The determination of presence/absence of Z-DNA binding domain in organisms was simply done by using GTOP database. GTOP is containing protein fold predictions based on homology search against protein sequences of known structure. If there was a homologous hit for the Z-DNA binding domain with an e-value less than 10⁻¹⁰, it is estimated that the organism has the Z-DNA binding domain. If there was no hit, it is considered that the Z-DNA binding domain is absent in the organism.

The amino acid sequences of Z-DNA binding domain of E. coli recA protein and that of human ADAR protein from GTOP are shown in Figure 1(a) and Figure 1(b), respectively. GTOP uses Swiss-Prot protein sequence database [22], so the Swiss-Prot codes are given for the sequences.

GTOP adopted SCOP classification of protein structures, the unit of classification is usually the protein domain. SCOP organizes protein structures according to evolutionary origin and structure similarity. Actually, protein domains are classified on hierarchical levels into four categories: class, fold, superfamily, and family. The 3D structure of recA protein from E. coli has a domain described as class: alpha and beta

protein, fold: p-loop containing nucleotide triphosphate hydrolases, superfamily: p-loop containing nucleotide triphosphate hydrolases, family: recA protein-like (ATPase-domain). This domain is described as c.37.1.11 in SCOP code, and this is used as a keyword in GTOP search. Another Z-DNA binding domain in ADAR from human is described as class: all alpha protein, fold: DNA/RNA-binding 3-helical bundle, superfamily: winged helix DNA-binding protein, and family: Z-DNA binding domain. This domain is expressed as a.4.5.19 in SCOP code and used as a keyword in GTOP search. As more genomic sequences become available, the survey of proteins becomes difficult without useful tools. GTOP has a tool of keyword search on the web. For example, we searched the Z-DNA binding domain in GTOP using c.37.1.11 as keyword, then the homologous proteins in an organism were displayed with e-values. Therefore, we can simply estimate the presence or absence of the Z-DNA binding domain.

We employed GTOP for the search of two types of Z-DNA binding domains. In GTOP, organisms are classified based on the annotation in the genome sequence according to hierarchy: three kingdoms (archaea, bacteria, and eukaryotes), phylum, and section. In GTOP, 68 organisms in archaea were divided into 5 phyla and 13 sections (Table 1(a)), 914 organisms in bacteria were divided into 21 phyla and 45 sections (Table 1(b)), and 199 organisms in eukaryotes were divided into 13 phyla and 21 sections (Table 1(c)).

The Z-DNA binding domain in recA protein from E. coli was observed in all the sections of archaea, bacteria, and eukaryotes in GTOP. Therefore, there is no need to distinguish the presence/absence of the Z-DNA binding domain in recA protein. This result indicated that this domain is essential for all the organisms.

Another Z-DNA binding domain in ADAR from human was observed in some organisms of archaea, bacteria, and eukaryotes, respectively. The representative organism in the column of organism in Tables 1(a)-(c) indicates the presence of Z-DNA binding domain in ADAR from human. The white space in the column of organism means the absence of this domain.

Comparisons of the ribosomal RNA sequences from various organisms are commonly used to deduce the phylogenetic trees [23]. The trees indicate clustered classification into three kingdoms, archaea, bacteria, and eukaryotes, and sub-clustered groups into phyla and sections according to their sequence similarities. The phylogenetic tree based on archaeal 16S small subunit ribosomal RNA sequences revealed that phylum Thaumachaeota may emerge before the divergence between Crenarchaeota and Euryarchaeota [24]. The presence of Z-DNA binding domain in the organisms belongs to the phyla Thaumachaeota, Crenarchaeota and Euryarchaeota suggested that the emergence of the Z-DNA binding domain was preceding to the branch between Crenarchaeota and Euryarchaeota (Table 1(a)). Four organisms, H. butylicus, A. fulgidus, T. onnurineus, and Candidatus K. cryptofilum are thermophiles, and M. burtonii and N. maritimus are mesophililes (Table 1(a)). This result suggested that the Z-DNA binding domain is favorable in thermophiles.

The evolutionary history of organisms of bacteria can be obtained by a comparison of conserved protein sequences of elongation factor-1 alpha/Tu or 70-kDa heat shock protein [25]. A clear separation of the Gram-positive and Gram-negative bacteria can be obtained. The phylum firmicutes only indicated the presence of Z-DNA binding domain in the Gram-positive bacteria (Table 1(b)), and the organisms have low G + C content. In Gram-negative bacteria, the phyla of aquificae, chlorobi, proteobacteria and thermotogae showed the presence of Z-DNA binding domain (Table 1(b)), and their G + C content ranged from 35% to 67% among the organisms. The organisms of P. marina, T. tengcongensis, Nitratiruptor sp., and F. nodosum are thermophiles.

The Z-DNA binding domain was observed only in the organisms belong to phylum metazoan, section eumetazoa in eukaryotes (Table 1(c)). It was interesting that only vertebrates indicated the presence of this domain and invertebrates indicated the absence of this domain. Mammalian genomes encode three ADAR genes, ADAR1, ADAR2 and ADAR3 [26 , 27]. ADAR1 contains two Z-DNA binding domains, but not ADAR2 nor ADAR3. The ADAR genes are present in Caenorhabditis elegans genome [28], Drosophila genome [29], squid nervous system [30], and their ADAR gene products have no Z-DNA binding domain [31]. The ADAR family catalyzes the conversion of adenosine to inosine in pre-mRNA, and the substrates require duplex RNA secondary structure. Adenosine to inosine editing modulates the calcium permeability of neural glutamate receptors [32] and reduces the G-protein coupling efficiency of serotonin 2C receptors [33].