Chemically Defined Medium

A medium is chemically defined when the chemical identity and the concentration of every component in it are known.

That criterion excludes, by definition:

• Serum (fetal bovine, newborn calf, human), which contains thousands of proteins, lipids, hormones and metabolites at concentrations that vary between lots and between donor animals
• Protein hydrolysates and peptones — soy, wheat, yeast, casein — which are enzymatic digests containing undefined mixtures of peptides of unknown sequence
• Tissue and plant extracts, conditioned medium, platelet lysate, and any other biological preparation of uncharacterised composition

It does not exclude protein. A chemically defined medium may contain recombinant human insulin, recombinant transferrin, recombinant albumin or a recombinant growth factor, because each is a single molecule of known sequence added at a known concentration. This is the point most often misunderstood, and it is why "chemically defined" and "protein-free" are two different labels that happen to overlap.

Note also what the definition does not promise. Chemically defined says nothing about whether a medium is animal-free — bovine insulin purified from pancreas is a defined molecule but is animal-derived. And it says nothing about performance. A chemically defined medium is a medium whose composition you know, which is a prerequisite for control and reproducibility, not a guarantee of higher yield.

These four terms are used loosely in catalogues and rigorously in regulatory filings. They are not a single scale, and understanding how they intersect prevents expensive procurement mistakes.

Serum-containing. The classical approach: a basal medium such as DMEM or RPMI 1640 plus 5–20% fetal bovine serum. Serum supplies everything the basal medium omits — lipids and cholesterol, transferrin and iron, albumin as a carrier and detoxifier, growth factors, hormones, attachment factors and protease inhibitors. It is convenient and universally permissive, and it is also the largest single source of variability in cell culture.

Serum-free (SFM). No serum. Serum's functions are replaced with specific additions, which may include purified or recombinant proteins, hydrolysates, lipid supplements and defined hormone cocktails such as ITS (insulin, transferrin, selenium) or its variants. Serum-free does not mean defined: a serum-free medium containing soy hydrolysate is undefined.

Animal-origin-free (AOF) / animal-component-free (ACF) / animal-derived-component-free (ADCF). No component of animal origin anywhere in the formulation or its manufacture, including the raw materials of raw materials. Plant hydrolysates and recombinant proteins expressed in non-animal hosts are permitted. This is a sourcing claim about biosafety and adventitious agent risk, not a composition claim — an AOF medium containing soy hydrolysate is animal-free but not defined. Xeno-free is the related term used in human cell therapy, meaning free of non-human animal components; a xeno-free medium may contain human-derived materials such as human serum albumin or human platelet lysate.

Protein-free. No protein of any kind, including recombinant. Peptides and hydrolysates may still be present, so protein-free is not automatically defined either.

Chemically defined (CD). Every component known by identity and concentration. Always serum-free and hydrolysate-free, by construction. May contain recombinant proteins. Protein-free chemically defined (PFCD) is the strictest common category: defined and containing no protein at all, with function supplied by small molecules, synthetic peptides, lipid emulsions and chelated metal ions.

The relationships are worth stating explicitly, because they are the source of most confusion: every chemically defined medium is serum-free, but most serum-free media are not chemically defined. A protein-free medium is not necessarily chemically defined. An animal-free medium is not necessarily chemically defined. And a chemically defined medium is not necessarily animal-free.

Since a chemically defined medium is defined largely by what it removes, it helps to be precise about what serum is.

Serum is the liquid fraction of clotted blood, with cells and clotting factors removed. Fetal bovine serum (FBS), the most-used, is collected from fetal calves at slaughter of pregnant cows; newborn calf serum, adult bovine serum, horse serum and human serum are also used. It is a biological fluid, not a formulation, and nobody has a complete inventory of its contents.

What it contributes to a culture:

• Albumin, at roughly 20–40 g/L in the neat serum, acting as a carrier for fatty acids, hormones and trace metals, as a buffer against shear, and as a sink for toxic species
• Transferrin, delivering iron in a form cells can take up
• Growth factors and hormones — insulin, IGF-1 and IGF-2, EGF, PDGF, FGF, hydrocortisone, thyroid hormones — at concentrations that vary between lots and between donor animals
• Lipids — cholesterol, phospholipids, free fatty acids — which most cultured cells cannot synthesise in adequate quantity
• Attachment factors including fibronectin and vitronectin
• Protease inhibitors, which is why serum-containing medium neutralises trypsin
• Trace elements, vitamins and metabolites at uncharacterised levels

At 10% in a culture, serum is supplying several grams per litre of protein. That is why it works so well and why it is such a problem: it is a permissive, broad-spectrum supplement whose composition you neither know nor control. Two lots from the same supplier can differ measurably in growth-promoting activity, which is why laboratories lot-test serum and then buy years of a lot that works.

Serum also carries adventitious agent risk — viruses, mycoplasma and, historically, transmissible spongiform encephalopathy concerns — which is what drives the regulatory pressure toward defined and animal-free processes for anything intended for human use.

Building a defined medium means replacing each function serum was performing, one at a time. In practice this is what the additions are for.

Iron delivery. Cells need iron and cannot take up free ferric ion efficiently; serum supplies transferrin. Defined media use recombinant human transferrin, or in protein-free formulations a synthetic chelator such as ferric ammonium citrate, ferric citrate or a hydroxypyridinone iron chelate.

Insulin signalling. Recombinant human insulin at roughly 5–10 mg/L, or the more stable analogue LONG-R3 IGF-1 in bioproduction media. This is usually the largest single driver of growth in a defined formulation.

Selenium. Sodium selenite at low nanomolar to micromolar levels, required as a cofactor for glutathione peroxidase and thioredoxin reductase. Without it, cells in a serum-free medium accumulate oxidative damage.

Lipids. Serum supplies cholesterol, fatty acids and phospholipids. Defined media use a chemically defined lipid concentrate — typically cholesterol, and arachidonic, linoleic, linolenic, myristic, oleic, palmitic and stearic acids solubilised with a synthetic surfactant such as Pluronic F-68 and, where permitted, a cyclodextrin carrier. Suspension CHO and NS0 lines are particularly cholesterol-dependent.

Antioxidant and carrier capacity. Albumin's role in serum is partly to bind and detoxify free fatty acids and reactive species. Recombinant albumin covers this in defined media that permit protein; protein-free formulations use small-molecule antioxidants, additional selenium, and careful control of trace metals instead.

Trace elements. Copper, zinc, manganese, molybdenum, vanadium, nickel and others, at nanomolar levels. Serum supplies these invisibly; a defined medium must specify them, and getting the trace element package wrong is a common reason a defined formulation underperforms.

Shear protection. Pluronic F-68 at 0.1–1 g/L for suspension and bioreactor culture, protecting cells from sparging and agitation damage in the absence of serum protein.

Attachment factors. For adherent cells, recombinant fibronectin, vitronectin or laminin, or a defined synthetic surface coating.

The basal medium underneath all this is usually not a classical one. Most defined formulations use a nutritionally rich, high-osmolality proprietary base — often DMEM/F-12-derived — because classical media like DMEM and RPMI were designed on the assumption that serum would fill their gaps.

Reproducibility. Serum lot variability is the largest uncontrolled variable in most cell culture. Two lots of FBS can differ substantially in growth-promoting activity, hormone content and lipid profile, which is why laboratories lot-test and then buy years of stock. A defined medium removes this variable entirely — the same formulation performs the same way in 2026 as it did in 2020.

Regulatory and biosafety position. Serum carries adventitious agent risk: viruses, mycoplasma and, historically, transmissible spongiform encephalopathy concerns. Regulators expect risk mitigation and documentation for animal-derived raw materials in any product intended for human use. A chemically defined, animal-origin-free process removes an entire section of that burden. For cell and gene therapy in particular, defined and xeno-free media are effectively an expectation rather than an option.

Downstream purification. Serum contributes grams per litre of protein that must be separated from a recombinant product. A defined medium — especially a protein-free one — dramatically simplifies capture and polishing and improves product purity.

Experimental interpretability. In signalling, metabolism, endocrinology and drug response studies, serum introduces hormones, growth factors, binding proteins and lipids that confound the measurement. Serum albumin binds hydrophobic compounds and shifts effective drug concentrations. Defined media make dose–response data mean what you think it means.

Supply and cost at scale. Serum supply is finite, price-volatile and ethically contested. At production scale a defined medium is both more predictable and, at volume, usually cheaper.

Where serum still makes sense: exploratory primary cell culture, established protocols where a defined alternative has not been validated, and any situation where the cost of adaptation exceeds the value of removing the variable. Defined media are not universally better; they are better when control matters.

Pluripotent stem cell culture is where chemically defined media became unavoidable, because undefined feeder layers and serum replacements made differentiation experiments impossible to reproduce. It is also the clearest published example of what a defined formulation actually looks like.

The Essential 8 (E8) formulation, published by Chen and colleagues in the Thomson laboratory in 2011, supports human iPSC derivation and culture with eight components — a deliberate demonstration that most of what had been in earlier stem cell media was unnecessary. The concentrations below are those given in the original publication:

Three things are worth noting. The basal medium is DMEM/F-12, not DMEM or RPMI, because F-12's trace elements and lipid precursors are needed once serum is gone. Ascorbate is supplied as the 2-phosphate salt, not free ascorbic acid, because free ascorbate oxidises in solution within days. And the formulation contains recombinant proteins — insulin, transferrin, FGF2, TGF-β1 — which is exactly why it is chemically defined rather than protein-free.

Mesenchymal stromal cell media follow a similar logic on a different base, typically alpha-MEM or DMEM/F-12 with insulin, transferrin, selenium, FGF2 and PDGF, replacing the 10–20% FBS of classical MSC protocols. For clinical-grade work these are usually specified xeno-free as well as defined.

When no catalogue product fits, media are formulated to order. This is more common than it looks — a large share of bioproduction runs on custom formulations, and it is worth understanding what the process involves before requesting one.

When a custom formulation is justified:

• An existing medium performs well but needs one or two changes: a different glucose level, glutamine removed, phenol red removed, a buffer swapped, a component omitted for a metabolic tracer study
• A process has been optimised at bench scale with a base plus several supplements, and consolidating them into a single formulation would remove handling steps and variability
• A published formulation is needed that no supplier stocks
• Scale-up requires a powdered or concentrated format of something currently bought as liquid
• Regulatory requirements demand an animal-origin-free or fully defined version of a medium currently containing hydrolysates

What the process usually involves. A specification is agreed covering the component list with concentrations, pH and osmolality targets, the format (liquid, powder, or concentrate), fill volume, and the sterility, endotoxin and performance-testing requirements. A pilot lot is manufactured and tested against the target specification and, importantly, against your cells — a formulation that meets its chemical specification can still underperform. Once the pilot is accepted, the formulation is locked and subsequent lots are made against it with a certificate of analysis per lot.

What to specify and often gets forgotten. Whether glutamine is included or supplied separately. Whether sodium bicarbonate is in the powder or added at reconstitution, and which CO2 atmosphere the formulation targets. The osmolality target, not just the component list — small changes to salt or glucose move it. The phenol red decision. Whether animal-origin-free status is required, which constrains raw material sourcing upstream of the medium itself. And the storage format, since powder and liquid have very different shelf lives and shipping costs.

For most laboratories the sensible sequence is to start from the closest catalogue formulation, prove the modification at bench scale with supplements, and only then consolidate into a custom lot — rather than commissioning a bespoke formulation from theory.

Selection is cell-line-specific. There is no general-purpose chemically defined medium. CD media are developed and optimised against a specific host — CHO, HEK293, BHK-21, NS0, hybridoma, Vero, iPSC — and a formulation that performs well for CHO will frequently not support HEK293 at all. Start from your host, then your application: expansion, transient transfection, stable production, virus production or differentiation each favour different formulations.

Adaptation is a process, not a swap. Cells maintained in serum for many passages have downregulated pathways they did not need. Two approaches are standard:

Sequential adaptation. Reduce serum stepwise — 10%, 5%, 2.5%, 1%, 0.5%, 0% — holding at each level for two to three passages until growth rate and viability recover before the next reduction. Slower, but higher success rate and less genetic drift.

Direct adaptation. Seed at high density directly into the defined medium and passage through the recovery. Faster where it works, and it works more often for robust suspension lines than for adherent or primary cells.

In both cases seed at a higher density than usual — cells in a defined medium are far more sensitive to conditioning factors they secrete themselves — and expect the culture to look poor for several passages before it recovers. Keep a frozen back-up of the serum-grown cells at every stage. Adaptation typically takes four to eight weeks.

Validate afterwards, not just growth rate. Adapted cells can differ from their parent in doubling time, metabolic profile, glycosylation of secreted protein, surface marker expression and transfection efficiency. Re-baseline the assays that matter to you and bank a new master cell bank in the defined medium.

Practical notes. Defined media are often supplied without L-glutamine so it can be added fresh or substituted with the stable dipeptide L-alanyl-L-glutamine. Many are supplied as powder for large-scale reconstitution. Because there is no serum protein to buffer against pH, osmolality and shear excursions, defined-medium cultures are less tolerant of process error — control your feeds, your CO2 and your agitation more tightly than you would with serum.

"Chemically defined means protein-free." It does not. A CD medium may contain recombinant insulin, transferrin or albumin. Protein-free chemically defined (PFCD) is the separate, stricter category.

"Serum-free means defined." It does not. Most serum-free media contain plant or yeast hydrolysates, which are undefined mixtures of peptides.

"Animal-free means defined." It does not. Soy hydrolysate is animal-free and thoroughly undefined.

"Chemically defined means animal-free." Not necessarily. Bovine insulin or animal-sourced cholesterol are defined molecules of animal origin. If you need both properties, look for a medium labelled chemically defined and animal-origin-free or ADCF.

"A defined medium will grow anything." Defined media are host-specific and generally less permissive than serum-containing media. This is the direct cost of removing serum's broad, undefined support.

"Adaptation is a one-week job." Four to eight weeks is typical, and the adapted line should be treated as a new line requiring re-characterisation.