Fine Epitope Mapping of the CD19 Extracellular Domain Promotes Design.

The B-cell surface protein CD19 is present throughout the cell life cycle and is uniformly expressed in leukemias, making it a target for chimeric antigen receptor engineered immune cell therapy. Identifying the sequence dependence of the binding of CD19 to antibodies empowers fundamental study and more tailored development of CD19-targeted therapeutics. To identify the antibody-binding epitopes on CD19, we screened a comprehensive single-site saturation mutation library of the human CD19 extracellular domain to identify mutations detrimental to binding FMC63-the dominant CD19 antibody used in chimeric antigen receptor development-as well as 4G7-2E3 and 3B10, which have been used in various types of CD19 research and development. All three antibodies had partially overlapping, yet distinct, epitopes near the published epitope of antibody B43. The FMC63 conformational epitope spans spatially adjacent, but genetically distant, loops in exons 3 and 4. The 3B10 epitope is a linear peptide sequence that binds CD19 with 440 pM affinity. Along with their primary goal of epitope mapping, the mutational tolerance data also empowered additional CD19 variant design and analysis. A designed CD19 variant with all N-linked glycosylation sites removed successfully bound antibody in the yeast display context, which provides a lead for aglycosylated applications. Screening for thermally stable variants identified mutations to guide further CD19 stabilization for fusion protein applications and revealed evolutionary affinity-stability trade-offs. These fundamental insights into CD19 sequence-function relationships enhance our understanding of antibody-mediated CD19-targeted therapeutics.

Retargeting CD19 Chimeric Antigen Receptor T Cells via Engineered CD19-Fusion Proteins.

CD19-targeted chimeric antigen receptor (CAR) T-cells (CAR19s) show remarkable efficacy in the treatment of relapsed/refractory acute lymphocytic leukemia and Non-Hodgkin's lymphoma. However, the use of CAR T-cell therapy against CD19-negative hematological cancers and solid tumors has been challenging. We propose CD19-fusion proteins (CD19-FPs) to leverage the benefits of CAR19s while retargeting this validated cellular therapy to alternative tumor antigens. We demonstrate the ability of a fusion of CD19 extracellular domain (ECD) and a human epidermal growth factor receptor 2 (HER2) single-chain antibody fragment to retarget CAR19s to kill HER2+ CD19- tumor cells. To enhance the modularity of this technology, we engineered a more robust CD19 ECD via deep mutational scanning with yeast display and flow cytometric selections for improved protease resistance and anti-CD19 antibody binding. These enhanced CD19 ECDs significantly increase, and in some cases recover, fusion protein expression while maintaining target antigen affinity. Importantly, CD19-FPs retarget CAR19s to kill tumor cells expressing multiple distinct antigens, including HER2, CD20, EGFR, BCMA, and Clec12A as N- or C-terminal fusions and linked to both antibody fragments and fibronectin ligands. This study provides fundamental insights into CD19 sequence-function relationships and defines a flexible and modular platform to retarget CAR19s to any tumor antigen.

An Automated Data-Driven Pipeline for Improving Heterologous Enzyme Expression.

Enzymes are the ultimate entities responsible for chemical transformations in natural and engineered biosynthetic pathways. However, many natural enzymes suffer from suboptimal functional expression due to poor intrinsic protein stability. Further, stability enhancing mutations often come at the cost of impaired function. Here we demonstrate an automated protein engineering strategy for stabilizing enzymes while retaining catalytic function using deep mutational scanning coupled to multiple-filter based screening and combinatorial mutagenesis. We validated this strategy by improving the functional expression of a Type III polyketide synthase from the Atropa belladonna biosynthetic pathway for tropane alkaloids. The best variant had a total of 8 mutations with over 25-fold improved activity over wild-type in E. coli cell lysates, an improved melting temperature of 11.5 ± 0.6 °C, and only minimal reduction in catalytic efficiency. We show that the multiple-filter approach maintains acceptable sensitivity with homology modeling structures up to 4 Å RMS. Our results highlight an automated protein engineering tool for improving the stability and solubility of difficult to express enzymes, which has impact for biotechnological applications.

Improved mutant function prediction via PACT: Protein Analysis and Classifier Toolkit.

MOTIVATION: Deep mutational scanning experiments have enabled the measurement of the sequence-function relationship for thousands of mutations in a single experiment. The Protein Analysis and Classifier Toolkit (PACT) is a Python software package that marries the fitness metric of a given mutation within these experiments to sequence and structural features enabling downstream analyses. PACT enables the easy development of user sharable protocols for custom deep mutational scanning experiments as all code is modular and reusable between protocols. Protocols for mutational libraries with single or multiple mutations are included. To exemplify its utility, PACT assessed two deep mutational scanning datasets that measured the tradeoff of enzyme activity and enzyme stability. RESULTS: PACT efficiently evaluated classifiers that predict protein mutant function tested on deep mutational scanning screens. We found that the classifiers with the lowest false positive and highest true positive rate assesses sequence homology, contact number and if mutation involves proline. AVAILABILITY AND IMPLEMENTATION: PACT and the processed datasets are distributed freely under the terms of the GPL-3 license. The source code is available at GitHub (https://github.com/JKlesmith/PACT). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Trade-offs between enzyme fitness and solubility illuminated by deep mutational scanning.

Proteins are marginally stable, and an understanding of the sequence determinants for improved protein solubility is highly desired. For enzymes, it is well known that many mutations that increase protein solubility decrease catalytic activity. These competing effects frustrate efforts to design and engineer stable, active enzymes without laborious high-throughput activity screens. To address the trade-off between enzyme solubility and activity, we performed deep mutational scanning using two different screens/selections that purport to gauge protein solubility for two full-length enzymes. We assayed a TEM-1 beta-lactamase variant and levoglucosan kinase (LGK) using yeast surface display (YSD) screening and a twin-arginine translocation pathway selection. We then compared these scans with published experimental fitness landscapes. Results from the YSD screen could explain 37% of the variance in the fitness landscapes for one enzyme. Five percent to 10% of all single missense mutations improve solubility, matching theoretical predictions of global protein stability. For a given solubility-enhancing mutation, the probability that it would retain wild-type fitness was correlated with evolutionary conservation and distance to active site, and anticorrelated with contact number. Hybrid classification models were developed that could predict solubility-enhancing mutations that maintain wild-type fitness with an accuracy of 90%. The downside of using such classification models is the removal of rare mutations that improve both fitness and solubility. To reveal the biophysical basis of enhanced protein solubility and function, we determined the crystallographic structure of one such LGK mutant. Beyond fundamental insights into trade-offs between stability and activity, these results have potential biotechnological applications.

Fine Epitope Mapping of Two Antibodies Neutralizing the Bordetella Adenylate Cyclase Toxin.

Adenylate cyclase toxin (ACT) is an important Bordetella pertussis virulence factor that is not included in current acellular pertussis vaccines. We previously demonstrated that immunization with the repeat-in-toxin (RTX) domain of ACT elicits neutralizing antibodies in mice and discovered the first two antibodies to neutralize ACT activities by occluding the receptor-binding site. Here, we fully characterize these antibodies and their epitopes. Both antibodies bind ACT with low nanomolar affinity and cross-react with ACT homologues produced by B. parapertussis and B. bronchiseptica. Antibody M1H5 binds B. pertussis RTX751 ∼100-fold tighter than RTX751 from the other two species, while antibody M2B10 has similar affinity for all three variants. To initially map the antibody epitopes, we generated a series of ACT chimeras and truncation variants, which implicated the repeat blocks II-III. To identify individual epitope residues, we displayed randomly mutated RTX751 libraries on yeast and isolated clones with decreased antibody binding by flow cytometry. Next-generation sequencing identified candidate epitope residues on the basis of enrichment of clones with mutations at specific positions. These epitopes form two adjacent surface patches on a predicted structural model of the RTX751 domain, one for each antibody. Notably, the cellular receptor also binds within blocks II-III and shares at least one residue with the M1H5 epitope. The RTX751 model supports the notion that the antibody and receptor epitopes overlap. These data provide insight into mechanisms of ACT neutralization and guidance for engineering more stable RTX variants that may be more appropriate vaccine antigens.

Plasmid-based one-pot saturation mutagenesis.

Deep mutational scanning is a foundational tool for addressing the functional consequences of large numbers of mutants, but a more efficient and accessible method for construction of user-defined mutagenesis libraries is needed. Here we present nicking mutagenesis, a robust, single-day, one-pot saturation mutagenesis method performed on routinely prepped plasmid dsDNA. The method can be used to produce comprehensive or single- or multi-site saturation mutagenesis libraries.

High-throughput evaluation of synthetic metabolic pathways.

A central challenge in the field of metabolic engineering is the efficient identification of a metabolic pathway genotype that maximizes specific productivity over a robust range of process conditions. Here we review current methods for optimizing specific productivity of metabolic pathways in living cells. New tools for library generation, computational analysis of pathway sequence-flux space, and high-throughput screening and selection techniques are discussed.

Producing glucose 6-phosphate from cellulosic biomass: structural insights into levoglucosan bioconversion.

The most abundant carbohydrate product of cellulosic biomass pyrolysis is the anhydrosugar levoglucosan (1,6-anhydro-ß-d-glucopyranose), which can be converted to glucose 6-phosphate by levoglucosan kinase (LGK). In addition to the canonical kinase phosphotransfer reaction, the conversion requires cleavage of the 1,6-anhydro ring to allow ATP-dependent phosphorylation of the sugar O6 atom. Using x-ray crystallography, we show that LGK binds two magnesium ions in the active site that are additionally coordinated with the nucleotide and water molecules to result in ideal octahedral coordination. To further verify the metal binding sites, we co-crystallized LGK in the presence of manganese instead of magnesium and solved the structure de novo using the anomalous signal from four manganese atoms in the dimeric structure. The first metal is required for catalysis, whereas our work suggests that the second is either required or significantly promotes the catalytic rate. Although the enzyme binds its sugar substrate in a similar orientation to the structurally related 1,6-anhydro-N-acetylmuramic acid kinase (AnmK), it forms markedly fewer bonding interactions with the substrate. In this orientation, the sugar is in an optimal position to couple phosphorylation with ring cleavage. We also observed a second alternate binding orientation for levoglucosan, and in these structures, ADP was found to bind with lower affinity. These combined observations provide an explanation for the high Km of LGK for levoglucosan. Greater knowledge of the factors that contribute to the catalytic efficiency of LGK can be used to improve applications of this enzyme for levoglucosan-derived biofuel production.

Comprehensive Sequence-Flux Mapping of a Levoglucosan Utilization Pathway in E. coli.

Synthetic metabolic pathways often suffer from low specific productivity, and new methods that quickly assess pathway functionality for many thousands of variants are urgently needed. Here we present an approach that enables the rapid and parallel determination of sequence effects on flux for complete gene-encoding sequences. We show that this method can be used to determine the effects of over 8000 single point mutants of a pyrolysis oil catabolic pathway implanted in Escherichia coli. Experimental sequence-function data sets predicted whether fitness-enhancing mutations to the enzyme levoglucosan kinase resulted from enhanced catalytic efficiency or enzyme stability. A structure of one design incorporating 38 mutations elucidated the structural basis of high fitness mutations. One design incorporating 15 beneficial mutations supported a 15-fold improvement in growth rate and greater than 24-fold improvement in enzyme activity relative to the starting pathway. This technique can be extended to improve a wide variety of designed pathways.

High-resolution sequence-function mapping of full-length proteins.

Comprehensive sequence-function mapping involves detailing the fitness contribution of every possible single mutation to a gene by comparing the abundance of each library variant before and after selection for the phenotype of interest. Deep sequencing of library DNA allows frequency reconstruction for tens of thousands of variants in a single experiment, yet short read lengths of current sequencers makes it challenging to probe genes encoding full-length proteins. Here we extend the scope of sequence-function maps to entire protein sequences with a modular, universal sequence tiling method. We demonstrate the approach with both growth-based selections and FACS screening, offer parameters and best practices that simplify design of experiments, and present analytical solutions to normalize data across independent selections. Using this protocol, sequence-function maps covering full sequences can be obtained in four to six weeks. Best practices introduced in this manuscript are fully compatible with, and complementary to, other recently published sequence-function mapping protocols.

The interrelationship between promoter strength, gene expression, and growth rate.

In exponentially growing bacteria, expression of heterologous protein impedes cellular growth rates. Quantitative understanding of the relationship between expression and growth rate will advance our ability to forward engineer bacteria, important for metabolic engineering and synthetic biology applications. Recently, a work described a scaling model based on optimal allocation of ribosomes for protein translation. This model quantitatively predicts a linear relationship between microbial growth rate and heterologous protein expression with no free parameters. With the aim of validating this model, we have rigorously quantified the fitness cost of gene expression by using a library of synthetic constitutive promoters to drive expression of two separate proteins (eGFP and amiE) in E. coli in different strains and growth media. In all cases, we demonstrate that the fitness cost is consistent with the previous findings. We expand upon the previous theory by introducing a simple promoter activity model to quantitatively predict how basal promoter strength relates to growth rate and protein expression. We then estimate the amount of protein expression needed to support high flux through a heterologous metabolic pathway and predict the sizable fitness cost associated with enzyme production. This work has broad implications across applied biological sciences because it allows for prediction of the interplay between promoter strength, protein expression, and the resulting cost to microbial growth rates.