Several features of the GT-factor trihelix domain resemble those of the Myb DNA-binding domain.

It is generally believed that GT-factors, DNAbinding proteins with specificity for GT-elements, are present only in plants. However, the current study shows that several features of the GT-factor trihelix domain resemble those of the Myb DNA-binding domain. The GT-element was first identified as a regulatory component of a light up-regulated element (Box II or GT-1 binding site), which is present within the upstream region of pea rbcS-3A (for review, see Zhou, 1999). Subsequently, sequences similar to the GT-1 binding site were found in the regulatory DNA sequences of many plant genes. Many analyses have shown that GT elements have diverse and important functions. Affinity screening using GT-1 binding sites has isolated the GT-binding factor, named GT-1a or B2F (Gilmartin et al., 1992; Perisic and Lam, 1992). These studies have postulated that three putative a-helices (trihelix domain) in GT-1a might be involved in DNA binding. Subsequent deletion analyses of GT-1a have shown that the trihelix domain of GT-1a is certainly involved in DNA binding (Hiratsuka et al., 1994; Lam, 1995). GT-2 was the first GT-factor to be isolated (Dehesh et al., 1990) and contains two separate trihelix domains (Ni et al., 1996). Each trihelix domain is involved in DNA binding. Additional members of this family have been isolated from the cDNA library (Smalle et al., 1998) or predicted from the genome sequence of Arabidopsis (Zhou, 1999). To date, GT-factors have been found only in plants. The Myb gene was originally identified as the oncogene (for review, see Introna et al., 1994). The c-Myb proto-oncogene has an important role in controlling the proliferation and differentiation of cells. The DNA-binding domain consists of three imperfect tandem repeats of 51 to 52 amino acids (R1, R2, and R3 from the N terminus). The solution structure of a specific DNA complex of the Myb DNA-binding domain has been determined (Ogata et al., 1994). Each R domain contains three helices. A hydrophobic core that includes three regularly spaced Trp residues maintains the trihelix structure (Ogata et al., 1992). The three regularly spaced Trp residues are characteristic of Myb proteins. More than 100 Myb genes have been found in Arabidopsis (Kranz et al., 1998). Myb proteins from animals have three Myb motifs (R1, R2, and R3), whereas most plant Myb proteins have two Myb motifs corresponding to R2 and R3 and forming the R2R3 Myb gene family. The proposed trihelix structure of GT-factors somewhat resembles the solution structure of the Myb DNA-binding domain. I therefore performed the sequence similarity search using the UCSC SAM-T98 program (http://www.cse.ucsc.edu/research/compbio/HMM-apps/T98-query.html) with default parameters (Karplus et al., 1998). This method uses an iterative hidden-Markov model. The proposed trihelix domain (from 66–153) of tobacco GT-1a (Gilmartin et al., 1992) was used as a query. The SAM-T98 search detected many Myb proteins (e.g. mouse c-Myb; E-value 5 7.9 3 10). This result shows that amino acid sequences of GT-factors and Myb proteins are related. A multiple alignment of homologous proteins was constructed based on the SAM-T98 alignment, as shown in Figure 1. In addition to the GT-factors (shown in red), the alignment includes several intermediate proteins (shown in black) between GT-factors and Myb proteins. Furthermore, the alignment includes maize myb protein C1 (PazAres et al., 1987) and mouse c-Myb protein (shown in blue). The alignment shows that many residues are well conserved between GT-factors and Myb proteins. The protein sequence-structure alignment, an approach also known as protein threading, provided convincing evidence that the GT-factor trihelix domain had a fold similar to that of the Myb DNAbinding domain. The GenTHREADER program (Jones, 1999; http://globin.bio.warwick.ac.uk/psiform.html) is able to more clearly discriminate between true and false positives than were the threading methods developed previously. A search using the same query predicted that the GT-factor trihelix domain contained a region that had a fold similar to those of the Myb DNA-binding domains (e.g. c-Myb R1 domain [probability of correct match, P 5 0.943], c-Myb R2 domain [P 5 0.930], and c-Myb R3 domain [P 5 0.829]). Furthermore, running a trihelix domain 1 This work was supported by grants from the Japanese Ministry of Education, Science, Sports and Culture. * E-mail [email protected]; fax 81–52–789 – 4296.