Characterization of an Endonuclease IV 3′-5′ Exonuclease Activity*

Previous characterization of Escherichia coli endonuclease IV has shown that the enzyme specifically cleaves the DNA backbone at apurinic/apyrimidinic sites and removes 3′ DNA blocking groups. By contrast, and unlike the major apurinic/apyrimidinic endonuclease exonuclease III, negligible exonuclease activity has been associated with endonuclease IV. Here we report that endonuclease IV does possess an intrinsic 3′-5′ exonuclease activity. The activity was detected in purified preparations of the endonuclease IV protein from E. coli and from the distantly related thermophile Thermotoga maritima; it co-eluted with both enzymes under different chromatographic conditions. Induction of either endonuclease IV in an E. coli overexpression system resulted in induction of the exonuclease activity, and the E. coli exonuclease activity had similar heat stability to the endonuclease IV AP endonuclease activity. Characterization of the exonuclease activity showed that its progression on substrate is sensitive to ionic strength, metal ions, EDTA, and reducing conditions. Substrates with 3′ recessed ends were preferred substrates for the 3′-5′ exonuclease activity. Comparison of the relative apurinic/apyrimidinic endonuclease and exonuclease activity of endonuclease IV shows that the relative exonuclease activity is high and is likely to be significant in vivo.

Apurinic and apyrimidinic (AP) 1 sites threaten genetic stability because they block replication and are mutagenic (1,2). They arise in DNA through the spontaneous loss of normal or damaged bases or through the release of modified or mismatched bases from DNA by DNA glycosylases (3,4). The first general step of base excision repair following base loss is the recognition and cleavage of DNA AP sites by an AP endonuclease. This is conserved from bacteria to humans.
There are two characterized conserved AP endonuclease families. These enzymes cleave the DNA backbone immediately 5Ј of an AP site, generating a 5Ј-deoxyribose phosphate group and a 3Ј deoxyribose-hydroxyl group that primes DNA repair synthesis (4). Removal of the 5Ј-deoxyribose phosphate group by a 5Ј-deoxyribose phosphodiesterase is considered to create a single-nucleotide gap, and repair is completed by DNA polymerase-mediated DNA resynthesis and rejoining via DNA ligase (5,6). The first enzyme family is typified by exonuclease (Exo) III from Escherichia coli (7,8) and the homologous APE-1 enzyme in humans (9), which are major AP endonucleases in these organisms. The second conserved AP endonuclease family is typified by E. coli endonuclease (Endo) IV (10,11) and includes the APN-1 protein from Saccharomyces cerevisiae (12) and Schizosaccharomyces pombe (13) and the CeAPN1 gene from the nematode Caenorhabditis elegans (14). In E. coli, Endo IV expression is induced by superoxide anion generators (15), but in S. cerevisiae, APN-1 is the predominant constitutive AP endonuclease.
Genetic studies indicate that Exo III and Endo IV have overlapping but distinctive repair specificities in vivo. In E. coli Exo III is encoded by the xth gene (16) and is a constitutive enzyme accounting for 80 -90% of the total AP activity in the cell. Endo IV is encoded by the nfo gene (17) and accounts for 5-10% of the total cellular AP activity (8). Exo III is a divalent metal ion-dependent enzyme and is inactivated by metal chelating agents (18). In contrast, the AP activity of E. coli Endo IV is resistant to inactivation by EDTA in normal assay conditions (11). Exo III and Endo IV also have a 3Јphosphatase and a 3Ј-repair phosphodiesterase in common. These activities are responsible for removing a multitude of blocking groups, including 3Ј-phosphoglycolate and 3Ј-phosphate, that are present at single-stranded breaks in DNA induced by oxidative agents (8,18). Endo IV is the only known enzyme that is active against damaged nucleotides with bases in the ␣ configuration (19). Exo III has a 3Ј-5Ј exonuclease activity, the functional significance of which is unknown (20). Even though E. coli Endo IV is a minor AP endonuclease, its expression can be induced more than 20fold by superoxide-generating agents, such as paraquat (15), which thus enhances the capability of the cells for repairing oxidative DNA damage or damage that is refractory to enzymatic processing by Exo III. nfo-like endonucleases have also been reported to nick DNA on the 5Ј side of various oxidatively damaged bases (21).
The Endo IV active site contains a trinuclear zinc center that is ligated by conserved protein side chains that cluster at the center of a deep, crescent shaped groove (22). Biochemical experiments have suggested a role for manganese (9), although the crystal structure of the Endo IV complexed with DNA indicates that manganese is not needed for activity (22). Two of the zinc atoms are partially buried in the enzyme, whereas the third atom is relatively accessible. The high resolution structures suggest that the geometry of the Endo IV trinuclear zinc cluster is exquisitely tuned for cleaving phosphodiester bonds, with all three zinc ions participating in catalysis (8,22). To date, no significant exonuclease activity has been associated with Endo IV (10,11). Here we report detection and characterization of a significant Endo IV 3Ј-5Ј exonuclease activity.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP (3000 Ci/mmol) was obtained from Amersham Biosciences. E. coli Endo IV was obtained from Epicentre Technologies and Fermentas and generously supplied by Bruce Demple (Harvard University, Boston, MA).
Overexpression of the E. coli and Thermotoga maritima Endo IV Genes-The E. coli Endo IV gene (Ec-Endo IV) was amplified by PCR from E. coli K12 genomic DNA using the following forward and reverse primers: 5Ј-ATGAAATACATTGGAGCGCA-3Ј and 5Ј-GGCTACCCGC-TTTTTCAGT-3Ј, designed from the published sequence accession number M22591 (16). The T. maritima Endo IV gene (Tm-Endo IV) was amplified from T. maritima genomic DNA using the primers 5Ј-ATGA-TAAAAATAGGAGCTCACA-3Ј and 5Ј-ATCGACCTCTATACCGAAT-T-3Ј designed from sequence data obtained from the Institute for Genomic Research (www.tigr.org). The reverse primers used for the amplification of both genes were designed to exclude the native stop codon. Amplification of the genes was performed by PCR (95°C for 1 min, 50°C for 1 min, and 72°C for 1 min for 35 cycles followed by 72°C for 10 min) in a reaction volume of 100 l containing 300 nM of each primer, 200 M of each deoxynucleoside triphosphate, 500 ng of T. maritima or E. coli genomic DNA, and 2.6 units of expand high fidelity DNA polymerase (Roche Molecular Biochemicals) in the supplied buffer. The amplified products of the anticipated sizes, 855 bp for the Ec-Endo IV gene and 861 bp for the Tm-Endo IV gene, were generated and cloned in frame with a C-terminal His tag in the expression vector pBAD-TOPO and transformed into E. coli TOP10 cells according to the manufacturer's instructions (Invitrogen).
Purification of E. coli and T. maritima Endonuclease IV-E. coli TOP10 cells harboring the pBAD-TOPO/Ec-Endo IV or Tm-Endo IV constructs were grown overnight at 37°C in 10 ml of LB containing 50 g/ml ampicillin (LB-amp). This overnight culture was used to inoculate a 1-liter LB-amp culture, which was grown to an A 600 of 0.5, at which point arabinose was added to a final concentration of 0.2% w/v. The cells were allowed to grow for a further 6 h at 37°C and were then collected by centrifugation at 3000 ϫ g for 20 min at 4°C. Following this they were resuspended in 20 ml of buffer A (50 mM phosphate buffer, pH 7.5, 500 mM NaCl, 20 mM imidazole) and lysed by sonication (four 10-s bursts at a medium setting). The crude extract was clarified by centrifugation a 12,000 ϫ g for 30 min at 4°C. Batch application of the supernatant to 1 ml of Pro-Bond resin (Invitrogen), pre-equilibrated with buffer A was performed at 4°C for 1 h with gentle agitation. Unbound proteins were collected in the flow through, which was followed by two subsequent 10-ml washes with buffer A and then buffer B (50 mM phosphate buffer, pH 7.5, 50 mM NaCl, 35 mM imidazole). The Ec-Endo IV and Tm-Endo IV His-tagged proteins were collected as 0.5-ml fractions following application of buffer C (50 mM phosphate buffer, pH 7.5, 50 mM NaCl, 100 mM imidazole). Fast protein liquid chromatography was then used to purify the proteins further. Fractions containing the 35.5-kDa Ec-Endo IV fusion protein fractions were applied to a MonoQ (DIONEX), whereas the 36.9-kDa Tm-Endo IV fusion protein fractions were applied to a MonoS HR 5/5 column (Pharmacia Corp.). The proteins were eluted with a 20-ml linear gradient from buffer D (10 mM Hepes, pH 7.2) to buffer D containing 1 M NaCl at a flow rate of 1 ml/min. Fractions of 0.5 ml were collected, and samples were analyzed on a 15% SDS-polyacrylamide gel electrophoresis and visualized by staining with Coomassie Brilliant Blue. The Ec-Endo IV fusion protein was eluted with a salt concentration of 337-375 mM, whereas the Tm-Endo IV fusion protein was eluted with a salt concentration of 600 -625 mM. For comparison of induced and noninduced extracts, E. coli TOPO10 cells harboring the empty inducible pBADTOPO expression vector or the vector bearing the E. coli or T. maritima Endo IV genes were inoculated into 10 ml of LB medium containing 100 g/ml ampicillin and were grown overnight at 37°C. 1 ml of the overnight cultures were used to inoculate 100 ml of LB-amp medium and were grown at 37°C to an optical density of 0.5 at 600 nm. Half of each culture was induced with 0.2% arabinose, and all cultures were grown for an additional 6 h. Crude extracts were then prepared and heattreated at 65°C for 7 min as described previously (23). Following 30 min on ice, the heat-treated extracts were clarified by centrifugation at 12,000 ϫ g for 30 min at 4°C.
DNA Substrates-Oligonucleotides were purchased from Sigma Genosys apart from the oligonucleotide containing a central synthetic (tetrahydrofuranyl) AP site, which was purchased from Integrated DNA Technologies (Coralville, IA). 5Ј end labeling was performed in a 50-l reaction at 37°C for 30 min using 20 units of T4 polynucleotide kinase (New England Biolabs) and [␥-32 P]ATP (3000Ci/mmol) in the kinase buffer supplied by the manufacturer. Nucleotides were then removed from the reaction mix using a mini-quick spin oligo column (Roche Molecular Biochemicals). The end-labeled oligonucleotide and its complement were mixed in equimolar quantities and annealed by heating to 95°C for 5 min followed by cooling to room temperature. The following substrates were used in enzymatic assays. SubSS-19AP was the single-stranded oligonucleotide 5Ј-GGCGAACGAGACGAGGG(C/-A)(P/G)CTGGAAAGG-3Ј bearing the synthetic AP site, tetrahydrofuranyl. The double-stranded form of the substrate, subDS-19AP, was generated by annealing with its complement 3Ј-CCGCTTGCTCTGCTCC-CGTCGACCTTTCC-5Ј. SubSS-18R was the single-stranded oligonucleotide 5Ј-GGCGAACGAGACGAGGGC-3Ј. The double-stranded form of the substrate, subDS-18R, was generated by annealing subSS-18R with the oligonucleotide 3Ј-CCGCTTGCTCTGCTCCCGTCGACCTTTCC-5Ј. Substrate subDS-20R was generated by annealing the oligonucleotides 5Ј-ATGTAACATCTTGCAGTCGG-3Ј and 3Ј-TACATTGTAGAACGTCA-GCCTATTCACGAA-5Ј. Substrate subDS-20O was generated by annealing the oligonucleotides 5Ј-ATGTAACATCTTGCAGTCGG-3Ј and 3ЈTACATTGTAGAACGTCAG-5Ј and substrate subDS-30B was generated by annealing 5Ј-ATGTAACATCTTGCAGTCGGATAAGTGCTT-3Ј and 3Ј-TACATTGTAGAACGTCAGCCTATTCACGAA-5Ј.
Enzymatic Assays-The assays were performed using 500 fmol of the 5Ј end-labeled substrate in a total volume of 25 l. All of the reactions were incubated at either 37°C (Ec-Endo IV) or 60°C (Tm-Endo IV) for 15 min. Prior to optimization, all of the assays were carried out in 10 mM Tris/HCl, pH 8.3, 10 mM KCl. For determination of optimal pH, the assays were performed in buffers (25 mM) of varying pH: sodium acetate/acetic acid, pH 5, MES/NaOH with pH 5.5, 6.0, or 6.5, Tris/HCl with pH 7.0, 7.5, 8.0, or 8.5, and glycine/NaOH with pH 9.0, 9.5, 10.0, or 10.5. Following pH optimization, all of the subsequent reactions were in 25 mM glycine/NaOH, pH 9.0. The reactions were stopped by transferring tubes to ice and adding an equal volume of loading solution (98% formamide, 10 mM EDTA, 0.5% xylene cyanole, and 0.5% bromphenol blue). The reaction products were analyzed by denaturing gel electrophoresis (20% polyacrylamide, 7 M urea) gels and quantified by phosphorimaging analysis using ImageQuant Software (Molecular Dynamics, Inc.).

Overexpression and Purification of E. coli and T. maritima
Endonuclease IV-The E. coli and T. maritima Endo IV genes were amplified from genomic DNA using primers designed from the published sequence and cloned into the pBAD-TOPO expression vector under the control of the E. coli arabinoseinducible pBAD promoter. The primers were designed so that the Endo IV open reading frames were cloned into the vectors in frame with a N-terminal leader sequence, encoding a 1.5-kDa polypeptide, and a C-terminal sequence, encoding a 3-kDa polypeptide, which includes the V5 epitope tag and a polyhistidine region. Following growth of E. coli strains harboring the pBAD-TOPO-Ec-Endo IV and pBAD-TOPO-Tm-Endo IV and induction of the fusion genes by arabinose, crude extracts were prepared, and the 35.5-kDa Ec-Endo IV and 36.9-kDa Tm-Endo IV fusion proteins were purified by immobilized metal affinity chromatography. Pooled samples bearing the eluted proteins of the expected sizes were further purified using ion exchange chromatography, anion exchange (MonoQ) for the E. coli protein, and cation exchange MonoS for the (T. maritima) protein.
Analyses of the purified fractions by SDS-PAGE and Coomassie Blue staining showed that the Ec-Endo IV and the Tm-Endo IV fusion proteins were Ͼ95% pure (Fig. 1A).
Evaluation of the Exonuclease Activity-Exonuclease activity in the MonoQ and MonoS Ec-Endo IV and Tm-Endo IV fractions was initially evaluated by incubation of the enzymes with a double-stranded 5Ј end-labeled 3Ј recessed 18-nt substrate (subDS-18R) at 37 and 60°C, respectively. The buffer used in this initial assay was different from that used previously in reports investigating Endo IV activity (11). In particular, EDTA and DTT were absent, and the salt concentration was low. Analysis of samples incubated with either Endo IV revealed multiple bands smaller in size than the labeled 18-nt fragment and indicating the presence of exonuclease activity. The band sizes decreased approximately one base at a time, consistent with the presence of a 3Ј-5Ј exonuclease activity (Fig.  1B). The fact that significant exonuclease activity was present in the MonoQ-purified Ec-Endo IV and the MonoS-purified Tm-Endo IV showed that the activity was co-purified with the two Endo IV proteins under very different ion exchange conditions and indicated that the activity might be intrinsic to the Endo IV enzymes. In agreement with this, the exonuclease activity associated with the purified Ec-Endo IV sample has a heat stability (stable after incubation at 65°C for 5 min) similar to that described for the AP endonuclease activity of the Ec-Endo IV (data not shown) (23). The possibility that the exonuclease activity was an arabinose-inducible E. coli gene was investigated. E. coli harboring the empty inducible pBAD-TOPO expression vector or the vector bearing the Ec-Endo IV or Tm-Endo IV genes were grown under identical conditions and induced with arabinose. Crude cell extracts were preincubated at 65°C for 7 min and assayed for exonuclease activity. A low level of background activity was detected in cells harboring the empty vector, but no significant difference in activity was observed between extracts from noninduced and induced cells ( Fig. 2A). Extracts from noninduced cells harboring the Endo IV genes had levels of exonuclease activity similar to that observed in the control extracts. By contrast, significant exonuclease activity was present in extracts from the induced cells harboring either Endo IV gene, confirming that the activity is associated with the Endo IV proteins ( Fig. 2A).
The natural substrate for the Endo IV enzymes is an AP site within a double-stranded DNA context. To determine whether the exonuclease activity was active at cleaved AP sites, a 5Ј end-labeled double-stranded 29-nt bearing a central synthetic (tetrahydrofuranyl) AP site at position 19 (subDS-19AP) was incubated with heat-treated extracts prepared from noninduced and induced cells harboring the plasmids, pBAD-TOPO, pBAD-TOPO-Ec-Endo IV, or pBAD-TOPO-Tm-Endo IV. Cleav-  age at the AP site in this substrate generates two primary cleavage products: an end-labeled 18-nt fragment and an unlabelled 11-nt fragment. Significant AP endonuclease activity was present in each extract and the substrate was cleaved to completion in each case. The 18-nt primary cleavage product (PCP) was further digested in all cases. For extracts prepared from cells harboring the empty vector or from noninduced cells bearing the vectors with the Endo IV genes, digestion was predominantly by one nucleotide. By contrast, exonuclease digestion of the 18-nt PCP by the extracts from the induced cells bearing the vector with either Endo IV gene was much more extensive (Fig. 2B). The 18-nt PCP was almost entirely degraded, and the 17-, 16-, and 15-nt products were clearly visible. To confirm the size of cleavage fragments, cleavage products were analyzed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry. The sizes generated were consistent with an 18-nt PCP and 17-, 16-, and 15-nt degradation products (data not shown). A further control was carried out to investigate the possibility that the exonuclease activity was a contaminant protein that binds to non-Endo IV parts of the purified Endo IV fusion proteins. Essentially a uracil DNA glycosylase gene from T. maritima was cloned into the pBADTOPO expression vector and induced and purified in an identical fashion to the E. coli and T. maritima Endo IV genes/proteins. Exonuclease activity above background levels was not detected when the purified glycosylase from T. maritima was incubated at either 37 or 60°C with either substrate (data not shown).
Purified E. coli Endo IV proteins were obtained from different sources and assayed for exonuclease activity using the double-and single-stranded substrates, subDS-19AP and subSS-19AP, respectively. All of the Endo IV proteins cleaved the substrates at the AP site and exhibited progressive exonuclease activity of varying levels on the double-stranded substrate ( Fig. 3 and Table I). Quantitation of the level of cleavage at the AP site in the substrates showed that the AP activity of the Endo IV enzymes was almost 2-fold more active on doublestranded DNA in comparison with single-stranded DNA (Table  I). Some exonuclease activity was detected on the singlestranded substrate, but the extent of 3Ј-5Ј digestion of the 18-nt fragment was significantly less than that observed for the double-stranded substrate. In addition, the 18-nt PCP generated from cleavage of the AP site on the single-stranded substrates was not generally digested by more than one nucleotide in a 3Ј-5Ј direction by the exonuclease activity (Fig. 3). Exonuclease digestion of the 18-nt PCP resulting from cleavage of the AP site in the double-stranded substrate was substantial. For example, in the case of the purified Ec-Endo IV MonoQ fraction, when the AP site was cleaved to ϳ73%, 73% of product was expected as the 18-nt PCP. However, only ϳ33% was present, indicating that ϳ55% of the 18-nt PCP was further digested by at least one nucleotide (Table I).
Characterization of the Exonuclease Activity-The effect of pH on the exonuclease activity of E. coli Endo IV was characterized. A larger amount of enzyme was used for this characterization to facilitate examination of the effect of pH on both the activity and extent to which the enzyme progressed on the substrate. The E. coli Endo IV displayed both AP activity and exonuclease activity over a wide pH range with optimal exonuclease activity at pH 9.0 (Fig. 4A). At this pH, there was complete cleavage of the AP site in the substrate. In the absence of exonuclease activity, the 18-nt PCP should account for 100% of the product. At pH 9.0, the 18-nt PCP only accounted for 11% of the cleaved product. The smallest digestion products were detectable at this pH, showing that the enzyme progressed furthest on the substrate at this pH. Both the exonuclease activity and the extent of progression on the substrate decreased as pH was increased above 9.0 (Fig. 4A). Prior to pH optimization, all of the exonuclease assays were performed in a reaction mixture containing 10 mM Tris/HCl, pH 8.3, 10 mM KCl. Following pH optimization, all of the subsequent assays were performed using 25 mM glycine-NaOH buffer, pH 9.0.
The Endo IV exonuclease activity was very sensitive to the salt concentration, with little activity observed at NaCl concentrations of Ͼ150 mM (Fig. 4B). Similar results were obtained for KCl (data not shown). In contrast, the AP endonuclease activity of Endo IV was active at low and high salt concentrations with Ͼ95% substrate cleavage occurring in the absence of added NaCl and ϳ75% cleavage occurring in the presence of 1 M NaCl (Fig. 4B).
The effect of magnesium, the metal ions associated with Endo IV (zinc and manganese) (9), and EDTA on the exonuclease activity was analyzed using both substrates (Fig. 5). 100 ng of the Ec-Endo IV cleaved ϳ65% of the AP sites in the subDS-19AP substrate, whereas in the presence of MgCl 2 , MnCl 2 , ZnCl 2 , MgSO 4 , MnSO 4 , and ZnSO 4 , cleavage was ϳ95, 67, 93, 90, 90, and 93%, respectively. It was difficult to estimate exonuclease activity using this substrate because the level of AP site cleavage varied. However, it was clear that the exonuclease activity was most progressive in the presence of zinc (Fig. 5). Progression of the enzyme on the substrate was also high in the presence of MnSO 4 but not in the presence of MnCl 2 . Although the presence of MgCl 2 or MgSO 4 appeared to stimulate AP activity, the progression of the exonuclease was lower than that for ZnCl 2 , ZnSO 4 , or MnSO 4 . Interestingly in the presence of EDTA, the highest level of digestion of the 18-nt PCP was observed, but progression was lowest. Overall, the metal ion effects were reproducible; however, significant variation was observed between experiments. The reason for this is unclear.
The effect of the reducing agent DTT on the exonuclease activity of the Endo IV was investigated (Fig. 6). Addition of DTT to the reaction mixture reduced the AP activity of the enzyme by 2-20% depending on the buffer used and the concentration of agent. Addition of 1 mM DTT to the standard reaction containing subDS-19AP and Endo IV in the glycine/ NaOH reaction buffer, pH 9.0, did not appear to significantly alter the level of 18-nt PCP generated. However, in the presence of DTT, the number and intensity of fragments smaller than 17 nt were much lower than in the absence of DTT. This indicates that the progression of the exonuclease on the substrate was impaired by the DTT. The addition of the same amount of DTT to the standard reaction in Tris/HCl buffer, pH 9.0, had less effect on the progress of the exonuclease activity. The inhibitory effect was much more pronounced in 100 mM DTT. Using a Tris/HCl buffer, pH 7.6, or a Hepes/KOH buffer, pH 7.6, no inhibitory effect on exonuclease activity was observed in the presence of 1 mM DTT, but impairment progression of the exonuclease was high at a 100 mM concentration (Fig. 6).
Inhibition of the Exonuclease Activity-Ec-Endo IV has previously been characterized extensively. However, a significant exonuclease activity associated with the enzyme has not been reported. In the majority of characterizations, EDTA, DTT, and KCl have been included in reaction buffers (10,11). The effect of these agents, singularly and combined, on the exonuclease activity of the Endo IV was investigated (Fig. 7). The addition of EDTA (1 mM), DTT (1 mM), or KCl (100 mM) to the standard reaction inhibited the exonuclease activity significantly but not completely. The addition of all three agents to the reaction mixture inhibited the activity to the extent that only the 18-nt PCP of the AP endonuclease activity was detected.
Determination of the Relative Activity of the Exonuclease Activity of Ec-Endo IV on Different Substrates-To determine the relative activity of the exonuclease activity of Ec-Endo IV on different substrates, 100 ng of the enzyme was incubated with 20 and 18-nt duplex oligonucleotides with 3Ј recessed ends, a 30-nt duplex substrate with blunt ends, and a 20-nt duplex substrate with a 3Ј overhang (Fig. 8). Comparison of the extent of digestion of the substrates showed that ϳ20% of the labeled 20-nt oligonucleotide of subDS-20R was digested by one nucleotide or more, and more than 45% of the 18-nt oligonucleotide of subDS-18R was digested. By contrast, the activity on the labeled 30-nt oligonucleotide of the substrate subDS-30B, was significantly less at ϳ12%, and the enzyme had little or no activity on 3Ј overhangs (Fig. 8). Digestion of the 18-nt PCP of subDS-19AP was greatest, with ϳ72% of the 18-nt product digested by one or more nucleotides.
Comparison of the Relative AP Endonuclease and Exonuclease Activity of Endo IV-The relative AP endonuclease and exonuclease activities of Ec-Endo IV were compared ( Fig. 9 and Table II)  low. For this substrate, 78.6% of substrate remained uncleaved after incubation with 100 ng of Endo IV under the conditions used. Of the 21.4% of cleaved product, the single-stranded 18-nt PCP accounted for ϳ17.2%, and a 17-nt product accounted for ϳ4%. This shows that the 18-nt PCP generated was subsequently digested further by one nucleotide to a level of ϳ20%. Additional smaller products were not observed. Thus, exonuclease activity on the single-stranded substrate is relatively low. By comparison, the level of exonuclease activity on the double-stranded substrate was significantly higher. Only 40% of substrate remained uncleaved after incubation with 100 ng of Endo IV. Of the 60% of cleaved product, the 18-nt PCP accounted for ϳ28%, a 17-nt product accounted for ϳ23%, a 16-nt product accounted for 4%, and additional smaller products were visible. DISCUSSION Previous characterization of Endo IV has shown that the enzyme specifically cleaves the DNA backbone at AP sites and also removes 3Ј DNA blocking groups such as 3Ј phosphates, 3Јphosphoglycolates, and 3Ј ␣,␤-unsaturated aldehydes that arise from oxidative base damage and the activity of combined glycosylase/lyase enzymes (1,2,4). By contrast, and unlike the major AP endonuclease Exo III, negligible exonuclease activity has been associated with Endo IV. The results reported here show that Endo IV does possess an intrinsic 3Ј-5Ј exonuclease activity; the activity was detected in purified preparations of the Endo IV protein from E. coli and from the distantly related thermophile T. maritima; it co-eluted with both enzymes under different chromatographic conditions; induction of either Endo IV in an E. coli overexpression system resulted in induction of the exonuclease activity; and the E. coli exonuclease activity had similar heat stability to the Endo IV AP endonuclease activity.
Considering that Ec-Endo IV is a highly characterized and widely explored enzyme, we sought to identify the reason that this activity was not detected previously. Characterization of the exonuclease activity was carried out using a variety of substrates under a variety of conditions. The Endo IV exonuclease was active over a broad pH range with labeled fragments lower than 5 nt detectable at the optimum pH, indicating that the exonuclease progresses extensively on the substrate and is active on substrates with small duplexed regions. The Endo IV exonuclease activity was very sensitive to the ionic strength of the reaction mixture. As the salt concentration increased, the portion of substrate digested to smaller fragments decreased. At 0 mM added NaCl, the 18-nt PCP was extensively digested into several smaller fragments. By 100 mM NaCl, it was pre-dominantly digested by one nucleotide; by 150 mM, significant inhibition was observed; and by 200 mM NaCl, the exonuclease activity was almost completely inhibited. By contrast, the AP activity appeared relatively insensitive to the NaCl concentration. The progression of the exonuclease on the substrate was enhanced by the presence of excess zinc, indicating that zinc may be important for processivity of the Endo IV on its substrate in addition to its known role in phosphodiester bond cleavage (22). By contrast, magnesium and manganese appeared to inhibit progression of the exonuclease activity on the substrate (although MnSO 4 appeared to have no effect). EDTA inhibited progression of the exonuclease activity and also had an inhibitory effect on the AP activity. The addition of the reducing agent DTT inhibited the progression of the enzyme on the substrate. This was an unexpected finding and suggests that the exonuclease activity of the enzyme may be more active when the cell is under oxidative stress. The Endo IV exonuclease activity may be processive or distributive, and further experiments are needed to clarify this issue.
Typical Endo IV assays in previous reports contained 100 mM KCl, 1 mM DTT, and 1 mM EDTA in the reaction mixture (10,11). The effect of cumulative addition of the agents to the reaction mixture on the Endo IV exonuclease activity was assessed. Addition of any of the agents reduced progress of the enzyme on the substrate, and addition of all three agents completely inhibited the exonuclease activity. Because previous characterizations of the Endo IV enzyme have predominantly used buffers including all three agents, this may explain why the exonuclease activity was not observed previously.
The results shown indicate that the Endo IV exonuclease activity works very well within an AP site cleavage context and has a strong preference for substrates with either 3Ј recessed ends or an incised AP site (Fig. 8) and raises the question as to what the extent of exonuclease activity in vivo is at such sites. Reconstitution of the E. coli base excision-repair pathway using crude or purified enzymes, including Endo IV showed no detectable resynthesis of DNA 5Ј of AP sites (5). However, the conditions used are likely to have been refractory to Endo IV exonuclease activity. Comparison of the relative AP endonuclease and exonuclease activity of Endo IV indicates that the exonuclease activity is high and that exonuclease action is likely to follow AP incision and have a functional significance in vivo. Although the function of the 3Ј-5Ј exonuclease activity of Exo III is unknown and has been considered redundant, the discovery of substantial exonuclease activity associated with Endo IV argues in favor of a significant functional importance for this activity in vivo.