Location of Putative Binding and Catalytic Sites of NaeI by Random Mutagenesis*

Endonuclease NaeI is a prototype for an unusual group of type I1 restriction endonucleases that must bind two DNArecognition sequences to cleave DNA. The naem gene, expressed from a Ptac promotor construct, was toxic to Escherichia coli in the absence of NaeI- sequence specific methylases. The nuem gene was mutagenized with N-methyl-N-nitrosoguanidine; four classes of NaeI variants were isolated in the absence of protecting methylase activity. Class I variants (T601, E70K) lacked detectable cleavage activity, but displayed good sequence-specific DNA binding. Class I1 variants (D95N, G141D) displayed 1 4 % of the wild-type cleavage activity and normal DNA binding. Class I11 variants (G131E, G131R, G155D, 62453) displayed significantly attenuated cleavage and binding activities. Class IV variants (G197D, G214RIA219T, G236S, L241P, 62453, G245R, G250E, G270E) lacked both cleavage and binding activities. These results imply two amino acids (Thr-60, Glu-70) essential for catalysis. In addition, two domains are indicated in NaeI: one (Thr-60 to Gly-155) mediates substrate binding and catalysis, the other (Gly-197 to Gly-270) may mediate binding of the activating DNA se- quence. Our results are compared with the active site residues of EcoRI, on a model 373A DNA Sequencer (Applied Biosystems) using the Taq DyeDeoxyTM terminator cycle sequencing kit (Applied Biosystems).

The NaeI restriction endonuclease, isolated from Nocardia aerocolonigenes, is a member of the unique type IIe class of endonucleases. This family includes BspMI, HpaII, NaeI, NurI, and Sac11 (Conrad and Topal, 1989;Oller et al., 1991) as well as EcoRII (Kriiger et al., 1988;Gabbara and Bhagwat, 1992). NaeI is characterized by the strong sigmoidal dependence of its cleavage velocity on substrate concentration. This sigmoidal dependence is indicative of the need to bind two DNA recognition sequences for cleavage. The two DNA-binding sites manifest different preferences for sequences flanking the recognition sequence (Yang and Topal, 1992). This gives the interesting situation that DNAs containing sequences without sufficient affinity to occupy one or the other sites are resistant to cleavage (Conrad and Topal, 1989;Oller et al., 1991;Yang and Topal, 1992). NaeI recognizes and cleaves the sequence GCC/GGC. Binding of the second DNA recognition sequence assembles an active form ofNaeI homodimer. Dimer formation in the absence of two-site binding gives an inactive conformation ofNaeI (Baxter and Topal, 1993). Although a significant amount of work has been done to understand the NaeI reaction scheme (Yang and ~~ * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom correspondence should be addressed. Tel.: 919-966-5040; Fax: 919-966-3015. Topal, 1992;Baxter and Topal, 1993), little is known about the structure of the protein and the amino acid residues that mediate its functions.
Together with the crystal structures for EcoRI, EcoRV, and BarnHI, identification of cleavage-deficient variants that bind DNA have defined a Mg2' binding motif. Such variants also provided critical information to help assign the amino acids involved in DNA recognition by EcoRI (reviewed by Heitman, 1992 and1993).
Here we employ random mutagenesis with MNNG' to isolate cleavage-deficient variants of NaeI endonuclease. Bacteria were selected by their ability to survive in the presence of mutagenized NaeI but lacking protecting methylases. Analysis of cleavage and DNA binding ability of the resulting NaeI variants enabled the identification of four classes of NaeI variants with reduced cleavage activity.

27286
This is an Open Access article under the CC BY license. restriction endonuclease were from Promega. Ampicillin, MNNG, X-gal, and IPTG were from Sigma. Protein dye reagent was from Stratagene. Sep-Pak plus C,, cartridges were from Waters.

Methods
with MNNG according to Miller (1972). 200 pl of MNNG (1 mg/ml) were MNNG Mutagenesis-Random mutations were generated in vivo added to 5 ml of log-phase growth CAAl (Me') cells containing the naeZR-containing plasmid pCY786. After treatment, the cells were grown overnight at 37 "C in 5 ml of LB/ampicillin medium. Minipreparations of naeIR-plasmid DNA were transformed into either WA802 or ER1992 competent cells, which lack specific methylases to protect against NaeI cleavage. Cleavage defective naeZR-containing variants were selected based on their ability to form colonies on ampicillin plates. Mutants were selected a t random and numbered in chronological order. Control cells containing wild type naeZR gave no colonies under these conditions when plated at a density of lo6 colony-forming unitdplate.
Temperature-sensitive Mutants-MNNG-treated pCY786 was transformed into WA802 cells and grown overnight a t 42 "C. Colonies that survived were replica-plated (Miller, 1972); replicas were grown at 25 "C overnight. Temperature-sensitive variants were selected that grew at 42 "C but not at 25 "C.
DNA Isolation-For minipreparations, putative NaeI mutant colonies were grown overnight in LB/ampicillin medium. 3 ml of the culture were pelleted and resuspended in STET buffer (8% sucrose, 5% Triton X-100, 50 mM EDTA, pH 8.0, 10 m~ Tris, pH 8.0). The cells were boiled for 1 min, and the cell debris was pelleted by spinning for 15 min in a microcentrifuge. DNA was precipitated from the resulting supernatant by adding isopropyl alcohol to 50% and centrifuging an additional 15 min. The DNA pellet was resuspended in 100 pl of H,O.
For larger-scale DNA preparations, naelR mutants were grown overnight in 100 ml of LB/ampicillin medium. The DNA was isolated using Qiagen-tip 100 columns according to the manufacturer, except that the DNA was resuspended in 100 pl of H,O instead of Tris-EDTA for DNA sequencing purposes.
Cell-free Extracts-Overnight cell cultures were subcultured, grown approximately 2 h to mid-log phase, induced with IPTG (220 pg/ml), and grown to saturation. Cells from 2.5 ml of growth were pelleted and then resuspended in 1 ml of buffer (20 mM KPO,, 0.1 mM EDTA, 1 mM p-mercaptoethanol, 5% glycerol). Cells were lysed by sonicating three times for 15 s, and cell debris was pelleted. The total protein concentration of the cell-free extracts was determined by the Bradford (1976) method.
Cleavage Activity Assays-A volume of cell-extract containing 3 pg (or serial dilutions) of total protein from mutant and wild type NaeIcontaining cells was incubated with 200 ng to 1 pg of pBR322 at 37 "C for 1 h in a 20-pl reaction volume containing 50 mM /3-mercaptoethanol, 200 mM NaCl, 100 mM MgCl,, 100 mM Tris (pH 8.0), and bovine serum albumin (50 pglml). Reactions were stopped by incubating for 10 min at 65 "C. SDS was added to 0.2% to each sample, and samples were incubated for 10 min at 65 "C and then electrophoresed on 1% agarose gels. Assay conditions for the temperature-sensitive mutants were the same as above except the temperature-sensitive mutants were incubated with 400 ng of pBR322 at both 25 "C and 42 "C.
Gel Mobility Shift Assays--100 n~ of oligodeoxyribonucleotide was radioactively labeled by incubation with 0.02 mCi of [yP321dATP and 16 units of T4 Kinase, and 1 x kinase buffer in a 50-111 reaction volume for 30 min at 37 "C. The reaction was stopped by incubating at 65 "C for 10 min. Unincorporated label was removed with G-50 Sephadex spin columns. The recovered radiolabeled oligoribodeoxnucleotide was annealed to its complement by heating at 70 "C for 5 min in buffer and slowly cooling to 25 "C. Cell-free extracts (3 pg of total protein) from mutant and wild type NaeI-containing cells were incubated with 10 nM [y3'P1dATP labeled double-stranded probe for 10 min a t 37 "C in a 10-pl reaction volume containing 100 IXIM Tris (pH 8.0), 50 m~ /3-mercaptoethanol, 200 m~ NaC1, 20 mM CaCl,, 100 ng/ml bovine serum albumin, 10% glycerol, 300 ng of poly(d1-dC). Samples were immediately loaded onto a 6% (37.5:l) polyacrylamide gel and run for 2 h at 200 V.
The NaeI mutants that bound to DNA were tested for specific versus nonspecific binding. Assay conditions were the same as the general binding assays except either 1 PM unlabeled sequence-specific competitor or 1 unlabeled nonspecific competitor were added to the reactions.
DNA Sequencing-Oligonucleotide primers for sequencing were designed based upon the naeZR gene sequence (Taron et al., 1994). The primers were desalted using the Sep-Pak plus C,, cartridges. 40 nM primer was resuspended in 500 pl of H,O, and NaCl was added to 0.7 M.
The cartridges were rinsed with 5 ml of MeOH followed by 5 ml of H,O.
The primer was applied to cartridge, then washed with 2 ml of H,O. The primer was eluted four times with 1 ml of 60% MeOWH,O. The fractions containing primer were pooled together, dried in a speed vac, and resuspended in 100 pl of H,O.
DNA was sequenced at the UNC-CH Automated DNA sequencing facility on a model 373A DNA Sequencer (Applied Biosystems) using the Taq DyeDeoxyTM terminator cycle sequencing kit (Applied Biosystems).

RESULTS
Mutagenesis of the NueZ Restriction Endonuclease Gene-To analyze the domains ofNueI involved in endonuclease function, we isolated NueI variants defective in DNA cleavage, DNA binding or both. NueI is lethal to the host cell in the absence of protecting methylase activity. Random mutagenesis of nuem by MNNG was used to select for defective variants.
To optimize conditions we determined a n MNNG survival curve and the amount ofNueI mutagenesis at each MNNG dose used. We isolated plasmid DNA from cells treated with 10 and 50% survival doses of MNNG, transfected the DNA into cells lacking methylase, and randomly picked surviving colonies. CAAl (MspI.me+) was used as the host cell strain for mutagenesis of nuem. MspI activity (mCCGG) is highly protective against cleavage by NueI (McClelland and Nelson, 1985).
To avoid DNA rearrangements and deletions, minipreparations of nuem-plasmid were prepared from each of the mutants, cleaved with restriction enzyme, and checked for appropriate fragment size by gel electrophoresis. Only those mutants showing expected fragment patterns were selected.
DNA Cleavage Assays-We compared the sequence-specific DNA cleavage activity of mutant and wild type NueI protein.
Serial dilutions of wild type and mutant cell-free extracts were incubated with pBR322, a plasmid DNA substrate containing four NueI cognate sites. Activity was measured from the intensity and banding pattern of the pBR322 cleavage products (Fig.  1). Of the 81 mutants compared to the wild type cell-free extract, 76 had no detectable activity and 5 (174,175,178,188,209) had partial activity.
The temperature-sensitive mutants TS8 and TS42 were tested for in vitro cleavage activities at the permissive (25 "C) and non-permissive (42 "C) temperatures. Both of the temperature-sensitive mutants showed small but detectable amounts of cleavage at both temperatures (not shown). Only very small differences in cleavage were detectable at the two temperatures.
Gel Mobility Shift Assays-To determine whether the variants with reduced cleavage activity resulted from changes in catalysis or changes in DNAbinding ability, NueI variants were also analyzed for DNA binding using the gel mobility-shift assay (Fig. 2). This assay measures protein-DNA binding because NueI.DNA complexes migrate slower than the free DNA during electrophoresis. Ca2+, instead of Me, was used as the divalent cation in the gel mobility shifts because NueI binds but does not cleave DNA in the presence of Ca2+ (Baxter and Topal, 1993). The DNA probe used in these assays was a 14-bp "specific" fragment containing the 6-bp NueI cognate recognition site (GCCGGC). Of 51 variants tested, 41 did not shift the mobility of the DNA probe, four (variants 60, 174, 175,209) bound similar t o the wild type NueI cell-free extract, two (203,212) bound tighter and shifted the mobility to that of higher apparent molecular weight complexes, and four (61,178,188,196) showed reduced but measurable binding.
Addition of large concentrations of cold NueI-specific probe (100-fold excess over labeled probe) to the binding reaction competed away the mobility shift. Addition of similar concentrations of a probe lacking the NueI recognition sequence did not. Therefore, the gel mobility shift depended on the presence of an NueI recognition sequence, which implies that the Cleavage activity of wild-type and m u t a n t NaeI extracts. Serially diluted ( 1 X = 3 pg of total cell free protein) wild type (lanes 3-7, upper panel) and mutant NaeI cell-free extracts were assayed as described under "Experimental Procedures" using pBR322 DNAas the substrate. Untreated DNAis shown in lane 1 ofboth panels. Separation of the reaction products by gel electrophoresis on 1% agarose for several example mutants and wild type is shown. Production of linear product is used as the indicator of NaeI activity. Note, for example, that dilution for mutant 178 is approximately equivalent to dilution 10"' of wild type. This indicates that this variant has approximately 1% the activity of wild type.
FIG. 2. Gel mobility shift assays of wild-type and v a r i a n t NaeI. Wild type and variant NaeI cell free extracts were assayed for DNA binding as described under "Experimental Methods." Specificity ofbinding was tested by competition with specific and nonspecific competitor DNAs as indicated. Variants203 (E70K), 178 (G245E), and 189 (G250E) are shown as examples of tight binding, reduced binding, and no binding, respectively. Identical amounts of total protein, as determined by Bradford assay, were used in each lane; the higher amounts of binding in the NueI controls (compare lunes 2 and 3) are caused by the larger amounts of pure NaeI in the controls relative to that in the cell extracts. All binding reactions were reproducible based on at least two determinations using extracts prepared from a different bacterial colony for each determination.
protein-DNA interactions were sequence-specific.
DNA Sequence Analysis of the NaeI Mutants-To link the biochemical phenotypes to specific amino acid changes, we se-  Table I . This included 31 mutants: all 10 of the variants that showed binding but reduced cleavage, 19 randomly chosen non-binding mutants, and the 2 temperature-sensitive mutants. Five oligonucleotide primers were designed to completely sequence the naeIR gene and to sequence both strands of regions that gave sequence ambiguity. Three primers, spaced approximately 300 bp apart, sequence the template (+) strand. Two primers sequence the first and last thirds of the complementary (-) strand of the NueI gene. Although MNNG tends to give a high frequency of double mutations (Foster, 1991), only 8 out of the 31 mutants sequenced contained double mutations and 2 contained triple mutations. The triple mutations were not considered because of ambiguity in assigning structure-function relationships. Only one of the double mutants is considered for the same reason. Results of both the DNA sequencing data and the biochemical analysis of the remaining 22 variants (including the two temperature-sensitive variants) are compiled in Table I . DISCUSSION We randomly mutated the NaeI gene using MNNG and selected for cells that survive in the presence of mutated naeIR. The NueI variants isolated display deficiencies in DNA cleavage, DNA binding, or both. By this method, we have started to characterize substrate and effector DNA binding domains of NaeI and amino acids essential for catalysis.
Mutations Involved in DNA Binding--Two clusters of mutations between amino acids Thr-60 and Gly-155 and between amino acids Gly-197 and Gly-250 in the NaeI amino acid sequence indicate potential DNA binding sites (Fig. 3). The first region contains an intermingling of amino acids that attenuate catalysis with substitutions that attenuate binding (Fig. 3). The second region contains only substitutions that attenuate binding; most of these reduced binding to below detectable levels. All, but one, of the variants (178: G245E) in the second region lack detectable DNA cleavage and binding activities ( Table I ) .
Temperature-sensitive Mutations-The amino acid changes responsible for the two temperature sensitive variants were determined. The amino acid changes fell close to or within the second putative DNA binding region (Fig. 3). We could detect low levels of cleavage by these variants in vitro. The cleavage in vitro, however, was not temperature sensitive. Binding of the temperature-sensitive variants to DNAcould not be detected by gel mobility shift assay at either temperature. Thus, as one would expect for amino acid changes in this region of the protein, cleavage apparently was reduced by reduction in the affinity of the protein for DNA. Mutations Involved in Catalysis-Protein variants that bind DNA, but have no detectable cleavage activity are useful to identify mutations involved in catalysis (Fig. 3). NaeI mutations T60I and E70K may be essential for catalysis: they bind DNA with differing capabilities, but have no detectable cleavage activity. The E70K mutation increases binding and produces higher mobility complexes than those produced by wild type protein. The increased binding phenotype has been seen with variants of other restriction enzymes (Wright et al., 1989;Xu and Schildkraut, 1991) and is usually associated with creation of additional positive charges in the protein.
NaeI mutations D95N, G141D, G155D, and G245E bind DNA and display weak cleavage activity. NaeI mutations G131R and G131E bind DNA so weakly that the lack of detectable cleavage can be attributed to the low binding activity.
The NaeI amino acids apparently essential for catalysis have similarity to the developing signature for the endonuclease catalytic site (Fig. 4). The roles these residues play in EcoRI, EcoRV, and BamHI have been deduced from their crystal structure (McClarin et al., 1986;Kim et al., 1990;Winkler et al., 1993;Newman et al., 1994) and from each protein's apparent inability to tolerate substitutions a t these positions (King et al.,  EcoRV, and BamHI (McClarin et al., 1986;Kim et al., 1990;Winkler et al., 1993;Newman, 1994). NaeI may contain some elements of this signature: Glu-70 (E70) appears essential for catalysis and substitution of Asp-95 (D95) gives reduced cleavage but normal binding (see Table I 1989; Xu and Schildkraut, 1991;Thielking et al., 1992). The EcoRV catalytic center shares a similar structure with EcoRI (Selent et al., 1992). Residues Asp-74 and Asp-90 in EcoRV, and Glu-111 and possibly Asp-91 in EcoRI are involved in chelating Mg2' (Heitman, 1992). When Asp-90 chelates Mg2' in EcoRV, it distorts the DNA and activates the catalytic center of the endonuclease (Thielking et al., 1992). Lys-92 contacts the scissile phosphate in EcoRV; the Lys-113 residue in EcoRI may have a similar role (Selent et al., 1992). BamHI has a slight variation on this model. The Asp-94, Glu-111, and Glu-113 residues of BamHI are conserved with those in EcoRI and EcoRV. However, it also requires Glu-77 for catalysis.
NaeI amino acids Glu-70, Asp-95, and Lys-97 appear similar to the catalytic site signature discussed above (Fig. 4). Glu-70 appears to be essential for catalysis by NaeI, Asp-95 also appears to be involved in catalysis, but is not essential since reduced cleavage was observed in the D95N variant. In addition, Thr-60 is apparently essential for catalytic function of NaeI, but has no apparent counterpart in the other restriction enzymes mentioned.
The NueI catalytic domain, defined by the cleavage-deficient variants Thr-60, Glu-70, Asp-95, and Gly-141, is intermingled with substitutions that significantly affect binding to DNA (Fig.  3). The substrate binding domains for BamH1, EcoRI, and EcoRV also contain the amino acids essential for catalysis. Thus, this region in NaeI appears be the substrate binding domain, which also contains the catalytic site. The second domain is independent of the first and only affects DNA binding. This domain appears to be the activation domain ofNaeI, which kinetic studies imply is not directly involved in catalysis.' It is informative to lay the two putative NueI protein domains over a map of the basic and acidic regions of NaeI (Fig.  5). The substrate-binding domain coincides with an acidic and a basic region. These two different regions correlate with the catalytic and binding variants defining the first domain, respectively. The activation domain coincides with another significant basic region in NaeI. This basic region supports the notion that this is the second DNA binding site.
In summary, we have used random mutagenesis to locate amino acids involved in catalysis and DNA binding. These amino acids define two domains in the protein separated by significant linear distance in the protein sequence, but not necessarily in three-dimensional space. The domain closest to the amino terminus shares relative position and acidic amino acids in common with BamHI, EcoRI, and EcoRV. Two amino acids in this domain, T60 and E70, appear to be essential for catalysis. This domain also contains amino acids that when substituted attenuate DNA binding. These amino acids are K. J. and M. D. Topal, unpublished results. the DNA sequence shown in Fig. 3. associated with a basic region located within the domain. These results imply that this domain is the substrate-binding site of NueI. The second domain contains almost exclusively amino acids that are required for DNA binding. These, amino acids coincide with a larger, second basic region in NueI. This result suggests that this domain may be the effector-binding site of NaeI (Yang and Topal, 1992;Baxter and Topal, 1993).