intragenic suppression of an active site mutation in the human apurinic/apyrimidinic endonuclease

Article No. jmbi1999.2573 available online at http://www.idealibrary.com on J. Mol. Biol. (1999) 287, 47±57

Intragenic Suppression of an Active Site Mutation inthe Human Apurinic/Apyrimidinic Endonuclease

Tadahide Izumi1,2, Jedrzej Malecki2, M. Ahmad Chaudhry3

Michael Weinfeld3, Jeff H. Hill1, J. Ching Lee2 and Sankar Mitra1,2*

1Sealy Center for MolecularScience and 2Department ofHuman Biological Chemistryand Genetics, University ofTexas Medical BranchGalveston, TX 77555-1079, USA3Experimental OncologyCross Cancer InstituteEdmonton, Canada T6G 1Z2

E-mail address of the [email protected]

Abbreviations used: AP site, apusite; BER, DNA base excision repairEDTA, ethylenediaminetetra-aceticapurinic/apyrimidinic endonucleasmethanesulfonate; nfo, endonucleasbuffered saline; 30-PG, 30-phosphoglexonuclease III; APE, apyrimidinic

0022-2836/99/110047±11 $30.00/0

The apurinic/apyrimidinic endonucleases (APE) contain several highlyconserved sequence motifs. The glutamic acid residue in a consensusmotif, LQE96TK98 in human APE (hAPE-1), is crucial because of its rolein coordinating Mg2�, an essential cofactor. Random mutagenesis of theinactive E96A mutant cDNA, followed by phenotypic screening inEscherichia coli, led to isolation of an intragenic suppressor with a secondsite mutation, K98R. Although the Km of the suppressor mutant wasabout sixfold higher than that of the wild-type enzyme, their kcat valueswere similar for AP endonuclease activity. These results suggest that theE96A mutation affects only the DNA-binding step, but not the catalyticstep of the enzyme. The 30 DNA phosphoesterase activities of the wild-type and the suppressor mutant were also comparable. No global changeof the protein conformation is induced by the single or double mutations,but a local perturbation in the structural environment of tryptophan resi-dues may be induced by the K98R mutation. The wild-type and suppres-sor mutant proteins have similar Mg2� requirement for activity. Theseresults suggest a minor perturbation in conformation of the suppressormutant enabling an unidenti®ed Asp or Glu residue to substitute forGlu96 in positioning Mg2� during catalysis. The possibility that Asp70 issuch a residue, based on its observed proximity to the metal-binding sitein the wild-type protein, was excluded by site-speci®c mutation studies.It thus appears that another acidic residue coordinates with Mg2� in themutant protein. These results suggest a rather ¯exible conformation ofthe region surrounding the metal binding site in hAPE-1 which is notobvious from the X-ray crystallographic structure.

# 1999 Academic Press

Keywords: DNA repair; AP endonuclease; 30phosphoesterase; missensemutation; site-directed mutagenesis*Corresponding author

Introduction

Apurinic and apyrimidinic (AP) sites in DNA,generated either by spontaneous and oxidativedepurination or as intermediate products duringbase excision repair in DNA (BER), are genotoxicand mutagenic (Wallace, 1994). AP endonucleases(APE) incise DNA strands 50 to the AP sites and

ing author:

rinic/apyrimidinic; DTT, dithiothreitol;acid; hAPE-1, humane 1; MMS, methyle IV; PBS, phosphateycolate; xth,exonuclease.

produce 30 OH termini (Wallace, 1994; Doetsch &Cunningham, 1990). Such enzymes are distin-guished from AP lyases which were originallyclassi®ed as type I and generate 30 phospho a,bunsaturated aldehyde via b-elimination (Doetsch &Cunningham, 1990). APEs also possess a 30 DNAphosphoesterase activity and thus remove abnor-mal 30 termini such as 30-phosphoglycolate (30-PG)which are generated by ROS at DNA strand breaksand prevent DNA repair synthesis (Doetsch &Cunningham, 1990). Escherichia coli has two APEgenes, exonuclease III (Xth) and endonuclease IV(Nxfo), encoded by xth and nfo, respectively. Thexth nfo double mutant is highly sensitive to suchgenotoxic agents as methyl methanesulfonate(MMS) and reactive oxygen species which lead tothe production of AP sites and 30-blocked termini

# 1999 Academic Press

mailto:[email protected]

48 Intragenic Suppression of Human APE-1

(Doetsch & Cunningham, 1990; Cunningham et al.,1986).

Only one APE gene, APE-1 (originally namedAPEX, APE, HAP1, and Ref-1), has been identi®edand cloned in mammalian cells; the protein is ahomolog of E. coli Xth (Demple et al., 1991; Robson& Hickson; 1991; Seki et al., 1992; Xanthoudakiset al., 1992). This enzyme accounts for most of thecellular APE activity, and its activation in responseto oxidative stress has been recently reported(Fung et al., 1998; Ramana et al., 1998). Besides theDNA repair function, the human APE-1 (hAPE-1)possesses two other unrelated functions. The pro-tein was identi®ed as a redox-enhancing factor,Ref-1, which reductively activates AP-1 and p53transcription factors (Xanthoudakis et al., 1992;Jayaraman et al., 1997). Indeed, the N terminusdomain has recently been reported to interact withthioredoxin which appears to be needed for redoxreactivation of the Ref-1 activity (Qin et al., 1996;Hirota et al., 1997). The hAPE-1 was also identi®edas a corepressor of the parathyroid hormone geneand of its own gene (Okazaki et al., 1994; Izumiet al., 1996). Although the biological signi®canceof the multiple, unrelated properties of hAPE-1is not yet clear, the fact that the gene is essentialfor the early embryonic development in mice(Xanthoudakis et al., 1996) indicates its essentialfunctions in cells.

Distinct regions in the hAPE-1 polypeptide areresponsible for the endonuclease/30-phosphoester-ase and gene regulatory activities (Xanthoudakiset al., 1994). The 60 amino acid residues at the Nterminus, which are not conserved in Xth, areinvolved in AP-1 activation and dispensable forthe APE activity (Walker et al., 1993; Izumi &Mitra, 1998). Further truncation of the protein fromboth N and C termini (80 and ®ve amino acid resi-

dues, respectively) led to total loss of the APEactivity (Izumi & Mitra, 1998).

Alignment of the amino acid sequences of themembers of the APE family shows several highlyconserved motifs (Seki et al., 1992). One of these isthe GXDHCP sequence in which the histidine resi-due was identi®ed as the catalytic residue(Figure 1). Another motif, LQETK, is also con-served in most APE enzymes (Figure 1). Indeed,elucidation of the tertiary structure of hAPE-1 andXth by X-ray crystallography led to the predictionthat Glu96 (E96) in the motif LQE96TK98 in hAPE-1 is essential for endonuclease activity because ofits role in coordinating the cofactor Mg2� (Barzilayet al., 1995a,b; Gorman et al., 1997). Substitution ofE96 with alanine (E96A) caused loss of the endonu-clease activity. However, a small activity of E96Awas observed in the E96A mutant protein (about400 times less than that of the wild-type) in con-trast to other active site mutants, such as H309N,of which the activity was at least 2000-fold lessthan that of the wild-type protein (Barzilay et al.,1995a,b). Because the metal ion cofactor is absol-utely required for the reaction (Gorman et al.,1997), we considered it likely that other amino acidresidues were also involved in the binding of Mg2�

with six coordinate bonds, and thus may beresponsible for the residual activity of the E96Amutant. Participation of multiple residues in themetal binding was also reported in E. coli RNaseH (Kashiwagi et al., 1996; Uchiyama et al., 1994).We entertained the possibility that a second mis-sense mutation in E96A could induce a subtle con-formational change in the neighborhood of theactive site surface that would allow positioning ofother residues in order to coordinate Mg2� moreef®ciently, and thereby restore the catalytic activityof the protein. Such a mutant protein could pro-

Figure 1. Homology of polypep-tide sequences of APEs. Only thehomology of residues 61 to 99 andthe C-terminal segment of hAPE-1are shown. Sequences in Swiss-plotdatabase were aligned with GCG9.0. Names on the left are the entrynames in Swiss-Plot. The numbersin the second column denote thestarting residue number. The APendonuclease activity in hAPE-1 islost after deletion of the shadedregions (Izumi & Mitra, 1998). TheE96 and K98 residues of hAPE-1and the two highly conservedmotifs are shown in bold letters.

Figure 3. Phenotypic rescue of E. coli BW528 withwild-type (W) and mutant APEs. Details of MMS treat-ment were given in Materials and Methods. Vec, theempty vector control.

Intragenic Suppression of Human APE-1 49

vide signi®cant insight into the nature of metalbinding to the essential DNA repair enzyme inmammals. Here, we report isolation and character-ization of such a missense suppressor of the E96Amutant which restored the AP endonuclease and30-PG-removing activity. Our results support themodel of participation of multiple acidic residuesin coordinating Mg2� and thereby the presence ofa ¯exible conformation of the protein.

Results

Isolation of intragenic suppressor mutant ofE96A; phenotypic rescue of E. coli by hAPE-1

APE is involved in the repair of DNA damageinduced by MMS, because AP sites are producedas intermediates during repair process(Cunningham et al., 1986). The human APE-1 wasshown to be able to complement xth nfo-negativeE. coli (Demple et al., 1991; Robson & Hickson,1991; Izumi & Mitra, 1998). We have recentlyestablished a phenotypic rescue system in E. colithat provides a sensitive screening procedure foridentifying active versus inactive hAPE-1 (Izumi &Mitra, 1998).

Figure 2. (a) Site-directed mutagenesis of the hAPE-1by PCR. The 50 and 30 vector primers ¯anking thecDNA and two primers for each mutagenesis areshown. Newly introduced EcoRV (E96A) and NsiI(H309A) sites, as well as an endogenous site (BglII forD70N), are underlined. Silent mutations are shown inbold letters. After ampli®cation by PCR, the fragmentswere inserted into the NdeI-SalI site in pIZ42 (Izumi &Mitra, 1998). (b) A diagram of the random mutagenesisprocedure. The same vector primers as in (a) were used.

We examined APE activity of two mutant pro-teins, E96A and H309A, in the cell toxicity assay(Figures 2 and 3). Both cDNAs failed to provideresistance to the E. coli strain against MMS. Noactivity was detected in either crude extracts(Figure 4(b)). We thus concluded that the E96A aswell as H309A missense mutants were inactive(Barzilay et al., 1995b). Then we utilized the pheno-typic rescue strategy to examine the possibility ofintragenic suppression of E96A. An expressionplasmid cDNA containing E96A was randomlymutagenized by PCR (Spee et al., 1993), and intro-duced into xthÿ nfoÿ E. coli by transformation.Resistant clones were isolated after challenging thebacteria with MMS treatment (Figure 2). Our pre-liminary experiments showed that the original pro-tocol by Spee et al. (1993) yielded high frequency ofmultiple mutations in control experiments (datanot shown). Because multiple mutations are unde-sirable for our purpose, we modi®ed the protocolby omitting dITP and reducing the concentrationof one dNTP to 40 mM, while maintaining the con-centration of the other three dNTPs at 200 mM(Spee et al., 1993). Four independent PCR reactionswith limiting concentration of different dNTPswere carried out so that misincorporation couldoccur at all sites. Of the approximately 2 � 104

independent clones screened, 16 clones wereselected on the basis of MMS resistance of the hostbacteria. A high level of cellular protection wasobserved in the selected clones compared to thebacteria harboring the E96A mutant plasmid(Figure 3). The DNA sequence was determined forfour of them, and in all cases Lys98 was found to

Figure 4. AP endonuclease activity of wild-type andmutant proteins of hAPE-1. (a) Immunoblot assay ofwild-type, E96A, and E96A K98R expressed in E. coliBW528. The same amount of crude extracts was ana-lyzed by SDS-PAGE (12 % acrylamide) and subsequentimmunoblot assay (Izumi & Mitra, 1998). Total and sol-uble fractions were analyzed separately. (b) AP endonu-clease activity was measured in crude lysates using atetrahydrofuran-containing 43mer oligonucleotide.

�


be replaced with an arginine residue (K98R, AAAto AGA), while the original missense mutationremained unchanged. When expressed in E. coli,

the level of expression and solubility of the mutantprotein were indistinguishable from those of thewild-type (Figure 4(a)), suggesting that no signi®-cant change in expression of the protein or in itsglobal conformation was caused by the mutation;this was further con®rmed by circular dichroismand intrinsic ¯uorescence studies of the wild-typeand mutant proteins as described below. AP endo-nuclease activity in the crude extract of E. coliexpressing the suppressor mutant protein wasfound to be about half that in bacteria expressingthe wild-type protein (Figure 4(b)). We have con-cluded therefore that the E96A K98R doublemutant regained the AP endonuclease activitydespite the loss of E96, the residue predicted to beessential for Mg2� coordination (Gorman et al.,1997; Barzilay et al., 1995b).

To examine if such intragenic suppression is acommon phenomenon among the APEs, the corre-sponding glutamic acid (E34) and lysine (K36) resi-dues in the LQETK motif of the E. coli Xth proteinwere changed to Ala and Arg, respectively, by site-speci®c mutagenesis and the mutants were testedin the cell protection assay. While the E34A mutantwas inactive in cell survival assay as expected(Figure 5(a)), the defect was again suppressed byintroduction of the secondary missense, K36R,although the effect was not as pronounced as thatobserved for hAPE-1.

Characterization of the mutant protein and theeffect of Mg2� on incision activity

To further characterize the suppressor mutant onhAPE-1, we puri®ed the wild-type and the mutantproteins to near homogeneity (Figure 6(a)). TheE. coli xth nfo strain (BW528) was lysogenized withl-DE3, a lambda phage containing the T7 RNApolymerase gene inducible by IPTG. Thus, the pur-i®ed APE protein did not contain any background

Figure 5. Phenotypic rescue ofE. coli BW528 with wild-type (W)and mutant APEs. (a) E. coli BW528with wild-type xth gene, E34A, andE34A K36R in pIZ42 were treatedwith MMS. (b) hAPE-1 cDNA withmissense mutations were subjectedto phenotypic rescue assay.W, wild-type hAPE-1; Vec, controlempty vector.

Figure 6. Analysis of the puri®ed wild-type and missense mutant proteins. (a) Coomassie blue staining of puri®edwild-type (W), and mutant proteins after SDS-PAGE. (b) Mg2� dependency for the AP endonuclease activity of thewild-type and mutant proteins. The 43mer substrate oligonucleotide was incubated with 20 pg of APEs for one min-ute at 37�C. The activity in 3 mM Mg2� was used for normalization. (c) Release of 30-PG by wild-type and suppressormutant APE-1. The indicated amounts of irradiated DNA were incubated with the enzymes and the residual 30-PGwas determined by a postlabeling procedure as described in Materials and Methods. (d) A 13mer oligo containing 30-PG was incubated with 0 (lane 1), 1 mg (lanes 2, 5, 8 and 11), 0.5 mg (lanes 3, 6, 9 and 12), or 0.1 mg (lanes 4, 7, 10and 13) of hAPE-1 (lanes 2-4, wild-type; lanes 5-7, E96A; lanes 8-10, K98R; lanes 11-13, E96A K98R). Lane 14, amixture of size markers for 13mer with 30-PG and 30-OH as indicated by arrows.


activity of class II AP endonucleases from E. coli(Cunningham et al., 1986). The DNA strand-clea-vage assay was highly speci®c for the endonu-clease activity of the recombinant APEs because ofthe use of tetrahydrofuran as a substrate. This sub-strate, unlike an intact AP site, is resistant to clea-vage via b elimination by contaminating AP lyases(Takeshita et al., 1987). Barzilay et al. (1995b)showed that although the E96A mutant lostactivity in low concentration of Mg2� (0.1 mM), theactivity was restored when the Mg2� concentrationin the reaction was increased to 2 mM. We exam-ined the effect of Mg2� concentration on theactivity of the mutant protein. As shown inFigure 6(b), the relative activity of the suppressormutant protein in different concentrations of Mg2�

was essentially the same as that of the wild-typeprotein. Increasing the concentration of Mg2� tomore than 1 mM did not affect the activity of thewild-type and the suppressor mutant protein,whose activities were comparable (Figure 6(b)).The activity of E96A was approximately 1000times lower than that of the active proteins (data

not shown). Thus, our results indicate that thedouble mutant protein regained a metal-coordinat-ing motif which functions nearly as ef®ciently asthe wild-type enzyme.

We then inquired whether the missensemutation affected the kinetic properties of theenzyme (Table 1). The Km value of the suppressormutant was found to be about sixfold higher thanthat of the wild-type protein, which in turn wascomparable to the reported value (Wilson et al.,1995). However, the kcat of the wild-type proteinand the mutant were comparable.

The 30-PG removing activity of the suppressormutant protein

We used a postlabeling assay (Weinfeld &Soderlind, 1991; Weinfeld et al., 1997) to examinethe enzymatic activity for removal of 30-PG, a 30-blocking damage in DNA that prevents DNA syn-thesis by DNA polymerases. The mutant proteinremoved the 30-PG residues from g-irradiatedDNA as ef®ciently as the wild-type protein with

Table 1. Kinetic parameters for AP endonuclease activity of wild-type andsuppressor mutant proteins

Km (nM) kcat (sÿ1) kcat/Km

Wild-type 11.9 � 3.6 1.80 � 0.14 0.151E96A K98R 69.5 � 9 1.76 � 0.1 0.025Wild-type mutant 0.17 1.02 6.04

Figure 7. Fluorescence spectroscopy. Emission spectraof the wild-type (*) and mutant hAPE-1 proteins (*,E96A; ~, K98R; &, E96A K98R) were measured asdescribed in Materials and Methods. The excitationwavelength was 295 nm.


three different concentrations of the substrate DNA(Figure 6(c)). Then we examined the 30-PG-remov-ing activity with a 13mer oligonucleotide contain-ing a PG at the 30 end (Figure 6(d)). The activitywas observed in wild-type, K98R, and E96A K98Rproteins with the latter two exhibiting approxi-mately 80 % activity of the wild-type protein,whereas the activity of E96A was undetectable(Figure 6(d)). Taken together, it is evident that thesecond mutation of K98R restored the DNA 30-phosphoesterase activity as well as the endonu-clease activity of the E96A mutant.

Asp70 is dispensable in theintragenic suppression

The recently reported X-ray crystallographicstructure of hAPE-1 shows that Asp70 is located inthe vicinity of the metal-binding locus (Gormanet al., 1997; Figure 1(b)). Because the Mg2� is absol-utely required for activity of the E96A K98R pro-tein like that of the wild-type protein, weconsidered the possibility that the position ofAsp70 was shifted closer to the metal-binding sitein the suppressor mutant so as to hold Mg2� stablyeven in the absence of the E96. In order to test thispossibility we created a triple-missense mutant,D70N E96A K98R, to examine its activity using thecell survival assay (Figure 5(b)). The sensitivity ofE. coli expressing D70N E96A K98R triple mutanthAPE-1 was similar to that expressing the E96AK98R double mutant. Moreover, the single mis-sense mutant, D70N, was as active as the wild-type in this assay (Figure 5(b)). These results indi-cate a lack of involvement of D70 in suppressingthe phenotype of E96A K98R, or the catalytic func-tion of the wild-type.

Investigation of the secondary structure ofhAPE-1 with circular dichroism andfluorescence spectroscopic analysis

To test whether any signi®cant change in the sec-ondary structure occurred as a result of mutations,the CD spectra of the wild-type and mutant hAPE-1 were obtained. The mutants included in thisinvestigation were E96A, K98R and the doublemutant of E96A K98R. Regardless of the nature ofthe mutation, the far-UV CD spectra of thesemutants were identical with that of wild-typehAPE-1 (data not shown). These results indicatethat these mutations do not signi®cantly perturbthe secondary structure of hAPE-1 and that theloss and restoration of activity of the mutant pro-

teins are not due to a net change in their secondarystructure.

CD spectroscopy provides global structuralinformation and may not be able to detect thestructural alterations that are subtle but may causesigni®cant functional changes. Therefore, intrinsictryptophan ¯uorescence was measured to probefor small local perturbations around the varioustryptophan residues (Figure 7). It is interesting tonote that the emission intensity was altered in themutant without a detectable change in the shape ofthe emission spectra. The K98R and E96A K98Rmutants exhibited identical emission spectra withlower intensity as compared to those of the wild-type and E96A mutant. The latter two proteinsexhibited identical emission spectra. These resultsindicate that the K98R mutation exerts an effect onthe structural environment surrounding the trypto-phan residues of hAPE-1. The decrease in emissionintensity without a change in the shape of theemission spectra implies that the net effect of themutation is a decrease in quantum yield of one ormore tryptophan residues. A decrease in quantumyield could be the consequence of an increase incollisional quenching of the ¯uorophore by the sol-vent or a decrease in the apolar nature of theenvironment surrounding the ¯uorophore or both.


It is interesting to note that the crystallographicstructure of hAPE-1 shows that K98 interacts withD70 (Gorman et al., 1997) which is bracketed byW67 and W75. Hence, it is possible that thechanges in ¯uorescence of these tryptophan resi-dues resulted from K98 mutation.

Discussion

Our observation that the K98R missensemutation suppressed the inactivating E96Amutation has important implications in the struc-ture-function relationship of hAPE-1, because thissuggests the presence of an alternative active con-formation of the APE polypeptide. That this possi-bility is not unique to the human APE wassupported by similar results obtained with theE. coli Xth protein, although the effect for Xth wasless striking than that for hAPE-1. Moreover, res-toration of both AP endonuclease and 30-phos-phoesterase activities in the double mutantindicates that intragenic suppression restores afully functional active site.

The E96A K98R suppressor mutant and thewild-type protein have similar kcat values,although the mutant has a higher Km than thewild-type protein. We may explain the results byassuming that the double mutant as well as theinactive E96A mutant retain an intact catalyticdomain, and the loss of the E96 results in itsinef®cient recognition of or binding to thedamaged DNA site. How substitution of Lys98by an Arg residue restores the DNA bindingactivity is not obvious. One simple interpretationcould be that Mg2� is coordinated by a watermolecule held in space by hydrogen bondingwith Arg98 in the suppressor mutant. Thus therequirement for direct interaction with Glu96 iseliminated. However, the possibility is unlikelybecause of the lack of acidic amino acid residuesinvolved in the metal coordination with Mg2�.Apparently E96 is the only acidic residue foundin the metal coordination (Gorman et al., 1997).Other studies indicate that at least one or twoacidic residues are involved for the metal binding(Mol et al., 1995; Kashiwagi et al., 1996; Kostrewaet al., 1996). Another possibility is that Arg, beinga stronger base than Lys, may eliminate the needfor Mg2� in holding the DNA in place. However,this possibility was also eliminated by our obser-vation that the Mg2� requirement was nearlyidentical for the wild-type and the doublemutant. The role of Mg2� in the catalytic activityof APE is not completely understood. The hAPE-1 protein has three distinct enzymatic activities,namely AP-endonuclease, 30 phosphoesterase andRNase H, which appear to have the same activesite residues and all require Mg2� (Barzilay et al.,1995a). The observed presence of a single Sm2�

(Mg2�) bound to E96 in the hAPE-1 (and E34 inXth) after soaking the crystals in the metal saltsolution suggested a single site binding of the

metal ion to the protein in the absence of DNA(Mol et al., 1994; Gorman et al., 1997). We utilizedatomic absorption spectroscopy to quantify theamount of Mg2� bound to the wild-type, E96A,K98R, and E96A K98R mutant proteins afterequilibrium dialysis of 10 mM enzyme solutionsagainst 10 mM Mg2�. No binding was detectedfor any of the proteins (data not shown), while asigni®cant activity could be observed at 10 mMMg2� for the active enzymes (Figure 6(b)).A similar lack of Mg2� binding to the wild-typeenzyme has been reported (Barzilay et al., 1995b).Our results support the earlier conclusion thatMg2� does not have a signi®cant structural rolein hAPE-1 (Barzilay et al., 1995b; Gorman et al.,1997). The structural studies of E. coli RNase H,whose catalytic mechanism involves an Asp-Hispair like that of hAPE-1, suggest that Mg2�

forms a hexa-coordinated complex that includesan ionic bond with a Glu residue (Kashiwagiet al., 1996; Uchiyama et al., 1994). While theoverall mechanism proposed for RNase H issimilar to that proposed by Barzilay et al. (1995b)for hAPE-1 in regard to the role of activated H2Oin phosphodiester bond cleavage, the bindingof Mg2� to the single side-chain of E96 inhAPE-1 has been proposed only by the lattergroup. Site-speci®c mutation studies have clearlyestablished the role of E96 in Mg2� binding.However, based on the present study and ana-logy from the RNase H study, it appears likelythat another acidic residue is involved in Mg2�

binding.Barzilay et al. (1995b) offered several possible

explanations for the presence of low residualactivity of the E96A mutant protein. Here, weshow that the af®nity for Mg2� in the absence ofDNA was not apparently altered in the single anddouble mutants. Furthermore, the Mg2� require-ment was not affected by the suppressormutations. These results are consistent with theinvolvement of a second acidic residue in the pro-tein which can substitute for the missing E96.A potential candidate for such a residue is D70,because the residue is known to be close to thedivalent cation binding site in the original motif(Gorman et al., 1997) and is likely to be affected bya subtle change in the polypeptide conformation.We have, however, excluded this possibility,because (i) we have shown that the triple-missensemutant D70N E96A K98R is as active as the doublemutant (Figure 5(b)) and (ii) this residue is not con-served in other APEs. An Asn residue, for exampleis present in the corresponding site of the Xth pro-tein (Gorman et al., 1997). Furthermore, a singlemissense mutant, D70N, showed the same level ofprotection of E. coli cells as the wild-type(Figure 5(b)). Thus, we have yet to identify the resi-due(s) involved in restoration of the activity in thesuppressor mutant. D308 in hAPE-1 is another can-didate for substituting the E96 residue (Gormanet al., 1997), although it is not as close to the metalion as E96 and D70 as determined from the X-ray


crystallographic structure of the wild-type protein(Gorman et al., 1997). In fact, introduction of a mis-sense mutation at D308 (D308A) caused a 25-foldreduction in the endonuclease activity and alsoaffected the preference for Mn2� (Barzilay et al.,1995b). However, it would be dif®cult to use ourapproach to show that the residue is indeedinvolved in the metal binding. Because D308 isnext to the catalytic residue H309, substitution ofthe original residue could affect the nucleolyticactivity itself but not the metal coordination.Rather, a direct approach in this case would beX-ray crystallographic analysis with the doublemutant protein to determine the distance of theD308 relative to the metal ion.

Finally, our study has implications regardingstructure-function relationship of enzymes beyondthe speci®c case of hAPE-1. Our results predictthat alternative conformations of proteins withidentical catalytic function can exist which couldnot be predicted from the X-ray crystallographicstructure of these proteins. It is interesting to notethat the tertiary structure of the active site ofRNase H, particularly of the loop containing theactive site residue His, was found to be ¯exible(Kashiwagi et al., 1996). We propose that hAPE-1and possibly many other enzymes have ¯exibleconformation around the active site, although oneconformation predominates in the wild-type pro-tein. In the case of hAPE-1, an alternative confor-mation is favored in the K98R mutant of hAPE-1over that present in the wild-type protein whichallows positioning of a second, unidenti®ed acidicresidue to hold Mg2� and to help stabilize the tern-ary complex without signi®cant distortion of theactive site pocket. Many DNA repair enzymes areactive on multiple substrates with signi®cantlydivergent structures (Krokan et al., 1997; Hanget al., 1996). These substrates may be recognized bydistinct conformers of the enzyme. One way to testthis prediction is by elucidating the structure of co-crystals of substrate-enzyme-Mg2� ternary complexof APE-1. Such structural studies will directlyidentify the alternative residues involved in Mg2�

binding.

Materials and Methods

Plasmid DNA and E. coli strains

E. coli xth nfo (BW528) and the cloned xth gene werekindly provided by Dr B. Weiss (Cunningham et al.,1986). The human APEX cDNA was originally a giftfrom Dr S. Seki, and was manipulated further for thepresent study (Seki et al., 1992; Izumi & Mitra, 1998).Missense mutations in the hAPE-1 cDNA were generatedby PCR using Pfu DNA polymerase (Stratagene;Figure 2). Primers for construction of missense mutantsof Xth protein are: E34A, 50GGCCTGCAGGCGACAA-AAGTTCATGAC30; and for E34A K36R, 50GGC-CTGCAGGCGACACGCGTTCATGACGAT30. The PstIsites are underlined and changed codons are in boldletters. In all cases, DNA fragments ampli®ed by PCRwere digested with appropriate enzymes and inserted

into the original vector at the corresponding sites. Allmutations in DNA generated by PCR were con®rmed byDNA sequencing.

Phenotypic rescue (survival) studies

The hAPE-1 cDNA used for the rescue experimentswas cloned into pIZ42 (Izumi & Mitra, 1998) to expressintact proteins and did not contain a His-tag region.Plasmid-bearing E. coli BW528 (xth nfo) was grown over-night at 28�C, diluted 25-fold in fresh LB medium andincubated further at 28�C until A600 reached about 0.6.The cells were centrifuged, washed twice in phosphate-buffered saline (PBS, pH 7.3), and resuspended in anequal volume of PBS. Immediately after addition of var-ious amounts of MMS, cells were incubated at 37�C withshaking for one hour and then diluted in PBS and platedon LB-agar at 28�C for colony counting (Izumi & Mitra,1998). All the rescue experiments were repeated reprodu-cibly more than twice.

Measurement of AP-endonuclease activity in E. colicrude extracts

A 43mer oligonucleotide containing a single tetra-hydrofuran residue (Glen Research), an AP site-analogue, was synthesized (Izumi & Mitra, 1998).The tetrahydrofuran-containing oligonucleotide wasannealed with its complementary strand, puri®ed bynon-denaturing 20 % PAGE, labeled at its 50 end with[g-32P]ATP by T4 polynucleotide kinase (Pharmacia), andused for the incision assay. Crude extracts (10 ng) ofE. coli were mixed with about 50 fmol of the oligonucleo-tide in a reaction mixture (20 ml) containing 60 mM Tris(pH 8.0), 1 mM MgCl2, 0.2 mM EDTA, 1 mM DTT, and100 mg/ml bovine serum albumin (BSA). After incu-bation for ten minutes at 37�C, the reaction was stoppedby the addition of a stop buffer (88 % formamide, 0.5 %bromophenol blue and xylene cyanol). The product wasquanti®ed by analysis in a PhosphorImager 425 (Molecu-lar Dynamics) after electrophoretic separation on 20 %polyacrylamide gels containing 7 M urea.

Isolation of intragenic suppressor mutant of E96A

Random mutations were introduced in the hAPE-1cDNA using a PCR-based method (Spee et al., 1993).Brie¯y, four PCR reactions were carried out in each setin which about 0.1 ng of the cDNA was ampli®ed usingtwo vector primers (Figure 2) with �Taq polymerase(Amersham) and 200 mM of each dNTP except for onereduced to 40 mM. The ampli®ed fragment was cleavedby NdeI and SalI, and then cloned into pIZ42 (Figure 2).The plasmid DNA was transformed into E. coli BW528and resistant clones were screened by serial exposure toMMS, as described above.

Purification of the wild-type and missensemutant proteins

E. coli strain BW528 (xth nfo) was lysogenized withDE3 l-phage using the l-phage lysogenization kit(Novagen). The hAPE-1 cDNAs were introduced intopET15b plasmid vector (Novagen) using its NdeI/XhoIsites in order to express the proteins with the T7 RNApolymerase system (Studier & Moffatt, 1986). Each con-struct carried 20 extra amino acid residues at the N ter-minus containing a histidine hexamer, which allowed


puri®cation of the protein by af®nity chromatography ona nickel column. E. coli BW528(DE3) carrying the cDNAswere grown in a fermenter at 37�C, and 0.5 mM IPTGwas added when A600 reached 1.5, and incubated furtherat 25�C for four hours. Cells were harvested and resus-pended in buffer A (20 mM Tris-Cl (pH 8.0), 300 mMNaCl) and lysed with a French press. The extracts wereapplied on 10 ml TALON (Clontech), washed with100 ml of buffer A, and eluted with buffer A containing10 mM imidazole. The eluents were then applied onto3 ml of Ni-NTA resin (Qiagen), washed with 30 ml ofbuffer A containing 40 mM imidazole, and eluted with6 ml of buffer A containing 200 mM imidazole. Proteinswere concentrated to about 6-10 mg/ml by Centricon 10(Amicon) and applied to a Superdex column (Pharmacia)equilibrated with buffer containing 20 mM Tris-Cl(pH 8.0), 300 mM NaCl, 5 % (v/v) glycerol, and 1 mMDTT. Fractions containing the hAPE-1 proteins were col-lected and stored at ÿ80�C.

Characterization of purified enzymes

The Km and kcat values of the puri®ed enzymes weredetermined using the same 43mer duplex DNA asdescribed above. The substrate at various concentration(Figure 6(d)) was incubated with 50 pg of protein at37�C in a reaction buffer (66 mM Tris-Cl (pH 8.0), 1 mMMg2�, 100 mM NaCl, 1 mM DTT, and 0.1 mg/ml BSA)and the reaction was stopped by the addition of the stopbuffer. The reaction was analyzed on PhosphorImagerand values were calculated by a method described bySakoda & Hiromi (1976). The effect of Mg2� on endonu-clease activity was examined by using the same reactionbuffer with 20 pg of puri®ed proteins and differentamounts of MgCl2. The products were analyzed by Phos-phorImager 425 (Molecular Dynamics) after separationby electrophoresis in denaturing 20 % polyacrylamidegel containing 7 M urea.

Measurement of residual phosphoglycolate

Calf thymus DNA (0.5 mg/ml in 10 mM sodiumphosphate, pH 7.4) was irradiated with 50 Gy g-rays(60Co Gammacell, AECL, Ottawa, ON). This introducesapproximately 1 pmol of phosphoglycolate per 1 mg ofDNA (Weinfeld & Soderlind, 1991). After precipitation,the DNA (0.33, 0.1 and 0.033 mg/ml) was incubated at37�C for one hour with 300 ng of either the wild-typehAPE-1 or the E96A K98R mutant in 30 ml of buffer con-taining 60 mM Tris-Cl (pH 8.0), 100 mM NaCl, 1 mMMgCl2 and 1 mM DTT. The removal of phosphoglycolatefrom 1 mg of DNA from each of these reactions was thendetermined by a postlabeling assay (Weinfeld &Soderlind, 1991).

Measurement of 30-PG removal from anoligonucleotide substrate

A 13mer oligonucleotide with a 30-phosphoglycerylterminus was synthesized in UTMB's Recombinant DNALaboratory (RDL) with glyceryl CPG (control pore glass;Glen Research). The sequence was 50CCTCAGTTT-CACT30. The 30-PG at the end was prepared as described(Urata & Akagi, 1993). Brie¯y, the glyceryl CPG-contain-ing oligo was incubated in 5 mM NaIO4 (pH 6) on icefor two hours to produce 30-phosphoglycolaldehyde. Thereaction was stopped by adding L-methionine, followedby precipitation with ethanol and washing with 70 %

ethanol. Then the oligomer was incubated in 0.5 MNaClO2 in the presence of 36 % dimethyl sulfoxide atroom temperature for ®ve hours to produce 30-PG. The30-PG containing the 13mer oligo was annealed with28mer complementary sequence (50CATGCGGTGCAGA-AGTGAAACTGAGGAG30) after 50 end-labeling with[g-32P]ATP (Amersham) by T4 polynucleotide kinase(Pharmacia), and puri®ed by Sephadex G25 (Pharmacia)gel ®ltration. The substrate oligomer, approximately25 fmol, was incubated at 37�C for ®ve minutes in 10 mlof reaction buffer (20 mM Tris (pH 8.0), 100 mM NaCl,1 mM MgCl2, 0.1 mM EDTA, 1 mM DTT, 100 mg/mlBSA) with puri®ed hAPE-1 protein. The reaction wasstopped by the addition of the stop buffer. The productswere analyzed in 20 % acrylamide gel containing 7 Murea after boiling and by subsequent exposure to Phos-phorImager cassette (Molecular Dynamics).

Circular dichroism spectra analysis

The secondary structure of the wild-type and mutanthAPE-1 in 50 mM potassium phosphate (pH 7.5),0.5 mM NaCl was monitored with an AVIV 62 DS circu-lar dichroism spectropolarimeter. The far-UV CD spectrawere obtained in fused quartz cuvettes with 0.1 cm pathlength and protein solutions with absorbance at 280 nmranging from 0.20 to 0.26. Each spectrum was recordedwith a 0.5 nm increment and one second interval. Foreach sample, ®ve repetitive scans were obtained andaveraged.

Fluorescence spectroscopy

The structure of the wild-type and mutant hAPE-1was also monitored by determining the emission spectrawith a Perkin Elmer LS-50 ¯uorimeter. The excitationwavelength was 295 nm. The emission spectra wereacquired by scanning between 300 and 400 nm. Fourrepetitive scans were obtained and averaged. The spectrawere normalized to the same protein concentration.

Atomic absorption spectroscopy

The amount of magnesium bound to wild-type andmutant hAPE-1 proteins was measured by ¯ame atomicabsorption spectrophotometry using a Perkin Elmermodel 5100 Zeeman spectrometer. The protein sampleswere prepared by dialyzing 10 mM (about 0.4 mg/ml)protein solutions against a buffer containing 20 mM Tris(pH 8.0), 150 mM NaCl, 0.2 mM DTT, and 10 mMMgCl2. The absorption was measured from a magnesiumhollow cathode lamp at 285.2 nm in the presence of 1 %lanthanum to suppress ionization.

Other materials and methods

DNA sequencing was carried out in the RecombinantDNA Laboratory (RDL) in UTMB using T7 and T3 uni-versal primers and internal primers in the hAPE-1 cDNAsynthesized in RDL.

Acknowledgments

We thank Dr Seki for human APEX cDNA and Dr B.Weiss for the E. coli strain. Incisive and helpful sugges-tions of Drs M. Dodson, C. D. Mol, J. A. Tainer, and


K. Morikawa are greatly appreciated. We also thank DrT. Wood of the Recombinant DNA Laboratory at UTMBfor DNA sequencing and oligonucleotide synthesis, andDr N. Alcock for atomic absorption spectroscopic anal-ysis. We thank Dr R. P. Hodge for advice on chemicalrearrangement of oligonucleotide. Dr D. Konkel's edi-torial help and Ms W. Smith's secretarial assistance aregratefully acknowledged.

This work was supported by the U.S. Public HealthScience grants CA53791 and ES08457 (to S.M.); by agrant from the National Cancer Institute of Canada withfunds from the Terry Fox Run (to M.W.); also supportedby NIH grant GM-45579, Robert A. Welch Foundationgrants H-0013 and H-1238 (to J.C.L.) and by NIEHS Cen-ter grant ES06676. The costs of publication of this articlewere defrayed in part by the payment of page charges.This article must therefore be hereby marked ``advertise-ment'' in accordance with 18 U.S.C. Section 1734 solelyto indicate this fact.

References

Barzilay, G., Walker, L. J., Robson, C. N. & Hickson,I. D. (1995a). Site-directed mutagenesis of thehuman DNA repair enzyme HAP1: identi®cation ofresidues important for AP endonuclease and RNaseH activity. Nucl. Acids Res. 23, 1544-1550.

Barzilay, G., Mol, C. D., Robson, C. N., Walker, L. J.,Cunningham, R. P., Tainer, J. A. & Hickson, I. D.(1995b). Identi®cation of critical active-site residuesin the multifunctional human DNA repair enzymeHAP1. Nature Struct. Biol. 2, 561-568.

Cunningham, R. P., Saporito, S. M., Spitzer, S. G. &Weiss, B. (1986). Endonuclease IV (nfo) mutant ofEscherichia coli. J. Bacteriol. 168, 1120-1127.

Demple, B., Herman, T. & Chen, D. S. (1991). Cloningand expression of APE, the cDNA encoding themajor human apurinic endonuclease: de®nition of afamily of DNA repair enzymes. Proc. Natl Acad. Sci.USA, 88, 11450-11454.

Doetsch, P. W. & Cunningham, R. P. (1990). The enzy-mology of apurinic/apyrimidinic endonucleases.Mutat. Res. 236, 173-201.

Fung, H., Kow, Y. W., Van Houten, B., Taatjes, D. J.,Hatahet, Z., Janssen, Y. M., Vacek, P., Faux, S. P. &Mossman, B. T. (1998). Asbestos increases mamma-lian AP-endonuclease gene expression, proteinlevels, and enzyme activity in mesothelial cells.Cancer Res. 58, 189-194.

Gorman, M. A., Morera, S., Rothwell, D. G., de LaFortelle, E., Mol, C. D., Tainer, J. A., Hickson, I. D.& Freemont, P. S. (1997). The crystal structure ofthe human DNA repair endonuclease HAP1suggests the recognition of extra-helical deoxyriboseat DNA abasic sites. EMBO J. 16, 6548-6558.

Hang, B., Chenna, A., Fraenkel-Conrat, H. & Singer, B.(1996). An unusual mechanism for the majorhuman apurinic/apyrimidinic (AP) endonucleaseinvolving 50 cleavage of DNA containing a ben-zene-derived exocyclic adduct in the absence of anAP site. Proc. Natl Acad. Sci. USA, 93, 13737-13741.

Hirota, K., Matsui, M., Iwata, S., Nishiyama, A., Mori,K. & Yodoi, J. (1997). AP-1 transcriptional activity isregulated by a direct association between thiore-doxin and Ref-1. Proc. Natl Acad. Sci. USA, 94, 3633-3638.

Izumi, T. & Mitra, S. (1998). Deletion analysis of humanAP-endonuclease: minimum sequence required for

the endonuclease activity. Carcinogenesis, 19, 525-527.

Izumi, T., Henner, W. D. & Mitra, S. (1996). Negativeregulation of the major human AP-endonuclease, amultifunctional protein. Biochemistry, 35, 14679-14683.

Jayaraman, L., Murthy, K. G., Zhu, C., Curran, T.,Xanthoudakis, S. & Prives, C. (1997). Identi®cationof redox/repair protein Ref-1 as a potent activatorof p53. Genes Dev. 11, 558-570.

Kashiwagi, T., Jeanteur, D., Haruki, M., Katayanagi, K.,Kanaya, S. & Morikawa, K. (1996). Proposal fornew catalytic roles for two invariant residues inEscherichia coli ribonuclease HI. Protein Eng. 9, 857-867.

Kostrewa, D. & Winkler, F. K. (1995). Mg2� binding tothe active site of EcoRV endonuclease: a crystallo-graphic study of complexes with substrate and pro-duct DNA at 2A resolution. Biochemistry, 34, 683-696.

Krokan, H. E., Standal, R. & Slupphaug, G. (1997). DNAglycosylases in the base excision repair of DNA.Biochem. J. 325, 1-16.

Mol, C. D., Kuo, C. F., Thayer, M. M., Cunningham,R. P. & Tainer, J. A. (1995). Structure and functionof the multifunctional DNA-repair enzyme exonu-clease III. Nature, 374, 381-386.

Okazaki, T., Chung, U., Nishishita, T., Ebisu, S., Usuda,S., Mishiro, S., Xanthoudakis, S., Igarashi, T. &Ogata, E. (1994). A redox factor protein, ref1, isinvolved in negative gene regulation by extracellu-lar calcium. J. Biol. Chem. 269, 27855-27862.

Qin, J., Clore, G. M., Kennedy, W. P., Kuszewski, J. &Gronenborn, A. M. (1996). The solution structure ofhuman thioredoxin complexed with its target fromRef-1 reveals peptide chain reversal. Structure, 4,613-620.

Ramana, C. V., Boldogh, I., Izumi, T. & Mitra, S. (1998).Activation of apurinic/apyrimidinic endonucleasein human cells by reactive oxygen species and itscorrelation with their adaptive response to geno-toxicity of free radicals. Proc. Natl Acad. Sci. USA,95, 5061-5066.

Robson, C. N. & Hickson, I. D. (1991). Isolation ofcDNA clones encoding a human apurinic/apyrimi-dinic endonuclease that corrects DNA repair andmutagenesis defects in E. coli xth (exonuclease III)mutants. Nucl. Acids Res. 19, 5519-5523.

Sakoda, M. & Hiromi, K. (1976). Determination of thebest-®t values of kinetic parameters of the Michae-lis-Menten equation by the method of least squareswith the Taylor expansion. J. Biochem. 80, 547-555.

Seki, S., Hatsushika, M., Watanabe, S., Akiyama, K.,Nagao, K. & Tsutsui, K. (1992). cDNA cloning,sequencing, expression and possible domain struc-ture of human APEX nuclease homologous toEscherichia coli exonuclease III. Biochim. Biophys.Acta, 1131, 287-299.

Spee, J. H., de Vos, W. M. & Kuipers, O. P. (1993). Ef®-cient random mutagenesis method with adjustablemutation frequency by use of PCR and dITP. Nucl.Acids Res. 21, 777-778.

Studier, F. W. & Moffatt, B. A. (1986). Use of bacterio-phage T7 RNA polymerase to direct selective high-level expression of cloned genes. J. Mol. Biol. 189,113-130.

Takeshita, M., Chang, C. N., Johnson, F., Will, S. &Grollman, A. P. (1987). Oligodeoxynucleotides con-taining synthetic abasic sites. Model substrates for


DNA polymerases and apurinic/apyrimidinic endo-nucleases. J. Biol. Chem. 262, 10171-10179.

Uchiyama, Y., Miura, Y., Inoue, H., Ohtsuka, E., Ueno,Y., Ikehara, M. & Iwai, S. (1994). Studies of theinteractions between Escherichia coli ribonuclease HIand its substrate. J. Mol. Biol. 243, 782-791.

Urata, H. & Akagi, M. (1993). A convenient synthesis ofoligonucleotides with a 30-phosphoglycolate and 30-phosphoglycaldehyde terminus. Tetrahedron Letters,34, 4015-4018.

Walker, L. J., Robson, C. N., Black, E., Gillespie, D. &Hickson, I. D. (1993). Identi®cation of residues inthe human DNA repair enzyme HAP1 (Ref-1) thatare essential for redox regulation of Jun DNA bind-ing. Mol. Cell. Biol. 13, 5370-5376.

Wallace, S. S. (1994). DNA damages processed by baseexcision repair: biological consequences. Int. J.Radiat. Biol. 66, 579-589.

Wilson, D. M., III, Takeshita, M., Grollman, A. P. &Demple, B. (1995). Incision activity of human apuri-nic endonuclease (Ape) at abasic site analogs inDNA. J. Biol. Chem. 270, 16002-16007.

Weinfeld, M. & Soderlind, K. J. (1991). 32P-postlabelingdetection of radiation-induced DNA damage:identi®cation and estimation of thymine glycols andphosphoglycolate termini. Biochemistry, 30, 1091-1097.

Weinfeld, M., Lee, J., Ruiqi, G., Karimi-Busheri, F.,Chen, D. & Allalunis-Turner, J. (1997). Use of apostlabelling assay to examine the removal of radi-ation-induced DNA lesions by puri®ed enzymesand human cell extracts. Mutat. Res. 378, 127-133.

Xanthoudakis, S., Miao, G. G., Wang, F., Pan, Y. C. &Curran, T. (1992). Redox activation of Fos-Jun DNAbinding activity is mediated by a DNA repairenzyme. EMBO J. 11, 3323-3335.

Xanthoudakis, S., Miao, G. G. & Curran, T. (1994). Theredox and DNA-repair activities of Ref-1 areencoded by nonoverlapping domains. Proc. NatlAcad. Sci. USA, 91, 23-27.

Xanthoudakis, S., Smeyne, R. J., Wallace, J. D. & Curran,T. (1996). The redox/DNA repair protein, Ref-1, isessential for early embryonic development in mice.Proc. Natl Acad. Sci. USA, 93, 8919-8923.

Edited by P. E. Wright

(Received 11 August 1998; received in revised form 17 December 1998; accepted 18 January 1999)

intragenic suppression of an active site mutation in the human apurinic/apyrimidinic endonuclease

Documents