Protein aggregation is linked with neurodegeneration and numerous other diseases by mechanisms that are not well understood. Here, we have analyzed the gain-of-function toxicity of artificial β sheet proteins that were designed to form amyloid-like fibrils. Using quantitative proteomics, we found that the toxicity of these proteins in human cells correlates with the capacity of their aggregates to promote aberrant protein interactions and to deregulate the cytosolic stress response. The endogenous proteins that are sequestered by the aggregates share distinct physicochemical properties: They are relatively large in size and significantly enriched in predicted unstructured regions, features that are strongly linked with multifunctionality. Many of the interacting proteins occupy essential hub positions in cellular protein networks, with key roles in chromatin organization, transcription, translation, maintenance of cell architecture and protein quality control. We suggest that amyloidogenic aggregation targets a metastable subproteome, thereby causing multifactorial toxicity and, eventually, the collapse of essential cellular functions.

A general strategy is described for the de novo design of proteins. In this strategy the sequence locations of hydrophobic and hydrophilic residues were specified explicitly, but the precise identities of the side chains were not constrained and varied extensively. This strategy was tested by constructing a large collection of synthetic genes whose protein products were designed to fold into four-helix bundle proteins. Each gene encoded a different amino acid sequence, but all sequences shared the same pattern of polar and nonpolar residues. Characterization of the expressed proteins indicated that most of the designed sequences folded into compact alpha-helical structures. Thus, a simple binary code of polar and nonpolar residues arranged in the appropriate order can drive polypeptide chains to collapse into globular alpha-helical folds.

Amyloid deposits are associated with several neurodegenerative diseases, including Alzheimer's disease and the prion diseases. The amyloid fibrils isolated from these different diseases share similar structural features. However, the protein sequences that assemble into these fibrils differ substantially from one disease to another. To probe the relationship between amino acid sequence and the propensity to form amyloid, we studied a combinatorial library of sequences designed de novo. All sequences in the library were designed to share an identical pattern of alternating polar and nonpolar residues, but the precise identities of these side chains were not constrained and were varied combinatorially. The resulting proteins selfassemble into large oligomers visible by electron microscopy as amyloid-like fibrils. Like natural amyloid, the de novo fibrils are composed of ␤-sheet secondary structure and bind the diagnostic dye, Congo red. Thus, binary patterning of polar and nonpolar residues arranged in alternating periodicity can direct protein sequences to form fibrils resembling amyloid. The model amyloid fibrils assemble and disassemble reversibly, providing a tractable system for both basic studies into the mechanisms of fibril assembly and the development of molecular therapies that interfere with this assembly.

No abstract

Combinatorial libraries of de novo amino acid sequences can provide a rich source of diversity for the discovery of novel proteins with interesting and important activities. Randomly generated sequences, however, rarely fold into well-ordered proteinlike structures. To enhance the quality of a library, features of rational design must be used to focus sequence diversity into those regions of sequence space that are most likely to yield folded structures. This review describes how focused libraries can be constructed by designing the binary pattern of polar and nonpolar amino acids to favor proteins that contain abundant secondary structure, while simultaneously burying hydrophobic side chains and exposing hydrophilic side chains to solvent. The "binary code" for protein design was used to construct several libraries of de novo proteins, including both ␣-helical and ␤-sheet structures. The recently determined solution structure of a binary patterned four-helix bundle is well ordered, thereby demonstrating that sequences that have neither been selected by evolution (in vivo or in vitro) nor designed by computer can form nativelike proteins. Examples are presented demonstrating how binary patterned libraries have successfully produced well-ordered structures, cofactor binding, catalytic activity, self-assembled monolayers, amyloid-like nanofibrils, and proteinbased biomaterials.

The protein Felix was designed de novo to fold into an antiparallel four-helix bundle of specific topology. Its sequence of 79 amino acid residues is not homologous to any known protein sequence, but is "native-like" in that it is nonrepetitive and contains 19 of the 20 naturally occurring amino acids. Felix has been expressed from a synthetic gene cloned in Escherichia coli, and the protein has been purified to homogeneity. Physical characterization of the purified protein indicates that Felix (i) is monomeric in solution, (ii) is predominantly alpha-helical, (iii) contains a designed intramolecular disulfide bond linking the first and fourth helices, and (iv) buries its single tryptophan in an apolar environment and probably in close proximity with the disulfide bond. These physical properties rule out several alternative structures and indicate that Felix indeed folds into approximately the designed three-dimensional structure.

The tendency of a polypeptide chain to form ar-helical or ,-strand secondary structure depends upon local and nonlocal effects. Local effects reflect the intrinsic propensities of the amino acid residues for particular secondary structures, while nonlocal effects reflect the positioning of the individual residues in the context of the entire amino acid sequence. In particular, the periodicity of polar and nonpolar residues specifies whether a given sequence is consistent with amphiphilic a-helices or ,8-strands. The importance of intrinsic propensities was compared to that of polar/nonpolar periodicity by a direct competition. Synthetic peptides were designed using residues with intrinsic propensities that favored one or the other type of secondary structure. The polar/nonpolar periodicities of the peptides were designed either to be consistent with the secondary structure favored by the intrinsic propensities of the component residues or in other cases to oppose these intrinsic propensities. Characterization ofthe synthetic peptides demonstrated that in all cases the observed secondary structure correlates with the periodicity of the peptide sequence-even when this secondary structure differs from that predicted from the intrinsic propensities of the component amino acids. The observed secondary structures are concentration dependent, indicating that oligomerization ofthe amphiphilic peptides is responsible for the observed secondary structures. Thus, for selfassembling oligomeric peptides, the polar/nonpolar periodicity can overwhelm the intrinsic propensities of the amino acid residues and serves as the major determinant of peptide secondary structure.The folded structures of proteins are stabilized by a variety of different features, including hydrogen bonding, van der Waals interactions, electrostatic interactions, the hydrophobic effect, and the intrinsic propensities of amino acid side chains for particular secondary structures (1). In recent years, the importance of each of these types of interactions individually has been probed in model peptide systems and in mutagenically altered proteins (2-15). The importance of one type of interaction relative to another type of interaction has received far less attention. This is not surprising since natural proteins typically are stabilized by the concerted action of numerous interrelated features, and it is not possible to isolate any one of these features from all the others. The study of proteins, however, is no longer limited to natural proteins. It is now possible to construct proteins that are designed entirely de novo (16)(17)(18)(19)(20). With the ability to design proteins from first principles comes the possibility to design structures by focusing on one type of interaction with the hope that optimizing this type will compensate for the mistakes that may result from an incomplete understanding of other features. Thus, the emerging field of de novo proteinThe publication costs of this article were defrayed in part by page charge payment. This art...

Repeated cycles of freezing and thawing are sufficient to separate highly expressed recombinant proteins away from the cellular milieu of E. coli. Freezing and thawing liberates recombinant proteins from the bacterial cytoplasm, but does not release the bulk of endogenous E. coli proteins. Furthermore, protein secretion is not required. Fractionation of overexpressed proteins by freeze/thaw treatment does not depend on the identity of the recombinant protein and has been observed for thirty-five different recombinant proteins expressed in E. coli. These include proteins originally found in plant, animal or microbial sources, as well as several proteins designed de novo. Freezing and thawing typically yields approximately 50% of the recombinant protein in relatively pure form. Thus the freeze/thaw treatment can be utilized as a general method for the isolation of recombinant proteins from E. coli.