MINIREVIEW Properties and Functions of GAF Domains in Cyclic Nucleotide Phosphodiesterases and Other Proteins

This review focuses primarily on our current understanding of the structure and function of GAF domains in cyclic nucleotide phosphodiesterases (PDEs). The GAF domain was originally identified by Aravind and Ponting (1997) using the position-specific iterative BLAST method (Altschul et al., 1997). The consensus sequence defining GAF domains is taken from a characteristic primary sequence of chemically conserved amino acids (AAs) that is associated with a particular pattern of predicted secondary structures. GAF domains occur in a variety of proteins and in association with a diverse collection of other functional domains. There are no invariant AAs in the predicted motifs that characterize these domains. The predicted GAF domains in closely related GAF-containing proteins contain sequence similarities; even among these, however, there are few invariant AAs. Among the 14 predicted GAF domains in human PDEs, a Phe is the only invariant amino acid. Twelve of the 14 predicted GAF domains contain the NK/RXnFX3DE signature sequence that we first described derived from the sequences in PDEs 2, 5, and 6 (McAllister-Lucas et al., 1995; Turko et al., 1996). Using site-directed mutagenesis of these AAs in PDE5, our laboratory demonstrated that each contributes importantly to the structural requirements for the allosteric cGMP-binding function in PDE5 (McAllister-Lucas et al., 1995; Turko et al., 1996). However, this Asn, Lys, and Asp arrangement occurs in GAF domains for which no ligand-binding function has been described, and its function is not understood. The GAF acronym is derived from the names of the first three classes of proteins recognized to contain this domain: mammalian cGMP-binding PDEs, Anabaena adenylyl cyclases, and Escherichia coli FhlA (Aravind and Ponting, 1997). There are now more than 1400 proteins in the nonredundant database that are predicted to contain a GAF domain (Schultz et al., 1998; Letunic et al., 2002; see http://smart. embl-heidelberg.de). GAFs have been shown to be associated with gene regulation in bacteria (Aravind and Ponting, 1997), light-detection and signaling pathways in plant and cyanobacterial phytochromes (Sharrock and Quail, 1989; Montgomery and Lagarias, 2002), ethylene detection and signaling in plants (Sato-Nara et al., 1999), nitrogen fixation in bacteria (Joerger et al., 1989), feedback control of a cyanobacterial adenylyl cyclase by cAMP-binding (Kanacher et al., 2002), and the two-component sensor histidine kinase in viruses, bacteria, and plants (Table 1) (Kaneko et al., 2001; Urao et al., 2001). Notably, sequences predicted to form GAF domains are found in PDEs from diverse organisms including trypanosomatids, nematodes, sponges, insects, and mammals (Koyanagi et al., 1998; Schultz et al., 1998; Letunic et al., 2002; Rascon et al., 2002; Zoraghi and Seebeck, 2002; see http://smart.embl-heidelberg.de). GAF domains are described as one of the largest families of small-molecule-binding regulatory (R) domains, but direct evidence of ligand binding has only been demonstrated for a very few GAFs. The first demonstration of GAF domain binding of ligand was binding of cGMP to mammalian PDE5 (Lincoln et al., 1976; Francis et al., 1980). Increasingly, roles other than ligand binding are being documented, and it seems likely that all of the functions provided by GAF domains have not yet been fully appreciated. Distribution of GAF Domains in PDEs. GAF motifs are composed of 110 AA and are present in one to four (and even partial) copies in all living organisms from archaea to mammals; they are particularly abundant in plants and bacteria (Table 1). To date, the predicted GAF domains are distributed as follows: archaea (45), bacteria (463), viruses (1), fungi (11), plants (432), arthropods (9), nematodes (2), This work was supported by National Institutes of Health research grants DK40299 and DK58277 and American Heart Association Postdoctoral Fellowship 032525B.