On the Possibilities and Limitations of Rational Protein Design to Expand the Speciﬁcity of Restriction Enzymes: a Case Study Employing Ecorv As the Target

Protein Engineering vol.13 no.4 pp.275–281, 2000

On the possibilities and limitations of rational protein design to expand the speciﬁcity of restriction enzymes: a case study employing EcoRV as the target

Thomas Lanio, Albert Jeltsch and Alfred Pingoud1 a 6 base pair (bp) recognition site (Roberts and Macelis, 1999). The biological function of restriction enzymes is to cleave Institut fu¨r Biochemie, FB 08, Justus-Liebig-Universita¨t, Heinrich-Buff-Ring phage DNA invading a bacterium and thereby to protect the 58, D-35392 Giessen, Germany. bacterial cell from infection. As palindromic sites compris- 1To whom correspondence should be addressed ing 8 bp statistically occur only every 65 536 bp and phage E-mail: [email protected] genomes usually are short (104–105 bp), a restriction enzyme The restriction endonuclease EcoRV has been characterized with an 8 or even 10 bp recognition site could not fulfil this in structural and functional terms in great detail. Based role efficiently. Restriction endonucleases having a recognition on this detailed information we employed a structure- sequence of ജ8 bp, therefore, in general do not provide a guided approach to engineer variants of EcoRV that should selective advantage for the bacterial host and in consequence be able to discriminate between differently flanked EcoRV to date only 3.5% of all possible 8 bp cutters are available recognition sites. In crystal structures of EcoRV complexed from natural sources (Roberts and Macelis, 1999) and most with d(CGGGATATCCC)2 and d(AAAGATATCTT)2, likely not many more will be found by screening bacterial Lys104 and Ala181 closely approach the two base pairs strains. On the other hand, rare cutting enzymes could be flanking the GATATC recognition site and thus were particularly useful for the manipulation of large DNA frag- proposed to be a reasonable starting point for the rational ments, e.g. human genes, which are often larger than 5 kb. A extension of site specificity in EcoRV [Horton,N.C. and possible way to overcome the shortage of naturally occurring Perona,J.J. (1998) J. Biol. Chem., 273, 21721–21729]. To ‘rare cutters’ is the engineering of existing restriction enzymes test this proposal, several single (K104R, A181E, A181K) to recognize a more extended sequence than their natural and double mutants of EcoRV (K104R/A181E, K104R/ recognition sequence. A181K) were generated. A detailed characterization of all The restriction endonuclease EcoRV (recognition site: GAT- variants examined shows that only the substitution of ATC) is one of the best characterized type II restriction Ala181 by Glu leads to a considerably altered selectivity enzymes and, therefore, within this class of enzymes is an with both oligodeoxynucleotide and macromolecular DNA ideal target for a rational protein design. Besides detailed substrates, but not the predicted one, as these variants biochemical analysis (review: Pingoud and Jeltsch, 1997), a prefer cleavage of a TA flanked site over all other sites, wealth of structural information is available for this enzyme under all conditions tested. The substitution of Lys104 by including the structure of the free enzyme (Winkler et al., Arg, in contrast, which appeared to be very promising on 1993), a structure of the enzyme bound non-specifically to the basis of the crystallographic analysis, does not lead to DNA (Winkler et al., 1993), different structures of specific variants which differ very much from the EcoRV wild- EcoRV substrate (Kostrewa and Winkler, 1995; Perona and type enzyme with respect to the flanking sequence prefer- Martin, 1997; Horton and Perona, 1998a) and EcoRV product ences. The K104R/A181E and K104R/A181K double complexes (Kostrewa and Winkler, 1995; Horton and Perona, mutants show nearly the same preferences as the A181E 1998b) as well as structures of EcoRV variants bound to and A181K single mutants. We conclude that even for specific substrates (Horton and Perona, 1998b) and of EcoRV the very well characterized restriction enzyme EcoRV, bound to modified substrates (Martin et al., 1999). All these properties that determine specificity and selectivity are structural studies as well as biochemical studies using chemic- difficult to model on the basis of the available structural ally modified substrates (Thorogood et al., 1996) and EcoRV information. variants (Wenz et al., 1996) show that EcoRV in addition to Keywords: DNA recognition/EcoRV/protein engineering/ contacts with its recognition sequence interacts also with base rational protein design/site-directed mutagenesis/specificity pairs upstream and downstream of its recognition site. These additional contacts may explain the site preferences of EcoRV observed under certain conditions (Taylor and Halford, 1992; Introduction Lanio et al., 1998; Scho¨ttler et al., 1998). Hence it seems to Type II restriction endonucleases belong to the most important be possible to create new protein–DNA contacts to the base enzymes in molecular biology and molecular medicine. These pairs flanking the GATATC recognition site which could enable homodimeric enzymes recognize and cleave DNA in short the nuclease to recognize an extended site comprising up palindromic sequences comprising 4–8 base pairs with very to 10 bp. Crystallographic analyses of EcoRV with short high accuracy. Independent of the sequence context, the oligodeoxynucleotides differing in the base pairs adjacent to canonical site is cleaved several orders of magnitude faster the recognition site, suggest a promising starting point for a than all other sites, including sites which differ at only one rational protein engineering project. Two amino acid residues, base pair from the canonical recognition site (reviews: Roberts Lys104 and Ala181, form water-mediated contacts to the and Halford, 1993; Pingoud and Jeltsch, 1997). Today, more neighboring base pairs upstream (Ala181) and downstream than 2000 restriction endonucleases with ~200 different speci- (Lys104) of the recognition site (Horton and Perona, 1998a). ficities are known, among them only a few with longer than In the complex with d(CGGGATATCCC)2, Lys104 interacts © Oxford University Press 275 T.Lanio et al. through water molecules with the exocyclic N-4 amino group the 5-methyl group of thymine (Horton and Perona, 1998a). of the flanking cytosines on the 3Ј-side of the recognition These water-mediated contacts to a base next to the recognition sequence. These contacts are not seen with the d(AAAGAT- sequence could in principle be replaced by a specific hydrogen ATCTT)2 substrate, presumably because they are prevented bond, if the lysine is replaced by the slightly larger arginine. by steric exclusion of water molecules due to the presence of Changing a water-mediated contact to a specific hydrogen bond should enable the resulting variant to discriminate against substrates which could not provide that additional contact. In the case of Lys104 a change to arginine has been suggested in order to create a direct contact on the 3Ј-side with the O-4 of a flanking thymine or the O-6 of a flanking guanine (Horton and Perona, 1998a). On the 5Ј-side, the side chain of Ala181 points towards the base pair flanking the recognition site. This residue has been investigated by site-directed mutagenesis recently and some of the variants generated were shown to display an extended specificity towards differently flanked sites, such as A181E, A181F, A181I and A181K (Scho¨ttler et al., 1998). Therefore, combination of variants at positions 104 and 181 appeared to be very promising for the design of an 8 or 10 bp cutter, as discussed by Horton and Perona (1998a) and illustrated in Figure 1. As shown in Table I, combinations of different substitutions at positions 104 and 181 should result in several EcoRV variants with an extended specificity towards differently flanked substrates (Horton and Perona, 1998a). For example, combination of the Lys104 to Arg with the Ala181 to Glu substitution should result in a double mutant with a preference for CGATATCG. We have produced and characterized all the variants given in Table I. While some of the variants show preferences that are different from the wild-type enzyme, none of them displays the postulated preferences. Furthermore, the combination of amino acid substitutions does not lead to synergistic effects.

Table I. Predicted preferences of single and double mutants of EcoRV with amino acid substitutions at positions 104 and 181 according to Horton and Perona (1998a)

Variant Possible contact Predicted preferences

K104R 3Ј O4Tor3Ј O6GNGATATC(G/T) A181E 5Ј N4CCGATATCN A181K 5Ј N7A and 5Ј N7/O6G (A/G)GATATCN K104R/A181E 5Ј N4C and 3Ј O6GCGATATCG K104R/A181K 5Ј N7A and 3Ј O4TAGATATCT

Fig. 1. Models of possible interactions of EcoRV mutants with flanking DNA. These models were constructed by Horton and Perona (1998a) using subunit II of the dGC cocrystal structure as the starting point. Mutations in the protein were introduced and torsion angles varied systematically. Optimal least-squares superpositions of alternative base pairs were carried out using atoms in the glycosidic bonds and the glycosidic torsion angles were adjusted to match those in the dGC structure. (A) The K104R/A181E mutant interacting with flanking CG base pair. Here a 3Ј-G (GUA10) replaces the 3Ј-C visualized in the dGC structure. CYT9 is the 3Ј-base of the target site GATATC. Dotted black lines indicate modeled hydrogen bonds or nearest approach distances with the distance between the two electronegative atoms indicated in Å. The specific recognition of the outer base pair of the target site by Gly182 and Gly184 is also shown. (B)The K104R/A181K mutant interacting with a flanking TA base pair, where a 3Ј- T (THY10) replaces the 3Ј-C in the dGC structure. (C) The K104R/A181K mutant interacting with a flanking GC base pair. Here the flanking pair is as visualized in the dGC structure with a 3Ј-C nucleotide adjacent to the target site (reproduced with permission of the authors). 276 Restriction enzymes: EcoRV as the target

Materials and methods Table II. Kinetic parameters, Km and kcat, for the cleavage of four Site-directed mutagenesis of the ecoRV gene, overexpression differently flanked oligodeoxynucleotides by wild-type EcoRV and EcoRV and purification of mutant proteins variants Mutagenesis was performed using a PCR megaprimer tech- WTa K104R A181Ka A181Ea K104R K104R nique as described (Roth et al., 1998). Together with the A181K A181E mutation, one of the PCR primers introduces a characteristic restriction site into the gene to allow for a fast screening of Km (nM) AT 32 6.4 47 n.d.c. 87 n.d.c. the presence of the mutation. The mutant genes were cleaved CG 13 8 52 2700 48 2100 GC 5.5 20 59 n.d.c. 45 n.d.c. with NsiI and SalI and cloned into the plasmid pHis (Wenz TA 5.6 9 51 3100 46 3900 –1 et al., 1994). Both strands of the complete ecoRV gene of all kcat (min ) AT 1 2 4.8 n.d.c. 7 n.d.c. clones were sequenced using an ABI 373 sequencer (Applied CG 3 1.6 0.7 1.5 0.5 1.3 Biosystems). Expression of the EcoRV mutants was induced GC 0.8 2 10 n.d.c. 3.5 n.d.c. TA 1 4.4 0.7 3.5 0.8 8.5 in 3 ml of LB medium at a cell density of 1 OD600 by addition kcat/Km AT 0.53 5.2 1.7 0.0015 1.4 0.004 of IPTG to a final concentration of 1 mM. The His6-tagged (M–1 s–1ϫ106) mutants were purified using Ni-NTA magnetic agarose beads CG 3.8 3.4 0.23 0.009 0.16 0.01 (5% suspension) (Qiagen, Hilden, Germany). Harvested cells GC 2.5 1.5 3.1 0.001 1.3 0.002 were resuspended in 600 µl of binding buffer [30 mM TA 3.1 8.4 0.23 0.019 0.3 0.03 potassium phosphate, pH 7.5, 0.1 mM DTE, 0.01% (w/v) n.d.c., No detectable cleavage. lubrol, 500 mM NaCl, 20 mM imidazole] and after sonication aTaken from Scho¨ttler et al. (1998). (2ϫ30 s on ice), cell debris was removed by centrifugation and 200 µl of the supernatant were transferred into the wells of a 96-well microplate. A volume of 20 µl of a suspension velocities (v0) were calculated from the linear part of individual of agarose beads was added to the supernatant and vortex reaction progress curves obtained at five or more different mixed at 600 r.p.m. for 30 min at 4°C. The microplate was substrate concentrations. The Km and kcat values were calculated then placed on a 96-well magnet (Qiagen) for 1 min and the from v0 vs c diagrams by a best fit to the Michaelis–Menten supernatant was removed with a pipet. The beads were washed equation. Errors were calculated to be within 10–30%. The oligodeoxynucleotides used as substrates for EcoRV twice with binding buffer. To elute the His6-tagged protein, 50 µl of elution buffer [30 mM potassium phosphate, pH 7.5, (see above) were cloned into the EcoRI site of pUC8 as 0.1 mM DTE, 0.01% (w/v) lubrol, 500 mM NaCl, 200 mM described (Scho¨ttler et al., 1998). For the DNA cleavage imidazole] were added and the eluates were collected after kinetics, 15 nM plasmid was digested with appropriate dilutions placing the microplate on the magnet for 1 min. Protein of EcoRV variants (1.5 and 15 nM) at 37°C in the standard preparations obtained contained Ͼ90% pure EcoRV and were cleavage buffer containing 1 or 10 mM MgCl2. After defined time intervals, 15 µl aliquots were withdrawn, the reaction free of contaminating nuclease activity. Typical yields were µ 5 µgin50µl (corresponding to a 1–2 µM EcoRV solution) stopped with 5 l 5-fold gel loading buffer containing 250 mM starting froma3mlofculture. For detailed characterization, EDTA and analyzed by agarose gel electrophoresis. After larger amounts of certain variants were prepared as described ethidium bromide staining, quantification of substrate and (Wenz et al., 1994). product concentrations was carried out by integration of the intensities of the respective bands on a digitized image. Initial DNA-cleavage assays rates (k) were obtained from the linear part of the reaction We employed the following 20mer oligodeoxynucleotides as progress curves. substrates for EcoRV: 20AT, d(GATCGAAGATATCTTCG- ATC)2; 20CG, d(GATCGACGATATCGTCGATC)2; 20GC, Results d(GATCGAGGATATCCTCGATC)2; and 20TA, d(GATCGA- Mutagenesis and expression of the EcoRV variants TGATATCATCGATC) (the EcoRV recognition sequence is 2 Eco highlighted in bold face and the different nucleotides flanking RV mutants at positions 104 and 181 were generated by the recognition site are underlined). All oligodeoxynucleotides PCR mutagenesis. All mutants and the wild-type enzyme carry were purchased from Interactiva (Ulm, Germany). After puri- an N-terminal His6 tag which allows purification by affinity fication by denaturing polyacrylamide gel electrophoresis, the chromatography using Ni-NTA agarose beads. A volume of Ј 3mlofEscherichia coli cell culture on average yielded 5 µg oligodeoxynucleotides were labeled at the 5 -end using T4 Ͼ polynucleotide kinase. If not stated otherwise, all cleavage of a recombinant protein preparation of 90% purity as judged reactions were carried out in 20 mM Tris–HCl, pH 7.5, 50 by SDS–PAGE. mM NaCl, 10 mM MgCl2, 100 µg/ml BSA (bovine serum Cleavage of oligodeoxynucleotides albumin) at ambient temperature. To measure the preferences of wild-type EcoRV and the EcoRV Reaction mixtures for determination of Km and kcat values variants for flanking sequences, cleavage experiments were contained 0.01–10 µM of radioactively labeled oligodeoxy- performed with four self-complementary 20mer oligodeoxynu- nucleotide. Reactions were started by addition of enzyme at a cleotides which differ from each other only in the neighboring concentration which was at least 10-fold lower than the positions 5Ј and 3Ј to the GATATC recognition site. The results substrate concentration. After defined time intervals, 2 µl of the cleavage experiments are summarized in Table II. As aliquots were withdrawn from the reaction mixture, spotted on described previously (Scho¨ttler et al., 1998), the wild-type to a DEAE-cellulose plate (Macherey–Nagel, Du¨ren, Germany) enzyme cleaves the oligodeoxynucleotides with a CG, GC and subjected to homochromatography (Brownlee and Sanger, and TA flanked recognition site with almost the same rate 1969). For the quantification of substrate and product concen- (kcat/KM) and the oligodeoxynucleotide with the AT flanked trations, an Instant Imager was used (Canberra Packard). Initial recognition site more slowly. Unlike the wild-type enzyme, 277 T.Lanio et al.

–1 –1 6 Fig. 2. kcat/Km values (M s ϫ10 ) for the cleavage of oligodeoxynucleotides containing differently ﬂanked EcoRV sites by EcoRV variants carrying substitutions at positions 104 and 181. Note the different scales in (A) and (B).

Table III. Rate constants, k (min–1), for the cleavage of plasmids with four differently ﬂanked EcoRV recognition sites by EcoRV and EcoRV variants

Plasmid Steady state Single turnover 10 mM MgCl2

10 mM MgCl2 1mMMgCl2

WTa A181Ka A181Ea WTa A181Ka A181Ea WTa A181Ka A181Ea

AT 1.6 0.45 0.6 41 0.65 2.8 7.7 0.091 1.9 CG 1.9 0.36 0.09 42 0.53 0.74 9.1 0.024 0.32 GC 1.5 0.38 0.1 63 0.44 0.31 7.1 0.081 1.2 TA 3.5 0.2 7.2 33 0.23 13.2 6.2 0.013 4.6

Plasmid Steady state Single turnover 10 mM MgCl2

10 mM MgCl2 1mMMgCl2

K104R K104R K104R K104R K104R K104R K104R K104R K104R A181K A181E A181K A181E A181K A181E

AT 2.3 4.5 1.2 7.5 2.8 5.4 7.9 0.58 2.8 CG 3.5 2.4 0.3 6.6 2.4 0.8 10 0.9 0.65 GC 2.5 4.6 0.3 7.9 3.7 0.8 10 1.1 0.53 TA 4.8 1.9 8.5 9.5 1.5 13 13 0.65 7.4 aTaken from Scho¨ttler et al. (1998). the K104R variant cleaves the AT flanked site with a high site at least 10 times faster than sites with a purine at the activity and slightly disfavors the GC flanked site (Figure 2). 5Ј-side, albeit displaying a very low catalytic activity compared It had been predicted by Horton and Perona (1998a) that this with the wild-type enzyme (Table II, Figure 2). The decrease variant forms a contact to the O4 of a thymine or to the O6 in catalytic activity can be explained by the additional charge of a guanine flanking the recognition sequence at the 3Ј-side. of a deprotonated glutamic acid residue within the protein– However, whereas CG-flanked sites are preferred over AT- DNA interface, which is likely to disturb substrate binding flanked sites by the wild-type enzyme, this preference is lost (increase in KM) but not necessarily the catalytic step (unaltered for the K104R variant, indicating that K104R presumably does kcat). The residual activity and the high specificity of the not form the postulated contact to the O6 of guanine or, if it variant can be explained by a hydrogen bond between the does, that this contact is not accompanied by a preference for partially protonated carboxylate of Glu181 and O4 of thymine AT- and CG-flanked EcoRV sites. A different result had been (Scho¨ttler et al., 1998). obtained for the A181K and A181E variants (Scho¨ttler et al., According to the suggestion of Horton and Perona (1998a), 1998): the A181K variant prefers sites flanked by a purine at the double mutant K104R/A181E should strongly prefer the the 5Ј-side, most probably due to a hydrogen bond between CG-flanked substrate, whereas K104R/A181K should cleave the amino group of the lysine side chain and N7 of purine and an AT-flanked site much faster than all other sites (Table I, cleaves the CG- and TA-flanked sites more than one order of Figure 1). As shown in Table II and Figure 2, the double magnitude more slowly than the GC-flanked site (Table II, mutants A181K/K104R and A181E/K104R show nearly the Figure 2). In contrast, the A181E variant strongly prefers sites same kinetic parameters as the single mutants A181K and with a pyrimidine at the 5Ј-side and cleaves the TA-flanked A181E. Hence a combination of the substitution Lys104 → 278 Restriction enzymes: EcoRV as the target

Fig. 3. Initial rates (min–1) for the cleavage of plasmid substrates containing differently ﬂanked EcoRV recognition sites by EcoRV variants carrying substitutions at positions 104 and 181.

279 T.Lanio et al.

Arg with the substitution Ala181 → Glu and Ala181 → Lys, in particular in the case of EcoRV, which is one of the best respectively, leads only to minor alterations in the kinetic characterized enzymes acting on DNA. Structural analyses of parameters of the EcoRV variants. wild-type enzyme complexed with differently flanked sub- Cleavage of macromolecular substrates strates reveals that two amino acid side chains closely approach the bases flanking the recognition site (Winkler et al., 1993; To analyze whether macromolecular substrates are cleaved with the same preferences as oligodeoxynucleotide substrates, Horton and Perona, 1998a). Based on these crystallographic steady-state and pre-steady-state cleavage experiments with data, a proposal for the design of EcoRV variants with an plasmids containing the four different 20mer oligodeoxynucle- enhanced selectivity had been put forward by Horton and otide sequences were performed (Table III, Figure 3). The Perona (1998a). They had suggested generating single and wild-type enzyme, the K104R and the A181K variants display double EcoRV mutants with amino acid substitutions at posi- hardly any selectivity towards differently flanked EcoRV sites, tions Lys104 and Ala181. The resulting variants should be under both steady-state and single-turnover conditions. Only able to distinguish between differently flanked GATATC sites. All these variants have been generated by us and thoroughly at low MgCl2 concentrations does A181K display the same preference with plasmid substrates as with oligodeoxynucleo- characterized. However, none of them has the preferences Ј tide substrates. The A181K variant, however, is considerably expected by Horton and Perona. A181E predicted to prefer 5 Ј less active than the wild-type enzyme with macromolecular C-flanked sites in fact best cleaves 5 T-flanked sites. A181K Ј substrates. The enzymatic activity of the A181K variant is seems to prefer a purine 5 adjacent to the recognition site as restored to wild-type level by the additional substitution of proposed, but this preference is observed only under certain cleavage conditions. The Lys104 to arginine exchange has Lys104 by Arg whereas the selectivity observed at low MgCl2 concentration is lost. The resulting double mutant has wild- almost no influence on the DNA cleavage of EcoRV.Combining type activity under steady-state conditions and displays, like arginine at position 104 with lysine and glutamic acid at the wild-type enzyme and the single mutant, little preference position 181 does not improve the specificity of the single regarding the 5Ј- and 3Ј-flanking regions. Under-single turnover mutants but leads only to minor differences regarding site preferences. Taken together, the results of our cleavage experi- conditions with 10 mM MgCl2, the K104R single mutant is less active than the wild-type enzyme, but, like the wild-type ments show that the predicted contacts do not exist or at least enzyme, has almost no preferences for flanking regions. In do not have the expected effect, i.e. do not influence the contrast, the activity and selectivity of the A181E mutant are transition state. higher with macromolecular substrates than with oligo- We conclude that on the basis of structures of enzyme– deoxynucleotide substrates. The low activity of A181E with substrate complexes, it seems to be very difficult to predict oligodeoxynucleotides is mainly due to a high Km. Presumably, amino acid exchanges that would lead to an extended specifi- this deficiency is not observed with plasmids, because macro- city. One reason for the problems encountered could be that molecular DNA has a higher surface density of negative charge in the co-crystals obtained so far, EcoRV cannot be activated ϩ than oligodeoxynucleotides (Zhang et al., 1996), resulting in by Mg2 ions soaked into the crystals. Thus, structures of an improved substrate binding. The additional substitution of wild-type EcoRV or certain variants complexed to their specific Lys104 by arginine in the A181E context does not have a substrate do not necessarily represent the ground state of the major influence regarding the cleavage of plasmid substrates. enzymatic reaction. If they did, it could be that the designed The strong increase of the wild-type activity under single- mutations cannot provide much additional discrimination, turnover conditions compared with steady-state conditions possibly because one or two additional hydrogen bonding reveals product dissociation to be the rate-limiting step for the contacts using long side chains are not sufficient. More cleavage reaction. In contrast, the rate of DNA cleavage by important, however, no structural information on the transition all variants is not increased under single-turnover conditions, state of EcoRV is available, such that predictions regarding indicating that for them the catalytic step or a step preceding the effects of single amino acid substitutions are unreliable. it is rate limiting. We believe that generating EcoRV variants with extended specificity by protein engineering requires the design of Discussion cooperative conformational changes that couple recognition to In spite of the availability of detailed structural data, rational catalysis. The amino acid substitutions required to facilitate protein design with type II restriction endonucleases as targets such conformational changes can hardly be predicted. At turns out to be very difficult. The recognition process that present, our lack of understanding of the key steps, together leads to the activation of the catalytic centers of restriction with a thorough evaluation of their entropic and enthalpic enzymes is highly redundant and involves not only hydrogen contributions, in the mechanism of action of EcoRV which bonds and van der Waals contacts to the bases of the recognition lead to the activation of the catalytic centers prevents a site (direct readout), but also a multitude of interactions, some successful rational protein design for target site expansion of of them water mediated, between the protein and the sugar– EcoRV and requires random mutagenesis/selection approaches. phosphate backbone of the DNA (indirect readout) (Rosenberg, In fact, directed evolution approaches have already been 1991; Kostrewa and Winkler, 1995; Perona and Martin, 1997; demonstrated to be successful for the generation of EcoRV Wah et al., 1997). This may be the reason why all attempts to variants with extended specificity (Lanio et al., 1998) change the specificity of restriction enzymes have failed so far (Jeltsch et al., 1996), with the notable exception of Acknowledgements specificity changes directed towards chemically modified We thank Dr J.Perona for communicating results prior to publication. The substrates (Lanio et al., 1996). In contrast, designing new financial support of the Deutsche Forschungsgemeinschaft, the Bundes- contacts of the protein to the DNA outside the original ministerium fu¨r Bildung, Wissenschaft, Forschung und Technologie and the recognition sequence appears to be a more manageable task, European Community is gratefully acknowledged. 280 Restriction enzymes: EcoRV as the target

References Brownlee,G.G. and Sanger,F. (1969) Eur J Biochem., 11, 395–399. Horton,N.C. and Perona,J.J. (1998a) J. Biol. Chem., 273, 21721–21729. Horton,N.C. and Perona,J.J. (1998b) J. Mol. Biol., 277, 779–787. Jeltsch,A., Wenz,C., Wende,W., Selent,U. and Pingoud,A. (1996) Trends Biotechnol., 14, 235–238. Kostrewa,D. and Winkler,F.K. (1995) Biochemistry, 34, 683–696. Lanio,T., Selent,U., Wenz,C., Wende,W., Schulz,A., Adiraj,M., Katti,S.B. and Pingoud,A. (1996) Protein Engng, 9, 1005–1010. Lanio,T., Jeltsch,A. and Pingoud,A. (1998) J. Mol. Biol., 283,59–69. Martin,A.M., Horton,N.C., Lusetti,S., Reich,N.O. and Perona,J.J. (1999) Biochemistry, 38, 8430–8439. Newman,M., Strzelecka,T., Dorner,L.F., Schildkraut,I. and Aggarwal,A.K. (1995) Science, 269, 656–663. Newman,M., Lunnen,K., Wilson,G., Greci,J., Schildkraut,I. and Phillips,S.E. (1998) EMBO J., 17, 5466–5476. Perona,J.J. and Martin,A.M. (1997) J. Mol. Biol., 273, 207–225. Pingoud,A. and Jeltsch,A. (1997) Eur. J. Biochem., 246,1–22. Roberts,R.J. and Halford,S.E. (1993) Nucleases. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 35–88. Roberts,R.J. and Macelis,D. (1999) Nucleic Acids Res., 27, 312–313. Rosenberg,J.M. (1991) Curr. Opin. Struct. Biol., 1, 104–113. Roth,M., Helm-Kruse,S., Friedrich,T. and Jeltsch,A. (1998) J. Biol. Chem., 273, 17333–17342. Scho¨ttler,S., Wenz,C., Lanio,T., Jeltsch,A. and Pingoud,A. (1998) Eur. J. Biochem., 258, 184–191. Taylor,J.D. and Halford,S.E. (1992) Biochemistry, 31,90–97. Thorogood,H., Grasby,J.A. and Connolly,B.A. (1996) J. Biol. Chem., 271, 8855–8862. Wah,D.A., Hirsch,J.A., Dorner,L.F., Schildkraut,I. and Aggarwal,A.K. (1997) Nature, 388,97–100. Wenz,C., Selent,U., Wende,W., Jeltsch,A., Wolfes,H. and Pingoud,A. (1994) Biochim. Biophys. Acta, 1219,73–80. Wenz,C., Jeltsch,A. and Pingoud,A. (1996) J. Biol. Chem., 271, 5565–5573. Winkler,F.K., Banner,D.W., Oefner,C., Tsernoglou,D., Brown,R.S., Heathman,S.P., Bryan,R.K., Martin,P.D., Petratos,K. and Wilson,K.S. (1993) EMBO J., 12, 1781–1795. Zhang,W., Bond,J.P., Anderson,C.F., Lohman,T.M. and Record,M.T., Jr (1996) Proc. Natl Acad. Sci. USA, 93, 2511–2516.

Received November 1, 1999; revised January 18, 2000; accepted February 8, 2000

281