Neural network extrapolation to distant regions of the protein fitness landscape.

Machine learning (ML) has transformed protein engineering by constructing models of the underlying sequence-function landscape to accelerate the discovery of new biomolecules. ML-guided protein design requires models, trained on local sequence-function information, to accurately predict distant fitness peaks. In this work, we evaluate neural networks' capacity to extrapolate beyond their training data. We perform model-guided design using a panel of neural network architectures trained on protein G (GB1)-Immunoglobulin G (IgG) binding data and experimentally test thousands of GB1 designs to systematically evaluate the models' extrapolation. We find each model architecture infers markedly different landscapes from the same data, which give rise to unique design preferences. We find simpler models excel in local extrapolation to design high fitness proteins, while more sophisticated convolutional models can venture deep into sequence space to design proteins that fold but are no longer functional. Our findings highlight how each architecture's inductive biases prime them to learn different aspects of the protein fitness landscape.

Directed evolution of angiotensin-converting enzyme 2 peptidase activity profiles for therapeutic applications.

Angiotensin-converting enzyme 2 (ACE2) has been investigated for its ability to beneficially modulate the angiotensin receptor (ATR) therapeutic axis to treat multiple human diseases. Its broad substrate scope and diverse physiological roles, however, limit its potential as a therapeutic agent. In this work, we address this limitation by establishing a yeast display-based liquid chromatography screen that enabled use of directed evolution to discover ACE2 variants that possess both wild-type or greater Ang-II hydrolytic activity and improved specificity toward Ang-II relative to the off-target peptide substrate Apelin-13. To obtain these results, 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 wild-type 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 the directed evolution screen. At physiologically relevant substrate concentrations, T371L/Y510Ile hydrolyzes as much or more Ang-II than wild-type 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.

Yeast surface display-based identification of ACE2 mutations that modulate SARS-CoV-2 spike binding across multiple mammalian species.

Understanding how severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) interacts with different mammalian angiotensin-converting enzyme II (ACE2) cell entry receptors elucidates determinants of virus transmission and facilitates development of vaccines for humans and animals. Yeast display-based directed evolution identified conserved ACE2 mutations that increase spike binding across multiple species. Gln42Leu increased ACE2-spike binding for human and four of four other mammalian ACE2s; Leu79Ile had an effect for human and three of three mammalian ACE2s. These residues are highly represented, 83% for Gln42 and 56% for Leu79, among mammalian ACE2s. The above findings can be important in protecting humans and animals from existing and future SARS-CoV-2 variants.

Yeast surface display-based identification of ACE2 mutations that modulate SARS-CoV-2 spike binding across multiple mammalian species.

Understanding how SARS-CoV-2 interacts with different mammalian angiotensin-converting enzyme II (ACE2) cell entry receptors elucidates determinants of virus transmission and facilitates development of vaccines for humans and animals. Yeast display-based directed evolution identified conserved ACE2 mutations that increase spike binding across multiple species. Gln42Leu increased ACE2-spike binding for human and four of four other mammalian ACE2s; Leu79Ile had a effect for human and three of three mammalian ACE2s. These residues are highly represented, 83% for Gln42 and 56% for Leu79, among mammalian ACE2s. The above findings can be important in protecting humans and animals from existing and future SARS-CoV-2 variants.

Neural networks to learn protein sequence-function relationships from deep mutational scanning data.

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.

Discovery of human ACE2 variants with altered recognition by the SARS-CoV-2 spike protein.

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.

Discovery of human ACE2 variants with altered recognition by the SARS-CoV-2 spike protein.

Understanding how human ACE2 genetic variants differ in their recognition by SARS-CoV-2 can have a major impact in leveraging 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 can have a strong influence upon the ACE2-spike interaction. These findings illuminate new links between ACE2 sequence and spike recognition, and will find wide-ranging utility in SARS-CoV-2 fundamental research, epidemiological analyses, and clinical trial design.

De Novo Isolation & Affinity Maturation of yeast-displayed Virion-binding human fibronectin domains by flow cytometric screening against Virions.

BACKGROUND: The promise of biopharmaceuticals comprising one or more binding domains motivates the development of novel methods for de novo isolation and affinity maturation of virion-binding domains. Identifying avenues for overcoming the challenges associated with using virions as screening reagents is paramount given the difficulties associated with obtaining high-purity virus-associated proteins that retain the conformation exhibited on the virion surface. RESULTS: Fluorescence activated cell sorting (FACS) of 1.5 × 107 clones taken from a naïve yeast surface-displayed human fibronectin domain (Fn3) against whole virions yielded two unique binders to Zika virions. Construction and FACS of site-directed binding loop mutant libraries based on one of these binders yielded multiple progeny clones with enhanced Zika-binding affinities. These affinity-matured clones bound Zika virions with low double- or single-digit nanomolar affinity in ELISA assays, and expressed well as soluble proteins in E. coli shake flask culture, with post-purification yields exceeding 10 mg/L. CONCLUSIONS: FACS of a yeast-displayed binding domain library is an efficient method for de novo isolation of virion-binding domains. Affinities of isolated virion-binding clones are readily enhanced via FACS screening of mutant progeny libraries. Given that most binding domains are compatible with yeast display, the approach taken in this work may be broadly utilized for generating virion-binding domains against many different viruses for use in passive immunotherapy and the prevention of viral infection.

Shotgun Proteomic Profiling of Bloodborne Nanoscale Extracellular Vesicles.

Analyses of bloodborne nanoscale extracellular vesicles (nsEVs) have shown tremendous promise in enabling the development of noninvasive blood-based clinical diagnostic tests, predicting and monitoring the efficacy of treatment programs, and identifying new drug targets in the context of health conditions such as cancer and Alzheimer's disease. In this chapter we present a protocol for generating global nsEV proteomic profiles that can further the utility of nsEV analysis for the above biomedical applications by enlightening us of differences in protein abundance across normal and disease state nsEVs. This protocol features the use of magnetic particle-based immunoprecipitation to enrich highly purified populations of nsEVs directly from plasma or serum samples. The constituent proteins of these vesicles are subsequently characterized using a comparative shotgun proteomics approach that entails bottom-up, tandem mass spectrometric analysis of peptides generated by proteolytic digestion of nsEV-derived proteins. The methods described here are compatible with parallel processing of dozens of plasma or serum samples and can be valuable tools in enabling nsEV biomarker discoveries that have high translational relevance in the development of both novel therapeutics and blood sample diagnostic assays.

Magnetic Particle-Based Immunoprecipitation of Nanoscale Extracellular Vesicles from Biofluids.

Analysis of nanoscale extracellular vesicles (nsEVs) present in blood, cell culture media, and other biofluids has shown tremendous promise in enabling the development of noninvasive blood-based clinical diagnostic tests, predicting and monitoring the efficacy of treatment programs, and providing molecular level insights into pathology that can enlighten new drug targets in the contexts of health conditions such as cancer and Alzheimer's Disease (AD). In this chapter, we present methods for using magnetic particle-based immunoprecipitation to enrich highly purified populations of nsEVs directly from plasma, serum, and other biofluids. These methods enable downstream analysis of nsEV protein and nucleic acid constituents in the contexts of both global omics profiling and quantification of individual protein or nucleic acid species of interest. Additionally, these methods allow the researcher to either enrich total nsEV populations or enrich nsEVs derived from a particular tissue type from the overall nsEV population. The methods described here are compatible with parallel processing of dozens of biofluid samples and can be valuable tools for enabling nsEV analyses that have high translational relevance in the development of both novel therapeutics and noninvasive diagnostic assays.

Nanoscale Extracellular Vesicle Analysis in Alzheimer's Disease Diagnosis and Therapy.

Diagnostic assays that leverage bloodborne neuron-derived (neuronal) nanoscale extracellular vesicles (nsEVs) as "windows into the brain" can predict incidence of Alzheimer's Disease (AD) many years prior to onset. Beyond diagnostics, bloodborne neuronal nsEVs analysis may have substantial translational impact by revealing mechanisms of AD pathology; such knowledge could enlighten new drug targets and lead to new therapeutic approaches. The potential to establish three-dimensional nsEV analysis methods that characterize highly purified bloodborne nsEV populations in method of enrichment, cell type origin, and protein or RNA abundance dimensions could bring this promise to bear by yielding nsEV "omics" datasets that uncover new AD biomarkers and enable AD therapeutic development. In this review we provide a survey of both the current status of and new developments on the horizon in the field of neuronal nsEV analysis. This survey is supplemented by a discussion of the potential to translate such neuronal nsEV analyses to AD clinical diagnostic applications and drug development.

Engineering pH responsive fibronectin domains for biomedical applications.

BACKGROUND: Engineered antibodies with pH responsive cell surface target antigen-binding affinities that decrease at the acidic pH (5.5-5.8) within the endosomes have been found to have reduced susceptibility to degradation within the lysosomes and increased serum half-life. Such pH responsive recombinant antibodies have been developed for the treatment of cancer and cardiovascular disease. Engineered tenth type III human fibronectin (Fn3) domains are emerging as a class of target antigen-binding biopharmaceuticals that could complement or be superior to recombinant antibodies in a number of biomedical contexts. As such, there is strong motivation for demonstrating the feasibility of engineering Fn3s with pH responsive antigen binding behavior that could lead to improved Fn3 pharmacokinetics. RESULTS: A yeast surface-displayed Fn3 histidine (His) mutant library screening approach yielded epidermal growth factor receptor (EGFR)-binding Fn3 domains with EGFR binding affinities that markedly decrease at endosomal pH; the first reported case of engineering Fn3s with pH responsive antigen binding. Yeast surface-displayed His mutant Fn3s, which contain either one or two His mutations, have equilibrium binding dissociation constants (KDs) that increase up to four-fold relative to wild type when pH is decreased from 7.4 to 5.5. Assays in which Fn3-displaying yeast were incubated with soluble EGFR after ligand-free incubation in respective neutral and acidic buffers showed that His mutant Fn3 pH responsiveness is due to reversible changes in Fn3 conformation and/or EGFR binding interface properties rather than irreversible unfolding. CONCLUSIONS: We have established a generalizable method for efficiently constructing and screening Fn3 His mutant libraries that could enable both our laboratory and others to develop pH responsive Fn3s for use in a wide range of biomedical applications.

Cytokine refacing effect reduces granulocyte macrophage colony-stimulating factor susceptibility to antibody neutralization.

Crohn's Disease (CD) afflicts over half a million Americans with an annual economic impact exceeding $10 billion. Granulocyte macrophage colony-stimulating factor (GM-CSF) can increase patient immune responses against intestinal microbes that promote CD and has been effective for some patients in clinical trials. We have made important progress toward developing GM-CSF variants that could be more effective CD therapeutics by virtue of being less prone to neutralization by the endogenous GM-CSF autoantibodies that are highly expressed in CD patients. Yeast display engineering revealed mutations that increase GM-CSF variant binding affinity by up to ∼3-fold toward both GM-CSF receptor alpha and beta subunits in surface plasmon resonance experiments. Increased binding affinity did not reduce GM-CSF half-maximum effective concentration (EC50) values in conventional in vitro human leukocyte proliferation assays. Affinity-enhancing mutations did, however, promote a 'refacing effect' that imparted all five evaluated GM-CSF variants with increased in vitro bioactivity in the presence of GM-CSF-neutralizing polyclonal antisera. The most improved variant, H15L/R23L, was 6-fold more active than wild-type GM-CSF. Incorporation of additional known affinity-increasing mutations could augment the refacing effect and concomitant bioactivity improvements described here.

pH responsive granulocyte colony-stimulating factor variants with implications for treating Alzheimer's disease and other central nervous system disorders.

Systemic injection of granulocyte colony-stimulating factor (G-CSF) has yielded encouraging results in treating Alzheimer's Disease (AD) and other central nervous system (CNS) disorders. Making G-CSF a viable AD therapeutic will, however, require increasing G-CSF's ability to stimulate neurons within the brain. This objective could be realized by increasing transcytosis of G-CSF across the blood brain barrier (BBB). An established correlation between G-CSF receptor (G-CSFR) binding pH responsiveness and increased recycling of G-CSF to the cell exterior after endocytosis motivated development of G-CSF variants with highly pH responsive G-CSFR binding affinities. These variants will be used in future validation of our hypothesis that increased BBB transcytosis can enhance G-CSF therapeutic efficacy. Flow cytometric screening of a yeast-displayed library in which G-CSF/G-CSFR interface residues were mutated to histidine yielded a G-CSF triple His mutant (L109H/D110H/Q120H) with highly pH responsive binding affinity. This variant's KD, measured by surface plasmon resonance (SPR), increases ∼20-fold as pH decreases from 7.4 to below histidine's pKa of ∼6.0; an increase 2-fold greater than for previously reported G-CSF His mutants. Cell-free protein synthesis (CFPS) enabled expression and purification of soluble, bioactive G-CSF triple His variant protein, an outcome inaccessible via Escherichia coli inclusion body refolding. This purification and bioactivity validation will enable future identification of correlations between pH responsiveness and transcytosis in BBB cell culture model and animal experiments. Furthermore, the library screening and CFPS methods employed here could be applied to developing other pH responsive hematopoietic or neurotrophic factors for treating CNS disorders.

Engineering superactive granulocyte macrophage colony-stimulating factor transferrin fusion proteins as orally-delivered candidate agents for treating neurodegenerative disease.

Intravenously injected granulocyte macrophage colony-stimulating factor (GM-CSF) has shown efficacy in Alzheimer's Disease (AD) and Parkinson's Disease (PD) animal studies and is undergoing clinical evaluation. The likely need for dosing of GM-CSF to patients over months or years motivates pursuit of avenues for delivering GM-CSF to circulation via oral administration. Flow cytometric screening of 37 yeast-displayed GM-CSF saturation mutant libraries revealed residues P12, H15, R23, R24, and K72 as key determinants of GM-CSF's CD116 and CD131 GM-CSF receptor (GM-CSFR) subunit binding affinity. Screening combinatorial GM-CSF libraries mutated at positions P12, H15, and R23 yielded variants with increased affinities toward both CD116 and CD131. Genetic fusion of GM-CSF to human transferrin (Trf), a strategy that enables oral delivery of other biopharmaceuticals in animals, yielded bioactive wild type and variant cytokines upon secretion from cultured Human Embryonic Kidney cells. Surface plasmon resonance (SPR) measurements showed that all evaluated variants possess decreases in CD116 and CD131 binding KD values of up to 2.5-fold relative to wild type. Improved affinity led to increased in vitro bioactivity; the most bioactive variant, P12D/H15L/R23L, had a leukocyte proliferation assay EC50 value 3.5-fold lower than the wild type GM-CSF/Trf fusion. These outcomes are important first steps toward our goal of developing GM-CSF/Trf fusions as orally available AD and PD therapeutics.

Efficient sampling of SCHEMA chimera families to identify useful sequence elements.

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.

Efficient screening of fungal cellobiohydrolase class I enzymes for thermostabilizing sequence blocks by SCHEMA structure-guided recombination.

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.

SCHEMA recombination of a fungal cellulase uncovers a single mutation that contributes markedly to stability.

A quantitative linear model accurately (R(2) = 0.88) describes the thermostabilities of 54 characterized members of a family of fungal cellobiohydrolase class II (CBH II) cellulase chimeras made by SCHEMA recombination of three fungal enzymes, demonstrating that the contributions of SCHEMA sequence blocks to stability are predominantly additive. Thirty-one of 31 predicted thermostable CBH II chimeras have thermal inactivation temperatures higher than the most thermostable parent CBH II, from Humicola insolens, and the model predicts that hundreds more CBH II chimeras share this superior thermostability. Eight of eight thermostable chimeras assayed hydrolyze the solid cellulosic substrate Avicel at temperatures at least 5 degrees C above the most stable parent, and seven of these showed superior activity in 16-h Avicel hydrolysis assays. The sequence-stability model identified a single block of sequence that adds 8.5 degrees C to chimera thermostability. Mutating individual residues in this block identified the C313S substitution as responsible for the entire thermostabilizing effect. Introducing this mutation into the two recombination parent CBH IIs not featuring it (Hypocrea jecorina and H. insolens) decreased inactivation, increased maximum Avicel hydrolysis temperature, and improved long time hydrolysis performance. This mutation also stabilized and improved Avicel hydrolysis by Phanerochaete chrysosporium CBH II, which is only 55-56% identical to recombination parent CBH IIs. Furthermore, the C313S mutation increased total H. jecorina CBH II activity secreted by the Saccharomyces cerevisiae expression host more than 10-fold. Our results show that SCHEMA structure-guided recombination enables quantitative prediction of cellulase chimera thermostability and efficient identification of stabilizing mutations.

A family of thermostable fungal cellulases created by structure-guided recombination.

SCHEMA structure-guided recombination of 3 fungal class II cellobiohydrolases (CBH II cellulases) has yielded a collection of highly thermostable CBH II chimeras. Twenty-three of 48 genes sampled from the 6,561 possible chimeric sequences were secreted by the Saccharomyces cerevisiae heterologous host in catalytically active form. Five of these chimeras have half-lives of thermal inactivation at 63 degrees C that are greater than the most stable parent, CBH II enzyme from the thermophilic fungus Humicola insolens, which suggests that this chimera collection contains hundreds of highly stable cellulases. Twenty-five new sequences were designed based on mathematical modeling of the thermostabilities for the first set of chimeras. Ten of these sequences were expressed in active form; all 10 retained more activity than H. insolens CBH II after incubation at 63 degrees C. The total of 15 validated thermostable CBH II enzymes have high sequence diversity, differing from their closest natural homologs at up to 63 amino acid positions. Selected purified thermostable chimeras hydrolyzed phosphoric acid swollen cellulose at temperatures 7 to 15 degrees C higher than the parent enzymes. These chimeras also hydrolyzed as much or more cellulose than the parent CBH II enzymes in long-time cellulose hydrolysis assays and had pH/activity profiles as broad, or broader than, the parent enzymes. Generating this group of diverse, thermostable fungal CBH II chimeras is the first step in building an inventory of stable cellulases from which optimized enzyme mixtures for biomass conversion can be formulated.