Combining machine learning with structure-based protein design to predict and engineer post-translational modifications of proteins.

Post-translational modifications (PTMs) of proteins play a vital role in their function and stability. These modifications influence protein folding, signaling, protein-protein interactions, enzyme activity, binding affinity, aggregation, degradation, and much more. To date, over 400 types of PTMs have been described, representing chemical diversity well beyond the genetically encoded amino acids. Such modifications pose a challenge to the successful design of proteins, but also represent a major opportunity to diversify the protein engineering toolbox. To this end, we first trained artificial neural networks (ANNs) to predict eighteen of the most abundant PTMs, including protein glycosylation, phosphorylation, methylation, and deamidation. In a second step, these models were implemented inside the computational protein modeling suite Rosetta, which allows flexible combination with existing protocols to model the modified sites and understand their impact on protein stability as well as function. Lastly, we developed a new design protocol that either maximizes or minimizes the predicted probability of a particular site being modified. We find that this combination of ANN prediction and structure-based design can enable the modification of existing, as well as the introduction of novel, PTMs. The potential applications of our work include, but are not limited to, glycan masking of epitopes, strengthening protein-protein interactions through phosphorylation, as well as protecting proteins from deamidation liabilities. These applications are especially important for the design of new protein therapeutics where PTMs can drastically change the therapeutic properties of a protein. Our work adds novel tools to Rosetta's protein engineering toolbox that allow for the rational design of PTMs.

How to Design Peptides.

Novel design of proteins to target receptors for treatment or tissue augmentation has come to the fore owing to advancements in computing power, modeling frameworks, and translational successes. Shorter proteins, or peptides, can offer combinatorial synergies with dendrimer, polymer, or other peptide carriers for enhanced local signaling, which larger proteins may sterically hinder. Here, we present a generalized method for designing a novel peptide. We first show how to create a script protocol that can be used to iteratively optimize and screen novel peptide sequences for binding a target protein. We present a step-by-step introduction to utilizing file repositories, data bases, and the Rosetta software suite. RosettaScripts, an .xml interface that allows for sequential functions to be performed, is used to order the functions for repeatable performance. These strategies may lead to more groups venturing into computational design, which may result in synergies from artificial intelligence/machine learning (AI/ML) to phage display and screening. Importantly, the beginner is expected to be able to design their first peptide ligand and begin their journey in peptide drug discovery. Generally, these peptides potentially could be used to interact with any enzyme or receptor, for example, in the study of chemokines and their interactions with glycosoaminoglycans and their receptors.

Accurate de novo design of membrane-traversing macrocycles.

We use computational design coupled with experimental characterization to systematically investigate the design principles for macrocycle membrane permeability and oral bioavailability. We designed 184 6-12 residue macrocycles with a wide range of predicted structures containing noncanonical backbone modifications and experimentally determined structures of 35; 29 are very close to the computational models. With such control, we show that membrane permeability can be systematically achieved by ensuring all amide (NH) groups are engaged in internal hydrogen bonding interactions. 84 designs over the 6-12 residue size range cross membranes with an apparent permeability greater than 1 × 10-6 cm/s. Designs with exposed NH groups can be made membrane permeable through the design of an alternative isoenergetic fully hydrogen-bonded state favored in the lipid membrane. The ability to robustly design membrane-permeable and orally bioavailable peptides with high structural accuracy should contribute to the next generation of designed macrocycle therapeutics.

Ensuring scientific reproducibility in bio-macromolecular modeling via extensive, automated benchmarks.

Each year vast international resources are wasted on irreproducible research. The scientific community has been slow to adopt standard software engineering practices, despite the increases in high-dimensional data, complexities of workflows, and computational environments. Here we show how scientific software applications can be created in a reproducible manner when simple design goals for reproducibility are met. We describe the implementation of a test server framework and 40 scientific benchmarks, covering numerous applications in Rosetta bio-macromolecular modeling. High performance computing cluster integration allows these benchmarks to run continuously and automatically. Detailed protocol captures are useful for developers and users of Rosetta and other macromolecular modeling tools. The framework and design concepts presented here are valuable for developers and users of any type of scientific software and for the scientific community to create reproducible methods. Specific examples highlight the utility of this framework, and the comprehensive documentation illustrates the ease of adding new tests in a matter of hours.

XENet: Using a new graph convolution to accelerate the timeline for protein design on quantum computers.

Graph representations are traditionally used to represent protein structures in sequence design protocols in which the protein backbone conformation is known. This infrequently extends to machine learning projects: existing graph convolution algorithms have shortcomings when representing protein environments. One reason for this is the lack of emphasis on edge attributes during massage-passing operations. Another reason is the traditionally shallow nature of graph neural network architectures. Here we introduce an improved message-passing operation that is better equipped to model local kinematics problems such as protein design. Our approach, XENet, pays special attention to both incoming and outgoing edge attributes. We compare XENet against existing graph convolutions in an attempt to decrease rotamer sample counts in Rosetta's rotamer substitution protocol, used for protein side-chain optimization and sequence design. This use case is motivating because it both reduces the size of the search space for classical side-chain optimization algorithms, and allows larger protein design problems to be solved with quantum algorithms on near-term quantum computers with limited qubit counts. XENet outperformed competing models while also displaying a greater tolerance for deeper architectures. We found that XENet was able to decrease rotamer counts by 40% without loss in quality. This decreased the memory consumption for classical pre-computation of rotamer energies in our use case by more than a factor of 3, the qubit consumption for an existing sequence design quantum algorithm by 40%, and the size of the solution space by a factor of 165. Additionally, XENet displayed an ability to handle deeper architectures than competing convolutions.

Anchor extension: a structure-guided approach to design cyclic peptides targeting enzyme active sites.

Despite recent success in computational design of structured cyclic peptides, de novo design of cyclic peptides that bind to any protein functional site remains difficult. To address this challenge, we develop a computational "anchor extension" methodology for targeting protein interfaces by extending a peptide chain around a non-canonical amino acid residue anchor. To test our approach using a well characterized model system, we design cyclic peptides that inhibit histone deacetylases 2 and 6 (HDAC2 and HDAC6) with enhanced potency compared to the original anchor (IC50 values of 9.1 and 4.4 nM for the best binders compared to 5.4 and 0.6 µM for the anchor, respectively). The HDAC6 inhibitor is among the most potent reported so far. These results highlight the potential for de novo design of high-affinity protein-peptide interfaces, as well as the challenges that remain.

Current directions in combining simulation-based macromolecular modeling approaches with deep learning.

Introduction: Structure-guided drug discovery relies on accurate computational methods for modeling macromolecules. Simulations provide means of predicting macromolecular folds, of discovering function from structure, and of designing macromolecules to serve as drugs. Success rates are limited for any of these tasks, however. Recently, deep neural network-based methods have greatly enhanced the accuracy of predictions of protein structure from sequence, generating excitement about the potential impact of deep learning.Areas covered: This review introduces biologists to deep neural network architecture, surveys recent successes of deep learning in structure prediction, and discusses emerging deep learning-based approaches for structure-function analysis and design. Particular focus is given to the interplay between simulation-based and neural network-based approaches.Expert opinion: As deep learning grows integral to macromolecular modeling, simulation- and neural network-based approaches must grow more tightly interconnected. Modular software architecture must emerge allowing both types of tools to be combined with maximal versatility. Open sharing of code under permissive licenses will be essential. Although experiments will remain the gold standard for reliable information to guide drug discovery, we may soon see successful drug development projects based on high-accuracy predictions from algorithms that combine simulation with deep learning - the ultimate validation of this combination's power.

MHCEpitopeEnergy, a Flexible Rosetta-Based Biotherapeutic Deimmunization Platform.

As non-"self" macromolecules, biotherapeutics can trigger an immune response that can reduce drug efficacy, require patients to be taken off therapy, or even cause life-threatening reactions. To enable the flexible and facile design of protein biotherapeutics while reducing the prevalence of T-cell epitopes that drive immune recognition, we have integrated into the Rosetta protein design suite a new scoring term that allows design protocols to account for predicted or experimentally identified epitopes in the optimized objective function. This flexible scoring term can be used in any Rosetta design trajectory, can be targeted to specific regions of a protein, and can be readily extended to work with a variety of epitope predictors. By performing extensive design runs with varied design parameter choices for three case study proteins as well as a larger diverse benchmark, we show that the incorporation of this scoring term enables the effective exploration of an alternative, deimmunized sequence space to discover diverse proteins that are potentially highly deimmunized while retaining physical and chemical qualities similar to those yielded by equivalent nondeimmunizing sequence design protocols.

Computationally designed peptide macrocycle inhibitors of New Delhi metallo-ß-lactamase 1.

The rise of antibiotic resistance calls for new therapeutics targeting resistance factors such as the New Delhi metallo-ß-lactamase 1 (NDM-1), a bacterial enzyme that degrades ß-lactam antibiotics. We present structure-guided computational methods for designing peptide macrocycles built from mixtures of l- and d-amino acids that are able to bind to and inhibit targets of therapeutic interest. Our methods explicitly consider the propensity of a peptide to favor a binding-competent conformation, which we found to predict rank order of experimentally observed IC50 values across seven designed NDM-1- inhibiting peptides. We were able to determine X-ray crystal structures of three of the designed inhibitors in complex with NDM-1, and in all three the conformation of the peptide is very close to the computationally designed model. In two of the three structures, the binding mode with NDM-1 is also very similar to the design model, while in the third, we observed an alternative binding mode likely arising from internal symmetry in the shape of the design combined with flexibility of the target. Although challenges remain in robustly predicting target backbone changes, binding mode, and the effects of mutations on binding affinity, our methods for designing ordered, binding-competent macrocycles should have broad applicability to a wide range of therapeutic targets.

Computational design of mixed chirality peptide macrocycles with internal symmetry.

Cyclic symmetry is frequent in protein and peptide homo-oligomers, but extremely rare within a single chain, as it is not compatible with free N- and C-termini. Here we describe the computational design of mixed-chirality peptide macrocycles with rigid structures that feature internal cyclic symmetries or improper rotational symmetries inaccessible to natural proteins. Crystal structures of three C2- and C3-symmetric macrocycles, and of six diverse S2-symmetric macrocycles, match the computationally-designed models with backbone heavy-atom RMSD values of 1 Å or better. Crystal structures of an S4-symmetric macrocycle (consisting of a sequence and structure segment mirrored at each of three successive repeats) designed to bind zinc reveal a large-scale zinc-driven conformational change from an S4-symmetric apo-state to a nearly inverted S4-symmetric holo-state almost identical to the design model. These symmetric structures provide promising starting points for applications ranging from design of cyclic peptide based metal organic frameworks to creation of high affinity binders of symmetric protein homo-oligomers. More generally, this work demonstrates the power of computational design for exploring symmetries and structures not found in nature, and for creating synthetic switchable systems.

Better together: Elements of successful scientific software development in a distributed collaborative community.

Many scientific disciplines rely on computational methods for data analysis, model generation, and prediction. Implementing these methods is often accomplished by researchers with domain expertise but without formal training in software engineering or computer science. This arrangement has led to underappreciation of sustainability and maintainability of scientific software tools developed in academic environments. Some software tools have avoided this fate, including the scientific library Rosetta. We use this software and its community as a case study to show how modern software development can be accomplished successfully, irrespective of subject area. Rosetta is one of the largest software suites for macromolecular modeling, with 3.1 million lines of code and many state-of-the-art applications. Since the mid 1990s, the software has been developed collaboratively by the RosettaCommons, a community of academics from over 60 institutions worldwide with diverse backgrounds including chemistry, biology, physiology, physics, engineering, mathematics, and computer science. Developing this software suite has provided us with more than two decades of experience in how to effectively develop advanced scientific software in a global community with hundreds of contributors. Here we illustrate the functioning of this development community by addressing technical aspects (like version control, testing, and maintenance), community-building strategies, diversity efforts, software dissemination, and user support. We demonstrate how modern computational research can thrive in a distributed collaborative community. The practices described here are independent of subject area and can be readily adopted by other software development communities.

De novo design of bioactive protein switches.

Allosteric regulation of protein function is widespread in biology, but is challenging for de novo protein design as it requires the explicit design of multiple states with comparable free energies. Here we explore the possibility of designing switchable protein systems de novo, through the modulation of competing inter- and intramolecular interactions. We design a static, five-helix 'cage' with a single interface that can interact either intramolecularly with a terminal 'latch' helix or intermolecularly with a peptide 'key'. Encoded on the latch are functional motifs for binding, degradation or nuclear export that function only when the key displaces the latch from the cage. We describe orthogonal cage-key systems that function in vitro, in yeast and in mammalian cells with up to 40-fold activation of function by key. The ability to design switchable protein functions that are controlled by induced conformational change is a milestone for de novo protein design, and opens up new avenues for synthetic biology and cell engineering.

A systematic study of minima in alanine dipeptide.

The alanine dipeptide is a standard system to model dihedral angles in proteins. It is shown that obtaining the Ramachandran plot accurately is a hard problem because of many local minima; depending on the details of geometry optimizations, different Ramachandran plots can be obtained. To locate all energy minima, starting from geometries from MD simulations, 250,000 geometry optimizations were performed at the level of RHF/6-31G*, followed by re-optimizations of the located 827 minima at the level of MP2/6-311++G**, yielding 30 unique minima, most of which were not previously reported in literature. Both in vacuo and solvated structures are discussed. The minima are systematically categorized based on four backbone dihedral angles. The Gibbs energies are evaluated and the structural factors determining the relative stabilities of conformers are discussed. © 2018 Wiley Periodicals, Inc.

Programmable design of orthogonal protein heterodimers.

Specificity of interactions between two DNA strands, or between protein and DNA, is often achieved by varying bases or side chains coming off the DNA or protein backbone-for example, the bases participating in Watson-Crick pairing in the double helix, or the side chains contacting DNA in TALEN-DNA complexes. By contrast, specificity of protein-protein interactions usually involves backbone shape complementarity1, which is less modular and hence harder to generalize. Coiled-coil heterodimers are an exception, but the restricted geometry of interactions across the heterodimer interface (primarily at the heptad a and d positions2) limits the number of orthogonal pairs that can be created simply by varying side-chain interactions3,4. Here we show that protein-protein interaction specificity can be achieved using extensive and modular side-chain hydrogen-bond networks. We used the Crick generating equations5 to produce millions of four-helix backbones with varying degrees of supercoiling around a central axis, identified those accommodating extensive hydrogen-bond networks, and used Rosetta to connect pairs of helices with short loops and to optimize the remainder of the sequence. Of 97 such designs expressed in Escherichia coli, 65 formed constitutive heterodimers, and the crystal structures of four designs were in close agreement with the computational models and confirmed the designed hydrogen-bond networks. In cells, six heterodimers were fully orthogonal, and in vitro-following mixing of 32 chains from 16 heterodimer designs, denaturation in 5 M guanidine hydrochloride and reannealing-almost all of the interactions observed by native mass spectrometry were between the designed cognate pairs. The ability to design orthogonal protein heterodimers should enable sophisticated protein-based control logic for synthetic biology, and illustrates that nature has not fully explored the possibilities for programmable biomolecular interaction modalities.

Accurate computational design of multipass transmembrane proteins.

The computational design of transmembrane proteins with more than one membrane-spanning region remains a major challenge. We report the design of transmembrane monomers, homodimers, trimers, and tetramers with 76 to 215 residue subunits containing two to four membrane-spanning regions and up to 860 total residues that adopt the target oligomerization state in detergent solution. The designed proteins localize to the plasma membrane in bacteria and in mammalian cells, and magnetic tweezer unfolding experiments in the membrane indicate that they are very stable. Crystal structures of the designed dimer and tetramer-a rocket-shaped structure with a wide cytoplasmic base that funnels into eight transmembrane helices-are very close to the design models. Our results pave the way for the design of multispan membrane proteins with new functions.

Comprehensive computational design of ordered peptide macrocycles.

Mixed-chirality peptide macrocycles such as cyclosporine are among the most potent therapeutics identified to date, but there is currently no way to systematically search the structural space spanned by such compounds. Natural proteins do not provide a useful guide: Peptide macrocycles lack regular secondary structures and hydrophobic cores, and can contain local structures not accessible with l-amino acids. Here, we enumerate the stable structures that can be adopted by macrocyclic peptides composed of l- and d-amino acids by near-exhaustive backbone sampling followed by sequence design and energy landscape calculations. We identify more than 200 designs predicted to fold into single stable structures, many times more than the number of currently available unbound peptide macrocycle structures. Nuclear magnetic resonance structures of 9 of 12 designed 7- to 10-residue macrocycles, and three 11- to 14-residue bicyclic designs, are close to the computational models. Our results provide a nearly complete coverage of the rich space of structures possible for short peptide macrocycles and vastly increase the available starting scaffolds for both rational drug design and library selection methods.

De novo design of covalently constrained mesosize protein scaffolds with unique tertiary structures.

The folding of natural proteins typically relies on hydrophobic packing, metal binding, or disulfide bond formation in the protein core. Alternatively, a 3D structure can be defined by incorporating a multivalent cross-linking agent, and this approach has been successfully developed for the selection of bicyclic peptides from large random-sequence libraries. By contrast, there is no general method for the de novo computational design of multicross-linked proteins with predictable and well-defined folds, including ones not found in nature. Here we use Rosetta and Tertiary Motifs (TERMs) to design small proteins that fold around multivalent cross-linkers. The hydrophobic cross-linkers stabilize the fold by macrocyclic restraints, and they also form an integral part of a small apolar core. The designed CovCore proteins were prepared by chemical synthesis, and their structures were determined by solution NMR or X-ray crystallography. These mesosized proteins, lying between conventional proteins and small peptides, are easily accessible either through biosynthetic precursors or chemical synthesis. The unique tertiary structures and ease of synthesis of CovCore proteins indicate that they should provide versatile templates for developing inhibitors of protein-protein interactions.

Simultaneous Optimization of Biomolecular Energy Functions on Features from Small Molecules and Macromolecules.

Most biomolecular modeling energy functions for structure prediction, sequence design, and molecular docking have been parametrized using existing macromolecular structural data; this contrasts molecular mechanics force fields which are largely optimized using small-molecule data. In this study, we describe an integrated method that enables optimization of a biomolecular modeling energy function simultaneously against small-molecule thermodynamic data and high-resolution macromolecular structural data. We use this approach to develop a next-generation Rosetta energy function that utilizes a new anisotropic implicit solvation model, and an improved electrostatics and Lennard-Jones model, illustrating how energy functions can be considerably improved in their ability to describe large-scale energy landscapes by incorporating both small-molecule and macromolecule data. The energy function improves performance in a wide range of protein structure prediction challenges, including monomeric structure prediction, protein-protein and protein-ligand docking, protein sequence design, and prediction of the free energy changes by mutation, while reasonably recapitulating small-molecule thermodynamic properties.

Accurate de novo design of hyperstable constrained peptides.

Naturally occurring, pharmacologically active peptides constrained with covalent crosslinks generally have shapes that have evolved to fit precisely into binding pockets on their targets. Such peptides can have excellent pharmaceutical properties, combining the stability and tissue penetration of small-molecule drugs with the specificity of much larger protein therapeutics. The ability to design constrained peptides with precisely specified tertiary structures would enable the design of shape-complementary inhibitors of arbitrary targets. Here we describe the development of computational methods for accurate de novo design of conformationally restricted peptides, and the use of these methods to design 18-47 residue, disulfide-crosslinked peptides, a subset of which are heterochiral and/or N-C backbone-cyclized. Both genetically encodable and non-canonical peptides are exceptionally stable to thermal and chemical denaturation, and 12 experimentally determined X-ray and NMR structures are nearly identical to the computational design models. The computational design methods and stable scaffolds presented here provide the basis for development of a new generation of peptide-based drugs.

Computational design of an unnatural amino acid dependent metalloprotein with atomic level accuracy.

Genetically encoded unnatural amino acids could facilitate the design of proteins and enzymes of novel function, but correctly specifying sites of incorporation and the identities and orientations of surrounding residues represents a formidable challenge. Computational design methods have been used to identify optimal locations for functional sites in proteins and design the surrounding residues but have not incorporated unnatural amino acids in this process. We extended the Rosetta design methodology to design metalloproteins in which the amino acid (2,2'-bipyridin-5yl)alanine (Bpy-Ala) is a primary ligand of a bound metal ion. Following initial results that indicated the importance of buttressing the Bpy-Ala amino acid, we designed a buried metal binding site with octahedral coordination geometry consisting of Bpy-Ala, two protein-based metal ligands, and two metal-bound water molecules. Experimental characterization revealed a Bpy-Ala-mediated metalloprotein with the ability to bind divalent cations including Co(2+), Zn(2+), Fe(2+), and Ni(2+), with a Kd for Zn(2+) of ∼40 pM. X-ray crystal structures of the designed protein bound to Co(2+) and Ni(2+) have RMSDs to the design model of 0.9 and 1.0 Šrespectively over all atoms in the binding site.