DMEM (Dulbecco's Modified Eagle Medium)

DMEM was described by Renato Dulbecco and Gordon Freeman in 1959 as a modification of Harry Eagle's medium for plaque assays of polyoma virus. The change was deliberately simple: take Eagle's formulation and use a "fourfold concentration of amino acids and vitamins." Eagle's own Minimal Essential Medium, published the same year, had already established the minimum set of nutrients a mammalian cell needs in culture — thirteen essential amino acids, eight vitamins, glucose, and the six inorganic salts of Earle's balanced salt solution.

Raising those concentrations mattered because Eagle's MEM was designed to be minimal. Cultures exhausted it quickly and had to be refed often. Quadrupling the amino acids and vitamins gave a medium that supports higher cell densities between feeds, which is exactly what a virologist counting plaques over several days needs.

Modern commercial DMEM is not literally 4x MEM in every component. Suppliers added glycine, serine and ferric nitrate, and shifted the salt balance: sodium bicarbonate went from 2,200 to 3,700 mg/L and sodium chloride dropped from 6,800 to 6,400 mg/L to compensate osmotically. The medium sold today as "DMEM" is a stable, well-characterised standard, but it is a commercial standard rather than the literal 1959 recipe.

These are the two most-used basal media in cell culture, and the question of which to use comes up constantly. They are not interchangeable, and the differences are structural rather than cosmetic.

The decision usually reduces to three questions.

Is the culture adherent or in suspension? DMEM's 1.8 mM calcium supports firm attachment; RPMI's 0.42 mM was deliberately kept low so lymphoblastoid cells would stay in suspension. Adherent fibroblast and epithelial lines go in DMEM; suspension lymphoid and haematopoietic cells go in RPMI.

What does the line's reference protocol say? For established lines this is not a preference. Switching a line between DMEM and RPMI changes doubling time, saturation density and often protein expression, which invalidates comparison with the published literature unless you re-characterise the line afterwards.

How metabolically demanding is the culture? DMEM at 25 mM glucose and roughly four times RPMI's amino acid concentration is built for dense, fast, high-demand cultures. RPMI is built for cells growing at moderate rates over days, and dense cultures will exhaust it sooner.

Neither medium is simply richer than the other. DMEM wins on amino acids and glucose; RPMI wins on vitamins, phosphate buffering and its unusual reduced-glutathione content.

The table below is the standard high-glucose DMEM formulation (the reference formulation used across major suppliers, corresponding to Gibco 11965 and Sigma D5796-class products). Low-glucose DMEM is identical except that D-glucose is 1,000 mg/L instead of 4,500 mg/L.

Inorganic salts and bulk components

Amino acids (mg/L)

Glycine 30; L-arginine·HCl 84; L-cystine·2HCl 63; L-histidine·HCl·H2O 42; L-isoleucine 105; L-leucine 105; L-lysine·HCl 146; L-methionine 30; L-phenylalanine 66; L-serine 42; L-threonine 95; L-tryptophan 16; L-tyrosine disodium salt dihydrate 104; L-valine 94.

Vitamins (mg/L)

Choline chloride 4; D-calcium pantothenate 4; folic acid 4; niacinamide 4; pyridoxine·HCl 4; thiamine·HCl 4; i-inositol 7.2; riboflavin 0.4.

Note what is absent. DMEM contains no non-essential amino acids beyond glycine and serine — no alanine, asparagine, aspartate, glutamate or proline. It contains no nucleosides, no biotin, no vitamin B12 and no lipoic acid. Its only trace metal is a token amount of ferric nitrate. Everything else a cell needs is expected to come from serum. This is the single most important structural fact about DMEM: it is a rich amino-acid and vitamin base built on the assumption that serum will supply lipids, trace elements, carrier proteins and growth factors.

The two glucose levels are not interchangeable defaults, and picking one on habit is a common source of irreproducible data.

High glucose (4,500 mg/L, 25 mM) is roughly five times physiological blood glucose. Use it for fast-growing transformed lines, for transient transfection and virus production in HEK293, for hybridomas and for any culture you intend to push to high density without frequent feeding. The extra glucose acts as a nutrient reservoir. The cost is that cells grown at 25 mM glucose lean heavily on glycolysis, acidify the medium faster through lactate accumulation, and can show a glycolytic phenotype that is an artefact of the medium rather than of the biology under study.

Low glucose (1,000 mg/L, 5.6 mM) is close to normal human blood glucose. Use it for primary hepatocytes, for mesenchymal stromal cells where high glucose has documented effects on differentiation, for long-term or differentiation cultures, and for any metabolic study where you do not want glycolysis running at maximum. Low-glucose DMEM is also the more appropriate choice when you are comparing oxidative and glycolytic metabolism, because 25 mM glucose suppresses oxidative phosphorylation through the Crabtree effect in many transformed lines.

Glucose-free DMEM exists for the same reason: it lets you set the carbon source yourself, substituting galactose to force oxidative metabolism (a standard approach in mitochondrial toxicity screening) or titrating glucose to a defined level.

If you are switching an established culture between glucose levels, do it gradually and re-establish your growth curve. Doubling time, confluence at passage and transfection efficiency all shift.

Most of the confusion around ordering DMEM comes from four independent options that suppliers combine freely.

Glutamine. L-glutamine at 4 mM is standard but degrades in solution — it cyclises to pyrrolidone carboxylic acid and releases ammonia, with a half-life of a few weeks at 4 °C and much less at 37 °C. Glutamine-free DMEM lets you add fresh glutamine at use, or substitute a dipeptide form (L-alanyl-L-glutamine), which is stable in solution and releases glutamine slowly as cells cleave it. For anything cultured longer than a week without a full medium change, the dipeptide is the better engineering choice.

Sodium pyruvate. At 1 mM, pyruvate provides an additional energy substrate and a small antioxidant benefit. It is genuinely useful for cells at low density or low serum, where it aids plating efficiency. Leave it out if you are doing metabolic flux work, where an unlabelled pyruvate pool complicates tracer interpretation.

Phenol red. A pH indicator at 15 mg/L, not a nutrient. Remove it when it interferes with your readout — fluorescence and absorbance assays in the 550–600 nm region, flow cytometry — and remember that phenol red has weak oestrogenic activity, which matters for hormone-responsive lines such as MCF-7.

HEPES. Usually 10–25 mM, added to stabilise pH during handling outside the incubator. HEPES buffers well near pH 7.2–7.4 and does not depend on CO2. The trade-off is real: HEPES is phototoxic under light, generating hydrogen peroxide, and at 25 mM it is cytotoxic to some primary and stem cell cultures. Use it when your protocol involves extended time on the bench, not as a default.

DMEM/F-12 is a 1:1 mixture of DMEM and Ham's F-12, and it is the most useful thing in the DMEM family for anyone moving toward reduced serum.

Ham's F-12, published by Richard Ham in 1965, was designed as a chemically defined medium for clonal growth and is nutritionally the opposite of DMEM: modest amino acid concentrations, but a broad set of components DMEM lacks — trace elements (zinc, copper, iron as ferrous sulfate), lipid precursors (linoleic acid, lipoic acid, putrescine), nucleoside precursors (hypoxanthine, thymidine) and additional vitamins (biotin, B12). Mixing the two gives a base with DMEM's nutrient concentrations and F-12's breadth.

The standard DMEM/F-12 formulation contains 3,151 mg/L glucose (17.5 mM), 2,438 mg/L sodium bicarbonate (29 mM), 116.6 mg/L calcium chloride, 55 mg/L sodium pyruvate (0.5 mM), 365 mg/L L-glutamine (2.5 mM) and 8.1 mg/L phenol red, plus the F-12 trace elements at half their F-12 concentration. A widely used HEPES variant substitutes 15 mM HEPES and drops the bicarbonate to about 1,200 mg/L.

DMEM/F-12 is the standard base for serum-free and reduced-serum formulations, for keratinocyte and epithelial culture, for neural stem cells (with N-2 or B-27 supplement), and for hybridoma and iPSC work. If you are trying to lower serum in a DMEM-based process, moving the base to DMEM/F-12 is usually the first productive step, because the trace elements and lipid precursors that serum was quietly supplying are already in the medium.

The working pH of DMEM is 7.2 ± 0.2, the standard specification for the medium and the range in which most mammalian cells are cultured. That figure refers to complete medium equilibrated in its intended CO2 atmosphere, which is the only condition under which the number means anything.

DMEM is buffered by the bicarbonate/CO2 system. Dissolved CO2 forms carbonic acid, which dissociates to bicarbonate and a proton; pH is set by the ratio of bicarbonate concentration to CO2 partial pressure. The consequence is that the bicarbonate level in a medium and the CO2 level in the incubator are a matched pair.

This is why measuring the pH of a bottle straight from the fridge tells you very little. Outside a CO2 atmosphere the medium loses dissolved CO2 and reads alkaline — often 7.6 or above — and returns to specification within an hour or two in the incubator. If you need a meaningful reading, equilibrate an aliquot in the incubator with the cap loosened first, or use a HEPES-buffered formulation whose pH does not depend on CO2 at all.

DMEM's 3,700 mg/L (44 mM) bicarbonate was formulated for 10% CO2, not 5%. This is the most frequently ignored fact about the medium. At 5% CO2 the bicarbonate is in excess and the medium sits alkaline, typically pH 7.6–7.8 when freshly placed in the incubator, visibly pink-to-purple. Most laboratories run DMEM at 5% CO2 anyway and get away with it, because metabolising cells produce lactate and CO2 that pull the pH back down within hours. But a freshly seeded flask at low density has no such correction, and plating efficiency suffers. If you culture at 5% CO2 and see persistently purple medium, either move to 10% CO2, use a formulation with reduced bicarbonate, or add HEPES.

The reverse problem — medium turning yellow — is straightforward acidification. Phenol red is yellow below about pH 6.8. Overgrowth, high glucose driving lactate production, a leaking or empty CO2 supply, or a bacterial contamination will all do it. Yellow medium in a flask that was seeded yesterday and looks sparse under the microscope suggests contamination rather than metabolism.

Because bicarbonate escapes as CO2 whenever a bottle sits open at atmospheric pressure, medium left on the bench slowly becomes alkaline. This is normal and reverses in the incubator. It is a reason to keep bottle-open time short, not a reason to discard the bottle.

DMEM is the standard base for adherent, anchorage-dependent cells: HEK293 and its derivatives, HeLa, NIH/3T3 and other fibroblast lines, COS-7, Vero, MDCK, C2C12 myoblasts, primary fibroblasts, smooth muscle cells and endothelial cells (usually with additional supplements), and adherent CHO. Typical serum supplementation is 10% fetal bovine serum, or 5–10% newborn calf serum for hardier lines.

DMEM's high calcium (1.8 mM) is a real selection criterion, not a footnote. Calcium at that level promotes cadherin-mediated cell–cell adhesion and drives keratinocyte differentiation; keratinocyte and some epithelial protocols therefore specify a low-calcium base instead. Conversely, cells that need firm attachment do better in DMEM than in RPMI, whose calcium is only about 0.4 mM.

DMEM is a poor first choice for suspension lymphoid cultures — most human leukaemia and lymphoma lines, PBMC, hybridomas at low density — where RPMI 1640 is the established base. It is also not the medium for clonal growth at very low density without serum, which is what Ham's F-12 and DMEM/F-12 were built for.

Serum. 10% FBS is standard. Heat inactivation (56 °C, 30 min) is conventional for complement-sensitive work but is not universally necessary and degrades some growth factors; decide deliberately rather than by habit.

Common additions. Non-essential amino acids at 1x (100x stock) are worth adding to any DMEM culture at reduced serum, because DMEM's own non-essential amino acid content is limited to glycine and serine. Sodium pyruvate at 1 mM if not already present. Penicillin/streptomycin only where genuinely required — routine antibiotics mask low-level contamination and change the mycoplasma risk profile.

Storage. Liquid DMEM is stored at 2–8 °C, protected from light, and is typically stable for the labelled shelf life (commonly 12–24 months unopened). Once glutamine is added, treat the medium as having a working life of about four weeks at 2–8 °C. Riboflavin, folic acid, tryptophan and HEPES are all light-sensitive; leaving medium under laboratory lighting generates hydrogen peroxide and degradation products that reduce plating efficiency. Store bottles in the dark, not on an open shelf.

Warming. Warm only the aliquot you need. Repeatedly cycling a 500 mL bottle to 37 °C accelerates glutamine breakdown and bicarbonate loss. Warming to room temperature is sufficient for most work; medium does not need to reach 37 °C before it goes on cells.

Powdered DMEM is the economical option at scale, but it ships without sodium bicarbonate, which must be added at reconstitution (3.7 g/L for the standard formulation), the pH adjusted to roughly 0.1–0.2 units below target to allow for the rise during filtration, and the solution sterile-filtered through 0.22 µm. Use water of at least 18.2 MΩ·cm resistivity with low endotoxin.

Because DMEM is phenol-red buffered and transparent, it reports its own condition better than most reagents. Learning to read it saves incubators.

Bacterial contamination. The medium turns yellow rapidly — often overnight — and becomes visibly turbid or cloudy. Under the microscope you see small motile or refractile particles between cells, and at higher density a shimmering granularity. The diagnostic pairing is yellow plus turbid plus a sparse or recently seeded culture: metabolism alone cannot acidify a flask that has few cells in it.

Yeast contamination. Turbidity with visible ovoid budding particles, often with a slower pH drop than bacteria. Frequently appears as discrete clumps rather than uniform cloudiness.

Fungal or mould contamination. Visible filamentous structures, sometimes a fuzzy floating colony discernible by eye. The pH may stay near normal for some time, so mould can be missed on a colour check alone.

Mycoplasma. The dangerous one, because it produces no turbidity, no colour change and no visible organism under a standard light microscope. Cultures look normal while growth rate declines, saturation density falls and experimental results drift. Mycoplasma cannot be diagnosed by inspecting the medium — it requires PCR, a luminescence-based enzymatic assay or DNA staining, and it is the reason routine testing exists rather than being optional.

What is not contamination. Medium turning purple after standing on the bench is bicarbonate loss, not infection. Medium turning yellow in a confluent flask that was fed three days ago is normal metabolism. Fine crystalline precipitate after a cold-to-warm cycle is usually calcium phosphate coming out of solution rather than microbial growth.

Reducing the risk. Aliquot medium into working volumes so a single contamination event does not cost a whole bottle. Never return decanted medium to the stock bottle. Keep antibiotic use deliberate — routine penicillin/streptomycin suppresses low-level bacterial contamination into an invisible chronic state and does nothing against mycoplasma, so it can hide the problem it appears to solve.

Running high-glucose DMEM by default in metabolic work. 25 mM glucose is a pharmacological, not physiological, concentration and it changes metabolic phenotype.

Assuming DMEM is complete. It is not. Without serum or a designed supplement, DMEM lacks lipids, trace elements, transferrin and growth factors.

Ignoring the bicarbonate/CO2 mismatch. See above; this quietly costs plating efficiency in many laboratories.

Using DMEM/F-12 and DMEM interchangeably. They differ in glucose, calcium, bicarbonate, osmolality and trace element content. Data from one does not transfer to the other.

Adding glutamine to a bottle and forgetting the date. Glutamine decay produces ammonia, which is toxic and accumulates. Label the bottle at the moment of supplementation.

Sterile-filtering serum-containing medium through the wrong membrane. Protein binds and blocks; use a low-protein-binding membrane and prefilter.

Storing on a lit shelf. Photodegradation of riboflavin and HEPES is a real and measurable source of cytotoxicity.