How Engineering Is Transforming Cell Culture Optimization

Traditional cell culture media development relies on experience, intuition, and incremental improvements through experimentation. Researchers usually adjust one or two media components at a time, assess the impact on cell growth or productivity, and gradually refine formulations over multiple rounds of testing. This approach has underpinned many of the industry’s successes, particularly in Chinese hamster ovary (CHO) cells, which remain the industry standard and primary workhorse of monoclonal antibody (mAb) production. While this trial-and-error approach served the industry well for many years, recent rapid advances in biomanufacturing have outpaced media development, leading to bottlenecks and inefficiencies.

Impetus for change

Historically, there was little reason or capability to do things differently. The industry focused on a relatively small number of well-characterized production cell lines, and high-throughput experimentation using automation and AI was still in its infancy. Media optimization was largely a manual process because there were few viable alternatives. Over time, this iterative approach produced highly effective media, but it took decades of continuous refinement to reach today’s performance.

Today, mAb manufacturers are striving for higher titers, greater process consistency, and lower manufacturing costs. In addition to this, the industry is using more diverse cell lines to produce increasingly novel therapeutic modalities, each with distinct nutritional requirements. This shift is exposing the limitations of current media development approaches. The one-size-fits-all media design and legacy formulations are no longer suitable for an evolving industry. Traditional optimization methods are slow, resource-intensive, and increasingly difficult to apply as biological complexity grows. Rather than refining a single formulation for one production cell line, researchers are now faced with designing numerous media for a diverse range of cells, many of which have never been cultured at commercial scale.

When media becomes the bottleneck

One of the clearest signs that media is becoming a limiting factor is inconsistency. Experimental results become noisy, batch-to-batch variation increases, and processes become harder to reproduce or scale. Biology will always exhibit some degree of variability, but poorly optimized media can introduce additional uncertainty, making it difficult to distinguish genuine biological effects from problems caused by the culture environment. This issue is exacerbated by the use of undefined or animal-derived components—such as fetal bovine serum (FBS) and human platelet lysate (HPL) —which introduce unquantified variability, as well as safety concerns and supply chain vulnerabilities. This makes manufacturing less reliable, complicates regulatory submissions, and increases timelines, costs, and process risks.

Limitations of incremental optimization

Modern biomanufacturing approaches require well-defined, complex media. A modern chemically defined medium can contain 50 or more individual components, each with its own concentration range. Even testing three levels of each would generate more combinations than could be screened in several lifetimes, so trial-and-error optimization simply cannot navigate a design space this large. This is forcing the industry to begin treating media optimization as an engineering problem, not just a biological challenge. The engineering insight is that you do not need a complete mechanistic model of cellular metabolism to improve it. By treating the cell as a system—systematically varying media composition and measuring the response—data-driven models learn the input-to-output relationships that matter, and use them to predict which formulations to test next. Combining this with automation and high-throughput experimentation enables researchers to explore far more of the design space than manual methods ever could.

Modern culture media development

Modern media optimization begins by defining the objective. Depending on the application, this might be increasing cell growth, improving mAb titers, reducing manufacturing costs, or enhancing overall process robustness. Increasingly, this also means designing formulations that are chemically defined and animal-free from the outset, eliminating the variability, safety, and supply chain risks of components like FBS and HPL, rather than engineering around them. Drawing on both scientific expertise and computational tools, researchers identify the ingredients and formulation parameters most likely to influence performance. From there, computational models can be used to generate hundreds, or even thousands, of candidate media formulations for iterative testing.

Rather than preparing these manually, automated liquid handling systems can be used to produce each test formulation with a high degree of precision, before robotic platforms culture the cells under controlled conditions. Automated analytical instruments then measure key performance indicators, from cell growth and viability to productivity and other process-specific outputs, generating large, consistent datasets. This information is fed back into the optimization model, which analyzes how cells responded, identifying patterns that inform the next round of formulations for experimental testing. After only a handful of optimization cycles, researchers can converge on formulations that would have taken far longer to discover using conventional trial-and-error approaches.

Partnering for engineering-led development

Building this level of automation, data infrastructure, and computational expertise in house is neither practical nor cost-effective for most biomanufacturers. Many organizations are therefore choosing to work with specialist partners that have the capabilities for engineering-led media development, rather than relying on traditional trial-and-error optimization. Platforms that combine automation and computational modeling into an iterative workflow allow novel media to be developed faster and with greater confidence than conventional approaches. For example, an AI-driven platform was used to screen 150 de novo CHO DG44 feed formulations in just two months, doubling mAb expression while reducing media costs by 50-60 percent.¹ With this approach, media optimization can become a repeatable, data-driven process, rather than an exercise in trial and error. This is exactly what is needed in an industry where reproducibility, speed, and scalability are increasingly important.

Engineering the future of cell culture media

The demands of mAb manufacturing continue to evolve, and media development must progress alongside them. Combining biological expertise with engineering, automation, and data-driven optimization enables researchers to explore far more potential formulations than ever before. Rather than replacing biology, engineering is providing the tools to navigate its growing complexity and, in doing so, shape how the next generation of cell culture media will be developed. Ultimately, the goal is to make media development more predictable, not more complicated.

Reference

1. Multus. (2026). High-Throughput CHO Media Optimization with MediOP™ Platform and Valita Titer IgG Assay Integration. Available at: https://www.multus.bio/resources/high-throughput-cho-media-optimization-with-mediop-tm-platform-and-valita-titer-igg-assay-integration. Accessed August 5, 2026.

Kevin Pan is the Co-founder and Chief Technology Officer at Multus