The role of Mg2+ and specific amino acid residues in the catalytic reaction of the major human abasic endonuclease: new insights from EDTA-resistant incision of acyclic abasic site analogs and site-directed mutagenesis1

Article No. jmbi.1999.2888 available online at http://www.idealibrary.com on

J. Mol. Biol. (1999) 290, 447±457

The Role of Mg2‡ and Specific Amino Acid Residues in the Catalytic Reaction of the Major Human Abasic Endonuclease: New Insights from EDTA-resistant Incision of Acyclic Abasic Site Analogs and Sitedirected Mutagenesis Jan P. Erzberger and David M. Wilson III* Molecular and Structural Biology Division, Lawrence Livermore National Laboratory P.O. Box 808, L-452 Livermore, CA 94551, USA

Ape1, the major protein responsible for excising apurinic/apyrimidinic (AP) sites from DNA, cleaves 50 to natural AP sites via a hydrolytic reaction involving Mg2‡. We report here that while Ape1 incision of the AP site analog tetrahydrofuran (F-DNA) was 7300-fold reduced in 4 mM EDTA relative to Mg2‡, cleavage of ethane (E-DNA) and propane (PDNA) acyclic abasic site analogs was only 20 and 30-fold lower, respectively, in EDTA compared to Mg2‡. This ®nding suggests that the primary role of the metal ion is to promote a conformational change in the ringcontaining abasic DNA, priming it for enzyme-mediated hydrolysis. Mutating the proposed metal-coordinating residue E96 to A or Q resulted in a 600-fold reduced incision activity for both P and F-DNA in Mg2‡ compared to wild-type. These mutants, while retaining full binding activity for acyclic P-DNA, were unable to incise this substrate in EDTA, pointing to an alternative or an additional function for E96 besides Mg2‡coordination. Other residues proposed to be involved in metal coordination were mutated (D70A, D70R, D308A and D308S), but displayed a relatively minor loss of incision activity for F and P-DNA in Mg2‡, indicating a non-essential function for these amino acid residues. Mutations at Y171 resulted in a 5000-fold reduced incision activity. A Y171H mutant was fourfold less active than a Y171F mutant, providing evidence that Y171 does not operate as the proton donor in catalysis and that the additional role of E96 may be in establishing the appropriate active site environment via a hydrogen-bonding network involving Y171. D210A and D210N mutant proteins exhibited a 25,000-fold reduced incision activity, indicating a critical role for this residue in the catalytic reaction. A D210H mutant was 15 to 20-fold more active than the mutants D210A or D210N, establishing that D210 likely operates as the leaving group proton donor. # 1999 Academic Press

*Corresponding author

Keywords: DNA repair; Ape1; Hap1; nuclease; phosphate hydrolysis

Introduction Abbreviations used: AP, apurinic/apyrimidinic; EDTA, ethylenediaminetetraacetic acid; BER, base excision repair; E-DNA, ethane-DNA; P-DNA, propaneDNA; F-DNA, tetrahydrofuran-DNA; Wt, wild-type; ExoIII, exonuclease III (E. coli); EndoIV, endonuclease IV (E. coli); Apn1, apurinic endonuclease 1 (S. cerevisiae); DNase I, deoxyribonuclease 1; EMSA, electrophoretic mobility shift assay; ss, single-stranded; ds, doublestranded; BSA, bovine serum albumin. E-mail address of the corresponding author: [email protected] 0022-2836/99/270447±11 $30.00/0

Ape1, the major apurinic/apyrimidinic (AP) endonuclease in mammalian cells (Demple et al., 1991; Robson & Hickson 1991), initiates the repair of abasic sites in DNA and is an essential component of the base excision repair (BER) pathway (Xanthoudakis et al., 1996). AP sites are generated by spontaneous hydrolysis of the N-glycosyl bond (Lindahl, 1993) or via the removal of damaged or inappropriate bases by DNA glycosy# 1999 Academic Press

448 lases (Krokan et al., 1997). The 10,000 AP sites generated per mammalian genome per day are potentially mutagenic or lethal lesions, due largely to their non-coding nature (Loeb & Preston, 1986). Ape1 initiates the repair of such damages by hydrolyzing the DNA backbone immediately 50 to an AP site, generating a 30 -OH group and a 50 deoxyribose moiety (Demple & Harrison, 1994). Following incision, components of the BER pathway complete the process (Wilson & Thompson, 1997; Wilson, 1998). AP endonucleases have been classi®ed into two families based on homology to either Escherichia coli exonuclease III (ExoIII) or E. coli endonuclease IV (EndoIV) (Demple et al., 1997). Ape1 is homologous to ExoIII and displays the characteristic Mg2‡-responsiveness seen for the ExoIII family members (Kane & Linn, 1981; Barzilay et al., 1995). The EndoIV family, which includes Apn1 of Saccharomyces cerevisiae, acts in a Mg2‡-independent manner, but little is known about the catalytic mechanism of this family of proteins. The crystal structures of ExoIII (Mol et al., 1995), Ape1 (Gorman et al., 1997), and the non-speci®c bovine endonuclease DNase I with and without DNA (Suck et al., 1986; Weston et al., 1992), have provided important insights into the binding pockets and catalytic mechanisms of these structurally related proteins. Combined with the results of sitespeci®c mutagenesis, there is strong evidence that a D-H-H2O triad operates to generate the catalytic nucleophile within the active site of the enzyme (Mol et al., 1995; Barzilay et al., 1995; Jones et al., 1996). Speci®cally, residues D283 and H309 of Ape1 (and the corresponding conserved residues of ExoIII and DNase I) form a hydrogen bond that allows the histidine residue to act as a general base, accepting a proton from water to generate the reactive OHÿ nucleophile. Structural studies have also revealed a role for residue E96 of Ape1 and the equivalent glutamates of ExoIII and DNase I in metal coordination. In each case, the divalent metal ion soaked into the protein crystals (Mn2 ‡ for ExoIII, Ca2 ‡ for DNase I and Sm2 ‡ for Ape1) was found to speci®cally associate with the glutamate residue. Moreover, biochemical studies have shown that an Ape1 E96A mutant displays a reduced AP siteincision activity and an altered Mg2 ‡ -dependency when compared to the wild-type protein (Barzilay et al., 1995), further supporting the notion that this amino acid contributes to metal coordination. However, based on the active site pocket structure of Ape1, other potential metal coordinating residues have been identi®ed, including D210, D308 and D70 (Gorman et al., 1997). The speci®c role of these residues in recognition and catalysis remains largely unde®ned. D210 and D308 are conserved in ExoIII, Ape1 and DNase I , while D70 is replaced by N in ExoIII and by R in bovine DNase I, as determined by alignments and their positions within the active site. D70 was suggested as a potential metal-coor-

Ape1 Incision Mechanism

dinating residue due to its proximity to the scissile phosphate group in an Ape1
AP-DNA model (Gorman et al., 1997), but, as noted above, a coordinating role for this residue would be unique to Ape1. Previous studies have reported a ®ve- to 25fold reduction in AP site incision activity for an Ape1 D308A mutant and a preference for Mn2‡ over Mg2 ‡ (Barzilay et al., 1995; Masuda et al., 1998), indicating a potential role in metal positioning. A D210N mutation has been reported to abolish the incision activity of Ape1 (Hang et al., 1997), as has the equivalent mutation in DNase I (Jones et al., 1996), but the contribution of this conserved residue has not been de®ned. Suggested roles include Mg2 ‡ -coordination in DNase I (Jones et al., 1996) and leaving group protonation in ExoIII (Mol et al., 1995) and Ape1 (Gorman et al., 1997). A variety of leaving group stabilization mechanisms have been described for enzyme-catalyzed nucleophilic substitution reactions, including stabilization by a metal ion (Beese & Steitz, 1991) or direct protonation via histidine (Hondal et al., 1998) or aspartic acid residues (Wu & Zhang, 1996; Denu et al., 1996). It remains unclear how the ExoIII family members stabilize the leaving group. For DNaseI, a Y-E-H triad has been suggested where the histidine residue acts as a general acid to protonate the leaving group (Weston et al., 1992; Jones et al., 1996). This triad is not conserved in ExoIII or Ape1, however. Instead, residue Y171, which occupies a similar position in the active site of ExoIII and Ape1 as the histidine residues does in DNase I, has been proposed as the leaving group protonator for the human AP endonuclease (Jones et al., 1996). As noted earlier, D210 has also been suggested as a possible proton donor due to its proximity to the 50 scissile phosphate in an Ape1
AP-DNA model (Gorman et al., 1997). The data presented here provides new insights into the catalytic reaction mechanism of Ape1 as it pertains to the role of Mg2‡ and the contribution(s) of speci®c amino acid residues in metal-coordination and leaving group protonation.

Results Binding of Ape1 to acyclic AP site structures It has been proposed that recognition of Ape1 is mediated by a speci®c interaction of the protein with an extrahelical deoxyribose of the target AP site (Gorman et al., 1997). However, previous studies have found that Ape1 cleaves duplex DNA substrates harboring acyclic (E-DNA and P-DNA) or cyclic (F-DNA) abasic site analogs (Figure 1) with nearly the same catalytic ef®ciency (
Ape1 Incision Mechanism

Figure 1. Structures of acyclic and cyclic abasic site analogs. Chemical structures of the abasic site analogs ethane (E-DNA), propane (P-DNA) and tetrahydrofuran (F-DNA). The sequence of the abasic site-containing DNA was 50 -GTCACCGTCXTACGACTC-30 and its complementary strand was 50 -GAGTCGTACGACGGTGAC-30 , where X represents the abasic site.

mobility shift assays (EMSA), we found that, while puri®ed recombinant Ape1 protein formed detectable complexes with F-containing DNA substrates, complexes with E-DNA and P-DNA were much less prominent (Figure 2). Moreover, the amount of stable protein-DNA complexes observed with E and P-DNA did not signi®cantly change with increasing amounts of protein, while an additional band (not observed in the F-DNA reactions), previously shown to be incised DNA (Wilson et al., 1997), was readily detected (Figure 2). These results suggested that Ape1 was incising acyclic AP sites in the absence of Mg2‡. Mg2‡-independent incision at acyclic abasic sites by Ape1 In order to determine whether our binding results were consistent with E and P-DNA cleavage in EDTA, Ape1 incision assays were per-

449 formed in buffers containing either 10 mM Mg2 ‡ or 4 mM EDTA (a metal chelator). Consistent with previous ®ndings (Wilson et al., 1995), Ape1 cleaved E, P and F-containing substrates in the presence of 10 mM MgCl2 with similar ef®ciencies (Figure 3(a)). The rate of Ape1 incision of E and PDNA was 20 and 30-fold lower, respectively, in EDTA compared to Mg2 ‡ (Figure 3(a)), whereas incision of F-DNA was 7300-fold lower in EDTA than in Mg2 ‡ (summarized in Table 1). This EDTA-resistant incision of E and P-DNA was protein concentration and time-dependent, and was seen with both recombinant Ape1 protein and Ape1 isolated from HeLa cells. The differential incision activities observed between the cyclic and acyclic AP sites in EDTA provides evidence that the Ape1 protein samples were not carrying contaminating Mg2‡, consistent with the previous observation that Ape1 does not stably bind the metal ion (Barzilay et al., 1995). Moreover, presoaking Ape1 in buffer containing 4 mM EDTA for up to 60 minutes on ice prior to carrying out the incision reactions in either EDTA or Mg2‡-containing buffer did not alter the incision rates for any of these substrates. To determine if P-DNA incision in EDTA was an Ape1-speci®c phenomenon, we performed endonuclease assays with the structurally related enzymes ExoIII and DNase I. ExoIII only exhibited a fourfold higher incision activity for P-DNA compared to F-DNA in EDTA (Figure 3(b)), in contrast to the 160-fold difference observed with Ape1 (Table 1). Furthermore, this incision of P-DNA in EDTA by ExoIII was >1000-fold lower than the enzyme's activity in Mg2‡, again a different trend from that observed with Ape1. No nuclease activity was observed for DNase I in EDTA (Figure 3(b)). The ®nding that neither ExoIII nor DNase I exhibit a signi®cant EDTA-resistant activity is evidence that Mg2‡ is not a contaminant of the P-DNA stocks.

Figure 2. Binding of Ape1 to E-, P- and F-DNA substrates. EMSAs were performed by incubating 100 fmol of 32Plabeled duplex E, P or F-DNA with 2 or 10 ng of Ape1 (56 or 280 fmol) in EDTA and fractionating the 10 ml reactions on a non-denaturing gel. Arrows indicate the position of Ape1-DNA complexes, unbound double-stranded (ds) or single-stranded (ss) DNA and incised DNA. The ssDNA migrates slightly slower than the incised DNA and represents <5 % of the total labeled DNA.

450

Ape1 Incision Mechanism

Figure 3. Incision of cyclic and acyclic abasic sites in Mg2‡ and EDTA. (a) Ape1 incision. An aliquot of 0.5 pmol labeled E, P, or F-DNA was incubated in 10 ml with 20 or 200 pg (0.56 or 5.6 fmol) of Ape1 in buffers containing either 10 mM Mg2‡ or 4 mM EDTA for ®ve minutes at 37  C. Reactions were loaded onto 16 % denaturing polyacrylamide gels and bands were visualized by phosphorimager analysis. (b) ExoIII and DNase I incision. 0.5 pmol of labeled P or F-DNA was incubated with either 500 pg ExoIII (18 fmol) or 500 pg bovine DNase I (17 fmol) and analyzed as described above. Reaction conditions were identical with those used for Ape1. Arrows indicate full length 18-mer substrate and incised 9-mer product. Control samples without protein are indicated by - throughout.

Incision and binding of F and P-DNA by Ape1 E96 mutants Given the Mg2 ‡ -independent susceptibility of P-DNA to Ape1-promoted incision, this substrate presented a potential tool for identifying the residue(s) of Ape1 that is/are involved solely in Mg2‡ coordination. In this context, mutating a residue that only serves to coordinate the divalent metal during catalysis should, while reducing F-DNA incision considerably, not affect the overall ability of the Ape1 mutant to incise P-DNA in Mg2‡ or EDTA. However, if a speci®c residue is mutated and activity for both P and F-DNA is reduced signi®cantly, it would suggest that the residue is either not involved in metal coordination or that it possesses an additional function in catalysis.

Since E96 of Ape1 represents the most likely candidate for metal coordination (Barzilay et al., 1995), this residue was tested ®rst by mutating it to A and Q. Both mutants exhibited 600-fold reduced endonuclease activity for F and P-DNA in Mg2 ‡ containing buffer, and neither mutant retained the ability to incise P-DNA in EDTA (Figure 4). EMSA performed in EDTA revealed that both mutants bound F-DNA with similar or greater af®nity than the wild-type, and that these mutants complexed with P-DNA with af®nities nearly identical with FDNA (Figure 5). These results establish the overall structural integrity of the Ape1 E96 mutant proteins and imply that this residue is either not involved in metal coordination or has an additional function in the incision reaction. The E96 mutant proteins also formed protein-DNA complexes with E-DNA (data not shown), but with

Table 1. Speci®c activities of Ape1 incision of E, P and F-DNA Substrate E-DNA P-DNA F-DNA

Mg2‡

EDTA

Ratio

3647  280 3482  134 5101  458

184  27 114  29 0.7  0.05

20:1 30:1 7300:1

Speci®c activities and standard deviations of at least ®ve different experiments are given for each substrate and are expressed as units  103/mg Ape1). The ratio of activity in Mg2‡ relative to activity in EDTA is also given. All assays were performed in the linear range of Ape1 enzymatic activity as described in Materials and Methods.

451

Ape1 Incision Mechanism

Figure 4. Activities of the mutant proteins for P and F-DNA substrates in Mg2‡ and EDTA. The speci®c activities and the standard deviations have been determined from at least four independent experiments for each of the mutants. The values represent units (pmol minÿ1)/mg protein  103. The lower limit of detection for our assays is 5 units/mg and this value indicates that the activity is below the threshold of detection.

a twofold lower af®nity than that observed for P and F-DNA. Incision and binding of F and P-DNA by other potential metal-binding residue mutants We tested other potential metal coordinating residues (D70, D210, and D308) for their role in

Figure 5. E96A and E96Q binding to P and F-DNA substrates. Graphic representation of EMSA binding data obtained for E96Q
P-DNA (!); E96Q
F-DNA (!); E96A
P-DNA (*); E96A
F-DNA (*) and wild-type (wt) Ape1
F-DNA (&). A 100 fmol sample of labeled DNA was incubated with protein amounts varying from 0.5 ng to 16 ng (14 fmol - 448 fmol). The percentage of shifted DNA was quanti®ed by phosphorimager analysis. Note that wtApe1 binding to P-DNA is not measurable due to incision.

this step of the catalytic reaction. Mutations at positions D70 and D308 led only to a moderate reduction in AP endonuclease activity for F-DNA in Mg2‡, indicating a non-essential function for these residues (Figure 4). However, several notable trends were observed during the course of these experiments: The D70A mutant retained signi®cant EDTA-resistant incision activity for P-DNA, whereas mutations of D308 to A or S and the D70R mutation resulted in a more substantial reduction in the ability to incise P-DNA in EDTA (Figure 4). A reduced activity was observed for the D308 mutants for P-DNA relative to F-DNA in Mg2‡ (Figure 4). D308A, D308S and D70R mutants were unable to stably bind P-DNA in EDTA, but retained normal F-DNA binding af®nity (Figure 6; data not shown for D308S). Consistent with the incision pro®les of the various mutants in EDTA, incised P-DNA product was only observed prominently with the D70A protein in the EMSAs. Mutating residue D210 to an A or N residue essentially abolished incision activity (Figure 4), but did not affect binding activity for either substrate (Figure 6; data not shown for D210N), indicating a critical role for this amino acid residue in catalysis. Thus, none of the alternative residues appears to be an obvious metal-binding candidate. Y171 and D210: potential roles in the active site chemistry and the catalytic reaction To examine the potential of Y171 as the leaving group proton donor in the catalytic reaction and to

452

Ape1 Incision Mechanism

Figure 6. Binding of D70A, D70R, D308A and D210A mutants to P and F-DNA. EMSAs containing 100 fmol of P or F-DNA and 2, 10 or 25 ng (56, 140 or 700 fmol) mutant protein were performed as described in the legend to Figure 2. Arrows indicate the position of mutant protein-DNA complexes, unbound dsDNA and incised DNA product.

test the signi®cance of the hydrogen bond between Y171 and E96 present in the crystal structure of Ape1 (Gorman et al., 1997), Y171 was mutated to F to remove the hydroxyl group and to H to introduce the proposed proton donor residue of DNase I (Weston et al., 1992; Jones et al., 1996). Both mutants displayed a signi®cantly reduced incision activity while retaining normal binding activity

(data not shown), indicating an important role for this residue in incision. Moreover, Y171F (928(45) units/mg) was found to exhibit a higher incision activity than Y171H (280(20) units/mg) for FDNA in Mg2‡ (Figure 7), a pattern not consistent with a role in leaving group protonation. Since Y171 seemed unlikely to operate as the proton donor and since it had been suggested from

Figure 7. Incision of F-DNA by D210 and Y171 mutants. (a) A 0.5 pmol sample of labeled F-DNA was incubated with 1, 5 or 25 ng (28, 140 or 700 fmol) mutant protein as described in the legend to Figure 3. Arrows indicate full length 18-mer substrate and incised 9-mer product. (b) Graphical representation of the speci®c activities and standard deviations. WtApe1 activity is indicated in Table 1.

Ape1 Incision Mechanism

the crystal structure of Ape1 that D210 may function in this capacity (Gorman et al., 1997), we investigated the role of this residue in leaving group protonation. A D210H mutant was able to incise P and F-DNA in Mg2‡ with speci®c activities 15 and 20-fold higher (7874(680) units/mg for PDNA and 3228(110) units/mg for F-DNA) than those observed for D210A and D210N (Figure 7), indicating that a proton donating group at this position retains activity. pH titration studies were performed to further examine the role of this residue as a proton donor. Unfortunately, EMSAs revealed that Ape1-DNA complex stability is severely compromised at pH values >8.0 (data not shown), and therefore no useful information about pKa shifts of residues acting as general acids in the catalytic reaction could be obtained. Similar observations have been reported for ExoIII (Black & Cowan, 1997).

Discussion The ExoIII family of AP endonucleases, and the structurally related protein DNase I, employ an acid-base catalytic reaction mechanism in order to incise the phosphodiester linkage of DNA. This hydrolytic incision reaction is promoted by Mg2‡, which is thought to stabilize the developing charge on the phosphate oxygen atoms, strategically positioning them for the subsequent nucleophilic attack by the active site-generated hydroxyl radical (Gerlt, 1993). Our ®nding that E and P-DNA are incised in the presence of 4 mM EDTA suggests that, for these substrates, Ape1 promotes hydrolysis in the absence of divalent metal ions. Since there is no reason to expect a Mg2 ‡ -dependent difference in the nucleophile generation or the leaving group reaction of E and P-DNA relative to F-DNA, we propose that the primary role of Mg2 ‡ is to facilitate the formation of a conformational state in FDNA that is primed for enzyme-catalyzed hydrolysis. Thus, in this context, Ape1 would be capable of inducing this hydrolysis-prone conformational state in E and P-DNA without the aid of Mg2 ‡ . NMR studies have shown that the phospho-diester backbone immediately adjacent to an acyclic AP site analog assumes a different conformational state than the corresponding linkage in DNAs harboring the ring-containing tetrahydrofuran residue (Kalnick et al., 1989), and is indeed more ¯exible in nature (Takeshita et al., 1987). Moreover, an analysis of the co-crystal structure of DNase I bound to DNA in the absence of the metal cofactor reveals that a considerable conformational change is necessary to move the leaving group of the target phosphate in line with the active site nucleophile. This observation may indicate that Mg2‡ plays a similar role in establishing the transition state/intermediate substrate conformation in DNase I.

453 If, as we suggest, the major role for Mg2‡ is the establishment of an ``activated'' DNA backbone conformation, our results indicate that this step of the reaction is rate limiting. Once the hydroxyl ion and the PÐO(30 ) bond have been aligned within the active site pocket, nucleophilic attack and leaving group protonation would presumably occur rapidly. Quantitatively, the contribution of Mg2‡ to the incision rate is over 7300-fold for F-DNA, compared to only 20 or 30-fold for the acyclic substrates. It is noteworthy that Ape1 appears to retain some incision activity in EDTA for cyclic AP sites, implying that even these substrates may occasionally undergo a protein-induced conformational change that leads to incision in the absence of Mg2‡. Natural AP sites exist predominantly in a ring-closed state (i.e. cyclic in nature), possibly explaining the Mg2‡-dependence of incision for these substrates. Whether there are a subset of in vivo substrates (e.g. ring-opened AP sites) that are susceptible to Mg2‡-independent incision merits further investigation. Mutating residue E96 led to a reduction in activity for both F and P-DNA in Mg2‡, and to the inability of the mutant Ape1 protein to incise PDNA in EDTA. These ®ndings indicated that E96 is not solely involved in metal coordination and that this residue has either an alternative or additional function in the catalytic reaction. Given that: (1) samarium is bound to E96 in the Ape1 crystals (Gorman et al., 1997); (2) an E96A mutation results in a metal ion concentration-sensitive incision activity (Barzilay et al., 1995); and (3) mutagenesis of the other potential metal-coordinating residues D70 and D308 had only a minor effect on the incision capacity of Ape1, the existing data still supports the notion that E96 is the main metalcoordinating residue, but that this residue must have an additional function. Since mutations at either E96 or Y171 dramatically reduced the incision capacity of Ape1, and since these two residues are within hydrogen-bonding distance in the crystal structure of Ape1 (Gorman et al., 1997), we propose that the additional role for E96 is in establishing the precise active site chemistry for catalysis via a hydrogen-bonding network including Y171. Recent results indicating that an E96A mutation causes a change in the ¯uorescence of Ape1 (Izumi et al., 1999) offers supportive evidence that E96 mutations lead to a distortion that affects active site residues, reducing the incision activity of these mutants without affecting substrate binding (Figure 5). The leaving group reaction of DNase I has been suggested to involve a triad of residues (Y76-E78H134), with H134 operating as the proton donor. In the ExoIII family members, a tyrosine residue (Y109 in ExoIII and Y171 in Ape1) is present at the equivalent position to H134, leading to the suggestion that Y171 may function as the leaving group proton donor in Ape1 (Jones et al., 1996). However, the Y171F mutant protein retained signi®cant incision activity and was fourfold more active than

454

Ape1 Incision Mechanism

Figure 8. Proposed catalytic mechanism of Ape1. In our Ape1 reaction scheme, two conformational states exist for the substrate DNA, one that is speci®c to the recognition complex (ÿMg2‡; Wilson et al., 1997) and one that is speci®c to the incision complex (‡Mg2‡; data presented here). That is, upon formation of the initial recognition Ape1-DNA complex, an additional DNA rearrangement is needed to permit the ef®cient execution of the hydrolytic reaction. We propose that Mg2‡, which is likely positioned within the active site by E96, functions to promote this rearrangement. Once this conformational state has been established in AP-DNA, the scissile phosphate group is readily prone to nucleophilic attack by the D283-H309-water triad and leaving group protonation by residue D210. Other residues in Ape1, such as N212, N68, E96 and Y171, are likely to be crucial for establishing the active site environment, including residue positions and pKas, through a hydrogen-bonding network.

a Y171H mutant, which should possess an increased proton-donating capability. A Y171H/ G130E double mutant, which in theory should fully duplicate the triad chemistry of DNase I in Ape1, also displayed an activity lower than Y171F, even at pH 6.5 where DNase I is fully active (J.P.E. & D.M.W., unpublished results). These results suggest that Y171 is not involved in leaving group protonation, but clearly indicate an important role for this residue in catalysis, perhaps in the establishment of the appropriate active site chemistry as suggested above. It is worth mentioning that Y171, and in particular a ring structure at this position, may also contribute to protein-DNA complex formation, as a Y171Q mutant displays a 15-fold reduced binding af®nity relative to wild-type as determined by EMSAs (J.P.E. & D.M.W., unpublished results). The ®nding that D210H is 15 to 20-fold more active than D210A and D210N mutants provides the ®rst evidence that this residue operates as the leaving group proton donor (the updated reaction scheme for Ape1 is summarized in Figure 8). A considerable pKa shift would be required for protonation to occur at this aspartic acid residue within the active pH range (6.5 - 9.0) of Ape1 (Kane & Linn, 1981). Thus, the residues immediately surrounding D210 (e.g. N68 and N212), which are perfectly conserved in each member of this structural family, probably help establish the required active site environment. N212 has previously been shown to contribute to DNA binding (Rothwell et al., 1996), and we are currently investigating the potential role of these residues in catalysis. Interestingly, EDTA-resistant cleavage of acyclic AP sites was not observed with DNase I or ExoIII, perhaps re¯ecting the substrate-dependent nucleolytic rate differences observed between these struc-

turally related proteins. The surprising observation that Ape1 D308A, D308S and D70R mutants bind F-DNA, but not P-DNA, indicates that these residues play some role in complex stability and/or in promoting the DNA conformational change, yet their precise functions are presently dif®cult to predict. As noted earlier, there is evidence that D308 may contribute to metal ion coordination (Barzilay et al., 1995; Masuda et al., 1998). One noteworthy observation is that residue D70 of Ape1 is substituted by an asparagine residue in ExoIII (N9), a glutamine residue in human DNase I (Q9) or an arginine residue in bovine DNase I (R9). It will be interesting to ®nd out whether the residue differences at this position in these proteins correlate with their ability to bind and incise P-DNA in EDTA. An analysis of the DNase I
DNA structure leads us to speculate that the residues corresponding to D70 and D308 of Ape1 (Q9 and D251 of human DNase I) engage in minor groove interactions near the scissile phosphate group or contact the DNA backbone on the 30 side of the damaged site. Finally, the observation that catalytically inactive Ape1 mutants bind to P-DNA and F-DNA with similar af®nities is conclusive evidence that the ring structure of the AP site, while in¯uencing certain aspects of catalysis, is not essential for damagespeci®c recognition or protein-DNA complex stability. Based on previous work (Erzberger et al., 1998, and references therein), we favor the idea that it is the unique compatibility of Ape1 and AP-DNA that permits site-speci®c recognition and complex formation. Recently, a study has demonstrated that a loop structure present in the ExoIII-like endonucleases can impart increased AP site speci®city when introduced into DNase I (Cal et al., 1998).

455

Ape1 Incision Mechanism

Materials and Methods Enzymes and oligonucleotides The [g-32P]ATP, T4 polynucleotide kinase and DraIII restriction enzyme were obtained from Amersham (Arlington Heights, IL). BamH1, ExoIII and DNase I and Pfu polymerase were obtained from Stratagene (La Jolla, CA). All oligonucleotides were obtained from Operon (Alameda, CA), except for P-DNA and E-DNA, which were provided by Dr Masaru Takeshita (SUNY, Stony Brook, USA; Takeshita et al., 1987). The following oligonucleotides were used for site-directed mutagenesis: Bam3 0 Ape1, CGGGATCCTCACAGTGCTAGGTA; Ape1DraIII, AAACCTCACCCAGTGCGAAAC; Ape1E96A, CTGTGCCTTCAAGCGACCAAATGTTCAG; Ape1E96Q, CTGTGCCTTCAACAGACCAAATGTTCAG; Ape1D210A, GTGCTGTGTGGAGCCCTCAATGTGGC; Ape1D210N, GTGCTGTGTGGAAACCTCAATGTGGC; Ape1D210H, GTGCTGTGTGGACACCTCAATGTGGC; Ape1D70A, CTCTTGGAATGTGGCTGGGCTTCGAGCCTG; Ape1D70R, CTCTTGGAATGTGCGTGGGCTTCGAGCCTG; BamApe1D308A, CGGGATCCTCACAGTGCTAGGTATAGGGTGATAGGACAGTGAGCACTG-CCGAGGGC; BamApe1D308S, CGGGATCCTCACAGTGCTAGGTATAGGGTGATAGGACAGTGAGAACTGCCGAGGGC; Ape1Y171F, GCTGGTAACAGCATTTGTACCTAATGC; Ape1Y171H, GCTGGTAACAGCACATGTACCTAATGC.

Figure 9. Puri®ed wtApe1 and Ape1 mutant proteins. Approximately 1 mg of wtApe1 and various Ape1 mutants was loaded on a SDS-12 % polyacrylamide gel, separated and stained with Coomassie brilliant blue R 250. Lanes correspond to M, molecular weight standards (weight in kDa is indicated); 1, wtApe1; 2, D70A; 3, D70R; 4, D308A; 5, D308S; 6, E96Q; 7, E96A; 8, D210A; 9, D210N; 10, D210H; 11, Y171F and 12, Y171H.

similar chromatographic pro®les and bound F-DNA with af®nities nearly identical to wild-type (see Results), indicating that all proteins retain signi®cant structural integrity.

Generation and purification of Ape1 mutant proteins

Incision and binding assays

With the exception of the D308A and D308S mutants (where the mutant codon was incorporated into an extended 30 primer and only one round of PCR was required), site-directed mutagenesis was performed using the PCR overlap method (Ausubel, 1997). Two rounds of PCR were required, with the ®rst round consisting of two separate standard PCR reactions. In reaction 1, a fragment of the Ape1 cDNA was ampli®ed using pETApe1 (Erzberger et al., 1998) as the template and two oligonucleotides containing speci®c restriction sites, Ape1DraIII and Bam30 Ape1. In reaction 2, a fragment shorter than the one produced in reaction 1 was generated using the pETApe1 template, and one of the site-speci®c mutant primers (see above) with Bam30 Ape1. These two PCR products were then mixed at a 1:10 ratio (reaction 1 product:reaction 2 (mutant) product), and the second round of PCR was performed with Ape1DraIII and Bam30 Ape1 primers. All PCR reactions were carried out with Pfu DNA polymerase. Final PCR products and the pETApe1 plasmid were digested with DraIII and BamH1, the appropriate restricted fragments were gelpuri®ed, and the DNAs were ligated and transformed into DH5a bacteria. Recombinant plasmids were puri®ed using the alkaline lysis method (Ausubel, 1997) and sequenced by dye-terminator sequencing (Amersham, Arlington Heights, IL) to con®rm the presence of the mutation. All mutant proteins were expressed in bacteria and puri®ed as described for the wild-type protein (Erzberger et al., 1998), except that the IPTG-induction step was carried out at 28  C to maximize protein solubility. The levels of protein expression and solubility were comparable to the wild-type protein at 28  C with the exception of the D210 mutants, which were ten- to 25fold less soluble than wild-type Ape1. Proteins were >95 % pure as determined by Coomassie blue staining of SDS-polyacrylamide gels (Figure 9). All mutants had

Nuclease activity assays and EMSAs were performed as described elsewhere (Erzberger et al., 1998), except that the binding and dilution buffers contained 100 mg/ ml bovine serum albumin (BSA) and 0.01 % (v/v) Triton X-100. These modi®cations helped prevent loss of protein activity at low concentrations and reduced the amount of denatured DNA formed during electrophoresis. Activities of wild-type protein were measured in the linear range of the incision reaction as previously determined (Wilson et al., 1995). Activity units are de®ned as pmol of abasic DNA incised per minute at 37  C. It should be noted that to obtain detectable levels of incision for the E96, Y171, and D210 mutants, we needed to increase protein concentrations 50 to 1000-fold relative to the protein concentrations used in the wild-type reactions, and therefore these speci®c activities should be used for relative comparisons only. We should also note that no mutant protein completely lacked enzymatic activity, but whether the residual activity observed is intrinsic to Ape1 itself or the result of a minor contaminant cannot be con®dently determined at the elevated protein concentrations employed. As a reference point, an H309S mutant (residue H309 is considered essential for catalysis and therefore mutations at this position should essentially inactivate the enzyme), has a residual activity of 0.1(0.02)  103 units/mg at a protein concentration 1000-fold higher than wild-type (J.P.E. & D.M.W., unpublished results).

Acknowledgments We thank Ms Tina Xi and Dr Matthew Coleman for invaluable sequencing support, Drs Felice Lightstone, Daniel Barsky, Michael Colvin and members of the

456 Wilson lab for helpful discussions and Dr James George for critical input. Drs Masaru Takeshita and Arthur Grollman kindly provided the main E and primary PDNA stocks. This work was carried out under the auspices of the US Department of Energy by Lawrence Livermore National Laboratory under contract no. W7405-ENG-48 and supported by LDRD (97-ERD-002) and NIH (CA79056) grants to D.M.W.

References Ausubel, F. M. (1997). Editor of Current Protocols in Molecular Biology, John Wiley & Sons, New York, NY. Barzilay, G., Mol, C. D., Robson, C. N., Walker, L. J., Cunningham, R. P., Tainer, J. A. & Hickson, I. D. (1995). Identi®cation of critical active-site residues in the multifunctional human DNA repair enzyme HAP1. Nature Struct. Biol. 2, 561-568. Beese, L. S. & Steitz, T. A. (1991). Structural basis for the 30 -50 exonuclease activity of Escherichia coli DNA polymerase I: a two metal ion mechanism. EMBO J. 10, 25-33. Black, C. B. & Cowan, J. A. (1997). Inert chromium and cobalt complexes as probes of magnesium-dependent enzymes. Evaluation of the mechanistic role of the essential metal cofactor in Escherichia coli exonuclease III. Eur. J. Biochem. 243, 684-689. Cal, S., Tan, K. L., McGregor, A. & Connolly, B. A. (1998). Conversion of bovine pancreatic DNase I to a repair endonuclease with a high selectivity for abasic sites. EMBO J. 17, 7128-7138. Demple, B. & Harrison, L. (1994). Repair of oxidative damage to DNA: enzymology and biology. Annu. Rev. Biochem. 3, 915-948. Demple, B., Herman, T. & Chen, D. S. (1991). Cloning and expression of APE1, the cDNA encoding the major human apurinic endonuclease: de®nition of a family of DNA repair enzymes. Proc. Natl Acad. Sci. USA, 88, 11450-11454. Demple, B., Harrison, L., Wilson, D. M., III, Bennett, R. A., Takagi, T. & Ascione, A. G. (1997). Regulation of eukaryotic abasic endonucleases and their role in genetic stability. Environ. Health Perspect. 105, 931-934. Denu, J. M., Lohse, D. L., Vijayalakshmi, J., Saper, M. A. & Dixon, J. E. (1996). Visualization of intermediate and transition-state structures in protein-tyrosine phosphatase catalysis. Proc. Natl Acad. Sci. USA, 93, 2493-2498. Erzberger, J. P., Barsky, D., SchaÈrer, O. D., Colvin, M. E. & Wilson, D. M., III (1998). Elements in abasic site recognition by the major human and Escherichia coli apurinic/apyrimidinic endonucleases. Nucl. Acids Res. 26, 2771-2778. Gerlt, J. A. (1993). Mechanistic principles of enzymecatalyzed cleavage of phosphodiester bonds. In Nucleases, 2nd edit., pp. 1-34, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. Gorman, M. A., Morera, S., Rothwell, D. G., de La Fortelle, E., Mol, C. D., Tainer, J. A., Hickson, I. D. & Freemont, P. S. (1997). The crystal structure of the human DNA repair endonuclease HAP1 suggests the recognition of extra-helical deoxyribose at DNA abasic sites. EMBO J. 16, 6548-6558. Hang, B., Rothwell, D. G., Sagi, J., Hickson, I. D. & Singer, B. (1997). Evidence for a common active site

Ape1 Incision Mechanism for cleavage of an AP site and the benzene-derived exocyclic adduct, 3, N4-benzetheno-dC, in the major human AP endonuclease. Biochemistry 36, 1541115418. Hondal, R. J., Zhao, Z., Kravchuk, A. V., Liao, H., Riddle, S. R., Yue, X., Bruzik, K. S. & Tsai, M. D. (1998). Mechanism of phosphatidylinositol-speci®c phospholipase C: a uni®ed view of the mechanism of catalysis. Biochemistry, 37, 4568-4580. Izumi, T., Malecki, J., Chaudhry, M. A., Weinfeld, M., Hill, J. H., Lee, J. C. & Mitra, S. (1999). Intragenic suppression of an active site mutation in the human apurinic/apyrimidinic endonuclease. J. Mol. Biol. 287, 47-57. Jones, S. J., Worrall, A. F. & Connolly, B. A. (1996). Sitedirected mutagenesis of the catalytic residues of bovine pancreatic deoxyribonuclease I. J. Mol. Biol. 264, 1154-1163. Kalnik, M. W., Chang, C. N., Johnson, F., Grollman, A. P. & Patel, D. J. (1989). NMR studies of abasic sites in DNA duplexes: deoxyadenosine stacks into the helix opposite acyclic lesions. Biochemistry, 28, 3373-3383. Kane, C. M. & Linn, S. (1981). Puri®cation and characterization of an apurinic/apyrimidinic endonuclease from HeLa cells. J. Biol. Chem. 256, 3405-3414. Krokan, H. E., Standal, R. & Slupphaug, G. (1997). DNA glycosylases in the base excision repair of DNA. Biochem. J. 325, 1-16. Lindahl, T. (1993). Instability and decay of the primary structure of DNA. Nature, 362, 709-715. Loeb, L. A. & Preston, B. D. (1986). Mutagenesis by apurinic/apyrimidinic sites. Annu. Rev. Genet. 20, 201-230. Masuda, Y., Bennett, R. A. O. & Demple, B. (1998). Rapid dissociation of human apurinic endonuclease (Ape1) from incised DNA induced by magnesium. J. Biol. Chem. 273, 30360-30365. Mol, C. D., Kuo, C. F., Thayer, M. M., Cunningham, R. P. & Tainer, J. A. (1995). Structure and function of the multifunctional DNA-repair enzyme exonuclease III. Nature, 374, 381-386. Robson, C. N., Hochhauser, D., Craig, R., Rack, K., Buckle, V. J. & Hickson, I. D. (1992). Structure of the human DNA repair gene HAP1 and its localisation to chromosome 14q 11.2-12. Nucl. Acids Res. 20, 4417-4421. Robson, C. N. & Hickson, I. D. (1991). Isolation of cDNA clones encoding a human apurinic/apyrimidinic endonuclease that corrects DNA repair and mutagenesis defects in E. coli xth (exonuclease III) mutants. Nucl. Acids Res. 19, 5519-5523. Suck, D. & Oefner, C. (1986). Structure of DNase I at Ê resolution suggests a mechanism for binding 2.0 A to and cutting DNA. Nature, 321, 620-625. Takeshita, M., Chang, C. N., Johnson, F., Will, S. & Grollman, A. P. (1987). Oligodeoxynucleotides containing synthetic abasic sites. Model substrates for DNA polymerases and apurinic/apyrimidinic endonucleases. J. Biol. Chem. 262, 10171-10179. Weston, S. A., Lahm, A. & Suck, D. (1992). X-ray structure of the DNase I-d(GGTATACC)2 complex at Ê resolution. J. Mol. Biol. 226, 1237-1256. 2.3 A Wilson, S. H. (1998). Mammalian base excision repair and DNA polymerase beta. Mutat. Res. 407, 203215. Wilson, D. M., III & Thompson, L. H. (1997). Life without DNA repair. Proc. Natl Acad. Sci. USA, 94, 12754-12757.

Ape1 Incision Mechanism Wilson, D. M., III, Takeshita, M., Grollman, A. P. & Demple, B. (1995). Incision activity of human apurinic endonuclease (Ape1) at abasic site analogs in DNA. J. Biol. Chem. 270, 16002-16007. Wilson, D. M., III, Takeshita, M. & Demple, B. (1997). Abasic site binding by the human apurinic endonuclease, Ape1, and determination of the DNA contact sites. Nucl. Acids Res. 25, 933-939.

457 Wu, L. & Zhang, Z. Y. (1996). Probing the function of Asp128 in the lower molecular weight protein-tyrosine phosphatase-catalyzed reaction. A pre-steadystate and steady-state kinetic investigation. Biochemistry, 35, 5426-5434. Xanthoudakis, S., Smeyne, R. J., Wallace, J. D. & Curran, T. (1996). The redox/DNA repair protein, Ref-1, is essential for early embryonic development in mice. Proc. Natl Acad. Sci. USA, 93, 8919-8923.

Edited by T. Richmond (Received 14 December 1998; received in revised form 3 May 1999; accepted 12 May 1999)