Search

Protein engineering has nearly limitless applications across chemistry, energy, and medicine, but creating new proteins with improved or novel functions remains slow, labor-intensive, and inefficient. In this work, we present theSelf-driving Autonomous Machines for Protein Landscape Exploration(SAMPLE) platform for fully autonomous protein engineering. SAMPLE is driven by an intelligent agent that learns protein sequence-function relationships, designs new proteins, and sends designs to a fully automated robotic system that experimentally tests designed proteins and provides feedback to improve the agent’s understanding of the system. We deployed four SAMPLE agents with the goal of engineering glycoside hydrolase enzymes with enhanced thermal tolerance. Despite showing individual differences in their search behavior, all four agents quickly converged on thermostable enzymes that were at least 12 °C more stable than the starting sequences. Self-driving laboratories automate and accelerate the scientific discovery process and hold great potential for the fields of protein engineering and synthetic biology.

Alcohol-forming fatty acyl reductases (FARs) catalyze the reduction of thioesters to alcohols and are key enzymes for microbial production of fatty alcohols. Many metabolic engineering strategies utilize FARs to produce fatty alcohols from intracellular acyl-CoA and acyl-ACP pools; however, enzyme activity, especially on acyl-ACPs, remains a significant bottleneck to high-flux production. Here, we engineer FARs with enhanced activity on acyl-ACP substrates by implementing a machine learning (ML)-driven approach to iteratively search the protein fitness landscape. Over the course of ten design-test-learn rounds, we engineer enzymes that produce over twofold more fatty alcohols than the starting natural sequences. We characterize the top sequence and show that it has an enhanced catalytic rate on palmitoyl-ACP. Finally, we analyze the sequence-function data to identify features, like the net charge near the substrate-binding site, that correlate with in vivo activity. This work demonstrates the power of ML to navigate the fitness landscape of traditionally difficult-to-engineer proteins., Fatty acyl reductases (FARs) are critical enzymes in the biosynthesis of fatty alcohols and have the ability to directly acces acyl-ACP substrates. Here, authors couple machine learning-based protein engineering framework with gene shuffling to optimize a FAR for the activity on acyl-ACP and improve fatty alcohol production.

Directed laboratory evolution applies iterative rounds of mutation and selection to explore the protein fitness landscape and provides rich information regarding the underlying relationships between protein sequence, structure, and function. Laboratory evolution data consist of protein sequences sampled from evolving populations over multiple generations and this data type does not fit into established supervised and unsupervised machine learning approaches. We develop a statistical learning framework that models the evolutionary process and can infer the protein fitness landscape from multiple snapshots along an evolutionary trajectory. We apply our modeling approach to dihydrofolate reductase (DHFR) laboratory evolution data and the resulting landscape parameters capture important aspects of DHFR structure and function. We use the resulting model to understand the structure of the fitness landscape and find numerous examples of epistasis but an overall global peak that is evolutionarily accessible from most starting sequences. Finally, we use the model to perform an in silico extrapolation of the DHFR laboratory evolution trajectory and computationally design proteins from future evolutionary rounds.

Angiotensin Converting-Enzyme 2 (ACE2) is currently being investigated for its ability to beneficially modulate the Angiotensin receptor (ATR) therapeutic axis to treat multiple human diseases, but its broad substrate scope and diverse physiological roles limit its potential as a therapeutic agent. In this work we engineer ACE2 variants with enhanced hydrolytic activity and specificity toward Angiotensin-II (Ang-II) that may lead to new therapeutic candidates with improved efficacy and reduced off-target side effects. We established a yeast display-based liquid chromatography screen that enabled use of directed evolution to discover ACE2 variants with improved Ang-II activity and specificity relative to the off-target peptide substrates Apelin-13 and Angiotensin-I (Ang-I). We screened ACE2 active site libraries to reveal three substitution-tolerant positions (M360, T371 and Y510) that can be mutated to enhance ACE2’s activity profile and followed up on these hits with focused double mutant libraries to further improve the enzyme. Relative to wildtype ACE2, our top variant (T371L/Y510Ile) displayed a sevenfold increase in Ang-II turnover number (kcat), a sixfold diminished catalytic efficiency (kcat/Km) on Apelin-13, and an overall decreased activity on other ACE2 substrates that were not directly assayed in directed evolution screen. At physiologically relevant substrate concentrations, T371L/Y510Ile hydrolyzes more Ang-II than wildtype ACE2 with concomitant Ang-II:Apelin-13 specificity improvements reaching 30-fold. Our efforts have delivered ATR axis-acting therapeutic candidates with relevance to both established and unexplored ACE2 therapeutic applications and provide a foundation for further ACE2 engineering efforts.Significance StatementThe Angiotensin Converting Enzyme 2 (ACE2) carboxypeptidase is being clinically trialed for treatment of acute respiratory distress syndrome (ARDS) related to traumatic injuries and viral infections. We developed an enzyme engineering platform that enabled discovery of ACE2 variants possessing enhanced peptide hydrolysis activity and specificity profiles relevant to ARDS treatment. Beyond their potential utility as ARDS therapeutics, our improved ACE2 variants could be further developed for new unexplored ACE2 therapeutic contexts such as treating Alzheimer’s Disease.

Understanding how SARS-CoV-2 interacts with different mammalian angiotensin-converting enzyme II (ACE2) cell entry receptors will help elucidate determinants of intra- and cross-species virus transmission, facilitate development of effective new vaccines for both humans and livestock animals, and guide livestock farming and coronavirus screening procedures to ensure food supply security. In this work we applied laboratory directed evolution to several mammalian ACE2s with the goal of identifying conserved ACE2 mutations that increase spike binding affinity across multiple species. We found the Gln42Leu mutation increased ACE2-spike binding for human as well as four of four other mammalian ACE2s, while the Leu79Ile mutation had a similar effect for human and three of three mammalian ACE2 orthologs. These results are especially notable given the residues' high levels of representation, i.e, 83% for Gln42 and 56% for Leu79, among annotated mammalian ACE2s. We also found that substitutions at ACE2 position 34, which is relatively variable across mammalian ACE2s, increased binding for multiple ACE2 orthologs. Taken together, these results speak strongly to the plausibility of SARS-CoV-2 strains with increased ability to cross species transmission barriers. Our results can guide further computational and experimental studies to develop biomedical technologies and animal husbandry practices that help protect both humans and livestock from existing and future SARS-CoV-2 variants.

Significance Understanding the relationship between protein sequence and function is necessary to design new and useful proteins with applications in bioenergy, medicine, and agriculture. The mapping from sequence to function is tremendously complex because it involves thousands of molecular interactions that are coupled over multiple lengths and timescales. We show that neural networks can learn the sequence–function mapping from large protein datasets. Neural networks are appealing for this task because they can learn complicated relationships from data, make few assumptions about the nature of the sequence–function relationship, and can learn general rules that apply across the length of the protein sequence. We demonstrate that learned models can be applied to design new proteins with properties that exceed natural sequences., The mapping from protein sequence to function is highly complex, making it challenging to predict how sequence changes will affect a protein’s behavior and properties. We present a supervised deep learning framework to learn the sequence–function mapping from deep mutational scanning data and make predictions for new, uncharacterized sequence variants. We test multiple neural network architectures, including a graph convolutional network that incorporates protein structure, to explore how a network’s internal representation affects its ability to learn the sequence–function mapping. Our supervised learning approach displays superior performance over physics-based and unsupervised prediction methods. We find that networks that capture nonlinear interactions and share parameters across sequence positions are important for learning the relationship between sequence and function. Further analysis of the trained models reveals the networks’ ability to learn biologically meaningful information about protein structure and mechanism. Finally, we demonstrate the models’ ability to navigate sequence space and design new proteins beyond the training set. We applied the protein G B1 domain (GB1) models to design a sequence that binds to immunoglobulin G with substantially higher affinity than wild-type GB1.

Single-nucleotide polymorphisms (SNPs) are the most common source of genetic variation between individuals and have implications in human disease, pathogen drug resistance, and agriculture. SNPs are typically detected using DNA sequencing, which requires advanced sample preparation and instrumentation, and thus cannot be deployed for on-site testing or in low-resource settings. In this work we have developed a simple and robust assay to rapidly detect SNPs in nucleic acid samples. Our approach combines LAMP-based target amplification with fluorescent probes to detect SNPs with high specificity in a one-pot reaction format. A competitive “sink” strand preferentially binds to off-target products and shifts the free energy landscape to favor specific activation by SNP products. We demonstrated the broad utility and reliability of our SNP-LAMP method by detecting three distinct SNPs across the human genome. We also designed an assay to rapidly detect highly transmissible SARS-CoV-2 variants. This work demonstrates that competitive SNP-LAMP is a powerful and universal method that could be applied in point-of-care settings to detect any target SNP with high specificity and sensitivity.

The human caspase family comprises 12 cysteine proteases that are centrally involved in cell death and inflammation responses. The members of this family have conserved sequences and structures, highly similar enzymatic activities and substrate preferences, and overlapping physiological roles. In this paper, we present a deep mutational scan of the executioner caspases CASP3 and CASP7 to dissect differences in their structure, function, and regulation. Our approach leverages high-throughput microfluidic screening to analyze hundreds of thousands of caspase variants in tightly controlled in vitro reactions. The resulting data provides a large-scale and unbiased view of the impact of amino acid substitutions on the proteolytic activity of CASP3 and CASP7. We use this data to pinpoint key functional differences between CASP3 and CASP7, including a secondary internal cleavage site, CASP7 Q196 that is not present in CASP3. Our results will open avenues for inquiry in caspase function and regulation that could potentially inform the development of future caspase-specific therapeutics.

The human caspase family comprises 12 cysteine proteases that are centrally involved in cell death and inflammation responses. The members of this family have conserved sequences and structures, highly similar enzymatic activities and substrate preferences, and overlapping physiological roles. In this paper, we present a deep mutational scan of the executioner caspases CASP3 and CASP7 to dissect differences in their structure, function, and regulation. Our approach leverages high-throughput microfluidic screening to analyze hundreds of thousands of caspase variants in tightly controlled in vitro reactions. The resulting data provides a large-scale and unbiased view of the impact of amino acid substitutions on the proteolytic activity of CASP3 and CASP7. We use this data to pinpoint key functional differences between CASP3 and CASP7, including a secondary internal cleavage site, CASP7 Q196 that is not present in CASP3. Our results will open avenues for inquiry in caspase function and regulation that could potentially inform the development of future caspasespecific therapeutics.

Understanding how human ACE2 genetic variants differ in their recognition by SARS-CoV-2 can facilitate the leveraging of ACE2 as an axis for treating and preventing COVID-19. In this work, we experimentally interrogate thousands of ACE2 mutants to identify over one hundred human single-nucleotide variants (SNVs) that are likely to have altered recognition by the virus, and make the complementary discovery that ACE2 residues distant from the spike interface influence the ACE2-spike interaction. These findings illuminate new links between ACE2 sequence and spike recognition, and could find substantial utility in further fundamental research that augments epidemiological analyses and clinical trial design in the contexts of both existing strains of SARS-CoV-2 and novel variants that may arise in the future.

In order to make renewable fuels and chemicals from microbes, new methods are required to engineer microbes more intelligently. Computational approaches, to engineer strains for enhanced chemical production typically rely on detailed mechanistic models (e.g., kinetic/stoichiometric models of metabolism)—requiring many experimental datasets for their parameterization—while experimental methods may require screening large mutant libraries to explore the design space for the few mutants with desired behaviors. To address these limitations, we developed an active and machine learning approach (ActiveOpt) to intelligently guide experiments to arrive at an optimal phenotype with minimal measured datasets. ActiveOpt was applied to two separate case studies to evaluate its potential to increase valine yields and neurosporene productivity in Escherichia coli. In both the cases, ActiveOpt identified the best performing strain in fewer experiments than the case studies used. This work demonstrates that machine and active learning approaches have the potential to greatly facilitate metabolic engineering efforts to rapidly achieve its objectives.

Machine learning (ML) is revolutionizing our ability to understand and predict the complex relationships between protein sequence, structure, and function. Predictive sequence-function models are enabling protein engineers to efficiently search the sequence space for useful proteins with broad applications in biotechnology. In this review, we highlight the recent advances in applying ML to protein engineering. We discuss supervised learning methods that infer the sequence-function mapping from experimental data and new sequence representation strategies for data-efficient modeling. We then describe the various ways in which ML can be incorporated into protein engineering workflows, including purely in silico searches, ML-assisted directed evolution, and generative models that can learn the underlying distribution of the protein function in a sequence space. ML-driven protein engineering will become increasingly powerful with continued advances in high-throughput data generation, data science, and deep learning.

The mapping from protein sequence to function is highly complex, making it challenging to predict how sequence changes will affect a protein’s behavior and properties. We present a supervised deep learning framework to learn the sequence-function mapping from deep mutational scanning data and make predictions for new, uncharacterized sequence variants. We test multiple neural network architectures, including a graph convolutional network that incorporates protein structure, to explore how a network’s internal representation affects its ability to learn the sequence-function mapping. Our supervised learning approach displays superior performance over physics-based and unsupervised prediction methods. We find networks that capture nonlinear interactions and share parameters across sequence positions are important for learning the relationship between sequence and function. Further analysis of the trained models reveals the networks’ ability to learn biologically meaningful information about protein structure and mechanism. Finally, we demonstrate the models’ ability to navigate sequence space and design new proteins beyond the training set. We applied the GB1 models to design a sequence that binds to IgG with substantially higher affinity than wild-type GB1. Our software is available from https://github.com/gitter-lab/nn4dms.SignificanceUnderstanding the relationship between protein sequence and function is necessary to design new and useful proteins with applications in bioenergy, medicine, and agriculture. The mapping from sequence to function is tremendously complex because it involves thousands of molecular interactions that are coupled over multiple lengths and timescales. In this work, we show neural networks can learn the sequence-function mapping from large protein datasets. Neural networks are appealing for this task because they can learn complicated relationships from data, make few assumptions about the nature of the sequencefunction relationship, and can learn general rules that apply across the length of the protein sequence. We demonstrate the learned models can be applied to design new proteins with properties that exceed natural sequences.

SummaryMachine learning can infer how protein sequence maps to function without requiring a detailed understanding of the underlying physical or biological mechanisms. It’s challenging to apply existing supervised learning frameworks to large-scale experimental data generated by deep mutational scanning (DMS) and related methods. DMS data often contain high dimensional and correlated sequence variables, experimental sampling error and bias, and the presence of missing data. Importantly, most DMS data do not contain examples of negative sequences, making it challenging to directly estimate how sequence affects function. Here, we develop a positive-unlabeled (PU) learning framework to infer sequence-function relationships from large-scale DMS data. Our PU learning method displays excellent predictive performance across ten large-scale sequence-function data sets, representing proteins of different folds, functions, and library types. The estimated parameters pinpoint key residues that dictate protein structure and function. Finally, we apply our statistical sequence-function model to design highly stabilized enzymes.

Directed evolution continues to expand the capabilities of complex biomolecules for a range of applications, such as adeno-associated virus vectors for gene therapy; however, advances in library design and selection strategies are key to develop variants that overcome barriers to clinical translation. To address this need, we applied structure-guided SCHEMA recombination of the multimeric adeno-associated virus (AAV) capsid to generate a highly diversified chimeric library with minimal structural disruption. A stringent in vivo Cre-dependent selection strategy was implemented to identify variants that transduce adult neural stem cells (NSCs) in the subventricular zone. A novel variant, SCH9, infected 60% of NSCs and mediated 24-fold higher GFP expression and a 12-fold greater transduction volume than AAV9. SCH9 utilizes both galactose and heparan sulfate as cell surface receptors and exhibits increased resistance to neutralizing antibodies. These results establish the SCHEMA library as a valuable tool for directed evolution and SCH9 as an effective gene delivery vector to investigate subventricular NSCs.

The spatial organization of microbial communities arises from a complex interplay of biotic and abiotic interactions, and is a major determinant of ecosystem functions. Here we design a microfluidic platform to investigate how the spatial arrangement of microbes impacts gene expression and growth. We elucidate key biochemical parameters that dictate the mapping between spatial positioning and gene expression patterns. We show that distance can establish a low-pass filter to periodic inputs and can enhance the fidelity of information processing. Positive and negative feedback can play disparate roles in the synchronization and robustness of a genetic oscillator distributed between two strains to spatial separation. Quantification of growth and metabolite release in an amino-acid auxotroph community demonstrates that the interaction network and stability of the community are highly sensitive to temporal perturbations and spatial arrangements. In sum, our microfluidic platform can quantify spatiotemporal parameters influencing diffusion-mediated interactions in microbial consortia., The spatial organisation of microbial communities is caused by the interplay of biotic and abiotic factors. Here the authors design a microfluidic platform to quantify the spatiotemporal parameters influencing diffusion-mediated interactions, and use this device to investigate information transmission and metabolic cross-feeding in synthetic microbial consortia.

Experimental methods that capture the individual properties of single cells are revealing the key role of cell-to-cell variability in countless biological processes. These single-cell methods are becoming increasingly important across the life sciences in fields such as immunology, regenerative medicine, and cancer biology. Existing single-cell analysis methods are often limited by their low analysis throughput, their inability to profile high-dimensional phenotypes, and complicated experimental workflows with slow turnaround times. In this work, we present Single-cell Nucleic Acid Profiling in Droplets (SNAPD) to analyze the transcriptional states of hundreds of thousands of single mammalian cells. Individual cells are encapsulated in aqueous droplets on a microfluidic chip and the content of each cell is profiled by amplifying a targeted panel of transcriptional markers. Molecular logic circuits then integrate this multi-dimensional information to categorize cells based on their transcriptional profile and produce a detectable fluorescence output. SNAPD analyzes over 100,000 cells per hour and can be used to quantify distinct cell types within populations, detect rare cells at frequencies down to 0.1%, and enrich specific cell types using microfluidic sorting. SNAPD provides a simple, rapid, low cost, and scalable approach to study complex phenotypes in heterogeneous cell populations.

Microbial interactions are major drivers of microbial community dynamics and functions. However, microbial interactions are challenging to decipher due to limitations in parallel culturing of sub-communities across many environments and accurate absolute abundance quantification of constituent members of the consortium. To this end, we developed Microbial Interaction Network Inference in microdroplets (MINI-Drop), a high-throughput method to rapidly infer microbial interactions in microbial consortia in microfluidic droplets. Fluorescence microscopy coupled to automated computational droplet and cell detection was used to rapidly determine the absolute abundance of each strain in hundreds to thousands of droplets per experiment. We show that MINI-Drop can accurately infer pairwise as well as higher-order interactions using a microbial interaction toolbox of defined microbial interactions mediated by distinct molecular mechanisms. MINI-Drop was used to investigate how the molecular composition of the environment alters the interaction network of a three-member consortium. To provide insight into the variation in community states across droplets, we developed a probabilistic model of cell growth modified by microbial interactions. In sum, we demonstrate a robust and generalizable method to probe cellular interaction networks by random encapsulation of sub-communities into microfluidic droplets.

Summary Microbial interactions are major drivers of microbial community dynamics and functions but remain challenging to identify because of limitations in parallel culturing and absolute abundance quantification of community members across environments and replicates. To this end, we developed Microbial Interaction Network Inference in microdroplets (MINI-Drop). Fluorescence microscopy coupled to computer vision techniques were used to rapidly determine the absolute abundance of each strain in hundreds to thousands of droplets per condition. We showed that MINI-Drop could accurately infer pairwise and higher-order interactions in synthetic consortia. We developed a stochastic model of community assembly to provide insight into the heterogeneity in community states across droplets. Finally, we elucidated the complex web of interactions linking antibiotics and different species in a synthetic consortium. In sum, we demonstrated a robust and generalizable method to infer microbial interaction networks by random encapsulation of sub-communities into microfluidic droplets.

We introduce a method for identifying elements of a protein structure that can be shuffled to make chimeric proteins from two or more homologous parents. Formulating recombination as a graph-partitioning problem allows us to identify noncontiguous segments of the sequence that should be inherited together in the progeny proteins. We demonstrate this noncontiguous recombination approach by constructing a chimera of β-glucosidases from two different kingdoms of life. Although the protein's alpha–beta barrel fold has no obvious subdomains for recombination, noncontiguous SCHEMA recombination generated a functional chimera that takes approximately half its structure from each parent. The X-ray crystal structure shows that the structural blocks that make up the chimera maintain the backbone conformations found in their respective parental structures. Although the chimera has lower β-glucosidase activity than the parent enzymes, the activity was easily recovered by directed evolution. This simple method, which does not rely on detailed atomic models, can be used to design chimeras that take structural, and functional, elements from distantly-related proteins.

Arginases catalyze the divalent cation-dependent hydrolysis of l-arginine to urea and l-ornithine. There is significant interest in using arginase as a therapeutic anti-neogenic agent against l-arginine auxotrophic tumors and in enzyme replacement therapy for treating hyperargininemia. Both therapeutic applications require enzymes with sufficient stability under physiological conditions. To explore sequence elements that contribute to arginase stability we used SCHEMA-guided recombination to design a library of chimeric enzymes composed of sequence fragments from the two human isozymes Arginase I and II. We then developed a novel active learning algorithm that selects sequences from this library that are both highly informative and functional. Using high-throughput gene synthesis and our two-step active learning algorithm, we were able to rapidly create a small but highly informative set of seven enzymatically active chimeras that had an average variant distance of 40 mutations from the closest parent arginase. Within this set of sequences, linear regression was used to identify the sequence elements that contribute to the long-term stability of human arginase under physiological conditions. This approach revealed a striking correlation between the isoelectric point and the long-term stability of the enzyme to deactivation under physiological conditions.

We describe an efficient SCHEMA recombination-based approach for screening homologous enzymes to identify stabilizing amino acid sequence blocks. This approach has been used to generate active, thermostable cellobiohydrolase class I (CBH I) enzymes from the 390 625 possible chimeras that can be made by swapping eight blocks from five fungal homologs. Constructing and characterizing the parent enzymes and just 32 ‘monomeras’ containing a single block from a homologous enzyme allowed stability contributions to be assigned to 36 of the 40 blocks from which the CBH I chimeras can be assembled. Sixteen of 16 predicted thermostable chimeras, with an average of 37 mutations relative to the closest parent, are more thermostable than the most stable parent CBH I, from the thermophilic fungus Talaromyces emersonii. Whereas none of the parent CBH Is were active >65°C, stable CBH I chimeras hydrolyzed solid cellulose at 70°C. In addition to providing a collection of diverse, thermostable CBH Is that can complement previously described stable CBH II chimeras (Heinzelman et al., Proc. Natl Acad. Sci. USA 2009;106:5610–5615) in formulating application-specific cellulase mixtures, the results show the utility of SCHEMA recombination for screening large swaths of natural enzyme sequence space for desirable amino acid blocks.

Directed evolution circumvents our profound ignorance of how a protein's sequence encodes its function by using iterative rounds of random mutation and artificial selection to discover new and useful proteins. Proteins can be tuned to adapt to new functions or environments via simple adaptive walks involving small numbers of mutations. Directed evolution studies have demonstrated how rapidly at least some proteins can evolve under strong selection pressures, and, because the entire ‘fossil record’ of evolutionary intermediates is available for detailed study, they have provided new insight into the relationship between sequence and function. Directed evolution has also shown how mutations that are functionally neutral can set the stage for further adaptation.

Our awe about shapes and colors do not stop with our eyes; bound with imagination and curiosity, humanity has peered into these materials as symbols of strength, wisdom, as well as tranquility. We peer into its luster as our forefathers have done in the past and our children will in the future, to revel in what we see from its reflection in the present. We can rest to the solace and calm of its organization and its permanence will be present for generations to come. Crystals come in all forms, sizes, and colors. Whether they be of the metallic, ionic, covalent, or molecular kind, the word “crystal” denotes order and stability. A material that is “crystalline” is composed of well-organized and “packed” constituent subunits. The crystals from a biological source intercalate this crystalline order with organic macromolecules to endow the assemblies with other physical characteristics to regulate form and function (i.e., size, hydrophobicity, electronegativity, and functional bonding). In this chapter, ionic crystals of biogenic origin as well as the biological mechanisms involved in the formation and growth of these diverse crystalline materials from Nature will be explored.

Natural enzymes are incredibly proficient catalysts, but engineering them to have new or improved functions is challenging due to the complexity of how an enzyme's sequence relates to its biochemical properties. Here, we present an ultrahigh-throughput method for mapping enzyme sequence-function relationships that combines droplet microfluidic screening with next-generation DNA sequencing. We apply our method to map the activity of millions of glycosidase sequence variants. Microfluidic-based deep mutational scanning provides a comprehensive and unbiased view of the enzyme function landscape. The mapping displays expected patterns of mutational tolerance and a strong correspondence to sequence variation within the enzyme family, but also reveals previously unreported sites that are crucial for glycosidase function. We modified the screening protocol to include a high-temperature incubation step, and the resulting thermotolerance landscape allowed the discovery of mutations that enhance enzyme thermostability. Droplet microfluidics provides a general platform for enzyme screening that, when combined with DNA-sequencing technologies, enables high-throughput mapping of enzyme sequence space.

SCHEMA structure-guided recombination is an effective method for producing families of protein chimeras having high sequence diversity, functional diversity, and thermostabilities greater than any of the parent proteins from which the chimeras are made. A key feature of SCHEMA chimera families is their amenability to a "sample, model, and predict" operation that allows one to characterize members of a small chimera sample set and use those data to construct models that accurately predict the properties of every member of the family. In this chapter, we describe applications of this "sample, model, and predict" approach and outline methods for designing chimera sample sets that enable efficient construction of models to identify useful sequence elements. With these models we can also predict the sequences and properties of the most desirable chimeras.

Knowing how protein sequence maps to function (the “fitness landscape”) is critical for understanding protein evolution as well as for engineering proteins with new and useful properties. We demonstrate that the protein fitness landscape can be inferred from experimental data, using Gaussian processes, a Bayesian learning technique. Gaussian process landscapes can model various protein sequence properties, including functional status, thermostability, enzyme activity, and ligand binding affinity. Trained on experimental data, these models achieve unrivaled quantitative accuracy. Furthermore, the explicit representation of model uncertainty allows for efficient searches through the vast space of possible sequences. We develop and test two protein sequence design algorithms motivated by Bayesian decision theory. The first one identifies small sets of sequences that are informative about the landscape; the second one identifies optimized sequences by iteratively improving the Gaussian process model in regions of the landscape that are predicted to be optimized. We demonstrate the ability of Gaussian processes to guide the search through protein sequence space by designing, constructing, and testing chimeric cytochrome P450s. These algorithms allowed us to engineer active P450 enzymes that are more thermostable than any previously made by chimeragenesis, rational design, or directed evolution.

SCHEMA structure-guided recombination is an effective method for producing families of protein chimeras having high sequence diversity, functional diversity, and thermostabilities greater than any of the parent proteins from which the chimeras are made. A key feature of SCHEMA chimera families is their amenability to a “sample, model, and predict” operation that allows one to characterize members of a small chimera sample set and use those data to construct models that accurately predict the properties of every member of the family. In this chapter, we describe applications of this “sample, model, and predict” approach and outline methods for designing chimera sample sets that enable efficient construction of models to identify useful sequence elements. With these models we can also predict the sequences and properties of the most desirable chimeras.

The production of contrast agents sensitive to neuronal signaling events is a rate-limiting step in the development of molecular-level functional magnetic resonance imaging (molecular fMRI) approaches for studying the brain. High-throughput generation and evaluation of potential probes are possible using techniques for macromolecular engineering of protein-based contrast agents. In an initial exploration of this strategy, we used the method of directed evolution to identify mutants of a bacterial heme protein that allowed detection of the neurotransmitter dopamine in vitro and in living animals. The directed evolution method involves successive cycles of mutagenesis and screening that could be generalized to produce contrast agents sensitive to a variety of molecular targets in the nervous system.

Microdroplets are typically generated by one of two microfluidic geometries, the T-junction and flow focusing. These two geometries are often thought to form drops through different mechanisms. Here, by directly measuring the pressures in the drop maker, we show that flow focus devices exhibit pressure fluctuations that are essentially identical to those found in T-junctions, suggesting that, in these devices and low-to-moderate capillary number, the drop formation process is also dominated by interfacial stresses and proceeds through a plugging–squeezing process.

We introduce a method for identifying elements of a protein structure that can be shuffled to make chimeric proteins from two or more homologous parents. Formulating recombination as a graph-partitioning problem allows us to identify noncontiguous segments of the sequence that should be inherited together in the progeny proteins. We demonstrate this noncontiguous recombination approach by constructing a chimera of β-glucosidases from two different kingdoms of life. Although the protein's alpha–beta barrel fold has no obvious subdomains for recombination, noncontiguous SCHEMA recombination generated a functional chimera that takes approximately half its structure from each parent. The X-ray crystal structure shows that the structural blocks that make up the chimera maintain the backbone conformations found in their respective parental structures. Although the chimera has lower β-glucosidase activity than the parent enzymes, the activity was easily recovered by directed evolution. This simple method, which does not rely on detailed atomic models, can be used to design chimeras that take structural, and functional, elements from distantly-related proteins.

The development of molecular probes that allow in vivo imaging of neural signaling processes with high temporal and spatial resolution remains challenging. Here we applied directed evolution techniques to create magnetic resonance imaging (MRI) contrast agents sensitive to the neurotransmitter dopamine. The sensors were derived from the heme domain of the bacterial cytochrome P450-BM3 (BM3h). Ligand binding to a site near BM3h's paramagnetic heme iron led to a drop in MRI signal enhancement and a shift in optical absorbance. Using an absorbance-based screen, we evolved the specificity of BM3h away from its natural ligand and toward dopamine, producing sensors with dissociation constants for dopamine of 3.3-8.9 microM. These molecules were used to image depolarization-triggered neurotransmitter release from PC12 cells and in the brains of live animals. Our results demonstrate the feasibility of molecular-level functional MRI using neural activity-dependent sensors, and our protein engineering approach can be generalized to create probes for other targets.

Background Many of the mutations accumulated by naturally evolving proteins are neutral in the sense that they do not significantly alter a protein's ability to perform its primary biological function. However, new protein functions evolve when selection begins to favor other, "promiscuous" functions that are incidental to a protein's original biological role. If mutations that are neutral with respect to a protein's primary biological function cause substantial changes in promiscuous functions, these mutations could enable future functional evolution. Results Here we investigate this possibility experimentally by examining how cytochrome P450 enzymes that have evolved neutrally with respect to activity on a single substrate have changed in their abilities to catalyze reactions on five other substrates. We find that the enzymes have sometimes changed as much as four-fold in the promiscuous activities. The changes in promiscuous activities tend to increase with the number of mutations, and can be largely rationalized in terms of the chemical structures of the substrates. The activities on chemically similar substrates tend to change in a coordinated fashion, potentially providing a route for systematically predicting the change in one activity based on the measurement of several others. Conclusion Our work suggests that initially neutral genetic drift can lead to substantial changes in protein functions that are not currently under selection, in effect poising the proteins to more readily undergo functional evolution should selection favor new functions in the future. Reviewers This article was reviewed by Martijn Huynen, Fyodor Kondrashov, and Dan Tawfik (nominated by Christoph Adami).