer to the UvrA, UvrB, and UvrC subunits, and helicase 11, respectively. As UvrA has a 10-fold higher affinity for ADP (Ki = 20 pM) compared to its affinity for ATP (Kacinski et al., 1981;Seeberg and Steinum, 1982), we conclude that under our experimental conditions all UvrA is bound to ADP and that the ADP-bound form of UvrA makes essentially the same contacts with substrate DNA (no qualitative change in the footprint) but with somewhat lower affinity. ADP also lowers the affinity to nonsubstrate DNA (compare lanes 6 and 9). A UvrA. UvrB .DNA complex also formed in the presence of ADP (Fig. 2, compare lanes 20 and 21); however, this complex lacks the specific hypersensitive bond that is formed in the presence of ATP and thus is a nonproductive complex similar to the one formed in the absence of any cofactor. ATP-yS inhibited specific binding of UvrA at low concentrations of the protein (compare lanes 7 and 14) and at high protein that ATP hydrolysis is necessary for the formation of the productive complex.
Dissociation of the Postincision Complex-Earlier studies indicated that ABC excinuclease remains bound to DNA following incision and that the three subunits along with the excised oligomer are released by the joint action of DNA polymerase I and helicase I1 (Husain et al., 1985;Caron et al. 1985). We analyzed the possible protein-protein and DNAprotein interactions during this process.
As a first step, the effect of helicase I1 on the ABC excinuclease complex was investigated. It has been suggested that helicase I1 alone may interact with the complex and release the UvrC subunit (Caron et al., 1985). Fig. 3 shows the results of our footprinting studies with helicase 11. As is apparent from this figure helicase I1 does not bind to HMT-modified DNA specifically (lanes 5 and 10) nor does it have any effect on the UvrA or UvrA-UvrB footprint (lanes 2 and 3 uersus  lanes 6 and 7). It seems, however, that helicase I1 has a minor but reproducible effect on the footprint of the postincision complex: a new DNase I-sensitive site is produced on the nonadducted strand across from the 5"incision site of the adducted strand ( l a n e 8, arrow). This may be taken as evidence for (but not as a proof of) binding of helicase I1 near the 5"incision site as has been proposed by Matson (1986).

5"
DNoseI: ---------   figure. Thus, it appears that under our experimental conditions Pol I does not bind stably to the proteins in the postincision complex nor to the two nicks produced by ABC excinuclease. Under similar experimental conditions it has been shown that Pol I produces a 20-bp-long footprint around a nonprotected nick at a defined location (Joyce et al., 1986). An attempt to obtain such a complex by including into the reaction mixture the first nucleotide incorporated into the excision gap (dCTP) also failed to elicit any Pol I footprint even though the nucleotide was efficiently incorporated into the gap (data not shown). Similarly, in the presence of all 4 dNTPs Pol I binds to the 5' nick and fills in the gap (see below).

-+ + + + + + + + + + + -A A A A A A A A -A A A A A A A A A
To study the interactions of all six proteins thought to be necessary for complete nucleotide excision repair, Pol I, helicase 11, and ligase were added to the postincision ABC exci- nuclease complex separately or in combination (under resynthesis conditions), and the resulting changes in the two nuclease incisions and the DNase I footprints were examined. By following the fate of the bands which result from the "normal" 5"incision and the "uncoupled" 3'-incision (with 5"terminally labeled DNA a band corresponding to the 3' incision is seen only when this incision is not coupled to the incision on the 5' side), the accessibility of these two cleavage sites in the postincision complex can be inferred (Fig. 5A,  lanes 1-9). Pol I, when added to the postincision complex in the presence of dNTPs, reduces the intensity of the 5'incision band by filling in the gap. During this repair synthesis there is a significant level of premature terminations resulting in a "ladder" within the repair patch (Fig. 5A, lane 4 ) . Ligase when added alone to the postincision complex reduces the intensity of the normal 5"incision band and increases the intensity of the uncoupled 3"incision band (Fig. 5A, lane 5 ) . This result suggests that the 5"incision site in the postincision complex is more accessible to DNA ligase than the 3'incision site. However, the intensity of the uncoupled 3'incision band does not increase proportionally to the decrease of the 5"incision band suggesting that the 3"incision site is also accessible, yet is more sterically hindered than the 5' site, to ligase. When helicase I1 and ligase are added together to the postincision complex the decrease in intensity of the 5"incision band is less compared to that obtained with ligase alone ( l u n e 5 versus lane 7). This observation is consistent with the suggestion that helicase I1 binds near the 5'-incision site (Matson, 1986) partly blocking accessibility of this nick to ligase. The Pol I + helicase I1 and Pol I + ligase combinations were not significantly different from Pol I alone with regard to filling in the gap in the postincision complex. However, the addition of helicase I1 and ligase together with Pol I decreases the intensity of the gap-filling ladder generated by the stalling of Pol I (compare the band intensities in lane   9 versus lanes 4,6, and 8), suggesting that helicase I1 increases the processivity of Pol I in this reaction. DNase I footprinting of the top strand in the presence of various protein combinations was not very informative because it is not possible to examine the region 3' to the 5'-incision site of ABC excinuclease under our experimental conditions (lanes 10-20) where more than half of the substrate has been incised by the ABC excinuclease. In contrast, the bottom strand footprint was quite informative regarding the fate of the postincision complex. Helicase alone resulted in the appearance of an additional DNase I band as noted above (Fig. 5B, lane 6 ) . Ligase alone had no discernible effect (data not shown), Pol I + ligase and especially Pol I + helicase I1 + ligase combinations resulted in partial disappearance of the ABC excinuclease footprint (lanes 7 and 8) suggesting that under these conditions the excision gap was filled in and ligated resulting in release of the ABC excinuclease complex from a fraction of the molecules. That Pol I could perform repair synthesis in the absence of helicase I1 and in so doing displace the excised oligomer and ABC excinuclease was unexpected in view of the fact that it had been reported (Husain et al., 1985) that the joint action of Pol I + helicase I1 was necessary for turnover of ABC excinuclease. We, therefore, conducted the experiments described in the following section to examine the repair synthesis in the presence and absence of helicase I1 in more detail.
Repair Synthesis Reaction-We conducted these studies with a circular substrate because the linear substrate used in the previous experiments is only 5040% repaired under optimal conditions, presumably because the ABC excinuclease makes nonspecific complexes at the termini which interfere and [o-~'P]~CTP (30 pCi, 6000 Ci/mmol). Following removal of unincorporated radioactivity the DNA was divided into 10 aliquots which were digested with the indicated restriction enzymes. The resultant products were analyzed by separation on a nondenaturing 8% polyacrylamide gel followed by autoradiography. Panel B, the amount of radioactivity incorporated into each band in panel A was determined by scanning properly exposed autoradiographs and/or directly counting the excised bands. These values were used to generate the physical map of the distribution of label incorporated into the repair patches. The given percentages (mean of four experiments) represent the distribution of label in the repair patches and were calculated as follows: 83% of the label was incorporated in the region 5' to the SmaI incision site, 93% into the region 5' to the XbaI site, 97% 5' to the Sal1 site, 99% 5' to the PstI site, and 100% into the region 5' to the Hind111 site. These percentages are not weighted for the relative frequencies of Cs at the corresponding intervals. with formation of specific complexes.' The circular substrate used for the repair patch assay is M13mp19 replicative form containing on the minus strand a HMT furan side monoadduct to thymidine at the KpnI site in the polylinker region (Kodadek and Gamper, 1988). The presence of a psoralen at this location inhibits KpnI incision. Removal of the adduct, followed by repair synthesis, regenerates the KpnI site, making the DNA susceptible to digestion by KpnI (Fig. 6)  The HMT-modified M13mp19 DNA was repaired as described in Fig.  7, except all four dNTPs were cold and 100 ng of M13mp19 DNA was used per reaction. Following repair reaction the DNA was digested with the indicated restriction enzymes and analyzed by electrophoresis on a 0.8% agarose gel. Lanes I and 7 contain unmodified M13mp19 digested with ClaI; lanes 2 and 8 contain unmodified M13mp19 digested with ClaI and KpnI; lanes 3 and 4 contain HMTmodified M13mp19 digested with ClaI or ClaI and KpnI, respectively (see Fig. 6). Lanes 5 and 6 contain HMT-modified M13mp19 DNA which was digested with ClaI and KpnI following treatment with ABC excinuclease, Pol I, and ligase in the absence or presence of helicase 11, respectively. study the repair reaction, the HMT-modified M13mp19 DNA is first incubated with ABC excinuclease and then with Pol I + ligase or Pol I + helicase I1 + ligase in the presence of ATP and dNTPs (one of which is radiolabeled). The plasmid is then digested with PuuII to yield a 322-bp fragment containing the KpnI site. Digestion of this PuuII fragment with other enzymes that cut within the polylinker region provides a map of the distribution of label incorporated during repair synthesis. From this distribution the size of the repair patch can be determined.
The results of such a' repair patch assay are shown in Fig.  FIG. 10. Models for functions of helicase I1 and DNA polymerase I and helicase I1 in catalytic turnover of ABC excinuclease.

I-W
7A. As is apparent from this figure nearly all the label incorporated into the M13 substrate is within the PuuII fragment carrying the HMT-adducted T. Also, nearly all of the radiolabeled PuuII fragment has become sensitive to KpnI suggesting almost complete restoration of this site. Digestion with SmaI, which incises at the 3"incision site of ABC excinuclease, reveals that 83% of the label is incorporated into the 12-nucleotide gap of ABC excinuclease. Digestion with other enzymes gave the results shown in Fig. 7A and summarized in Fig. 7B. From this data it appears that under the conditions used here, Pol I simply fills in the gap to produce a dodecanucleotide patch; only in 10% of the molecules the repair patch is 12-20 nucleotides, and in 4%, 20-38 nucleotides. No molecules were detected with patches longer than 45 nucleotides.
The role of helicase I1 in the gap filling reaction was studied by conducting similar experiments with Pol I plus ligase alone. The results of such an experiment are shown in Fig. 8. Nearly the same level of repair synthesis is obtained with and without helicase, supporting the conclusion arrived at earlier with linear substrates that helicase I1 is not essential for repair synthesis and ligation (see also Fig. 5).
To demonstrate that the repair synthesis observed in the previous experiments in the presence or absence of helicase I1 was not due to repair in only a fraction of the molecules, we conducted similar experiments to regenerate the KpnI site and analyzed the reaction products by ethidium bromide staining instead of autoradiography (Fig. 9). Polymerase I restores the KpnI site nearly quantitatively even in the absence of helicase I1 (lane 5 ) . However, a subtle helicase effect was apparent, as even under these experimental conditions where ABC excinuclease was in molar excess over substrate, helicase I1 increased the efficiency of overall repair (compare lanes 5 and 6). Attempts at conducting these experiments with substantially lower concentrations of ABC excinuclease in order to see a more pronounced stimulation of repair by helicase I1 were unsuccessful. We believe this to be due to inactivation of the enzyme by the large dilutions of the ABC excinuclease subunits which were necessary to achieve substoichiometric concentrations (10-25 pM). However, in assays with randomly damaged DNA where these experiments are conducted with higher concentrations of substrate and ABC excinuclease, helicase I1 has a dramatic effect on incision (Husain et al., 1985).

DISCUSSION
In this paper we have addressed three issues regarding the mechanism of nucleotide excision in E. coli: the function of ATP in the assembly and function of ABC excinuclease, the role of helicase in repair synthesis, and the exact size of the repair patch.
Our finding that ATP is not necessary for formation of specific UvrA . DNA complexes is in agreement with the conclusion of Seeberg and Steinum (1982)   . The fact that we obtain what appears to be an UvrA-UvrB footprint even when ATP is absent suggests that there are three different types of UvrA-UvrB complexes: 1) the complex formed in the presence of ATP and absence of DNA; 2) that formed in the presence of DNA and absence of ATP; and 3) the complex formed in the presence of both. What is the role of ATP in the action mechanism of ABC excision nuclease? We know that ATP hydrolysis (not just binding) is required for the formation of AzBl and productive UvrA.UvrB.DNA com-plexes, and it has also been reported that the UvrA .UvrB complex acts as a helicase to dissociate small oligonucleotides from DNA duplexes (Oh and Grossman, 1987). It is therefore possible that the energy of ATP hydrolysis is used in the formation of a UvrA. UvrB .DNA ternary complex in which the DNA is in an altered, possibly strained, conformation, such that the addition of the UvrC subunit results in the hydrolysis of the 8th phosphodiester bond 5' and the 5th phosphodiester bond 3' to the damaged nucleotide.
We also attempted to uncover the mechanism by which Pol I and helicase I1 participate in the catalytic turnover of ABC excinuclease. Based on our data, we consider the following three models (Fig. 10). 1) Pol I alone fills in the gap and ligase seals the resulting nick but ABC excinuclease remains attached to this postrepair 3-stranded structure; helicase I1 releases the ABC excinuclease and the excised oligomer. 2) Pol I fills in the gap and releases the ABC excinuclease (still attached to the excised oligomer); helicase I1 dissociates-the excision oligomer from the enzyme, allowing the Uvr proteins to enter new rounds of repair. 3) Pol I and helicase I1 bind to the 5"incision site and move in the same direction, polymerase synthesizing the incised strand 5' to 3' while helicase I1 moves 3' to 5' on the complementary single strand (Matson, 1986) displacing the excised oligomer and ABC excinuclease. The net effect is filling in the gap and releasing the ABC excinuclease subunits and the excised dodecamer. Although model 3 seems attractive and it is supported by the evidence that helicase I1 appears to bind to the 5"incision site in our footprints, the fact that we obtain complete gap filling and sealing in the absence of helicase I1 suggests that this model cannot be entirely correct. Model 1 predicts that the postincision footprint of ABC excinuclease + helicase I1 and ABC excinuclease + helicase I1 + Pol I should be different having a distinct footprint in the former but not in the latter case. We do not see such a difference and therefore think this model highly unlikely. Model 2 is consistent with nearly all the data in this paper and with the previously published accounts of helicase I1 effects. However, it is inconsistent with the evidence for the apparent binding of helicase I1 to the postincision complex. Perhaps, a hybrid of Models 2 and 3 would account for all the observation made so far; helicase I1 acts in a secondary role during the filling in of the gap by Pol I and then releases the components of the ABC excinucleaseexcised oligomer complex that has been displaced as a result of gap filling. However, it must be pointed out that even though our data suggest that Pol I and helicase I1 bind to the 5'-incision site and travel in the same direction during resynthesis, we have failed to obtain any evidence for the presence of a supramolecular "repairosome" complex consisting of the incision and resynthesis-ligation proteins. It appears that the interaction of postincision proteins with the incision complex is too transient for detection by DNase I footprinting.
Finally, in this paper we have measured the size of the repair patch made at a defined adduct site and obtained values for the relative frequencies of the different size classes. We have found that Pol I can carry out repair synthesis even in the absence of helicase I1 and that 80-90% of the gaps are filled in without any nick translation. No patches as long as 50 nucleotides were detected. Thus, it appears that the primary repair process in nucleotide excision repair generates only short patches and that the long patches observed in vivo (Hanawalt et al., 1979) may be the result of joint excisionrecombination repair processes.