Author Correction: A transgenic approach for controlling Lygus in cotton.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Quantifying the nativeness of antibody sequences using long short-term memory networks.

Antibodies often undergo substantial engineering en route to the generation of a therapeutic candidate with good developability properties. Characterization of antibody libraries has shown that retaining native-like sequence improves the overall quality of the library. Motivated by recent advances in deep learning, we developed a bi-directional long short-term memory (LSTM) network model to make use of the large amount of available antibody sequence information, and use this model to quantify the nativeness of antibody sequences. The model scores sequences for their similarity to naturally occurring antibodies, which can be used as a consideration during design and engineering of libraries. We demonstrate the performance of this approach by training a model on human antibody sequences and show that our method outperforms other approaches at distinguishing human antibodies from those of other species. We show the applicability of this method for the evaluation of synthesized antibody libraries and humanization of mouse antibodies.

A Proliferation Inducing Ligand (APRIL) targeted antibody is a safe and effective treatment of murine IgA nephropathy.

IgA nephropathy (IgAN) is the most prevalent primary chronic glomerular disease for which no safe disease-specific therapies currently exist. IgAN is an autoimmune disease involving the production of autoantigenic, aberrantly O-glycosylated IgA1 and ensuing deposition of nephritogenic immune complexes in the kidney. A Proliferation Inducing Ligand (APRIL) has emerged as a key B-cell-modulating factor in this pathogenesis. Using a mouse anti-APRIL monoclonal antibody (4540), we confirm both the pathogenic role of APRIL in IgAN and the therapeutic efficacy of antibody-directed neutralization of APRIL in the grouped mouse ddY disease model. Treatment with 4540 directly translated to a reduction in relevant pathogenic mechanisms including suppressed serum IgA levels, reduced circulating immune complexes, significantly lower kidney deposits of IgA, IgG and C3, and suppression of proteinuria compared to mice receiving vehicle or isotype control antibodies. Furthermore, we translated these findings to the pharmacological characterization of VIS649, a highly potent, humanized IgG2κ antibody targeting and neutralizing human APRIL through unique epitope engagement, leading to inhibition of APRIL-mediated B-cell activities. VIS649 treatment of non-human primates showed dose-dependent reduction of serum IgA levels of up to 70%. A reduction of IgA+, IgM+, and IgG+ B cells was noted in the gut-associated mucosa of VIS649-treated animals. Population-based modeling predicted a favorable therapeutic dosing profile for subcutaneous administration of VIS649 in the clinical setting. Thus, our data highlight the potential therapeutic benefit of VIS649 for the treatment of IgAN.

Development of enzymes for robust aryloxyphenoxypropionate and synthetic auxin herbicide tolerance traits in maize and soybean crops.

BACKGROUND: Effective management of weedy species in agricultural fields is essential for maintaining favorable growing conditions and crop yields. The introduction of genetically modified crops containing herbicide tolerance traits has been a successful additional tool available to farmers to better control weeds. However, weed resistance challenges present a need for additional herbicide tolerance trait options. RESULTS: To help meet this challenge, a new trait that provides tolerance to an aryloxyphenoxypropionate (FOP) herbicide and members of the synthetic auxin herbicide family, such as 2,4-dichlorophenoxyacetic acid (2,4-D), was developed. Development of this herbicide tolerance trait employed an enzyme engineered with robust and specific enzymatic activity for these two herbicide families. This engineering effort utilized a microbial-sourced dioxygenase scaffold to generate variants with improved enzymatic parameters. Additional optimization to enhance in-plant stability of the enzyme enabled an efficacious trait that can withstand the higher temperature conditions often found in field environments. CONCLUSION: Optimized herbicide tolerance enzyme variants with enhanced enzymatic and temperature stability parameters enabled robust herbicide tolerance for two herbicide families in transgenic maize and soybeans. This herbicide tolerance trait for FOP and synthetic auxin herbicides such as 2,4-D could be useful in weed management systems, providing additional tools for farmers to control weeds. © 2019 Society of Chemical Industry.

Structural prediction of antibody-APRIL complexes by computational docking constrained by antigen saturation mutagenesis library data.

IgA nephropathy (IgAN) is the most prevalent cause of primary glomerular disease worldwide, and the cytokine A PRoliferation-Inducing Ligand (APRIL) is emerging as a key player in IgAN pathogenesis and disease progression. For a panel of anti-human APRIL antibodies with known antagonistic activity, we sought to define their structural mode of engagement to understand molecular mechanisms of action and aid rational antibody engineering. Reliable computational prediction of antibody-antigen complexes remains challenging, and experimental methods such as X-ray co-crystallography and cryoEM have considerable technical, resource, and throughput barriers. To overcome these limitations, we implemented an integrated and accessible experimental-computational workflow to more accurately predict structures of antibody-APRIL complexes. Specifically, a yeast surface display library encoding site-saturation mutagenized surface positions of APRIL was screened against a panel of anti-APRIL antibodies to rapidly obtain a comprehensive biochemical profile of mutational impact on binding for each antibody. The experimentally derived mutational profile data were used as quantitative constraints in a computational docking workflow optimized for antibodies, resulting in robust structural models of antibody-antigen complexes. The model results were confirmed by solving the cocrystal structure of one antibody-APRIL complex, which revealed strong agreement with our model. The models were used to rationally select and engineer one antibody for cross-species APRIL binding, which quite often aids further testing in relevant animal models. Collectively, we demonstrate a rapid, integrated computational-experimental approach to robustly predict antibody-antigen structures information, which can aid rational antibody engineering and provide insights into molecular mechanisms of action.

Extending human IgG half-life using structure-guided design.

Engineering of antibodies for improved pharmacokinetics through enhanced binding to the neonatal Fc receptor (FcRn) has been demonstrated in transgenic mice, non-human primates and humans. Traditionally, such approaches have largely relied on random mutagenesis and display formats, which fail to address related critical attributes of the antibody, such as effector functions or biophysical stability. We have developed a structure- and network-based framework to interrogate the engagement of IgG with multiple Fc receptors (FcRn, C1q, TRIM21, FcγRI, FcγRIIa/b, FcγRIIIa) simultaneously. Using this framework, we identified features that govern Fc-FcRn interactions and identified multiple distinct pathways for enhancing FcRn binding in a pH-specific manner. Network analysis provided a novel lens to study the allosteric impact of half-life-enhancing Fc mutations on FcγR engagement, which occurs distal to the FcRn binding site. Applying these principles, we engineered a panel of unique Fc variants that enhance FcRn binding while maintaining robust biophysical properties and wild type-like binding to activating receptors. An antibody harboring representative Fc designs demonstrates a half-life improvement of > 9 fold in transgenic mice and > 3.5 fold in cynomolgus monkeys, and maintains robust effector functions such as antibody-dependent cell-mediated cytotoxicity and complement-dependent cytotoxicity.

A transgenic approach for controlling Lygus in cotton.

Lygus species of plant-feeding insects have emerged as economically important pests of cotton in the United States. These species are not controlled by commercial Bacillus thuringiensis (Bt) cotton varieties resulting in economic losses and increased application of insecticide. Previously, a Bt crystal protein (Cry51Aa2) was reported with insecticidal activity against Lygus spp. However, transgenic cotton plants expressing this protein did not exhibit effective protection from Lygus feeding damage. Here we employ various optimization strategies, informed in part by protein crystallography and modelling, to identify limited amino-acid substitutions in Cry51Aa2 that increase insecticidal activity towards Lygus spp. by >200-fold. Transgenic cotton expressing the variant protein, Cry51Aa2.834_16, reduce populations of Lygus spp. up to 30-fold in whole-plant caged field trials. One transgenic event, designated MON88702, has been selected for further development of cotton varieties that could potentially reduce or eliminate insecticide application for control of Lygus and the associated environmental impacts.

Safety and Upper Respiratory Pharmacokinetics of the Hemagglutinin Stalk-Binding Antibody VIS410 Support Treatment and Prophylaxis Based on Population Modeling of Seasonal Influenza A Outbreaks.

BACKGROUND: Seasonal influenza is a major public health concern in vulnerable populations. Here we investigated the safety, tolerability, and pharmacokinetics of a broadly neutralizing monoclonal antibody (VIS410) against Influenza A in a Phase 1 clinical trial. Based on these results and preclinical data, we implemented a mathematical modeling approach to investigate whether VIS410 could be used prophylactically to lessen the burden of a seasonal influenza epidemic and to protect at-risk groups from associated complications. METHODS: Using a single-ascending dose study (n = 41) at dose levels from 2 mg/kg-50 mg/kg we evaluated the safety as well as the serum and upper respiratory pharmacokinetics of a broadly-neutralizing antibody (VIS410) against influenza A (ClinicalTrials.gov identifier NCT02045472). Our primary endpoints were safety and tolerability of VIS410 compared to placebo. We developed an epidemic microsimulation model testing the ability of VIS410 to mitigate attack rates and severe disease in at risk-populations. FINDINGS: VIS410 was found to be generally safe and well-tolerated at all dose levels, from 2-50 mg/kg. Overall, 27 of 41 subjects (65.9%) reported a total of 67 treatment emergent adverse events (TEAEs). TEAEs were reported by 20 of 30 subjects (66.7%) who received VIS410 and by 7 of 11 subjects (63.6%) who received placebo. 14 of 16 TEAEs related to study drug were considered mild (Grade 1) and 2 were moderate (Grade 2). Two subjects (1 subject who received 30 mg/kg VIS410 and 1 subject who received placebo) experienced serious AEs (Grade 3 or 4 TEAEs) that were not related to study drug. VIS410 exposure was approximately dose-proportional with a mean half-life of 12.9 days. Mean VIS410 Cmax levels in the upper respiratory tract were 20.0 and 25.3 µg/ml at the 30 mg/kg and 50 mg/kg doses, respectively, with corresponding serum Cmax levels of 980.5 and 1316 µg/mL. Using these pharmacokinetic data, a microsimulation model showed that median attack rate reductions ranged from 8.6% (interquartile range (IQR): 4.7%-11.0%) for 2% coverage to 22.6% (IQR: 12.7-30.0%) for 6% coverage. The overall benefits to the elderly, a vulnerable subgroup, are largest when VIS410 is distributed exclusively to elderly individuals, resulting in reductions in hospitalization rates between 11.4% (IQR: 8.2%-13.3%) for 2% coverage and 30.9% (IQR: 24.8%-35.1%) for 6% coverage among those more than 65 years of age. INTERPRETATION: VIS410 was generally safe and well tolerated and had good relative exposure in both serum and upper respiratory tract, supporting its use as either a single-dose therapeutic or prophylactic for influenza A. Including VIS410 prophylaxis among the public health interventions for seasonal influenza has the potential to lower attack rates and substantially reduce hospitalizations in individuals over the age of 65. FUNDING: Visterra, Inc.

Optimization of the in-silico-designed kemp eliminase KE70 by computational design and directed evolution.

Although de novo computational enzyme design has been shown to be feasible, the field is still in its infancy: the kinetic parameters of designed enzymes are still orders of magnitude lower than those of naturally occurring ones. Nonetheless, designed enzymes can be improved by directed evolution, as recently exemplified for the designed Kemp eliminase KE07. Random mutagenesis and screening resulted in variants with >200-fold higher catalytic efficiency and provided insights about features missing in the designed enzyme. Here we describe the optimization of KE70, another designed Kemp eliminase. Amino acid substitutions predicted to improve catalysis in design calculations involving extensive backbone sampling were individually tested. Those proven beneficial were combinatorially incorporated into the originally designed KE70 along with random mutations, and the resulting libraries were screened for improved eliminase activity. Nine rounds of mutation and selection resulted in >400-fold improvement in the catalytic efficiency of the original KE70 design, reflected in both higher k(cat) values and lower K(m) values, with the best variants exhibiting k(cat)/K(m) values of >5×10(4) s(-)(1) M(-1). The optimized KE70 variants were characterized structurally and biochemically, providing insights into the origins of the improvements in catalysis. Three primary contributions were identified: first, the reshaping of the active-site cavity to achieve tighter substrate binding; second, the fine-tuning of electrostatics around the catalytic His-Asp dyad; and, third, the stabilization of the active-site dyad in a conformation optimal for catalysis.

Kemp elimination catalysts by computational enzyme design.

The design of new enzymes for reactions not catalysed by naturally occurring biocatalysts is a challenge for protein engineering and is a critical test of our understanding of enzyme catalysis. Here we describe the computational design of eight enzymes that use two different catalytic motifs to catalyse the Kemp elimination-a model reaction for proton transfer from carbon-with measured rate enhancements of up to 10(5) and multiple turnovers. Mutational analysis confirms that catalysis depends on the computationally designed active sites, and a high-resolution crystal structure suggests that the designs have close to atomic accuracy. Application of in vitro evolution to enhance the computational designs produced a >200-fold increase in k(cat)/K(m) (k(cat)/K(m) of 2,600 M(-1)s(-1) and k(cat)/k(uncat) of >10(6)). These results demonstrate the power of combining computational protein design with directed evolution for creating new enzymes, and we anticipate the creation of a wide range of useful new catalysts in the future.

Structure prediction for CASP7 targets using extensive all-atom refinement with Rosetta@home.

We describe predictions made using the Rosetta structure prediction methodology for both template-based modeling and free modeling categories in the Seventh Critical Assessment of Techniques for Protein Structure Prediction. For the first time, aggressive sampling and all-atom refinement could be carried out for the majority of targets, an advance enabled by the Rosetta@home distributed computing network. Template-based modeling predictions using an iterative refinement algorithm improved over the best existing templates for the majority of proteins with less than 200 residues. Free modeling methods gave near-atomic accuracy predictions for several targets under 100 residues from all secondary structure classes. These results indicate that refinement with an all-atom energy function, although computationally expensive, is a powerful method for obtaining accurate structure predictions.

The role of quantum mechanics in structure-based drug design.

Herein we will focus on the use of quantum mechanics (QM) in drug design (DD) to solve disparate problems from scoring protein-ligand poses to building QM QSAR models. Through the variational principle of QM we know that we can obtain a more accurate representation of molecular systems than classical models, and while this is not a matter of debate, it still has not been shown that the expense of QM approaches is offset by improved accuracy in DD applications. Objectively validating the improved applicability and performance of QM over classical-based models in DD will be the focus of research in the coming years along with research on the conformational sampling problem as it relates to protein-ligand complexes.

Assessment of Semiempirical Quantum Mechanical Methods for the Evaluation of Protein Structures.

The ability to discriminate native structures from computer-generated misfolded ones is key to predicting the three-dimensional structure of a protein from its amino acid sequence. Here we describe an assessment of semiempirical methods for discriminating native protein structures from decoy models. The discrimination of decoys entails an analysis of a large number of protein structures, and provides a large-scale validation of quantum mechanical methods and their ability to accurately model proteins. We combine our analysis of semiempirical methods with a comparison of an AMBER force field to discriminate decoys in conjunction with a continuum solvent model. Protein decoys provide a rigorous and reliable benchmark for the evaluation of scoring functions, not only in their ability to accurately identify native structures but also to be computationally tractable to sample a large set of non-native models.

Prediction of structures of multidomain proteins from structures of the individual domains.

We describe the development of a method for assembling structures of multidomain proteins from structures of isolated domains. The method consists of an initial low-resolution search in which the conformational space of the domain linker is explored using the Rosetta de novo structure prediction method, followed by a high-resolution search in which all atoms are treated explicitly and backbone and side chain degrees of freedom are simultaneously optimized. The method recapitulates, often with very high accuracy, the structures of existing multidomain proteins.

Haptic applications for molecular structure manipulation.

We describe the application of haptic technology to enhance the information available in chemical systems, specifically related to computational drug design. These methods are designed to build upon the visual information presented by molecular viewers and add the sensation of touch, or force feedback. The addition of sensory input can aid in the analysis of molecular structures and the understanding of intermolecular interactions by delivering chemically relevant forces to the end user.

New algorithms and an in silico benchmark for computational enzyme design.

The creation of novel enzymes capable of catalyzing any desired chemical reaction is a grand challenge for computational protein design. Here we describe two new algorithms for enzyme design that employ hashing techniques to allow searching through large numbers of protein scaffolds for optimal catalytic site placement. We also describe an in silico benchmark, based on the recapitulation of the active sites of native enzymes, that allows rapid evaluation and testing of enzyme design methodologies. In the benchmark test, which consists of designing sites for each of 10 different chemical reactions in backbone scaffolds derived from 10 enzymes catalyzing the reactions, the new methods succeed in identifying the native site in the native scaffold and ranking it within the top five designs for six of the 10 reactions. The new methods can be directly applied to the design of new enzymes, and the benchmark provides a powerful in silico test for guiding improvements in computational enzyme design.

Development of a Parametrized Force Field To Reproduce Semiempirical Geometries.

Here we describe the development of a classical force field parameter set to reproduce the geometry of proteins minimized at the semiempirical quantum mechanical level. The overall goal of the development of this new force field is to provide an inexpensive, yet reliable, method to arrive at geometries that are more consistent with a semiempirical treatment of protein structures. Since the minimization of a large number of protein structures at the semiempirical level can become cost-prohibitive, a "preminimization" with an appropriately parametrized classical treatment could potentially lead to more computationally efficient methods for studying protein structures through semiempirical means. Here we demonstrate that this force field allows for more rapid and stable geometry optimizations at the semiempirical level and can aid in the adoption of quantum mechanical calculations for large biological systems.