High cell density culture systems

Every reactor geometry in this cluster is an answer to one problem, and it is worth stating the problem once. Oxygen is barely soluble in water. Henry’s law fixes the equilibrium concentration: medium in contact with air at 37 °C holds roughly 7 mg per litre, and enriching the gas raises that ceiling proportionally but does nothing else. A vessel therefore cannot store oxygen for a dense culture; it can only deliver it continuously, and the supply is a rate: OTR = kLa × (C* − C_L). The driving force is small and fixed by physics, so the only real variable is kLa — the liquid-film coefficient multiplied by gas–liquid interfacial area per unit volume. That is why the answers all look like ways of making more interface, or of moving liquid past an interface faster. The interfacial mechanics themselves — spargers, coalescence, antifoam depressing kLa, bubble-rupture damage, dissolved carbon dioxide — are treated in bioreactor gas mixing systems; this page sets out what density does to the rest of the process.

Why scale-up has no correct answer#

Geometric scale-up is over-constrained. Hold power per unit volume constant and impeller tip speed rises with scale; hold tip speed constant and power per volume falls, taking kLa with it. Mixing time grows with vessel size under any of these rules, so a large tank has gradients — in dissolved oxygen, in pH, in feed concentration — that a two-litre vessel does not, and cells circulating through it experience an oscillating environment rather than a mean one. Volumetric gas flow per unit volume falls as the tank grows, which is what strands carbon dioxide in large vessels. No rule preserves all of these at once. Scale-up therefore begins by choosing which criterion to keep and continues by predicting what the discarded ones will cost.

Where the limit moves at high density#

Push density hard enough and oxygen stops being the binding constraint. Mammalian cells growing on high glucose run aerobic glycolysis and excrete lactate; glutamine catabolism releases ammonium. Both inhibit growth well before any nutrient is exhausted — ammonium at a few millimolar, lactate at tens of millimolar, with the added burden that base added to neutralise lactic acid drives osmolality up. Perfusion exists for that reason. It is a removal technology first and a feeding technology second: continuously exchanging medium keeps inhibitors below threshold, which is what allows densities above 10⁸ cells per millilitre that fed-batch cannot approach.

What limits retention#

Perfusion requires holding cells back while medium leaves, and the retention device is where these systems fail. Hollow-fibre filtration, in tangential flow or in the alternating-flow variant that reverses direction each cycle to sweep the membrane, is limited by fouling: a gel layer of cells, debris and DNA builds on the fibre wall, transmembrane pressure climbs, and product sieving falls until the protein being made is retained along with the cells. Alternating flow was designed specifically to slow that; it does not abolish it. Retention run time, not oxygen transfer, is usually what sets how long a perfusion campaign lasts.