Role of ATP in removal of psoralen cross-links from DNA of Escherichia coli permeabilized by treatment with toluene.

Removal of interstrand cross-linked from DNA was examined in Escherichia coli permeabilized by treatment with toluene. Under these conditions, the reaction requires ATP and Mg2+, and the mechanism appears to be similar to that occurring in whole cells. Under optimum conditions, the rate constant was 0.06 min-1. Genetical, physical, and biochemical analysis of the repair process suggest the following mechanism. In an ATP-dependent reaction, the uvrA and uvrB gene products cleave a phosphodiester bond on the 5' side of one arm of the cross-link, producing a 3'-OH terminus. Subsequently, DNA polymerase I (5'-3' exonuclease activity) makes a second strand cut on the 3' side of the cross-link in the same DNA strand, completing removal of the covalent link between complementary strands. The second reaction did not occur in a uvrD- strain, which had normal levels of DNA polymerizing activity. The uvrD gene may regulate the specificity or activity of the 5'-3' exonuclease of DNA polymerase I in vivo.


Removal of interstrand
cross-links from DNA was examined in Escherichia coli permeabilized by treatment with toluene. lJnder these conditions, the reaction requires ATP and Mg"+, and the mechanism appears to be similar to that occurring in whole cells. IJnder optimum conditions, the rate constant was 0.06 min I. Genetical, physical, and biochemical analysis of the repair process suggest the following mechanism. In an ATP-dependent reaction, the uurA and uvrB gene products cleave a phosphodiester bond on the 5' side of one arm of the cross-link, producing a 3'-OH terminus. Subsequently, DNA polymerase I (5'-3' exonuclease activity) makes a second strand cut on the 3' side of the cross-link in the same DNA strand, completing removal of the covalent link between complementary strands. The second reaction did not occur in a uurD strain, which had normal levels of DNA polymerizing activity. The uvrD gene may regulate the specificity or activity of the 5'-3' exonuclease of DNA polymerase I in vim the role of small molecules in metabolic processes of replication (3,4) and repair (5)(6)(7)(8)(9)(10)(11)  The presence of the endAl mutation in each strain was confirmed by the methyl green staining procedure (12). However, when cells were treated with toluene prior to cross-linking treatment, psoralen was added at the first wash step, and incubation at 0" was continued for 5 min to equilibrate the drug with the cells. Toluene-treated cells were then exposed to light, washed once and resuspended for incubation as described above.
The order of treatments did not affect eventual cross-link removal. However, the yield of cross-links in cellular DNA is increased 4 to 8-fold by a prior toluene treatment. Cross-links were induced in cellular DNA by addition of 4,5',8trimethylpsoralen (which is referred to as psoralen throughout this paper) to cell suspensions which were then exposed to 300 to 400 nm light as described previously (14). analogues acted as inhibitors blocking the reaction when ATP was included.

Cross-link
Removal in Toluene-treated Escherichia coli-E. coli were grown in medium containing 1:'HJthymidine to label the DNA uniformly, and permeabilized by treatment with toluene as described under "Materials and Methods." After treatments producing 60 to 88 cross-links/E. coli chromosome (XI), cells were incubated in different reaction mixtures as described. DNA isolated after various incubation times was denatured, and the number of interstrand crosslinks remaining was measured by the hydroxylapatite chromatography technique (2). Results in Fig. 1 show that crosslinks were removed selectively from DNA by a reaction that occurred with apparent first order kinetics. The rate constant of 6 x lo-" min ' determined here was nearly equal to that reported for the reaction in whole cells (2). Significantly, cross-link removal did not occur in permeabilized cells in the absence of ATP or Mg'+. Although ATP and Mg'+ appear to be the only small molecules required, results in Fig. 1 show that the reaction was slightly faster when the four deoxyribonucleotide triphosphates (dNTPs) were included at 36 PM. Under these conditions, 90 to 95% of the cross-links were removed.
In whole cells, the time required for total cross-link removal increased with the extent of psoralen and light exposures, suggesting that treatments producing more than 200 crosslinks/chromosome somehow saturated the cell's capacity to repair these damages (2). As shown in Fig. 3, a similar effect was noted for toluene-treated cells exposed to treatments producing between 80 and 190 cross-links/chromosome. During a given incubation time, the fraction of cross-links removed decreased with higher psoralen and light exposures. However, the initial rate constant for cross-link removal was similar to that determined for lower yields of cross-links ( Fig.  2 and 3). One interpretation of these data is that toluenetreated E. coli. and perhaps whole cells, have less than 80 active repair enzymes available for each damaged chromosome.
It has been reported that DNA ligase of E. coli can rejoin DNA strands cut by UV endonuclease acting on DNA containing pyrimidine dimers (29, 30). The ligation reaction requires nicotinamide adenine dinucleotide (NAD) as a cofactor, and is inhibited by a reaction product, nicotinamide mononucleotide (NMN) (31). To determine whether ligation affected cross-link removal, the reaction was studied in the presence of 2.2 mM NAD or 5 mM NMN. Neither of these reagents affected cross-link removal ( removed when guanosine or cytosine triphosphate were sub- (2). As shown in Table II, cross-link removal occurred norstituted for ATP (similar data were obtained for KMBL1482, mally in permeabilized cells containing recBZ1, recC22 (data data not shown). ATP may be hydrolyzed during the strand-not shown) or polA1 (data not shown) mutations. No reaction cutting reaction or it may regulate the activity of UV endo-was detected in uurA6 or uurB5 (similar to uurA6, data not nuclease, or both. To distinguish among these possibilities, shown) mutants and in polAIO7 (similar to uurDlO1, data methylene-substituted derivatives of ATP which are resistant not shown) or uurDlO1 cross-link removal was barely detectato hydrolysis were tested. As shown in Table I, the reaction ble or occurred at a greatly reduced rate. The apparent slow did not occur in the presence of either u,P-methylene ATP or removal in whole cells with a polAlO7 mutant was attributed P,y-methylene ATP at 1.35 mM. Additionally, neither of these to nonspecific DNA breakdown ( Table III, less than 1% of reversibly renatura One strand cut on the 5' side of each of two adjacent cross-links will produce fragments which can diffuse independently at denaturation only when the cuts are placed in one of four possible relative orientations. This can be visualized by considering complementary DNA strands, an upper (U) and a lower CL) oriented with the 3' end of L to the right.
Designating adjacent cross-links as 1 and 2, the possible relative orientations of strand cuts are: U(l)U(2), U(l)L(2), L(l)U(2), and L(l)L(2). If 2 is situated to the right of 1, then only U(lJL (2) will produce separating fragments. If 1 is situated to the right of 2, then U (2)LCl) is the only combination allowing separating fragments.
Thus, an average of only one out of four possible combinations of strand cuts placed at random will produce fragments which separate at denaturation. The average size of such fragments will be eight times the average single strand spacing between crosslinks. After denaturation, each fragment will have two single strand tails equal in length to the cross-link spacing, and these tails will comprise 25% of the total mass.  Table III  are results   from  reversibly  renaturing  DNA  of the uvrA6 mutant, in which no single strand regions were detected. Substantial amounts of X-sensitive DNA accumulated and persisted for up to 40 min in both the uvrDlO1 and polA107 mutants (data not shown). The 12 to 15% of the reversibly renaturing material that was digested by Sl nuclease appears to be substantially less than the 25% predicted as shown in Fig. 5. This apparent discrepancy can be accounted for by the limited activity of the UV endonuclease during cross-link removal after more intense psoralen and light treatments. In order to be valid, the enzyme analysis of the 3' end required at least 200 cross-links/chromosome, or more than 4 crosslinks/DNA fragment isolated. However, only 93 cross-links had been removed after 40 min. Data in Table III show that after 40 min, there are few strand cuts at the remaining 107 cross-links, which would produce single strand tails upon denaturation.

Removal of DNA Cross-links
Thus, strand cutting occurred to a maximum of 50% under these conditions, and a similar limiting value would be expected for polA107 and uvrD mutants. Strand cutting at only one-half of the cross-links would effectively double the size of segments released during denaturation, and would reduce the amount of material existing as single strand tails to 12.5%, similar to what was detected in these experiments.
To determine whether material sensitive to Sl nuclease was a tail attached to a double-stranded segment, and whether it terminated with a 3'-hydroxyl group, reversibly renaturing DNA was treated with snake venom phosphodiesterase. This enzyme is an exonuclease which initiates digestion specifically at 3'-hydroxyl termini of single-stranded DNA, releasing mononucleotides processively (26). Percentages of reversibly renaturing DNA susceptible to hydrolysis by snake venom phosphodiesterase or Sl nuclease are compared in Table III  for several selected strains   and incubation times or conditions. It was found that nearly all of the DNA susceptable to Sl was also hydrolyzed by snake venom phosphodiesterase. This indicates that the single-stranded DNA regions attached to rapidly renaturing DNA existed as tails terminating with 3'hydroxyl groups, as shown in Fig. 5. These molecules were not formed in uurA6 or uurB5 mutants or in wild type cells in the absence of ATP. They accumulated and persisted in the uvrDlO1 and polAlO7 strains.
DNA Polymerase I Activity of Brij Extracts -As described above, partially cut DNA with similar structure accumulated in both polAlO7 and uvrDlO1 mutants, and DNA polymerase I catalyzed the second strand-cutting reaction completing cross-link removal. One possible interpretation might be that the uvrD gene product regulates the activity or stability of DNA polymerase I. To test whether these cells contained normal levels of DNA polymerase I, extracts were prepared from Brij lysates of wild type, uurDlO1 and polA1 strains. "Activated" calf thymus DNA, the four dNTPs and :'H-labeled thymidine triphosphate were added to the extracts, and after various incubation times, incorporation of radioactivity into acid-insoluble material was determined (27,28). These data