Biotech Manufacturing Process: The Key Phases Explained

Biotech manufacturing turns living cells into medicines through a sequence of tightly controlled phases, from growing the cells that produce a therapeutic protein all the way through purifying, formulating, and packaging the final drug product. The process broadly splits into upstream work (everything involved in producing the protein inside cells), downstream work (extracting and purifying it), and formulation and fill-finish (getting it into a vial or syringe in a form that stays stable). Each phase has its own engineering challenges, and a misstep at any point can compromise the entire batch. What follows is a walk through those phases in roughly the order they happen on a manufacturing floor, along with the practical considerations that make biotech production so different from conventional pharmaceutical manufacturing.

Choosing the Right Expression System

Before anything else, manufacturers have to pick the biological system that will actually make the protein. The choice matters more than it might seem, because the host organism determines how the protein folds, what chemical modifications it carries, and how much of it you can produce. For many therapeutic antibodies, Chinese hamster ovary (CHO) cells are the workhorse. Mammalian cells like CHO can add the sugar structures and other modifications that human proteins need to function properly in the body.

Bacterial systems like E. coli are faster and cheaper to grow, but they lack the machinery for many of the modifications found in more complex organisms. That makes them a poor fit for large antibodies but a reasonable choice for smaller proteins that do not need those modifications. Yeast systems such as Pichia pastoris and insect cell systems sit somewhere in between, offering better protein folding and more modifications than bacteria while remaining easier to culture than mammalian cells.¹ The cell line you pick cascades through every subsequent decision, from bioreactor design to purification strategy, so getting it right at the start saves enormous headaches later.

Upstream Processing and Bioreactor Culture

Once a cell line is established, upstream processing is about growing those cells at scale and coaxing them to produce as much protein as possible. Cells are expanded from small flasks through progressively larger vessels until they reach production-scale bioreactors, which can range from a few hundred liters to 20,000 liters or more. Inside the bioreactor, conditions like temperature, pH, dissolved oxygen, and nutrient supply are all tightly controlled. Even seemingly minor details, such as how fast the impeller spins, can make a significant difference. Higher impeller speeds improve oxygen delivery but also create mechanical shear forces that can damage cells, so operators need to balance mixing with cell survival.²

The two dominant culture strategies are fed-batch and perfusion. In fed-batch, cells grow in a closed vessel with nutrients added periodically, and the entire batch is harvested at the end. In perfusion, fresh media flows continuously through the bioreactor while spent media (containing the product) is continuously removed, and the cells are retained. Perfusion cultures can achieve much higher productivity per unit time and tend to produce a more consistent product, partly because the protein spends less time sitting in a soup of cell debris and enzymes. One head-to-head comparison found that average productivity in perfusion was roughly seven and a half times greater than in fed-batch, and that exposure of the product to cell breakdown material was seven to ten times lower.³

Fed-batch is still widely used, though, because the equipment and operations are simpler, and water and consumable usage tend to be lower. An economic analysis of monoclonal antibody manufacturing found that perfusion using alternating tangential flow filtration could offer cost savings of around 20 percent compared to fed-batch when cell densities were high enough, but fed-batch had a better environmental sustainability profile in terms of water and consumable use.⁴ In practice, the choice between the two depends on the product, the facility, and the manufacturer’s comfort with operational complexity.

Harvest and Primary Recovery

When upstream culture is done (or continuously, in perfusion), the next job is separating the protein product from the cells and cellular debris. This step, called primary recovery or clarification, is the bridge between upstream and downstream processing. It sounds straightforward, but mammalian cell cultures are messy. The fluid contains intact cells, dead cell fragments, DNA, lipids, host cell proteins, and the target protein all mixed together.

The two main approaches are centrifugation and microfiltration, and many facilities combine them. Centrifugation spins the culture fluid to separate cells and heavy debris from the liquid, and depth filtration then catches finer particles that the centrifuge missed. This combination of centrifugation with depth filtration has become the preferred approach for handling the complex fluids from mammalian cell culture and bacterial fermentation alike.⁵ Microfiltration can also be used on its own as a primary recovery step.⁶ The goal at this stage is not purity so much as getting rid of the bulk material that would clog or foul the chromatography columns used in the next steps.

Chromatographic Purification

Once the fluid is clarified, the real purification begins. For monoclonal antibodies, the first chromatography step is almost always Protein A affinity chromatography. Protein A is a bacterial surface protein that binds tightly and specifically to a particular region on antibodies. When the clarified fluid passes through a column packed with Protein A resin, the antibodies stick while most impurities flow through. Then the conditions are shifted (typically by lowering the pH), and the antibodies release from the resin in a highly concentrated, much purer form. This single step clears away host cell proteins, DNA, and molecular weight variants with remarkable efficiency.⁷

Getting the best performance out of Protein A requires careful tuning. Parameters like the pH used for binding, the pH used for elution, and the flow rate through the column all affect how much product you recover and how pure it is.⁸ Different antibodies and Fc-fusion proteins also bind to Protein A with different efficiencies, largely because of differences in how big the molecules are and how they crowd each other on the resin surface.⁹

After Protein A capture, the product typically goes through one or two “polishing” chromatography steps to remove remaining trace impurities. These polishing steps are especially important for clearing host cell proteins that can co-purify with the antibody and potentially trigger immune responses in patients. Multimodal chromatography resins, which interact with proteins through more than one type of chemical interaction, have shown strong results here. One study using a multimodal resin with a targeted wash step achieved an 80 percent reduction in total host cell protein content, including several particularly stubborn impurities that had resisted removal in earlier steps.¹⁰ Ongoing development of new multimodal resins with different ligand structures and pore sizes continues to push the boundaries of what polishing can achieve.¹¹

Viral Safety

Because many biotech drugs are made in mammalian cells, there is an inherent risk that viruses could contaminate the product, either from the cell line itself or from adventitious sources during manufacturing. Regulatory agencies require manufacturers to demonstrate robust viral clearance across multiple steps, using different mechanisms, so that the overall process provides a very high margin of safety.

Two key viral clearance steps are typically built into the downstream process. Low pH viral inactivation exposes the intermediate product to acidic conditions that destroy the lipid envelopes of many viruses. This is often done right after the Protein A step, since that elution already uses low pH. Continuous versions of this step have been developed that perform equivalently to traditional batch operations, inactivating industry-standard model viruses effectively within about 15 minutes at low pH.¹²

Virus filtration is the other major step, and it works by a different principle entirely: size-based exclusion. Specialized nanofiltration membranes physically block viral particles while allowing the much smaller antibody molecules to pass through. Because the retention mechanism is largely mechanical rather than chemical, virus filters provide a robust safety layer regardless of the specific virus type. This step is a standard component of nearly every recombinant protein purification process.¹³

Formulation and Fill-Finish

By the time the protein has been purified, it is in a buffer solution that is great for chromatography but not necessarily suitable for injecting into a patient. Formulation is the process of getting the protein into a stable, deliverable form. A common step here is ultrafiltration/diafiltration (UF/DF), where membranes concentrate the protein and swap the surrounding buffer to the final formulation buffer.¹⁴ This sounds like a simple buffer exchange, but in practice it is tricky. The protein itself interacts with the excipients (stabilizers, sugars, surfactants) in the formulation, so the concentrations of those excipients on either side of the membrane do not always end up where you expect them.¹⁵ Even the choice of buffer itself can affect how much protein passes through the membrane and is lost.¹⁶

Some proteins are stable enough to be stored as liquid solutions, and those are filled directly into vials or pre-filled syringes. Others are not, and for those, lyophilization (freeze-drying) is the standard solution. Lyophilization removes water from the formulation under vacuum, leaving a stable dry cake that can be reconstituted before use. The process is slow and energy-intensive, so there is strong industry interest in optimizing it.¹⁷ Lyophilization has also become the leading approach for stabilizing lipid-based nanoparticles, such as those used in mRNA vaccine delivery, which are notoriously unstable in liquid form.¹⁸

Fill-finish itself takes place in highly controlled cleanroom environments. Vials are washed, depyrogenated (heated to destroy bacterial endotoxins), filled with the drug product under aseptic conditions, stoppered, and sealed. Contamination at this final stage would ruin a batch that may represent months of upstream and downstream work, so the environmental controls and monitoring here are some of the strictest in any manufacturing industry.

Real-Time Monitoring and Process Analytical Technology

Traditionally, quality in biotech manufacturing was verified by testing samples taken from the process at defined points and analyzed in a lab after the fact. The problem is that by the time you get results back, the batch has already moved on, and if something went wrong, it is too late to fix it. The industry has been shifting toward process analytical technology, or PAT, which embeds sensors and analytical tools directly into the manufacturing process so operators can monitor product quality and process conditions in real time.¹⁹

Spectroscopic techniques like UV-Vis, Raman, and near-infrared spectroscopy can measure protein concentration, detect aggregates, and track impurity clearance without ever pulling a sample out of the process stream. Biosensors offer high specificity for monitoring particular quality attributes. These tools feed data into control systems that can adjust process parameters on the fly, keeping the process within its optimal range rather than discovering deviations after the fact. PAT is also expanding beyond traditional antibody production: recent work has demonstrated inline UV-Vis and optical density measurements for continuous monitoring during mRNA purification by precipitation and tangential flow filtration.²⁰

Scaling Up Without Losing the Product

A process that works beautifully in a 3-liter benchtop bioreactor does not automatically work at 2,000 liters. Scale-up is one of the most challenging aspects of biotech manufacturing because the physics of mixing, gas transfer, and heat removal all change as vessel size increases. A strategy that has proven effective is scaling based on keeping specific power input and air flow rates proportional. One approach demonstrated across five different monoclonal antibody projects successfully transferred processes from 3-liter to 2,000-liter bioreactors across four different manufacturing sites by maintaining a minimum volumetric air flow sufficient to strip carbon dioxide produced by cell metabolism, combined with matched power input per unit volume.²¹

Scale-up is not just about the bioreactor, though. Every downstream step also has to scale, and the relationships are not always linear. A chromatography column that performs perfectly at bench scale might show different flow distribution patterns at manufacturing scale, affecting resolution and yield. Filtration steps face analogous challenges with membrane area and pressure management. Each unit operation needs its own scale-up strategy, and the interactions between steps add another layer of complexity.

Raw Material Variability

One of the less glamorous but deeply consequential challenges in biotech manufacturing is variability in raw materials. Cell culture media, the nutrient solution that cells grow in, is a complex mixture of amino acids, vitamins, sugars, trace metals, and other components. Lot-to-lot variation in these materials can change how cells grow, how productive they are, and even the quality profile of the protein they produce.²²

Hydrolysates, which are broken-down proteins from plant or yeast sources used as media supplements, are particularly problematic. They remain poorly characterized, and although a detailed metabolomics study found that the batch-to-batch variability is driven by a relatively small fraction of the compounds present, identifying and controlling those compounds is an ongoing effort.²³ This is why many manufacturers have moved toward chemically defined media, where every ingredient is known and controlled, though not all processes can use them.

Process Validation and Regulatory Expectations

Before a biotech manufacturing process can be used commercially, it must be validated to demonstrate that it consistently produces product meeting predetermined quality specifications. Regulatory frameworks typically describe this as a lifecycle with multiple stages: process design and characterization, process performance qualification, and continued process verification.

Process performance qualification, or PPQ, involves running a defined number of batches under commercial conditions to confirm the process works as expected. Determining how many PPQ batches are needed is itself a scientific exercise, drawing on knowledge from earlier development stages and historical variability for similar products to establish a statistical confidence level.²⁴ For newer continuous manufacturing technologies, validation approaches have had to adapt. A recent study demonstrated how process characterization and PPQ could be applied to a continuous chromatography system, using a UV-based PAT tool to supervise quality in real time during the qualification runs.²⁵

Process changes after approval, whether upgrading equipment, switching raw material suppliers, or optimizing a purification step, trigger comparability assessments. Manufacturers have to demonstrate that the product made after the change is equivalent to the product made before. These assessments rely on an array of analytical techniques that probe everything from higher-order protein structure to biological activity.²⁶

Single-Use Versus Stainless Steel Equipment

A facility-level decision that shapes the entire manufacturing process is whether to use traditional stainless steel equipment or single-use (disposable) systems. Single-use bioreactors, bags, tubing, and filters eliminate the need for cleaning and sterilization between batches, which saves water and reduces the risk of cross-contamination. At laboratory scale, one life cycle assessment found that a single-use bioreactor system used only about 5.3 kilograms of water per gram of protein produced, compared to nearly 12 kilograms for a stainless steel system, with the difference driven largely by the cleaning and sterilization water that stainless steel requires.²⁷

The carbon footprint picture is more complicated than it used to seem, however. A detailed bottom-up analysis at the 2,000-liter scale found that the carbon footprint of single-use technologies was higher than earlier estimates had assumed, with filters and bags identified as key emission hotspots. As electricity grids get cleaner, the energy penalty of cleaning stainless steel shrinks, and some stainless steel process steps now have a lower carbon footprint than their single-use counterparts. The study concluded that a hybrid facility design, mixing single-use and stainless steel where each makes the most sense, can cut both plastic waste and emissions.²⁸

The Push Toward Continuous Manufacturing

Conventional biotech manufacturing is batch-based: each step runs to completion before material moves to the next. Continuous manufacturing links unit operations together so that material flows through the entire process without stopping. The potential advantages include smaller equipment, more consistent product quality, and lower cost of goods. There is growing application of integrated and continuous bioprocessing for mammalian cell-derived therapeutics, and frameworks for connecting upstream perfusion to downstream purification are being developed and tested.²⁹

Some groups have demonstrated integrated platforms that run perfusion culture with single-column continuous chromatography capture, producing antibodies at steady state for extended periods.³⁰ The reality, though, is that fully continuous end-to-end manufacturing remains rare in commercial production. Challenges persist in integrating traditionally separate steps, monitoring product quality across a continuously flowing process, and satisfying regulatory expectations that were designed with batch operations in mind.³¹ Most facilities that have adopted continuous elements use them for specific steps, such as perfusion upstream or continuous chromatography downstream, while keeping the rest of the process in batch mode.

Adapting the Platform for New Modalities

Much of the biotech manufacturing infrastructure was built around monoclonal antibodies, which have been the dominant product class for decades. As the industry moves into gene therapies, cell therapies, and mRNA-based products, the manufacturing toolkit is being stretched and adapted. Adeno-associated virus (AAV) vectors for gene therapy, for example, use many of the same unit operations originally developed for antibodies, including similar bioreactor culture, chromatography capture, and filtration steps. But AAV particles behave very differently from antibodies under process stress, and the analytical tools and stress-testing approaches developed for antibodies do not always translate well.³²

mRNA manufacturing represents an even more dramatic departure. The upstream process involves in vitro transcription rather than cell culture, and the purification steps rely on precipitation and chromatographic methods that differ substantially from antibody platforms. The development of PAT for mRNA purification, as mentioned in the context of real-time monitoring, is one example of how the tools and principles of established bioprocessing are being adapted rather than discarded for these newer products. The general architecture of biotech manufacturing, moving from production through purification to formulation with quality monitoring at each stage, remains intact. The specific methods inside each box are what is evolving.