The spatial architecture of protein function and adaptation

Abstract

Statistical analysis of protein evolution suggests a design for natural proteins in which sparse networks of coevolving amino acids (termed sectors) comprise the essence of three-dimensional structure and function. However, proteins are also subject to pressures deriving from the dynamics of the evolutionary process itself--the ability to tolerate mutation and to be adaptive to changing selection pressures. To understand the relationship of the sector architecture to these properties, we developed a high-throughput quantitative method for a comprehensive single-mutation study in which every position is substituted individually to every other amino acid. Using a PDZ domain (PSD95(pdz3)) model system, we show that sector positions are functionally sensitive to mutation, whereas non-sector positions are more tolerant to substitution. In addition, we find that adaptation to a new binding specificity initiates exclusively through variation within sector residues. A combination of just two sector mutations located near and away from the ligand-binding site suffices to switch the binding specificity of PSD95(pdz3) quantitatively towards a class-switching ligand. The localization of functional constraint and adaptive variation within the sector has important implications for understanding and engineering proteins.

Figure 1. Sector architecture in the PDZ domain family

a, b, The PDZ sector (blue spheres) shown in a cartoon (a) or space filling (b) representation of the structure of rat PSD95^pdz3 (Protein Data Bank (PDB) accession 1BE9). Yellow stick bonds represent the co-crystallized peptide ligand, with ligand positions numbered (0, −1, −2). The sector comprises a sparse network of residues built around the ligand-binding pocket and connecting to a distant surface site (marked with asterisk) through a subset of amino acid interactions within the protein core.

Figure 2. Complete single mutagenesis in PSD95^pdz3

a, The data matrix showing $\mathrm{\Delta }{E}_i^x$—the functional cost of every mutation x at each position i relative to wild-type PSD95^pdz3—colorimetrically, with blue representing loss-of-function and red representing gain-of-function mutations. The wild-type amino acid at each position is indicated by bold squares in the grid. The average functional cost of each amino acid substitution over all positions ( ${\langle \mathrm{\Delta }{E}_i^x\rangle }_i$) is shown at the right. b, The functional cost of all amino acid substitutions at each position shown as the average taken over each column ( ${\langle \mathrm{\Delta }{E}_i^x\rangle }_x$). c, A histogram of the data in panel b indicates positions with a significant effect (>2σ, 20 out of 83). d, Mapping of the 20 functionally significant positions on the PSD95^pdz3 structure, the peptide ligand shown as yellow stick bonds. These positions comprise a distributed, physically contiguous network built around the binding pocket and extending through the protein structure.

Figure 3. The relationship of mutational sensitivity of positions to the protein sector

a, The distribution of mutational effect in PSD95^pdz3 (grey), overlaid with distributions (in black) of sector, core (solvent accessibility < 0.15), positionally conserved (relative entropy >1, the mean value over all positions in the MSA) and ligand contacting positions (within 4 Å shell of ligand atoms). b–d, Slices through the core of PSD95^pdz3, showing mutationally significant core positions (dark blue), mutationally non-significant core positions (cyan), and the sector (orange mesh). All non-core positions are in grey, with the peptide ligand shown in yellow stick bonds.

Figure 4. Adaptation through sector variation

a, Average mutational effect in PSD95^pdz3 when binding the wild-type ligand CRIPT ( ${\langle \mathrm{\Delta }{E}_{i\mid WT}^x\rangle }_x$, top) or a class-switching T₋₂F variant ( ${\langle \mathrm{\Delta }{E}_{i\mid {\mathrm{T}}_{-2}\mathrm{F}}^x\rangle }_x$, bottom). T₋₂F involves a Thr to Phe mutation at position minus two of the peptide ligand (Fig. 1a). b, The difference, or epistasis, between mutational effects for binding wild-type or T₋₂F ligands, shown either averaged over amino acids at each position (top, ${\langle \mathrm{\Delta }\mathrm{\Delta }{E}_i^x\rangle }_x$) or broken down by amino acid (bottom, $\mathrm{\Delta }\mathrm{\Delta }{E}_i^x=\mathrm{\Delta }{E}_{i\mid WT}^x-\mathrm{\Delta }{E}_{i\mid T_{-2}F}^x$). The nine positions showing statistically significant epistasis (c) are numbered, and the asterisks mark positions where mutations on average can be positively selected for the T₋₂F ligand. c, A histogram of epistasis between mutations at each position in PSD95^pdz3 and the T₋₂F ligand variation. d, A mapping of epistatic positions (red) on the structure of PSD95^pdz3; the wild-type peptide ligand is shown in yellow stick bonds, the T₋₂ position is indicated in red and sector positions are in blue mesh. The positions showing mutational epistasis with T₋₂F comprise a physically distributed network propagating from the T₋₂ position, and are entirely composed of sector positions. The asterisks are as in b. e, The B2H and sequencing data for sector positions 330 and 372 when binding CRIPT (Fig. 2c) or T₋₂F (Supplementary Fig. 7) ligands suggest mutations for altering the specificity of PSD95^pdz3 towards T₋₂F; the wild-type amino acid is underlined. f, Binding affinities for purified PSD95^pdz3 carrying the Gly330Thr and His372Ala mutations both singly and together. Gly330Thr displays high affinity but non-specific binding for CRIPT and T₋₂F ligands, His372Ala shows a partial specificity switch towards T₋₂F and the double mutant represents a complete specificity switch for these two ligands.