Iiniin nsfi inf i n mrnmliniWifiiiiim Twtar

Fig. 1. Amino acid sequence comparison of Sec4p and Ypt1p. Loop2, Loop7, and hypervariable domains are indicated. The numbering is for Sec4p (gray box indicates gap in the alignment). Identical amino acids and divergent amino acids are indicated with black and white marks on the top and bottom rows beneath the ruler, respectively.

The algorithm we have developed is based on the method of Casari et al. (1995) who take a theoretical approach to classify residues according to their potential to provide specific function ("tree-determining" residues), without making any prior assumptions as to their positional location. This output of this method generates a large number of potentially important functional residues and can be combined with other considerations to create a smaller list of possibilities that can be experimentally addressed. One adaptation is to incorporate structural considerations, such as solvent accessibility and positions responsive to the nucleotide status of the GTPase (Bauer et al., 1999). We have made use of the adaptation of Heo and Meyer (2003), who incorporated experimental data to provide functional classifications. The main assumption of this algorithm, illustrated in Fig. 2, is that homologous proteins in two different functional classes have diverged from an ancestral gene. Each amino acid position

Low Poly Animals Mask Template

Fig. 2. Schematic illustrating evolutionary assumption of algorithm. The algorithm assumes that homologous proteins that share a common catalytic mechanism and overall fold but that have different cellular functions have diverged from a common ancestor. Residues, or loops that do not participate in the catalytic mechanism (black loops of the core domain), have evolved to provide unique functions A or B.

Fig. 2. Schematic illustrating evolutionary assumption of algorithm. The algorithm assumes that homologous proteins that share a common catalytic mechanism and overall fold but that have different cellular functions have diverged from a common ancestor. Residues, or loops that do not participate in the catalytic mechanism (black loops of the core domain), have evolved to provide unique functions A or B.

relevant for the divergent function of two classes has been under selective pressure to remain conserved within one functional class and to become divergent in the other functional class. The algorithm then calculates a ''conservation distance'' and a ''divergence distance'' for each amino acid position.

2. Each of the two functional classes is named according to convenience; in this example we use ''Ypt1p group'' and ''Sec4p group.'' Each group contains sequences that have been experimentally determined to provide the function of either class. For Ypt1p and Sec4p, we define function as the ability to complement a deletion of the gene, or to restore functionality in cells containing a thermosensitive allele of the gene. Such sequences have been acquired through a meta-analysis, with a literature search of genes that have been documented to functionally substitute for either YPT1 or SEC4 (Clement et al, 1998; Dietmaier et al., 1995; Dumas et al., 2001; Fabry et al., 1993; Haubruck et al., 1989; Pertuiset et al., 1995; Saloheimo et al., 2004). Accession numbers for the ''Ypt1p group'' are AAF33844, CRU13168, L08128, and TRE277108; and for the ''Sec4p group'' are AF015306, CLI272025, RYL1_YARLI, and VVCYPTV1. A global alignment file is then created using Clustal X (Thompson et al., 1997) with the core groups of amino acid sequences and sequences related to both Ypt1p and Sec4p that were identified through BLAST searches (total number of sequences = 114).

3. The global alignment file is then perused for potentially relevant residues. We include amino acid positions as potentially relevant for consideration only if they are conserved for all members in one core group and divergent from all members in the other core group (and vice versa). The conservation criteria used are amino acid identity, conservation of charge, and conservation of aromatic amino acids. An example demonstrating the residues considered potentially relevant is shown in Fig. 3A where the alignment of Loop2 between Ypt1p and Sec4p is shown. Out of a total of 12 amino acids, 7 are divergent between the two sequences and thus are potentially relevant amino acids. As can be readily appreciated, this is an impractical number of combinations to check experimentally. However, with the inclusion of the core groups, shown in Fig. 3B, it is immediately apparent, for instance, that residue S48 of Sec4p, which might otherwise attract attention because it encodes a neutral residue whereas the corresponding residue of Ypt1p is charged, is most likely insignificant. The reason for this is that in the other sequences that encode Ypt1p function (Haubruck et al., 1990), the corresponding residue is identical to that of Sec4p. Similar considerations would suggest that Sec4p N46 is most probably not functionally relevant and also perhaps K44 because the corresponding residue is not conserved among members of each group.

0 0

Post a comment