The surprising applications of tangential flow filtration

Tangential flow filtration (TFF) is used to concentrate conditioned medium on a laboratory bench. It is also used to process tens of thousands of litres of milk, whey or antibody in a manufacturing plant. The principle is the same at both ends of that range. This article looks at where TFF came from, how it became established in industry, and how it is used today, from protein and antibody manufacturing through to newer applications such as viral vectors, mRNA lipid nanoparticles and extracellular vesicles (EVs).

What is tangential flow filtration?

In conventional filtration, often called dead-end or normal flow filtration, the liquid is pushed straight at the filter. Anything that is retained builds up on the surface as a cake, and the flow rate falls until the filter blocks. In TFF the feed is pumped across the membrane, parallel to its surface. Part of the liquid passes through the membrane as permeate (also called filtrate), and the rest, the retentate, returns to the feed reservoir. Flow across the surface sweeps away much of the material that would otherwise accumulate, so a membrane can be used for far longer and with much larger volumes.

The driving force is the transmembrane pressure (TMP), the average pressure difference between the feed side and the permeate side of the membrane. Two operations are carried out. In concentration, permeate is removed and the volume of retained product falls. In diafiltration, fresh buffer is added at the same rate as permeate is removed, washing out small molecules and exchanging the buffer. Most processes combine the two, which is why TFF in bioprocessing is often referred to as UF/DF (ultrafiltration/diafiltration).

TFF describes the flow, not the membrane. The same mode of operation is used with microfiltration membranes, which retain cells and particles, and with ultrafiltration membranes, which retain proteins and other macromolecules.

Where did TFF come from?

Separating colloids from small molecules with a membrane was a laboratory technique long before it was an industrial one. In 1907 Heinrich Bechhold described filters made by impregnating filter paper with gelatin hardened with formaldehyde, or with collodion (cellulose nitrate). By varying their composition he was able to control pore size, and he is generally credited with coining the term ultrafiltration.^1,2 Richard Zsigmondy patented a membrane filter in 1918 that became the basis of the membrane filters later manufactured by Sartorius.² These early membranes were used in what would now be called dead-end mode, mainly to study colloids and estimate particle size, and flow through them was slow.

The turning point came from research on desalination. Around 1960, Sidney Loeb and Srinivasa Sourirajan at the University of California, Los Angeles made cellulose acetate membranes with an asymmetric structure, a very thin, dense skin on a much thicker porous support.³ Because the selective layer was so thin, the membranes combined good salt rejection with a water flux high enough to make reverse osmosis practical.

Alan Michaels, a chemical engineer at the Massachusetts Institute of Technology, founded Amicon Corporation in 1962.⁴ Building on the Loeb-Sourirajan membrane and on work at MIT on polyelectrolyte complex hydrogels, Amicon developed in 1963 what Michaels later described as the first synthetic, high-permeability ultrafiltration membrane able to retain macromolecules. Michaels himself thought it curious that the membrane came out of a collaboration with the engineering company Dorr-Oliver, started in 1963 to develop a large-scale industrial separation process.⁵ Membranes made from non-cellulosic synthetic polymers followed later in the decade, and Michaels' 1968 review in Chemical Engineering Progress introduced ultrafiltration to the wider chemical process industries.⁶

The new membranes were permeable enough to reveal a different problem. As water passes through, retained solutes accumulate in a thin layer at the membrane surface. At high concentrations this layer can form a gel. This concentration polarisation, rather than the membrane itself, often limited the flux. In the laboratory the solution was to stir the liquid above the membrane, as in the stirred ultrafiltration cells that Amicon sold for many years. At process scale, the feed was instead pumped at high velocity through thin channels, tubes or hollow fibres, so that the boundary layer was continually swept away. That is tangential flow, or cross-flow, filtration, and it remains the way ultrafiltration is operated at scale.

The first industrial applications

One of the first large-scale uses of ultrafiltration had nothing to do with biology. In electrodeposition painting, or electrocoat, a car body or other metal part is immersed in a water-based paint and coated by passing a current through it. The automotive industry adopted the process during the 1960s, but paint carried out of the tank on the coated parts was lost in the rinse water, and salts and other contaminants built up in the bath. From 1969 PPG Industries patented the use of ultrafiltration to treat electrocoat baths.⁷ The membrane retains the paint resin and pigment and passes water and dissolved salts, giving a clean permeate that can be used for rinsing while the recovered paint goes back to the tank. Electrocoat lines still use ultrafiltration in much the same way.

The dairy industry followed. Whey, the liquid left over from cheese making, contains valuable proteins at low concentration and was largely treated as waste. Ultrafiltration allowed the proteins to be concentrated and the lactose and salts removed, and whey protein concentrates became a major product. In 1969 Jean-Louis Maubois, Germain Mocquot and Louis Vassal at INRA in France developed what became known as the MMV process. Milk is ultrafiltered to give a liquid with roughly the composition of the finished cheese, and this is then coagulated with rennet.⁸ Because the whey proteins are retained in the cheese instead of being drained off with the whey, yields are higher.

Polymer membranes limited the temperatures and cleaning chemicals that could be used. Ceramic membranes made from materials such as zirconia and titania became commercially available in the mid-1980s² and can be cleaned with hot acid and alkali and sterilised with steam. Together with improved polymer membranes, they helped extend cross-flow filtration into fruit juice, wine and beer clarification. Microfiltration and ultrafiltration are now also used widely in drinking water and wastewater treatment, where plants can contain many thousands of square metres of hollow fibre membrane.

TFF in biopharmaceutical manufacturing

Recombinant proteins brought TFF into pharmaceutical manufacturing, with rather different requirements from paint or dairy. Membranes had to bind as little protein as possible, withstand repeated cleaning in sodium hydroxide, and give retention that was reproducible enough to be validated. Regenerated cellulose and polyethersulfone became the main ultrafiltration membrane materials. They are used either in flat sheet cassettes, in which layers of membrane are separated by mesh screens that promote mixing in the feed channel, or in hollow fibre cartridges with open lumens.⁹

UF/DF is now used at the end of almost every protein purification process to bring the drug substance to its final concentration and exchange it into the formulation buffer, and it is often used between chromatography steps as well. Microfiltration in tangential flow is used to harvest cells and clarify culture fluid. As antibody doses have increased and subcutaneous delivery has become more common, target concentrations have risen to the point where the viscosity of the concentrated protein, rather than the membrane, sets the limit on what can be achieved.

Several variations of TFF have been developed for particular problems:

High performance TFF. Conventional TFF can only separate solutes that differ in size by around ten-fold. In 1997, a team led by Robert van Reis at Genentech showed that proteins of similar size could be separated by operating in the pressure-dependent part of the flux curve, controlling start-up conditions carefully and using membranes with narrower pore size distributions. They were able to separate IgG from bovine serum albumin, and albumin monomers from oligomers.¹⁰ Later work added charged membranes and optimised buffer pH and ionic strength.

Alternating tangential flow. In perfusion culture, cells are kept in the bioreactor while spent medium is continuously removed and replaced. Recirculating the culture through a conventional TFF loop can damage cells and foul the membrane. Alternating tangential flow (ATF), developed by Jerry Shevitz at Refine Technology, uses a diaphragm pump to move culture back and forth through a hollow fibre filter, giving rapid, alternating flow with low shear.¹¹ ATF is now one of the most widely used cell retention methods in perfusion processes.

Single-pass TFF. SPTFF removes the recirculation loop altogether. Membrane modules are connected in series to create a long flow path, so that the target concentration is reached in a single pass. In the first published study, bovine IgG at about 45 g/L was concentrated to between 190 and 225 g/L with recoveries of more than 95%.¹² Because the retentate leaves the system rather than returning to a tank, SPTFF can be connected directly to other steps in a continuous process. It has also been used to reduce the volume of intermediate product pools that would otherwise be too large for the available tanks.¹³

Viruses, nanoparticles and vesicles

The products processed by TFF have become larger and, in many cases, more fragile. Viral vaccines and gene therapy vectors such as adeno-associated virus (AAV) and lentivirus are much larger than antibodies, so more open membranes can be used to retain them while host cell proteins and other impurities pass through. Enveloped viruses such as lentivirus can lose infectivity through shear in the pump and in the membrane channels, and membrane type, module format and flow rates have to be chosen with this in mind.¹⁴

In mRNA vaccine manufacture, lipid nanoparticles form when lipids dissolved in ethanol are mixed rapidly with mRNA in an acidic buffer. The ethanol then has to be removed, the particles exchanged into a neutral buffer and the product concentrated. TFF is the standard way of doing all three at scale.

EVs are one of the most recent additions. For many years the usual method of isolating EVs was differential ultracentrifugation, which is difficult to scale beyond a few hundred millilitres and can cause vesicles to aggregate. In 2014, Heinemann and colleagues described a sequential filtration method that combined dead-end prefiltration, TFF and filtration through a track-etched membrane to isolate exosomes from cell culture supernatants and large volumes of biofluid.¹⁵ In 2018 Busatto and colleagues compared TFF with ultracentrifugation for EVs from conditioned medium. TFF gave a higher yield of EVs with similar physical characteristics, fewer free macromolecules and aggregates, and better batch-to-batch consistency, in about half the processing time.¹⁶

Also in 2018, Haraszti and colleagues produced exosomes from mesenchymal stem cells grown on microcarriers. Isolating these by TFF with a 500 kDa cartridge gave seven times as many exosomes as ultracentrifugation of the same cultures. Exosomes produced by the combination of three-dimensional culture and TFF were also seven times more potent at delivering siRNA to neurons than exosomes from conventional two-dimensional culture isolated by ultracentrifugation.¹⁷

TFF concentrates EVs very effectively, but on its own it does not separate them from particles of similar size, such as lipoproteins and protein aggregates. It is therefore often followed by size exclusion chromatography or another method. How much additional purification is needed is still being debated, and the answer may well be different for biomarker studies and for EV therapeutics.

From the laboratory to the plant

One reason TFF is so widely used in industry is that a process developed with a few hundred square centimetres of membrane can be scaled up to many square metres by keeping a small number of parameters constant. These are the membrane and module type, the length of the flow path, the volume processed per unit area of membrane (the membrane loading, in L/m²), the cross-flow rate per unit area or per fibre, and the TMP. Manufacturers design their cassette and hollow fibre ranges so that these can be held constant from small devices to process modules.

Other things do not scale so simply. The hold-up volume, the liquid that remains in tubing, pump and module and cannot be drained, is proportionally much larger in a small system, which limits how far a small sample can be concentrated and affects recovery. Pumps differ in the shear they generate. Heat from the pump builds up in a recirculating loop, more quickly in some systems than others. Choosing a suitable filter for the sample type and volume is covered in more detail in an earlier article on choosing the right TFF filter.

In research laboratories, TFF is increasingly used in place of centrifugal concentrators and ultracentrifugation for volumes from a few hundred millilitres to a few litres. At this scale it is often run by hand, with a peristaltic pump and a clamp on the retentate line, and without measuring pressures or permeate flow. That works, but it makes it difficult to tell whether flux is falling because the membrane is fouling or because the retentate is becoming more viscous, and harder to reproduce a run or hand the method on to a larger system.

The EVlution™ TFF system from Cell Guidance Systems uses hollow fibre cartridges available in ten pore sizes, so that one set-up can be used to concentrate EVs, viruses, proteins or antibodies. The recently launched EVlution™ Pro instruments add monitoring of volumes and pressures during each run, using the same cartridges.

Where next?

Much of the current development in TFF is driven by continuous manufacturing. Single-pass and staged configurations for continuous concentration and diafiltration are being developed so that TFF can be linked directly to perfusion culture and continuous chromatography, and single-use TFF assemblies are now common in clinical and multi-product facilities.¹⁴ High-concentration antibody formulations continue to push against the limits set by viscosity.

For EVs and other complex biological products the questions are less about engineering and more about the product. EV preparations are heterogeneous, and it is not yet clear how membrane type, cut-off and shear affect which subpopulations are recovered, or whether those differences matter for function. Answering that will need the same kind of careful process data that made TFF routine for proteins.

Frequently asked questions

What is the difference between tangential flow filtration and ultrafiltration?
Ultrafiltration describes the type of membrane, one that retains macromolecules such as proteins while passing water, salts and small molecules. Tangential flow filtration describes how the liquid flows, across the membrane surface rather than straight through it. Ultrafiltration membranes can be used in dead-end devices such as centrifugal concentrators, and TFF can be carried out with microfiltration as well as ultrafiltration membranes. Most process-scale ultrafiltration is, however, carried out in tangential flow.

Who invented tangential flow filtration?
There is no single inventor. Heinrich Bechhold coined the term ultrafiltration in 1907.¹ The asymmetric membranes developed by Loeb and Sourirajan around 1960 made high flux possible,³ and Alan Michaels and Amicon Corporation produced the first synthetic high-permeability ultrafiltration membranes in 1963.⁵ Flowing the feed across the membrane became the standard way of operating these membranes at scale because it limits the build-up of retained material at the membrane surface.

Why is TFF used to isolate extracellular vesicles?
TFF can process large volumes of conditioned medium or biofluid, scales well and exerts relatively gentle forces on vesicles. In direct comparisons with ultracentrifugation, TFF gave higher EV yields, fewer aggregates and better batch-to-batch consistency in less time.^16,17 It is usually combined with a second step such as size exclusion chromatography when higher purity is needed.

References

1. Bechhold H. Kolloidstudien mit der Filtrationsmethode. Z Phys Chem. 1907;60:257-318.
2. Tamime AY, ed. Membrane Processing: Dairy and Beverage Applications. Chichester: Wiley-Blackwell; 2013.
3. Loeb S, Sourirajan S. Sea water demineralization by means of an osmotic membrane. Adv Chem Ser. 1963;38:117-132.
4. National Academy of Engineering. Alan S. Michaels. In: Memorial Tributes: Volume 12. Washington, DC: National Academies Press; 2008. doi:10.17226/12473.
5. Michaels AS. Fifteen years of ultrafiltration: problems and future promises of an adolescent technology. In: Cooper AR, ed. Ultrafiltration Membranes and Applications. Polymer Science and Technology, vol 13. New York: Plenum Press; 1980. doi:10.1007/978-1-4613-3162-9_1.
6. Michaels AS. New separation technique for the CPI. Chem Eng Prog. 1968;64(12):31-43.
7. PPG Industries Inc. Ultrafiltration of electrodepositable compositions. US Patent 3,663,405. 1972.
8. Maubois JL, Mocquot G, Vassal L. Preparation of cheese using ultrafiltration. US Patent 4,205,090. 1980.
9. van Reis R, Zydney A. Bioprocess membrane technology. J Membr Sci. 2007;297(1-2):16-50.
10. van Reis R, Gadam S, Frautschy LN, et al. High performance tangential flow filtration. Biotechnol Bioeng. 1997;56(1):71-82.
11. Shevitz J. Fluid filtration system. US Patent 6,544,424. 2003.
12. Casey C, Gallos T, Alekseev Y, Ayturk E, Pearl S. Protein concentration with single-pass tangential flow filtration (SPTFF). J Membr Sci. 2011;384(1-2):82-88.
13. Dizon-Maspat J, Bourret J, D'Agostini A, Li F. Single pass tangential flow filtration to debottleneck downstream processing for therapeutic antibody production. Biotechnol Bioeng. 2012;109(4):962-970.
14. Zydney AL. New developments in membranes for bioprocessing: a review. J Membr Sci. 2021;620:118804.
15. Heinemann ML, Ilmer M, Silva LP, et al. Benchtop isolation and characterization of functional exosomes by sequential filtration. J Chromatogr A. 2014;1371:125-135.
16. Busatto S, Vilanilam G, Ticer T, et al. Tangential flow filtration for highly efficient concentration of extracellular vesicles from large volumes of fluid. Cells. 2018;7(12):273.
17. Haraszti RA, Miller R, Stoppato M, et al. Exosomes produced from 3D cultures of MSCs by tangential flow filtration show higher yield and improved activity. Mol Ther. 2018;26(12):2838-2847.

Cell Guidance Systems supplies EVlution™ TFF systems and hollow fibre cartridges for concentrating EVs, proteins, antibodies and viruses, alongside Exo-spin™ purification, ExoLISA™ quantification and a range of EV and exosome services that can help accelerate your research with confidence.

Learn more about powerful technologies that are enabling research: