Design principles of protein switches

Computational design of de novo bioenergetic membrane proteins

The major energy-producing reactions of biochemistry occur at biological membranes. Computational protein design now provides the opportunity to elucidate the underlying principles of these processes and to construct bioenergetic pathways on our own terms. Here, we review recent achievements in this endeavour of 'synthetic bioenergetics', with a particular focus on new enabling tools that facilitate the computational design of biocompatible de novo integral membrane proteins. We use recent examples to showcase some of the key computational approaches in current use and highlight that the overall philosophy of 'surface-swapping' - the replacement of solvent-facing residues with amino acids bearing lipid-soluble hydrophobic sidechains - is a promising avenue in membrane protein design. We conclude by highlighting outstanding design challenges and the emerging role of AI in sequence design and structure ideation.

Conformation-Dependent Hydrogen-Bonding Interactions in a Switchable Artificial Metalloprotein

Hydrogen-bonding (H-bonding) interactions in metalloprotein active sites can critically regulate enzyme function. Changes in the protein structure triggered by interplay with substrates, products, and partner proteins are often translated to the metallocofactor by way of specific changes in H-bond networks connected to the active site. However, the complexities of metalloprotein architecture and mechanism often preclude our ability to define the precise molecular interactions giving rise to these intricate regulatory pathways. To address this shortcoming, we have developed conformationally switchable artificial metalloproteins (swArMs) in which allosteric Gln-binding triggers protein conformational changes that impact the microenvironment surrounding an installed metallocofactor. Herein, we report a combined structural, spectroscopic, and computational approach to enhance the conformation-dependent changes in H-bond interactions surrounding the metallocofactor site of a swArM. Structure-informed molecular dynamics simulations were employed to predict point mutations that could enhance active site H-bond interactions preferentially in the Gln-bound holo-conformation of the swArM. Testing our predictions via the unique infrared spectral signals associated with the metallocofactor site, we have identified three key residues capable of imparting conformational control over the metallocofactor microenvironment. The resultant swArMs not only model biologically relevant structural regulation but also provide an enhanced Gln-responsive biological probe to be leveraged in future biosensing applications.

Deep learning guided design of dynamic proteins

Deep learning has greatly advanced design of highly stable static protein structures, but the controlled conformational dynamics that are hallmarks of natural switch-like signaling proteins have remained inaccessible to de novo design. Here, we describe a general deep-learning-guided approach for de novo design of dynamic changes between intra-domain geometries of proteins, similar to switch mechanisms prevalent in nature, with atom-level precision. We solve 4 structures validating the designed conformations, show microsecond transitions between them, and demonstrate that the conformational landscape can be modulated by orthosteric ligands and allosteric mutations. Physics-based simulations are in remarkable agreement with deep-learning predictions and experimental data, reveal distinct state-dependent residue interaction networks, and predict mutations that tune the designed conformational landscape. Our approach demonstrates that new modes of motion can now be realized through de novo design and provides a framework for constructing biology-inspired, tunable and controllable protein signaling behavior de novo .

Photocontrolled Reversible Amyloid Fibril Formation of Parathyroid Hormone-Derived Peptides

Peptide fibrillization is crucial in biological processes such as amyloid-related diseases and hormone storage, involving complex transitions between folded, unfolded, and aggregated states. We here employ light to induce reversible transitions between aggregated and nonaggregated states of a peptide, linked to the parathyroid hormone (PTH). The artificial light-switch 3-{[(4-aminomethyl)phenyl]diazenyl}benzoic acid (AMPB) is embedded into a segment of PTH, the peptide PTH25-37, to control aggregation, revealing position-dependent effects. Through in silico design, synthesis, and experimental validation of 11 novel PTH25-37-derived peptides, we predict and confirm the amyloid-forming capabilities of the AMPB-containing peptides. Quantum-chemical studies shed light on the photoswitching mechanism. Solid-state NMR studies suggest that β-strands are aligned parallel in fibrils of PTH25-37, while in one of the AMPB-containing peptides, β-strands are antiparallel. Simulations further highlight the significance of π-π interactions in the latter. This multifaceted approach enabled the identification of a peptide that can undergo repeated phototriggered transitions between fibrillated and defibrillated states, as demonstrated by different spectroscopic techniques. With this strategy, we unlock the potential to manipulate PTH to reversibly switch between active and inactive aggregated states, representing the first observation of a photostimulus-responsive hormone.

An integrative approach to protein sequence design through multiobjective optimization

With recent methodological advances in the field of computational protein design, in particular those based on deep learning, there is an increasing need for frameworks that allow for coherent, direct integration of different models and objective functions into the generative design process. Here we demonstrate how evolutionary multiobjective optimization techniques can be adapted to provide such an approach. With the established Non-dominated Sorting Genetic Algorithm II (NSGA-II) as the optimization framework, we use AlphaFold2 and ProteinMPNN confidence metrics to define the objective space, and a mutation operator composed of ESM-1v and ProteinMPNN to rank and then redesign the least favorable positions. Using the two-state design problem of the foldswitching protein RfaH as an in-depth case study, and PapD and calmodulin as examples of higher-dimensional design problems, we show that the evolutionary multiobjective optimization approach leads to significant reduction in the bias and variance in RfaH native sequence recovery, compared to a direct application of ProteinMPNN. We suggest that this improvement is due to three factors: (i) the use of an informative mutation operator that accelerates the sequence space exploration, (ii) the parallel, iterative design process inherent to the genetic algorithm that improves upon the ProteinMPNN autoregressive sequence decoding scheme, and (iii) the explicit approximation of the Pareto front that leads to optimal design candidates representing diverse tradeoff conditions. We anticipate this approach to be readily adaptable to different models and broadly relevant for protein design tasks with complex specifications.

An integrative approach to protein sequence design through multiobjective optimization

With recent methodological advances in the field of computational protein design, in particular those based on deep learning, there is an increasing need for frameworks that allow for coherent, direct integration of different models and objective functions into the generative design process. Here we demonstrate how evolutionary multiobjective optimization techniques can be adapted to provide such an approach. With the established Non-dominated Sorting Genetic Algorithm II (NSGA-II) as the optimization framework, we use AlphaFold2 and ProteinMPNN confidence metrics to define the objective space, and a mutation operator composed of ESM-1v and ProteinMPNN to rank and then redesign the least favorable positions. Using the multistate design problem of the foldswitching protein RfaH as an in-depth case study, we show that the evolutionary multiobjective optimization approach leads to significant reduction in the bias and variance in RfaH native sequence recovery, compared to a direct application of ProteinMPNN. We suggest that this improvement is due to three factors: (i) the use of an informative mutation operator that accelerates the sequence space exploration, (ii) the parallel, iterative design process inherent to the genetic algorithm that improves upon the ProteinMPNN autoregressive sequence decoding scheme, and (iii) the explicit approximation of the Pareto front that leads to optimal design candidates representing diverse tradeoff conditions. We anticipate this approach to be readily adaptable to different models and broadly relevant for protein design tasks with complex specifications.

De novo protein design-From new structures to programmable functions

Methods from artificial intelligence (AI) trained on large datasets of sequences and structures can now "write" proteins with new shapes and molecular functions de novo, without starting from proteins found in nature. In this Perspective, I will discuss the state of the field of de novo protein design at the juncture of physics-based modeling approaches and AI. New protein folds and higher-order assemblies can be designed with considerable experimental success rates, and difficult problems requiring tunable control over protein conformations and precise shape complementarity for molecular recognition are coming into reach. Emerging approaches incorporate engineering principles-tunability, controllability, and modularity-into the design process from the beginning. Exciting frontiers lie in deconstructing cellular functions with de novo proteins and, conversely, constructing synthetic cellular signaling from the ground up. As methods improve, many more challenges are unsolved.

Development of epistatic YES and AND protein logic gates and their assembly into signalling cascades

The construction and assembly of artificial allosteric protein switches into information and energy processing networks connected to both biological and non-biological systems is a central goal of synthetic biology and bionanotechnology. However, designing protein switches with the desired input, output and performance parameters is challenging. Here we use a range of reporter proteins to demonstrate that their chimeras with duplicated receptor domains produce YES gate protein switches with large (up to 9,000-fold) dynamic ranges and fast (minutes) response rates. In such switches, the epistatic interactions between largely independent synthetic allosteric sites result in an OFF state with minimal background noise. We used YES gate protein switches based on β-lactamase to develop quantitative biosensors of therapeutic drugs and protein biomarkers. Furthermore, we demonstrated the reconfiguration of YES gate switches into AND gate switches controlled by two different inputs, and their assembly into signalling networks regulated at multiple nodes.

Functional regulation of an intrinsically disordered protein via a conformationally excited state

A longstanding goal in the field of intrinsically disordered proteins (IDPs) is to characterize their structural heterogeneity and pinpoint the role of this heterogeneity in IDP function. Here, we use multinuclear chemical exchange saturation (CEST) nuclear magnetic resonance to determine the structure of a thermally accessible globally folded excited state in equilibrium with the intrinsically disordered native ensemble of a bacterial transcriptional regulator CytR. We further provide evidence from double resonance CEST experiments that the excited state, which structurally resembles the DNA-bound form of cytidine repressor (CytR), recognizes DNA by means of a "folding-before-binding" conformational selection pathway. The disorder-to-order regulatory switch in DNA recognition by natively disordered CytR therefore operates through a dynamical variant of the lock-and-key mechanism where the structurally complementary conformation is transiently accessed via thermal fluctuations.

Computational structural-based GPCR optimization for user-defined ligand: Implications for the development of biosensors

Organisms have developed effective mechanisms to sense the external environment. Human-designed biosensors exploit this natural optimization, where different biological machinery have been adapted to detect the presence of user-defined molecules. Specifically, the pheromone pathway in the model organism Saccharomyces cerevisiae represents a suitable candidate as a synthetic signaling system. Indeed, it expresses just one G-Protein Coupled Receptor (GPCR), Ste2, able to recognize pheromone and initiate the expression of pheromone-dependent genes. To date, the standard procedure to engineer this system relies on the substitution of the yeast GPCR with another one and on the modification of the yeast G-protein to bind the inserted receptor. Here, we propose an innovative computational procedure, based on geometrical and chemical optimization of protein binding pockets, to select the amino acid substitutions required to make the native yeast GPCR able to recognize a user-defined ligand. This procedure would allow the yeast to recognize a wide range of ligands, without a-priori knowledge about a GPCR recognizing them or the corresponding G protein. We used Monte Carlo simulations to design on Ste2 a binding pocket able to recognize epinephrine, selected as a test ligand. We validated Ste2 mutants via molecular docking and molecular dynamics. We verified that the amino acid substitutions we identified make Ste2 able to accommodate and remain firmly bound to epinephrine. Our results indicate that we sampled efficiently the huge space of possible mutants, proposing such a strategy as a promising starting point for the development of a new kind of S.cerevisiae-based biosensors.

Fluorometric biosensing of α-amylase using an artificial allosteric biosensor immobilized on nanostructured interface

Protein biosensors hold a promise to transform the way we collect physiological data by enabling quantification of biomarkers outside of specialized laboratory environment. However, achieving high specificity and sensitivity in homogeneous assay format remains challenging. Here we report construction of fluorescent biosensor arrays based on artificial allosteric α-amylase-activated PQQ-dependent glucose dehydrogenase (Amy-GDH). Amy-GDH was covalently immobilized on silica nanoparticles that were then arrayed on fiberglass sheets. The activity of the biosensor was monitored using a smartphone camera via emergence of bright fluorescence (λ_ex 365 nm) originating from reduced phenazine methosulfate upon glucose oxidation by Amy-GDH. We show that such biosensor arrays demonstrate an apparent K_d of 115 pM for α-amylase with a detection limit of 2 pM. Using the developed biosensor arrays, we were able to specifically and accurately quantify the concentration of α-amylase in biological fluids such as serum and saliva. We propose that the presented approach can enable construction of ultrasensitive point-of-care diagnostic arrays.

Design and characterization of a protein fold switching network

To better understand how amino acid sequence encodes protein structure, we engineered mutational pathways that connect three common folds (3α, β-grasp, and α/β-plait). The structures of proteins at high sequence-identity intersections in the pathways (nodes) were determined using NMR spectroscopy and analyzed for stability and function. To generate nodes, the amino acid sequence encoding a smaller fold is embedded in the structure of an ~50% larger fold and a new sequence compatible with two sets of native interactions is designed. This generates protein pairs with a 3α or β-grasp fold in the smaller form but an α/β-plait fold in the larger form. Further, embedding smaller antagonistic folds creates critical states in the larger folds such that single amino acid substitutions can switch both their fold and function. The results help explain the underlying ambiguity in the protein folding code and show that new protein structures can evolve via abrupt fold switching.

Smart genetically-encoded biosensors for the chemical monitoring of living systems

Genetically-encoded biosensors provide the all-optical and non-invasive visualization of dynamic biochemical events within living systems, which has allowed the discovery of profound new insights. Twenty-five years of biosensor development has steadily improved their performance and has provided us with an ever increasing biosensor repertoire. In this feature article, we present recent advances made in biosensor development and provide a perspective on the future direction of the field.

Metamorphic proteins under a computational microscope: Lessons from a fold-switching RfaH protein

Metamorphic proteins constitute unexpected paradigms of the protein folding problem, as their sequences encode two alternative folds, which reversibly interconvert within biologically relevant timescales to trigger different cellular responses. Once considered a rare aberration, metamorphism may be common among proteins that must respond to rapidly changing environments, exemplified by NusG-like proteins, the only transcription factors present in every domain of life. RfaH, a specialized paralog of bacterial NusG, undergoes an all-α to all-β domain switch to activate expression of virulence and conjugation genes in many animal and plant pathogens and is the quintessential example of a metamorphic protein. The dramatic nature of RfaH structural transformation and the richness of its evolutionary history makes for an excellent model for studying how metamorphic proteins switch folds. Here, we summarize the structural and functional evidence that sparked the discovery of RfaH as a metamorphic protein, the experimental and computational approaches that enabled the description of the molecular mechanism and refolding pathways of its structural interconversion, and the ongoing efforts to find signatures and general properties to ultimately describe the protein metamorphome.

Redox- and metal-directed structural diversification in designed metalloprotein assemblies

Herein we describe a designed protein building block whose self-assembly behaviour is dually gated by the redox state of disulphide bonds and the identity of exogenous metal ions. This protein construct is shown - through extensive structural and biophysical characterization - to access five distinct oligomeric states, exemplifying how the complex interplay between hydrophobic, metal-ligand, and reversible covalent interactions could be harnessed to obtain multiple, responsive protein architectures from a single building block.

Exploring Performance Parameters of Artificial Allosteric Protein Switches

Biological information processing networks rely on allosteric protein switches that dynamically interconvert biological signals. Construction of their artificial analogues is a central goal of synthetic biology and bioengineering. Receptor domain insertion is one of the leading methods for constructing chimeric protein switches. Here we present an in vitro expression-based platform for the analysis of chimeric protein libraries for which traditional cell survival or cytometric high throughput assays are not applicable. We utilise this platform to screen a focused library of chimeras between PQQ-glucose dehydrogenase and calmodulin. Using this approach, we identified 50 chimeras (approximately 23% of the library) that were activated by calmodulin-binding peptides. We analysed performance parameters of the active chimeras and demonstrated that their dynamic range and response times are anticorrelated, pointing to the existence of an inherent thermodynamic trade-off. We show that the structure of the ligand peptide affects both the response and activation kinetics of the biosensors suggesting that the structure of a ligand:receptor complex can influence the chimera's activation pathway. In order to understand the extent of structural changes in the reporter protein induced by the receptor domains, we have analysed one of the chimeric molecules by CD spectroscopy and hydrogen-deuterium exchange mass spectrometry. We concluded that subtle ligand-induced changes in the receptor domain propagated into the GDH domain and affected residues important for substrate and cofactor binding. Finally, we used one of the identified chimeras to construct a two-component rapamycin biosensor and demonstrated that core switch optimisation translated into improved biosensor performance.

Design and discovery of metamorphic proteins

Metamorphic proteins are single amino acid sequences that reversibly interconvert between multiple, dramatically different native structures, often with distinct functions. Since the discovery of the first metamorphic proteins in the early 2000s, several additional metamorphic proteins have been identified, and it was suggested that up to 4% of proteins in the PDB may switch folds. Metamorphic proteins have been found to share common features such as marginal thermostability and inconsistencies in predicted secondary structures. Outstanding challenges in the field include the search for more metamorphic proteins and the design of new proteins that switch folds. Identification of novel metamorphic proteins in nature will improve therapeutic targeting of fold-switching proteins involved in human pathology and will enhance the design of protein-based therapies. Designed fold switching proteins have applications as biosensors, molecular switches, molecular machines, and self-assembling systems.

Nanostructured Interface Loaded with Chimeric Enzymes for Fluorimetric Quantification of Cyclosporine A and FK506

Advances in protein engineering resulted in increased efforts to create protein biosensors that can replace instrumentation-heavy analytical and diagnostic methods. Sensitivity, amenability to multiplexing, and manufacturability remain to be among the key issues preventing broad utilization of protein biosensors. Here, we attempt to address these by constructing arrays utilizing protein biosensors based on the artificial allosteric variant of PQQ-glucose dehydrogenase (GDH). We demonstrated that the silica nanoparticle-immobilized GDH protein could be deposited on fiberglass sheets without loss of activity. The particle-associated GDH activity could be monitored using changes in the fluorescence of the commonly used electron mediator phenazine methosulfate. The constructed biosensor arrays of macrocyclic immunosuppressant drugs cyclosporine A and FK-506 displayed very low background and a remarkable dynamic range exceeding 300-fold that resulted in a limit of detection of 2 pM for both analytes. This enabled us to quantify both drugs in human blood, serum, urine, and saliva. The arrays could be stored in dry form and quantitatively imaged using a smartphone camera, demonstrating the method's suitability for field and point-of-care applications. The developed approach provides a generalizable platform for biosensor array development that is compatible with inexpensive and potentially scalable manufacturing.

Engineering and exploiting synthetic allostery of NanoLuc luciferase

Allostery enables proteins to interconvert different biochemical signals and form complex metabolic and signaling networks. We hypothesize that circular permutation of proteins increases the probability of functional coupling of new N- and C- termini with the protein's active center through increased local structural disorder. To test this we construct a synthetically allosteric version of circular permutated NanoLuc luciferase that can be activated through ligand-induced intramolecular non-covalent cyclisation. This switch module is tolerant of the structure of binding domains and their ligands, and can be used to create biosensors of proteins and small molecules. The developed biosensors covers a range of emission wavelengths and displays sensitivity as low as 50pM and dynamic range as high as 16-fold and could quantify their cognate ligand in human fluids. We apply hydrogen exchange kinetic mass spectroscopy to analyze time resolved structural changes in the developed biosensors and observe ligand-mediated folding of newly created termini.

Engineering a Fluorescent Protein Color Switch Using Entropy-Driven β-Strand Exchange

Protein conformational switches are widely used in biosensing. They are often composed of an input domain (which binds a target ligand) fused to an output domain (which generates an optical readout). A central challenge in designing such switches is to develop mechanisms for coupling the input and output signals via conformational changes. Here, we create a biosensor in which binding-induced folding of the input domain drives a conformational shift in the output domain that results in a sixfold green-to-yellow ratiometric fluorescence change in vitro and a 35-fold intensiometric fluorescence increase in cultured cells. The input domain consists of circularly permuted FK506 binding protein (cpFKBP) that folds upon binding its target ligand (FK506 or rapamycin). cpFKBP folding induces the output domain, an engineered green fluorescent protein (GFP) variant, to replace one of its β-strands (containing T203 and specifying green fluorescence) with a duplicate β-strand (containing Y203 and specifying yellow fluorescence) in an intramolecular exchange reaction. This mechanism employs the loop-closure entropy principle, embodied by the folding of the partially disordered cpFKBP domain, to couple ligand binding to the GFP color shift. This study highlights the high-energy barriers present in GFP folding which cause β-strand exchange to be slow and are also likely responsible for the shift from the β-strand exchange mechanism in vitro to ligand-induced chromophore maturation in cells. The proof-of-concept design has the advantages of full genetic encodability and potential for modularity. The latter attribute is enabled by the natural coupling of binding and folding and circular permutation of the input domain, which theoretically allows different binding domains to be compatible for insertion into the GFP surface loop.

OUP accepted manuscript

Potyviridae, the largest family of known RNA viruses (realm Riboviria), belongs to the picorna-like supergroup and has important agricultural and ecological impacts. Potyvirid genomes are translated into polyproteins, which are in turn hydrolyzed to release mature products. Recent sequencing efforts revealed an unprecedented number of potyvirids with a rich variability in gene content and genomic layouts. Here, we review the heterogeneity of non-core modules that expand the structural and functional diversity of the potyvirid proteomes. We provide a family-wide classification of P1 proteinases into the functional Types A and B, and discuss pretty interesting sweet potato potyviral ORF (PISPO), putative zinc fingers, and alkylation B (AlkB)—non-core modules found within P1 cistrons. The atypical inosine triphosphate pyrophosphatase (ITPase/HAM1), as well as the pseudo tobacco mosaic virus-like coat protein (TMV-like CP) are discussed alongside homologs of unrelated virus taxa. Family-wide abundance of the multitasking helper component proteinase (HC-pro) is revised. Functional connections between non-core modules are highlighted to support host niche adaptation and immune evasion as main drivers of the Potyviridae evolutionary radiation. Potential biotechnological and synthetic biology applications of potyvirid leader proteinases and non-core modules are finally explored.

Deep learning approaches for conformational flexibility and switching properties in protein design

Following the hugely successful application of deep learning methods to protein structure prediction, an increasing number of design methods seek to leverage generative models to design proteins with improved functionality over native proteins or novel structure and function. The inherent flexibility of proteins, from side-chain motion to larger conformational reshuffling, poses a challenge to design methods, where the ideal approach must consider both the spatial and temporal evolution of proteins in the context of their functional capacity. In this review, we highlight existing methods for protein design before discussing how methods at the forefront of deep learning-based design accommodate flexibility and where the field could evolve in the future.