INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Endonuclease V (Endo V) of Escherichia coli cleaves near many types of lesions in DNA. It was originally described as preferring DNA that was treated with acid, alkali, OsO₄, UV radiation, or 7-bromomethylbenz(a)anthracene, as well as uracil-containing DNA and even untreated single-stranded DNA (9, 12). The enzyme was recently found to be identical to deoxyinosine 3' endonuclease (13, 39), thereby increasing its repertoire to include dIMP residues, abasic sites, urea residues, single-base mismatches, pseudo-Y structures, and flap structures in DNA (36-39).

Endo V cleaves the second phosphodiester bond 3' to deoxyinosine in DNA (37). In E. coli, there are three other known DNA repair pathways that are initiated by cleavage of a phosphodiester bond, as opposed to cleavage of a glycosylic bond (25). The UvrABC complex excises pyrimidine dimers and nucleotides containing bulky adducts. The MutSLH system removes regions containing mismatched bases. The VSP (very short patch) repair system repairs regions containing deaminated 5-methylcytosine. The broad specificity of Endo V is rivaled only by that of UvrABC, a large protein complex. By contrast, however, Endo V is a relatively small (25-kDa) monomeric protein (37).

In this work, we produce an nfi mutant and study some of its biological properties, with an emphasis on its response to DNA-damaging agents.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Strains and their construction. The bacterial strains, plasmids, and phages used in this study are listed in Table 1. Generalized transduction was performed with phage P1 dam rev6 (29). Bacterial transformations with intact plasmids (5) and with linear DNA (26) were performed as previously described. After the introduction of the nfi-1::kan mutation into any strain, its genotype was confirmed by PCR (2) with the following two primers, which are complementary to the gene: ATGGATCTCGCGTCATTAC and CAGTTTACCTGAATTAGGG.

Nomenclature. AU^r, Val^r, and Str^r signify resistance to 6-azauracil, valine, and streptomycin, respectively.

Media. Luria-Bertani (LB) media (20) were used for routine growth. Medium E (32), supplemented with glucose (2%) and thiamine (1 µg/ml), was employed as a minimal medium. Tetracycline was used at 15 µg/ml, ampicillin was used at 100 µg/ml, chloramphenicol was used at 25 µg/ml, and amino acid supplements were used at 100 µg/ml. Strains carrying the dut-1 mutation were propagated at 25°C in LB medium supplemented with 1 mM thymidine (30). Light-sensitive hemH mutants were grown in foil-covered tubes.

Nitrite sensitivity and mutagenesis. Saturated cultures were centrifuged, and the cells were washed with the original volume of 0.1 M sodium acetate buffer, pH 4.6. One-milliliter samples (about 3 × 10⁹ cells) were pelleted by centrifugation and resuspended in 0.5 ml of either the sodium acetate buffer or a fresh solution of 40 mM NaNO₂ in the same buffer. After incubation of the cells at 37°C for 6, 12, or 18 min, 5 ml of cold medium A (20) was added. A small sample (0.2 ml) was diluted and plated on LB agar for measurement of survival, which was 95% for the buffer-treated controls. The remainder was centrifuged, resuspended in 10 ml of LB medium, and grown overnight to saturation on a shaker at 37°C to permit expression of mutations. Before being plated on any minimal selective medium, the cells were washed by centrifugation and diluted in 10 mM MgSO₄.

Mutation frequencies. Spontaneous mutation frequencies were determined by means of a modified fluctuation test (16). For each strain, 10 2-ml cultures were grown to saturation from inocula of about 100 cells. For the scoring of prototrophic revertants and Val^r mutants, the cells were washed (by centrifugation) in 10 mM MgSO₄ and grown for 2 days at 37°C. AU^r and Str^r mutants were counted after 1 day. The following selective media were used: minimal medium containing 80 µg of L-valine per ml, minimal medium containing 40 µg of azauracil per ml plus 0.2% Casamino Acids (Difco) that had been treated with Norit (30), and LB medium containing 200 µg of streptomycin sulfate per ml.

Sensitivity tests. The following were as previously described: gradient plate sensitivity tests (6), UV irradiation with a germicidal lamp (6), gamma irradiation in oxygenated cell suspensions (7), and H₂O₂ treatment (8). A calibrated ⁶⁰Co source (Nuclear Reactor Laboratory, University of Michigan) was used for the gamma radiation. For exposure to visible light, saturated cultures were diluted in LB broth (with ampicillin as needed) and 10 ml containing about 10⁴ cells was placed in a petri dish at a distance of 15 cm from two 15-W cool white fluorescent tubes (Westinghouse F15T8/CW). Exposure was at ambient temperature for 4 h. Controls were covered with aluminum foil.

Other methods. Endo V assays (13); general cloning methods (2); and growth and titration of phages M13mp18 (2, 35), X174 (28), and c60 (2) were as described previously. Competent cells for transfection assays were prepared by treatment with CaCl₂ (2). PCR products were analyzed by electrophoresis in a 2% agarose gel containing ethidium bromide (2). DNA concentrations were measured with Hoechst 33528 dye (4).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Construction of nfi-1 mutants. Plasmid pGG13 contains a chloramphenicol acetyltransferase (cat) gene cassette inserted at nucleotide (nt) 443 of the 669-nt nfi gene. The insertion (nfi-1::cat) caused a 99% reduction in plasmid-specified Endo V activity without affecting the expression of the neighboring genes (13). The mutation was transferred to the chromosome by transformation of the recD strain BW853 with plasmid DNA that had been linearized by digestion with endonucleases ApaI and SmaI. A chloramphenicol-resistant, ampicillin-sensitive transformant (BW1160) was then used as a donor to transduce the nfi-1::cat mutation into other strains (Table 1).

Construction of strains that overexpress Endo V. The high-copy-number nfi⁺ plasmid pGG42 (Table 1) also contains yjaG, a gene of unknown function downstream of nfi. The plasmid was digested with endonuclease AflIII and religated, thereby deleting yjaG by removing a segment extending from the nfi-yjaG intergenic region to a point within the vector. The remaining cloned DNA extended from 433 nt upstream of nfi to 41 nt downstream of it. The resulting plasmid, pGG48, contained nfi as the only intact cloned gene and conferred a 46-fold increase in Endo V activity. Because the standard assay overestimates the activity in plasmid-free cells, the actual increase in Endo V may be over 100-fold.

Sensitivity to visible light. nfi is 12 nt downstream from hemE, the gene for uroporphyrinogen decarboxylase (23). Both genes are missing from an otherwise homologous region in Hemophilus influenzae. This physical and evolutionary association of the genes suggested a physiological relationship. hemE is involved in the biosynthesis of photosensitizing metabolites (34), and overproducers of HemE are sensitive to visible light (13), probably through the photochemical production of singlet oxygen or superoxide (22). Therefore, we hypothesized (13) that Endo V might have evolved to repair DNA damage from active oxygen species generated photochemically by protoporphyrin precursors. To test our hypothesis, we combined the nfi mutation with a hemH mutation (34) that causes photosensitivity due to accumulation of these metabolites. The strains were exposed to visible light at a level that reduced the survival of a hemH mutant (BW1187) to 18% without affecting that of the hemH⁺ parent (KL16). The addition of an nfi mutation to each strain produced no further detectable effect on survival. Moreover, this treatment did not produce a significant increase in mutations for resistance to azauracil, valine, or streptomycin (data not shown). Therefore, Endo V does not appear to be essential for DNA repair under these conditions.

Action on single-stranded DNA in vivo. Endo V was originally described as a single-strand-specific DNase, and its standard assay was based on its ability to cleave the single-stranded DNA of a filamentous phage (12). However, in the cell, such activity might cause lethal damage to the replicating chromosome. To see if the enzyme manifests this activity in vivo, we tested the plating efficiencies of two phages containing single-stranded DNA: the filamentous phage M13 and the icosahedral phage X174. M13mp18 (35) and X174 phages were titrated on derivatives of strains KL16 and CR63.1, respectively. Similar results (±10%) were obtained for each of the following: the nfi⁺ parent, the corresponding nfi mutant, the nfi⁺ strain carrying the vector pUC19, and a strain carrying plasmid pGG48 (pUC19::nfi⁺). Because the infecting M13 DNA might carry some protective capsid protein into the cell, we measured the infectivity of the naked DNA as well. To compensate for variations in the levels of competence of the cell preparations used for these transfection assays, for each strain tested, the specific infectivity (PFU per nanogram) of the single-stranded viral DNA was divided by that obtained with the double-stranded plasmid, or replicative form, of M13. The resulting ratios were 42, 59, 39, and 39, respectively, for the four strains described above. Again, there was no marked difference between wild-type, mutant, and overproducing hosts. Therefore, we were unable to demonstrate any significant activity of the enzyme on single-stranded DNA in vivo.

Action on uracil-containing DNA in vivo. Endo V also specifically cleaves uracil-containing double-stranded DNA. Bacteriophage  that is grown on a dut-1 ung host contains uracil in place of some of its DNA thymine residues (10). The leaky (nonlethal) dut (dUTPase) mutation reduces the synthesis of TTP and permits the accumulation of dUTP, which is then incorporated into the DNA in place of some of the TTP. The ung (uracil-DNA glycosylase) mutation blocks the base-excision repair of this DNA, which would otherwise produce lethal double-strand breaks. Therefore, the uracil-containing  phage has a high plating efficiency only on an ung mutant (10). To test if Endo V is active on uracil-containing  DNA in vivo, we determined the effect of Endo V deficiency and overproduction on the plating efficiency of the  phage (Table 2). The hosts were either ung⁺ or had a tight insertion mutation in ung. Whereas an ung mutation increased the survival of the  phage 10⁴-fold, an nfi mutation had little effect (experiment 1). In experiment 2 (Table 2), we used an ung mutation to eliminate the high background of bacteriophage restriction initiated by the glycosylase. However, even under these circumstances, neither the absence nor the overproduction of Endo V affected the plating efficiencies. Therefore, we were unable to demonstrate any significant activity of Endo V on uracil-containing phage DNA in vivo.

View this table: [in this window] [in a new window]   TABLE 2.   Effect of nfi gene dose on the growth of uracil-containing  bacteriophages

Action at apyrimidinic sites in vivo. Endo V has AP endonuclease activity (37), i.e., it cleaves near abasic (apyrimidinic or apurinic) sites in DNA. Abasic sites may be generated by the action of uracil-DNA glycosylase on the uracil residues in the DNA of a dut mutant. Because a tight dut mutation is lethal (11), we chose to use the dut-1 allele, which specifies a dUTPase that appears to be temperature sensitive in vivo (30). Even at 42°C, however, there is enough residual activity of the mutant enzyme to enable a high level of survival, unless the cells also contain a mutation affecting the repair of abasic sites. One such mutation is in xth, the gene for exonuclease III, the enzyme that possesses the major AP endonuclease activity in E. coli. Consequently, a dut-1 xth double mutant has a low level of survival at 42°C, whereas a dut-1, an xth, or a dut-1 xth ung triple mutant is almost fully viable (30). These findings had indicated that although unrepaired abasic sites (in the dut xth mutant) are lethal, the same number of single-strand breaks (in the dut mutant) or persistent uracil residues (in the dut xth ung mutant) are not. Previously, it was found that an nfo (endonuclease IV) mutation increased the lethality of the dut xth combination (6), indicating that endonuclease IV also has a significant role in the repair of apyrimidinic sites. We now tested our nfi mutation to see if it would have a similar effect (Table 3). Although the nfi-1 mutation did not significantly increase the temperature sensitivity of a dut-1 or an xth mutant, it produced a fourfold decrease in the survival of the double mutant, suggesting that Endo V contributes to the repair of apyrimidinic sites in vivo.

Sensitivity to DNA-damaging agents. Endo V can recognize DNAs that have been damaged by a variety of agents. To assess the relative importance of the enzyme in some DNA repair pathways, we measured the sensitivity of nfi mutants to DNA-damaging compounds. Because individual DNases may be redundant in some pathways, we also tested the nfi mutation in combination with others. We asked, for example, if an nfi mutation enhances the known sensitivity of an xth (exonuclease III) mutant to H₂O₂ or of an nfo (endonuclease IV) mutant to mitomycin. The cells were exposed to the lethal agents either in liquid cultures (Fig. 1) or during growth on an agar plate containing a linear concentration gradient of the compound (Table 4). The nfi mutation specified an increased sensitivity to nitrofurantoin and to bleomycin (Table 4), both of which produce free radical damage to DNA. However, the nfi mutant did not appear to be significantly more sensitive to mitomycin, methyl methanesulfonate, or tert-butyl hydroperoxide under conditions in which either the nfo or xth mutant was more sensitive than the wild type (Table 4).

View larger version (29K): [in this window] [in a new window]   FIG. 1.   Survival curves. Cell suspensions were exposed to DNA-damaging agents at the following levels: H2O2, 10 mM;  rays, 1 krad/min; UV light (peak at 254 nm), 1 W/m2; and HNO2, 40 mM. Dashed lines and open symbols are used for nfi mutants, and solid lines with filled symbols denote the corresponding nfi+ parents. The following sets of congenic strains were used for the experiments whose results are shown in panels A to C: AB1157 (wild type), BW1161 (nfi), BW9109 (xth), BW1166 (xth nfi), and BW1163 (xth nfo nfi). (D) The strains were KL16 (wild type) and BW1185 (nfi). Each datum point in the graphs in panels A, C, and D represents the average of results for two experiments in which each set of strains demonstrated the same order of sensitivities. (B) Each point is the result obtained with a separate oxygenated cell suspension; therefore, each set of vertical points represents the results of an independent experiment consisting of equally treated samples of four different strains. wt, wild type.

Frequencies of spontaneous mutation. In vitro, Endo V cleaves DNA near bases that are mispaired, oxidized, or hydrolytically deaminated. All of these lesions can occur in the chromosome during normal growth in the absence of external mutagens. Spontaneous mutations were scored for several markers (Table 5). The his and arg mutations were ochre (TAA) codons that could revert by intragenic mutations only of A · T base pairs or by suppressor mutations that might occur at G · C base pairs as well. The Val^r and AU^r traits represent a broad spectrum of mutational changes. Mutations to Val^r occur by loss of function of any of five genes involved in the transport of branched-chain amino acids or by alterations in either the structure or expression of any of six others involved in the feedback-inhibitable biosynthesis of isoleucine and valine (19). AU^r results from loss of function of the upp (uracil phosphoribosyltransferase) gene (19). Relative frequencies of spontaneous mutation were analyzed by a modified fluctuation test that minimizes jackpot effects. An nfi mutant demonstrated about a twofold increase in the mutation frequencies for three of the traits tested (Table 5).

Nitrite-induced mutation. An nfi⁺ strain and an nfi mutant strain were compared with respect to their levels of mutagenicity after treatment with nitrous acid (Table 6). The nfi mutant consistently displayed a greater survival than the wild type, confirming the results shown in Fig. 1. Nitrous acid produced about two to three times as many AU^r and Val^r mutants of the nfi mutant strain as of the wild type. The results were more striking for streptomycin resistance. The nitrite-treated nfi⁺ samples showed great variability, because their low frequency of Str^r mutations and low rates of survival combined to produce occasional jackpot results. Nevertheless, the average frequency of nitrite-induced Str^r mutants was 12-fold higher in the nfi mutant strain than in the wild type. The difference was about 1,000-fold if jackpot results were offset by comparing the median (rather than the mean) values (Table 6, footnote d).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The most striking effect of an Endo V deficiency is on the ability of the cell to repair nitrous acid-induced DNA damage. Compared to wild-type strains, nfi mutants displayed a greater frequency of nitrous acid-induced mutations but, paradoxically, the nfi mutants had an increased rate of survival. Apparently, at high doses of the agent, the strand breakage caused by the enzyme might be more lethal than the unrepaired mutagenic DNA lesions. As reviewed in reference 15, nitrous acid oxidatively deaminates the bases in DNA. Adenine is deaminated to hypoxanthine (in deoxyinosine), which pairs mainly with cytosine, resulting in A · T  G · C transitions. The deamination of cytosine to uracil results in G · C  A · T transitions. Xanthine produced from guanine can pair with thymine and has been blamed for G · C  A · T transitions; however, we do not know how well the DNA replication machinery of E. coli discriminates among alternative xanthine-containing base pairs. Of the traits tested in this study, the greatest increase in mutation frequency was that seen for Str^r after nitrite exposure. All known mutations to Str^r occur at one of two AAA (lysine) codons in the rpsL gene and are predominantly A · T  G · C transitions (31). Therefore, our result is consistent with the nfi mutant being defective in the repair of deaminated deoxadenosine (i.e., deoxyinosine) in DNA. Experiments are now in progress to determine the precise base pair changes induced by nitrous acid in the nfi mutant.

Considering the specificity of Endo V, the nature of its role in dealing with nitrous acid-induced lesions is not surprising but the magnitude of its role is, given that there are other enzymes in E. coli that can deal with dIMP in DNA. There is a hypoxanthine-DNA glycosylase activity (18) that is now known to be a property of 3-methyladenine-DNA glycosylase II, the AlkA protein (27). A putative second hypoxanthine-DNA glycosylase, with a reportedly different molecular weight, was also described (14); however, it may be the same protein because an alkA mutant had no detectable hypoxanthine glycosylase activity (27). In addition, the MutSLH mismatch repair system might be able to recognize the unstable hypoxanthine · thymine base pairs as a mismatch, but only in newly replicated DNA (21). Our results indicate that whereas these other systems can deal effectively with most spontaneous deamination mutations, they are not adequate to protect the cell from mutagenesis by nitrous acid.

Endo V also helps to protect the cell against killing by bleomycin and nitrofurantoin. Both agents bind to DNA and produce free-radical-mediated DNA damage, which has been better characterized for bleomycin (24) than for nitrofurantoin (17, 40). However, we do not know what specific lesions the enzyme recognizes. Experiments with dut mutants suggested that Endo V might make a significant contribution to the repair of apyrimidinic sites, even though there are several other E. coli endonucleases and endolyases that cleave at such sites (25). Although our apyrimidinic sites were generated by transient uracil incorporation, other experiments in this study indicated that Endo V had little effect on the DNA uracil itself.

Some of our negative results were of equal importance to the positive results. In the first studies of Endo V, it was uniquely characterized as specific for single-stranded DNA and for uracil-containing DNA. However, we were unable to demonstrate either activity in vivo. Its cleavage of single-stranded DNA, although it formed the basis for a standard assay (12), is admittedly so slow relative to its activity on dIMP-containing DNA that it was not detected at first in some studies (37). It is likely that its specificity for single-stranded DNA results from its ability to recognize base mismatches (38, 39); thus, it may cleave at unpaired or mispaired bases produced by transient hydrogen bonding of nearly homologous regions. (If this is true, the enzyme should have no activity on a homopolymer, a prediction that has not yet been tested.) Opportunities for such annealing would be extensive in a large single-stranded M13 or X174 DNA molecule, upon which Endo V appeared to have no activity in vivo. Moreover, if the enzyme were to cleave single-stranded regions of the chromosome, such as those near replication forks, it would produce lethal double-strand breaks. Therefore, we presume that either single-stranded DNA is protected in the cell by proteins that bind to it or that the specificity of Endo V in vivo may be different from that in vitro, perhaps as a result, for example, of its being part of a repair complex.

Similarly, we were unable to demonstrate activity of Endo V on uracil-containing DNA. Unlike uracil-DNA glycosylase (10), Endo V requires a high level of uracil substitution in its DNA substrates (9), and we do not know to what extent this was achieved in our  phages. Therefore, we can conclude only that the activity of Endo V, even in overproducers, is not enough to affect the viability of a uracil-containing  phage under conditions in which an ung mutation does affect its viability. Studies in progress, of specific mutation rates (e.g., G · C  A · T), may provide a more sensitive determination of the relative roles of Endo V and uracil-DNA glycosylase in the repair of uracil in DNA.