Maintaining sufficient volumes of high-quality seed culture is a major challenge, especially as more automated production platforms come online. These platforms typically demand 10-100L of cell culture with =99% viability each week, which at this scale is labor-intensive and difficult to automate. Here we are introducing a fully-automated platform to resolve these issues and bring reliability and scalability to traditional cell culture work.

Next-Generation biologics for biotherapeutics and drug development requires advanced and complex characterization assays. We present a novel device that enables complete sample preparation, including thermal heating, centrifugation, filtration, viscosity independent ML/AI integrated liquid handling, vacuum washing, and microfluidic electrophoresis procedures. This device enables higher complexity microfluidic characterization assays, particularly protein-based assays for biotherapeutic development to run independently, efficiently, and reproducibly.

Improving protein production efficiency and increasing capacity to produce large screening sets is crucial to accelerate early-stage discovery and lead optimization. This talk will highlight the integration of Design of Experiments (DOE) methodologies with automated liquid handling to systematically optimize a transient Chinese Hamster Ovary (CHO) expression system. Key factors explored include plasmid DNA usage, transfection reagent, and transfection and feeding strategies. The automated DOE workflow leads to improved cell viability, increased protein production titer, and decreased transfection DNA usage without compromising product quality, while maintaining culture scale and compatibility with high-throughput automated protein production.

Adaptyv Bio operates an automated cloud laboratory for high-throughput protein validation, serving over 50 companies, from major biopharmas to frontier AI labs. We present our experience expressing over 20,000 AI-designed proteins using cell-free systems, systematically evaluating expression tags, codon optimization strategies, and system variants. We describe integrating these workflows into a fully automated platform capable of expressing arbitrary customer proteins at scales from sub-microgram to milligram quantities.

Join a Think Tank group to share and experience and hear what others have learned.

Think Tank 1: Tactics for Minimizing Production Time and Costs

•    What are the keys to enhance cross-functional collaborations that lead to optimized CEPA workflows?

•    How do we effectively and realistically prioritize requests?

•    Standard workflow for non-standard proteins

Co-Hosts:

Richard Altman, MS, Field Application Scientist, Thomson Instrument Company

James Bowman, PhD, CTO, AI Proteins

Anita Ghosh, PhD, Associate Director, Protein Engineering and Production, AstraZeneca 

Think Tank 2: Eliminating Bottlenecks

•    What/where are the bottlenecks?

•    Doing more with less 

•    Delivering peak performance from researchers to hosts to instruments 

Co-Hosts:

Mark Ellis, Senior Principal Scientist, UCB Pharma

Mansi Malhotra, PhD, Advisor, Product Development, Elanco Animal Health

Daniel Poole, PhD, Senior Scientist, Biologics HTP Expression Sciences, Johnson & Johnson Innovative Medicine

Think Tank 3: Transition from Benchtop to R&D 

•    Outsourcing versus insourcing for capacity flexibility

•    Miniaturized construct optimization

•    Balancing faster throughput with downstream needs

Co-Hosts: 

Heather Lopes, Director & Head, Protein Sciences, Aera Therapeutics

Ayla Sessions, Associate Director, AstraZeneca

Jarrod Shilts, PhD, Group Leader, ExpressionEdits Ltd.

5:20 Think Tank Report Outs: Listen and Learn

​During the Think Tank Table discussions, we shared our experiences and working solutions for end-to-end protein production workflows. Now as a collective community, let’s hear from the table facilitators as they share key discussion points and strategies. What can we take away and apply?

• Where have you seen AI transforming the production value chain the most (yield optimization, purification, formulation, etc.)?
• What parts of protein production remain unchanged by the implementation of AI tools and why do you think that is the case?
• If you could magically have AI improve one part of the process, what would it be?​

• Map Your Bottlenecks - Where Does Time Actually Get Lost?
• Automate Smarter, Not Harder: Automating for Speed and Consistency
• Turning Data into Actionable Insights 
• Faster Testing Without Compromising Quality
• Designing Workflows That Scale: AI-Enabled Predictive Developability​

High-throughput protein production remains a time-consuming, expensive bottleneck that relies heavily on trial-and-error experiments. Here we present a holistic, automation-friendly solution to maximize the efficiency, by leveraging synthetic biology and AI techniques. It involves the application of a massive library for multiplexed expression improvement and AI-aided optimization, and chromatography-free purification via synthetic organelles in vivo. This integrated pipeline de-risks discovery by making protein production a predictable, scalable, and automated engineering task.

Production of multi-specific antibodies is often complicated by product related impurities, such as mispaired species and half molecules. Without repeated productions and process optimization, it is hard to predict the impurities and downstream process. We will share an analytics platform that is implemented at harvest, prior to affinity chromatography consisting of HT MS, cGE, and analytical HPLC, to help quantitate impurities and guide purification strategies.

Mapping the human intracellular chemoproteome is a major frontier for drug discovery.  I will describe an optimized high-throughput protein expression and purification workflow that seamlessly integrates with our next-generation DEL screening platform. By systematically identifying and eliminating key noise sources through reengineered methodologies, we have dramatically improved DEL data quality. We are now approaching 1,000 human proteins profiled — an unprecedented scale that reveals clearer protein–small-molecule interaction patterns and reduces hit-set artifacts common in small DEL experiments. I will conclude with perspectives on how large, high-quality experimental datasets are essential for training truly predictive ML/AI models in drug discovery.

Protein Sciences at Aera Therapeutics has developed a robust multi-production platform that encompasses multiple expression systems, utilization of complex chromatography over automated 4-step and 2-step preparative purifications, and a robust internally-built yeast display platform.

Mammalian cell transfection enables the rapid generation of diverse large-molecule modalities for drug discovery. We have built an automated ecosystem to produce highly pure, complex antibodies suitable for cell-based functional assays and developability assessments. Although automation of protein production has grown exponentially in recent years, automated QC and characterization platforms remain challenging. Here, we present a high-throughput workflow for protein quantification and purity evaluation that accelerates hit-to-lead progression.