Development and Manufacture of Therapeutic Bispecific Antibodies

Glycosylated bispecific immunoglobulin antibodies — heavy chains in green and pink and light chains in blue and yellow — engineered to target two different antigens. (HTTPS://STOCK.ADOBE.COM)

To meet the ongoing need for new and improved drugs, the biopharmaceutical community strives to create molecules with new functions. Bispecific antibodies (bsAbs), which can simultaneously home in on two different targets, illustrate the scientific ingenuity needed for this task. The basic proof of concept for these complex molecules was established in 1960 (1), and their application to the redirection of effector cells was reported in the mid-1980s (2–4), but producing them has proved to be challenging.

Many technical advances, both large and small, were needed before bsAbs could be prepared in quantities that would be sufficient for their clinical development. Key among these advances was the invention of the single-chain variable fragment (scFv) and the creation of methods that could enable antibody protein chains to form two different target binding sites rather than two identical ones, as occurs in naturally produced antibodies (5). A T-cell–redirecting bsAb composed of two different scFvs was produced in the mid-1990s (6). The “knobs-into-holes” method of engineering antibody domains for heavy chain heterodimerization debuted about the same time (7). Bispecifics were constructed with two different heavy chains and two identical light chains in 1998 (8). Building on those early advances, antibody engineers since have designed more than 100 types of bsAbs (9–11).

To date, four bsAbs have been granted marketing approvals. Early scientific discoveries relating to scFvs and T-cell redirection served as a foundation for development of the first two approved products, Removab (catumaxomab) and Blincyto (blinatumomab). Catumaxomab was first approved in Europe for malignant ascites in 2009 but subsequently withdrawn for commercial reasons. By contrast, the 2014 approval of blinatumomab, which comprises a tandem scFv targeting CD19 and CD3 on T cells, sparked substantial interest in the T-cell redirection approach. The more recently approved bsAbs — Hemlibra (emicizumab) and Rybrevant (amivantamab) — rely on different designs. The former’s bispecificity derives from the common light-chain approach; the latter was constructed using Genmab’s proprietary DuoBody technology. By that method, two antibodies (each with differing specificity) are expressed individually, then purified and mixed together under conditions that favor formation of a bispecific molecule.

Clinical Pipeline
Since 2000, more than 200 bsAbs sponsored by commercial entities have been evaluated in clinical studies, although most of them have entered clinical study only relatively recently. That is not surprising given the industry’s relative inexperience with methods available early on. BsAbs entered clinical study at a rate of only about one per year during the 2000s, increasing to about six per year in the first half of the 2010s. Successes and expansion in construction and production methods, however, have enabled remarkable growth in the pipeline since then. In the past six years, an average of 27 bsAbs have entered clinical study each year. As a consequence of that growth, the commercial clinical pipeline now includes ~160 bsAb-based therapeutics. Most are in early stage development, with only about a dozen in late-stage clinical studies).

The commercial clinical pipeline of bsAbs is focused primarily on cancer treatments. Of the ~160 bispecifics in clinical studies, about 90% are undergoing evaluation for one or more types of cancer. Two approaches dominate: T-cell engagement and immune-checkpoint modulation. They respectively make up ~47% and ~28% of the anti-cancer bispecifics currently in clinical trials. T-cell engaging bispecifics such as blinatumomab target a tumor-associated antigen (TAA) and CD3 on T cells, thereby bringing cytotoxic T cells into proximity with tumor cells. Of the bispecifics that target CD3, the two most frequent TAA targets are B-cell maturation antigen (BCMA) and CD20. Of the bispecific immune-checkpoint modulators, common targets include PD-1 and PD-L1, CTLA-4, 4-1BB, OX40, TIM3, TIGIT, LAG-3, and CD47.

Manufacturing Requirements
Development of suitable biomanufacturing processes is a core element of commercializing new bsAb candidates. Companies defining processes for their bispecific candidates invariably want to maximize productivity using expression systems and upstream processes that generate high titers coupled with efficient purification processes that maximize yield while providing required purity. Such processes must be robust and meet the same regulatory requirements as other biological medicines do.
Developers of bsAbs have noted that production scales for these molecules often match those of standard monoclonal antibodies (MAbs) (13). We find that bsAb sponsors generally prefer manufacturing early phase clinical material at smaller scales than are common for conventional MAb production because of higher potencies that provide for lower dosing requirements. Nevertheless, companies could scale up their processes to large cell-culture production facilities for late-phase clinical studies and commercialization. Scalability of the production processes for next-generation antibodies is as important as it is for standard MAbs.

Platform manufacturing processes often are highly desirable for companies with pipelines that include multiple bispecific candidates. Such approaches enable rapid early stage process development; provide time and cost efficiencies; reduce raw-material and consumables lead-times; and facilitate process transfers, analytical development, and regulatory filings. A platform approach typically is adopted when candidates are structurally similar, but it also can be applied to disparate molecules. For example, Bristol-Myers Squibb (BMS) implemented a platform for bsAbs despite the candidates’ having different structures and formats (14).

Expression Systems
Single-chain variable fragments have been expressed in bacterial, yeast, mammalian-cell, insect-cell (baculovirus based), plant-cell, and cell-free expression systems. Escherichia coli is a common choice for expressing scFv proteins because bacteria grow quickly on inexpensive media and can produce proteins in large quantities. However, expressed scFv molecules can be misfolded or form inclusion bodies that require additional downstream processing steps to solubilize and refold proteins.

Using mammalian cell lines such as Chinese hamster ovary (CHO) cells for scFv production can overcome those problems because of the advanced internal protein-folding processes in eukaryotic cells and their ability to perform complex posttranslational modifications. Often used for biomanufacturing of MAbs and many other proteins, highly scalable CHO cell cultures can express proteins stably and at sufficiently high titers to compete with prokaryotic expression systems (12). Mammalian cell lines grow more slowly than prokaryotes do, however, making production take longer.

IgG-like bsAbs are expressed predominantly in mammalian cell lines for the same reasons. However, a manufacturing challenge arises in ensuring the correct assembly of antibody fragments into IgG-like bispecifics. Four different polypeptide chains can be combined in 16 possible ways, only two of which (~12.5% of what is expressed) have a desired asymmetric heterodimeric structure, with the rest (~87.5%) remaining as impurities (12).

That problem has been evident particularly with the quadroma technology used to produce some of the first bsAbs. Quadromas are created from the fusion of two hybridoma cell lines that express monospecific, bivalent antibodies. The resulting cells produce heavy and light chains from both parent lines. The molecules assemble randomly, so only a small proportion of them form the desired bsAbs. The others become product-related variants, including free heavy chains, light chains, homodimers, half molecules, and mispaired antibodies that are impurities to remove through a complex purification process. The combination of poor expression levels for desired antibody structures and intensive purification schemes ultimately result in low overall process productivities (10).

Protein/Cell Engineering Improves Manufacturability
To address that challenge, protein engineers developed a number of different strategies for designing bsAb molecules with improved manufacturability characteristics. Fc heterodimerization techniques ensure that asymmetric molecules are produced through a combination of complementary heavy chains, which substantially reduces the number of combinatorial possibilities (16). The knobs-into-holes technology enables Fc heterodimerization and reportedly improves antibody stability, making proteins with improved tolerance for the types of pH conditions that fluctuate during purification by protein A chromatography (17). CrossMab technology (15) and the common light-chain method are protein-engineering approaches that reduce further the number of structural combinations produced by ensuring correct pairings of light chain and heavy chains (16).

Such methods have enabled biopharmaceutical companies to use production schemes for bispecific candidates that are very similar to those used to make MAbs. For example, Merus has manufactured a bispecific, humanized IgG1 antibody (zenocutuzomab) at 2,000-L scale using a DG44 CHO cell line with a cell culture process that took 14 days in a production bioreactor. A familiar downstream process for MAbs was used to purify the bulk drug substance, with a protein A capture step followed by low-pH hold for virus inactivation, flow-through anion-exchange chromatography, and a final cation-exchange chromatography step to remove minor residual homodimers and half-IgG contaminants. Eluate was filtered through a 20-nm virus-removal filter before solution concentration and drug-substance formulation to 20 g/L. The Merus process provided bioreactor titers in the range of 1.0–1.5 g/L and downstream yields of 60–75% (16). Although such process performance is not especially high by MAb standards, it would be considered acceptable for many such products in early phase development.

Despite improvements in candidate design that lead to forced pairings, generating cell lines that produce heterodimers at high concentrations without simultaneous increases in product-related impurities remains an important goal in the development of economically viable processes that will yield safe and efficacious products (17).

One group of cell-line engineers have defined a process for generating cell lines that express bsAbs comprising three polypeptide chains (18). To increase the likelihood of correct protein assembly, the scientists introduced three genes into separate vectors, enabling developers to optimize their ratios of the different plasmids that resulted. Transfected pools were analyzed with a cell-secretion assay to determine which of them contained high proportions of high producers. The team used capillary electrophoresis with sodium-dodecyl sulfate (CE-SDS) to identify the pool containing the highest proportion of a desired bispecific heterodimer form. Clones were created from the pool with the most optimized plasmid ratio and grown in microscale bioreactors to assess growth performance, productivity, and heterodimer purity. Ultimately the team selected a clone that expressed heterodimers to 97% purity on CE-SDS and 98% purity by size-exclusion high-performance liquid chromatography (SEC-HPLC) with <2% aggregates. The resulting cell line expressed a bsAb at ~5 g/L and proved to be stable beyond 60 generations (18).

Not all development groups have adopted the approach of integrating each gene coding for a bsAb into separate vectors. A team from Eli Lilly integrated two heavy-chain and two light-chain genes into a single-plasmid quad vector that gave titers comparable to multiplasmid systems while facilitating generation of stable clones (19).

Production and Processing
Upstream: Because the composition of cell culture media has been shown to affect product quality, media design should be considered as a means to control levels of partial products, aggregates, and product variants in cell culture harvests (17). Scientists at Amgen have shown that cell culture temperature can modulate half-antibody and aggregate formation in a CHO cell line. The team reduced product-related impurity formation by adopting a biphasic culture process with a growth phase at 36 °C followed by a cooler phase that improved product quality to increase yields through the downstream process (20).

Downstream: BMS has used a standard three-stage filtration process in harvesting cell cultures with its bsAb platform. The process includes a protein A affinity capture step, but the team performed additional development work to optimize loading capacities and maximize step yields. Elution-buffer screening helped optimize yields and minimize aggregation. Product-related impurities were removed effectively following two further polishing steps of ion-exchange and mixed-mode chromatography. Although step yields across one or two polishing steps were lower than might be expected for a MAb, the platform allowed for purification of three different bsAbs to meet product-quality regulatory requirements for clinical trials (14).

Because of fundamental structural similarities between monoclonal and bispecific antibodies, many current purification methods for bispecifics are based on established purification processes for conventional MAbs. Affinity, charge, size, hydrophobicity, and mixed modes of chromatography often are used, with additional strategies applied to overcome the unique challenges presented by bsAbs (21). Those include aggregation, fragment formation, and mispaired products.

For example, researchers at Regeneron have used differential protein A chromatography to remove homodimeric heavy-chain mispaired molecules after introducing a point mutation into the Fc region of one heavy chain to prevent protein A binding. Homodimers with the mutation thus flow through the capture column during loading. Fully bsAbs can be separated from those without the mutation by leveraging their decreased avidity for the protein A matrix (22).

Although protein A chromatography still might be the primary means for bsAb capture, TeneoBio and its partners have reported replacing that affinity resin in the capture step for a bispecific CD3-TAA candidate with a another based on a llama VHH “nanobody” fragment immobilized on agarose. The ligand binds specifically to the CH1 domain of IgG heavy chains. That enabled the team to reduce copurification of product variants with intact Fc domains and also to elute the product under mild conditions that minimize the aggregation of both active and inactive variants (23).

Moving beyond the initial capture step, another group of purification-development scientists have adapted a mixed-mode polishing step to facilitate removal of a by-product resulting from incomplete chain pairing of a symmetric bsAb (24). This is accomplished by operating the column in a weak partitioning mode, which provides for high throughput, good yield, and effective by-product removal (24).

Pushing Antibody Technologies Forward
Generating antibodies with two or more specificities is one of the most innovative fields in therapeutic antibody development. It has tremendous potential for use in creating new treatments for patients who currently have unmet needs. Bispecific-antibody development also is stimulating innovations in bioprocessing techniques from expression through upstream processing and candidate purification. Wherever possible, process-development scientists and engineers are borrowing techniques honed in mature MAb platforms and applying them to bsAb manufacturing. Nevertheless, the unique qualities of bispecifics make developing biomanufacturing strategies for them a still-emerging art form. Much remains to be done toward creating high-producing and high-yielding processes for this class of products, and challenges are likely to increase as bsAbs become more sophisticated and complex in the future.

References
1 Nisonoff A, Wissler FC, Lipman LN. Properties of the Major Component of a Peptic Digest of Rabbit Antibody. Science 132 (3441) 1960: 1770–1771; https://doi.org/10.1126/science.132.3441.1770.

2 Karpovsky B, et al. Production of Target-Specific Effector Cells Using Hetero-Cross-Linked Aggregates Containing Anti-Target Cell and Anti-Fc Gamma Receptor Antibodies. J. Exp. Med. 160 (6) 1984: 1686–1701; https://doi.org/10.1084/jem.160.6.1686.

3 Staerz UD, Kanagawa O, Bevan MJ. Hybrid Antibodies Can Target Sites for Attack By T Cells. Nature 314(6012) 1985: 628–631; https://doi.org/10.1038/314628a0.

4 Perez P, et al. Specific Targeting of Cytotoxic T Cells By Anti-T3 Linked to Anti-Target Cell Antibody. Nature 316(6026) 1985: 354–356; https://doi.org/10.1038/316354a0.

5 Huston JS, et al. Protein Engineering of Antibody Binding Sites: Recovery of Specific Activity in an Anti-Digoxin Single-Chain Fv Analogue Produced in Escherichia coli. Proc. Nat. Acad. Sci. USA 85, 1988: 5879–5883; https://www.pnas.org/content/pnas/85/16/5879.full.pdf.

6 Mack M, et al. A Small BsAb Construct Expressed As a Functional Single-Chain Molecule with High Tumor Cell Cytotoxicity. Proc. Nat. Acad. Sci. USA 92(15) 1995: 7021–7025; https://dx.doi.org/10.1073%2Fpnas.92.15.7021.

7 Ridgway JB, Presta LG, Carter P. “Knobs-into-Holes” Engineering of Antibody CH3 Domains for Heavy Chain Heterodimerization. Protein Eng. 9, 1996: 617–621; https://doi.org/10.1093/protein/9.7.617.

8 Merchant AM, et al. An Efficient Route to Human Bispecific IgG. Nature Biotechnol. 16(7) 1998: 677–681; https://www.nature.com/articles/nbt0798-677.

9 Riethmuller G. Symmetry Breaking: BsAbs, the Beginnings, and 50 Years On. Cancer Immunol. 12, 2012: epub; https://pubmed.ncbi.nlm.nih.gov/22896757.

10 Brinkmann U, Kontermann RE. The Making of BsAbs. mAbs 9(2) 2017: 182–212; https://dx.doi.org/10.1080%2F19420862.2016.1268307.

11 Labrijn AF, et al. BsAbs: A Mechanistic Review of the Pipeline. Nat. Rev. Drug Discov. 18(8) 2019: 585–608; https://doi.org/10.1038/s41573-019-0028-1.

12 Wang Q, et al. Design and Production of BsAbs. Antibodies 8(3) 2019: 43; https://doi.org/10.3390/antib8030043.

13 Rios M, Lundberg B. A Light-Chain Platform for Developing BsAbs. BioProcess Int. ebook, April 2020; https://bioprocessintl.com/manufacturing/monoclonal-antibodies/common-light-chain-platform-for-developing-bispecific-antibodies.

14 Zheng J, Waller JA, Ghose S. Recovery and Purification Processing for BsAb Production. Am. Pharm. Rev. 29 March 2021: https://www.americanpharmaceuticalreview.com/Featured-Articles/574699-Recovery-and-Purification-Processing-for-Bispecific-Antibody-Production/.

15 Schaefer W, et al. Immunoglobulin Domain Crossover As a Generic Approach for the Production of Bispecific IgG Antibodies. PNAS 108(27) 2011: 11187–11192; https://www.pnas.org/cgi/doi/10.1073/pnas.1019002108.

16 Nardis CD, et al. A New Approach for Generating BsAbs Based on a Common Light Chain Format and the Stable Architecture of Human Immunoglobulin G1. JBC 292(35) 2017: 14706–14717; https://doi.org/10.1074/jbc.M117.793497.

17 Bratt J, et al. Therapeutic Ig-G Like BsAbs: Modular Versatility and Manufacturing Challenges, Part 2. BioProcess Int. 16(1–2) 2018: 40–45; https://bioprocessintl.com/analytical/cell-line-development/therapeutic-igg-like-bispecific-antibodies-modular-versatility-and-manufacturing-challenges-part-2.

18 Fagete S, et al. Breakthrough CLD Platform for BsAbs Expression. ESACT (Poster) BMC Proc. 14(Suppl 5) 2020: 5; https://doi.org/10.1186/s12919-020-00188-y.

19 Rajendra Y, et al. Transient and Stable CHO Expression, Purification and Characterization of Novel Hetero-Dimeric Bispecific IgG Antibodies. Biotechnol. Progr. 33(20) 2016: 469–477: https://doi.org/10.1002/btpr.2414.

20 Gomez N, et al. Culture Temperature Modulates Half Antibody and Aggregate Formation in a Chinese Hamster Ovary Cell Line Expressing a BsAb. Biotechnol. Bioeng. 115(12) 2018: 2930–2940; https://doi.org/10.1002/bit.26803.

21 Chen SW, Zhang W. Current Trends and Challenges in the Downstream Purification of BsAbs. Antibody Therapeut. 4(2) 2021: 73–88; https://doi.org/10.1093/abt/tbab007.

22 Tustian AD, et al., Development of a Novel Affinity Chromatography Resin for Platform Purification of BsAbs with Modified Protein A Binding Avidity. Biotechnol. Progr. 24(30) 2018: 650–658; https://doi.org/10.1002/btpr.2622.

23 Chamow SM, et al. Capture of CH1-Contraining BsAbs: Evaluating an Alternative to Protein A. BioProcess Int. 18(5) 2020: 26–38; https://bioprocessintl.com/downstream-processing/chromatography/capture-of-ch1-containing-bispecific-antibodies-evaluating-an-alternative-to-protein-a.

24 Zhang T, et al. Removing a Light-Chain Missing Byproduct By MMC ImpRes Mixed Mode Chromatography Under Weak Partitioning Mode in Purifying a WuXiBody-Based BsAb. Prot. Expr. Purif. 186, 2021: 105927; https://doi.org/10.1016/j.pep.2021.105927.

Dr. Janice Reichert is editor in chief of mAbs (an Informa Taylor & Francis journal) and executive director of The Antibody Society, Inc., 247 Prospect Street, Framingham MA 01701; https://www.antibodysociety.org; [email protected]. Corresponding author Dr. Nick Hutchinson is deputy chair of The Antibody Society’s communications and membership committee and business steering group lead for mammalian cell culture at Fujifilm Diosynth Biotechnologies, Belasis Avenue, Stockton-on-Tees, Billingham, UK, TS23 1LH; https://fujifilmdiosynth.com; [email protected].

This article is abridged from BPI’s September 2021 eBook, “Bispecific Antibodies: Their Development and Manufacturing As Therapeutics” by Janice Reichert and Nick Hutchinson. Download the full version, including pipeline details and illustrative figures, on the BPI website here: https://bioprocessintl.com/manufacturing/emerging-therapeutics-manufacturing/ebook-bispecific-antibody-development-and-manufacture.