From Protein Structure to Function with Bioinformatics.pdf

134 P. TompaSequence-independence was also found in two other systems, linker histonesand prions. In the case of linker histones, the binding region of apoptotic nucleaseDNA fragmentation factor 40 (DFF40), was studied, and it was found that any segmentof the C-terminal domain (CTD) of sufficient length could bind and activatethe enzyme, regardless of its primary sequence and location in intact CTDs (Hansenet al. 2006). In the case of yeast prions Ure2p and Sup35p, amyloid formation providesyet another rather general example for the sequence-independence of recognition(Ross et al. 2005). These physiological prions probably provide selectiveadvantages for host cells (Wickner et al. 1999). They contain Q/N-rich disorderedprion domains, which can be randomized without the loss of the prion-like propertyof the protein (Ross et al. 2005).Partly based on these observations the concept of fuzziness has been developed(Tompa and Fuxreiter 2008). Fuzziness is a concept of disorder in the bound stateof IDPs, which have two, possibly interrelated, manifestations. In certain casesmolecular recognition and partner binding by an IDP occurs without an inducedfolding, i.e. ordering of the IDP, as described for T-cell receptor ζ-chains (Sigalovet al. 2004) and the product of the umuD gene (Simon et al. 2008). In other cases,binding occurs without the acquisition of a single dominant structure, insteadinvolving multiple states, which may be considered as polymorphism in the boundstate. This has been observed in the case of T-cell factor 4 (Tcf4) binding to β-catenin(Graham et al. 2001) and nuclear localization signal (NLS) to α-importin(Fontes et al. 2000). Evidently, this phenomenon limits the predictability of IDPfunction, because it contradicts the strict correspondence between sequence andfunction of the protein.5.6.3 Good News: Conservation and DisorderAs a final note, we should also mention that disorder is not completely opposed toconservation, as certain disordered regions appear to be evolutionarily conserved(Chen et al. 2006a, b). In a comprehensive study of domain families, Dunker andcolleagues have shown that many regions of at least 20 consecutive amino acids hadsignificant conservation. Such regions, termed conserved disorder predictions(CDP), were found in almost 30% of the domain families. Most CDPs are short,only 9% of them exceeding 30 residues in length, and they usually cover less than15% of the respective domain. The longest CDP, however, is 171 amino acids(in dentin matrix protein). Perhaps even more significantly, 8.7% of CDPs covermore than half of their respective domains, and 16 CPDs cover the entire domain.The functions of the domains harboring CDPs coincide with the general functionalpreferences of IDPs, such as DNA/RNA binding, ribosome structure, protein binding(both signalling/regulation and complex formation). Although this informationhas not yet been used in predicting function, it may provide the foundation of futureimportant applications.