N-methyl-D-aspartate (NMDA) receptors are ligand-gated ion channels which mediate the excitatory neurotransmission, and are associated with cancer and brain diseases, including epilepsy, schizophrenia, and Alzheimer’s disease. Thus, targeting NMDA receptors is a promising strategy for treating the diseases. Due to the instability and complexity of those large multi-oligomeric membrane protein complexes expressed in eukaryotic species, it remains challenging to yield properly assembled proteins. Here, I present the developed method to improve protein production, electrophysiological and cryo-EM studies of NMDA receptor complexes, and functional antibodies. Our work contributes to a better understanding of molecular mechanisms and strategies to target NMDA receptors.

Moderna is developing multiple mRNA vaccines and therapeutics. Since target protein expression from mRNA drugs is required for their activity, Moderna needs to produce lab-scale quantities of targeted proteins for various studies. Here, we expressed a single-pass transmembrane protein in multiple hosts and screened over 30 fusion constructs for solubility in Expi293, enabling production of the protein in soluble form with successful cleavage of the fusion partners/purification tags.

The endoplasmic reticulum (ER) is where approximately one-third of the proteome, including hormones, growth factors, and antibodies, is synthesized. The secretory export of these proteins risks leakage of the ER-resident proteome, which causes cell stress and interferes with biosynthesis. We have developed cost-effective strategies for enhanced prokaryotic and eukaryotic production of the WD40 domains of the COPI complex, crucial for retaining the ER-resident proteome during secretory trafficking. These strategies have broad applications since WD40 domains play key roles in the secretory trafficking of medically significant proteins, such as the SARS-CoV-2 spike protein produced from genetic vaccines.

Mutations in polycystin ion channel subunits cause Polycystic Kidney Disease (PKD), a genetic disorder that afflicts 1:1000 individuals worldwide with perfuse kidney cysts. We developed a mammalian protein expression and purification pipeline to produce recombinant polycystin ion channel variants. We applied this methodology in combination with CryoEM to evaluate the structural changes caused by missense PKD causing mutations in specific domains of the PKD2 polycystin homotetrameric channel assembly.

Pichia pastoris is a proven expression host that can be utilized for the production of difficult-to-express proteins. The GlycoFi platform humanized the Pichia glycosylation pathway which allows it to express glycoproteins with a human-like glycan profile. This platform gives the opportunity to select for the optimal glycan structure through genetic knockouts and over-expressions, while simultaneously tackling the production of many challenging-to-express proteins.

Cell-free expression systems reconstitute the core cellular functionalities of transcription, translation, and metabolism outside of the confines of the cell. This format enables experimentation simply by adding DNA or other components to a reaction, offering rapid experimental cycles compared to cells and higher control over conditions. Here we describe our work to leverage the high throughput capacity of this technology towards biomanufacturing and sensing applications.

We present a fast and cost-effective modular cell-free expression platform for the generation of high-resolution cryo-EM structures of GPCR/G protein complexes. We show with the full-length human ß1-adrenergic receptor and with the histamine-2 receptor the first molecular structures of GPCR/G protein complexes obtained from cell-free generated samples. Key advantages are (i) direct cotranslational GPCR insertion into tailored membranes of nanoparticles, (ii) no contacts to detergents, (iii) synthesis of full-length, non-engineered GPCRs without deletions or fusions, (iv) suitable for a wide array of GPCRs, oGPCRs, and other membrane proteins, and (v) free choice of ligands and lipid environment.

Artificial intelligence-enabled drug design produces a large number of possible therapeutic candidates that need to be down-selected for biological validation. High-throughput antibody specificity (on- and off-target) assays are often restricted to a limited subset of specific toxicity targets and lack customizability, which is even more critical for antibody-drug conjugates (ADCs) that may interact with intracellular targets. Sensor-integrated proteome on chip (SPOC) is a novel proteomic biosensor that provides qualitative, quantitative, and kinetic binding affinity data simultaneously against thousands of extracellular domains and intracellular protein targets on a single 1.5-sq cm surface plasmon resonance (SPR) biosensor chip. Real-time kinetic SPR analysis provides insight into transient interactions to identify off-targets earlier in the discovery pipeline for lead candidate identification, and to permit design of tunable on- and off-rates for binding interactions.

Cysteine-dense peptides (CDPs) are an attractive pharmaceutical scaffold that display extreme biochemical properties, low immunogenicity, and the ability to bind targets with high affinity and selectivity. Identifying CDPs that can be expressed in mammalian cells is crucial in predicting their compatibility with gene therapy and mRNA therapy. We developed CysPresso—a novel machine learning approach combining time series-inspired methods and embeddings—that predicts recombinant expression of CDPs based on their primary sequence. Our study showcases the applicability of deep learning-based protein representations, such as those provided by AlphaFold2, in tasks beyond structure prediction.

The inherent challenge of protein structural enablement is to obtain the target protein of sufficient quality and quantity to support various structural techniques. Our goal is to compile large datasets from a semi-automated, small-scale, data-rich characterization workflow into a minable database, allowing us to build correlations between construct designs and favorable outcomes and ultimately to predict successful designs. This workflow leads to fewer design iterations, greater probabilities of success and reduction in resource demand.

While N-glycosylation is a known critical quality attribute (CQA), the impact of O-glycosylation on therapeutic protein quality remains elusive. My presentation describes the establishment of an O-glycoengineered CHO cell line platform to modulate O-glycosylation and provides evidence that O-glycosylation of therapeutic glycoproteins, such as etanercept, impacts the physicochemical properties and biological activity. This O-glycoengineered production platform is valuable to identify O-glycosylation as a CQA for glycoprotein drugs to improve quality.

Through gene-specific cell engineering, we have developed a panel of glycoengineered CHO cells, (geCHO) enabling rapid production of bespoke glycoforms of therapeutic proteins. With this platform, we have produced and screened multiple therapeutic drug candidates (vaccines and enzymes) in vitro and in vivo to determine their optimal glycoform with respect to antigenicity, activity, and stability, etc.

Many next-generation therapeutics remain intrinsically challenging to produce in CHO cells. We exploited a cumate-inducible CHO platform allowing reduced expression of various classes of r-proteins during selection of stable pools. Fed-batch productions showed that pools generated without cumate (OFF-pools) were significantly more productive. Cell viability was lower and pool recovery was delayed during selection of ON-pools (mimicking constitutive expression), suggesting that high producers were likely lost or overgrown by faster-growing, low-producing cells. Using an inducible system to minimize r-protein expression during pool selection can contribute to reducing cellular stresses, including ER stress and metabolic burden, leading to improved productivity.