From Protein Structure to Function with Bioinformatics.pdf

12 Prediction of Protein Function from Theoretical Models 311Fig. 12.5 Catalytic site of E. coli 6-phosphofructokinase bound to fructose 6-phosphate (F6P)and adenosine diphosphate (ADP) (PDB code 4pfk). Ligands are shown as coloured sticks (F6Pon left, ADP on right). ATP-dependent enzymes, like that from E. coli, have a Gly at the catalyticsite (not shown). Modelling of an Asp residue at the same position (magenta), as found in PPidependentenzymes, shows that it is responsible for changing the specificity for phospho donor(see text)splicing events tend to lie both in coil regions (rather than within elements of regularsecondary structure) and at the protein surface. Splicing events were generally few innumber for a particular gene, 1 or 2, and small in size, with 60% affecting 50 residuesor fewer. These findings suggested that splicing tends to occur in positions and in waysthat perturb only minimally the protein tertiary structure, consistent with most alternativeisoforms having folding properties similar to the original form and thus potentialfunctionality. However, a later study (Tress et al. 2007), in which fewer transcriptswere analysed in structural terms, revealed that many alternatively spliced isoformswould have to have dramatically different structures to determined structures of otherisoforms. For fully 49 of 85 transcripts mapped onto homologous structures, theauthors inferred that isoform and principal sequences would adopt substantially differentstructures. An example, taken from Tress et al. (2007) and shown in Fig. 12.6,relates to an isoform of interleukin 4 lacking exon 2. The structural region encoded bythat exon contributes to both the folding core of the protein and to a disulphide bridge,showing that the 3D structure of the isoform must be substantially different to thedetermined structure of the complete protein. As yet, we have only a very incompletelarger scale picture of the functional consequences of structural changes – minor andmajor – due to alternative splicing. For example, for only 4 of 214 loci could experimentaldata illustrating functional differences between splice isoforms be found byTress et al. (2007).