LeGenD: determining N-glycoprofiles using an explainable AI-leveraged model with lectin profiling.

Glycosylation affects many vital functions of organisms. Therefore, its surveillance is critical from basic science to biotechnology, including biopharmaceutical development and clinical diagnostics. However, conventional glycan structure analysis faces challenges with throughput and cost. Lectins offer an alternative approach for analyzing glycans, but they only provide glycan epitopes and not full glycan structure information. To overcome these limitations, we developed LeGenD, a lectin and AI-based approach to predict N-glycan structures and determine their relative abundance in purified proteins based on lectin-binding patterns. We trained the LeGenD model using 309 glycoprofiles from 10 recombinant proteins, produced in 30 glycoengineered CHO cell lines. Our approach accurately reconstructed experimentally-measured N-glycoprofiles of bovine Fetuin B and IgG from human sera. Explanatory AI analysis with SHapley Additive exPlanations (SHAP) helped identify the critical lectins for glycoprofile predictions. Our LeGenD approach thus presents an alternative approach for N-glycan analysis.

Preparing glycomics data for robust statistical analysis with GlyCompareCT.

GlyCompareCT is a portable command-line tool to facilitate downstream glycomic data analyses, by addressing data inherent sparsity and non-independence. Inputting glycan abundances, users can run GlyCompareCT with one line of code to obtain the abundances of a minimal substructure set, named glycomotif, thereby quantifying hidden biosynthetic relationships between measured glycans. Optional parameters tuning and annotation are supported for personal preference. For complete details on the use and execution of this protocol, please refer to Bao et al. (2021).1.

Correction: StanDep: Capturing transcriptomic variability improves context-specific metabolic models.

[This corrects the article DOI: 10.1371/journal.pcbi.1007764.].

Systematic evaluation of parameters for genome-scale metabolic models of cultured mammalian cells.

Genome-scale metabolic models describe cellular metabolism with mechanistic detail. Given their high complexity, such models need to be parameterized correctly to yield accurate predictions and avoid overfitting. Effective parameterization has been well-studied for microbial models, but it remains unclear for higher eukaryotes, including mammalian cells. To address this, we enumerated model parameters that describe key features of cultured mammalian cells - including cellular composition, bioprocess performance metrics, mammalian-specific pathways, and biological assumptions behind model formulation approaches. We tested these parameters by building thousands of metabolic models and evaluating their ability to predict the growth rates of a panel of phenotypically diverse Chinese Hamster Ovary cell clones. We found the following considerations to be most critical for accurate parameterization: (1) cells limit metabolic activity to maintain homeostasis, (2) cell morphology and viability change dynamically during a growth curve, and (3) cellular biomass has a particular macromolecular composition. Depending on parameterization, models predicted different metabolic phenotypes, including contrasting mechanisms of nutrient utilization and energy generation, leading to varying accuracies of growth rate predictions. Notably, accurate parameter values broadly agreed with experimental measurements. These insights will guide future investigations of mammalian metabolism.

A genome-scale metabolic network model and machine learning predict amino acid concentrations in Chinese Hamster Ovary cell cultures.

The control of nutrient availability is critical to large-scale manufacturing of biotherapeutics. However, the quantification of proteinogenic amino acids is time-consuming and thus is difficult to implement for real-time in situ bioprocess control. Genome-scale metabolic models describe the metabolic conversion from media nutrients to proliferation and recombinant protein production, and therefore are a promising platform for in silico monitoring and prediction of amino acid concentrations. This potential has not been realized due to unresolved challenges: (1) the models assume an optimal and highly efficient metabolism, and therefore tend to underestimate amino acid consumption, and (2) the models assume a steady state, and therefore have a short forecast range. We address these challenges by integrating machine learning with the metabolic models. Through this we demonstrate accurate and time-course dependent prediction of individual amino acid concentration in culture medium throughout the production process. Thus, these models can be deployed to control nutrient feeding to avoid premature nutrient depletion or provide early predictions of failed bioreactor runs.

In situ detection of protein interactions for recombinant therapeutic enzymes.

Despite their therapeutic potential, many protein drugs remain inaccessible to patients since they are difficult to secrete. Each recombinant protein has unique physicochemical properties and requires different machinery for proper folding, assembly, and posttranslational modifications (PTMs). Here we aimed to identify the machinery supporting recombinant protein secretion by measuring the protein-protein interaction (PPI) networks of four different recombinant proteins (SERPINA1, SERPINC1, SERPING1, and SeAP) with various PTMs and structural motifs using the proximity-dependent biotin identification (BioID) method. We identified PPIs associated with specific features of the secreted proteins using a Bayesian statistical model and found proteins involved in protein folding, disulfide bond formation, and N-glycosylation were positively correlated with the corresponding features of the four model proteins. Among others, oxidative folding enzymes showed the strongest association with disulfide bond formation, supporting their critical roles in proper folding and maintaining the ER stability. Knockdown of disulfide-isomerase PDIA4, a measured interactor with significance for SERPINC1 but not SERPINA1, led to the decreased secretion of SERPINC1, which relies on its extensive disulfide bonds, compared to SERPINA1, which has no disulfide bonds. Proximity-dependent labeling successfully identified the transient interactions supporting synthesis of secreted recombinant proteins and refined our understanding of key molecular mechanisms of the secretory pathway during recombinant protein production.

StanDep: Capturing transcriptomic variability improves context-specific metabolic models.

Diverse algorithms can integrate transcriptomics with genome-scale metabolic models (GEMs) to build context-specific metabolic models. These algorithms require identification of a list of high confidence (core) reactions from transcriptomics, but parameters related to identification of core reactions, such as thresholding of expression profiles, can significantly change model content. Importantly, current thresholding approaches are burdened with setting singular arbitrary thresholds for all genes; thus, resulting in removal of enzymes needed in small amounts and even many housekeeping genes. Here, we describe StanDep, a novel heuristic method for using transcriptomics to identify core reactions prior to building context-specific metabolic models. StanDep clusters gene expression data based on their expression pattern across different contexts and determines thresholds for each cluster using data-dependent statistics, specifically standard deviation and mean. To demonstrate the use of StanDep, we built hundreds of models for the NCI-60 cancer cell lines. These models successfully increased the inclusion of housekeeping reactions, which are often lost in models built using standard thresholding approaches. Further, StanDep also provided a transcriptomic explanation for inclusion of lowly expressed reactions that were otherwise only supported by model extraction methods. Our study also provides novel insights into how cells may deal with context-specific and ubiquitous functions. StanDep, as a MATLAB toolbox, is available at https://github.com/LewisLabUCSD/StanDep.

The emerging impact of cell-free chemical biosynthesis.

Biomanufacturing has emerged as a promising alternative to chemocatalysis for green, renewable, complex synthesis of biofuels, medicines, and fine chemicals. Cell-free chemical biosynthesis offers additional advantages over in vivo production, enabling plug-and-play assembly of separately produced enzymes into an optimal cascade, versatile reaction conditions, and direct access to the reaction environment. In order for these advantages to be realized on the larger scale of industry, strategies are needed to reduce costs of biocatalyst generation, improve biocatalyst stability, and enable economically sustainable continuous cascade operation. Here we overview the advantages and remaining challenges of applying cell-free chemical biosynthesis for commodity production, and discuss recent advances in cascade engineering, enzyme immobilization, and enzyme encapsulation which constitute important steps towards addressing these challenges.

Rapid, portable detection of endocrine disrupting chemicals through ligand-nuclear hormone receptor interactions.

Endocrine disrupting chemicals (EDC) are structurally diverse compounds that can interact with nuclear hormone receptors, posing significant risk to human and ecological health. Unfortunately, many conventional biosensors have been too structure-specific, labor-intensive or laboratory-oriented to detect broad ranges of EDC effectively. Recently, several technological advances are providing more rapid, portable, and affordable detection of endocrine-disrupting activity through ligand-nuclear hormone receptor interactions. Here, we overview these recent advances applied to EDC biosensors - including cell lyophilization, cell immobilization, cell-free systems, smartphone-based signal detection, and improved competitive binding assays.

Efficient tRNA degradation and quantification in Escherichia coli cell extract using RNase-coated magnetic beads: A key step toward codon emancipation.

Emancipating sense codons toward a minimized genetic code is of significant interest to science and engineering. A key approach toward sense codon emancipation is the targeted in vitro removal of native tRNA. However, challenges remain such as the insufficient depletion of tRNA in lysate-based in vitro systems and the high cost of the purified components system (PURE). Here we used RNase-coated superparamagnetic beads to efficiently degrade E. coli endogenous tRNA. The presented method removes >99% of tRNA in cell lysates, while partially preserving cell-free protein synthesis activity. The resulting tRNA-depleted lysate is compatible with in vitro-transcribed synthetic tRNA for the production of peptides and proteins. Additionally, we directly measured residual tRNA using quantitative real-time PCR. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 33:1401-1407, 2017.

Rapid in vitro screening for the location-dependent effects of unnatural amino acids on protein expression and activity.

The incorporation of unnatural amino acids (uAA) can introduce novel functional groups into proteins site-specifically, with important applications in basic sciences and protein engineering. However, uAA incorporation can impact protein expression and functional activity depending on its location within the protein-a process that is not yet completely understood and difficult to predict. Therefore, practical applications often necessitate a time-consuming optimization of uAA location by individual gene cloning, expressions, purification, and evaluations for each location tested. To address this limitation, we introduce a streamlined and versatile in vitro system to rapidly express and screen uAA-containing proteins without cumbersome cell culturing or purification procedures. We utilized this technology to simultaneously screen 24 different t4-lysozyme mutants with different uAA incorporation sites in a matter of hours, compared to weeks-long workflow of conventional methods. Screening data offered a mechanistic explanation to some effects of uAA incorporation on expression and activity. Despite these insights, rational prediction of such effects remained challenging, further confirming the value of a rapid screening approach. Biotechnol. Bioeng. 2017;114: 2412-2417. © 2017 Wiley Periodicals, Inc.

Protein synthesis directly from PCR: progress and applications of cell-free protein synthesis with linear DNA.

A rapid, versatile method of protein expression and screening can greatly facilitate the future development of therapeutic biologics, proteomic drug targets and biocatalysts. An attractive candidate is cell-free protein synthesis (CFPS), a cell-lysate-based in vitro expression system, which can utilize linear DNA as expression templates, bypassing time-consuming cloning steps of plasmid-based methods. Traditionally, such linear DNA expression templates (LET) have been vulnerable to degradation by nucleases present in the cell lysate, leading to lower yields. This challenge has been significantly addressed in the recent past, propelling LET-based CFPS as a useful tool for studying, screening and engineering proteins in a high-throughput manner. Currently, LET-based CFPS has promise in fields such as functional proteomics, protein microarrays, and the optimization of complex biological systems.