The Interaction between Z-DNA and the Zab Domain of Double-stranded RNA Adenosine Deaminase Characterized Using Fusion Nucleases*

Zab is a structurally defined protein domain that binds specifically to DNA in the Z conformation. It consists of amino acids 133–368 from the N terminus of human double-stranded RNA adenosine deaminase, which is implicated in RNA editing. Zab contains two motifs with related sequence, Zα and Zβ. Zα alone is capable of binding Z-DNA with high affinity, whereas Zβ alone has little DNA binding activity. Instead, Zβ modulates Zα binding, resulting in increased sequence specificity for alternating (dCdG) n as compared with (dCdA/dTdG) n . This relative specificity has previously been demonstrated with short oligonucleotides. Here we demonstrate that Zab can also bind tightly to (dCdG) n stabilized in the Z form in supercoiled plasmids. Binding was assayed by monitoring cleavage of the plasmids using fusion nucleases, in which Z-DNA-binding peptides from the N terminus of double-stranded RNA adenosine deaminase are linked to the nuclease domain of FokI. A fusion nuclease containing Zα shows less sequence specificity, as well as less conformation specificity, than one containing Zab. Further, a construct in which Zβ has been replaced in Zab with Zα, cleaves Z-DNA regions in supercoiled plasmids more efficiently than the wild type but with little sequence specificity. We conclude that in the Zab domain, both Zα and Zβ contact DNA. Zα contributes contacts that produce conformation specificity but not sequence specificity. In contrast, Zβ contributes weakly to binding affinity but discriminates between sequences of Z-DNAs.

The biological role of Z-DNA has long awaited elucidation. New hypotheses were formed upon the discovery that the N terminus of ADAR1 1 binds to Z-DNA with high affinity (1). ADAR1 deaminates adenosines to inosine (functionally G) in double-stranded RNA in vitro (2), and has been suggested to edit pre-mRNA in vivo (3). The substrates for A to G RNA editing remain to be enumerated; they include modulating central nervous system function by changing glutamic acid receptors (4) and serotonin receptors (5). These editing events require the interaction between intronic sequences and the region surrounding the edited site to form the double-stranded RNA substrate; therefore editing must take place before splicing. Splicing has been shown to occur early in mRNA processing, sometimes before the termination of transcription. Movement of RNA polymerase during transcription leaves negatively supercoiled DNA in its wake, which transiently stabilizes Z-DNA. It has been proposed that Z-DNA binding may serve to target ADAR1 to regions of active transcription and thereby to nascent mRNA (6).
Although editing events have been characterized primarily in mammals, double-stranded RNA deaminase activity has also been shown in many animal phlya (3). ADAR1 has been identified from a variety of vertebrates (3); in every case the Z-DNA-binding motifs at the N terminus are conserved (1). These motifs, called Z␣ and Z␤, are separated by a linker of conserved size but divergent sequence; a single known exception is in the human protein (hADAR1), where the linker has been precisely duplicated (1,7).
The Z-DNA-binding region of hADAR1 has been extensively characterized (1, 8 -10). Z␣ binds to Z-DNA stabilized by bromination (1) and in supercoiled plasmids with (dCdG) n inserts (9,11). When Z␣ is titrated into a solution containing short oligonucleotides with alternating purine-pyrimidine sequences, the CD spectrum converts from a B-DNA to a Z-DNA form (8 -10, 12). It has recently been shown by limited proteolysis that the Z-DNA-binding region of hADAR1 is structurally organized into a bipartite domain (12). This domain, named Zab, contains the originally identified Z-DNA-binding region, Z␣; a second motif, Z␤, identified by sequence similarity (1); and the intervening linker. Although this entire region fulfills the definition of a protein domain, it is also possible to identify a proteolytically stable subdomain containing only Z␣. The DNA binding of this subdomain, named Za, has been compared with that of Zab. Za and Zab were overexpressed in Escherichia coli and tested for DNA binding by gel mobility shift assays and by the ability to stabilize short oligonucleotides in the Z conformation as shown by CD (12). Both bind to Z-DNA in the presence of large amounts of competing B-DNA. Za binds to short stretches of both (dCdG) n and (dCdA/dTdG) n , and oligonucleotides of either sequence adopt the Z form in the presence of a 2:1 base pair:peptide ratio of protein. In contrast, Zab does not bind to (dCdA/dTdG) n in these assays.
Fusion nucleases have a particular utility for the study of Z-DNA-binding proteins, because they have separate domains for DNA binding and cleavage. Z-DNA in cells must exist embedded in long stretches of B-DNA. It is likely to be stabilized by supercoiling or by protein binding rather than base modification and high salt. A fusion nuclease with a specificity for binding Z-DNA will cut Z-DNA in a supercoiled plasmid at physiological salt concentrations and at modest protein:DNA ratios. The cleavage can easily be detected (11). This assay has allowed us to characterize the binding of several different Z-DNA-binding peptides.
A short peptide containing Z␣ (residues 121-197), which has been described previously (1). Two molecules of Z␣ appear to bind together on Z-DNA and display conformational but not sequence specificity (9). In addition, it has been shown previously that a construct with two Z␣ segments connected by a short flexible linker bind Z-DNA more tightly than the Z␣ construct alone and that two molecules of Z␣ bind to a 6-base pair Z-DNA substrate (9). When Z␣ is linked to Z␤ by the same flexible linker, the binding differed from that of the linked Z␣-Z␣ construct in character, with less sensitivity to competing Z-DNA (9). Under low salt conditions, which emphasize nonspecific binding, Za (residues 133-209) binds slightly more specifically than the Z␣ construct. In contrast, Zab (residues 133-368) remains extremely conformation-specific even under low stringency conditions. Both Za and Zab bind to (dCdG) n in the Z form; however, only the Za nuclease cleaves plasmids containing (dCdA/dTdG) n inserts. Further, Zaa a modified Zab construct in which Z␤ has been replaced with Z␣, cleaves dramatically more efficiently than Z␣, Za, or Zab and shows no sequence specificity. Zaa nuclease may thus be used as a reagent to detect Z-DNA both in vitro and in vivo. Because Zaa binds DNA with a much higher affinity than Za, we conclude that both copies of Z␣ contribute to the binding site, as has been suggested previously (9). These experiments support the previous interpretation of data about Zab, in which both motifs form a single binding site (12). Within this binding site, Z␣ contributes contacts that assure conformation specificity, whereas Z␤ contributes additional sequence specificity.
Construction of DNA-binding Domain Expression Clones-Clones expressing Za and Zab were constructed as described previously (12). To separate the effects of Z␣ and Z␤ contacts with DNA, an additional construct was made. In Zaa, the Z␤ motif is replaced with a second copy of Z␣ (see Fig. 1). To construct Zaa, the expression plasmid for Za, pZa77, was cleaved with NcoI and NdeI, resulting in linear DNA extending from the NdeI site at the 5Ј end of the Za gene to the NcoI site in the pET28a polylinker. DNA coding for Z␣ and the linker region was amplified using pZab236 as a template and the T7 promoter primer and 5Ј-GAGCGGATTAATAAACTCAAGAGGATCTTCC-3Ј primer. The latter primer contains complementary nucleotides for residues 287-293 in hADAR1, which are located in the junction between the linker repeat and Z␤ (see Fig. 1). The AseI site (underlined above) was used to generate NdeI compatible ends. The amplified DNA was cleaved with NcoI, which cuts within the polylinker region, and with AseI. When this DNA is ligated to NcoI-NdeI-cut pZa77, the resulting plasmid includes the restored pET28a sequences connected to Z␣, the linker region, and a second Z␣. The sequences of all plasmids were confirmed by sequencing (data not shown).
Expressed protein was purified, and the binding activity was verified by gel mobility shift assay and BIAcore measurements (data not shown). Zaa was found to bind Z-DNA tightly and specifically. In gel mobility shift assays, one major high molecular weight complex is seen, similar to that seen for Zab. In the case of Zaa, the expression level and the amount of intact size Zaa expressed was comparable with those of Zab. This suggests that Zaa is likely to adopt a well folded structure equivalent to Zab. This observation is further supported by the results from limited proteolysis, which are consistent with Zaa being stably folded into a structure very similar to that of Zab (data not shown).
Construction of Endonucleases Including Z-DNA-binding Domains-The construction of pET15b:Z␣ nuclease, the plasmid expressing Z␣F N, has been described previously (11). The three new nucleases were constructed by replacing the Z␣ region in this construct with the appropriate other DNA-binding domain as follows. pET15b:Z␣ nuclease was cleaved with NcoI and XbaI to produce linear DNA extending from the XbaI site 5Ј to the endonuclease domain to the NcoI site in the polylinker. pZa77, pZab236, and pZaa were used as templates to amplify DNA, with T7 primer and 3Ј primer containing a XbaI site that is homologous to the region coding for the C terminus of each of the Z-DNA-binding peptides, respectively. The amplified DNA was cut with NcoI, which cuts in the polylinker region, and XbaI, which cuts within the 3Ј primer. NcoI and XbaI-cleaved pET15b:Z␣ nuclease and amplified DNA were ligated to produce three constructs; each can be expressed as a Z-DNA-binding fusion nuclease (F N ). The nucleases are called ZaF N , ZaaF N , and ZabF N . Fig. 1 shows the schematic diagrams of fusion nuclease constructs including Z-DNA-binding peptides.
Expression and Purification of Fusion Nucleases-The expression of fusion nucleases has been described previously (11). Briefly, proteins were expressed in E. coli strain BL21(DE3) and were found in the supernatant after lysis. His-tagged proteins were enriched using His-Bind resin (Novagen). Fusion nucleases were further purified on MonoS cation exchange columns (Amersham Pharmacia Biotech) and, if necessary, by Superose 12 (Amersham Pharmacia Biotech) gel filtration chromatography. Nucleases were concentrated and stored at Ϫ80°C in 50% glycerol. Protein concentrations were determined spectroscopically by measurements of the absorbance maximum near 280 nm. Extinction coefficients were calculated according to Gill and Von Hippel (14).
Cleavage of Plasmid Substrates-Plasmids were isolated from XL1 Blue using the alkaline lysis method followed by Wizard Plus Mini-or Midiprep (Promega). DNA was further extracted with phenol/CHCl 3 to remove residual RNase. DNAs were then precipitated and dissolved in TE (10 mM Tris-Cl, 1 mM EDTA, pH 8.0). The concentration of DNA was determined by using ⑀ 260 (1 mg/ml) ϭ 20. More than 90% of the plasmids were negatively supercoiled, as judged by migration on a 1% agarose gel.
Digestion of plasmid DNA with fusion nucleases were carried out under two different salt concentrations. Digestion were performed in the presence of 10 mM Tris-Cl, pH 8.0, 3 mM dithiothreitol, 50 g/ml bovine serum albumin, 200 g/ml E. coli tRNA, and 5% glycerol. High salt buffer contained 75 mM NaCl, whereas low salt buffer contained 15 mM NaCl. Unless otherwise specified, all reactions were performed in high salt. The reactions were preincubated with nucleases for 20 min at 22°C prior to adding 10 mM MgCl 2 in the case of high salt digestions or 2.5 mM MgCl 2 for low salt digestions. The reactions were incubated for an additional 4 h at 22°C in the presence of MgCl 2 and were terminated by heating at 50°C for 30 min. A second nuclease, PstI, was added to allow the mapping of specific cleavage sites. The reaction was incubated an additional 3 h at 37°C. Proteinase K was added to a final concentration of 50 g/ml and incubated at 37°C for 1 h. The results were analyzed using agarose gel electrophoresis.
Primer Extension-The DNA fragments for primer extension were isolated using the Qiaex II gel extraction kit (Qiagen). The two primers for primer extension, EcoRI primer, 5Ј-GTATCACGAGGCCCT-3Ј (#1204), and SalI primer, 5Ј-AGTCATGCCCCGCGC-3Ј (#1208), were purchased from New England Biolabs. The primers were end-labeled with T4 polynucleotide kinase and [␥-32 P]ATP. The reaction contained 10 -20 ng of DNA, 6. sequenase (Amersham Pharmacia Biotech) in a total volume of 8 l. A single cycle of 4 min at 95°C, 1 min at 55°C, and 1 min at 72°C was used for primer extension. The reaction was quenched with 4 l of stop buffer (95% formamide, 20 mM EDTA, 0.05% bromphenol blue, 0.05% xylene cyanol FF). The results were analyzed by denaturing acrylamide gel electrophoresis, using sequencing lanes as markers.

Za and Zab Show Differences in the Specificity of Bind-
ing-Za and other similar constructs containing Z␣ bind to a variety of sequences in the Z conformation. At micromolar concentrations, Za is able to stabilize oligonucleotides with either (dCdG) n or (dCdA/dTdG) n sequences in the Z conformation equally well (9,12). In contrast, Zab binds poorly, if at all, to (dCdA/dTdG) n sequences. Because this occurs even at high concentrations of Zab, this binding may reflect true sequence specificity rather than energetic considerations.
It is possible to test the sequence specificity of Zab by using DNA substrates in which Z-DNA is stabilized by something other than protein binding. This is accomplished in supercoiled plasmids containing appropriate inserts (13). To compare the binding of Za and Zab, fusion nucleases were constructed with each of these proteins. A previously described Z␣ fusion nuclease, containing a Z-DNA-binding region more susceptible to proteolytic digestion (12), has been shown to cleave supercoiled plasmids containing (dCdG) 13 inserts (11). The newly constructed nucleases, ZaF N and ZabF N , respectively, both cleave the (dCdG) 13 -containing plasmid, pDHg16, to a similar degree, as shown in Fig. 2. Fig. 2 also shows the cleavage of pDHf14, a homologous plasmid with a (dCdA/dTdG) 30 insert. Although the extent of cleavage is small, it can be seen that ZaF N cleaves this plasmid at concentrations between 200 and 100 nM. The limited cleavage at the (dCdA/dTdG) 30 insert with ZaF N as compared with (dCdG) 13 may be accounted for because only some of the plasmids contain Z-DNA at bacterial superhelical density. A higher torsional free energy is required to cause a B-Z transition for (dCdA/dTdG) 30 than for (dCdG) 13 (13). In contrast to ZaF N , ZabF N shows less cleavage of pDHf14 in all concentrations. ZabF N cuts much less efficiently near the (dCdA/dTdG) 30 insert than ZaF N , although both nucleases cut (dCdG) 13 in similar amounts, supporting the hypothesis that Zab binds to Z-DNA in a sequence-specific manner. This sequence specificity is unlikely to result from contacts between the linker region, which is not conserved in ADAR1 from different species. Instead contacts between Z␤ and DNA may be implicated. It is also possible that the sequence specificity may result from altering the position of the DNA with respect to Z␣ in the larger construct, thereby altering the nature of the binding.
A Fusion Nuclease, ZaaF N , in Which Z␣ Takes the Place of Z␤, Is Very Active in Cleaving at Z-DNA and Shows No Sequence Specificity-Z␤ was first identified by its sequence similarity to Z␣ (1). 17 of 72 amino acid residues are identical, and 20 are conservative changes. If the differences between the binding of Z-DNA by Za and Zab are the result of rearranging the protein-DNA interaction, then replacing Z␤ with Z␣ in Zab may have little or no effect. On the other hand, if contacts made by Z␤ are responsible for these differences, such replacement should result in a peptide that combines the affinity of two Z␣s but remains sequence nonspecific. This replacement was carried out as described under "Experimental Procedures" to produce Zaa and ZaaF N . Zaa could be expressed and purified under identical conditions used for Za and Zab. Fig. 3 shows a comparison of the cleavage by ZaF N , ZabF N , and ZaaF N , as well as the previously characterized Z␣F N . Plasmids containing a (dCdG) 13 insert were incubated with enzymes and then subsequently cut with PstI. Two diagnostic bands (open arrows) are produced when the plasmid is also cleaved next to the Z-DNA insert by the fusion nuclease. Z␣F N cleaves at the insert but also produces a number of other bands as the result of cleavage at non-Z-DNA sites. ZaF N cleaves more effectively, as shown by the increased amount of specific product. Some nonspecific cleavage can also be detected but considerably less than that seen for Z␣F N . The amount of cleavage seen for ZabF N is roughly equivalent to that seen for ZaF N ; however, the nonspecific product bands are absent, and the overall level of nonspecific cleavage is dramatically reduced. ZaaF N shows increased activity; even at the lowest enzyme concentration, more than 50% of the substrate is specifically cleaved. Like ZabF N , ZaaF N shows decreased nonspecific cleavage.
Unlike ZabF N , ZaaF N cleaves at Z-DNA formed by different sequences (Fig. 4). Like ZaF N , ZaaF N cleaves a plasmid with a (dCdA/dTdG) 30 insert, albeit only to a limited extent. Both Z␣F N and ZaF N produce noticeable degradation of this plasmid at nonspecific sites, as visualized by the other bands near the full-length band. This nonspecific cutting is more pronounced on the (dCdA/dTdG) 30 -containing plasmid than on the same plasmid without the insert (Fig. 4, pDHf14). The explanation for this is unknown; however, it may be the result of an interaction between Z␣ or Za and the insert sequence, which is of insufficient duration or affinity to allow specific cleavage but still enough to raise the level of background cleavage.
Two Copies of a Z-DNA-binding Motif Result in Increased Conformational Specificity; Z␤ Is More Effective than Z␣-As demonstrated above, both ZaF N and Z␣F N cut plasmids with less specificity than either ZabF N or ZaaF N in high salt condi- tion. This is the result of lower specificity of binding by peptides containing only a single motif, as verified in direct measurements of binding by gel mobility shift assay and BIAcore (data not shown). In gel shift assays, Zab binds equally well in the presence or absence of 1,000-fold B-DNA competitor; the protein can be pre-incubated with the competitor without effect. In contrast, Z␣ and Za show some inhibition of binding in the presence of 1,000-fold B-DNA competitor. This inhibition is significantly enhanced by pre-incubating the protein with the B-DNA before the Z-DNA probe is added (data not shown).
To further demonstrate this difference, plasmid containing the (dCdG) 13 insert was cleaved with each fusion nuclease under low salt conditions. Lowering the salt enhances cleavage. However, pilot experiments with Z␣F N showed a high level of nonspecific cleavage; therefore, low salt conditions were considered suboptimal for that system. Fig. 5 shows the result of using the low stringency conditions. Z␣F N produces almost no specific product at any concentration. At 100 nM, the plasmid is found as small rapidly migrating nonspecific fragments at the bottom of the gel. The nuclease containing a more structurally defined binding domain, ZaF N , produces specific products at 50 nM, but the background remains high, and further dilution results in the loss of all activity. In contrast, ZaaF N has less nonspecific activity than Za at 100 nM and cuts with good specificity at 50 and 25 nM. Zab has the largest difference in binding specificity. The total activity of ZabF N is less than that seen for ZaaF N ; however, even at 100 nM the cleavage shows some specificity.
Z␣, Za, Zaa, and Zab All Direct Cleavage to the Same Sites- Fig. 6 shows the mapping of the plasmid cleavage at nucleotide resolution using each of the nucleases discussed. Cleaved plasmids were used as templates for primer extension. Two primers were used, one for each strand. Cleavage by the fusion nuclease results in preferential stop sites, which are mapped relative to sequencing lanes using the same primers. The experiments shown in Figs. 2-5 reveal only double-stranded breaks, whereas primer extension experiments show both double-and single-stranded breaks. Fig. 6 (A and B) shows the results using pDHg16, the plasmid with the (dCdG) 13 insert. The primer is positioned at the bottom in both cases. The overall pattern of cleavage is similar for all the nucleases. Strong sites map between 5 and 20 bases outside the region of alternating purine-pyrimidine sequence. This is in agreement with studies of other fusion nucleases, which show cleavage 1-2 helical turns away from the binding site for Ubx and zinc finger nucleases (15,16). The heterogeneity of sites may in part be the result of heterogeneity of B-Z junctions in this plasmid, as shown by Johnston and Rich (17). No cleavage is seen within the region of Z-DNA. Such cleavage might be expected if the binding domain were situated at random within the insert; however, the simplest interpretation is that the FokI nuclease, like many others, is unlikely to cut within Z-DNA (11). The differences in the extent of cleavage with the different nucleases are not unexpected in light of the results shown in Fig. 3. Sites on both sides of the insert can be seen using the SalI primer (Fig. 6B); using the EcoRI primer, only upstream sites are evident (Fig. 6A). This may be the result of extensive upstream cutting, as will be discussed for ZaaF N .
ZaF N and ZabF N show nearly identical patterns, with preferred sites visible on both sides of the insert in both directions. The total extent of cleavage is moderate. There are some differences in intensity between different sites when these two nucleases are compared. ZabF N cleaves distant sites more strongly than the other nucleases (Fig. 6, A and B, designated by asterisks). These differences are reproducible. They may be the result of differences in the size of the binding domain construct and the position on Z-DNA. It is also possible that the difference in preference reflects a difference in ability to bind to junction sequences that may adopt the Z-conformation.
ZaaF N cuts at the same sites as the other nucleases. Only sites upstream of the insert are visualized with the EcoRI primer (Fig. 6A), and downstream sites are underrepresented with the SalI primer as well. This can be explained if the higher level of cleavage with this nuclease results in plasmid molecules that are multiply nicked or cut. Such plasmids will only template for primer extension to the major upstream sites. In addition, cleavage with ZaaF N produces several additional bands, mapping very close to the B-Z junction (Fig. 6B, open  box). Perhaps the binding of Zaa to Z-DNA is stable enough to allow less favorable, slower cleavage by FokI nuclease very close to the Z-DNA or the B-Z junction. Fig. 6 (C and D) shows the map of cleavage sites around a (dCdA/dTdG) 30 insert. Only ZaaF N cuts well enough for data to be collected. Cutting is shown flanking the insert and even encroaching into it. The superhelical density of the plasmid is probably not sufficient to maintain the entire insert in the Z form (17). In addition, there is a notable asymmetry of cleavage for the (dCdA/dTdG) 30 insert (Fig. 6E). Cleavage is much stronger 5Ј to (dCdA) and 3Ј to (dTdG). This cleavage is likely to reflect an asymmetric binding between Z␣ and this DNA sequence, with both copies in Zaa aligned to reinforce this asymmetry.

DISCUSSION
Fusion nucleases are potent tools to characterize the DNA binding of protein domains. Cleavage by the nuclease moiety is a measure of the specificity and affinity of the binding domain (11,15,18,19). Here we characterize Z-DNA binding, using nucleases that combine the endonuclease domain of FokI restriction enzyme with various constructs from the N-terminal end of hADAR1. This region has been shown to bind Z-DNA with high affinity and specificity (1), and the domain structure has been mapped (12). Here we examine the role of subdomains in the binding to Z-DNA.
Any of several constructs containing the Z-DNA-binding motif Z␣ can bind to Z-DNA. High affinity binding does not require the entire subdomain, but Za binding is more Z-DNA-specific than that of truncated constructs. The complete domain, Zab, contains both Z␣ and Z␤ motifs with the intervening linker and binds with greater specificity as shown in experiments with B-DNA competitor. It also appears to be more selective for sequence. Our experiments make it unlikely that the specificity for sequence is the result of a requirement for different binding energies needed to stabilize different sequences in the Z form. These experiments were conducted using supercoiling to stabilize the Z conformation, a situation thought to reflect what occurs in vivo.
It has been shown previously that Z␣ provides most of the binding affinity in Zab (11). When Zab is bisected with chymotrypsin to separate Z␣ and Z␤ containing subdomains, the resulting DNA binding is identical to that of the Z␣-containing piece alone. This result is supported by our demonstration that when Z␤ is replaced in Zab with Z␣, the resulting construct Zaa binds much more tightly than either Za or Zab. The binding is appropriate for the additive effects of two copies of Za. Further, it had been suggested that Z␤ might modify the binding of Z␣ to produce the sequence specificity of Zab. Again, the results from Zaa support this theory; Zaa binds to Z-DNA of different sequences, as does Za. On the other hand, there is evidence that Zaa has sequence specificity; when the cleavage sites are mapped on the nonpalindromic substrate, (dCdA/dTdG) 30 , there is an asymmetry of cleavage. It is possible that Zaa makes some base-specific contacts that result in this bias.
The specific cleavage sites in a plasmid containing a Z-DNA insert are generally the same using ZaF N , ZabF N or ZaaF N . However, some differences in the frequency with which each site is used are observed. ZabF N prefers sites that are further from the edge of the insert, whereas ZaF N and ZaaF N prefer the nearer sites. This preference may be explained by differences in the C-terminal residues between Za and Zaa, which are the same, and Zab. These residues may orient the nuclease domain differently, such that it more easily makes contact with substrate DNA at different sites. Such a result of altered C-terminal residues has been shown previously for two zinc finger DNA-binding domains in fusion nuclease constructs (16,20). In addition, ZaaF N has unique cleavage sites, very close to the edge of the insert. These sites may result from the tighter binding seen for this construct, which could allow the nuclease to cleave at less favorable spots.
Different binding domains may favor slightly different positions on Z-DNA. Possibly, these positions reflect different B-Z junction points. However, it is likely that Z␣ and Z␤ are positioned in Zab similarly with respect to the DNA. When Z␤ is replaced with Z␣, the resulting construct remains stably folded without major changes to the structure, as shown by limited proteolysis (data not shown). The second copy of Z␣ probably binds Z-DNA at a position similar to that of Z␤.
One goal of research into fusion nucleases containing a Z-DNA-binding domain is the need for useful reagents to identify Z-DNA in vivo. The original construct, Z␣F N , showed some promise, but the specificity and affinity were not high enough to be useful. ZabF N , which demonstrates the highest specificity for the Z conformation, may prove useful; however, the sequence specificity is likely to limit its usefulness. ZaaF N is a more likely candidate for an all-purpose reagent to detect Z-DNA. It is capable of cleaving supercoiled (dCdG) n nearly to completion. It can recognize (dCdA/dTdG) n and mixed sequences (data not shown) in the Z form. Future experiments expressing ZaaF N in eukaryotic cells and examining both reporter genes and endogenous genes may prove of interest.