Next: Acknowledgements Up: Discussion Previous: Are metal ions

Secondary Structure Prediction and Domain Structure

The secondary structure prediction shown in Figure 1 suggests that the Ser/Thr phosphatases are made up of both both - helix and - strand. The most highly conserved region between eukaryotic and phage phosphatases, and E. coli diadenosine tetra-phosphatase comprises the first 143 positions in the alignment (Figure 1). It seems reasonable to suggest that this region makes up a single structural domain in the common phosphatase tertiary structure. The overall predicted secondary structure for the predicted domain is: ---(-/-)--, with ``*'' indicating the putative phosphate binding loops and parentheses showing uncertainty in the predicted state. This prediction is consistent with a single - sheet with - helices packing on either face. Positions 144-356 probably comprise a further structural domain but with a higher proportion of - helix and the summary secondary structure (-/-)--(-)--.

In common with all predictions, our prediction for the Ser/Thr phosphatases is likely to contain errors. Accordingly, we have presented the evidence for and against the assigned state of each predicted region. These data will provide a valuable guide for those planning site-directed mutagenesis experiments to probe the functional specificity of the phosphatases. Consideration of our results should avoid uneccesary effort performing mutations that are likely to disrupt the hydrophobic core of the protein. For example, mutations that alter the length of the protein chain should be restricted to regions strongly predicted as loop, while point mutations of conserved hydrophobic residues in predicted secondary structures should be avoided.

Protein Ser/Thr phosphatases are under investigation by X-ray crystallography in several laboratories. The prediction presented here may be helpful in guiding the interpretation of low resolution electron density maps, particularly when fitting the protein sequence to the observed density.

Recently, a number of blind tests of secondary structure prediction methods have been performed (for a recent review see [63]). Our prediction for the phosphatases is another in this series. Accordingly, when the first structure of a protein phosphatase is solved, comparison of prediction with experiment will provide further guidance for the development of improved secondary structure prediction methods.



Next: Acknowledgements Up: Discussion Previous: Are metal ions


gjb@bioch.ox.ac.uk