Further Characterization of Escherichia coliEndonuclease V

Endonuclease V from Escherichia colihas a wide substrate spectrum. In addition to deoxyinosine-containing DNA, the enzyme cleaves DNA containing urea residues, AP sites, base mismatches, insertion/deletion mismatches, flaps, and pseudo-Y structures. The gene coding for the enzyme was identified to beorf 225 or nfi(e n donuclease fi ve). Using enzyme purified from an overproducing strain, the deoxyinosine- and mismatch-specific activities of endonuclease V was found to have different divalent metal requirements. The affinity of the enzyme is greater than 20-fold higher for DNA containing deoxyinosine than deoxynebularine or base mismatches. Under optimal cleavage conditions, endonuclease V forms two stable complexes with DNA containing deoxyinosine, but not with DNA containing base mismatches or deoxynebularine, suggesting that the 6-keto group of hypoxanthine in DNA is critical for stable interactions with the protein. The enzyme recognizes deoxyuridine in DNA but exhibits a much lower affinity to DNA containing deoxyuridine compared with DNA containing deoxyinosine. Interestingly, deoxyuridine-specific endonuclease activity of endonuclease V has a divalent metal requirement similar to the mismatch activity. A model for the mechanism of substrate recognition is proposed to explain these different activities.

Recently, we and others (1,2) have shown that deoxyinosine 3Ј-endonuclease is identical to endonuclease V, previously characterized as an enzyme with endonuclease activity toward DNA treated with acid, alkali, OsO 4 , or 7-bromomethyl-benz[a]anthracen (3,4). The gene coding for endonuclease V (nfi, endonuclease five) was found to be identical to orf 225, located at 90 minutes of the Escherichia coli genome (GenBank TM ascension no. U00006) (5). In addition to the less defined substrates, we have shown that homogenous preparations of endonuclease V purified from the wild type E. coli cells recognize several well defined DNA lesions, including deoxyinosine (6), urea residues (6), AP sites (6), base mismatches (7), loops and hairpins (2), flaps, and pseudo-Y DNA structures (2). The ability of endonuclease V to recognize mismatches and abnormal replicative DNA structures suggests that the enzyme might play an important role in DNA metabolism (2).
Unlike DNA N-glycosylases, endonuclease V cleaves the DNA strand containing lesions at the second phosphodiester bond 3Ј to the lesion, leaving a nick with 3Ј-hydroxyl and 5Ј-phosphoryl groups. However, the lesion is not removed from the DNA by the enzyme. Endonuclease V forms stable complexes with DNA containing deoxyinosine both before and after cleavage, showing similar affinity to both the substrate and the product (8). Based on these results, we have proposed earlier (8) that, besides its endonuclease activity, the enzyme might function to target other repair protein(s) to the lesion. Therefore, it is possible that endonuclease V could initiate a novel repair pathway. Endonuclease V cleaves DNA containing base mismatches in a strand-specific manner; it cleaves the DNA strand containing mismatch closer to the 5Ј terminus (7). Furthermore, the mismatch-specific activity of the enzyme is reduced when the mismatch is flanked by GC pairs; however, its deoxyinosine-specific activity is not influenced by the sequence context. These results suggest that endonuclease V might have different modes of interaction between DNA containing deoxyinosine and mismatches. It is, therefore, important to understand the mechanism governing the recognition of various DNA lesions by this small protein (24,900 Da). To understand the possible mechanism that is involved in the substrate recognition by endonuclease V, we cloned the gene coding for the enzyme (2) so that a large quantity of the protein can be obtained rapidly for further biochemical and physical characterization.

MATERIALS AND METHODS
All reagents were of analytical grade of purity. Restriction enzymes were purchased from New England Biolabs. General molecular biology procedures used were adapted from the protocols described in Sambrook et al. (9).
Cloning, Overexpression, and Purification of Endonuclease V-Homogeneous endonuclease V was purified from 400 g of E. coli paste according to the procedures described previously (6). Fraction V, which was judged to be homogeneous by silver staining, was blotted onto a polyvinylidene difluoride membrane and sent to the University of Texas Medical Branch at Galveston for N-terminal amino acid sequencing. The first 24 amino acid residues obtained for the N terminus of endonuclease V were determined to be MDLASLRAQQIELASSVIREDRLD. When sequence homology for this amino acid sequence was searched through the GenBank TM using the Blast program (10), the N-terminal amino acid sequence was found to be almost identical to the N-terminal sequence of a hypothetical protein (24.9 kDa) translated from orf 225 (GenBank TM access no. U00006) (5). The only difference was that the hypothetical protein had two extra amino acid residues at the N terminus, that is, the translated N-terminal amino acid sequence derived from orf 225 was MIMDLASLRAQQIELASSVIREDRLD.
To confirm that orf 225 is the gene coding for endonuclease V, orf 225 was amplified from E. coli genomic DNA by the polymerase chain reaction using flanking primers derived from the genomic DNA sequences obtained from the GenBank TM . The 5Ј primers and the 3Ј primer contained a NdeI restriction site (underlined) and an AvaI restriction site (underlined), respectively. Since the genomic sequence of orf 225 obtained from the GenBank TM (access no. U00006) (5) revealed two potential start codons for orf 225, two different 5Ј primers were used: 5Ј primer 1, CTGGGAATTCCATATGATTATGGATCTCGC-* This work is supported by National Instiutes of Health Grant GM 37216 (to Y. W. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
GTCA, 5Ј primer 2, CTGGGAATTCCATATGGATCTCGCGTCATTAC-GC. The sequence for the 3Ј primer was CGACGATCCCTCGAGCTG-AATTAGGGCTGATTTGC. Polymerase chain reactions were carried out using Taq DNA polymerase (Boehringer Mannhein). The polymerase chain reaction fragment amplified with 5Ј primer 1 and the 3Ј primer was cloned into the NdeI and the AvaI sites of pET22b(ϩ) vector (Novagen) to generate the plasmid pETI-1. The polymerase chain reaction fragment amplified with 5Ј primer 2 and the 3Ј primer was cloned into pET22b(ϩ), yielding the plasmid pETI-2. Both the pETI-1 and the pETI-2 plasmids were transformed into E. coli strain BL21(DE3) by electroporation. When BL21(DE3) carrying plasmid pETI-1 or pETI-2 was grown and induced with isopropyl-1-thio-␤-D-galactopyranoside, the overexpression of a 25-kDa protein was observed to exhibit activities identical to endonuclease V (data not shown).
Based on the above observations, endonuclease V was then purified from an overexpressing host BL21(DE3) carrying pETI-2. The protein produced by the gene cloned in pETI-2 has the same N-terminal sequences as determined for endonuclease V. Four liters of BL21(DE3) carrying pETI-2 were grown in Luria broth containing 50 g/ml ampicillin to an absorbance at 600 nm of 0.4. Isopropyl-1-thio-␤-D-galactopyranoside was then added to the culture at a final concentration of 1 mM, followed by further incubation at 37°C for 3 h. Bacteria were harvested by centrifugation and endonuclease V was purified according to earlier published protocol (2,11).
Endonuclease Assay-The standard reaction mixture for assaying deoxyinosine-specific activity (10 l) contained 10 mM Tris-HCl, pH 7.5, 2 mM MgCl 2 , 20 fmol of end-labeled oligodeoxynucleotide, and an appropriate amount of enzyme. The reaction mixture was incubated at 37°C for 10 min and stopped by the addition of 10 l of loading buffer (90% formamide, 1 mM EDTA, 0.1% xylene cyanol, and 0.1% bromphenol blue). Three to five l of the reaction mixture were loaded onto a 12.5% denaturing polyacrylamide gel (National Diagnostics) and electrophoresed at 2000 V for 1.5 h. The polyacrylamide gel was then dried under vacuum and exposed to x-ray film or analyzed by Fuji Bio-Imaging analyzer.
The standard reaction mixture for assaying mismatch-specific activity was the same as that for deoxyinosine-specific activity assay, except 0.34 mM MnCl 2 was used instead of 2 mM MgCl 2 .
The polyacrylamide gel (10%) contained acrylamide/N,NЈ-methylenebisacrylamide at a 19.76/0.24 ratio in TBE buffer (89 mM Tris, 89 mM boric acid, pH 8.3, and 2.5 mM EDTA). The gels were preelectropheresed at 300 V (4°C) for 30 min, samples were applied, and electrophoresis was continued at 300 V (4°C) for 150 min. Following electrophoresis, the gels were dried under vacuum and exposed to x-ray film. The radioactive bands in the dried gels were quantified with a Fuji Bio-Imaging analyzer.

RESULTS
Optimal conditions for the endonuclease activities of endonuclease V-We have shown previously (6) that MnCl 2 can replace MgCl 2 and support the endonuclease activities of endonuclease V. An optimal MnCl 2 concentration (0.34 mM) supports about 60 -70% of deoxyinosine-specific endonuclease activity of endonuclease V at optimal MgCl 2 (6). In contrast, the mismatchspecific endonuclease activity of the enzyme is more than 20fold higher in the presence of Mn 2ϩ (0.34 mM) than Mg 2ϩ (2 mM) (data not shown), showing sharp differences in the divalent ion requirements for the deoxyinosine-specific activity and mismatch-specific activity.
Endonuclease V was shown to have an optimal pH above 9.0 by earlier investigators using partially purified enzyme preparations (3,4). However, highly purified endonuclease V prepared from an overproducing E. coli strain did not agree with these findings. Consistent with our results published earlier (6), endonuclease V cleaves deoxyinosine-containing DNA at a pH range from 6.0 to 9.5, with an optimal pH between 7.0 to 7.5. Less than 30% of deoxyinosine-specific activity remained when the pH was higher than 9.0. Similarly, the enzyme has a sharp pH optimum of 8.0 for the mismatch-specific endonuclease activity. Only 50% of mismatch-specific activity remained at either pH 7.5 or 8.5, and less than 10% of the optimal activity when the pH of the reaction was higher than 9.0.
Binding of Endonuclease V to DNA Containing Deoxyinosine and Base Mismatches-We have showed previously (8) that endonuclease V purified from the wild type E. coli strain forms two complexes with DNA containing deoxyinosine with observed K d1 and K d2 for complexes I and II of 4 and 400 nM, respectively. In agreement with the earlier results, endonuclease V purified from the overproducing strain formed two stable complexes with oligodeoxynucleotide containing deoxyinosine (Fig. 1). The K d1 and K d2 (2 nM and 60 nM, respectively, data not shown) were lower than those observed previously using endonuclease V purified from wild type E. coli cells (8). This is probably due to a higher amount of active fraction in the enzyme preparation prepared from the overproducing strain.
Under optimal cleavage conditions (i.e. in the presence of 2 mM MgCl 2 , 37°C) most of the dI-containing oligonucleotides remains bound with endonuclease V as protein-DNA complexes, with less than 10% of total radioactive material migrated as cleaved oligonucleotide (8) (Fig. 1). In contrast, DNAprotein complexes formed between endonuclease V and mismatch-containing DNA were relatively unstable, either in the presence of Mg 2ϩ or Mn 2ϩ . In the presence of Mn 2ϩ , the majority of the radioactive material migrated as cleaved oligonucleotides in an EMSA when duplex A/A was used as DNA substrate (Fig. 1). Fig. 2 showed the quantitative results of an EMSA for endonuclease V with duplex A/A (with a 5Ј-endlabeled top strand) in the presence of 0.34 mM MnCl 2 . It was interesting to note that there were two new electrophoretic migration species observed in the EMSA with duplex A/A as compared with that with duplex I/A (see Fig. 1). One appeared at high enzyme concentrations, migrated slower than the corresponding complex II. This higher complex was probably due to binding of additional protein molecule(s) to complex II and was quantified as complex II in Fig. 2B. Another species migrated slightly faster than did complex I. This might be complex I that assumed a slightly different conformation and was thus quantified as complex I. Fig. 2B shows that, under optimal cleavage conditions and at all enzyme concentrations examined, less than 10% of the radioactivity was found to be associated with complex I. When the enzyme concentration was increased to above 6.5 nM, the formation of complex II was observed. Increasing enzyme concentrations led to a higher percentage of complex II (as much as 40% of total radioactivity); however, the amount of complex I remained relatively unchanged. At all enzyme concentrations, the total amount of complexes formed was less than the amount of cleaved oligonucleotide. This is in sharp contrast to the interaction of endonuclease V with deoxyinosine-containing oligonucleotide in which most of oligonucleotides were found to be associated with the enzyme as complex I and complex II (Fig. 1) (8). In the absence of a divalent cation and in the presence of 100 mM NaCl, without substrate cleavage, both complexes I and II were formed with an apparent K d of 58 and 62 nM, respectively (data not shown) (Fig. 3). Endonuclease V exhibits similar binding pattern with duplexes G/A and N/A, both in the presence and absence of Mn 2ϩ (data not shown).
Endonucleolytic Activities of Endonuclease V on DNA Containing Deoxyuridine-In our earlier studies, we found that endonuclease V could not cleave deoxyuridine-containing oligodeoxynucleotides (6). Since partially purified endonuclease V preparations from earlier investigators were shown to recognize deoxyuridine-containing DNA (1, 3, 4), we decided to reexamine whether a highly purified endonuclease V preparation could recognize DNA containing deoxyuridine in various sequence contexts. Oligonucleotides U24 and U21 were annealed to single-stranded SK(Ϫ) DNA to form the following duplexes which are identical to duplexes F and H used in earlier studies (6):

DUPLEX H
The endonucleolytic activity of endonuclease V on these two DNA duplexes were compared with that on duplex I/SK which was formed by annealing oligo I to single-stranded SK(Ϫ) T/A (see Equation 4).   4 shows that in the presence of 2 mM MgCl 2 , the endonucleolytic activity of endonuclease V on deoxyuridine-containing DNA was much lower compared with that on deoxyinosinecontaining DNA. The enzyme cleaved deoxyuridine-containing DNA only at a high enzyme concentration, and the activity did not appear to increase proportionally with an increase in enzyme concentration. Since duplex F and duplex H contained U/T and U/C mispairs, respectively, cleavage activity on these duplexes could be due to the mismatch-specific activity of endonuclease V. We have shown earlier that the mismatch-specific activity of endonuclease V was much higher in the presence of Mn 2ϩ than Mg 2ϩ , and we thus examined whether endonuclease V had increased endonucleolytic activity on duplexes F and H when reactions were performed in the presence of Mn 2ϩ . In the presence of Mn 2ϩ , endonuclease V had a much higher activity on deoxyuridine-containing DNA, about 20-fold higher than in the presence of Mg 2ϩ at most enzyme concentrations tested (Fig. 5). Interestingly, in the presence of Mn 2ϩ , the endonucleolytic activity of endonuclease V on duplexes F and H increased proportionally to the increase in the amount of enzyme. In contrast to the earlier observations (7,8), the deoxyuridine-specific endonuclease activity of endonuclease V was found to have an optimal pH between 8.0 and 8.5, which is similar to the optimal pH observed for mismatch-and deoxyinosine-specific activity (data not shown). However, the deoxyuridine endonuclease activity of endonuclease V was less sensitive to higher pH values; 60 -70% of the deoxyuridine endonuclease activity remained when the pH of the reaction was increase to 9.5. At this pH, both the mismatch-and deoxyinosine-specific activity of the enzyme was inhibited to less than 20% of the optimum activity (data not shown).
Endonuclease V Exhibits Specific Recognition of Deoxyuridine in DNA-Since duplex F and duplex H contain a U/T and U/C mispair, respectively, it is uncertain whether the endonucleolytic activity observed on these duplexes is due to the specific recognition of dU or to the base mismatch. To determine whether endonuclease V has a specific recognition for deoxyuridine, we prepared the following oligonucleotide duplexes: FIG. 4. Comparison of endonucleolytic activity of endonuclease V on DNA containing deoxyinosine and deoxyuridine. DNA substrates were prepared as described under "Materials and Methods" and elsewhere in the text. Oligonucleotide containing deoxyinosine or deoxyuridine was 5Ј-end-labeled. Twenty femtomoles of DNA substrates were incubated with the indicated amount of endonuclease V (dI Endo) in a 10-l reaction mixture containing 10 mM Tris-HCl, pH 7.5, and 2 mM MgCl 2 at 37°C for 10 min. The reactions were then stopped by adding 10 l of loading buffer, and electrophoresis on a 12.5% denaturing polyacrylamide gel electrophoresis was done. The protein concentrations used for the cleavage reactions ranged from 0.65 to 130 nM. In these duplexes, X stands for either dA, dT, dC, or dG. Thus, Duplex U/X can contain either U/A pair or U/T, U/C, or U/G mispair and duplex X/U can contain either an A/U pair or a T/U, C/U, or G/U mispair. Fig. 6 shows that endonuclease V cleaved oligo U in these duplexes efficiently in the presence of Mn 2ϩ , no matter whether dU was paired with dA or mispaired with other bases. We have shown earlier (7) that if X/U or U/X (X being any of the four bases) in these duplexes were replaced by mismatched pairs, endonuclease V cleaved only the top strand whose 5Ј terminus is closer to the mismatches. However, if U/X or X/U was replaced with I/X or X/I, the enzyme cleaves the strand containing deoxyinosine whether it is on the top strand or the bottom strand. Since endonuclease V cleaved the deoxyuridine-containing strand in both duplex U/X and duplex X/U (Fig. 6, A  and B), the enzyme specifically recognizes deoxyuridine in a manner similar to the recognition of deoxyinosine. In addition, endonuclease V possesses a very weak activity on singlestranded DNA containing deoxyuridine (data not shown), further indicating that the enzyme recognizes deoxyuridine specifically in the DNA. However, it is interesting to note that the deoxyuridine-specific activity of the enzyme had a similar divalent ion requirement as the mismatch-specific activity. Even on Duplex U/A, which contains a U/A pair, the endonucleolytic activity of the enzyme was 5-fold more active in the presence of Mn 2ϩ than in the presence of Mg 2ϩ . Furthermore, in the presence of Mn 2ϩ , although endonuclease V binds more tightly to duplex U/A than to duplex A/A, more than 20% of radioactive material migrated as cleaved oligonucleotides in an EMSA when duplex U/A was used as DNA substrate (data not shown). Thus, the deoxyuridine endonuclease activity of endonuclease V exhibit features that is distinct from the deoxyinosine-and mismatch-specific endonuclease activity. DISCUSSION To further understand the mechanism involved in the substrate recognition by endonuclease V, the gene coding for the enzyme was cloned into an overproducing pET plasmid and large quantities were purified. In agreement with earlier stud-ies (8), purified endonuclease V formed two stable complexes with DNA containing deoxyinosine, both under optimal cleavage and noncleavage conditions. The estimated affinity of endonuclease V for dI-containing oligonucleotides in complexes I and II were 2 and 60 nM, respectively. In contrast, the enzyme exhibits a much weaker interaction with DNA containing base mismatches. Under noncleavage conditions, i.e. in the presence of NaCl, the K d1 and K d2 determined for complexes I and II with mismatched DNA was 58 and 62 nM, respectively. Under optimal cleavage conditions (i.e. in the presence of Mn 2ϩ ), most of the radioactive material was found to be associated with the cleaved products that were not complexed with endonuclease V. These data indicate that endonuclease V has a poor affinity to nicked mismatch-containing DNA, suggesting that the enzyme dissociates from mismatched DNA after cleavage. This is in sharp contrast to the interaction of the enzyme with deoxyinosine-containing DNA in which the enzyme remains bound to DNA containing deoxyinosine even after cleavage (Fig. 1) (8). The much larger K d1 of the enzyme for DNA containing base mismatches further demonstrates a weaker interaction of endonuclease V with DNA containing mismatches. Interestingly, the K d2 of the enzyme for DNA containing base matches is similar to that for DNA containing deoxyinosine. This is in agreement with our earlier studies (8) which suggest that complex II is formed through protein-protein interaction by binding of a second molecule of endonuclease V to the first molecule of endonuclease V in complex I. Thus, K d2 reflects the affinity of the second molecule of endonuclease V to complex I. It is interesting to note that, under optimal cleavage conditions, although less than 10% of complex I was formed between the enzyme and mismatch-containing DNA, close to 40% of complex II was formed at a high enzyme concentration (130 nM). These data suggest that the second molecule of the enzyme stabilizes the interaction between the enzyme and the nicked mismatch-containing DNA, probably by bridging together the DNAs 5Ј and 3Ј to the nick. The result also indicates that the formation of complex I is a critical step in the interaction of the enzyme with DNA containing mismatch.
The large difference in stability between the complex I derived from the interaction of endonuclease V with deoxyinosine-and deoxynebularine-containing DNA suggests that the 6-keto group of deoxyinosine is critical for the binding of the enzyme to the DNA. The role of the 6-keto group has been suggested earlier to be important for the specific recognition of deoxyinosine both in single-and double-stranded DNA (7). Consistent with this hypothesis, we showed here that the stability of complex I formed with different substrates were as follows, deoxyinosine Ͼ Ͼ deoxyuridine Ͼ deoxynebularine, single base mismatches. It is possible that this 6-keto group interacts specifically with positively charged group(s) of the protein through an ionic interaction. However, endonuclease V did not form stable complexes with DNA containing a guanine which possesses a 6-keto and a 2-amino group. It is possible that the 2-amino group of guanine extends outside the "binding pocket" of endonuclease V, thus providing steric hindrance to the interaction of endonuclease V with the 6-keto group. Furthermore, the ability of endonuclease V to recognize deoxyuridine in DNA, as demonstrated by the ability of endonuclease V to nick both single-and double-stranded DNA containing deoxyuridine, suggests that the 4-keto group of uracil can interact favorably with endonuclease V and deoxyuridine can also enter the binding pocket of the protein. Similar to guanine, the 5methyl group of thymine might also provide steric hindrance to the unique interaction of endonuclease V with dU. The inability of endonuclease V to bind to thymine is thus reminiscent of the failure of uracil DNA N-glycosylase to recognize thymine in DNA (11,12).
The interaction of endonuclease V with DNA can be best illustrated in Scheme 1 in which deoxyuridine is shown overlaying with deoxyinosine, deoxycytidine with deoxyadenosine, and thymidine with deoxyguanosine. Endonuclease V interacts specifically with the 6-keto group of hypoxanthine and 4-keto group of uracil, probably through an ionic interaction (Scheme 1A). The complexes formed by the enzyme with duplex U/A is much less stable than those formed with duplex I/A. The weaker interaction of the enzyme with DNA containing deoxyuridine can be explained by the fact that the keto group of uracil (4-keto) is located in a position toward the imidazole ring of hypoxanthine, perhaps further away from the group(s) in the protein that give rise to the strong interaction with the 6-keto group of hypoxanthine. When the keto groups are replaced by amino groups (Scheme 1B), no specific interaction of endonuclease V with the DNA was observed, demonstrating the importance of the keto groups for the specific interaction. For thymine and guanine (Scheme 1C), although both bases have the required keto group for specific interaction with endonuclease V, the presence of a 5-methyl (thymine) or a 2-amino group (guanine) seems to abolish this specific interaction, presumably due to steric hindrance. However, endonuclease V can still interact with these bases in a strand specific manner when they are present in a base mismatches, probably due to changes in the secondary structures of DNA. Based on this model, it is thus expected that mutation of some specific amino acid residues might affect only the deoxyinosine-specific endonuclease activity but not the mismatch-specific activities of endonuclease V.