Design of structurally distinct proteins using strategies inspired by evolution

Abstract

Natural recombination combines pieces of pre-existing proteins to create new tertiary structures and functions. We describe a computational protocol, called SEWING, which is inspired by this process and builds new proteins from connected or disconnected pieces of existing structures. Helical proteins designed with SEWING contain structural features absent from other de novo designed proteins and in some cases remain folded to over 100 °C. High resolution structures of the designed proteins CA01 and DA05R1 were solved by X-ray crystallography (2.2 Å resolution) and NMR respectively, and there was excellent agreement with the design models. This method provides a new strategy to rapidly create large numbers of diverse and designable protein scaffolds.

Principles for computational design of binding antibodies

“…different natural antibodies; second, these newly designed backbones are docked against a target antigenic surface; and, third, for each backbone segment in the designed antibody, different conformations from natural antibodies are sampled and the sequence is optimized by Rosetta design calculations. To solve the problem of simultaneously designing a protein fold and its binding activity, the last step optimizes both antibody stability and binding energy jointly (15); previous computational-design algorithms, by contrast, concentrated on only one feature, either stability or binding, depending on the application (5,6,8,9,16,17).…”

“…The antibody Fv is a larger and more complex structure than the proteins that have previously been the subject of fold design (5,8,12,16), yielding three principal complications for computational design. First, as the Fv is a heterodimer, accurate design of not only one but two domains, as well as their interaction, is required.…”

“…By contrast, the functional surfaces of most natural proteins, including antibodies, contain nonideal features, such as unpaired polar groups, buried charges, and long loops that are essential for function (10, 11). It has therefore been postulated that computational design of "nonideal" backbone and sequence features is of fundamental importance for understanding protein structure, stability, and function, and may open the way to the application of computational-design methodology to difficult problems in design of function (7,8,(11)(12)(13).The antibody variable fragment (Fv) served us as an exemplary target for design of nonideal backbones since it comprises six loop segments in the antigen-binding surface [complementaritydetermining regions (CDRs) L1-L3 in the light chain and CDRs H1-H3 in the heavy chain]; many other protein folds, including TIM-barrel enzymes and β-propellers, similarly use loops in active sites (7,11,12). Furthermore, the antibody Fv comprises two chains, light and heavy, adding a layer of complexity so far absent from fold-design studies.…”

“…By contrast, the functional surfaces of most natural proteins, including antibodies, contain nonideal features, such as unpaired polar groups, buried charges, and long loops that are essential for function (10, 11). It has therefore been postulated that computational design of "nonideal" backbone and sequence features is of fundamental importance for understanding protein structure, stability, and function, and may open the way to the application of computational-design methodology to difficult problems in design of function (7,8,(11)(12)(13).…”

“…Computational protein design has mostly targeted so-called "ideal" proteins with high secondary-structure content, where polar backbone atoms form regular, short-range hydrogen bonds (5)(6)(7)(8)(9); irregularities, such as those seen in long loop regions, were almost absent from these designs. By contrast, the functional surfaces of most natural proteins, including antibodies, contain nonideal features, such as unpaired polar groups, buried charges, and long loops that are essential for function (10,11).…”