Downstream bioprocessing: designing the train from harvest to bulk

The train, in order

1. Clarify the harvest without shearing it. Centrifugation then depth filtration is the conventional route for mammalian culture. Cell density and viability at harvest decide the filter area needed, and a culture harvested late shears more debris into the stream and raises the burden on every step after it.
2. Capture on affinity, and design for what it does not remove. Affinity capture takes the product out of a complex feed in one step at high purity. It does not remove aggregate, charge variants or leached ligand, which is what the polishing steps exist for. The eluate is also low pH, which is convenient for the next step.
3. Run viral inactivation downstream of capture, at low pH. Holding the low pH eluate for a defined time is the standard inactivation step for enveloped viruses, and placing it here uses the pH the capture elution already produced. The hold time, pH and temperature are validated parameters, not process convenience.
4. Polish on two orthogonal modes. Two steps selecting on different properties, typically an ion exchange and either a hydrophobic or a mixed mode step, remove aggregate, host cell protein and DNA. Orthogonality is the point: two steps selecting on the same property remove the same impurities twice.
5. Concentrate and exchange by tangential flow filtration. Tangential flow filtration concentrates the product and exchanges it into the formulation buffer in one operation. Membrane cut-off, flux and the number of diavolumes are the parameters, and the practical limit is usually the viscosity the product reaches at final concentration.

Yield accounting across the train

Step yields multiply. Five steps at ninety percent each leave under sixty percent, which is why an apparently small loss in a polishing step is worth as much attention as a large one in capture. Track step yield from the first development run and keep the record.

Hold steps between operations lose product too, through adsorption and through degradation. Hold time and container material belong in the development record rather than in the operator's judgement.

Single use against stainless

Single use equipment removes cleaning validation and changeover time and adds consumable cost and supply risk. Below a certain scale and campaign length it wins clearly; above it, the consumable bill dominates.

Extractables and leachables from single use assemblies are a product contact issue that has to be assessed for the actual buffers and contact times used, not read off a general vendor document.

What to fix before scale-up

Residence times, linear velocities, load challenges and buffer compositions are the parameters that must hold constant. Anything defined in terms of column volume or absolute time rather than in scalable terms will change when the column diameter does.

Establish the operating ranges rather than single set points. A process characterised only at its target values cannot demonstrate control, and the first deviation becomes an investigation with nothing to compare against.

Tangential filtration: concentration and buffer exchange in one operation

Removing permeate concentrates the retained product; adding fresh buffer at the rate permeate leaves exchanges the buffer at constant volume. The two are normally run in sequence, and the order affects both the time and the product's stability at the concentrations reached along the way.

Diafiltration removes small molecules exponentially with the volumes exchanged, so the number of exchange volumes rather than the time is what determines how complete it is. Working in exchange volumes makes a process transferable between scales; working in minutes does not.

Fouling, cleaning and reuse

Flux falls as material deposits on the membrane, and how much of that is recoverable by cleaning decides whether a membrane is reusable. A cleaning regime is validated by measuring water flux and integrity after cleaning, and a membrane that does not recover is a consumable rather than an asset.

Reuse saves money and adds a cleaning validation burden and a carryover question. Single-use assemblies cost more per batch and remove both, which is why they dominate where batches are frequent and different products share equipment.

Continuous operation and perfusion

In a perfusion culture the filter retains cells while medium is exchanged continuously, which makes the filter part of the bioreactor rather than a downstream step. Reversing the flow periodically keeps the membrane clear far longer than steady flow does, which is what makes long perfusion runs possible.

Used ahead of a production step to reach a high seed density, the same arrangement shortens the production run. It adds a pump, a controller and a consumable that has to last the whole campaign, and those are the practical constraints rather than the biology.

What comes before the membrane: sizing a depth filtration system

A feed carrying cells, debris or precipitate will foul any membrane quickly, so a clarification step comes first. A graded depth medium captures particles through its thickness rather than on a surface, which gives it far more capacity for a dirty feed than a membrane of the same area.

Sizing that step is done by running the real feed to a pressure limit and measuring the throughput per unit area. It is the least glamorous part of the train and the part most likely to stop a batch when it is undersized.

Sterilising filtration and its integrity test

A sterilising filter is only evidence if its integrity was tested, and the standard non-destructive test measures the pressure at which gas passes through a wetted membrane. Below that pressure gas moves only by diffusion; above it, bulk flow begins at the largest pore.

The measured pressure is compared against a value the manufacturer has correlated with retention of a challenge organism, and it is run after use as well as before where the filtration supports a sterility claim. A dye-based check is used where a visual confirmation of an assembly is wanted, and it does not replace the pressure test.

Dialysis is diafiltration without a pump

Putting a sample inside a semi-permeable membrane in a large volume of buffer exchanges small molecules by diffusion until equilibrium. It is slow, needs no equipment and no pressure, and it is the gentlest buffer exchange available, which is why it survives for precious and fragile material.

The arithmetic is the same as diafiltration: exchange is exponential in the volume ratio, so two or three changes of a hundredfold volume remove small molecules almost completely while one large change does not. Working in changes rather than in hours is what makes a protocol transferable.

Specifying tubing, and preparing it

Tubing is specified by its molecular weight cut-off, which is nominal and has a distribution, and by its flat width, which sets the volume it holds per unit length. A cut-off close to the product loses material, so a margin of several times below its mass is the usual rule.

Regenerated cellulose tubing is supplied with glycerol and sulphur compounds and has to be rinsed, and for trace-metal or nucleic acid work boiled in a chelating solution before use. Skipping that preparation puts the tubing's own contaminants into the sample, which is the commonest complaint about it.

Cassettes and devices with a fixed membrane area remove the tying, the clipping and the preparation, at a higher price per sample, and they are the right answer for routine small volumes. Tubing remains cheaper for large volumes and for anything unusual.

Common questions

What order should downstream bioprocessing steps go in?

Clarification, affinity capture, low pH viral inactivation, two orthogonal polishing steps, viral filtration, then tangential flow filtration into the formulation buffer. The order is chosen so each step's output suits the next one's input.

Why is viral inactivation downstream of capture?

Because the affinity elution is already at low pH, so the hold costs no extra buffer exchange, and because the feed is far cleaner than the harvest, which makes the step more robust and easier to validate.

What does tangential flow filtration (TFF) do that a column cannot?

It concentrates and exchanges buffer without binding the product, so there is no elution step and no capacity limit in the usual sense. The limits are membrane flux and the viscosity of the concentrated product.

How many polishing steps are needed?

Usually two, selecting on different properties. One is sometimes enough for a very clean capture and a tolerant specification; three suggests the capture step or the upstream process is doing less than it should.

How do I size the membrane area of a tangential flow filtration system?

From flux measured on a small device with your real feed, not from a water flux figure. Process time is area multiplied by flux, and a real feed is usually several times slower than water.

Tangential flow filtration cassette or hollow fibre?

Cassettes for high area in a small footprint where the product tolerates shear; hollow fibres for shear-sensitive material and feeds carrying cells or particulates, which is why perfusion uses them.

How is a sterilising filter's integrity proved?

By measuring the pressure at which gas passes through the wetted membrane and comparing it against the manufacturer's value correlated with organism retention. Where the filtration supports a sterility claim, the test is run after use as well as before.

How many buffer changes does dialysis need?

Exchange is exponential in the volume ratio, so two or three changes of a large volume remove small molecules almost completely while one change does not. Work in changes rather than in hours.

Does cellulose tubing need preparing?

Yes. It is supplied with glycerol and sulphur compounds and has to be rinsed, and boiled in a chelating solution for trace-metal or nucleic acid work. Skipping it contaminates the sample.

Browse by service class

• Compare Gene synthesis on advertised prices