De novo design of high-affinity binders of bioactive helical peptides.

Many peptide hormones form an α-helix on binding their receptors1-4, and sensitive methods for their detection could contribute to better clinical management of disease5. De novo protein design can now generate binders with high affinity and specificity to structured proteins6,7. However, the design of interactions between proteins and short peptides with helical propensity is an unmet challenge. Here we describe parametric generation and deep learning-based methods for designing proteins to address this challenge. We show that by extending RFdiffusion8 to enable binder design to flexible targets, and to refining input structure models by successive noising and denoising (partial diffusion), picomolar-affinity binders can be generated to helical peptide targets by either refining designs generated with other methods, or completely de novo starting from random noise distributions without any subsequent experimental optimization. The RFdiffusion designs enable the enrichment and subsequent detection of parathyroid hormone and glucagon by mass spectrometry, and the construction of bioluminescence-based protein biosensors. The ability to design binders to conformationally variable targets, and to optimize by partial diffusion both natural and designed proteins, should be broadly useful.

Substrates and Cyclic Peptide Inhibitors of the Oligonucleotide-Activated Sirtuin 7.

The sirtuins are NAD+ -dependent lysine deacylases, comprising seven isoforms (SIRT1-7) in humans, which are involved in the regulation of a plethora of biological processes, including gene expression and metabolism. The sirtuins share a common hydrolytic mechanism but display preferences for different ϵ-N-acyllysine substrates. SIRT7 deacetylates targets in nuclei and nucleoli but remains one of the lesser studied of the seven isoforms, in part due to a lack of chemical tools to specifically probe SIRT7 activity. Here we expressed SIRT7 and, using small-angle X-ray scattering, reveal SIRT7 to be a monomeric enzyme with a low degree of globular flexibility in solution. We developed a fluorogenic assay for investigation of the substrate preferences of SIRT7 and to evaluate compounds that modulate its activity. We report several mechanism-based SIRT7 inhibitors as well as de novo cyclic peptide inhibitors selected from mRNA-display library screening that exhibit selectivity for SIRT7 over other sirtuin isoforms, stabilize SIRT7 in cells, and cause an increase in the acetylation of H3 K18.

A context-dependent and disordered ubiquitin-binding motif.

Ubiquitin is a small, globular protein that is conjugated to other proteins as a posttranslational event. A palette of small, folded domains recognizes and binds ubiquitin to translate and effectuate this posttranslational signal. Recent computational studies have suggested that protein regions can recognize ubiquitin via a process of folding upon binding. Using peptide binding arrays, bioinformatics, and NMR spectroscopy, we have uncovered a disordered ubiquitin-binding motif that likely remains disordered when bound and thus expands the palette of ubiquitin-binding proteins. We term this motif Disordered Ubiquitin-Binding Motif (DisUBM) and find it to be present in many proteins with known or predicted functions in degradation and transcription. We decompose the determinants of the motif showing it to rely on features of aromatic and negatively charged residues, and less so on distinct sequence positions in line with its disordered nature. We show that the affinity of the motif is low and moldable by the surrounding disordered chain, allowing for an enhanced interaction surface with ubiquitin, whereby the affinity increases ~ tenfold. Further affinity optimization using peptide arrays pushed the affinity into the low micromolar range, but compromised context dependence. Finally, we find that DisUBMs can emerge from unbiased screening of randomized peptide libraries, featuring in de novo cyclic peptides selected to bind ubiquitin chains. We suggest that naturally occurring DisUBMs can recognize ubiquitin as a posttranslational signal to act as affinity enhancers in IDPs that bind to folded and ubiquitylated binding partners.

Correction: In vivo modulation of ubiquitin chains by N-methylated non-proteinogenic cyclic peptides.

[This corrects the article DOI: 10.1039/D0CB00179A.].

In vivo modulation of ubiquitin chains by N-methylated non-proteinogenic cyclic peptides.

Cancer and other disease states can change the landscape of proteins post-translationally tagged with ubiquitin (Ub) chains. Molecules capable of modulating Ub chains are potential therapeutic agents, but their discovery represents a significant challenge. Recently, it was shown that de novo cyclic peptides, selected from trillion-member random libraries, are capable of binding particular Ub chains. However, these peptides were overwhelmingly proteinogenic, so the prospect of in vivo activity was uncertain. Here, we report the discovery of small, non-proteinogenic cyclic peptides, rich in non-canonical features like N-methylation, which can tightly and specifically bind Lys48-linked Ub chains. These peptides engage three Lys48-linked Ub units simultaneously, block the action of deubiquitinases and the proteasome, induce apoptosis in vitro, and attenuate tumor growth in vivo. This highlights the potential of non-proteinogenic cyclic peptide screening to rapidly find in vivo-active leads, and the targeting of ubiquitin chains as a promising anti-cancer mechanism of action.

Structure-activity relationship of ipglycermide binding to phosphoglycerate mutases.

Catalysis of human phosphoglycerate mutase is dependent on a 2,3-bisphosphoglycerate cofactor (dPGM), whereas the nonhomologous isozyme in many parasitic species is cofactor independent (iPGM). This mechanistic and phylogenetic diversity offers an opportunity for selective pharmacologic targeting of glycolysis in disease-causing organisms. We previously discovered ipglycermide, a potent inhibitor of iPGM, from a large combinatorial cyclic peptide library. To fully delineate the ipglycermide pharmacophore, herein we construct a detailed structure-activity relationship using 280 substituted ipglycermide analogs. Binding affinities of these analogs to immobilized Caenorhabditis elegans iPGM, measured as fold enrichment relative to the index residue by deep sequencing of an mRNA display library, illuminated the significance of each amino acid to the pharmacophore. Using cocrystal structures and binding kinetics, we show that the high affinity of ipglycermide for iPGM orthologs, from Brugia malayi, Onchocerca volvulus, Dirofilaria immitis, and Escherichia coli, is achieved by a codependence between (1) the off-rate mediated by the macrocycle Cys14 thiolate coordination to an active-site Zn2+ in the iPGM phosphatase domain and (2) shape complementarity surrounding the macrocyclic core at the phosphotransferase-phosphatase domain interface. Our results show that the high-affinity binding of ipglycermide to iPGMs freezes these structurally dynamic enzymes into an inactive, stable complex.

Peptide Folding and Binding Probed by Systematic Non-canonical Mutagenesis.

Many proteins and peptides fold upon binding another protein. Mutagenesis has proved an essential tool in the study of these multi-step molecular recognition processes. By comparing the biophysical behavior of carefully selected mutants, the concert of interactions and conformational changes that occur during folding and binding can be separated and assessed. Recently, this mutagenesis approach has been radically expanded by deep mutational scanning methods, which allow for many thousands of mutations to be examined in parallel. Furthermore, these high-throughput mutagenesis methods have been expanded to include mutations to non-canonical amino acids, returning peptide structure-activity relationships with unprecedented depth and detail. These developments are timely, as the insights they provide can guide the optimization of de novo cyclic peptides, a promising new modality for chemical probes and therapeutic agents.

Initiating ribosomal peptide synthesis with exotic building blocks.

Ribosomal peptide synthesis begins almost exclusively with the amino acid methionine, across all domains of life. The ubiquity of methionine initiation raises the question; to what extent could polypeptide synthesis be realized with other amino acids, proteinogenic or otherwise? This highlight describes the breadth of building blocks now known to be accepted by the ribosome initiation machinery, from subtle methionine analogues to large exotic non-proteinogenic structures. We outline the key methodological developments that have enabled these discoveries, including the exploitation of methionyl-tRNA synthetase promiscuity, synthetase and tRNA engineering, and the utilization of artificial tRNA-loading ribozymes, flexizymes. Using these methods, the number and diversity of validated initiation building blocks is rapidly expanding permitting the use of the ribosome to synthesize ever more artificial polymers in search of new functional molecules.

Affinity Maturation of Macrocyclic Peptide Modulators of Lys48-Linked Diubiquitin by a Twofold Strategy.

Messenger RNA display of peptides containing non-proteinogenic amino acids, referred to as RaPID system, has become one of the leading methods to express libraries consisting of more than trillion-members of macrocyclic peptides, which allows for discovering de novo bioactive ligands. Ideal macrocyclic peptides should have dissociation constants (KD ) as low as single-digit values in the nanomolar range towards a specific target of interest. Here, a twofold strategy to discover optimized macrocyclic peptides within this affinity regime is described. First, benzyl thioether cyclized peptide libraries were explored to identify tight binding hits. To obtain more insights into critical sequence information, sequence alignment was applied to guide rational mutagenesis for the improvement of their binding affinity. Using this twofold strategy, benzyl thioether macrocyclic peptide binders against Lys48-linked ubiquitin dimer (K48-Ub2) were successfully obtained that display KD values in the range 0.3-1.2 nm, which indicate binding two orders of magnitude stronger than those of macrocyclic peptides recently reported. Most importantly, this macrocyclic peptide also showed an improved cellular inhibition of the K48-Ub2 recognition by deubiquitinating enzymes and the 26S proteasome, resulting in the promotion of apoptosis in cancer cells.

Ribosomal Incorporation of Aromatic Oligoamides as Peptide Sidechain Appendages.

Derivatives of 4-aminomethyl-l-phenylalanine with aromatic oligoamide foldamers as sidechain appendages were successfully charged on tRNA by means of flexizymes. Their subsequent incorporation both at the C-terminus of, and within, peptide sequences by the ribosome, was demonstrated. These results expand the registry of chemical structures tolerated by the ribosome to sidechains significantly larger and more structurally defined than previously demonstrated.

De novo macrocyclic peptides that specifically modulate Lys48-linked ubiquitin chains.

A promising approach in cancer therapy is to find ligands that directly bind ubiquitin (Ub) chains. However, finding molecules capable of tightly and specifically binding Ub chains is challenging given the range of Ub polymer lengths and linkages and their subtle structural differences. Here, we use total chemical synthesis of proteins to generate highly homogeneous Ub chains for screening against trillion-member macrocyclic peptide libraries (RaPID system). De novo cyclic peptides were found that can bind tightly and specifically to K48-linked Ub chains, confirmed by NMR studies. These cyclic peptides protected K48-linked Ub chains from deubiquitinating enzymes and prevented proteasomal degradation of Ub-tagged proteins. The cyclic peptides could enter cells, inhibit growth and induce programmed cell death, opening new opportunities for therapeutic intervention. This highly synthetic approach, with both protein target generation and cyclic peptide discovery performed in vitro, will make other elaborate post-translationally modified targets accessible for drug discovery.

Nonproteinogenic deep mutational scanning of linear and cyclic peptides.

High-resolution structure-activity analysis of polypeptides requires amino acid structures that are not present in the universal genetic code. Examination of peptide and protein interactions with this resolution has been limited by the need to individually synthesize and test peptides containing nonproteinogenic amino acids. We describe a method to scan entire peptide sequences with multiple nonproteinogenic amino acids and, in parallel, determine the thermodynamics of binding to a partner protein. By coupling genetic code reprogramming to deep mutational scanning, any number of amino acids can be exhaustively substituted into peptides, and single experiments can return all free energy changes of binding. We validate this approach by scanning two model protein-binding peptides with 21 diverse nonproteinogenic amino acids. Dense structure-activity maps were produced at the resolution of single aliphatic atom insertions and deletions. This permits rapid interrogation of interaction interfaces, as well as optimization of affinity, fine-tuning of physical properties, and systematic assessment of nonproteinogenic amino acids in binding and folding.

Publisher Correction: Ribosomal synthesis and folding of peptide-helical aromatic foldamer hybrids.

In the version of this Article originally published, in Fig.1f there was an erroneous 'Gly-Gly' label placed above the foldamer-peptide structure. Furthermore, in Fig. 2a, the expected target structures from substrates 9 and 10 were inadvertently swapped. These errors have been corrected in the online versions.

Ribosomal synthesis and folding of peptide-helical aromatic foldamer hybrids.

Translation, the mRNA-templated synthesis of peptides by the ribosome, can be manipulated to incorporate variants of the 20 cognate amino acids. Such approaches for expanding the range of chemical entities that can be produced by the ribosome may accelerate the discovery of molecules that can perform functions for which poorly folded, short peptidic sequences are ill suited. Here, we show that the ribosome tolerates some artificial helical aromatic oligomers, so-called foldamers. Using a flexible tRNA-acylation ribozyme-flexizyme-foldamers were attached to tRNA, and the resulting acylated tRNAs were delivered to the ribosome to initiate the synthesis of non-cyclic and cyclic foldamer-peptide hybrid molecules. Passing through the ribosome exit tunnel requires the foldamers to unfold. Yet foldamers encode sufficient folding information to influence the peptide structure once translation is completed. We also show that in cyclic hybrids, the foldamer portion can fold into a helix and force the peptide segment to adopt a constrained and stretched conformation.

Interplay between partner and ligand facilitates the folding and binding of an intrinsically disordered protein.

Protein-protein interactions are at the heart of regulatory and signaling processes in the cell. In many interactions, one or both proteins are disordered before association. However, this disorder in the unbound state does not prevent many of these proteins folding to a well-defined, ordered structure in the bound state. Here we examine a typical system, where a small disordered protein (PUMA, p53 upregulated modulator of apoptosis) folds to an α-helix when bound to a groove on the surface of a folded protein (MCL-1, induced myeloid leukemia cell differentiation protein). We follow the association of these proteins using rapid-mixing stopped flow, and examine how the kinetic behavior is perturbed by denaturant and carefully chosen mutations. We demonstrate the utility of methods developed for the study of monomeric protein folding, including ß-Tanford values, Leffler α, Φ-value analysis, and coarse-grained simulations, and propose a self-consistent mechanism for binding. Folding of the disordered protein before binding does not appear to be required and few, if any, specific interactions are required to commit to association. The majority of PUMA folding occurs after the transition state, in the presence of MCL-1. We also examine the role of the side chains of folded MCL-1 that make up the binding groove and find that many favor equilibrium binding but, surprisingly, inhibit the association process.

High thermodynamic stability of parametrically designed helical bundles.

We describe a procedure for designing proteins with backbones produced by varying the parameters in the Crick coiled coil-generating equations. Combinatorial design calculations identify low-energy sequences for alternative helix supercoil arrangements, and the helices in the lowest-energy arrangements are connected by loop building. We design an antiparallel monomeric untwisted three-helix bundle with 80-residue helices, an antiparallel monomeric right-handed four-helix bundle, and a pentameric parallel left-handed five-helix bundle. The designed proteins are extremely stable (extrapolated ΔGfold > 60 kilocalories per mole), and their crystal structures are close to those of the design models with nearly identical core packing between the helices. The approach enables the custom design of hyperstable proteins with fine-tuned geometries for a wide range of applications.

Coupled folding and binding of the disordered protein PUMA does not require particular residual structure.

Many cellular proteins are 'disordered' in isolation. A subset of these intrinsically disordered proteins (IDPs) can, upon binding another molecule, fold to a well-defined three-dimensional structure. In the structurally heterogeneous, unbound ensemble of these IDPs, conformations are likely to exist that, in part, resemble the final bound form. It has been suggested that these conformations, displaying 'residual structure', could be important for the mechanism of such coupled folding and binding reactions. PUMA, of the BCL-2 family, is an IDP in isolation but will form a single, contiguous α-helix upon binding the folded protein MCL-1. Using the helix-breaking residue proline, we systematically target each potential turn of helix of unbound PUMA and assess the binding to MCL-1 using time-resolved stopped-flow techniques. All proline-containing mutants bound, and although binding was weaker than the wild-type protein, association rate constants were largely unaffected. We conclude that population of particular residual structure, containing a specific helical turn, is neither required for the binding nor for fast association of PUMA and MCL-1.

Folding and binding of an intrinsically disordered protein: fast, but not 'diffusion-limited'.

Coupled folding and binding of intrinsically disordered proteins (IDPs) is prevalent in biology. As the first step toward understanding the mechanism of binding, it is important to know if a reaction is 'diffusion-limited' as, if this speed limit is reached, the association must proceed through an induced fit mechanism. Here, we use a model system where the 'BH3 region' of PUMA, an IDP, forms a single, contiguous α-helix upon binding the folded protein Mcl-1. Using stopped-flow techniques, we systematically compare the rate constant for association (k(+)) under a number of solvent conditions and temperatures. We show that our system is not 'diffusion-limited', despite having a k(+) in the often-quoted 'diffusion-limited' regime (10(5)-10(6) M(-1) s(-1) at high ionic strength) and displaying an inverse dependence on solvent viscosity. These standard tests, developed for folded protein-protein interactions, are not appropriate for reactions where one protein is disordered.

Protein folding: adding a nucleus to guide helix docking reduces landscape roughness.

The elongated three-helix-bundle spectrin domains R16 and R17 fold and unfold unusually slowly over a rough energy landscape, in contrast to the homologue R15, which folds fast over a much smoother, more typical landscape. R15 folds via a nucleation-condensation mechanism that guides the docking of the A and C-helices. However, in R16 and R17, the secondary structure forms first and the two helices must then dock in the correct register. Here, we use variants of R16 and R17 to demonstrate that substitution of just five key residues is sufficient to alter the folding mechanism and reduce the landscape roughness. We suggest that, by providing access to an alternative, faster, folding route over their landscape, R16 and R17 can circumvent their slow, frustrated wild-type folding mechanism.

Separating the effects of internal friction and transition state energy to explain the slow, frustrated folding of spectrin domains.

The elongated three-helix bundle domains spectrin R16 and R17 fold some two to three orders of magnitude more slowly than their homologue R15. We have shown that this slow folding is due, at least in part, to roughness in the free-energy landscape of R16 and R17. We have proposed that this roughness is due to a frustrated search for the correct docking of partly preformed helices. However, this accounts for only a small part of the slowing of folding and unfolding. Five residues on the A helix of R15, when inserted together into R16 or R17, increase the folding rate constants, reduce landscape roughness, and alter the folding mechanism to one resembling R15. The effect of each of these mutations individually is investigated here. No one mutation causes the behavior seen for the five in combination. However, two mutations, E18F and K25V, significantly increase the folding and unfolding rates of both R16 and R17 but without a concomitant loss in landscape roughness. E18F has the greatest effect on the kinetics, and a Φ-value analysis of the C helix reveals that the folding mechanism is unchanged. For both E18F and K25V the removal of the charge and resultant transition state stabilization is the main origin of the faster folding. Consequently, the major cause of the unusually slow folding of R16 and R17 is the non-native burial of the two charged residues in the transition state. The slowing due to landscape roughness is only about fivefold.