Recent advances in protein design rely on rational and computational approaches to create novel sequences that fold and function. In contrast, natural systems selected functional proteins without any design a priori. In an attempt to mimic nature, we used large libraries of novel sequences and selected for functional proteins that rescue Escherichia coli cells in which a conditionally essential gene has been deleted. In this way, the de novo protein SynSerB3 was selected as a rescuer of cells in which serB, which encodes phosphoserine phosphatase, an enzyme essential for serine biosynthesis, was deleted. However, SynSerB3 does not rescue the deleted activity by catalyzing hydrolysis of phosphoserine. Instead, SynSerB3 up-regulates hisB, a gene encoding histidinol phosphate phosphatase. This endogenous E. coli phosphatase has promiscuous activity that, when overexpressed, compensates for the deletion of phosphoserine phosphatase. Thus, the de novo protein SynSerB3 rescues the deletion of serB by altering the natural regulation of the His operon.

The availability of large collections of de novo designed proteins presents new opportunities to harness novel macromolecules for synthetic biological functions. Many of these new functions will require binding to small molecules. Is the ability to bind small molecules a property that arises only in response to biological selection or computational design? Or alternatively, is small molecule binding a property of folded proteins that occurs readily amidst collections of unevolved sequences? These questions can be addressed by assessing the binding potential of de novo proteins that are designed to fold into stable structures, but are “naïve” in the sense that they (i) share no significant sequence similarity with natural proteins and (ii) were neither selected nor designed to bind small molecules. We chose three naïve proteins from a library of sequences designed to fold into 4-helix bundles and screened for binding to 10,000 compounds displayed on small molecule microarrays. Several binders were identified, and binding was characterized by a series of biophysical assays. Surprisingly, despite the similarity of the three de novo proteins to one another, they exhibit selective ligand binding. These findings demonstrate the potential of novel proteins for molecular recognition and have significant implications for a range of applications in synthetic biology.Keywords: binary code; four helix bundle; molecular evolution; protein design; small molecule microarray;

To probe the potential for activity in unevolved amino acid sequence space, we created a third generation combinatorial library of de novo four-helix bundle proteins. The “artificial superfamily” of helical bundles was designed using binary patterning of polar and nonpolar residues, and expressed in Escherichia coli from a library of synthetic genes. WA20, picked from the library, is one of the most stable proteins in the superfamily, and has rudimentary activities such as esterase and lipase. Here we report the crystal structure of WA20, determined by the multiwavelength anomalous dispersion method. Unexpectedly, the WA20 crystal structure is not a monomeric four-helix bundle, but a dimeric four-helix bundle. Each monomer comprises two long α-helices that intertwist with the helices of the other monomer. The two monomers together form a 3D domain-swapped four-helix bundle dimer. In addition, there are two hydrophobic pockets, which may potentially provide substrate binding sites. Small-angle X-ray scattering shows that the molecular weight of WA20 is ∼25 kDa and the shape is rod-like (the maximum length, Dmax = ∼8 nm), indicating that WA20 forms a dimeric four-helix bundle in solution. These results demonstrate that our de novo protein library contains not only simple monomeric proteins, but also stable and functional multimeric proteins.

The aggregation of polypeptides into amyloid fibrils is associated with a number of human diseases. Because these fibrils—or intermediates on the aggregation pathway—play important roles in the etiology of disease, considerable effort has been expended to understand which features of the amino acid sequence promote aggregation. One feature suspected to direct aggregation is the π-stacking of aromatic residues. Such π-stacking interactions have also been proposed as the targets for various aromatic compounds that are known to inhibit aggregation. In the case of Alzheimer’s disease, the aromatic side chains Phe19 and Phe20 in the wild-type amyloid beta (Aβ) peptide have been implicated. To explicitly test whether the aromaticity of these side chains plays a role in aggregation, we replaced these two phenylalanine side chains with leucines or isoleucines. These residues have similar sizes and hydrophobicities as Phe but are not capable of π-stacking. Thioflavin-T fluorescence and electron microscopy demonstrate that replacement of residues 19 and 20 by Leu or Ile did not prevent aggregation, but rather enhanced amyloid formation. Further experiments showed that aromatic inhibitors of aggregation are as effective against Ile- and Leu-substituted versions of Aβ42 as they are against wild-type Aβ. These results suggest that aromatic π-stacking interactions are not critical for Aβ aggregation or for the inhibition of Aβ aggregation.

The amyloid-β (Aβ) aggregation pathway is a key target in efforts to discover therapeutics that prevent or delay the onset of Alzheimer’s disease. Efforts at rational drug design, however, are hampered by uncertainties about the precise nature of the toxic aggregate. In contrast, high-throughput screening of compound libraries does not require a detailed understanding of the structure of the toxic species, and can provide an unbiased method for the discovery of small molecules that may lead to effective therapeutics. Here, we show that small molecule microarrays (SMMs) represent a particularly promising tool for identifying compounds that bind the Aβ peptide. Microarray slides with thousands of compounds immobilized on their surface were screened for binding to fluorescently labeled Aβ. Seventy-nine compounds were identified by the SMM screen, and then assayed for their ability to inhibit the Aβ-induced killing of PC12 cells. Further experiments focused on exploring the mechanism of rescue for one of these compounds: Electron microscopy and Congo red binding showed that the compound enhances fibril formation, and suggest that it may rescue cells by accelerating Aβ aggregation past an early toxic oligomer. These findings demonstrate that the SMM screen for binding to Aβ is effective at identifying compounds that reduce Aβ toxicity, and can reveal potential therapeutic leads without the biases inherent in methods that focus on inhibitors of aggregation.

De novo proteins from designed combinatorial libraries were bound to heme terminated gold electrodes. The novel heme proteins were shown to possess peroxidase activity, and this activity was compared to that of horseradish peroxidase and bovine serum albumin when immobilized in a similar fashion. The various designed proteins from the libraries displayed distinctly different levels of peroxidase activity, thereby demonstrating that the sequence and structure of a designed protein can exert a substantial effect on the peroxidase activity of immobilized heme.

Heme proteins can perform a variety of electrochemical functions. While natural heme proteins carry out particular functions selected by biological evolution, artificial heme proteins, in principle, can be tailored to suit specified technological applications. Here we describe initial characterization of the electrochemical properties of a de novo heme protein, S824C. Protein S824C is a four-helix bundle derived from a library of sequences that was designed by binary patterning of polar and nonpolar amino acids. Protein S824C was immobilized on a gold electrode and the formal potential of heme–protein complex was studied as a function of pH and ionic strength. The binding of exogenous N-donor ligands to heme/S824C was monitored by measuring shifts in the potential that occurred upon addition of various concentrations of imidazole or pyridine derivatives. The response of heme/S824C to these ligands was then compared to the response of isolated heme (without protein) to the same ligands. The observed shifts in potential depended on both the concentration and the structure of the added ligand. Small changes in structure of the ligand (e.g. pyridine versus 2-amino pyridine) produced significant shifts in the potential of the heme–protein. The observed shifts correlate to the differential binding of the N-donor molecules to the oxidized and reduced states of the heme. Further, it was observed that the electrochemical response of the buried heme in heme/S824C differed significantly from that of isolated heme. These studies demonstrate that the structure of the de novo protein modulates the binding of N-donor ligands to heme.

One hundred years ago, Alois Alzheimer observed a relationship between cognitive impairment and the presence of plaque in the brains of patients suffering from the disease that bears his name. The plaque was subsequently shown to be composed primarily of a 42-residue peptide called amyloid β (Aβ) 42. Despite the importance of Aβ42 aggregation in the molecular etiology of Alzheimer's disease, the amino acid sequence determinants of this process have yet to be elucidated. Although stretches of hydrophobic residues in the C-terminal half of Aβ42 have been implicated, the mechanism by which these residues promote aggregation remains unclear. In particular, it is not known whether the side chains of these hydrophobic residues mediate specific interactions that direct self-assembly or, alternatively, whether hydrophobicity per se at these positions is sufficient to promote aggregation. To distinguish between these two possibilities, we substituted 12 hydrophobic residues in the C-terminal half of Aβ42 with random nonpolar residues. The mutant sequences were screened by using a fusion of Aβ42 to GFP. Because aggregation of Aβ42 prevents folding of the GFP reporter, this screen readily distinguishes aggregating from nonaggregating variants of Aβ42. Application of the screen demonstrated that, despite the presence of 8–12 mutations, all of the sequences aggregated. To confirm these results, several of the mutant sequences were prepared as synthetic peptides and shown to form amyloid fibrils similar to those of WT Aβ42. These findings indicate that hydrophobic stretches in the sequence of Aβ42, rather than specific side chains, are sufficient to promote aggregation.

We previously reported the de novo design of combinatorial libraries of proteins targeted to fold into four-helix bundles. The sequences of these proteins were designed using a binary code strategy in which each position in the linear sequence is designated as either polar or nonpolar, but the exact identity of the amino acid at each position is varied combinatorially. We subsequently reported that approximately half of these binary coded proteins were capable of binding heme. These de novo heme-binding proteins showed CO binding characteristics similar to natural heme proteins, and several were active as peroxidases. Here we analyze the midpoint reduction potentials and heme binding stoichiometries of several of these de novo heme proteins. All the proteins bound heme with a 1:1 stoichiometry. The reduction potentials ranged from −112 to −176 mV. We suggest that this represents an estimate of the default range of potentials for heme proteins that have neither been prejudiced by rational design nor selected by evolution.

Combinatorial libraries of de novo amino acid sequences can provide a rich source of diversity for the discovery of novel proteins. Randomly generated sequences, however, rarely fold into well ordered protein-like structures. To enhance the quality of a library, diversity must be focused into those regions of sequence space most likely to yield well folded structures. We have constructed focused libraries of de novo sequences by designing the binary pattern of polar and nonpolar amino acids to favor structures that contain abundant secondary structure, while simultaneously burying hydrophobic side chains in the protein interior and exposing hydrophilic side chains to solvent. Because binary patterning specifies only the polar/nonpolar periodicity, but not the identities of the side chains, detailed structural features, including packing interactions, cannot be designed a priori. Can binary patterned libraries nonetheless encode well folded proteins? An unambiguous answer to this question requires determination of a 3D structure. We used NMR spectroscopy to determine the structure of S-824, a novel protein from a recently constructed library of 102-residue sequences. This library is “naïve” in that it has not been subjected to high-throughput screens or directed evolution. The experimentally determined structure of S-824 is a four-helix bundle, as specified by the design. As dictated by the binary-code strategy, nonpolar side chains are buried in the protein interior, and polar side chains are exposed to solvent. The polypeptide backbone and buried side chains are well ordered, demonstrating that S-824 is not a molten globule and forms a unique structure. These results show that amino acid sequences that have neither been selected by evolution, nor designed by computer, nor isolated by high-throughput screening, can form native-like structures. These findings validate the binary-code strategy as an effective method for producing vast collections of well folded de novo proteins.

Amyloid fibrils are associated with a variety of neurodegenerative maladies including Alzheimer's disease and the prion diseases. The structures of amyloid fibrils are composed of β-strands oriented orthogonal to the fibril axis (“cross β” structure). We previously reported the design and characterization of a combinatorial library of de novo β-sheet proteins that self-assemble into fibrillar structures resembling amyloid. The libraries were designed by using a “binary code” strategy, in which the locations of polar and nonpolar residues are specified explicitly, but the identities of these residues are not specified and are varied combinatorially. The initial libraries were designed to encode proteins containing amphiphilic β-strands separated by reverse turns. Each β-strand was designed to be seven residues long, with polar (○) and nonpolar (●) amino acids arranged with an alternating periodicity (○●○●○●○). The initial design specified the identical polar/nonpolar pattern for all of the β-strands; no strand was explicitly designated to form the edges of the resulting β-sheets. With all β-strands preferring to occupy interior (as opposed to edge) locations, intermolecular oligomerization was favored, and the proteins assembled into amyloid-like fibrils. To assess whether explicit design of edge-favoring strands might tip the balance in favor of monomeric β-sheet proteins, we have now redesigned the first and/or last β-strands of several sequences from the original library. In the redesigned β-strands, the binary pattern is changed from ○●○●○●○ to ○●○K○●○ (K denotes lysine). The presence of a lysine on the nonpolar face of a β-strand should disfavor fibrillar structures because such structures would bury an uncompensated charge. The nonpolar → lysine mutations, therefore, would be expected to favor monomeric structures in which the ○●○K○●○ sequences form edge strands with the charged lysine side chain accessible to solvent. To test this hypothesis, we constructed several second generation sequences in which the central nonpolar residue of either the N-terminal β-strand or the C-terminal β-strand (or both) is changed to lysine. Characterization of the redesigned proteins shows that they form monomeric β-sheet proteins.

A variety of naturally occurring biomaterials owe their unusual structural and mechanical properties to layers of β-sheet proteins laminated between layers of inorganic mineral. To explore the possibility of fabricating novel two-dimensional protein layers, we studied the self-assembly properties of de novo proteins from a designed combinatorial library. Each protein in the library has a distinct 63 amino acid sequence, yet they all share an identical binary pattern of polar and nonpolar residues, which was designed to favor the formation of six-stranded amphiphilic β-sheets. Characterization of proteins isolated from the library demonstrates that (i) they self assemble into monolayers at an air/water interface; (ii) the monolayers are dominated by β-sheet secondary structure, as shown by both circular dichroism and infrared spectroscopies; and (iii) the measured areas (500- 600 Å2) of individual protein molecules in the monolayers match those expected for proteins folded into amphiphilic β-sheets. The finding that similar structures are formed by distinctly different protein sequences suggests that assembly into β-sheet monolayers can be encoded by binary patterning of polar and nonpolar amino acids. Moreover, because the designed binary pattern is compatible with a wide variety of different sequences, it may be possible to fabricate β-sheet monolayers by using combinations of side chains that are explicitly designed to favor particular applications of novel biomaterials.

•Well-ordered protein structure is commonly assumed to be an essential feature of protein function.•Here we investigated the biophysical and structural properties of a family of de novo designed proteins that provide life-sustaining functions in E. coli.•We discovered that the de novo proteins tested here do not form well-ordered structures and instead form dynamic dimer structures.•These results highlight the importance of protein dynamics in protein function and also suggest that dynamic structures may have been important in early evolution.Designing and producing novel proteins that fold into stable structures and provide essential biological functions are key goals in synthetic biology. In initial steps toward achieving these goals, we constructed a combinatorial library of de novo proteins designed to fold into 4-helix bundles. As described previously, screening this library for sequences that function in vivo to rescue conditionally lethal mutants of Escherichia coli (auxotrophs) yielded several de novo sequences, termed SynRescue proteins, which rescued four different E. coli auxotrophs. In an effort to understand the structural requirements necessary for auxotroph rescue, we investigated the biophysical properties of the SynRescue proteins, using both computational and experimental approaches. Results from circular dichroism, size-exclusion chromatography, and NMR demonstrate that the SynRescue proteins are α-helical and relatively stable. Surprisingly, however, they do not form well-ordered structures. Instead, they form dynamic structures that fluctuate between monomeric and dimeric states. These findings show that a well-ordered structure is not a prerequisite for life-sustaining functions, and suggests that dynamic structures may have been important in the early evolution of protein function.Download high-res image (70KB)Download full-size image

Aggregation of the amyloid β (Aβ) peptide plays a key role in the molecular etiology of Alzheimer’s disease. Despite the importance of this process, the relationship between the sequence of Aβ and the propensity of the peptide to aggregate has not been fully elucidated. The sequence determinants of aggregation can be revealed by probing the ability of amino acid substitutions (mutations) to increase or decrease aggregation. Numerous mutations that decrease aggregation have been isolated by laboratory-based studies. In contrast, very few mutations that increase aggregation have been reported, and most of these were isolated from rare individuals with early-onset familial Alzheimer’s disease. To augment the limited data set of clinically derived mutations, we developed an artificial genetic screen to isolate novel mutations that increase aggregation propensity. The screen relies on the expression of Aβ–green fluorescent protein fusion in Escherichia coli. In this fusion, the ability of the green fluorescent protein reporter to fold and fluoresce is inversely correlated with the aggregation propensity of the Aβ sequence. Implementation of this screen enabled the isolation of 20 mutant versions of Aβ with amino acid substitutions at 17 positions in the 42-residue sequence of Aβ. Biophysical studies of synthetic peptides corresponding to sequences isolated by the screen confirm the increased aggregation propensity and amyloidogenic behavior of the mutants. The mutations were isolated using an unbiased screen that makes no assumptions about the sequence determinants of aggregation. Nonetheless, all 16 of the most aggregating mutants contain substitutions that reduce charge and/or increase hydrophobicity. These findings provide compelling evidence supporting the hypothesis that sequence hydrophobicity is a major determinant of Aβ aggregation.