Search

There has been considerable progress in the development of computational methods for designing protein-protein interactions, but engineering high-affinity binders without extensive screening and maturation remains challenging. Here, we test a protein design pipeline that uses iterative rounds of deep learning (DL)-based structure prediction (AlphaFold2) and sequence optimization (ProteinMPNN) to design autoinhibitory domains (AiDs) for a PD-L1 antagonist. Inspired by recent advances in therapeutic design, we sought to create autoinhibited (or masked) forms of the antagonist that can be conditionally activated by proteases. Twenty-threede novodesigned AiDs, varying in length and topology, were fused to the antagonist with a protease sensitive linker, and binding to PD-L1 was tested with and without protease treatment. Nine of the fusion proteins demonstrated conditional binding to PD-L1 and the top performing AiDs were selected for further characterization as single domain proteins. Without any experimental affinity maturation, four of the AiDs bind to the PD-L1 antagonist with equilibrium dissociation constants (KDs) below 150 nM, with the lowest KDequal to 0.9 nM. Our study demonstrates that DL-based protein modeling can be used to rapidly generate high affinity protein binders.Significance statementProtein-protein interactions are critical to most processes in biology, and improved methods for designing protein binders will enable the creation of new research reagents, diagnostics, and therapeutics. In this study, we show that a deep learning-based method for protein design can create high-affinity protein binders without the need for extensive screening or affinity maturation.

Many protein therapeutics are competitive inhibitors that function by binding to endogenous proteins and preventing them from interacting with native partners. One effective strategy for engineering competitive inhibitors is to graft structural motifs from a native partner into a host protein. Here, we develop and experimentally test a computational protocol for embedding binding motifs in de novo designed proteins. The protocol uses an “inside-out” approach: Starting with a structural model of the binding motif docked against the target protein, the de novo protein is built by growing new structural elements off the termini of the binding motif. During backbone assembly, a score function favors backbones that introduce new tertiary contacts within the designed protein and do not introduce clashes with the target binding partner. Final sequences are designed and optimized using the molecular modeling program Rosetta. To test our protocol, we designed small helical proteins to inhibit the interaction between Gαqand its effector PLC-β isozymes. Several of the designed proteins remain folded above 90°C and bind to Gαqwith equilibrium dissociation constants tighter than 80 nM. In cellular assays with oncogenic variants of Gαq, the designed proteins inhibit activation of PLC-β isozymes and Dbl-family RhoGEFs. Our results demonstrate that computational protein design, in combination with motif grafting, can be used to directly generate potent inhibitors without further optimization via high throughput screening or selection.statement for broader audienceEngineered proteins that bind to specific target proteins are useful as research reagents, diagnostics, and therapeutics. We used computational protein design to engineer de novo proteins that bind and competitively inhibit the G protein, Gαq, which is an oncogene for uveal melanomas. This computational method is a general approach that should be useful for designing competitive inhibitors against other proteins of interest.

Biomolecular structure drives function, and computational capabilities have progressed such that the prediction and computational design of biomolecular structures is increasingly feasible. Because computational biophysics attracts students from many different backgrounds and with different levels of resources, teaching the subject can be challenging. One strategy to teach diverse learners is with interactive multimedia material that promotes self-paced, active learning. We have created a hands-on education strategy with a set of 16 modules that teach topics in biomolecular structure and design, from fundamentals of conformational sampling and energy evaluation to applications, such as protein docking, antibody design, and RNA structure prediction. Our modules are based on PyRosetta, a Python library that encapsulates all computational modules and methods in the Rosetta software package. The workshop-style modules are implemented as Jupyter Notebooks that can be executed in the Google Colaboratory, allowing learners access with just a Web browser. The digital format of Jupyter Notebooks allows us to embed images, molecular visualization movies, and interactive coding exercises. This multimodal approach may better reach students from different disciplines and experience levels, as well as attract more researchers from smaller labs and cognate backgrounds to leverage PyRosetta in science and engineering research. All materials are freely available at https://github.com/RosettaCommons/PyRosetta.notebooks.

Chimeric antigen receptor (CAR) tonic signaling, defined as spontaneous activation and release of proinflammatory cytokines by CAR-T cells, is considered a negative attribute because it leads to impaired antitumor effects. Here, we report that CAR tonic signaling is caused by the intrinsic instability of the mAb single-chain variable fragment (scFv) to promote self-aggregation and signaling via the CD3ζ chain incorporated into the CAR construct. This phenomenon was detected in a CAR encoding either CD28 or 4-1BB costimulatory endodomains. Instability of the scFv was caused by specific amino acids within the framework regions (FWR) that can be identified by computational modeling. Substitutions of the amino acids causing instability, or humanization of the FWRs, corrected tonic signaling of the CAR, without modifying antigen specificity, and enhanced the antitumor effects of CAR-T cells. Overall, we demonstrated that tonic signaling of CAR-T cells is determined by the molecular instability of the scFv and that computational analyses of the scFv can be implemented to correct the scFv instability in CAR-T cells with either CD28 or 4-1BB costimulation.

The FastDesign protocol in the molecular modeling program Rosetta iterates between sequence optimization and structure refinement to stabilize de novo designed protein structures and complexes. FastDesign has been used previously to design novel protein folds and assemblies with important applications in research and medicine. To promote sampling of alternative conformations and sequences, FastDesign includes stages where the energy landscape is smoothened by reducing repulsive forces. Here, we discover that this process disfavors larger amino acids in the protein core because the protein compresses in the early stages of refinement. By testing alternative ramping strategies for the repulsive weight, we arrive at a scheme that produces lower energy designs with more native-like sequence composition in the protein core. We further validate the protocol by designing and experimentally characterizing over 4000 proteins and show that the new protocol produces higher stability proteins. This article is protected by copyright. All rights reserved.

We report the measurement and analysis of sulfonium-π, thioether-π, and ammonium-π interactions in a β-hairpin peptide model system, coupled with computational investigation and PDB analysis. These studies indicated that the sulfonium-π interaction is the strongest and that polarizability contributes to the stronger interaction with sulfonium relative to ammonium. Computational studies demonstrate that differences in solvation of the trimethylsulfonium versus the trimethylammonium group also contribute to the stronger sulfonium-π interaction. In comparing sulfonium-π versus sulfur-π interactions in proteins, analysis of SAM- and SAH-bound enzymes in the PDB suggests that aromatic residues are enriched in close proximity to the sulfur of both SAM and SAH, but the populations of aromatic interactions of the two cofactors are not significantly different, with the exception of the Me-π interactions in SAM, which are the most prevalent interaction in SAM but are not possible for SAH. This suggests that the weaker interaction energies due to loss of the cation-π interaction in going from SAM to SAH may contribute to turnover of the cofactor.

Here we introduce Z-lock, an optogenetic approach for reversible, light-controlled steric inhibition of protein active sites. The light oxygen voltage (LOV) domain and Zdk, a small protein that binds LOV selectively in the dark, are appended to the protein of interest where they sterically block the active site. Irradiation causes LOV to change conformation and release Zdk, exposing the active site. Computer-assisted protein design was used to optimize linkers and Zdk-LOV affinity, for both effective binding in the dark, and effective light-induced release of the intramolecular interaction. Z-lock cofilin was shown to have actin severing ability in vitro, and in living cancer cells it produced protrusions and invadopodia. An active fragment of the tubulin acetylase αTAT was similarly modified and shown to acetylate tubulin on irradiation.

Engineered, light-sensitive protein switches are used to interrogate a broad variety of biological processes. These switches are typically constructed by genetically fusing naturally occurring light-responsive protein domains with functional domains from other proteins. Protein activity can be controlled using a variety of mechanisms including light-induced colocalization, caging, and allosteric regulation. Protein design efforts have focused on reducing background signaling, maximizing the change in activity upon light stimulation, and perturbing the kinetics of switching. It is common to combine structure-based modeling with experimental screening to identify ideal fusion points between domains and discover point mutations that optimize switching. Here, we introduce commonly used light-sensitive domains and summarize recent progress in using them to regulate protein activity.

Phosphatidylinositol-3-kinases (PI3Ks) are lipid kinases that phosphorylate phosphatidylinositol 4,5-bisphosphate to generate a key lipid second messenger, phosphatidylinositol 3,4,5-bisphosphate. PI3Kα and PI3Kγ require activation by RAS proteins to stimulate signaling pathways that control cellular growth, differentiation, motility and survival. Intriguingly, RAS binding to PI3K isoforms likely differ, as RAS mutations have been identified that discriminate between PI3Kα and PI3Kγ, consistent with low sequence homology (23%) between their RAS binding domains (RBDs). As disruption of the RAS/PI3Kα interaction reduces tumor growth in mice with RAS- and epidermal growth factor receptor driven skin and lung cancers, compounds that interfere with this key interaction may prove useful as anti-cancer agents. However, a structure of PI3Kα bound to RAS is lacking, limiting drug discovery efforts. Expression of full-length PI3K isoforms in insect cells has resulted in low yield and variable activity, limiting biophysical and structural studies of RAS/PI3K interactions. This led us to generate the first RBDs from PI3Kα and PI3Kγ that can be expressed at high yield in bacteria and bind to RAS with similar affinity to full-length PI3K. We also solved a 2.31 A X-ray crystal structure of the PI3Kα-RBD, which aligns well to full-length PI3Kα. Structural differences between the PI3Kα and PI3Kγ RBDs are consistent with differences in thermal stability and may underly differential RAS recognition and RAS-mediated PI3K activation. These high expression, functional PI3K RBDs will aid in interrogating RAS interactions and could aid in identifying inhibitors of this key interaction.

Recombinant T cell receptors (TCRs) can be used to redirect naïve T cells to eliminate virally infected or cancerous cells; however, they are plagued by low stability and uneven expression. Here, we use molecular modeling to identify mutations in the TCR constant domains (Cα/Cβ) that increase the unfolding temperature of Cα/Cβ by 20 °C, improve the expression of four separate α/β TCRs by 3- to 10-fold, and improve the assembly and stability of TCRs with poor intrinsic stability. The stabilizing mutations rescue the expression of TCRs destabilized through variable domain mutation. The improved stability and folding of the TCRs reduces glycosylation, perhaps through conformational stabilization that restricts access to N-linked glycosylation enzymes. The Cα/Cβ mutations enables antibody-like expression and assembly of well-behaved bispecific molecules that combine an anti-CD3 antibody with the stabilized TCR. These TCR/CD3 bispecifics can redirect T cells to kill tumor cells with target HLA/peptide on their surfaces in vitro., Recombinant T-cells receptors can redirect naïve T cells but often have low or uneven expression. Here, the authors model mutations in TCR constant domains to increase T cell receptor expression, assembly, and stability and generate bispecific molecules.

Optogenetic techniques use light-responsive proteins to study dynamic processes in living cells and organisms. These techniques typically rely on repurposed naturally occurring light-sensitive proteins to control sub-cellular localization and activity. We previously engineered two optogenetic systems, the Light Activated Nuclear Shuttle (LANS) and the Light-Inducible Nuclear eXporter (LINX), by embedding nuclear import or export sequence motifs into the C-terminal helix of the light-responsive LOV2 domain of Avena sativa phototropin 1, thus enabling light-dependent trafficking of a target protein into and out of the nucleus. While LANS and LINX are effective tools, we posited that mutations within the LOV2 hinge-loop, which connects the core PAS domain and the C-terminal helix, would further improve the functionality of these switches. Here, we identify hinge-loop mutations that favourably shift the dynamic range (the ratio of the on- to off-target subcellular accumulation) of the LANS and LINX photoswitches. We demonstrate the utility of these new optogenetic tools to control gene transcription and epigenetic modifications, thereby expanding the optogenetic ‘tool kit’ for the research community.

Sactipeptides are ribosomally-synthesized peptides that contain a characteristic thioether bridge (sactionine bond) that is installed posttranslationally and is absolutely required for their antibiotic activity. Sactipeptide biosynthesis requires a unique family of radical SAM enzymes, which contain multiple [4Fe-4S] clusters, to form the requisite thioether bridge between a cysteine and the α-carbon of an opposing amino acid through radical-based chemistry. Here we present the structure of the sactionine bond-forming enzyme CteB, from Clostridium thermocellum ATCC 27405, with both SAM and an N-terminal fragment of its peptidyl-substrate at 2.04 Å resolution. CteB has the (β/α)(6)-TIM barrel fold that is characteristic of radical SAM enzymes, as well as a C-terminal SPASM domain that contains two auxiliary [4Fe-4S] clusters. Importantly, one [4Fe-4S] cluster in the SPASM domain exhibits an open coordination site in absence of peptide substrate, which is coordinated by a peptidyl-cysteine residue in the bound state. The crystal structure of CteB also reveals an accessory N-terminal domain that has high structural similarity to a recently discovered motif present in several enzymes that act on ribosomally-synthesized and post-translationally modified peptides (RiPPs), known as a RiPP precursor peptide recognition element (RRE). This crystal structure is the first of a sactionine bond forming enzyme and sheds light on structures and mechanisms of other members of this class such as AlbA or ThnB.

Light-sensitive proteins can be used to perturb signaling networks in living cells and animals with high spatiotemporal resolution. We recently engineered a protein heterodimer that dissociates when irradiated with blue light and demonstrated that by fusing each half of the dimer to termini of a protein that it is possible to selectively block binding surfaces on the protein when in the dark. Upon activation with light, the dimer dissociates and exposes the binding surface, allowing the protein to bind its partner. Critical to the success of this system, called Z-lock, is that the linkers connecting the dimer components to the termini are engineered so that the dimer forms over the appropriate binding surface. Here, we develop and test a protocol in the Rosetta molecular modeling program for designing linkers for Z-lock. We show that the protocol can predict the most effective linker sets for three different light-sensitive switches, including a newly designed switch that binds the Rho-family GTPase Cdc42 upon stimulation with blue light. This protocol represents a generalized computational approach to placing a wide variety of proteins under optogenetic control with Z-lock.

We used the molecular modeling program Rosetta to identify clusters of amino acid substitutions in antibody fragments (scFvs and scAbs) that improve global protein stability and resistance to thermal deactivation. Using this methodology, we increased the melting temperature (T(m)) and resistance to heat treatment of an antibody fragment that binds to the Clostridium botulinum hemagglutinin protein (anti-HA33). Two designed antibody fragment variants with two amino acid replacement clusters, designed to stabilize local regions, were shown to have both higher T(m) compared to the parental scFv and importantly, to retain full antigen binding activity after 2 hours of incubation at 70 °C. The crystal structure of one thermostabilized scFv variants was solved at 1.6 Å and shown to be in close agreement with the RosettaAntibody model prediction.

The prediction of protein three-dimensional structure from amino acid sequence has been a grand challenge problem in computational biophysics for decades, owing to its intrinsic scientific interest and also to the many potential applications for robust protein structure prediction algorithms, from genome interpretation to protein function prediction. More recently, the inverse problem - designing an amino acid sequence that will fold into a specified three-dimensional structure - has attracted growing attention as a potential route to the rational engineering of proteins with functions useful in biotechnology and medicine. Methods for the prediction and design of protein structures have advanced dramatically in the past decade. Increases in computing power and the rapid growth in protein sequence and structure databases have fuelled the development of new data-intensive and computationally demanding approaches for structure prediction. New algorithms for designing protein folds and protein-protein interfaces have been used to engineer novel high-order assemblies and to design from scratch fluorescent proteins with novel or enhanced properties, as well as signalling proteins with therapeutic potential. In this Review, we describe current approaches for protein structure prediction and design and highlight a selection of the successful applications they have enabled.

The Rosetta software for macromolecular modeling, docking and design is extensively used in laboratories worldwide. During two decades of development by a community of laboratories at more than 60 institutions, Rosetta has been continuously refactored and extended. Its advantages are its performance and interoperability between broad modeling capabilities. Here we review tools developed in the last 5 years, including over 80 methods. We discuss improvements to the score function, user interfaces and usability. Rosetta is available at ., This Perspective reviews tools developed over the past five years in the macromolecular modeling, docking and design software Rosetta.

Cells have evolved a variety of mechanisms to regulate the enormous complexity of processes taking place inside them. One mechanism consists in tightly controlling the localization of macromolecules, keeping them away from their place of action until needed. Since a large fraction of the cellular response to external stimuli is mediated by gene expression, it is not surprising that transcriptional regulators are often subject to stimulus-induced nuclear import or export. Here we review recent methods in chemical biology and optogenetics for controlling the nuclear localization of proteins of interest inside living cells. These methods allow researchers to regulate protein activity with exquisite spatiotemporal control, and open up new possibilities for studying the roles of proteins in a broad array of cellular processes and biological functions.

Protein ubiquitination involves E1, E2, and E3 trienzyme cascades. E2 and RING E3 enzymes often collaborate to first prime a substrate with a single ubiquitin (UB) and then achieve different forms of polyubiquitination: multiubiquitination of several sites and elongation of linkage-specific UB chains. Here, cryo-EM and biochemistry show that the human E3 anaphase-promoting complex/cyclosome (APC/C) and its two partner E2s, UBE2C (aka UBCH10) and UBE2S, adopt specialized catalytic architectures for these two distinct forms of polyubiquitination. The APC/C RING constrains UBE2C proximal to a substrate and simultaneously binds a substrate-linked UB to drive processive multiubiquitination. Alternatively, during UB chain elongation, the RING does not bind UBE2S but rather lures an evolving substrate-linked UB to UBE2S positioned through a cullin interaction to generate a Lys11-linked chain. Our findings define mechanisms of APC/C regulation, and establish principles by which specialized E3-E2-substrate-UB architectures control different forms of polyubiquitination.

Building new proteins from the old Proteins are the workhorses of biology. Designing new, stable proteins with functions desirable in biotechnology or biomedicine remains challenging. Jacobs et al. developed a computational method called SEWING that designs proteins using pieces of existing structures (see the Perspective by Netzer and Fleishman). The new proteins can contain structural features such as pockets or grooves that are required for function. The solved structures of two designed proteins agreed well with the design models. The method allows rapid design of a diverse set of structures that will facilitate functional design. Science , this issue p. 687 ; see also p. 657

Protein binding specificities can be manipulated by redesigning contacts that already exist at an interface or by expanding the interface to allow interactions with residues adjacent to the original binding site. Previously, we developed a strategy, called AnchorDesign, for expanding interfaces around linear binding epitopes. The epitope is embedded in a loop of a scaffold protein, in our case a monobody, and then surrounding residues on the monobody are optimized for binding using directed evolution or computational design. Using this strategy, we have increased binding affinities by over 100-fold, but we have not tested whether it can be used to control protein binding specificities. Here, we test whether AnchorDesign can be used to engineer a monobody that binds specifically to the Mitogen-activated protein kinase (MAPK) p38α, but not to the related MAPKs ERK2 and JNK. To anchor the binding interaction, we used a small (D) docking motif from the Mitogen-activated protein kinase kinase (MAP2K) MKK6 that interacts with similar affinity to p38α and ERK2. Our hypothesis was that by embedding the motif in a larger protein that we could expand the interface and create contacts with residues that are not conserved between p38α and ERK2. Molecular modeling was used to inform insertion of the D motif into the monobody and a combination of phage and yeast display were used to optimize the interface. Binding experiments demonstrate that the engineered monobody binds to the target surface on p38α and does not exhibit detectable binding to ERK2 or JNK.

We have developed a set of protocols in the molecular modeling program Rosetta for performing requirement-driven protein design. First, the user specifies a set of structural features that need to be present in the designed protein. These requirements can be general (e.g., "create a protein with five helices"), or they can be very specific and require the correct placement of a set of amino acids to bind a ligand. Next, a large set of protein models are generated that satisfy the design requirements. The models are built using a method that we recently introduced into Rosetta, called SEWING, that rapidly assembles novel protein backbones by combining pieces of naturally occurring proteins. In the last step of the process, rotamer-based sequence optimization and backbone refinement are performed with Rosetta, and a variety of quality metrics are used to pick sequences for experimental characterization. Here we describe the input files and user options needed to run SEWING and perform requirement-driven design and provide detailed instructions for two specific applications of the process: the design of new structural elements at a protein-protein interface and the design of ligand binding sites.

Light-inducible dimers are powerful tools for cellular optogenetics, as they can be used to control the localization and activity of proteins with high spatial and temporal resolution. Despite the generality of the approach, application of light-inducible dimers is not always straightforward, as it is frequently necessary to test alternative dimer systems and fusion strategies before the desired biological activity is achieved. This process is further hindered by an incomplete understanding of the biophysical/biochemical mechanisms by which available dimers behave and how this correlates to in vivo function. To better inform the engineering process, we examined the biophysical and biochemical properties of three blue-light-inducible dimer variants (cryptochrome2 (CRY2)/CIB1, iLID/SspB, and LOVpep/ePDZb) and correlated these characteristics to in vivo colocalization and functional assays. We find that the switches vary dramatically in their dark and lit state binding affinities and that these affinities correlate with activity changes in a variety of in vivo assays, including transcription control, intracellular localization studies, and control of GTPase signaling. Additionally, for CRY2, we observe that light-induced changes in homo-oligomerization can have significant effects on activity that are sensitive to alternative fusion strategies.

Summary Dendritic spines are the major loci of synaptic plasticity and are considered as possible structural correlates of memory. Nonetheless, systematic manipulation of specific subsets of spines in the cortex has been unattainable, and thus, the link between spines and memory has been correlational. We developed a novel synaptic optoprobe, AS-PaRac1 (activated synapse targeting photoactivatable Rac1), which can label recently potentiated spines specifically, and induce the selective shrinkage of AS-PaRac1-containing spines. In vivo imaging of AS-PaRac1 revealed that a motor learning induced substantial synaptic remodelling in a small subset of neurons. The acquired motor learning was disrupted by the optical shrinkage of the potentiated spines, whereas it was not affected by the identical manipulation of spines evoked by a distinct motor task in the same cortical region. Taken together, our results demonstrate that a newly acquired motor skill depends on the formation of a task-specific dense synaptic ensemble.

We engineered a photoactivatable system for rapidly and reversibly exporting proteins from the nucleus by embedding a nuclear export signal in the LOV2 domain from phototropin 1. Fusing the chromatin modifier Bre1 to the photoswitch, we achieved light-dependent control of histone H2B monoubiquitylation in yeast, revealing fast turnover of the ubiquitin mark. Moreover, this inducible system allowed us to dynamically monitor the status of epigenetic modifications dependent on H2B ubiquitylation.

Lipoprotein lipase (LPL) is a dimeric enzyme that is responsible for clearing triglyceride-rich lipoproteins from the blood. Although LPL plays a key role in cardiovascular health, an experimentally derived three-dimensional structure has not been determined. Such a structure would aid in understanding mutations in LPL that cause familial LPL deficiency in patients and help in the development of therapeutic strategies to target LPL. A major obstacle to structural studies of LPL is that LPL is an unstable protein that is difficult to produce in the quantities needed for nuclear magnetic resonance or crystallography. We present updated LPL structural models generated by combining disulfide mapping, computational modeling, and data derived from single-molecule Förster resonance energy transfer (smFRET). We pioneer the technique of smFRET for use with LPL by developing conditions for imaging active LPL and identifying positions in LPL for the attachment of fluorophores. Using this approach, we measure LPL-LPL intermolecular interactions to generate experimental constraints that inform new computational models of the LPL dimer structure. These models suggest that LPL may dimerize using an interface that is different from the dimerization interface suggested by crystal packing contacts seen in structures of pancreatic lipase.

Cellular signal transduction is often regulated at multiple steps in order to achieve more complex logic or precise control of a pathway. For instance, some signaling mechanisms couple allosteric activation with localization to achieve high signal to noise. Here, we create a system for light activated nuclear import that incorporates two levels of control. It consists of a nuclear import photoswitch, Light Activated Nuclear Shuttle (LANS), and a protein engineered to preferentially interact with LANS in the dark, Zdk2. First, Zdk2 is tethered to a location in the cytoplasm, which sequesters LANS in the dark. Second, LANS incorporates a nuclear localization signal (NLS) that is sterically blocked from binding to the nuclear import machinery in the dark. When activated with light, LANS both dissociates from its tethered location and exposes its NLS, which leads to nuclear accumulation. We demonstrate that this coupled system improves the dynamic range of LANS in mammalian cells, yeast, and C. elegans and provides tighter control of transcription factors that have been fused to LANS.

The computer-based design of PPIs (protein–protein interactions) is a challenging problem because large desolvation and entropic penalties must be overcome by the creation of favourable hydrophobic and polar contacts at the target interface. Indeed, many computationally designed interactions fail to form when tested in the laboratory. In the present article, we highlight strategies our laboratory has been pursuing to make interface design more tractable. Our general approach has been to make use of structural motifs found in native proteins that are predisposed to interact with a particular binding geometry, and then further bolster these anchor points with favourable hydrophobic contacts. We describe the use of three different anchor points, i.e. β-strand pairing, metal binding and the docking of α-helix into a groove, to successfully design new interfaces. In several cases, high-resolution crystal structures show that the design models closely match the experimental structure. In addition, we have tested the use of buried hydrogen-bond networks as a source of affinity and specificity at interfaces. In these cases, the designed complexes did not form, highlighting the challenges associated with designing buried polar interactions.

Over the past decade, the Rosetta biomolecular modeling suite has informed diverse biological questions and engineering challenges ranging from interpretation of low-resolution structural data to design of nanomaterials, protein therapeutics, and vaccines. Central to Rosetta’s success is the energy function: amodel parameterized from small molecule and X-ray crystal structure data used to approximate the energy associated with each biomolecule conformation. This paper describes the mathematical models and physical concepts that underlie the latest Rosetta energy function, beta_nov15. Applying these concepts,we explain how to use Rosetta energies to identify and analyze the features of biomolecular models.Finally, we discuss the latest advances in the energy function that extend capabilities from soluble proteins to also include membrane proteins, peptides containing non-canonical amino acids, carbohydrates, nucleic acids, and other macromolecules.

Cell growth and differentiation are controlled by growth factor receptors coupled to the GTPase Ras. Oncogenic mutations disrupt GTPase activity, leading to persistent Ras signaling and cancer progression. Recent evidence indicates that monoubiquitination of Ras leads to Ras activation. Mutation of the primary site of monoubiquitination impairs the ability of activated K-Ras (one of the three mammalian isoforms of Ras) to promote tumor growth. To determine the mechanism of human Ras activation, we chemically ubiquitinated the protein and analyzed its function by NMR, computational modeling and biochemical activity measurements. We established that monoubiquitination has little effect on the binding of Ras to guanine nucleotide, GTP hydrolysis or exchange-factor activation but severely abrogates the response to GTPase-activating proteins in a site-specific manner. These findings reveal a new mechanism by which Ras can trigger persistent signaling in the absence of receptor activation or an oncogenic mutation.

SummaryProtein design tests our understanding of protein stability and structure. Successful design methods should allow the exploration of sequence space not found in nature. However, when redesigning naturally occurring protein structures, most fixed backbone design algorithms return amino acid sequences that share strong sequence identity with wild-type sequences, especially in the protein core. This behavior places a restriction on functional space that can be explored and is not consistent with observations from nature, where sequences of low identity have similar structures. Here, we allow backbone flexibility during design to mutate every position in the core (38 residues) of a four-helix bundle protein. Only small perturbations to the backbone, 1–2 Å, were needed to entirely mutate the core. The redesigned protein, DRNN, is exceptionally stable (melting point >140°C). An NMR and X-ray crystal structure show that the side chains and backbone were accurately modeled (all-atom RMSD = 1.3 Å).

Pathways of ubiquitin-like (UBL) molecule transfer regulate a myriad of cellular cascades. Here, we report a high-throughput assay that correlates catalytic human NEDD8 transfer to bacterial survival. The assay was utilized to screen mutant NEDD8 and NAE (NEDD8-activating enzyme) libraries to engineer a more stable NEDD8 and redesign the NEDD8-NAE interface. This approach will be useful in understanding the specificities underlying UBL pathways.

Summary Mutation of surface residues to charged amino acids increases resistance to aggregation and can enable reversible unfolding. We have developed a protocol using the Rosetta computational design package that "supercharges" proteins while considering the energetic implications of each mutation. Using a homology model, a single-chain variable fragment antibody was designed that has a markedly enhanced resistance to thermal inactivation and displays an unanticipated ≈30-fold improvement in affinity. Such supercharged antibodies should prove useful for assays in resource-limited settings and for developing reagents with improved shelf lives.

Fluorescent biosensors for living cells currently require laborious optimization and a unique design for each target. They are limited by the availability of naturally occurring ligands with appropriate target specificity. Here we describe a biosensor based on an engineered fibronectin monobody scaffold that can be tailored to bind different targets via high throughput screening. This Src family kinase (SFK) biosensor was made by derivatizing a monobody specific for activated SFK with a bright dye whose fluorescence increases upon target binding. We identified sites for dye attachment and alterations to eliminate vesiculation in living cells, providing a generalizable scaffold for biosensor production. This approach minimizes cell perturbation because it senses endogenous, unmodified target, and because sensitivity is enhanced by direct dye excitation. Automated correlation of cell velocities and SFK activity revealed that SFK are activated specifically during protrusion. Activity correlates with velocity, and peaks 1–2 microns from the leading edge.

DNA methylation patterns regulate gene expression programs and are maintained through a highly coordinated process orchestrated by the RING E3 ubiquitin ligase UHRF1. UHRF1 controls DNA methylation inheritance by reading epigenetic modifications to histones and DNA to activate histone H3 ubiquitylation. Here, we find that all five domains of UHRF1, including the previously uncharacterized ubiquitin-like domain (UBL), cooperate for hemi-methylated DNA-dependent H3 ubiquitin ligation. Our structural and biochemical studies, including mutations found in cancer genomes, reveal a bifunctional requirement for the UBL in histone modification: (1) the UBL makes an essential interaction with the backside of the E2 and (2) the UBL coordinates with other UHRF1 domains that recognize epigenetic marks on DNA and histone H3 to direct ubiquitin to H3. Finally, we show UBLs from other E3s also have a conserved interaction with the E2, Ube2D, highlighting a potential prevalence of interactions between UBLs and E2s.

Hydrogen bond networks play a critical role in determining the stability and specificity of biomolecular complexes, and the ability to design such networks is important for engineering novel structures, interactions, and enzymes. One key feature of hydrogen bond networks that makes them difficult to rationally engineer is that they are highly cooperative and are not energetically favorable until the hydrogen bonding potential has been satisfied for all buried polar groups in the network. Existing computational methods for protein design are ill-equipped for creating these highly cooperative networks because they rely on energy functions and sampling strategies that are focused on pairwise interactions. To enable the design of complex hydrogen bond networks, we have developed a new sampling protocol in the molecular modeling program Rosetta that explicitly searches for sets of amino acid mutations that can form self-contained hydrogen bond networks. For a given set of designable residues, the protocol often identifies many alternative sets of mutations/networks, and we show that it can readily be applied to large sets of residues at protein-protein interfaces or in the interior of proteins. The protocol builds on a recently developed method in Rosetta for designing hydrogen bond networks that has been experimentally validated for small symmetric systems, but was not extensible to many larger protein structures and complexes. The sampling protocol we describe here not only recapitulates previously validated designs with performance improvements, but also yields viable hydrogen bond networks for cases where the previous method fails, such as the design of large, asymmetric interfaces relevant to engineering protein-based therapeutics.

SummaryDegradation by the ubiquitin-proteasome system requires assembly of a polyubiquitin chain upon substrate. However, the structural and mechanistic features that enable template-independent processive chain synthesis are unknown. We show that chain assembly by ubiquitin ligase SCF and ubiquitin-conjugating enzyme Cdc34 is facilitated by the unusual nature of Cdc34-SCF transactions: Cdc34 binds SCF with nanomolar affinity, nevertheless the complex is extremely dynamic. These properties are enabled by rapid association driven by electrostatic interactions between the acidic tail of Cdc34 and a basic ‘canyon’ in the Cul1 subunit of SCF. Ab initio docking between Cdc34 and Cul1 predicts intimate contact between the tail and the basic canyon, an arrangement confirmed by crosslinking and kinetic analysis of mutants. Basic canyon residues are conserved in both Cul1 paralogs and orthologs, suggesting that the same mechanism underlies processivity for all cullin-RING ubiquitin ligases. We discuss different strategies by which processive ubiquitin chain synthesis may be achieved.

The precise spatio-temporal dynamics of protein activity are often critical in determining cell behaviour, yet for most proteins they remain poorly understood; it remains difficult to manipulate protein activity at precise times and places within living cells. Protein activity has been controlled by light, through protein derivatization with photocleavable moieties or using photoreactive small-molecule ligands. However, this requires use of toxic ultraviolet wavelengths, activation is irreversible, and/or cell loading is accomplished via disruption of the cell membrane (for example, through microinjection). Here we have developed a new approach to produce genetically encoded photoactivatable derivatives of Rac1, a key GTPase regulating actin cytoskeletal dynamics in metazoan cells. Rac1 mutants were fused to the photoreactive LOV (light oxygen voltage) domain from phototropin, sterically blocking Rac1 interactions until irradiation unwound a helix linking LOV to Rac1. Photoactivatable Rac1 (PA-Rac1) could be reversibly and repeatedly activated using 458- or 473-nm light to generate precisely localized cell protrusions and ruffling. Localized Rac activation or inactivation was sufficient to produce cell motility and control the direction of cell movement. Myosin was involved in Rac control of directionality but not in Rac-induced protrusion, whereas PAK was required for Rac-induced protrusion. PA-Rac1 was used to elucidate Rac regulation of RhoA in cell motility. Rac and Rho coordinate cytoskeletal behaviours with seconds and submicrometre precision. Their mutual regulation remains controversial, with data indicating that Rac inhibits and/or activates Rho. Rac was shown to inhibit RhoA in mouse embryonic fibroblasts, with inhibition modulated at protrusions and ruffles. A PA-Rac crystal structure and modelling revealed LOV-Rac interactions that will facilitate extension of this photoactivation approach to other proteins.

We recently described two protein G variants (NuG1 and NuG2) with redesigned first hairpins that were almost twice as stable, folded 100-fold faster, and had a switched folding mechanism relative to the wild-type protein. To test the structural accuracy of our design algorithm and to provide insights to the dramatic changes in the kinetics and thermodynamics of folding, we have now determined the crystal structures of NuG1 and NuG2 to 1.8 A and 1.85 A, respectively. We find that they adopt hairpin structures that are closer to the computational models than to wild-type protein G; the RMSD of the NuG1 hairpin to the design model and the wild-type structure are 1.7 A and 5.1 A, respectively. The crystallographic B factor in the redesigned first hairpin of NuG1 is systematically higher than the second hairpin, suggesting that the redesigned region is somewhat less rigid. A second round of structure-based design yielded new variants of NuG1 and NuG2, which are further stabilized by 0.5 kcal/mole and 0.9 kcal/mole.

A challenge in the structure-based design of specificity is modeling the negative states, i.e., the complexes that you do not want to form. This is a difficult problem because mutations predicted to destabilize the negative state might be accommodated by small conformational rearrangements. To overcome this challenge, we employ an iterative strategy that cycles between sequence design and protein docking in order to build up an ensemble of alternative negative state conformations for use in specificity prediction. We have applied our technique to the design of heterodimeric CH3 interfaces in the Fc region of antibodies. Combining computationally and rationally designed mutations produced unique designs with heterodimer purities greater than 90%. Asymmetric Fc crystallization was able to resolve the interface mutations; the heterodimer structures confirmed that the interfaces formed as designed. With these CH3 mutations, and those made at the heavy-/light-chain interface, we demonstrate one-step synthesis of four fully IgG-bispecific antibodies.

Cellular optogenetic switches, a novel class of biological tools, have improved our understanding of biological phenomena that were previously intractable. While the design and engineering of these proteins has historically varied, they are all based on borrowed elements from plant and bacterial photoreceptors. In general terms, each of the optogenetic switches designed to date exploits the endogenous light-induced change in photoreceptor conformation while repurposing its effect to target a different biological phenomenon. We focus on the well-characterized light-oxygen-voltage 2 (LOV2) domain from Avena sativa phototropin 1 as our cornerstone for design. While the function of the LOV2 domain in the context of the phototropin protein is not fully elucidated, its thorough biophysical characterization as an isolated domain has created a strong foundation for engineering of photoswitches. In this chapter, we examine the biophysical characteristics of the LOV2 domain that may be exploited to produce an optogenetic switch and summarize previous design efforts to provide guidelines for an effective design. Furthermore, we provide protocols for assays including fluorescence polarization, phage display, and microscopy that are optimized for validating, improving, and using newly designed photoswitches.

Inducible dimers are powerful tools for controlling biological processes through colocalizing signaling molecules. To be effective, an inducible system should have a dissociation constant in the "off" state that is greater (i.e., weaker affinity) than the concentrations of the molecules that are being controlled, and in the "on" state a dissociation constant that is less (i.e., stronger affinity) than the relevant protein concentrations. Here, we reengineer the interaction between the light inducible dimer, iLID, and its binding partner SspB, to better control proteins present at high effective concentrations (5-100 μM). iLID contains a light-oxygen-voltage (LOV) domain that undergoes a conformational change upon activation with blue light and exposes a peptide motif, ssrA, that binds to SspB. The new variant of the dimer system contains a single SspB point mutation (A58V), and displays a 42-fold change in binding affinity when activated with blue light (from 3 ± 2 μM to 125 ± 40 μM) and allows for light-activated colocalization of transmembrane proteins in neurons, where a higher affinity switch (0.8-47 μM) was less effective because more colocalization was seen in the dark. Additionally, with a point mutation in the LOV domain (N414L), we lengthened the reversion half-life of iLID. This expanded suite of light induced dimers increases the variety of cellular pathways that can be targeted with light.

SummaryB. stearothermophilus tryptophanyl-tRNA synthetase catalysis proceeds via high-energy protein conformations. Unliganded MD trajectories of the pretransition-state complex with Mg2+•ATP and the (post) transition-state analog complex with adenosine tetraphosphate relax rapidly in opposite directions, the former regressing, the latter progressing along the structural reaction coordinate. The two crystal structures (rmsd 0.7 Å) therefore lie on opposite sides of a conformational free-energy maximum as the chemical transition state forms. SNAPP analysis illustrates the complexity of the associated long-range conformational coupling. Switching interactions in four nonpolar core regions are locally isoenergetic throughout the transition. Different configurations, however, propagate their effects to unfavorable, longer-range interactions at the molecular surface. Designed mutation shows that switching interactions enhance the rate, perhaps by destabilizing the ground state immediately before the transition state and limiting nonproductive diffusion before and after the chemical transition state, thereby reducing the activation entropy. This paradigm may apply broadly to energy-transducing enzymes.

The ability to manipulate protein binding affinities is important for the development of proteins as biosensors, industrial reagents, and therapeutics. We have developed a structure-based method to rationally predict single mutations at protein-protein interfaces that enhance binding affinities. The protocol is based on the premise that increasing buried hydrophobic surface area and/or reducing buried hydrophilic surface area will generally lead to enhanced affinity if large steric clashes are not introduced and buried polar groups are not left without a hydrogen bond partner. The procedure selects affinity enhancing point mutations at the protein-protein interface using three criteria: 1) the mutation must be from a polar amino acid to a non-polar amino acid or from a non-polar amino acid to a larger non-polar amino acid, 2) the free energy of binding as calculated with the Rosetta protein modeling program should be more favorable than the free energy of binding calculated for the wild type complex and 3) the mutation should not be predicted to significantly destabilize the monomers. The Rosetta energy function emphasizes short-range interactions: steric repulsion, Van der Waals forces, hydrogen bonding, and an implicit solvation model that penalizes placing atoms adjacent to polar groups. The performance of the computational protocol was experimentally tested on two separate protein complexes; Gαi1 from the heterotrimeric G-protein system bound to the RGS14 GoLoco motif, and the E2, UbcH7, bound to the E3, E6AP from the ubiquitin pathway. 12 single-site mutations that were predicted to be stabilizing were synthesized and characterized in the laboratory. 9 of the 12 mutations successfully increased binding affinity with 5 of these increasing binding by over 1.0 kcal/mol. To further assess our approach we searched the literature for point mutations that pass our criteria and have experimentally determined binding affinities. Of the 8 mutations identified, 5 were accurately predicted to increase binding affinity, further validating the method as a useful tool to increase protein-protein binding affinities.

Recent efforts to design de novo or redesign the sequence and structure of proteins using computational techniques have met with significant success. Most, if not all, of these computational methodologies attempt to model atomic-level interactions, and hence high-resolution structural characterization of the designed proteins is critical for evaluating the atomic-level accuracy of the underlying design force-fields. We previously used our computational protein design protocol RosettaDesign to completely redesign the sequence of the activation domain of human procarboxypeptidase A2. With 68% of the wild-type sequence changed, the designed protein, AYEdesign, is over 10 kcal/mol more stable than the wild-type protein. Here, we describe the high-resolution crystal structure and solution NMR structure of AYEdesign, which show that the experimentally determined backbone and side-chains conformations are effectively superimposable with the computational model at atomic resolution. To isolate the origins of the remarkable stabilization, we have designed and characterized a new series of procarboxypeptidase mutants that gain significant thermodynamic stability with a minimal number of mutations; one mutant gains more than 5 kcal/mol of stability over the wild-type protein with only four amino acid changes. We explore the relationship between force-field smoothing and conformational sampling by comparing the experimentally determined free energies of the overall design and these focused subsets of mutations to those predicted using modified force-fields, and both fixed and flexible backbone sampling protocols.

LOVTRAP is an optogenetic approach for reversible light-induced protein dissociation using protein A fragments that bind to the LOV domain only in the dark, with tunable kinetics and a >150-fold change in the dissociation constant (Kd). By reversibly sequestering proteins at mitochondria, we precisely modulated the proteins' access to the cell edge, demonstrating a naturally occurring 3-mHz cell-edge oscillation driven by interactions of Vav2, Rac1, and PI3K proteins.

The structurally defined ubiquitin-like homology fold (UBL) can engage in several unique protein–protein interactions and many of these complexes have been characterized with high-resolution techniques. Using Rosetta's structural classification tools, we have created the Ubiquitin Structural Relational Database (UbSRD), an SQL database of features for all 509 UBL-containing structures in the PDB, allowing users to browse these structures by protein–protein interaction and providing a platform for quantitative analysis of structural features. We used UbSRD to define the recognition features of ubiquitin (UBQ) and SUMO observed in the PDB and the orientation of the UBQ tail while interacting with certain types of proteins. While some of the interaction surfaces on UBQ and SUMO overlap, each molecule has distinct features that aid in molecular discrimination. Additionally, we find that the UBQ tail is malleable and can adopt a variety of conformations upon binding. UbSRD is accessible as an online resource at rosettadesign.med.unc.edu/ubsrd.

Accurate energy functions are critical to macromolecular modeling and design. We describe new tools for identifying inaccuracies in energy functions and guiding their improvement, and illustrate the application of these tools to the improvement of the Rosetta energy function. The feature analysis tool identifies discrepancies between structures deposited in the PDB and low-energy structures generated by Rosetta; these likely arise from inaccuracies in the energy function. The optE tool optimizes the weights on the different components of the energy function by maximizing the recapitulation of a wide range of experimental observations. We use the tools to examine three proposed modifications to the Rosetta energy function: improving the unfolded state energy model (reference energies), using bicubic spline interpolation to generate knowledge-based torisonal potentials, and incorporating the recently developed Dunbrack 2010 rotamer library (Shapovalov & Dunbrack, 2011). © 2013 Elsevier Inc.

We computationally designed a de novo protein-protein interaction between wild-type ubiquitin and a redesigned scaffold. Our strategy was to incorporate zinc at the designed interface to promote affinity and orientation specificity. A large set of monomeric scaffold surfaces were computationally engineered with three-residue zinc coordination sites, and the ubiquitin residue H68 was docked to the open coordination site to complete a tetrahedral zinc site. This single coordination bond was intended as a hotspot and polar interaction for ubiquitin binding, and surrounding residues on the scaffold were optimized primarily as hydrophobic residues using a rotamer-based sequence design protocol in Rosetta. From thousands of independent design simulations, four sequences were selected for experimental characterization. The best performing design, called Spelter, binds tightly to zinc (Kd 10 nM) and binds ubiquitin with a Kd of 20 µM in the presence of zinc and 68 µM in the absence of zinc. Mutagenesis studies and nuclear magnetic resonance chemical shift perturbation experiments indicate that Spelter interacts with H68 and the target surface on ubiquitin; however, H68 does not form a hotspot as intended. Instead, mutation of H68 to alanine results in tighter binding. Although a 3/1 zinc coordination arrangement at an interface cannot be ruled out as a means to improve affinity, our study led us to conclude that 2/2 coordination arrangements or multiple-zinc designs are more likely to promote high-affinity protein interactions.

Here we show that a recent computationally designed zinc-mediated protein interface is serendipitously capable of catalyzing carboxyester and phosphoester hydrolysis. Although the original motivation was to design a de novo zinc-mediated protein-protein interaction (called MID1-zinc), we observed in the homodimer crystal structure a small cleft and open zinc coordination site. We investigated if the cleft and zinc site at the designed interface were sufficient for formation of a primitive active site that can perform hydrolysis. MID1-zinc hydrolyzes 4-nitrophenyl acetate with a rate acceleration of 10(5) and a k(cat)/K(M) of 630 M(-1) s(-1) and 4-nitrophenyl phosphate with a rate acceleration of 10(4) and a k(cat)/K(M) of 14 M(-1) s(-1). These rate accelerations by an unoptimized active site highlight the catalytic power of zinc and suggest that the clefts formed by protein-protein interactions are well-suited for creating enzyme active sites. This discovery has implications for protein evolution and engineering: from an evolutionary perspective, three-coordinated zinc at a homodimer interface cleft represents a simple evolutionary path to nascent enzymatic activity; from a protein engineering perspective, future efforts in de novo design of enzyme active sites may benefit from exploring clefts at protein interfaces for active site placement.

SummaryPhotocontrol of functional peptides is a powerful tool for spatial and temporal control of cell signaling events. We show that the genetically encoded light-sensitive LOV2 domain of Avena Sativa phototropin 1 (AsLOV2) can be used to reversibly photomodulate the affinity of peptides for their binding partners. Sequence analysis and molecular modeling were used to embed two peptides into the Jα helix of the AsLOV2 domain while maintaining AsLOV2 structure in the dark but allowing for binding to effector proteins when the Jα helix unfolds in the light. Caged versions of the ipaA and SsrA peptides, LOV-ipaA and LOV-SsrA, bind their targets with 49- and 8-fold enhanced affinity in the light, respectively. These switches can be used as general tools for light-dependent colocalization, which we demonstrate with photo-activable gene transcription in yeast.