Search

The yeast Fus1p SH3 domain binds to peptides containing the consensus motif, R(S/T)(S/T)SL, which is a sharp contrast to most SH3 domains, which bind to PXXP-containing peptides. Here, we have demonstrated that this domain binds to R(S/T)(S/T)SL-containing peptides derived from two putative in vivo binding partners from yeast proteins, Bnr1p and Ste5p, with K(d) values in the low micromolar range. The R(S/T)(S/T)SL consensus motif is necessary, but not sufficient for binding to the Fus1p SH3 domain, as residues lying N-terminal to the consensus motif also play a critical role in the binding reaction. Through mutagenesis studies and comparisons to other SH3 domains, we have discovered that the Fus1p SH3 domain utilizes a portion of the same binding surface as typical SH3 domains. However, the PXXP-binding surface, which plays the predominant role in binding for most SH3 domains, is debilitated in the WT domain by the substitution of unusual residues at three key conserved positions. By replacing these residues, we created a version of the Fus1p SH3 domain that binds to a PXXP-containing peptide with extremely high affinity (K(d)= 40 nM). Based on our data and analysis, we have clearly delineated two distinct surfaces comprising the typical SH3-domain-binding interface and show that one of these surfaces is the primary mediator of almost every "non-canonical" SH3-domain-mediated interaction described in the literature. Within this framework, dramatic alterations in SH3 domain specificity can be simply explained as a modulation of the binding strengths of these two surfaces.

Selection of a solubilizing detergent for membrane proteins is typically based on its ability to maintain the native structure and/or function of the molecule of interest. Descriptors of detergents as ‘harsh’ or ‘mild’ in terms of their propensity to denature membrane protein structures may act as a qualitative guide to this process, but the basis of the variable effects of detergents on native membrane protein folds is not yet fully explained. Previous work by our group suggested that the ability of the ‘harsh’ detergent sodium dodecylsulfate (SDS) to denature a series of wild-type and mutant model helical membrane-soluble ‘hairpin’ (helix-loop-helix) proteins depends on their level of detergent binding, as manifested by significant variability in their electrophoretic mobilities on SDS-PAGE, and in circular dichroism (CD) spectra and hydrodynamic radii [Rath, Nadeau et al., PNAS USA 106, 1760-65 (2009)]. Here we have extended this work to the characterization of a corresponding library of hairpin proteins in sodium perfluorooctanoate (SPFO), a surfactant considered to be non-denaturing. The ‘hairpin’ library has been developed from helix-loop-helix constructs from transmembrane segments 3 and 4 from the cystic fibrosis transmembrane conductance regulator (CFTR). We find that SPFO-solubilized ‘hairpins’ exhibit significantly less variability vs. their SDS-solubilized counterparts in their electrophoretic mobilities, CD spectra, and hydrodynamic radii. In addition, SPFO favours more stable helical structure while binding hairpins in larger amounts than does SDS across hairpin sequences. Pyrene excimer fluorescence data indicate that SPFO additionally retains more tertiary contacts vs. SDS for all hairpin sequences tested. Solubilization in SPFO therefore appears to favour protein-protein over lipid-protein interactions. Our overall results imply that the ‘harshness’ of a detergent is proportional to its protein sequence specificity upon binding, and consequent tendency not to disrupt intra-protein contacts during micelle formation.