Harvest and Recovery of Monoclonal Antibodies from Large-Scale Mammalian Cell Culture

Published on: 
BioPharm International, BioPharm International-05-01-2008, Volume 21, Issue 5

A comparison of primary harvest techniques.


The objective of harvest and clarification unit operations is the removal of cells and cell debris to enable product capture on a chromatographic column. The first step in the harvest of monoclonal antibodies from mammalian cell culture is cell removal, followed by filtration unit operations for additional clarification. Centrifugation and microfiltration have been the primary harvest techniques adopted industrially. Although depth filtration can be used as a primary harvest method, it is more common to see this unit operation follow the primary harvest step to provide additional clarification. Flocculants are sometimes added before harvest to improve harvest and clarification operations. Filtration through absolute pore size membranes is typically the final clarification step before capture chromatography. Expanded bed chromatography has been developed as an integrated unit operation that combines harvest with product capture, but to date, practical limitations have kept this technique from being adopted in commercial-scale operations.

Several types of unit operations have been used for harvest and clarification of extracellularly expressed products from mammalian cell culture. Key techniques include centrifugation, microfiltration, depth filtration, filtration through absolute pore size membranes, the addition of flocculants to improve other harvest steps, and expanded bed chromatography. This article provides a basic review of the fundamentals of these unit operations and practical considerations during process development. Further details on these unit operations have been reviewed elsewhere.1 The final section describes how these unit operations fit together to form the harvest and clarification process for monoclonal antibody (MAb) products.


Centrifugation uses the density difference between solids and the surrounding fluid. The centrifugal force accelerates the settling that would normally occur during sedimentation. Most industrial applications use disk stack centrifuges to remove cells and cell debris.2 Disk stack centrifuges offer continuous operation, making their throughput consistent with the desire to limit the time for harvest operations. Figure 1 shows the schematic for a disk stack centrifuge.

Editors note: Figure 1 for this article is not available online. To obtain a copy of the figure by fax, please contact hhaniff@advanstar.com

The basic principles of centrifugation involve a balance between the buoyant force acting on solid particles and Stokes's law, which expresses the drag force. The ratio of flow rate to the effective settling area (termed Σ) is held constant during scale-up of centrifugal operations. For a disk stack centrifuge, Σ is expressed as:

in which n is the number of disks, r0 and ri are the outer and inner radii of the disk, and θ is the angle of the disks from the vertical.

Operating conditions are often first screened at laboratory scale in a tubular bowl centrifuge and then translated to a disk stack centrifuge. For the two centrifuges:

The operating conditions (flow rate, RPM) for the disk stack centrifuge still require some empirical optimization even once the appropriate flow rate has been determined. This has been ascribed to a certain degree of cell disruption because of shear during centrifugation, which generates smaller particles that cannot be efficiently removed by the centrifuge under standard operating conditions.3 In general, long residence times (slow flow rates) will lead to a clearer centrate but at the expense of process throughput. The clarification efficiency of a centrifuge can be determined by a relative measurement of turbidity in the feed stream and the centrate. Not all aspects of a disk stack centrifuge can be captured effectively by a laboratory-scale tubular bowl centrifuge. In particular, shear stresses during entry and discharge from the bowl are difficult to scale. A scaled-down version of a disk stack centrifuge that requires less than 10 L of broth to operate has been developed to accelerate experimental development of operating conditions.4

Other operating parameters that require optimization for a continuous disk stack centrifuge operation include the discharge frequency, the discharge type, and the predischarge flush solution and volume. These parameters are determined empirically. The effectiveness of a centrifuge in removing particulates decreases as the bowl fills with the deposited solid sludge. However, too-frequent discharges will risk decreasing product yield and increase operating time. A practical compromise is to discharge the bowl after it fills up to 50–70% of the bowl volume. The discharge frequency is determined by the solids content of the cell culture broth. Experimentation with the ideal case of a continuous discharge in a nozzle centrifuge continues for the harvest of mammalian cell culture media but has not yet found favor in large-scale operation.5 In the case of discontinuous bowl discharge, the discharge type can be full or partial depending on whether the entire bowl contents are ejected during the discharge. Before discharge, a flush volume is used to push the contents of the bowl through the centrifuge so that yield losses during discharge are minimized. A buffer or water is often used for the flush. An important consideration is whether the osmotic difference between the flush fluid and the cell culture broth will cause lysis of cells releasing cell debris, host cell protein contaminants, DNA, and proteases, which can impact product quality negatively.


Tangential flow microfiltration (MF, also called cross-flow microfiltration) competes with centrifugation for the harvest of therapeutic products from mammalian cell culture.6,7 One advantage this technique offers is the creation of a particle-free harvest stream that requires minimal additional filtration.

Mass transport limitations resulting from the formation of a concentration polarization layer of particles close to the membrane surface remain a significant limitation of MF harvest operations.8 Various alternative flow configurations have been proposed to mitigate the effects of concentration polarization. One alternative is to use rotating disk filters that augment the cross-flow velocity close to the membrane surface, thus sweeping away the concentration polarization layer.9 Further developments have used Taylor vortices.10 The use of Dean vortices created by flow patterns in the MF device offers the possibility of reducing concentration polarization without the need for moving parts in the MF system.11 Yet another strategy has been the use of periodic backflushing to sweep the membrane surface clean.12

A comprehensive review of the fundamentals of cross-flow microfiltration and the phenomena involved during concentration polarization and fouling has been provided elsewhere.13 The flux decline during microfiltration has been described in terms of a series of physical phenomena.14

Optimal operation of tangential flow MF has been investigated by several researchers.6,7 A typical flux-versus-trans-membrane pressure (TMP) relationship is shown in Figure 2. In general, this can be typified by two regimes: i) a pressure-dependent regime in which an increase in TMP results in an increase in flux, and ii) a pressure-independent regime in which increases in TMP do not further increase flux. As a general rule, it is recommended to operate at the transition between these regimes to maximize flux while not permitting the TMP to rise to a level that would cause increased pore plugging and fouling of the membrane. A similar relationship exists for the cross-flow velocity at a given TMP, the effect of which also levels off at a certain point. Because TMP and cross-flow velocity are interdependent, one can maintain constant TMP operation only by manipulating back-pressure on the membrane to vary the cross-flow velocity as the operation proceeds and concentration increases.6

Figure 2. Typical flux-versus-trans-membrane pressure profile for cross-flow microfiltration

Microfiltration membranes used for cell culture harvest are often plagued with the problem of membrane fouling (i.e., irrecoverable declines in membrane flux). The operating conditions for the MF operation and the cleaning regimen for the membranes after use are both significant ways to address this issue. Another important variable is the membrane chemistry, with more hydrophilic membranes generally being less susceptible to significant fouling.


Optimizing MF Harvest Operations

The development of an MF harvest process has been outlined as a step-by-step process.7 Two important determinants for using MF systems for mammalian cell culture harvest are the flux and the product yield. The flux determines the surface area of membrane needed to process the cell culture broth, which has significant economic implications, because too high a flux can foul the membrane and shorten membrane lifetime.

The measurements of flux versus TMP and cross-flow rate curves at various concentrations at laboratory scale typically are the first series of experiments that are conducted. An easy way to carry out these experiments is to operate under total recycle mode in which the permeate is fed back into the load tank to maintain a constant concentration level. Steady state flux can then be measured over a few different cross-flow velocities to produce the flux versus TMP plot shown earlier in Figure 2. In these initial experiments, the broth should be concentrated to a degree which will be representative of the final desired concentration. Various membranes can be screened to identify ones that are optimal for the application because both the chemistry and the pore size play an important role in determining flux and flux decay characteristics. It is usual to operate at the transition point between the zones of increasing flux versus TMP and the zone of TMP-independent flux to maximize flux and minimize detrimental effects of fouling and pore plugging.

From this point on, it is typical to carry out further optimization at pilot scale so that the membrane configuration and channel width are representative of the large-scale operation. These experiments are carried out under non-steady state operating conditions (i.e., no recycle) so that concentration varies over the course of the experiment. Membrane loading (i.e., volume of broth processed per unit membrane surface area) is another important parameter for optimization. An effort should be made to conduct experiments with loadings in the correct ballpark of what is ultimately going to be operated at large-scale. A curve that is typically measured is the flux decay profile, which plots flux versus time (Figure 3). It is often difficult to predict how TMP will influence flux decay under these conditions from the steady state experiments conducted earlier. A higher TMP might result in a higher initial flux but cause more rapid flux decay or might have a beneficial effect because a majority of the total permeate might be collected in the very initial stages of the filtration. It is typical to attempt to optimize the area under the flux-versus-time curve while maintaining an upper limit on the processing time. Cross-flow velocities influence the filtration in-line with what was observed during the steady state screening experiments with the caveat that excessively high cross-flow velocities can cause undesirable effects of cell breakage, which may influence product quality (i.e., Chinese hamster ovary proteins (CHOP) and DNA levels) even before a significant impact is seen on the filtration.

Figure 3. A typical flux decay profile for microfiltration systems


Depth filtration (also called prefiltration or media filtration) refers to the use of a porous medium that is capable of retaining particles from the mobile phase throughout its matrix rather than just on its surface.15 These filters are frequently used when the feedstream contains a high content of particles.16 In such cases, depth filters can remove larger, insoluble contaminants before final filtration through a microfiltration membrane that would otherwise clog relatively quickly—hence the term prefiltration.6,17

Depth filters used in bioprocessing typically are composed of a fibrous bed of cellulose or polypropylene fibers along with a filter aid (e.g., diatomaceous earth) and a binder that is used to create flat sheets of filter medium. The filter aids provide a high surface area to the filter and are sometimes used by themselves in clarification applications.18 An additional charge can be imparted to some depth filters, either from the binder polymer or from other charged polymers incorporated into the filter.19 Sometimes, a microfiltration membrane with an absolute pore size rating is integrated into the depth filter sheet as the bottommost layer. Porous depth filters can retain particles in their tortuous flow channels to a level that size-based screening alone cannot achieve. A schematic of how a depth filter operates is shown in Figure 4.

Figure 4. Schematic of depth filter operation in removing particulates

For process-scale applications, depth filters are often fabricated into cells consisting of two layers of filters separated from each other such that flow occurs from the outside into the space between the layers and is then collected. Multiple cells can be stacked into a housing in which pressure is used to drive flow through the assembly. Depth filters are usually single-use devices that enable a reduction in the extent of process validation required for their use in biopharmaceutical applications.

Adsorptive Properties of Depth Filters

The high surface area and the possibility of charged interactions imply that depth filters can possess adsorptive properties in addition to their ability to trap larger particles. Positively charged depth filters have been used for a variety of applications, including the removal of endotoxins from water; the removal of virus particles that were smaller than the effective pore size of the filter; removing DNA from a buffer solution; and removing retrovirus and parvovirus during viral spiking studies into a solution of pure protein.20–23 More recently, depth filters were shown to reduce host cell protein impurities that would otherwise precipitate during subsequent Protein A column elution.24 Thus, in addition to clarifying the mammalian broth, depth filters also can adsorb some otherwise soluble impurities from the feed streams.

Placement of Depth Filters in a Harvest and Clarification Scheme

In a harvest and clarification scheme for mammalian cell culture broths, depth filters typically are placed after a centrifugation step.25 Because centrifugation (unlike MF) cannot efficiently remove all particulates from the broth, a secondary clarification step is needed. This niche is nicely filled by depth filtration, because the use of dead-end microfilters is often costly and prone to failure in the event of an excursion in particle counts from the centrifugation operation. Depth filters can be used as the sole harvest and clarification step, but this is not usually the most robust or cost-effective solution for large-scale operations. Because most depth filters do not come with an absolute pore size cutoff rating, a dead-ended microfilter is used in-line after the depth filter to effectively remove any residual particulates that might clog subsequent chromatographic steps.


Flocculants have been used for many years to improve the filtration of fermentation broths.2 These agents, ranging from simple electrolytes to synthetic polyelectrolytes, can act by several means to cause clumping of smaller particulates to form larger solids that can be filtered more effectively.

Filter aids for mammalian cell culture harvest are most often diatomaceous earths or perlites. More recently, chitosan, a nontoxic food-grade material, has been used as a flocculant for mammalian cell culture clarification.26 The use of chitosan was reported to increase depth filter volumetric throughput by six to seven fold.

The use of a calcium chloride and potassium phosphate combination as a flocculant also has been reported for MAb harvest.27 These two compounds, when combined, produce calcium phosphate, which is insoluble and can interact with proteins through ionic and metal-chelate interactions. This was applied for removing impurities from the cell culture broth.

Flocculation for MAb cell culture harvest and clarification has not been applied very widely and should draw further interest in the coming years given that it can not only be an aid to cell removal but also reduce impurity levels, thus reducing the burden on the downstream chromatographic steps.


Absolute microfilters operated in the dead-end mode with pore sizes ranging from 0.2–1.0 μm can be applied for the removal of cells and cell debris from mammalian cell culture fluids. In practice, it is rare to see such filters used as the sole harvest technique beyond the laboratory scale because the surface area needed for filtration can be prohibitive. For large-scale operations, these filters are commonly used as a terminal polishing step during clarification to ensure the absence of particulates in the load material for the capture chromatographic step.


Expanded-bed adsorption chromatography (EBA) has drawn significant interest over the years because of its potential to eliminate the need for separate harvest and clarification steps.28 –30 In this technique, the cell broth containing cells and cell debris is introduced into a column packed with the EBA resin in the upward flow direction. The flow itself fluidizes the resin beads, causing them to float, thus allowing the cells and cell debris to pass through and exit through the top adaptor. The product adsorbs onto the resin. Following completion of loading, the column is washed and then allowed to settle. Product elution can take place in the downward flow direction.

Some of the important considerations during the development of EBA operations and some of the technique's limitations have been reviewed recently.31 Although this technique has elicited significant interest, practical issues with uniform flow distribution from the bottom of the column as column diameters increase and issues with frit and resin fouling have kept this from being adapted for commercial-scale operations. If these engineering issues are addressed in the future, EBA could perhaps re-emerge as a technique of choice for large-scale mammalian cell culture harvest.


The appropriate combination of harvest and clarification unit operations for mammalian cell culture harvest is obviously scale- and facility-dependent. Preferences also have evolved with increasing experience with scale-up of certain unit operations.

The early 1990s saw significant focus on MF-based harvest strategies because centrifugation was considered a significant capital investment and experience with controlling the shear exerted on mammalian cells was limited. Over time, better centrifuge design and efforts by both leading vendors (Westfalia and Alfa Laval) to optimize operating conditions for biopharmaceutical cell culture broths have resulted in this becoming the predominant harvest technique for cell culture facilities over a 2,000-L scale.

Also, the capital investment in large-scale centrifuges is no longer seen as cost prohibitive in multiproduct production facilities because the operating expenses with membranes for MF are significantly higher and the centrifuges can be easily changed over from one product to another.32 Disk stack centrifuges, on the other hand, are easier to clean and operate in a sanitary fashion than large-scale MF housings are. Bioburden control and cleaning validation also are simpler for centrifuges than for MF systems because the latter have complex flow paths which lead to the possibility of dead zones that can harbor microbial growth.

Unlike MF, which provides a clean filtrate stream, the use of centrifuges does, however, imply the need for a secondary clarification step. This niche is most commonly filled in MAb manufacturing processes by depth filtration. Large-scale depth filtration now can be readily scaled up with process-scale housings and disposable filter modules. Recent work with flocculants and filter aids might further increase the throughput of this step. It is rare to see depth filtration as the sole harvest technique beyond smaller production scales (e.g., a few hundred liters).

Terminal clarification is almost always provided by in-line filtration through microfilters with an absolute pore size rating. These terminal filters ensure a particle-free feedstock for the capture chromatographic step.

Harvest and clarification schemes for MAb production processes today (Figure 5) are the product of much evolution and evaluation carried out over the last 15 years. Harvest techniques for mammalian cell culture systems are now routinely expected to operate with high yields (>98%) and minimal cell disruption. The high titers that can now be achieved in cell culture operations mean that the challenge has now moved further downstream to improve purification throughput. This also implies that the current cell culture scales are likely to stay with us over the next decade. Instead of radical changes to the way harvest and clarification are carried out, improvements are likely to be in the form of gains in efficiency and throughput and improved understanding of existing unit operations.

Figure 5. Common harvest and clarification schemes for large-scale mammalian cell culture harvest


The authors would like to acknowledge many Bristol-Myers Squibb employees in the departments of manufacturing sciences, process sciences, and manufacturing operations at the Syracuse, NY, site for contributing to the development of robust harvest operations for several mammalian cell culture products. We also thank Dr. Steven S. Lee, VP and GM, for supporting this review.

This is an excerpt from the chapter in the forthcoming John Wiley and Sons book, Process Scale Purification of Antibodies, edited by Uwe Gottschalk.

Abhinav A. Shukla is an associate director of Manufacturing Sciences, and Jagannadha Rao Kandula is a process engineer of Manufacturing, both at Bristol-Myers Squibb, Co., East Syracuse, NY, 315.431.7926, abhinav.shukla@bms.com


1. Russell E, Wang A, Rathore AS. Harvest of a therapeutic protein product. In:Shukla AA, Etzel MR, Gadam S, editors. Process scale bioseparations for the biopharmaceutical industry. Florida: CRC Press; 2007. p. 1–58.

2. Belter PA, Cussler EL, Hu WS. Bioseparations: downstream processing for biotechnology. New York: John Wiley & Sons Inc.; 1988.

3. Hutchinson N, Bingham N, Murrell N, Farid S, Hoare M. Shear stess analysis of mammalian cell suspensions for prediction of industrial centrifugation and its verification. Biotechnol Bioeng. 2006;95:483–491.

4. Maybury JP, Mannweiler K, Tichener-Hooker NJ, Hoare M, Dunill P. The performance of a scaled-down industrial disk stack centrifuge with a reduced feed material requirement. Bioprocess Eng. 1998; 18:191–199.

5. Zhao X, Zhou J. Effect of solid ejection size on continuous centrifugation during mammalian cell culture product recovery. ACS National Meeting; 2007 Aug 19–23; Boston, MA.

6. Van Reis R, Leonard LC, Hsu CC, Builder S. Industrial scale harvest of proteins from mammalian cell culture by tangential flow filtration. Biotechnol. Bioeng. 1991;38:413–422.

7. Russotti G, Goklen K. Cross-flow membrane filtration of fermentation broth. Membrane Separations in Biotechnology. New York: Marcel Dekker; 2001. p. 85–159.

8. Van Reis R, Zydney A. Membrane separations in biotechnology. Curr Opin Biotechnol. 2001;12:208–211.

9. Lee SS, Burt A, Russotti G, Buckland B. Microfiltration of recombinant yeast cells using a rotating disk dynamic filtration system. Biotechnol Bioeng. 1995;48:386–400.

10. Parnham CS, Davis RH. Protein recovery from cell debris using rotary and tangential cross-flow microfiltration. Biotechnol Bioeng. 1995;47:155–164.

11. Luque S, Mallubhotla H, Gehlert G, Kuriyel R, Dzengelski S, Pearl S, Belfort G. A new coiled hollow fiber module design for enhanced microfiltration performance in biotechnology. Biotechnol Bioeng. 1999;63:247–257.

12. Davis R. Cross-flow microfiltration with backpulsing. In: Membrane Separations in Biotechnology. New York: Marcel Dekker; 2001. p. 161–188.

13. Belfort G, Davis R, Zydney A. The behavior of suspensions and macromolecular solutions in cross-flow microfiltration. J Membr Sci. 1994;96:1–58.

14. Nagata N, Herouvis K, Dziewulski D, Belfort G. Cross-flow membrane microfiltration of a bacterial fermentation broth. Biotechnol Bioeng. 1989;34:447–466.

15. Fiore JV, Olson WP, Holst SL. Depth filtration. In: Curling, JM, editor. Methods plasma protein fractionation. New York:Academic Press; 1980. p. 239–268.

16. Singhvi R, Schorr C, O'Hara C, Xie L, Wang D. Clarification of animal cell culture process fluids using depth microfiltration. Biopharm. 1996;4:35–41.

17. Badmington F. Prefiltration technology. Meltzer T, Jornitz M, editors. Filtration in the Biopharmaceutical Industry. New York:Marcel Dekke; 1998. p. 783–817.

18. Smith G. Filter aid filtration. In: Meltzer, TH, Jornitz, MW, editors. Filtration in the Biopharmaceutical Industry. New York:Marcel Dekker; 1998. p. 1–69.

19. Knight, R, Ostreicher, E. Charge-modified filter media. In: Meltzer, TH, Jornitz, MW, editors. Filtration in the Biopharmaceutical Industry. New York: Marcel Dekker; 1998. p. 95–125.

20. Gerba C, Hou K. Endotoxin removal by charge-modified filters. Appl Environ Microbiol. 1985; 50:1375–1377.

21. Hou K, Gerba C, Goyal S, Zerda K. Capture of latex beads, bacteria, endotoxin and viruses by charge-modified filters. Appl Environ Microbiol. 1980;40:892 –896.

22. Dorsey N, Eschrich J, Cyr G. The role of charge in the retention of DNA by charged cellulose based depth filters. Biopharm. 1997(1);46–49.

23. Tipton B, Boose JA, Larsen W, Beck J, O'Brien T. Retrovirus and parvovirus clearance from an affinity column product using adsorptive depth filtration. Biopharm. 2002(9):43–50.

24. Yigzaw Y, Tranh M, Piper R, Shukla A. Exploitation of the adsorptive properties of depth filters for host cell protein removal during monoclonal antibody purification. Biotechnol Prog. 2006;22:288–296.

25. Yavorsky D, Blanck R, Lambalot C, Brunkow R. The clarification of bioreactor cell cultures for biopharmaceuticals. Pharm Technol. 2003(3):62–74.

26. Riske F, Schroeder J, Belliveau J, Kang X, Kutzko J, Menon M. The use of chitosan as a flocculant in mammalian cell culture dramatically improves clarification throughput without adversely affecting monoclonal antibody recovery. J Biotechnol. 2007;128:813–823.

27. Coffman J. Flocculation of antibody-producing mammalian cells with precipitating solutions of soluble cations and anions. International Conference on Recovery of Biological Products XII. Abstact # M3; 2006 Apr 2 –7; Litchfield, Arizona.

28. Thommes J, Bader A, Halfar M, Karau A, Kula M. Isolation of mAbs from cell containing hybridoma broth using protein A coated adsorbent in expanded beds. J Chromatogr. 1996;153:111–122.

29. Hjorth R. Expanded-bed adsorption in industrial bioprocessing: recent developments. Tibtech. 1997;15:230–235.

30. Anspach F, Curbelo D, Hartmann R, Garke G, Deckwer WD. Expanded-bed chromatography in primary protein purification. J Chromatogr A. 1999;865:129–144.

31. Sonnenfeld A, Thommes J. Expanded bed adsorption for capture from crude solution. In: Shukla AA, Etzel MR, Gadam S, editors. Process scale bioseparations for the biopharmaceutical industry. Boca Raton (FL): CRC Press; 2007. p. 59–81.

32. Wang A, Lewos R, Rathore A. Comparisons of different options for harvest of a therapeutic protein product from high cell density yeast fermentation broth. Biotechnol Bioeng. 2006;94:91–104.