Predicting Manufacturability During Biopharmaceutical Discovery

Today’s biopharmaceutical products often consist of complex engineered synthetic protein sequences or fusions of sequences that are not normally found together in nature. These product characteristics stand in sharp contrast to the original products developed in our industry in the early 1980’s, where the promise of biotechnology was the ability to recreate naturally-occurring human proteins using the newly developed recombinant DNA technology and to produce these products in sufficient amounts to treat patients whose symptoms were, in effect, caused by deficiencies of such proteins. This new technology could bypass the requirement to isolate essential therapeutic proteins from a limited supply of natural sources. Products such as Factor VII and Factor VIII for hemophilia or insulin for diabetes had already changed the survival profile and quality of life for patients affected by these devastating diseases, but the risks of viral contamination of products isolated from human blood or immunogenicity against proteins isolated from other natural sources made the promise of recombinant production a very attractive alternative. By using the controlled conditions of recombinant protein production, global supply of safe and effective replacement proteins could be provided and patients worldwide could be treated.

True to the promise of developing replacement proteins for human diseases, some of the first products developed and approved by the biotechnology pioneers included recombinant human insulin [1], tissue plasminogen activator for treating strokes and cardiovascular disorders, [2] and ceredase for replacing the missing enzyme in Gaucher’s disease [3]. Hormones and growth factors soon followed, as products such as human growth hormone to treat dwarfism, erythropoietin to treat chemotherapy-induced anemia, and other replacement products emerged [4,5]. The industry quickly turned towards identification of additional naturally occurring products that could be produced with the new technology and could be used for novel medical interventions, such as immune system stimulation with chemokines and cytokines [6].

Following the production and commercialization of the “low-hanging fruit,” that is, the easy-to-identify recombinant human proteins for which there was a medical need and a waiting global market, scientists began to apply new logic to product development. One result was the initial efforts to fuse two functions from two different proteins together. The blockbuster product Enbrel contains the soluble portion of the normally membrane-bound TNF receptor that can bind to TNF and reduce the painful and devastating impact of an overactive inflammatory response due to autoimmunity [7]. The soluble TNF receptor is presented as a fusion protein linked to the Fc region of a human antibody, enabling dimerization to improve function and enabling a much longer biological half-life. Fusion products are still very much an active component of the biotechnology industry, with immunoglobulin Fc regions or human serum albumin frequently used as fusion partners to enhance half-life and bioavailability. Other fusion proteins use a targeting component to bring an enzyme or other factor to the disease site. In a fascinating implementation of this approach, scientists at the Burnham Institute built phage display libraries containing thousands of randomized peptide sequences and then injected these libraries into mice to identify specific peptides that target specific tissues or functions [8]. With this daring approach, Pasqualini et al. discovered peptides that were highly specific for new vasculature, and these peptides were reported just as new vasculature formation, angiogenesis, was being recognized as a requirement for solid tumor growth [9]. The oncology field focused on identifying compounds that could prevent angiogenesis and kill the tumor as it tried to expand without nutrient supply. What could be better than a peptide that could carry a toxic compound right to the new vasculature? Just such a product is in development now by an established biotechnology company that focuses on treatments for cancer [10, 11]. Other approaches to targeting include antibody-drug conjugates, bispecific antibodies, and fusions of naturally occurring ligands to native or engineered effector proteins.

The evolution of biopharmaceutical products from replacement products to natural proteins with new medical uses to fusions of different natural products led eventually to the realization that biopharmaceutical products could be engineered to be better than natural proteins, or in fact to be different from any protein that exists in nature. The first field in which this new paradigm took hold was antibody engineering. In 1984 Kohler and Milstein received the Nobel Prize in Medicine for their pioneering work in the 1970’s on the production of monoclonal antibodies [12]. One of the most significant advantages of this new monoclonal antibody technology over traditional techniques for producing antibodies was the creation of an immortalized cell line creating a continuous source of the same antibody with a single antigen specificity. This enabled the development of highly specific antibodies directed toward a single epitope on the target antigen. Initially, monoclonal antibodies were used as laboratory reagents but their use was quickly adopted as clinical diagnostic reagents and then as therapeutic agents. By 1986 the first monoclonal antibody for human use, Orthoclone OKT3, was approved for prevention of kidney transplant rejection. The low market penetration of this product was due, in part, to immunogenicity of the fully murine antibody [13], so scientists produced chimeric antibodies in which the constant region was from human IgG1 antibodies and the variable region was from the mouse antibody. Highly effective and successful chimeric antibodies include Rituxan, Erbitux, and Remicade [13].

These original protein therapeutics, including chimeric antibodies, were faithfully reproduced as they existed in nature. The ability to produce the products at high levels from recombinant systems was not assessed; process development scientists were expected to diligently work to achieve the best productivity and purity of the products. In today’s environment, the approach to discovery of new therapeutic products relies on creating engineered diversity, protein sequences that often do not exist in nature. Even the variable regions of antibodies that are originally isolated from immunized mice are subsequently engineered using in vitro technologies such as phage display to improve the affinity and efficacy of the antibodies, and often they are “humanized” as well [14].

These manipulations lead to libraries of potential candidates from which the final lead candidate can be selected for development and clinical evaluation. As the industry moved from natural proteins to screening libraries of engineered proteins, the focus of new molecule discovery programs remained on identifying the molecule with the greatest affinity, efficacy, or other desirable properties for a therapeutic product. There was still no or limited consideration of the ability to express the candidate molecules at economically feasible levels by fermentation or cell culture and to purify the products without sacrificing significant yield.

Technology Improvements to Enable Manufacturability Assessment

Traditionally, methods for final candidate selection have tended to be based on properties of the molecules produced directly from the discovery platform, which may not be consistent with properties of the product that will be manufactured in a different cell type and process. For phage display or in vitro discovery methods, final candidate selection has often been performed using antibody fragments such as Fabs or sFvs, with the hope that the whole antibody will perform the same way [15]. When the final candidate is selected from any discovery platform, the genes encoding this antibody must be isolated and transferred to an expression vector which is used to generate a stable production cell line. Antibodies that are selected in vitro or in bacteria are not selected at all for the ability to express at high levels in a mammalian production cell line, which adds significant risk to making the final candidate selection directly from the discovery platform.

The underlying assumption has always been that any molecule could be manufactured so the concept of assessing manufacturability has not been part of the lead candidate selection. In the past couple of years, this assumption has been replaced by a realization that the industry had a golden opportunity to greatly improve the cost of goods, shelf-life, and resource requirements by adding manufacturability assessment to the final selection of lead biopharmaceutical candidates.

Biopharmaceutical products are produced by expressing large levels of essentially foreign proteins in either mammalian cells such as Chinese Hamster Ovary (CHO) or microbial cells such as E. coli. While the concept of manufacturability applies to products produced in either mammalian or microbial systems, this article focuses on timelines, technologies, and criteria for assessing manufacturability in mammalian production systems.

A decade ago, creation of a CHO production cell line that enabled production of sufficient product for clinical trials and, eventually, commercial supply was a lengthy and arduous process that often took more than 12 months of transfection, selection, gene amplification, and determination of genetic stability [16]. Developing production cell lines forten or more lead candidates was completely unfeasible so screening for the ability of stable mammalian cells to produce different products was not performed. In many cases, transient transfection of cell lines such as HEK293 cells, a transformed human cell line that enables a short burst of high expression from transiently transfected genes, is used to generate small amounts of material for final screening and for initial formulation and analytical development, but this material may not be representative of the same protein produced in a stable CHO production cell line. Further, the level of expression in the HEK293 cell is not at all predictive of the potential expression levels in a stable CHO production cell line. HEK293 cells are not suitable for large-scale, high-density fermentation due to their sensitivity to environmental stress, so a clear mismatch between the generation of discovery material and clinical product exists.

Significant pressures on the industry to meet aggressive timelines along with significant changes in the regulatory acceptance of production cell line changes following early clinical trials have led to the emergence of several highly efficient cell line development technologies that enable stable CHO or other mammalian production cell lines to be developed in parallel and in less than four months. These advances mean that companies can now introduce the genes for tens of product candidates into CHO host lines, and can rapidly select production cell lines for all these candidates. The technologies that enable this advancement in product development fall into multiple categories. Some of the earliest approaches to improve the speed of developing high- productivity cell lines were the introduction of sensitive selectable markers that were placed under control of very weak transcriptional promoters [17]. When the desired gene is co-localized on an expression vector, selective pressure for the marker concurrently increases the target gene expression. Other genetic approaches include elements that impact the chromatin structure and allow the proteins that constitute the transcriptional machinery greater access to the transgene of interest [18, 19], elements that improve translation efficiency [20], or improved promoters and enhancers [21]. Vectors with improved transfection efficiency have also been implemented in the generation of mammalian production cell lines with high productivity, and these can enable parallel generation of production cell lines as well [22].

This ability to perform parallel cell line development at the latest stages of lead candidate selection enables a new paradigm in biopharmaceutical discovery, in which final candidate selection is performed using product expressed in the same type of stable mammalian cell lines that will be used for production of clinical material. The advantages of this approach are enormous. Integration of production cell line development into the final stages of discovery is expected to facilitate and streamline the entire development program and should substantially reduce overall development risk.

Key Issues that Define Manufacturability

Manufacturability consists of two main components, productivity and stability. Productivity is the ability to produce the product with the proper structure and post-translational modifications at high levels from a production cell line, in a form that enables purification of the intact product. Some protein sequences do not express or fold properly and some sequences cause extensive aggregation of the product in the cell culture media or following purification [23]. By screening product candidates in stable cell lines to determine their performance in the culture media and in storage, a lead candidate can be selected that avoids these typical issues.

Determining the ability of stable CHO cell lines to produce candidate molecules requires tools to assess total product content in the conditioned media from each candidate’s cell line. These tools have often been developed during the earlier stages of discovery or, if the molecules belong to a common product class such as antibodies, the tools exist as generic methods within the discovery labs. ELISA methods that detect the product specifically within a complex mixture are one approach to quantitating the production; SDS-PAGE is an alternative, less sensitive approach. While productivity is the first line of assessment, there are many additional areas that should be evaluated to determine manufacturability and those proteins that are expressed at somewhat lower levels in the stable CHO lines should not automatically be discarded. Only those candidates whose expression levels are so low that the cell line wouldn’t support production for early clinical development should be eliminated at this stage. The absolute expression level required will be determined by the intended indication, patient population, expected dose and length of treatment. Not all of these factors will be known at the time of lead candidate selection, so a conservative estimate of the requirements for early clinical development should be used to make decisions on lead candidate selection.

Another area that requires evaluation during manufacturability assessment is the biophysical properties and the stability of the candidate products. Even if they can be produced at high level and purified to meet suitable specifications, proteins that are prone to being oxidized, deamidated, aggregated, or that are highly susceptible to fragmentation do not make acceptable drug candidates. Often a lead candidate is selected and used in early clinical development before the issues with long-term storage are identified, and many programs have failed because of the protein’s instability rather than due to any performance issues in the clinic. Formulation development can help mitigate the extent of some of these negative chemical alterations in a drug product, but if an alternative product with the same biological performance and reduced tendency towards degradation is selected during discovery, the challenges of finding a suitable protective formulation are significantly reduced.

Manufacturing assessment clearly requires suitable, well-developed analytical methods that can be performed on many candidate molecules and that can reliably provide useful, quantitative information. These methods should, at a minimum, be capable of measuring total protein content, aggregate formation in the unpurified form found in culture media and in the purified forms if available, proteolysis, oxidation, deamidation, and improper folding. The complexity and diversity of analytical methods available to support evaluation of biopharmaceutical products during discovery and manufacturability assessment as well as throughout development and manufacturing is continuously increasing, and development and execution of analytical methods is an increasingly important core competency for biopharmaceutical companies. Manufacturing assessment demands robust and reliable analytical methods to ensure proper and timely results that can inform the decision making needed to select the lead candidate product from among the many that are under consideration.

Summary

Predicting manufacturability of candidate molecules has become an important component of lead selection for biopharmaceuticals. Emerging and established cell line development technologies enable the creation of stable cell lines or pools that provide more representative material for evaluation, and this evaluation should include productivity as well as biophysical characterization and stability assessments. By incorporating this additional selection criteria at the stage of lead identification, development programs will have reduced risk of failure due to product issues that are related to structure, stability, and ability to produce the molecule. This type of assessment is becoming more common in the biopharmaceutical industry, and this improvement in the screening paradigm has the potential to reduce the cost of valuable and useful products and to improve the timelines that it takes to move these products forward through development and onto the market.

References

1. Ladisch MR and Kohlmann KL “Recombinant human insulin”, Biotechnol. Prog. 1092, 8, 469-478.
   
2. Chien KR, Meidell R, and Gerard R “Tissue plasminogen activator: from molecular biology to myocardial infarction”, Am J Med Sci. 1987 Mar;293(3):201-7.
   
3. Niederau C, vom Dahl S, and Häussinger D. “First long-term results of imiglucerase therapy of type 1 Gaucher disease”, Eur J Med Res. 1998 Feb 21;3(1-2):25-30.
4. Thompson RG, Conforti P, and Holcombe J. “Biosynthetic human growth hormone: current status and future questions”, J Endocrinol Invest. 1989;12(8 Suppl 3):35-9.
   
5. Egrie JC, Strickland TW, Lane J, et al. Characterization and biological effects of recombinant human erythropoietin. Immunobiol. 1986;72:213-224.
   
6. Louie SG, and Jung B. “Clinical effects of biologic response modifiers”, Am J Hosp Pharm. 1993 Jul;50(7 Suppl 3):S10-8.
7. Murray K.M. and Dahl S.L. “Recombinant human tumor necrosis factor receptor (p75) Fc fusion protein (TNFR:Fc) in rheumatoid arthritis”, Ann Pharmacother. 1997 Nov;31(11):1335-8.
   
8. Pasqualini R, Ruoslahti E. “Organ targeting in vivo using phage display peptide libraries”, Nature. 1996 Mar 28;380(6572):364-6.
   
9. Folkman J. “Tumor angiogenesis: a possible control point in tumor growth”, Ann Intern Med. 1975 Jan;82(1):96-100.
   
10. Gregorc V., Zucali P., Ceresoli G. L., Noberasco C., Bajetta E., Santoro A., Viganò M. G., Caligaris-Cappio F., Lambiase A., and Bordignon C. “NGR-hTNF, a novel vascular targeting agent (VTA), as second-line therapy in malignant pleural mesothelioma (MPM): Preliminary results of multicenter phase II study”, Journal of Clinical Oncology, 2008 ASCO Annual Meeting Proceedings (Post-Meeting Edition). Vol 26, No 15S (May 20 Supplement), 2008: 8099.
11. Curnis F, Sacchi A, Corti A. Improving chemotherapeutic drug penetration in tumors by vascular targeting and barrier alteration. J Clin Invest. 2002 Aug;110(4):475-82.
12. Kohler G, Milstein C. Continuous cultures of fused cells secreting antibody of predefined specificity. Nature. 1975 Aug;256(5517):495-7.
13. Mather, J.P., Worsley, S., Fisher, J., and Vehar, G. “A View of Antibody Therapeutics Part 1: From Failure to Market Leader”, BioExecutive International (2005) 42.
    
14. Winter G, et al. Making antibodies by phage display technology. Annu Rev Immunol. 1994 Apr;12:433-55.
    
15. Hoogenboom HR. Selecting and screening recombinant antibody libraries. Nat Biotechnol. 2005 Sep;23(9):1105-16.
16. Schimke RT, et al. Amplification of genes in somatic mammalian cells. Methods Enzymol. 1987;151:85–104.
17. Wilson RH. Glutamine synthetase gene amplification in Chinese hamster ovary cells. In: Kellens RE, editor. Gene amplification in mammalian cells. New York: Marcel Dekker Inc; 1994. p. 301–11.
18. Fisch I. The role of matrix-attachment regions in increasing recombinant protein expression. BioProcess Int. 2007 Feb; 5(2):66–72.
19. Benton T, et al. The use of UCOE vectors in combination with a preadapted serum-free, suspension cell line allows for rapid production of large quantities of protein. Cytotechnology. 2002 Jan;38(1-3):43–6.
    
20. Chappell, SA, Edelman, GM, and Mauro, VP “Biochemical and Functional analysis of a 9-nucleotide RNA sequence that affects translation efficiency in eukaryotic cells”, Proc. Natl. Acad. Sci. USA 2004 9590-9594.
    
21. Running Deer J, Allison DS. High-level expression of proteins in mammalian cells using transcription regulatory sequences from Chinese hamster EF-1alpha gene. Biotechnol Prog. 2004 May- Jun;20(3):880-9.
    
22. Weis P, et al. Flexible methodology for developing mammalian cell lines. Biopharm Int. 2006 Feb;19(2):30-5.
    
23. Wang W. Instability, stabilization, and formulation of liquid protein pharmaceuticals. Int J Pharm. 1999 Aug 20;185(2):129-88.