Thermodynamic Model for B-Z Transition of DNA Induced by Z-DNA Binding Proteins

Z-DNA is stabilized by various Z-DNA binding proteins (ZBPs) that play important roles in RNA editing, innate immune response, and viral infection. In this review, the structural and dynamics of various ZBPs complexed with Z-DNA are summarized to better understand the mechanisms by which ZBPs selectively recognize d(CG)-repeat DNA sequences in genomic DNA and efficiently convert them to left-handed Z-DNA to achieve their biological function. The intermolecular interaction of ZBPs with Z-DNA strands is mediated through a single continuous recognition surface which consists of an α3 helix and a β-hairpin. In the ZBP-Z-DNA complexes, three identical, conserved residues (N173, Y177, and W195 in the Zα domain of human ADAR1) play central roles in the interaction with Z-DNA. ZBPs convert a 6-base DNA pair to a Z-form helix via the B-Z transition mechanism in which the ZBP first binds to B-DNA and then shifts the equilibrium from B-DNA to Z-DNA, a conformation that is then selectively stabilized by the additional binding of a second ZBP molecule. During B-Z transition, ZBPs selectively recognize the alternating d(CG)n sequence and convert it to a Z-form helix in long genomic DNA through multiple sequence discrimination steps. In addition, the intermediate complex formed by ZBPs and B-DNA, which is modulated by varying conditions, determines the degree of B-Z transition.


Introduction
Left-handed Z-DNA is a higher energy conformation than right-handed B-DNA. Z-DNA was first found in a polymer of alternating d(CG) n DNA duplexes observed in high salt conditions [1]; its crystal structure was reported in 1979 [2]. The Z-DNA helix is built from d(CG)-repeats, with the dC in the anti-conformation and the dG in the unusual syn-conformation, which causes the backbone to follow a zigzag path [3,4]. Z-DNA can also be stabilized by negative supercoiling generated behind a moving RNA polymerase during transcription [5].
A distinct biological function of Z-DNA is suggested by the discovery of various Z-DNA binding proteins (ZBPs). Double stranded (ds) RNA deaminase 1 (ADAR1) deaminates adenine in pre-mRNA to yield inosine, which codes as guanine [6][7][8]. ADAR1 has two left-handed Z-DNA binding domains (ZBDs), Zα and Zβ, at its NH 2 -terminus [7,9]. High binding affinity of this Zα domain to Z-DNA was shown by a band-shift assay and confirmed by CD and Raman spectroscopic measurement [10][11][12]. The DNA-dependent activator of IFN-regulatory factors (DAI; also known as ZBP1 or DLM-1) also contains two tandem ZBDs (Zα and Zβ) at the NH 2 -terminus, such as ADAR1 [13,14]. It has been shown that ZBDs regulate the localization of DAI and its association with stress granules [15,16]. All poxviruses have a gene called E3L that consists of two domains: An N-terminal ZBD and a C-terminal RNA binding domain [17,18]. This ZBD shows sequence homology to the Zα domains found in human ADAR1 and in the DAI of mammals (Figure 1a). The Z-DNA binding affinity of E3L protein is essential for pathogenesis in the poxviruses [17][18][19]. The RNA-dependent protein kinase (PKR) plays an important role in the innate immune response against viral infections by recognizing dsRNA in the cytosol [20][21][22]. In fish species, a functional analogue of PKR, PKZ contains two ZBDs instead of dsRNA binding domains [23][24][25][26]. Similar to PKR, the phosphorylation function of PKZ is activated by Z-DNA binding [24].
These biological data, along with the results of more recent structural studies of Z-DNA induced by various ZBPs, have provided insights into Z-DNA recognition and ZBP-induced B-Z transition carried out by the innate immune response, viral infection, and RNA editing that are influenced by the nature of the Z-DNA. Crystallographic and NMR studies have provided detailed three-dimensional (3D) structures and dynamic information of various ZBPs complexed with Z-DNA. The structural and dynamic data summarized in this review have yielded a rich understanding of the mechanisms by which ZBPs selectively recognize the d(CG)-repeat DNA sequences in genomic DNA and efficiently convert them to left handed Z-DNA to achieve their biological function. is essential for pathogenesis in the poxviruses [17][18][19]. The RNA-dependent protein kinase (PKR) plays an important role in the innate immune response against viral infections by recognizing dsRNA in the cytosol [20][21][22]. In fish species, a functional analogue of PKR, PKZ contains two ZBDs instead of dsRNA binding domains [23][24][25][26]. Similar to PKR, the phosphorylation function of PKZ is activated by Z-DNA binding [24]. These biological data, along with the results of more recent structural studies of Z-DNA induced by various ZBPs, have provided insights into Z-DNA recognition and ZBP-induced B-Z transition carried out by the innate immune response, viral infection, and RNA editing that are influenced by the nature of the Z-DNA. Crystallographic and NMR studies have provided detailed threedimensional (3D) structures and dynamic information of various ZBPs complexed with Z-DNA. The structural and dynamic data summarized in this review have yielded a rich understanding of the mechanisms by which ZBPs selectively recognize the d(CG)-repeat DNA sequences in genomic DNA and efficiently convert them to left handed Z-DNA to achieve their biological function.

hZα ADAR1 -Z-DNA Complex
In 1999, Alexander Rich and his colleagues first reported the crystal structure of the Zα domain of human ADAR1 (hZα ADAR1 ) complexed with a six-base-pair (bp), double-stranded (ds) DNA fragment d(TCGCGCG) 2 [7]. The monomeric hZα ADAR1 domain binds to one strand of the palindromic dsDNA, in which the conformation of the DNA substrate is very similar to the canonical Z-DNA structure. (Figure 1b) [7]. A second monomer binds to the opposite strand of DNA yielding two-fold symmetry with respect to the DNA helical axis (Figure 1b) [7]. The hZα ADAR1 domain has a compact α/β architecture containing a three-helix bundle (α1 to α3) and twisted antiparallel β sheets (β1 to β3) (Figure 1b). The arrangement of a β-hairpin with hydrogen-bonds (H-bonds) between the β2 (L185-A188) and β3 strands (P193-I197) is a common feature of helix-turn-helix (HTH) proteins with a/β topology [7]. Aliphatic residues from the three helices, together with the W195 in strand β3, form a hydrophobic core [7]. The NMR studies reported that the free hZα ADAR1 protein in the solution adopts a similar fold in its complex structure [9,27].
The contact of hZα ADAR1 with the Z-DNA strand is mediated through a single continuous recognition surface, which consists of residues from an α3 helix and a β-hairpin (β2-loop-β3, also called a β-wing) (Figure 1c) [7]. The electrostatic interactions in the complex are made between K169, K170, N173, R174, and Y177 in an α3 helix as well as between T191 and W195 in a β-hairpin and five consecutive phosphate-backbones of Z-DNA ( Figure 1c) [7]. K169 and N173 form the direct and water-mediated H-bonds to the dC3pdG4 and dG2pdC3 phosphates, respectively (Figure 2a) [7]. The N173A mutant displays the most dramatic decrease in Z-DNA binding affinity, suggesting that it plays an important role in the Zα function [27,28]. Similarly, the K169A mutant also has a significantly lower Z-DNA binding affinity than a wild-type protein [27,28]. K170 forms direct H-bonds to the dG4pdC5 and dC5pdG6 phosphates (Figure 2a) [7]. The K170A mutant binds to Z-DNA with a lower affinity than wild-type protein, but better than the K169A and N173A mutants [27,28]. Interestingly, R174 and T191 bind to the furanose oxygens of dG6 and dG2, respectively (Figure 2a) [7]. However, the R174A and T191A mutations have little effect on the B-Z transition activities of hZα ADAR1 [28].
In addition to polar interactions, the aromatic ring of Y177 and the side-chains of P192 and P193 make the important van der Waals interactions with Z-DNA (Figure 2a) [7]. Interestingly, Y177 displays CH-π interaction with the C8 position of dG4 (Figure 2a) [7]. The Y177A mutant, which is unable to form both H-bonding and hydrophobic interactions, exhibits a significantly low Z-DNA binding affinity [27]. The Y177I, and Y177F mutants, which are capable only of hydrophobic interactions, could bind better to Z-DNA than Y177A, but still worse than wild-type protein [17,27,28]. Furthermore, the aromatic ring of W195, which forms a water-mediated H-bond to the dG2pdC3 phosphate, is almost perpendicular to Y177 and positioned in the center of the hydrophobic core ( Figure 1c) [7]. The W195F mutant has 2-fold lower B-Z transition activity than wild type protein [28]. [31], (e) hZβDAI-dT(CG)3 [32], and (f) hZαADAR1-rU(CG)3 complexes [33]. Intermolecular H-bonds and van der Waals contacts are indicated by solid lines and open circles, respectively. The water molecules in key positions within the protein-DNA interface are indicated by ovals.
In addition to polar interactions, the aromatic ring of Y177 and the side-chains of P192 and P193 make the important van der Waals interactions with Z-DNA (Figure 2a) [7]. Interestingly, Y177 displays CH-π interaction with the C8 position of dG4 (Figure 2a) [7]. The Y177A mutant, which is unable to form both H-bonding and hydrophobic interactions, exhibits a significantly low Z-DNA binding affinity [27]. The Y177I, and Y177F mutants, which are capable only of hydrophobic interactions, could bind better to Z-DNA than Y177A, but still worse than wild-type protein [17,27,28]. Furthermore, the aromatic ring of W195, which forms a water-mediated H-bond to the dG2pdC3 phosphate, is almost perpendicular to Y177 and positioned in the center of the hydrophobic core ( Figure 1c) [7]. The W195F mutant has 2-fold lower B-Z transition activity than wild type protein [28].
The crystal structural study reported that hZαADAR1 is also able to bind to 6-bp Z-DNA duplexes with non-CG-repeat sequences, such as d(CACGTG)2, d(CGTACG)2, and d(CGGCCG)2 [34]. In these structures, N173, Y177, P192, P193, and W195 contribute to the recognition of Z-DNA-like CG-repeat DNA [34]. However, R174 and T191 did not show intermolecular interaction with Z-DNA in most structures [34]. Similarly, K169 and K170 are only in contact with Z-DNA within some structures of these complexes [34]. Thus, these four residues might play an important role in the sequence discrimination step for the B-Z transition of DNA.
The second Z-DNA binding domain of human ADAR1 (hZβADAR1) adopts a winged-HTH fold like hZαADAR1, with the addition of a C-terminal α4 helix (see sequence in Figure 1a) [35]. Superposition of the hZβADAR1 with the hZαADAR1 structure reveals that A327, A332, and I335 (corresponding to K169, R174, and Y177 in hZαADAR1, respectively) did not perform H-bonding to the backbone of Z-DNA. Mutagenesis studies of these residues have shown that all three residues are important for Z-DNA-binding by Zα proteins [17,27,28]. This study suggested that hZβADAR1 is unable to interact with nucleic acids in a manner similar to that seen in the hZαADAR1-Z-DNA complex [35]. Instead, this region participates in self-association protein-protein interactions [35]. 3 [29], (c) yabZα E3L -dT(CG) 3 [30], (d) caZα PKZ -dT(CG) 3 [31], (e) hZβ DAI -dT(CG) 3 [32], and (f) hZα ADAR1 -rU(CG) 3 complexes [33]. Intermolecular H-bonds and van der Waals contacts are indicated by solid lines and open circles, respectively. The water molecules in key positions within the protein-DNA interface are indicated by ovals.
The crystal structural study reported that hZα ADAR1 is also able to bind to 6-bp Z-DNA duplexes with non-CG-repeat sequences, such as d(CACGTG) 2 , d(CGTACG) 2 , and d(CGGCCG) 2 [34]. In these structures, N173, Y177, P192, P193, and W195 contribute to the recognition of Z-DNA-like CG-repeat DNA [34]. However, R174 and T191 did not show intermolecular interaction with Z-DNA in most structures [34]. Similarly, K169 and K170 are only in contact with Z-DNA within some structures of these complexes [34]. Thus, these four residues might play an important role in the sequence discrimination step for the B-Z transition of DNA.
The second Z-DNA binding domain of human ADAR1 (hZβ ADAR1 ) adopts a winged-HTH fold like hZα ADAR1 , with the addition of a C-terminal α4 helix (see sequence in Figure 1a) [35]. Superposition of the hZβ ADAR1 with the hZα ADAR1 structure reveals that A327, A332, and I335 (corresponding to K169, R174, and Y177 in hZα ADAR1 , respectively) did not perform H-bonding to the backbone of Z-DNA. Mutagenesis studies of these residues have shown that all three residues are important for Z-DNA-binding by Zα proteins [17,27,28]. This study suggested that hZβ ADAR1 is unable to interact with nucleic acids in a manner similar to that seen in the hZα ADAR1 -Z-DNA complex [35]. Instead, this region participates in self-association protein-protein interactions [35].

mZα DLM1 -Z-DNA Complex
The amino acid sequence of the Zα domain of murine DLM-1 (mZα DLM1 ) is~35% identical to that of the hZα ADAR1 (Figure 1a). The overall structures of the complexes with Z-DNA are very similar to each other [29]. The core of the Zα-DNA interface in both proteins consists of three identical residues: N173, Y177, and W195 in hZα ADAR1 and N46, Y50, and W66 in mZα DLM1 , respectively (Figure 2a,b) [7,29]. K43 and Q47 form direct or water-mediated H-bonds to three phosphates of the dC3pdG4pdC5pdG6 sequence like the corresponding K170 and R174 of hZα ADAR1 (Figure 2b) [29]. The structural differences between the two domains are found in the α1-β1 loop and the β-hairpin [29]. The β-hairpin of mZα DLM1 is two residues shorter than that of hZα ADAR1 (Figure 1a), indicating that the β-hairpin is apparently tolerant of greater sequence variability than the α3 helix without loss of function [29].

yabZα E3L -Z-DNA Complex
The Zα domain of the Yaba-like disease virus E3L (yabZα E3L ) stabilizes the Z-DNA conformation in a manner similar to that of hZα ADAR1 and mZα DLM1 [30], although it shares only 26% sequence identity with hZα ADAR1 (Figure 1a). The crystal structural study revealed that two yabZα E3L domains are found in the asymmetric unit, each bound to one strand of double-stranded DNA in the Zconformation [30]. The intermolecular interaction of one asymmetric unit with Z-DNA is summarized in Figure 2c [30]. Three residues, N47, Y51, and W69 (corresponding to N173, Y177, and W195 in hZα ADAR1 ), play central roles in the interaction with Z-DNA, as with other members of the Zα family ( Figure 2c). K43, K44, and Q48 also participate in DNA recognition via direct or water-mediated H-bonds to the phosphate backbone of Z-DNA (Figure 2c).

caZα PKZ -Z-DNA Complex
The Zα domain of PKZ from Carassius auratus (caZα PKZ ), which shows limited identity with other ZBPs (28% for hZα ADAR1 , 20% for hZα DAI , and 22% for yabZα E3L , respectively), is able to convert d(CG)-repeat DNA from B-DNA to Z-DNA [36,37]. The interaction between caZα PKZ and Z-DNA is mediated by five residues in the α3 helix and four residues in the β-hairpin, similar to other Zα proteins (Figure 2d) [31]. Unlike the positively charged Lys or Arg in other ZBPs, the S35 in the α3 helix forms electrostatic interaction with the dC3pdG4 phosphate (Figure 2d) [31]. Interestingly, K56 of caZα PKZ interacts not only with dC1pdG2 but also with dT0pdC1 ( Figure 2d) [31], whereas a polar residue, like Ser or Thr at the corresponding position in other mammalian ZBPs, could not form these interactions ( Figure 2). Generally, the B-Z transition activity by ZBPs was decreased when the ionic strength was increased. Surprisingly, the reduction of the B-Z transition rate is more severe in caZα PKZ than in hZα ADAR1 , suggesting that the effect of charge-charge interactions on B-to-Z transition activity plays a more critical role in the case of caZα PKZ [31].

hZβ DAI -Z-DNA Complex
The second ZBD of human DAI (hZβ DAI ) was also shown to bind Z-DNA based on its binding specificity for Z-DNA and its ability to convert B-DNA to Z-DNA [16]. Although hZβ DAI also has α/β topology with three helices packed against three β-stands, like other ZBPs, hZβ DAI has a 3 10 helix at the N terminus of α3, instead of the long continuous α3 helix [32]. In the hZβ DAI -Z-DNA complex, protein-DNA interactions are mediated by the most conserved core residues, N141, Y145, and W162 (corresponding to N173, Y177, and W195) [32]. However, except for those core residues, other interactions with Z-DNA seem to be different for hZβ DAI (Figure 2) [32]. For example, K138, located in the region between K169 and K170 of hZα ADAR1 , forms an H-bond to the dC3pdG4 phosphate of one Z-DNA strand, whereas the two Lys residues of hZα ADAR1 contact the 4 phosphate groups of Z-DNA ( Figure 2) [32]. Interestingly, K138 spans the length of the Z-DNA molecule and interacts with the dC5pdG6 phosphate on the opposite DNA strand (Figure 2e) [32]. An NMR study found that free hZβ DAI has notable alterations in the α3 helix, the β-hairpin, and Y145 which are critical in Z-DNA recognition [38]. These results indicate that, unlike some other Zα domains, structural flexibility of hZβ DAI is required for Z-DNA binding [38].

hZα ADAR1 -Z-RNA Complex
ADAR1 edits dsRNA in vitro at significantly higher levels when dsRNA contains the purine-pyrimidine repeat sequence in dsRNA [39]. The hZα ADAR1 protein can bind to Z-RNA like Z-DNA [40]. It was first reported that the crystal structure of hZα ADAR1 complexed with 6-bp dsRNA, r(UCGCGCG) 2 [33]. Interestingly, hZα ADAR1 exhibited significantly different binding modes when bound to Z-RNA versus Z-DNA ( Figure 2). First, in the Z-RNA binding conformation, Y177 showed H-bonding interaction with the rG2prC3 phosphate and the O2' of rG2, whereas it H-bonded with only the dG2pdC3 phosphate in the Z-DNA binding structure ( Figure 2) [7,33]. Second, in the Z-DNA binding structure, R174 showed a direct, water-mediated H-bonding interactions with the dC5pdG6 phosphate and the O4' of dG6, respectively (Figure 2a) [7]. However, when binding to Z-RNA, a water-mediated H-bond with the rC5prG6 phosphate as well as an H-bond with the E171 side-chain were formed (Figure 2f) [33]. Third, in the Z-RNA binding conformation, H159 exhibited a distinct orientation due to a water-mediated H-bonding interaction with the K169 side-chain compared to the Z-DNA binding conformation [33]. Fourth, in the Z-RNA binding conformation, T191 showed H-bonding interaction with only the rC3prG4 phosphate, whereas it formed H-bonds with both the phosphate and the O4' of dC3 in the Z-DNA binding structure ( Figure 2) [7,33].

B-Z Transition of a 6-bp CG-Repeat DNA by hZα ADAR1
NMR studies on the hZα ADAR1 -Z-DNA interaction first proposed the B-Z transition mechanism of a 6-bp DNA, d(CGCGCG) 2 , by hZα ADAR1 , in which the hZα ADAR1 plays two independent roles: (i) one molecule first binds to B-DNA and shifts the equilibrium from B-DNA to Z-DNA (BP to ZP, where B, Z, and P indicate B-DNA, Z-DNA, and protein); and (ii) the second molecule selectively binds to and stabilizes the Z-DNA conformation (Figure 3a) [41]. This study confirmed the existence of a one-to-one complex of Z-DNA and hZα ADAR1 (ZP) by gel filtration chromatography, NMR dynamics data, and diffusion coefficient values as functions of the [P] t /[N] t molar ratio, where [P] t and [N] t are the total concentrations of the hZα ADAR1 and DNA, respectively [41]. It was found that the Z-DNA produced was half the total amount of the added hZα ADAR1 when the [P] t /[N] t ratio was ≤2 (that is Z t = 1/2[P] t ), where Z t is the total amount of the Z-DNA conformation (Figure 3b where k ex,ZP and k ex,ZP2 are the exchange rate constants of the imino protons for the ZP and ZP 2 complexes, respectively, χ is the [P] t /[N] t ratio, and α (= K d,ZP2 /K d,BP ) is the ratio of the dissociation constants of BP and ZP 2 complexes [41]. The k ex dataset was fitted using Equation (1)  (a)

B-Z Transition of a 6-bp Non-CG-Repeat DNA by hZαADAR1
The hZαADAR1 protein can also convert the B-form of non-CG-repeat DNA, d(CACGTG)2 and d(CGTACG)2, to Z-form with lower activities compared to CG-repeat DNA, d(CGCGCG)2 ( Figure  3b) [41,42]. Equation (1) could not be used to analyze the hydrogen exchange data of these DNA complexed with hZαADAR1 because the non-CG-repeat DNA did not satisfy the relation, Zt

B-Z Transition of a 6-bp Non-CG-Repeat DNA by hZα ADAR1
The hZα ADAR1 protein can also convert the B-form of non-CG-repeat DNA, d(CACGTG) 2 and d(CGTACG) 2 , to Z-form with lower activities compared to CG-repeat DNA, d(CGCGCG) 2 (Figure 3b) [41,42]. Equation (1) could not be used to analyze the hydrogen exchange data of these DNA complexed with hZα ADAR1 because the non-CG-repeat DNA did not satisfy the relation, Z t = 1/2[P] t . Instead, the observed k ex value for the imino proton in the B-DNA (not Z-DNA) conformation could be expressed as a function of the relative Z-DNA population (f Z = Z t /[N] t ) by the following equation: where k ex,B and k ex,BP are the exchange rate constants of the imino protons for free B-DNA and the BP complex, respectively [42]. The NMR dynamics studies found that hZα ADAR1 binds to non-CG-repeat DNA with weak binding affinity through the α3 helix as well as through the loop-β1-loop (151-158) and the α3-loop-β2 regions (178-191) [46]. Then, the B-form helix of non-CG-repeat DNA duplexes can be converted to a Z-conformation via these multiple intermolecular interactions with hZα ADAR1 proteins [46]. These studies explained how hZα ADAR1 exhibited the sequence preference of d(CGCGCG) 2 >> d(CACGTG) 2 > d(CGTACG) 2 during the B-Z transition [42,46]. First, the P binds to the B, with a sequence preference of d(CGCGCG) 2 >> d(CACGTG) 2 > d(CGTACG) 2 [42], even though the structural features of these three DNA duplexes complexed with hZα ADAR1 are very similar to each other [34]. Second, the BP of d(CGCGCG) 2 and d(CACGTG) 2 convert to ZP. In d(CGTACG) 2 , however, this process is less efficient as a way of discriminating d(TA)-containing DNA sequences from alternating pyrimidine-purine sequences [42]. Third, the ZP of d(CGCGCG) 2 and d(CACGTG) 2 binds to the P and forms the stable ZP 2 complex with a sequence preference of d(CGCGCG) 2 >> d(CACGTG) 2 , which acts as the third sequence discrimination step [42]. Taken together, it was suggested that hZα ADAR1 selectively recognizes the alternating d(CG) n sequence and then converts it to a Z-form helix in long genomic DNA through its multiple sequence discrimination steps [42,46].

B-Z Transition of a 6-bp DNA by yabZα E3L
The yabZα E3L could efficiently change the B-form helix of the d(CGCGCG) 2 to left-handed Z-DNA like hZα ADAR1 (Figure 3c) [43]. In this study, because the B-Z transition activity of yabZα E3L did not satisfy the relation, Z t = 1/2[P] t , the observed k ex value for the imino proton in the Z-DNA conformation could be expressed by the following equation instead of Equation (1): The k ex dataset was fitted using Equation (3) to obtain the α value of 0.154 and the K BZ,1 value of 1.02 (Table 1) [43]. This K BZ,1 value means that yabZα E3L and hZα ADAR1 have the same B-Z transition efficiency [43], which is consistent with their structural similarity in complexes with Z-DNA [7,30].

B-Z Transition of a 6-bp DNA by hZβ DAI
NMR studies have revealed that hZβ DAI had significantly lower B-Z transition activity than hZα ADAR1 and yabZα E3L (Figure 3c) [44]. In addition, the imino proton and 31 P-NMR spectra of d(CGCGCG) 2 complexed with hZβ DAI are completely different from those of the d(CGCGCG) 2 -hZα ADAR1 complex [44]. These indicate that the base pair geometry and backbone conformation of the hZβ DAI -induced Z-DNA helix are significantly different from those of the Z-DNA-hZα ADAR1 complex, similar to their crystal structures [7,32]. The hydrogen exchange study of the d(CGCGCG) 2 -hZβ DAI complex found that the exchange rates of imino protons in B-DNA as well as Z-DNA conformations are not affected by complex formation [A4]. In addition, diffusion optimized spectroscopy experiments confirmed that the Z-form of d(CGCGCG) 2 complexed with hZβ DAI exhibited one major complex state (perhaps ZP 2 ), even at various [P] t /[N] t [44]. Based on these results, they proposed the distinct B-Z transition mechanism where two molecules of hZβ DAI initially bind directly to the B-form DNA and form the BP 2 complex; subsequently, there is a conformational change from BP 2 to ZP 2 . (Figure 3a) [44].

B-Z Transition of a 6-bp DNA by caZα PKZ
The caZα PKZ domain can convert the dsDNA, d(CGCGCG) 2 to Z-DNA with lower activity rates than hZα ADAR1 and yabZα E3L (Figure 3d) [45]. Instead, caZα PKZ exhibits full B-Z transition activity when binding to d(TCGCGCG) 2 (Figure 3d) [45]. This indicates that the H-bonding interaction of K56 with the dT0pdC1 phosphate plays an important role in the B-Z transition of DNA by caZα PKZ [45]. In this study, instead of the k ex value, the 1 H and 15 The closed-form solution of Equation (9) has been expressed by [45]: where θ = arccos In order to obtain accurate K d values, all 1 H and 15 N titration curves and the f Z data were globally fitted with Equation (4) and Equation (5), respectively. This approach successfully provides two binding constants, K d,BP and K d,ZP2 , not the relative ratio (α) of these two constants as in previous studies.
At 10 mM NaCl, the global fitting gave a K d,BP and a K d,ZP2 of 28 and 350 nM, respectively, and a K BZ,1 of 0.87 (Table 2) [45]. As [NaCl] was increased, the K d,BP and K d,ZP2 values became increased, but the K BZ,1 value became smaller (Table 2) [45,47]. The NMR dynamics studies found that increasing the ionic strength interferes more with the association of ZP with caZα PKZ via intermolecular electrostatic interactions rather than the dissociation of ZP 2 [45]. In addition, the global fitting method using Equation (4) also provides the 1 H and 15 N chemical shift differences between the free and the bound forms for both B-DNA (∆δ B ) and Z-DNA binding (∆δ Z ). At higher concentrations of NaCl, the B-DNA-bound state exhibited completely different results than at 10 mM NaCl, whereas the Z-DNA binding conformation was not affected by the change of ionic strength [45,47]. These results meant that the B-DNA binding state of caZα PKZ exhibited distinct structural features under high and low salt conditions which might be related to reduced B-Z transition activity at higher [NaCl]. Taken together, these studies suggest that the intermediate complex formed by caZα PKZ and B-DNA can be used as a molecular ruler to measure the degree to which DNA transitions to the Z isoform [45,47].

Conclusions
Z-DNA is induced by various Z-DNA binding proteins (ZBPs) that play important roles in RNA editing, innate immune response, and viral infection. We summarized the structural and dynamics data of various ZBPs complexed with Z-DNA to understand the mechanisms by which ZBPs selectively recognize d(CG)-repeat DNA sequences in genomic DNA and efficiently convert it to left handed Z-DNA to achieve their biological function. The contact of ZBPs with Z-DNA strands mediated through a single continuous recognition surface consists of an α3 helix and a β-hairpin. In the ZBP-Z-DNA complexes, three conserved identical residues of ZBPs (N173, Y177, and W195 in hZα ADAR1 ) play a central role in interactions with Z-DNA. ZBPs convert a 6-bp DNA to a Z-form helix via a B-Z transition mechanism in which the ZBP first binds to B-DNA and then shifts the equilibrium from B-DNA to Z-DNA, a conformation that is then selectively stabilized by the additional binding of a second ZBP molecule. During B-Z transition, ZBPs selectively recognize the alternating d(CG) n sequence and then convert it to a Z-form helix in long genomic DNA through its multiple sequence discrimination steps. In addition, the intermediate complex formed by ZBPs and B-DNA, which is modulated by varying conditions, determines the degree of B-Z transition.