Boosting Mammalian Cell-line Manufacturing Pilot Plant—A Case Report

More than 37 mammalian cell-derived therapeutics entered the pharmaceutical market since the first monoclonal antibody OKT 3 was approved in 1986.¹ Several hundred proteins including antibodies, growth factors, cytokines, and enzymes are currently in clinical trials. An extensive preclinical pipeline ensures sustainable product flow through clinical trials for future decades. The probability of successfully passing clinical trials is still less than 20 percent for this kind of product, and increasing competition exists among nearly all complex therapeutic proteins and in all indications. Thus flexibility and speed of Phase I and II protein supply at pilot plants are now major issues for the industry. Figure 1 illustrates the current product life cycle for mammalian cell-based proteins, which is often up to 15 years from concept to market. To ensure return on investment and reasonable profit, more than eight years of market exclusivity are required. Unfortunately, patents expire after 20 years. Therefore there is substantial pressure on timelines to obtain market approval. Availability and flexibility of pilot and commercial manufacturing plants play a crucial role in creating timelines, especially for mammalian cell products.

Rene Brecht, Ph.D

This article focuses on the construction and operation of a contract plant designed around the production of recombinant proteins for mammalian cell lines using disposable systems. With growing experience, a manifold increase of customer cell-line productivity could be achieved, boosting extrapolated plant capacities to a level of several kilograms per year. This case report summarizes our experience with the pilot plant over a three-year period and discusses challenges for commercial plants of disposable nature.

DISPOSABLE NATURE PILOT PLANT PROVIDES UNIQUE FLEXIBILITY

In 2000 we initiated a pilot plant project based on thorough analysis of market trends, product life cycles, and state-of-the-art technologies. Most "early" customer cell lines have a very low specific productivity of less than 5 pg therapeutic protein produced per cell per day (pg/c-d). As a result, about 50 percent of the customers are asking for a newly designed, high-yielding cell line. An excess of 25 g net protein are needed for toxicology tests and Phase I trials. High product throughput must be ensured to serve a large number of customers. Therefore short product change-over periods are mandatory. Facility design should also allow manufacturing of investigational viruses and cell-therapy products to broaden the company's contract manufacturing base. Consequently the following specifications were used for facilities design.

Figure 1. Life Cycle of Complex Proteins

Specifications for Facilities Design

• Cell line and process development capacity: 10 to 14 slots/year

• Portable bioreactor equipment with low space requirement

• Closed systems to reduce clean-room loads in upstream suites

• Annual capacity:more than 500 g crude product at 5 pg/c-d cell line productivity

• Flexibility: more than ten investigational medicinal products per year

To meet specifications it was decided to integrate separate cell line and process development units into the manufacturing site, and to entirely rely on bioreactor systems with disposable culture ware.

We completed the construction and validation of the facility within 18 months (Figure 2), thereby complying with the EU guideline 91/356/EWG (current 2003/94 EC) and the German drug law where applicable. 19,000 ft² came into operation with a manufacturing core unit of less that 5,000 ft² (25 percent of space). The remaining space was divided into functional units for cell line development (15 percent), up-and downstream development (20 percent), quality control and product characterization (20 percent), and offices and utilities (20 percent). Expansion space in excess of 6,000 ft² within the building allows for flexibility to expand functional units according to market demands. The manufacturing core unit comprises a cleanroom for preculture and two upstream manufacturing suites. Up to five portable bioreactor units with fully disposable culture ware can be simultaneously operated in those process suites. Welding equipment for sterile connection and disconnection of flexible tubes and installations in cleanroom walls allow for feed media and waste allocation in unclassified areas. The downstream process suite, equipped with purification systems sized for dozens of grams per purification cycle, serves for protein purification from bulk supernatant. A separate cleanroom area is available for final sterile and nanofiltration of bulk drug substance. With an effective quality management system in place, simultaneous manufacturing of different proteins in upstream suites can be performed. In addition air-flow layout, airlock systems, pressure regimen, and cleanroom classifications within the manufacturing unit allow the processing of investigational medicinal viruses and cell-therapy products.

Figure 2. Pilot Plant Realization Timelines at ProBioGen's Berlin Manufacturing Site for Investigational Medicinal Products (Drug Bulk Substances) Derived from Mammalian Cells

Advantages of the facility can be summarized as the following:

• Lack of media preparation areas

• Lack of centralized steam and WFI production

• Segregation of media and waste from process areas

• Portable, closed bioreactor systems operating in batch or perfusion modes

• Segregation of process personnel and material flow

• Simultaneous protein manufacturing in upstream suites

• Significantly reduced validation requirements

• Fast change over

DISPOSABLE BIOREACTORS SUPPORT TAILOR-MADE PROCESSES

A growing number of disposable- based culture systems are commercially available worldwide. Conventional disposable based culture systems such as roller bottles

²

and cell factories

³

are in use for low-volume, high-potency proteins such as erythropoetin and INF beta. Having been developed and explored in research application, most of these disposable bioreactor systems have deficiencies in industrial hardware design, reliable process control standards, and human interface and software performance. Two types of disposable-based bioreactors in use for investigational medicinal protein manufacturing are the Wave and Biovest's Acusyst hollow-fiber bioreactor families.

^4,5

Flexibility of the pilot plant and portability of the systems allow ProBioGen to apply both to customer processes, thereby providing both perfusion and batch modes of operation. Cell-line poductivity, cell growth characteristics, complexity of proteins, sensitivity to proteases, product titers in crude bulk, process economics, and other parameters contribute to the choice of culture mode and system.

The scalability of Acusyst hollow-fiber bioreactors within the reactor family is based on increasing numbers of a standard hollow-fiber cartridge in each reactor system. A specific cycling pressure regimen, differentiating the system from conventional hollow-fiber bioreactors creates active media perfusion through the cell culture space of the cartridge. Cells and proteins are separated in such a cartridge from the media stream by hollow fibers with a membrane cut of lower than 10 kD. Through this process media perfusion and harvest are separated from each other.

The AcuSyst hollow-fiber bioreactor family comprises four bioreactor systems. They support operation of 1, 2, 6, 10, 12, or 20 hollow-fiber cartridges per run and system. The AcuSystXCELL, with a maximum 12 cartridges, was investigated for GMP manufacturing at the pilot plant (Figure 3). System-specific process development parameters include the media feeding rate, cartridge perfusion rate, harvest rate, and a pressure cycling regimen. Technical details are described elsewhere.⁶ Proper process development allows defining parameters for steady state manufacturing over periods of 30 to 90 days. In contrast to other systems, media perfusion is separated from cell culture space by membranes with a cut off of <10 kD. Proteins are retained in the cell culture space and can be harvested independently from media perfusion. Harvest rates of 0.5 to 2 cell culture volumes per day are common and significantly reduce protein residence time in contrast to batch or fed-batch mode. Thus high titers of up to 1 g/L for low expression cell lines of 5 pg/c-d productivity can be generated in the cell culture space of a cartridge. This high protein concentration in crude bulk significantly reduces processing time and product losses downstream.

Table 1. AcuSystXCELL Raw Bulk and Net Capacities Extrapolated From Pilot Runs of Three Different Products

These findings are also reported for other types of hollow-fiber bioreactors.⁷ We observed a reliable correlation of specific cell-line productivity in standard T flasks and yields in a continuous perfusion bioreactor. Table 1 provides the extrapolated raw bulk and net capacities of the AcuSystXCELL bioreactor for three different products. The processes were developed at pilot scales. Crude and net product derived from these runs after purification was extrapolated to the capacity of the 12-cartridge system AcuSystXCELL run by multiplication with the respective scaling factor. Correlation allows one to roughly calculate the ranges of system capacity on the basis of the specific productivity of customer cell lines and the batch duration. Short product residence time, high titers, and process economy make the system well suited for the manufacturing of sensitive, complex, low-volume, high-potency glycoproteins. The importance of short residence times was addressed several times in the literature.⁸

Figure 3. Biovest AcuSystXCELL – a Pressure-Driven, Hollow Bioreactor Equipped with 12 Cartridges in Operation in a Cleanroom Setting

The Wave system is a bag-based bioreactor family for batch and repeated batch modes of operation. Seven bioreactor scales with net culture volume of 1, 5, 10, 25, 100, 250, and 500 L are available. Actually, the largest scale employed at ProBioGen is the Wave Bioreactor System200, with a nominal culture volume of 100 L (Figure 4). Filling level, rocking rate, and angle are system-specific process development parameters. The system is especially useful for cell lines secreting significant amounts at substrate limitations. Products resistant to proteases and glycosidases are preferred to obtain reasonable final titers in the system.

Figure 4. Portable WAVE BIOREACTOR System200 of 100-L Working Volume in Operation at a Process Development Lab

The use of Acusyst hollow-fiber systems in manufacturing of biopharmaceuticals is still limited. The remaining challenges for widespread use include:

• Sophisticated handling, requires dedicated personnel with long experiences

• Consistency of cell harvest over long periods.

As far as we know, the use of Wave bioreactors for biopharmaceutical protein manufacturing is still limited to investigational medicinal product manufacturing and inoculum production for approved products. The remaining challenges for increasing use in this area are:

• Complexity of linear process scale-up due to different bag geometries

• Limited flexibility to develop high performance fed-batch processes (filling level optimum).

SUPPORTING LATER MANUFACTURING CHANGES

The ability for later changes in manufacturing equipment and processes is a prerequisite for most customers. To meet that need, early product characterization efforts for investigational medicinal proteins were significantly increased at the pilot plant. A dedicated process science team was established. For complex, sensitive products, a wide pattern of analysis including in-process glycosylation analyses and bioactivity tests was agreed upon with the customers. Product-specific databases were produced to characterize final, purified product and to control process consistency in each type of system and culture mode.

Unique flexibility, speed of material supply, and significant savings in QC and QA efforts at the pilot plant far outweigh early increased product characterization activities and later reference comparability costs, if manufacturing changes in later phases of product development cycle are necessary.

HIGH-EXPRESSION CELL LINES BOOST PLANT CAPACITIES

One of the major obstacles for protein manufacturing in the early stages of the product life cycle is the low expression rate of early research cell lines of usually less than 5 pg/c-d in standard cell culture. Experienced teams for cell line and process development were integrated into the company's Berlin pilot plant to develop and realize strategies for fast expression enhancement.

Table 2. Features of the Disposable-Based Bioreactor Systems

At least three major issues contribute to cell specific and volumetric productivity in final mammalian manufacturing processes. Figure 5 schematically summarizes the contribution of each issue to the cell-specific productivity in the final manufacturing processes.

The issues are:

• Molecular engineering of the manufacturing cell line (bars 1, 2, and 3)

• Initial media selection

• Process optimization in bioreactors including media optimization.

Over the last 15 years, validated fed-batch processes became the gold standard for antibody production. Volumetric productivities of more than 100 mg/L-d were described for individual proprietary fed-batch processes in the late 90s.^9,10 More than 80 mg/L-d could be reached in generic fed-batch processes by 2000.¹¹ Corresponding cell-specific productivities referred to in different sources over the years constantly grew up to ranges of 20 to 60 pg/c-d. Key industrial players such as Amgen, Genentech, Lonza, Biogen IDEC, PDL, and Sigma-Aldrich Biotechnology share the latest developments in this area at meetings such as the European Society of Animal Cell Culture Technology (ESCAT).^12-17 Significantly lower cell-specific productivities were observed for other human proteins such as erythropoietin.¹⁸ Efforts in therapeutic protein production in cultivated mammalian cells were reviewed in 2004.¹⁹

Figure 5. Issues of Cell Specific Productivity in Mammalian Cell Culture

The cell-line development team at our pilot plant concentrated on engineering production-quality cell lines based on optimization and adaptation of the starting cell line; vector design including a sequential selection strategy, complex promoters and translation enhancement, as well as elements stabilizing expression. Care was taken to transfer the complete process to a serum-free environment. Moreover, we developed technologies for targeted gene insertion of Chinese hamster ovary (CHO) cells as well as for a human and mouse heterohybridoma. Based on a proprietary strategy, the IgH locus of heterohybridoma known to account for stable, high-level expression was used to express a protein of choice.

In addition to the work done with well-established cell lines, a designer cell program was incorporated to develop new cell lines of human animal or avian origin from scratch, using different tissues of the respective species. ProBioGen is confident that new cell lines based on rational design will be able to outcompete traditional cell lines and primary cells in the production of recombinant proteins and virus vaccine strains. Benchmark results of our cell engineering effort are shown in Table 3.

Table 3. Proven Strategies for Development of High-Yielding Cell Lines

Methods in Brief for the Three Referred Cell Development Programs

(1) CHO DHFR-cells from a qualified source were electroporated with the ProBioGen high-level expression vector containing the transgene (a human therapeutic glycoprotein) under the control of a proprietary hybrid promoter and flanked by two independently expressed selection markers. Clone pools were derived employing a dual-selection strategy. Highest producers were established in a small-scale screening (30 clones). The complete process — from selection of highly productive cell pools, to cloning of single cells, to selection of a high producer cell line — was performed in ADCF basic medium with GMP-compliant protein supplements. The secreted protein was measured in a specific sandwich ELISA and calculated on a per cell and day basis.

(2) The human mouse heteromyeloma CB03, secreting large amounts of IgM, was transfected by electroporation with a targeting vector containing the leptin Fc-fusion protein gene. The vector was designed to allow for homologous recombination within the Igµ locus. Through targeting, the endogenous Ig promoter was replaced with a CMV-EF1alpha promoter. Homo-logous recombinants were detected by the absence of Igµ expression and the presence of the fusion protein, verified by PCR. After recloning and identifying a stable, high-producer, the secreted fusion protein was measured in a specific sandwich ELISA test in a stationary culture.

(3) A human neuronal designer cell line, derived from primary cells in a fully documented process, was transfected with Effectene (Qiagen, Germany) for stable expression of AAT from a proprietary hybrid promoter. A standard selection marker (puromycin) in an independent transcription unit was used to screen for a small number of high-producer clones. The product, which secreted over 72 h from stationary culture in T flasks, was quantified by a specific sandwich ELISA.

Once 40 pg/c-d can be reached for a certain protein in standard cell culture in CHO, media selection and process optimization is expected to increase volumetric productivities in excess of optimum-performing, fed-batch processes. The future potential of designer cell lines derived from scratch is illustrated by the results obtained so far. A comprehensive summary of current experiences of the cell-development team is described in detail elsewhere.⁶

PILOT-PLANT CAPACITIES AT A GLANCE

Figure 6 illustrates the boosting of plant capacities from 650 g of crude product (without purification) for low-expression cell lines to more than 5 kg crude protein per year. We assumed the use of cell lines with cell specific productivities of 40 pg/c-d for batch manufacturing.

Figure 6. Comparison of Annual Plant Capacities Extrapolated for Low-and High-Yielding Cell Lines

Rational approaches for cell-line engineering, combined with a concise media selection program and straight process development efforts, significantly increase the probability of generating such high-yielding cell lines. In most cases, dozens of grams of clinical-grade protein can be provided for Phase I clinical trials 15 months after cell transfection. Growing the throughput of different types of proteins through the facility by applying a combination of different approaches described earlier increases the predictability of protein yields per batch. High flexibility, advanced process science, tailor-made processes, and significant capacities are the major performance criteria of the plant.

Table 4. Experiences with Different Proteins at Berlin Pilot Plant

CHALLENGES FOR PLANTS WITH DISPOSABLE BIOREACTORS

Various disposable bioreactor systems are currently on the market. We have summarized our experiences with two of them. The major obstacles both wave and hollow-fiber systems face are:

• Limitation in scale

• Limited supplier sources

• Limited experiences in facility design and operation.

The AcuSyst hollow-fiber bioreactor family is limited by its largest system (AcuSyst XCELLERATOR) to 20 simultaneous cartridges in continuous perfusion. The WAVE family is limited to 500 L net culture volume per batch. The basic design of a commercial plant, equipped with single-use technology, revealed low capital expenditures. This is in accord with independent estimates for a biomanufacturing modular concept facility combined with single-use technology that was presented by Stedim and Applicon Corporations at bioLOGICS 2004 in Geneva,²⁰ and with other reports in the literature.^{21, 22}

Validated processes are the prerequisites for construction and validation of such a commercial plant. Demands for low-volume, highly potent proteins of a few kilograms net product per year will favor single-use technology.

Within the next 10 to 15 years, market trends will lead to a growing demand for highly flexible, multi-product plants that can produce an average annual capacity of 10 to 100 kg of purified mammalian cell-expressed protein. A new generation of large-scale, disposable, linear- scalable, high-performance bioreactors will meet the industry's growing need for fully disposable manufacturing plants.

Rene Brecht, Ph.D., vice president process science and manufacturing, ProBioGen AG, Goethestrasse 54 13086 Berlin, Germany 49.30.924.006.40, fax 49.30.924.006.19 rene.brecht@probiogen.de

Volker Sandig, Ph.D., volker.sandig@probiogen.de

Susann Koch, susann.koch@probiogen.de

Uwe Marx, Ph.D., uwe.marx@probiogen.de

Marco Riedel, Ph.D., marco.riedel@probiogen.de.

REFERENCES

1. Melmer G, Biopharmaceuticals and the industrial environment. In. G. Gelissen: Production of recombinant proteins. Wiley-VCH, 2005; 361-383.

2. www.synthecon.com.

3. www.nuncbrand.com.

4. www.wavebiotech.com.

5. www.biovest.com.

6. Sandig V, Rose Th, Winkler K, Brecht R: Mammalian cells. In. G. Gelissen: Production of recombinant proteins. Wiley-VCH, 2005; 233-252.

7. Nagel A, Koch S, Valley U, Emmrich F, Marx U: Membrane based cell culture systems – an alternative to in vivo production of monoclonal antibodies. Dev. Biol. Stand., Karger, 1999; 101:57-64.

8. Yalcin E, Kloth C, Buchholz R, Emmrich F, Harnisch J, Lemke U, Gerlach J, Marx U: Increased production yields of native interferon-gamma in membrane bioreactors by continuous harvest. 2001, proceedings of ESACT meeting.

9. Xie LZ, Wang DIC, High cell – density and monoclonal antibody production through medium design and rational control in a bioreactor. Biotechnol Bioeng. 1996; 51:725-729.

10. Zhou WC, Chen CC, Buckland B, Aunins J: Fed batch culture of recombinant NS0 myeloma cells with high monoclonal antibody production. Biotech Bioeng. 1997; 783-792.

11. Sauer PW, Burky JE, Wesson MC, Sternard HD, Qu LA. A High-Yielding, Generic Fed-Batch Cell Culture Process for Production of Recombinant Antibodies, Biotechnol Bioeng. 2000 March 5; 67(5):585-97.

12. Davies J, Reff M. Chromosome localization and gene-copy-number quantification of three random integrations in Chinese-hamster ovary cells and their amplified cell lines using fluorescence in situ hybridization; Biotechnol. Appl. Biochem. (2001); 33:99-105.

13. Sauer P. et al. Optimization of a fed-batch process producing humanized antibodies: increasing product. PDL, Poster presentation ESACT Congress, 2003.

14. Rendall M. et al., Transfection to manufacturing: reducing timelines for high yielding GS-CHO processes, Lonza, poster presentation, ESACT Congress, 2003.

15. Deeds ZW, et al. Creating a new medium to help meet the variable nutritional requirements of Chinese hamster ovary (CHO) cell clones, Sigma-Aldrich Biotechnology, poster presentation, ESACT Congress, 2003.

16. Sautter K, Fieder J, Otto R, Enenkel B, HTS based development of high producing CHO cell lines, Boehringer Ingelheim, poster presentation, ESACT Congress, 2003.

17. Talabardon M, Ong M, Rahmati S, Chang DYH, Noe W. Phase I process development for monoclonal antibody production. IDEC Pharmaceuticals, poster presentation, ESACT Congress, 2003.

18. Wang M-D., Yang, M, Huzel, N, Butler B. Erythropoietin production from CHO cells grown by continuous culture in a fluidized-bed bioreactor. Biotechnology and Bioengineering, 2002; 77(2):194-203.

19. Wurm, F. Production of recombinant protein therapeutics in cultivated mammalian cells. Nature Biotechnology, 2004; 22(11):1393-1397.

20. Monge M, Hammarberg B. The Biomanufacturing Concept Facility. The Modular approach combined with single – use technology. bioLOGIC, 2004; Geneva 22-24 June (combined oral presentation of Stedim and Applicon).

21. Sinclair A, Monge M. Quantitative economic evaluation of single use disposables in bioprocessing. Pharmaceutical Engineering 2002; 22 (3):20-34.

22. Hodge G. Disposable Components enable a new approach to biopharmaceutical manufacturing, BioPharm International 2004; 17 (3):38-49.