Skip Navigation

Alternatives to Monoclonal Antibody Production (Proceedings)

Comparison Of Hollow Fiber Bioreactors And Modular Minifermentors For The Production Of Large Numbers Of Monoclonal Antibodies In Vitro

Ella Kagan, Eric Vieira, & Howard T. Petrie
Monoclonal Antibody Core Facility
 Memorial Sloan-Kettering Cancer Center
 Box 341, 1275 York Avenue, New York, NY 10021.

The main function of the Monoclonal Antibody Core Facility at Memorial Sloan-Kettering is to provide interested parties with an inexpensive and ready supply of monoclonal antibodies. Since the facility's inception in 1993, production has been performed in vitro, using various bioreactors. Initially, a hollow fiber system (10 ml nominal production volume) was used. A summary of production data from 17 clones showed an average production rate of 3.2 mg per day, at an average purity of 36% and an average concentration of 2.0 mg per ml. Subsequently, preliminary evaluation of a modular minifermentor (30 ml nominal production volume) suggested that this type of bioreactor might have several advantages over the hollow fiber system. Data from 15 additional clones grown in the minifermentor showed an average production rate of approximately 5.7 mg per day, with an average purity of 65% and an average concentration of 2.9 mg per ml. Since the protocols used during these runs had been periodically modified and improved, and since the hybridoma clones produced were not an intersecting set, a controlled comparison was recently undertaken. Four clones were simultaneously inoculated into both types of bioreactors, and the production of antibody, along with numerous related parameters, was continuously monitored for 30-50 days under virtually identical conditions. In these experiments, the hollow fiber bioreactor produced antibody at an average rate of 6.0 mg per day, with an average purity of 76% and an average concentration of 4.6 mg per ml. Production in the modular minifermentor was at an average of 7.4 mg per day, with an average purity of 61% and an average concentration of 1.7 mg per ml. However, by virtue of an accelerated peak of production, the total antibody produced during any given period was greater in the modular minifermentor than in the hollow fiber bioreactor in 3 out of the 4 cases. Together with the lower initial cost of the disposable bioreactor, this accelerated production rate resulted in substantially lower average costs for antibodies produced in the modular minifermentor. However, the increased concentration and purity resulting from hollow fiber antibody production may nonetheless make it the method of choice for applications where subsequent purification or modification of antibody is an issue.

While most antibody-secreting clones have produced antibody faithfully as described above, approximately 12-15% of the clones inoculated in this facility have either produced antibody poorly, or did not produce antibody at all. Of these cases, two primary behaviors are observed. In the first (representing approximately 8% of all clones), no appreciable antibody production is detected from the outset of culture. In the second, production is initially observed at low to moderate levels, but subsequently wanes to nothing over a period of days to a few weeks. In some cases, antibody production in vivo as ascites fluid has been documented from the same starting preparation of cryopreserved hybridoma cells. These findings suggest that not all hybridomas may be adaptable and/or appropriate for antibody production in vitro. Methods to adapt these clones for in vitro production are currently being sought.


A monoclonal antibody core facility was established at Memorial Sloan-Kettering in order to provide researchers with an efficient source of monoclonal antibodies. Requests for antibody production in this Facility generally range from a few milligrams to one gram. Through the adaptation of commercially-available equipment for production on this scale, we have been able to successfully produce antibodies in vitro in 87% of cases (48/55 clones tested), without deviation from a standard method. Two different types of bioreactors have been used, a hollow-fiber system, and a recently devised modular minifermenter. While both types of bioreactors have been successful in the production of monoclonal antibody, each has distinct advantages that may make them more or less suitable for individual application. The results from each type of production, and their relative advantages are discussed.


Antibody quantities on a scale of >1 mg are frequently required by many research laboratories. The costs of such purchases are frequently prohibitive; hence, many research laboratories turn to independent production for antibody requirements on this scale. Common methodologies for such production are static culture, roller bottle culture, or ascites fluid. The former methods generally result in antibody concentrations that are too low for many purposes, while the latter has disadvantages associated with bioactive contaminants (cytokines, endogenous antibodies) and animal welfare. Both methods suffer from a relatively low purity of unprocessed product.

In the past few years, several methods for improved production of antibodies for research purposes have evolved. The best known of these is the hollow-fiber type of bioreactor cartridge. In these types of systems, cells and antibodies are trapped in a 10-12 ml cylindrical cartridge, while nutrients, gases, and waste products are selectively exchanged across the pores of numerous hollow fibers which traverse the cartridge. Medium for this exchange is contained in a (typically) 500 ml bottle, and the contents recirculate through the lumen of the fibers.

Another type of bioreactor which has recently emerged is also applicable to this type of production. This latter bioreactor is referred to as a modular minifermenter. In this system, cells and antibodies are contained in a disc-shaped cylinder (nominal volume 35 ml), which is separated from the adjacent nutrient cylinder (volume ca. 500 ml) by a selective membrane. The entire module rotates around the axis of the adjacent cylinders on an apparatus similar to the ones used for standard roller culture.

We have recently completed a side-by-side comparison of these two systems for antibody production using four random antibody-secreting hybridoma cell lines. We find that each has distinct advantages that make them more or less useful for different purposes. Production in some cases rivals that of ascites, although overall, production in these bioreactors is 2-6 times more expensive. However, this is offset by the higher purity (average 70-75% without processing) and absence of bioactive contaminants. The results of this comparison are described.

Materials and methods

Hollow-fiber bioreactors were purchased from Cellco (Germantown, MD). Modular minifermenters were purchased from Hereaus (S. Plainfield, NJ). Medium was GIBCO Hybridoma serum-free medium (H-SFM) supplemented with 0.5% fetal bovine serum (FBS; GIBCO). Bioreactors were maintained essentially as recommended by the manufacturer, with some modification for the purposes of scale and for the purposes of monitoring. Cell number and viability was monitored by Trypan blue dye exclusion and hemacytometer counting. Glucose concentrations were monitored using standard blood test strips. Antibody purity was monitored by SDS-PAGE electrophoresis followed by Coomassie staining and scanning densitometric analysis. Protein confirmation was confirmed by a Bradford-based colorimetric assay (Bio-Rad).


Of the four clones tested, all produced antibody at appreciable levels throughout the culture period (30-50 days). Cell viabilities were constant throughout the culture period at approximately 70% viable. Average cell density was approximately 4 x 107 cells/ml in the minifermenter versus 6 x 107 cells/ml in the hollow fiber system. Average antibody concentration was approximately 1.7 mg/ml in the minifermenter versus 4.6 mg/ml in the hollow fiber system. Average purity was 61% in the minifermenter versus 76% in the hollow fiber system. Average daily production was 7.4 mg in the minifermenter versus 6.0 mg in the hollow fiber system.


The results of the paired comparison suggest that both systems are adaptable to research-scale production of monoclonal antibodies in vitro. Both systems were relatively simple to maintain, requiring an average of approximately 15 minutes of hands-on time each day. The primary advantage of the hollow fiber systems is the purity and concentration of antibody produced , which suggest that this system may be more suitable for production where downstream processing (e.G., purification) is an issue. The main advantage of the minifermenter is cost, which is approximately 1/3 of the cost of production in the hollow fiber system in our hands. This increase cost is mainly related to the initial cost of the disposable bioreactor component. The other advantage of the minifermenter is that production peaks much earlier, at approximately day 8-10 of culture, versus day 15 in the hollow fiber system.

A summary of data from a much larger panel of hybridomas (55 produced in the last two years) corroborates our finding that antibody production in vitro is practical and affordable in both types of systems. It is important to point out that approximately 13% of the clones submitted for production (7/55) did not produce antibody well in one or the other of these systems (or in two cases, in either system). It is not apparent at this point whether this is a liability of the systems, or whether the clones were simply not good producers; parallel production as ascites was not performed. However, it is also important to point out that our facility, for reasons of practicality, does not deviate from a fixed protocol in the production of antibodies in these systems. It is likely that in many of these cases, production could be rescued through subcloning, or by modification of the protocols used.

Our costs for antibody production in the least expensive system (minifermenter) are approximately twice that of production as ascites ( at this Institution) in similar quantities. However, the elevated purity in the in vitro system (65% in a total of 15 clones produced) versus that of ascites, as well as the absence of bioactive contaminants such as inflammatory cytokines or endogenous antibodies, easily balances the difference in costs in the hands of our users.

New ALTEX: 3/2017
ALTEX 3/2017 cover

Support ALTWEB, Make a Gift
Online Humane Science Course


FRAME Training School in Experimental Design and Statistical Analysis
May 31-June 2, 2017
Saskatoon, Canada

Advances in Cell and Tissue Culture 2017
May 22-24, 2017
Manchester, UK

CAAT Academy
In Vitro Skin and Eye Models Hands-on Training
June 1-2 2017
Brussels, Belgium

CAAT Academy
Tools for Read-Across
June 15-16, 2017
Helsinki, Finland

ecopa SSCT Workshop 2017
June 14-16, 2017
Helsinki, Finland
Thomas Hartung will deliver a keynote address

CAAT Academy
Updates on Hepatotoxicity AOP Landscape and on the ADME Field - Season 2
June 22-23, 2017
Rennes, France

9th Euro-Global Summit on Toxicology and Applied Pharmacology
June 22-24, 2017
Paris, France
Email for Details

6th Annual Meeting of the American Society for Cellular and Computational Toxicology (ASCCT)
September 21-22
College Park, MD

10th World Congress on Alternatives and Animal Use in the Life Sciences
August 20-24, 2017
Seattle, Washington

CAAT Academy
Hands-on Training in Quantitative Human Cell and Effect-based In Vitro Bioanalysis for Assessing Endocrine-disrupting Compounds (EDCs)
September 7-8. 2017

More Meetings...