apyrimidinic endonuclease1

J. Mol. Biol. (1999) 287, 47±57

Article No. jmbi1999.2573 available online at http://www.idealibrary.com on

Intragenic Suppression of an Active Site Mutation in the Human Apurinic/Apyrimidinic Endonuclease Tadahide Izumi1,2, Jedrzej Malecki2, M. Ahmad Chaudhry3 Michael Weinfeld3, Jeff H. Hill1, J. Ching Lee2 and Sankar Mitra1,2* 1

Sealy Center for Molecular Science and 2Department of Human Biological Chemistry and Genetics, University of Texas Medical Branch Galveston, TX 775551079, USA 3 Experimental Oncology Cross Cancer Institute Edmonton, Canada T6G 1Z2

The apurinic/apyrimidinic endonucleases (APE) contain several highly conserved sequence motifs. The glutamic acid residue in a consensus motif, LQE96TK98 in human APE (hAPE-1), is crucial because of its role in coordinating Mg2‡, an essential cofactor. Random mutagenesis of the inactive E96A mutant cDNA, followed by phenotypic screening in Escherichia coli, led to isolation of an intragenic suppressor with a second site mutation, K98R. Although the Km of the suppressor mutant was about sixfold higher than that of the wild-type enzyme, their kcat values were similar for AP endonuclease activity. These results suggest that the E96A mutation affects only the DNA-binding step, but not the catalytic step of the enzyme. The 30 DNA phosphoesterase activities of the wildtype and the suppressor mutant were also comparable. No global change of the protein conformation is induced by the single or double mutations, but a local perturbation in the structural environment of tryptophan residues may be induced by the K98R mutation. The wild-type and suppressor mutant proteins have similar Mg2‡ requirement for activity. These results suggest a minor perturbation in conformation of the suppressor mutant enabling an unidenti®ed Asp or Glu residue to substitute for Glu96 in positioning Mg2‡ during catalysis. The possibility that Asp70 is such a residue, based on its observed proximity to the metal-binding site in the wild-type protein, was excluded by site-speci®c mutation studies. It thus appears that another acidic residue coordinates with Mg2‡ in the mutant protein. These results suggest a rather ¯exible conformation of the region surrounding the metal binding site in hAPE-1 which is not obvious from the X-ray crystallographic structure. # 1999 Academic Press

0

*Corresponding author

Keywords: DNA repair; AP endonuclease; 3 phosphoesterase; missense mutation; site-directed mutagenesis

Introduction Apurinic and apyrimidinic (AP) sites in DNA, generated either by spontaneous and oxidative depurination or as intermediate products during base excision repair in DNA (BER), are genotoxic and mutagenic (Wallace, 1994). AP endonucleases (APE) incise DNA strands 50 to the AP sites and Abbreviations used: AP site, apurinic/apyrimidinic site; BER, DNA base excision repair; DTT, dithiothreitol; EDTA, ethylenediaminetetra-acetic acid; hAPE-1, human apurinic/apyrimidinic endonuclease 1; MMS, methyl methanesulfonate; nfo, endonuclease IV; PBS, phosphate buffered saline; 30 -PG, 30 -phosphoglycolate; xth, exonuclease III; APE, apyrimidinic exonuclease. E-mail address of the corresponding author: [email protected] 0022-2836/99/110047±11 $30.00/0

produce 30 OH termini (Wallace, 1994; Doetsch & Cunningham, 1990). Such enzymes are distinguished from AP lyases which were originally classi®ed as type I and generate 30 phospho a,b unsaturated aldehyde via b-elimination (Doetsch & Cunningham, 1990). APEs also possess a 30 DNA phosphoesterase activity and thus remove abnormal 30 termini such as 30 -phosphoglycolate (30 -PG) which are generated by ROS at DNA strand breaks and prevent DNA repair synthesis (Doetsch & Cunningham, 1990). Escherichia coli has two APE genes, exonuclease III (Xth) and endonuclease IV (Nxfo), encoded by xth and nfo, respectively. The xth nfo double mutant is highly sensitive to such genotoxic agents as methyl methanesulfonate (MMS) and reactive oxygen species which lead to the production of AP sites and 30 -blocked termini # 1999 Academic Press

48 (Doetsch & Cunningham, 1990; Cunningham et al., 1986). Only one APE gene, APE-1 (originally named APEX, APE, HAP1, and Ref-1), has been identi®ed and cloned in mammalian cells; the protein is a homolog of E. coli Xth (Demple et al., 1991; Robson & Hickson; 1991; Seki et al., 1992; Xanthoudakis et al., 1992). This enzyme accounts for most of the cellular APE activity, and its activation in response to oxidative stress has been recently reported (Fung et al., 1998; Ramana et al., 1998). Besides the DNA repair function, the human APE-1 (hAPE-1) possesses two other unrelated functions. The protein was identi®ed as a redox-enhancing factor, Ref-1, which reductively activates AP-1 and p53 transcription factors (Xanthoudakis et al., 1992; Jayaraman et al., 1997). Indeed, the N terminus domain has recently been reported to interact with thioredoxin which appears to be needed for redox reactivation of the Ref-1 activity (Qin et al., 1996; Hirota et al., 1997). The hAPE-1 was also identi®ed as a corepressor of the parathyroid hormone gene and of its own gene (Okazaki et al., 1994; Izumi et al., 1996). Although the biological signi®cance of the multiple, unrelated properties of hAPE-1 is not yet clear, the fact that the gene is essential for the early embryonic development in mice (Xanthoudakis et al., 1996) indicates its essential functions in cells. Distinct regions in the hAPE-1 polypeptide are responsible for the endonuclease/30 -phosphoesterase and gene regulatory activities (Xanthoudakis et al., 1994). The 60 amino acid residues at the N terminus, which are not conserved in Xth, are involved in AP-1 activation and dispensable for the APE activity (Walker et al., 1993; Izumi & Mitra, 1998). Further truncation of the protein from both N and C termini (80 and ®ve amino acid resi-

Intragenic Suppression of Human APE-1

dues, respectively) led to total loss of the APE activity (Izumi & Mitra, 1998). Alignment of the amino acid sequences of the members of the APE family shows several highly conserved motifs (Seki et al., 1992). One of these is the GXDHCP sequence in which the histidine residue was identi®ed as the catalytic residue (Figure 1). Another motif, LQETK, is also conserved in most APE enzymes (Figure 1). Indeed, elucidation of the tertiary structure of hAPE-1 and Xth by X-ray crystallography led to the prediction that Glu96 (E96) in the motif LQE96TK98 in hAPE1 is essential for endonuclease activity because of its role in coordinating the cofactor Mg2‡ (Barzilay et al., 1995a,b; Gorman et al., 1997). Substitution of E96 with alanine (E96A) caused loss of the endonuclease activity. However, a small activity of E96A was observed in the E96A mutant protein (about 400 times less than that of the wild-type) in contrast to other active site mutants, such as H309N, of which the activity was at least 2000-fold less than that of the wild-type protein (Barzilay et al., 1995a,b). Because the metal ion cofactor is absolutely required for the reaction (Gorman et al., 1997), we considered it likely that other amino acid residues were also involved in the binding of Mg2‡ with six coordinate bonds, and thus may be responsible for the residual activity of the E96A mutant. Participation of multiple residues in the metal binding was also reported in E. coli RNase H (Kashiwagi et al., 1996; Uchiyama et al., 1994). We entertained the possibility that a second missense mutation in E96A could induce a subtle conformational change in the neighborhood of the active site surface that would allow positioning of other residues in order to coordinate Mg2‡ more ef®ciently, and thereby restore the catalytic activity of the protein. Such a mutant protein could pro-

Figure 1. Homology of polypeptide sequences of APEs. Only the homology of residues 61 to 99 and the C-terminal segment of hAPE-1 are shown. Sequences in Swiss-plot database were aligned with GCG 9.0. Names on the left are the entry names in Swiss-Plot. The numbers in the second column denote the starting residue number. The AP endonuclease activity in hAPE-1 is lost after deletion of the shaded regions (Izumi & Mitra, 1998). The E96 and K98 residues of hAPE-1 and the two highly conserved motifs are shown in bold letters.

Intragenic Suppression of Human APE-1

49

vide signi®cant insight into the nature of metal binding to the essential DNA repair enzyme in mammals. Here, we report isolation and characterization of such a missense suppressor of the E96A mutant which restored the AP endonuclease and 30 -PG-removing activity. Our results support the model of participation of multiple acidic residues in coordinating Mg2‡ and thereby the presence of a ¯exible conformation of the protein.

Results Isolation of intragenic suppressor mutant of E96A; phenotypic rescue of E. coli by hAPE-1 APE is involved in the repair of DNA damage induced by MMS, because AP sites are produced as intermediates during repair process (Cunningham et al., 1986). The human APE-1 was shown to be able to complement xth nfo-negative E. coli (Demple et al., 1991; Robson & Hickson, 1991; Izumi & Mitra, 1998). We have recently established a phenotypic rescue system in E. coli that provides a sensitive screening procedure for identifying active versus inactive hAPE-1 (Izumi & Mitra, 1998).

Figure 2. (a) Site-directed mutagenesis of the hAPE-1 by PCR. The 50 and 30 vector primers ¯anking the cDNA and two primers for each mutagenesis are shown. Newly introduced EcoRV (E96A) and NsiI (H309A) sites, as well as an endogenous site (BglII for D70N), are underlined. Silent mutations are shown in bold letters. After ampli®cation by PCR, the fragments were inserted into the NdeI-SalI site in pIZ42 (Izumi & Mitra, 1998). (b) A diagram of the random mutagenesis procedure. The same vector primers as in (a) were used.

Figure 3. Phenotypic rescue of E. coli BW528 with wild-type (W) and mutant APEs. Details of MMS treatment were given in Materials and Methods. Vec, the empty vector control.

We examined APE activity of two mutant proteins, E96A and H309A, in the cell toxicity assay (Figures 2 and 3). Both cDNAs failed to provide resistance to the E. coli strain against MMS. No activity was detected in either crude extracts (Figure 4(b)). We thus concluded that the E96A as well as H309A missense mutants were inactive (Barzilay et al., 1995b). Then we utilized the phenotypic rescue strategy to examine the possibility of intragenic suppression of E96A. An expression plasmid cDNA containing E96A was randomly mutagenized by PCR (Spee et al., 1993), and introduced into xthÿ nfoÿ E. coli by transformation. Resistant clones were isolated after challenging the bacteria with MMS treatment (Figure 2). Our preliminary experiments showed that the original protocol by Spee et al. (1993) yielded high frequency of multiple mutations in control experiments (data not shown). Because multiple mutations are undesirable for our purpose, we modi®ed the protocol by omitting dITP and reducing the concentration of one dNTP to 40 mM, while maintaining the concentration of the other three dNTPs at 200 mM (Spee et al., 1993). Four independent PCR reactions with limiting concentration of different dNTPs were carried out so that misincorporation could occur at all sites. Of the approximately 2  104 independent clones screened, 16 clones were selected on the basis of MMS resistance of the host bacteria. A high level of cellular protection was observed in the selected clones compared to the bacteria harboring the E96A mutant plasmid (Figure 3). The DNA sequence was determined for four of them, and in all cases Lys98 was found to

50

Figure 4. AP endonuclease activity of wild-type and mutant proteins of hAPE-1. (a) Immunoblot assay of wild-type, E96A, and E96A K98R expressed in E. coli BW528. The same amount of crude extracts was analyzed by SDS-PAGE (12 % acrylamide) and subsequent immunoblot assay (Izumi & Mitra, 1998). Total and soluble fractions were analyzed separately. (b) AP endonuclease activity was measured in crude lysates using a tetrahydrofuran-containing 43mer oligonucleotide.

be replaced with an arginine residue (K98R, AAA to AGA), while the original missense mutation remained unchanged. When expressed in E. coli,

Intragenic Suppression of Human APE-1

the level of expression and solubility of the mutant protein were indistinguishable from those of the wild-type (Figure 4(a)), suggesting that no signi®cant change in expression of the protein or in its global conformation was caused by the mutation; this was further con®rmed by circular dichroism and intrinsic ¯uorescence studies of the wild-type and mutant proteins as described below. AP endonuclease activity in the crude extract of E. coli expressing the suppressor mutant protein was found to be about half that in bacteria expressing the wild-type protein (Figure 4(b)). We have concluded therefore that the E96A K98R double mutant regained the AP endonuclease activity despite the loss of E96, the residue predicted to be essential for Mg2‡ coordination (Gorman et al., 1997; Barzilay et al., 1995b). To examine if such intragenic suppression is a common phenomenon among the APEs, the corresponding glutamic acid (E34) and lysine (K36) residues in the LQETK motif of the E. coli Xth protein were changed to Ala and Arg, respectively, by sitespeci®c mutagenesis and the mutants were tested in the cell protection assay. While the E34A mutant was inactive in cell survival assay as expected (Figure 5(a)), the defect was again suppressed by introduction of the secondary missense, K36R, although the effect was not as pronounced as that observed for hAPE-1. Characterization of the mutant protein and the effect of Mg2‡ on incision activity To further characterize the suppressor mutant on hAPE-1, we puri®ed the wild-type and the mutant proteins to near homogeneity (Figure 6(a)). The E. coli xth nfo strain (BW528) was lysogenized with l-DE3, a lambda phage containing the T7 RNA polymerase gene inducible by IPTG. Thus, the puri®ed APE protein did not contain any background

‰ Figure 5. Phenotypic rescue of E. coli BW528 with wild-type (W) and mutant APEs. (a) E. coli BW528 with wild-type xth gene, E34A, and E34A K36R in pIZ42 were treated with MMS. (b) hAPE-1 cDNA with missense mutations were subjected to phenotypic rescue assay. W, wild-type hAPE-1; Vec, control empty vector.

Intragenic Suppression of Human APE-1

51

Figure 6. Analysis of the puri®ed wild-type and missense mutant proteins. (a) Coomassie blue staining of puri®ed wild-type (W), and mutant proteins after SDS-PAGE. (b) Mg2‡ dependency for the AP endonuclease activity of the wild-type and mutant proteins. The 43mer substrate oligonucleotide was incubated with 20 pg of APEs for one minute at 37 C. The activity in 3 mM Mg2‡ was used for normalization. (c) Release of 30 -PG by wild-type and suppressor mutant APE-1. The indicated amounts of irradiated DNA were incubated with the enzymes and the residual 30 -PG was 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, 10 and 13) of hAPE-1 (lanes 2-4, wild-type; lanes 5-7, E96A; lanes 8-10, K98R; lanes 11-13, E96A K98R). Lane 14, a mixture 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-cleavage assay was highly speci®c for the endonuclease activity of the recombinant APEs because of the use of tetrahydrofuran as a substrate. This substrate, unlike an intact AP site, is resistant to cleavage via b elimination by contaminating AP lyases (Takeshita et al., 1987). Barzilay et al. (1995b) showed that although the E96A mutant lost activity in low concentration of Mg2‡ (0.1 mM), the activity was restored when the Mg2‡ concentration in the reaction was increased to 2 mM. We examined the effect of Mg2‡ concentration on the activity of the mutant protein. As shown in Figure 6(b), the relative activity of the suppressor mutant protein in different concentrations of Mg2‡ was essentially the same as that of the wild-type protein. Increasing the concentration of Mg2‡ to more than 1 mM did not affect the activity of the wild-type and the suppressor mutant protein, whose activities were comparable (Figure 6(b)). The activity of E96A was approximately 1000 times lower than that of the active proteins (data

not shown). Thus, our results indicate that the double mutant protein regained a metal-coordinating motif which functions nearly as ef®ciently as the wild-type enzyme. We then inquired whether the missense mutation affected the kinetic properties of the enzyme (Table 1). The Km value of the suppressor mutant was found to be about sixfold higher than that of the wild-type protein, which in turn was comparable to the reported value (Wilson et al., 1995). However, the kcat of the wild-type protein and the mutant were comparable. The 30 -PG removing activity of the suppressor mutant protein We used a postlabeling assay (Weinfeld & Soderlind, 1991; Weinfeld et al., 1997) to examine the enzymatic activity for removal of 30 -PG, a 30 blocking damage in DNA that prevents DNA synthesis by DNA polymerases. The mutant protein removed the 30 -PG residues from g-irradiated DNA as ef®ciently as the wild-type protein with

52

Intragenic Suppression of Human APE-1 Table 1. Kinetic parameters for AP endonuclease activity of wild-type and suppressor mutant proteins Wild-type E96A K98R Wild-type mutant

Km (nM)

kcat (sÿ1)

11.9  3.6 69.5  9 0.17

1.80  0.14 1.76  0.1 1.02

three different concentrations of the substrate DNA (Figure 6(c)). Then we examined the 30 -PG-removing activity with a 13mer oligonucleotide containing a PG at the 30 end (Figure 6(d)). The activity was observed in wild-type, K98R, and E96A K98R proteins with the latter two exhibiting approximately 80 % activity of the wild-type protein, whereas the activity of E96A was undetectable (Figure 6(d)). Taken together, it is evident that the second mutation of K98R restored the DNA 30 phosphoesterase activity as well as the endonuclease activity of the E96A mutant. Asp70 is dispensable in the intragenic suppression The recently reported X-ray crystallographic structure of hAPE-1 shows that Asp70 is located in the vicinity of the metal-binding locus (Gorman et al., 1997; Figure 1(b)). Because the Mg2‡ is absolutely required for activity of the E96A K98R protein like that of the wild-type protein, we considered the possibility that the position of Asp70 was shifted closer to the metal-binding site in the suppressor mutant so as to hold Mg2‡ stably even in the absence of the E96. In order to test this possibility we created a triple-missense mutant, D70N E96A K98R, to examine its activity using the cell survival assay (Figure 5(b)). The sensitivity of E. coli expressing D70N E96A K98R triple mutant hAPE-1 was similar to that expressing the E96A K98R double mutant. Moreover, the single missense mutant, D70N, was as active as the wildtype in this assay (Figure 5(b)). These results indicate a lack of involvement of D70 in suppressing the phenotype of E96A K98R, or the catalytic function of the wild-type.

kcat/Km 0.151 0.025 6.04

teins are not due to a net change in their secondary structure. CD spectroscopy provides global structural information and may not be able to detect the structural alterations that are subtle but may cause signi®cant functional changes. Therefore, intrinsic tryptophan ¯uorescence was measured to probe for small local perturbations around the various tryptophan residues (Figure 7). It is interesting to note that the emission intensity was altered in the mutant without a detectable change in the shape of the emission spectra. The K98R and E96A K98R mutants exhibited identical emission spectra with lower intensity as compared to those of the wildtype and E96A mutant. The latter two proteins exhibited identical emission spectra. These results indicate that the K98R mutation exerts an effect on the structural environment surrounding the tryptophan residues of hAPE-1. The decrease in emission intensity without a change in the shape of the emission spectra implies that the net effect of the mutation is a decrease in quantum yield of one or more tryptophan residues. A decrease in quantum yield could be the consequence of an increase in collisional quenching of the ¯uorophore by the solvent or a decrease in the apolar nature of the environment surrounding the ¯uorophore or both.

Investigation of the secondary structure of hAPE-1 with circular dichroism and fluorescence spectroscopic analysis To test whether any signi®cant change in the secondary structure occurred as a result of mutations, the CD spectra of the wild-type and mutant hAPE1 were obtained. The mutants included in this investigation were E96A, K98R and the double mutant of E96A K98R. Regardless of the nature of the mutation, the far-UV CD spectra of these mutants were identical with that of wild-type hAPE-1 (data not shown). These results indicate that these mutations do not signi®cantly perturb the secondary structure of hAPE-1 and that the loss and restoration of activity of the mutant pro-

Figure 7. Fluorescence spectroscopy. Emission spectra of the wild-type (*) and mutant hAPE-1 proteins (*, E96A; ~, K98R; &, E96A K98R) were measured as described in Materials and Methods. The excitation wavelength was 295 nm.

Intragenic Suppression of Human APE-1

It is interesting to note that the crystallographic structure of hAPE-1 shows that K98 interacts with D70 (Gorman et al., 1997) which is bracketed by W67 and W75. Hence, it is possible that the changes in ¯uorescence of these tryptophan residues resulted from K98 mutation.

Discussion Our observation that the K98R missense mutation suppressed the inactivating E96A mutation has important implications in the structure-function relationship of hAPE-1, because this suggests the presence of an alternative active conformation of the APE polypeptide. That this possibility is not unique to the human APE was supported by similar results obtained with the E. coli Xth protein, although the effect for Xth was less striking than that for hAPE-1. Moreover, restoration of both AP endonuclease and 30 -phosphoesterase activities in the double mutant indicates that intragenic suppression restores a fully functional active site. The E96A K98R suppressor mutant and the wild-type protein have similar kcat values, although the mutant has a higher Km than the wild-type protein. We may explain the results by assuming that the double mutant as well as the inactive E96A mutant retain an intact catalytic domain, and the loss of the E96 results in its inef®cient recognition of or binding to the damaged DNA site. How substitution of Lys98 by an Arg residue restores the DNA binding activity is not obvious. One simple interpretation could be that Mg2‡ is coordinated by a water molecule held in space by hydrogen bonding with Arg98 in the suppressor mutant. Thus the requirement for direct interaction with Glu96 is eliminated. However, the possibility is unlikely because of the lack of acidic amino acid residues involved in the metal coordination with Mg2‡. Apparently E96 is the only acidic residue found in the metal coordination (Gorman et al., 1997). Other studies indicate that at least one or two acidic residues are involved for the metal binding (Mol et al., 1995; Kashiwagi et al., 1996; Kostrewa et al., 1996). Another possibility is that Arg, being a stronger base than Lys, may eliminate the need for Mg2‡ in holding the DNA in place. However, this possibility was also eliminated by our observation that the Mg2‡ requirement was nearly identical for the wild-type and the double mutant. The role of Mg2‡ in the catalytic activity of APE is not completely understood. The hAPE1 protein has three distinct enzymatic activities, namely AP-endonuclease, 30 phosphoesterase and RNase H, which appear to have the same active site 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 in Xth) after soaking the crystals in the metal salt solution suggested a single site binding of the

53 metal ion to the protein in the absence of DNA (Mol et al., 1994; Gorman et al., 1997). We utilized atomic absorption spectroscopy to quantify the amount of Mg2‡ bound to the wild-type, E96A, K98R, and E96A K98R mutant proteins after equilibrium dialysis of 10 mM enzyme solutions against 10 mM Mg2‡. No binding was detected for any of the proteins (data not shown), while a signi®cant activity could be observed at 10 mM Mg2‡ for the active enzymes (Figure 6(b)). A similar lack of Mg2‡ binding to the wild-type enzyme has been reported (Barzilay et al., 1995b). Our results support the earlier conclusion that Mg2‡ does not have a signi®cant structural role in hAPE-1 (Barzilay et al., 1995b; Gorman et al., 1997). The structural studies of E. coli RNase H, whose catalytic mechanism involves an Asp-His pair like that of hAPE-1, suggest that Mg2‡ forms a hexa-coordinated complex that includes an ionic bond with a Glu residue (Kashiwagi et al., 1996; Uchiyama et al., 1994). While the overall mechanism proposed for RNase H is similar to that proposed by Barzilay et al. (1995b) for hAPE-1 in regard to the role of activated H2O in phosphodiester bond cleavage, the binding of Mg2‡ to the single side-chain of E96 in hAPE-1 has been proposed only by the latter group. Site-speci®c mutation studies have clearly established the role of E96 in Mg2‡ binding. However, based on the present study and analogy from the RNase H study, it appears likely that another acidic residue is involved in Mg2‡ binding. Barzilay et al. (1995b) offered several possible explanations for the presence of low residual activity of the E96A mutant protein. Here, we show that the af®nity for Mg2‡ in the absence of DNA was not apparently altered in the single and double mutants. Furthermore, the Mg2‡ requirement was not affected by the suppressor mutations. These results are consistent with the involvement of a second acidic residue in the protein 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 the divalent cation binding site in the original motif (Gorman et al., 1997) and is likely to be affected by a subtle change in the polypeptide conformation. We have, however, excluded this possibility, because (i) we have shown that the triple-missense mutant D70N E96A K98R is as active as the double mutant (Figure 5(b)) and (ii) this residue is not conserved in other APEs. An Asn residue, for example is present in the corresponding site of the Xth protein (Gorman et al., 1997). Furthermore, a single missense mutant, D70N, showed the same level of protection of E. coli cells as the wild-type (Figure 5(b)). Thus, we have yet to identify the residue(s) involved in restoration of the activity in the suppressor mutant. D308 in hAPE-1 is another candidate for substituting the E96 residue (Gorman et al., 1997), although it is not as close to the metal ion as E96 and D70 as determined from the X-ray

54 crystallographic structure of the wild-type protein (Gorman et al., 1997). In fact, introduction of a missense mutation at D308 (D308A) caused a 25-fold reduction in the endonuclease activity and also affected the preference for Mn2‡ (Barzilay et al., 1995b). However, it would be dif®cult to use our approach to show that the residue is indeed involved in the metal binding. Because D308 is next to the catalytic residue H309, substitution of the original residue could affect the nucleolytic activity itself but not the metal coordination. Rather, a direct approach in this case would be X-ray crystallographic analysis with the double mutant protein to determine the distance of the D308 relative to the metal ion. Finally, our study has implications regarding structure-function relationship of enzymes beyond the speci®c case of hAPE-1. Our results predict that alternative conformations of proteins with identical catalytic function can exist which could not be predicted from the X-ray crystallographic structure of these proteins. It is interesting to note that the tertiary structure of the active site of RNase H, particularly of the loop containing the active site residue His, was found to be ¯exible (Kashiwagi et al., 1996). We propose that hAPE-1 and possibly many other enzymes have ¯exible conformation around the active site, although one conformation predominates in the wild-type protein. In the case of hAPE-1, an alternative conformation is favored in the K98R mutant of hAPE-1 over that present in the wild-type protein which allows positioning of a second, unidenti®ed acidic residue to hold Mg2‡ and to help stabilize the ternary complex without signi®cant distortion of the active site pocket. Many DNA repair enzymes are active on multiple substrates with signi®cantly divergent structures (Krokan et al., 1997; Hang et al., 1996). These substrates may be recognized by distinct conformers of the enzyme. One way to test this prediction is by elucidating the structure of cocrystals of substrate-enzyme-Mg2‡ ternary complex of APE-1. Such structural studies will directly identify 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 were kindly provided by Dr B. Weiss (Cunningham et al., 1986). The human APEX cDNA was originally a gift from Dr S. Seki, and was manipulated further for the present study (Seki et al., 1992; Izumi & Mitra, 1998). Missense mutations in the hAPE-1 cDNA were generated by PCR using Pfu DNA polymerase (Stratagene; Figure 2). Primers for construction of missense mutants of Xth protein are: E34A, 50 GGCCTGCAGGCGACAAAAGTTCATGAC30 ; and for E34A K36R, 50 GGCCTGCAGGCGACACGCGTTCATGACGAT30 . The PstI sites are underlined and changed codons are in bold letters. In all cases, DNA fragments ampli®ed by PCR were digested with appropriate enzymes and inserted

Intragenic Suppression of Human APE-1 into the original vector at the corresponding sites. All mutations in DNA generated by PCR were con®rmed by DNA sequencing. Phenotypic rescue (survival) studies The hAPE-1 cDNA used for the rescue experiments was cloned into pIZ42 (Izumi & Mitra, 1998) to express intact proteins and did not contain a His-tag region. Plasmid-bearing E. coli BW528 (xth nfo) was grown overnight at 28 C, diluted 25-fold in fresh LB medium and incubated further at 28 C until A600 reached about 0.6. The cells were centrifuged, washed twice in phosphatebuffered saline (PBS, pH 7.3), and resuspended in an equal volume of PBS. Immediately after addition of various amounts of MMS, cells were incubated at 37 C with shaking for one hour and then diluted in PBS and plated on LB-agar at 28 C for colony counting (Izumi & Mitra, 1998). All the rescue experiments were repeated reproducibly more than twice. Measurement of AP-endonuclease activity in E. coli crude extracts A 43mer oligonucleotide containing a single tetrahydrofuran residue (Glen Research), an AP siteanalogue, was synthesized (Izumi & Mitra, 1998). The tetrahydrofuran-containing oligonucleotide was annealed with its complementary strand, puri®ed by non-denaturing 20 % PAGE, labeled at its 50 end with [g-32P]ATP by T4 polynucleotide kinase (Pharmacia), and used for the incision assay. Crude extracts (10 ng) of E. coli were mixed with about 50 fmol of the oligonucleotide in a reaction mixture (20 ml) containing 60 mM Tris (pH 8.0), 1 mM MgCl2, 0.2 mM EDTA, 1 mM DTT, and 100 mg/ml bovine serum albumin (BSA). After incubation for ten minutes at 37 C, the reaction was stopped by the addition of a stop buffer (88 % formamide, 0.5 % bromophenol blue and xylene cyanol). The product was quanti®ed by analysis in a PhosphorImager 425 (Molecular 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-1 cDNA using a PCR-based method (Spee et al., 1993). Brie¯y, four PCR reactions were carried out in each set in which about 0.1 ng of the cDNA was ampli®ed using two vector primers (Figure 2) with
Taq polymerase (Amersham) and 200 mM of each dNTP except for one reduced to 40 mM. The ampli®ed fragment was cleaved by NdeI and SalI, and then cloned into pIZ42 (Figure 2). The plasmid DNA was transformed into E. coli BW528 and resistant clones were screened by serial exposure to MMS, as described above. Purification of the wild-type and missense mutant proteins E. coli strain BW528 (xth nfo) was lysogenized with DE3 l-phage using the l-phage lysogenization kit (Novagen). The hAPE-1 cDNAs were introduced into pET15b plasmid vector (Novagen) using its NdeI/XhoI sites in order to express the proteins with the T7 RNA polymerase system (Studier & Moffatt, 1986). Each construct carried 20 extra amino acid residues at the N terminus containing a histidine hexamer, which allowed

55

Intragenic Suppression of Human APE-1 puri®cation of the protein by af®nity chromatography on a nickel column. E. coli BW528(DE3) carrying the cDNAs were grown in a fermenter at 37 C, and 0.5 mM IPTG was added when A600 reached 1.5, and incubated further at 25 C for four hours. Cells were harvested and resuspended in buffer A (20 mM Tris-Cl (pH 8.0), 300 mM NaCl) and lysed with a French press. The extracts were applied on 10 ml TALON (Clontech), washed with 100 ml of buffer A, and eluted with buffer A containing 10 mM imidazole. The eluents were then applied onto 3 ml of Ni-NTA resin (Qiagen), washed with 30 ml of buffer A containing 40 mM imidazole, and eluted with 6 ml of buffer A containing 200 mM imidazole. Proteins were 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 mM DTT. Fractions containing the hAPE-1 proteins were collected and stored at ÿ80 C.

ethanol. Then the oligomer was incubated in 0.5 M NaClO2 in the presence of 36 % dimethyl sulfoxide at room temperature for ®ve hours to produce 30 -PG. The 30 -PG containing the 13mer oligo was annealed with 28mer complementary sequence (50 CATGCGGTGCAGAAGTGAAACTGAGGAG30 ) 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, approximately 25 fmol, was incubated at 37 C for ®ve minutes in 10 ml of reaction buffer (20 mM Tris (pH 8.0), 100 mM NaCl, 1 mM MgCl2, 0.1 mM EDTA, 1 mM DTT, 100 mg/ml BSA) with puri®ed hAPE-1 protein. The reaction was stopped by the addition of the stop buffer. The products were analyzed in 20 % acrylamide gel containing 7 M urea after boiling and by subsequent exposure to PhosphorImager cassette (Molecular Dynamics).

Characterization of purified enzymes

The secondary structure of the wild-type and mutant hAPE-1 in 50 mM potassium phosphate (pH 7.5), 0.5 mM NaCl was monitored with an AVIV 62 DS circular dichroism spectropolarimeter. The far-UV CD spectra were obtained in fused quartz cuvettes with 0.1 cm path length and protein solutions with absorbance at 280 nm ranging from 0.20 to 0.26. Each spectrum was recorded with a 0.5 nm increment and one second interval. For each sample, ®ve repetitive scans were obtained and averaged.

The Km and kcat values of the puri®ed enzymes were determined using the same 43mer duplex DNA as described above. The substrate at various concentration (Figure 6(d)) was incubated with 50 pg of protein at 37 C in a reaction buffer (66 mM Tris-Cl (pH 8.0), 1 mM Mg2‡, 100 mM NaCl, 1 mM DTT, and 0.1 mg/ml BSA) and the reaction was stopped by the addition of the stop buffer. The reaction was analyzed on PhosphorImager and values were calculated by a method described by Sakoda & Hiromi (1976). The effect of Mg2‡ on endonuclease activity was examined by using the same reaction buffer with 20 pg of puri®ed proteins and different amounts of MgCl2. The products were analyzed by PhosphorImager 425 (Molecular Dynamics) after separation by electrophoresis in denaturing 20 % polyacrylamide gel containing 7 M urea. Measurement of residual phosphoglycolate Calf thymus DNA (0.5 mg/ml in 10 mM sodium phosphate, pH 7.4) was irradiated with 50 Gy g-rays (60Co Gammacell, AECL, Ottawa, ON). This introduces approximately 1 pmol of phosphoglycolate per 1 mg of DNA (Weinfeld & Soderlind, 1991). After precipitation, the DNA (0.33, 0.1 and 0.033 mg/ml) was incubated at 37 C for one hour with 300 ng of either the wild-type hAPE-1 or the E96A K98R mutant in 30 ml of buffer containing 60 mM Tris-Cl (pH 8.0), 100 mM NaCl, 1 mM MgCl2 and 1 mM DTT. The removal of phosphoglycolate from 1 mg of DNA from each of these reactions was then determined by a postlabeling assay (Weinfeld & Soderlind, 1991). 0

Measurement of 3 -PG removal from an oligonucleotide substrate A 13mer oligonucleotide with a 30 -phosphoglyceryl terminus was synthesized in UTMB's Recombinant DNA Laboratory (RDL) with glyceryl CPG (control pore glass; Glen Research). The sequence was 50 CCTCAGTTTCACT30 . The 30 -PG at the end was prepared as described (Urata & Akagi, 1993). Brie¯y, the glyceryl CPG-containing oligo was incubated in 5 mM NaIO4 (pH 6) on ice for two hours to produce 30 -phosphoglycolaldehyde. The reaction was stopped by adding L-methionine, followed by precipitation with ethanol and washing with 70 %

Circular dichroism spectra analysis

Fluorescence spectroscopy The structure of the wild-type and mutant hAPE-1 was also monitored by determining the emission spectra with a Perkin Elmer LS-50 ¯uorimeter. The excitation wavelength was 295 nm. The emission spectra were acquired by scanning between 300 and 400 nm. Four repetitive scans were obtained and averaged. The spectra were normalized to the same protein concentration. Atomic absorption spectroscopy The amount of magnesium bound to wild-type and mutant hAPE-1 proteins was measured by ¯ame atomic absorption spectrophotometry using a Perkin Elmer model 5100 Zeeman spectrometer. The protein samples were 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 mM MgCl2. The absorption was measured from a magnesium hollow 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 Recombinant DNA Laboratory (RDL) in UTMB using T7 and T3 universal primers and internal primers in the hAPE-1 cDNA synthesized in RDL.

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

56 K. Morikawa are greatly appreciated. We also thank Dr T. Wood of the Recombinant DNA Laboratory at UTMB for DNA sequencing and oligonucleotide synthesis, and Dr N. Alcock for atomic absorption spectroscopic analysis. We thank Dr R. P. Hodge for advice on chemical rearrangement of oligonucleotide. Dr D. Konkel's editorial help and Ms W. Smith's secretarial assistance are gratefully acknowledged. This work was supported by the U.S. Public Health Science grants CA53791 and ES08457 (to S.M.); by a grant from the National Cancer Institute of Canada with funds from the Terry Fox Run (to M.W.); also supported by NIH grant GM-45579, Robert A. Welch Foundation grants H-0013 and H-1238 (to J.C.L.) and by NIEHS Center grant ES06676. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ``advertisement'' in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

References Barzilay, G., Walker, L. J., Robson, C. N. & Hickson, I. D. (1995a). Site-directed mutagenesis of the human DNA repair enzyme HAP1: identi®cation of residues important for AP endonuclease and RNase H 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 residues in the multifunctional human DNA repair enzyme HAP1. Nature Struct. Biol. 2, 561-568. Cunningham, R. P., Saporito, S. M., Spitzer, S. G. & Weiss, B. (1986). Endonuclease IV (nfo) mutant of Escherichia coli. J. Bacteriol. 168, 1120-1127. Demple, B., Herman, T. & Chen, D. S. (1991). Cloning and expression of APE, 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. Doetsch, P. W. & Cunningham, R. P. (1990). The enzymology 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 mammalian AP-endonuclease gene expression, protein levels, and enzyme activity in mesothelial cells. Cancer Res. 58, 189-194. 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., Chenna, A., Fraenkel-Conrat, H. & Singer, B. (1996). An unusual mechanism for the major human apurinic/apyrimidinic (AP) endonuclease involving 50 cleavage of DNA containing a benzene-derived exocyclic adduct in the absence of an AP 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 is regulated by a direct association between thioredoxin and Ref-1. Proc. Natl Acad. Sci. USA, 94, 36333638. Izumi, T. & Mitra, S. (1998). Deletion analysis of human AP-endonuclease: minimum sequence required for

Intragenic Suppression of Human APE-1 the endonuclease activity. Carcinogenesis, 19, 525527. Izumi, T., Henner, W. D. & Mitra, S. (1996). Negative regulation of the major human AP-endonuclease, a multifunctional protein. Biochemistry, 35, 1467914683. Jayaraman, L., Murthy, K. G., Zhu, C., Curran, T., Xanthoudakis, S. & Prives, C. (1997). Identi®cation of redox/repair protein Ref-1 as a potent activator of p53. Genes Dev. 11, 558-570. Kashiwagi, T., Jeanteur, D., Haruki, M., Katayanagi, K., Kanaya, S. & Morikawa, K. (1996). Proposal for new catalytic roles for two invariant residues in Escherichia coli ribonuclease HI. Protein Eng. 9, 857867. Kostrewa, D. & Winkler, F. K. (1995). Mg2‡ binding to the active site of EcoRV endonuclease: a crystallographic study of complexes with substrate and product DNA at 2A resolution. Biochemistry, 34, 683696. Krokan, H. E., Standal, R. & Slupphaug, G. (1997). DNA glycosylases 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 function of the multifunctional DNA-repair enzyme exonuclease 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, is involved in negative gene regulation by extracellular calcium. J. Biol. Chem. 269, 27855-27862. Qin, J., Clore, G. M., Kennedy, W. P., Kuszewski, J. & Gronenborn, A. M. (1996). The solution structure of human thioredoxin complexed with its target from Ref-1 reveals peptide chain reversal. Structure, 4, 613-620. Ramana, C. V., Boldogh, I., Izumi, T. & Mitra, S. (1998). Activation of apurinic/apyrimidinic endonuclease in human cells by reactive oxygen species and its correlation with their adaptive response to genotoxicity of free radicals. Proc. Natl Acad. Sci. USA, 95, 5061-5066. 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. Sakoda, M. & Hiromi, K. (1976). Determination of the best-®t values of kinetic parameters of the Michaelis-Menten equation by the method of least squares with 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 structure of human APEX nuclease homologous to Escherichia 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 adjustable mutation frequency by use of PCR and dITP. Nucl. Acids Res. 21, 777-778. Studier, F. W. & Moffatt, B. A. (1986). Use of bacteriophage T7 RNA polymerase to direct selective highlevel expression of cloned genes. J. Mol. Biol. 189, 113-130. Takeshita, M., Chang, C. N., Johnson, F., Will, S. & Grollman, A. P. (1987). Oligodeoxynucleotides containing synthetic abasic sites. Model substrates for

Intragenic Suppression of Human APE-1 DNA polymerases and apurinic/apyrimidinic endonucleases. J. Biol. Chem. 262, 10171-10179. Uchiyama, Y., Miura, Y., Inoue, H., Ohtsuka, E., Ueno, Y., Ikehara, M. & Iwai, S. (1994). Studies of the interactions between Escherichia coli ribonuclease HI and its substrate. J. Mol. Biol. 243, 782-791. Urata, H. & Akagi, M. (1993). A convenient synthesis of oligonucleotides 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 in the human DNA repair enzyme HAP1 (Ref-1) that are essential for redox regulation of Jun DNA binding. Mol. Cell. Biol. 13, 5370-5376. Wallace, S. S. (1994). DNA damages processed by base excision 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 apurinic endonuclease (Ape) at abasic site analogs in DNA. J. Biol. Chem. 270, 16002-16007.

57 Weinfeld, M. & Soderlind, K. J. (1991). 32P-postlabeling detection of radiation-induced DNA damage: identi®cation and estimation of thymine glycols and phosphoglycolate termini. Biochemistry, 30, 10911097. Weinfeld, M., Lee, J., Ruiqi, G., Karimi-Busheri, F., Chen, D. & Allalunis-Turner, J. (1997). Use of a postlabelling assay to examine the removal of radiation-induced DNA lesions by puri®ed enzymes and 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 DNA binding activity is mediated by a DNA repair enzyme. EMBO J. 11, 3323-3335. Xanthoudakis, S., Miao, G. G. & Curran, T. (1994). The redox and DNA-repair activities of Ref-1 are encoded by nonoverlapping domains. Proc. Natl Acad. Sci. USA, 91, 23-27. 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 P. E. Wright (Received 11 August 1998; received in revised form 17 December 1998; accepted 18 January 1999)