Production of therapeutic antibodies with controlled fucosylation

Abstract

The clinical success of therapeutic antibodies is demonstrated by the number of antibody therapeutics that have been brought to market and the increasing number of therapeutic antibodies in development. Recombinant antibodies are molecular-targeted therapeutic agents and represent a major new class of drugs. However, it is still very important to optimize and maximize the clinical efficacy of therapeutic antibodies, in part to help lower the cost of therapeutic antibodies by potentially reducing the dose or the duration of treatment. Clinical trials using therapeutic antibodies fully lacking core fucose residue in the Fc oligosaccharides are currently underway, and their remarkable physiological activities in humans in vivo have attracted attention as next-generation therapeutic antibody approaches with improved efficacy. Thus, an industrially applicable antibody production process that provides consistent yields of fully non-fucosylated antibody therapeutics with fixed quality has become a key goal in the successful development of next-generation therapeutic agents. In this article, we review the current technologies for production of therapeutic antibodies with control of fucosylation of the Fc N-glycans.

Introduction

The clinical successes of recombinant humanized therapeutic antibodies have included improvement of overall survival and time to disease progression in the treatment of human malignancies, such as breast, colon and hematological cancer.^1–4 Currently 22 therapeutic monoclonal antibodies are approved by the US Food and Drug Administration (FDA). The worldwide revenue of therapeutic antibodies has reached $20 billion in 2007 and eight therapeutic antibodies are “blockbusters” with more than $1 billion annual sales. More than 200 monoclonal antibodies are in clinical trials, and this number is predicted to increase to around 250 by 2010. Recombinant antibodies are molecular-targeted therapeutic agents, and these agents represent a major new class of drugs. However, improvement of in vivo efficacy of therapeutic antibodies continues to be a challenge. The in vivo physiological activity of therapeutic antibodies is mediated by two independent mechanisms, (1) the efficacy resulting from target antigen neutralization or apoptosis and (2) biological activities referred to as antibody effector functions, antibody-dependent cellular cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC), which are activated by the formation of immune complexes.^5–8 The importance of ADCC for the clinical efficacy of therapeutic antibodies, especially anticancer antibodies, has become clear from genetic analyses of leukocyte receptor (FcγR) polymorphisms in patients,^9–19 and ADCC enhancement technology including the modification of N-glycans attached to the constant region (Fc) of the antibody has been a focus of attention for the biopharmaceutical industry.

Glycoengineered therapeutic antibodies lacking core fucose residue from the Fc N-glycans exhibit strong ADCC at lower concentrations with much higher efficacy compared to fucosylated counterparts,^20–37 and can evade the inhibitory effect of serum immunoglobulin G (IgG) on ADCC through its high binding to gamma receptor IIIa (Fc FcγRIIIa).^35,38,39 Recently, two non-fucosylated therapeutic antibodies, KW-0761 and MEDI-563/BIW-8405, demonstrated the ability to effectively deplete target cells from human blood through dramatically enhanced ADCC at very low doses of 0.003–0.03 mg/kg in Phase 1 clinical studies.^40,41 Fully non-fucosylated antibodies are, therefore, expected to be among the most powerful and elegant approaches for the development of next-generation therapeutic antibodies with inproved efficacy.

Robust and stable production of fully non-fucosylated therapeutic antibodies with fixed quality is required for the development of therapeutic antibodies because the high level of ADCC efficacy of non-fucosylated therapeutic antibody molecules is reduced in vivo by fucosylated counterparts through competition for binding to the antigen on target cells.^35,38,39 Unfortunately, all therapeutic antibodies that are currently on the market are heavily fucosylated because they are produced by mammalian cell lines with intrinsic enzyme activity responsible for the core-fucosylation of the Fc N-glycans of the products. Unique mammalian host cell lines that can stably produce fully non-fucosylated therapeutic antibodies have been successfully established by glycoengineering technologies,^{22,23,33,34,36} and have started to be employed in the biopharmaceutical industry. In this article, we review current technologies for the production of therapeutic antibodies with controlled fucosylation.

Role of N-glycosylation in biopharmaceuticals.

The efficacy of therapeutic biologics is well recognized to be critically dependent on appropriate posttranslational modifications. All marketed recombinant therapeutic antibodies are of the IgG class; IgG is a glycoprotein bearing a tri-mannosyl core structure of complex-type N-glycosylation of the Fcγ A comprehensive analysis of the protein data bank has indicated that more than half of all proteins are glycoproteins, and that about 90% of well-characterized glycoproteins bear N-glycosylation,⁴² suggesting that most bioactive proteins with possible therapeutic use will be also N-glycosylated. N-glycosylation of biopharmaceuticals plays several essential roles, e.g., conformational stability, folding accuracy, heat stability, solubility, pharmacokinetics, protection against proteolytic degradation, and is important for biological activity as well. Regulatory authorities require consistency of glycosylation heterogeneity to minimize the content of undesirable glycoforms produced during the manufacturing process, irrespective of the production system. Glycoengineering to properly control sugar residue structures of the products is thus an essential requirement for biopharmaceutical production.

Generally, the major benefit brought about by glycoengineering of biotherapeutics is the improvement of pharmacokinetics for retaining in vivo biological activity. The study of recombinant erythropoietin (rEPO) illustrates the importance of N-glycosylation control in clinical development. EPO analogues, including novel erythropoiesis stimulating protein (NESP), with additional consensus sequences for N-glycan attachment (Asn-X-Ser/Thr), retain the structural integrity and physical stability of rEPO, but induce a faster and higher hematocrit increase compared to rEPO.^43,44 This increased in vivo efficacy is due to the extended terminal half-life, which is highly correlated with the sialic acid content of the additional N-glycans, and not with the higher binding affinity to EPO receptor, which is inversely correlated with the sialic acid content.⁴⁵

Similarly, follicle-stimulating hormone (FSH) variants with an additional N-glycan also increase half-life fourfold, and the area under the curve (AUC) sixfold, compared to unmodified recombinant FSH. One hypersialyted candidate, FSH1208, demonstrated reduced renal clearance, and was found to improve ovary weight augmentation and serum estradiol levels in rats.⁴⁶ Tissue plasminogen activator (t-PA) harboring a high-mannose-type N-glycan, responsible for hepatic clearance, has an extended half-life (3 mins versus 30 mins) due to translocation of the N-glycosylation site, which changes its glycoform to the complex-type.⁴⁷ The application of N-glycan addition is also being employed in the development of first-in-class new biologics including leptin and Mpl ligand. In preclinical studies, obese (ob/ob) mice treated with a leptin analog containing five additional N-glycans lost more weight and maintained the reduced weight for a longer period compared to those treated with a nonglycosylated recombinant leptin.⁴⁸ When modified by the addition of four or six N-glycans, the N-terminal EPO-like domain of Mpl ligand, an O-glycosylated hormone lacking N-glycan, improves the amount and duration of platelet production.⁴⁸

Other benefits of biotherapeutics glycoengineering include improvement of the physicochemical properties responsible for in vivo biological activity of the products. Deglycosylated interferon beta (IFN-β) was found to have reduced antiviral activity compared to the glycosylated form due to the formation of an insoluble disulfide-linked precipitate and large soluble aggregates.⁴⁹ The fully nonglycosylated EPO produced by E. coli unfolds and precipitates upon exposure to heat and is sensitive to pH-induced denaturation, while the glycosylated EPO remains soluble and stable.⁵⁰ In human EPO, two of three N-glycans contribute to effective secretion from the cells.⁵¹ Collectively, these studies suggest that maintaining and stabilizing the physical properties of therapeutic proteins is very important for storage and the in vivo efficacy of the products. In addition, the potential for reduced immunogenicity is another advantage of biotherapeutics glycoengineered products. Non-human-derived recombinant proteins hold a risk of severe immunogenicity due to foreign sugar residues, e.g., α-1,3 fucose and β-1,2 xylose residues of plant and hyper-mannose antenna of yeast.^52,53 Gycoengineering of non-mammalian organisms to provide human-type glycosylation with consistent quality is essential for the production of glycosylated biopharmaceuticals. Hence, most biopharmaceutical processes have adopted mammalian cell lines as production hosts to circumvent issues of glycobiopharmaceutical immunogenicity.

Traditional glycosylation control approaches ensure stabilization and duration of medicinal effects by enhancing physical stability or extending terminal half-life. In these cases, the properties of each molecule in the product composition contribute to the medical effects. There have been no case studies examining enhancement of innate in vivo potency, rather than in vivo stability, of candidates by selection of the most suitable glycoform from the naturally occurring heterogeneous forms. It should be emphasized that the enrichment of the most suitable glycoform among the naturally occurring varieties is very important for the development of next-genereation therapeutic antibodies because undesired glycoforms inhibit the in vivo efficacy of the desired form through competition for binding to targets.^35,38,39 In traditional approaches applied to non-antibody proteins, the presence of undesired glycoforms has not been a serious concern because the molecules do not inhibit the in vivo efficacy of the desired ingredients. The antibody defucosylation approach is fundamentally different from existing glycosylation control strategies in this regard.

Significance of Fucosylation Control.

Compared to fucosylated IgGs, non-fucosylated forms exhibit dramatically enhanced ADCC due to the enhancement of FcγRIIIa binding capacity without any detectable change in CDC or antigen binding capability.^20,22,24,35 Among the effector functions of antibody therapeutics, ADCC has been identified clinically as an important mechanism of anti-cancer therapeutic antibodies.^9–19 N-glycosylation of antibody Fc regions is essential for binding to FcγR, which engages antibody effector functions.⁵⁴ The N-glycans bound to Asn-297 in the Fc of human IgG1 are biantennary complex-type composed of a tri-mannosyl core structure with or without core fucose residues, bisect N-acetylglucosamine (GlcNAc) and terminal galactose (Gal), thus giving rise to structural heterogeneity.⁵⁵ Non-fucosylated forms of human IgGs are observed as natural components in normal human serum, although the majority of the IgGs is fucosylated. The non-fucosylated anti-bodies have much higher binding affinity for FcγRIIIa than fucosylated human serum IgG, which is an essential feature for overcoming the competition of human serum IgG for binding of therapeutic IgG1 to FcγRIIIa on natural killer (NK) cells,^56,57 and can exhibit high ADCC even in human whole blood (Figure 1).^38,39 Crystal structure analysis revealed that the ADCC enhancement by non-fucosylated IgG1s is attributed to a subtle conformational change in a limited region of IgG1-Fc,⁵⁸ and that the high affinity of non-fucosylated antibodies for FcγRIII is mediated by interactions formed between the carbohydrate at FcγRIII Asn-162 and regions of the Fc that are only accessible when the Fc N-glycans lack fucose residues.⁵⁹

Figure 1.

Therapeutic antibody-induced ADCC in human blood. Therapeutic antibodies show the same antigen binding activity irrespective of core fucosylation of the Fc. (a) Non-fucosylated antibodies overcome the competition with serum IgG to bind to the effector cells through much higher binding affinity to FcγRIIIa than serum IgG, and thereby induce high ADCC. (b) Fucosylated antibodies fail to recruit effector cells effectively due to low binding affinity to the FcγRIIIa. (c) The high ADCC of non-fucosylated antibodies is inhibited by the fucosylated counterparts through the competition for binding to the antigen on target cells.

The advantages of non-fucosylated antibodies include achieving therapeutic efficacy at low doses,^21,22 inducing high cellular cytotoxicity against tumor cells that express low levels of antigen,²⁸ triggering high effector function in NK cells with the low-affinity FcγRIIIa allotype for the IgGs,²⁶ and exhibiting strong and saturated ADCC identical to that of Fc amino acid mutants with much higher FcγRIIIa binding affinity.^32,38 The superior in vivo efficacy of non-fucosylated IgG1s has also been demonstrated using a human PBMC-engrafted mouse model.²⁵ Recently, the Phase I studies of two therapeutic antibodies have proved the clinical usefulness of non-fucosylated antibodies. A trial of MEDI-563/BIW-8405, a fully non-fucosylated humanized anti-IL-5 receptor IgG1, has showed long-duration, complete elimination of human peripheral eosinophils after one injection at a low dose (0.03 mg/kg) (Kolbeck R et al., the American Society of Hematology 49th Annual Meeting 2007).⁴¹ Another fully non-fucosylated anti-body KW-0761, a humanized anti-CCR4 IgG1, showed a strong killing effect against adult T-cell leukemia-lymphoma (ATLL) at a 0.01 mg/kg dose, without any severe adverse events in a CCR4-positive peripheral T-cell lymphoma patient.⁴⁰ In a study of healthy subjects, KW-0761 has been reported to reduce the number of peripheral CCR4-positive cells by more than 80%. The candidate also reduced the production of Th2 cell-mediated cytokines with no change in the production of Th1 cell-mediated cytokines at a 0.003 mg/kg dose.⁴¹ The effects at these low doses are remarkably impressive; doses of currently marketed antibody therapies are typically in the 1–10 mg/kg range. This ADCC enhancement by defucosylation was also shown to be applicable to all human IgG subclasses²⁷ and Fc-fusion molecules.^29–31

The results observed for MEDI-563/BIW-8405 and KW-0761 suggest that further examination of non-fucosylated therapeutic antibodies is warranted. However, in the case of products composed of both forms, the enhanced ADCC of non-fucosylated therapeutic antibodies is inhibited by the fucosylated counterparts through competition for binding to antigen targets (Figure 1).^38,39 Development of the next-generation therapeutic antibodies with fully non-fucosylated Fc N-glycans thus requires robust production processes that include a consistent glycoform profile to completely control the core fucosylation level.

Achieving Fucosylation Control.

Several methods that lead to successful production of relatively highly non-fucosylated therapeutics have been reported. Possible defucosylation approaches can be grouped into three methodologies: (1) conversion of the N-glycosylation pathway of nonmammalian cells to the ‘humanized’ non-fucosylation pathway; (2) inactivation of the N-glycan fucosylation pathway of mammalian cells and (3) in vitro chemical synthesis of non-fucosylated N-glycoprotein or enzymatic modification of N-glycans to non-fucosylated forms.

Approach (1) conversion of non-mammalian N-glycosylation pathway into the mammalian type.

Protein N-glycosylation occurs in the endoplasmic reticulum (ER) lumen and Golgi apparatus.^60,61 The process begins with a flip of a branched dolichol-linked oligosaccharide, Man5GlcNAc2, synthesized in the cytoplasm, into the ER lumen to form a core oligosaccharide, Glc₃Man₉GlcNAc₂. The oligosaccharide is then transferred to an asparagine residue of the N-glycosylation consensus sequence on the nascent polypeptide chain, and sequentially trimmed by α-glycosidases I and II, which remove the terminal glucose residues, and α-mannosidase, which cleaves a terminal mannose residue. The resultant oligosaccharide, Man₈GlcNAc₂, is the junction intermediate that may either be further trimmed to yield Man₅GlcNAc₂, an original substrate leading to a complex-type structure in higher eukaryotes including mammalian cells, or extended by the addition of a mannose residue to yield Man₉GlcNAc₂ in lower eukaryote, in the Golgi apparatus.

Yeasts may offer several advantages as host cells in simplicity and cost-effectiveness compared to mammalian cell cultures, as well as much higher growth rates and cell densities and a reduced risk of viral infection from animal-originated additives. In lower eukaryotes including yeasts, Man₈GlcNAc₂, the junction intermediate with the mammalian N-glycosylation pathway, is extended by the addition of a mannose residue to yield Man₉GlcNAc₂ followed by the formation of a hyper-mannosylated structure. Many attempts have been made to customize the yeast N-glycosylation pathway to the human-type manner as part of a strategy to eliminate non-human-specific N-glycosylation genes and to introduce human-type N-glycosylation genes. Chiba et al. have identified several key genes of the yeast hyper-mannosylation pathway, and first demonstrated that a hyper-mannosylation-deficient mutant lacking the genes for outer chain extension would convert Man₈GlcNAc₂ to Man₅GlcNAc₂, an origin of a complex-type structure, by the introduction of α-1,2 mannosidase using Saccaromyces cerevisiae.⁶² Domain-swapping libraries of the relevant genes encoding glycosyltransferases, glycosidases, sugar nucleotide synthesis enzymes and sugar nucleotide transporters in mammalian pathways have enabled the proper localization of each component in the Golgi compartment to maximize the individual functions,⁶³ resulting in engineered yeasts, Pichia pastris, that can produce afucosylated complex-type N-glycans involving terminal galactosylation and sialylation.^64,65 Further studies are required to confirm whether the engineered yeasts bearing numerous glycosylation genes will maintain constitutively the balanced appropriate expression of multiple heterologous genes.

Plants may also be attractive hosts in terms of both avoiding the risk of animal-derived viral infection and cost-effectiveness of biopharmaceutical production. Higher plants have similar N-glycosylation pathways compared to mammals, and mainly generate complex-type glycans with an α-1,3 fucose residue attached to the innermost GlcNAc, a β-1,2 xylose residue attached to the junction mannose of the tri-mannosyl core, neither of which is found in humans. The immunogenicity of the non-human glycosylation, α-1,3 fucosylation and β-1,2 xylosylation, is of concern to regulatory authorities.⁵² Several groups have established mutant lines lacking α-1,3 fucose residues and β-1,2 xylose residues on N-glycans. The first deficient line was derived from Arabidopsis thaliana by crossing two α-1,3 fucosyltransferase-deficient mutants and a β-1,2 xylosyltransferase mutant identified from the T-DNA tagging line libraries.⁶⁶ The next deficient line came from moss, in which homologous recombination events frequently occur, by gene knockout targeting an α-1,3 fucosyltransferase gene and a β-1,2 xylosyltransferase gene.⁶⁷ None of the products from the engineered moss, vascular endothelial growth factor (VEGF), IgG4, and IgG1, are reported to carry either α-1,3 fucose or β-1,2 xylose on N-glycans.^67–69 The double gene knockdown approach has been also performed in duckweed.⁷⁰ The engineered duckweed expressed IgG1 with almost total elimination of α-1,3 fucose and β-1,2 xylose on N-glycans (95.8%). These hosts require specialized facilities for each species, since existing facilities for mammalian cell hosts are not appropriate for plant-based hosts. In addition, the cost-effectiveness of plant systems is still unclear, in contrast to the well-known mammalian process. Aquatic species have good potential to secrete the products into culture medium, which is a substantial advantage in the downstream of the production process. However, enhancing the secretion yield of a high-molecular-weight protein such as antibodies from the cell wall remains a challenging issue.

Approach (2) modification of fucosylation pathways in mammalian host cells.

In mammals, most core-fucosylation on N-glycans is formed by α-1,6 linkage. FUT8 is the only α-1,6 fucosyltransferase that catalyzes the transfer of fucose residues from GDP-fucose to the innermost GlcNAc of the tri-mannosyl core structure via the α-1,6 linkage in the medial Golgi cisternae.⁷¹ The intracellular GDP-fucose, a donor substrate for fucose transfer, is synthesized in the cytoplasm in two manners: de novo and salvage pathways. In the de novo pathway, GDP-mannose, arising from extracellular D-glucose, is converted to GDP-fucose via a three-step enzymatic reaction by GDP-mannose 4,6-dehydratase (GMD)⁷² and GDP-keto-6-deoxymannose 3,5-epimerase/4-reductase (FX).⁷³ In the salvage pathway, l-fucose, incorporated from an extracellular or lysozomal source, is phosphorylated by fucokinase⁷⁴ and then converted to GDP-fucose by GDP-fucose pyrophosphorylase (GFPP).⁷⁵ Most GDP-fucose is produced via the de novo pathway, while free l-fucose is reused for the salvage pathway.⁷⁶ The cytoplasmic GDP-fucose is incorporated into the Golgi apparatus by GDP-fucose transporter (GFT)^77,78 anchored at the Golgi membrane to provide a fucosylation substrate.^76,79 Several reports have suggested that GMD,³⁶ FX,⁸⁰ GDP-fucose transporter^77,78,81 and FUT8^22,23,36,82 are candidates to reduce α-1,6 fucosylation by genetic engineering. Among these, GMD and FUT8 have proven to share no redundancy, and to be independent key enzymes almost solely responsible for α-1,6 fucosylation, without complementing one another.^22,34,36,82

There are four existing mammalian defucosylation methods: (1) use of the host cells with reduced intrinsic α-1,6 fucosylation ability, e.g., Lec13, a variant of CHO cells partially deficient in GMD function,²⁰ or YB2/0, a rat-rat hybridoma cell line with intrinsically reduced FUT8 activity;²¹ (2) introduction of small interfering RNA (siRNA) against the α-1,6 fucosylation relevant genes;^23,36,81 (3) co-introduction of β-1,4-N-acetylglucosaminyltransferase (GnTIII) and Golgi α-mannosidase II (ManII);^58,83,84 and (4) disruption of the genomic locus responsible for α-1,6 fucosylation. The first through third approaches do yield partially non-fucosylated forms, but cannot stably yield fully non-fucosylated composition. In fact, the non-fucosylation range in these systems greatly depends on the original character of the individual cell clone, culture environment and culture phase, resulting in a lack of both systemic reproducibility and glycoform consistency. In addition, the co-expression of GnTIII and ManII increases not only bisecting GlcNAc residues, but also nonfucosylated hybrid-type glycans, which are uncommonly observed forms in vivo and bear lower ADCC than non-fucosylated complex-type forms.³⁵ A highly effective siRNA approach demonstrated that double knockdown of FUT8 and GMD achieved almost full defucosylation of antibodies and stably maintained the defucosylation level during a certain culture period in a well-regulated small scale.³⁶ Today, two knockout cell lines derived from CHO/DG44 cells by the fourth approach are the most applicable host cells for production of fully non-fucosylated antibodies.

The GMD knockout CHO/DG44 cell line was isolated as a spontaneous mutant by a selective culture using Lens culinaris agglutinin (LCA), recognizing the α-1,6 fucosylated tri-mannnosyl core structure on N-glycans on the cell surface to commit the cells to a cell-death pathway.³⁴ This cell line is completely deficient in GMD function due to a genomic deletion corresponding to the murine exons 5, 6 and 7, resulting in a complete loss of the ability to form intracellular GDP-fucose for fucosylation in the absence of extracellular _l-fucose. In the presence of _l-fucose, the cell line can develop its fucosylation ability through the salvage pathway. The GMD knockout cell line stably produces fully non-fucosylated antibodies with enhanced ADCC during a serum-free fed-batch culture with consistent N-glycosylation forms.

The FUT8 knockout CHO/DG44 cell lines were established by targeted disruption of the FUT8 alleles.²² These cell lines lack the FUT8 genomic region, including the translation initiation sites on both FUT8 alleles by two rounds of homologous recombination, which is a very rare event in mammalian somatic cells.⁸⁵ The FUT8 knockout cell lines exhibit almost the same morphology and growth kinetics as the parent cells, and produce fully non-fucosylated antibodies with consistency in galactosylation pattern through a serum-free fed-batch culture using an 1L bioreactors at a productivity of 1.76 g/L.³³ Anti-CD20 therapeutic antibodies derived from FUT8 knockout cells possess equal antigen-binding and CDC and more than 100-fold higher ADCC than the parent-produced one.²² Fully non-fucosylated antibodies studied in clinical trials were produced by the FUT8 knockout cells with a fed-batch culture in a kiloliter-scale process without any scale-up problems. A few dozen therapeutic antibodies are currently under development are being produced using the FUT8 knockout cell systems.

Approach (3) in vitro fucosylation control system.

In vitro addition of N-glycan to nonglycosylated proteins is a well known approach to obtaining a desired glycoform product. Inazu et al. have demonstrated that non-fucosylated NANA₂Gal₂GlcNAc₂Man₂GlcNac₂ can be transferred to one of three N-glycosylation sites on nonglycosylated insulin by endo-β-N-acetylglucosaminidase (Endo-M), which hydrolyzes a GlcNAc-GlcNAc bond to transfer an N-glycan attached to the donor peptide via GlcNAc residue to the acceptor glycosylation site labeled with GlcNAc residue.⁸⁶ In vitro release of fucose residues on IgGs by a fucosidase also has been reported.⁸⁷ However, conformational analysis suggests that in antibodies, N-glycans are embedded between two Fc domains,^88,89 and the enzymatic defucosylation efficiency is much lower due to steric hindrance, i.e., access of fucosidase to fucose residues is blocked by potions of the Fc domains. These enzymatic strategies generally have production cost issues and the yields are heavily influenced by the composition of N-glycosylation substrates.

Recently, Yamamoto et al. have reported a very interesting approach, the world's first whole chemical synthesis of a human N-glycoprotein.⁹⁰ They synthesized monocyte chemotactic protein 3 (MCP-3), consisting of 76 amino acids and an N-glycan, by dividing the whole peptide sequence into three fragments for chemical synthesis and ligating all of the pieces, thereby generating a single isoform with non-fucosylated NANA₂Gal₂GlcNAc₂Man₂GlcNac₂. Although this technology requires further improvement in yield and more practical applications with complex structures involving folding, in the future it may prove to be a suitable method for stable protein production of a single glycoform, including therapeutic antibodies.

Conclusion

Non-fucosylated antibodies, harboring a tri-mannosyl core structure of complex-type N-glycans of Fc without fucose residue, have revolutionized antibody therapies by providing greatly enhanced ADCC. A number of clinical applications of therapeutic antibodies have proven the importance of antibody ADCC for in vivo efficacy, and fully non-fucosylated antibodies are following on this success. Non-fucosylated antibodies can eliminate human serum IgG's inhibitory effect on ADCC through its high binding activity to FcγRIII; the homogeneous glycoforms composed of fully non-fucosylated glycans are required for maximized in vivo efficacy of ADCC with avoidance of the inhibitory effects of the fucosylated counterparts through the competition for antigen binding. Therefore, the production of antibody therapeutics consisting of fully non-fucosylated forms is recognized as a critical factor in the development of antibody therapies with high efficacy and reduced cost. With regard to clinical efficacy and regulatory requirements, a key element is development of a robust production process to ensure the structural consistency of the product. As of now, the manufacturing system using FUT8 knockout CHO/DG44 cells is the only method that provides clinical samples approved by regulatory authority and is considered to be the most feasible and reliable strategy for manufacturing fully non-fucosylated therapeutic antibodies among the existing production systems.

Due to the improved properties of non-fucosylated antibodies, clinical study of these candidates will continue to increase. Application of a defucosylation approach to high-level production systems approved for marketed recombinant therapeutics, e.g., CHOK1SV and PerC.⁶ systems, is practical and useful, and the combination of high-level production of products with improved efficacy might have a great impact on the use of antibody treatments, especially in expansion of the therapeutic indications.

Abbreviations

ADCC

antibody-dependent cellular cytotoxicity

CDC

complement-dependent cytotoxicity

Fc

antibody constant region

FcγRIIIa

human Fcγ-receptor IIIa

IgG

immunoglobulin G

NK cell

natural killer cell

CHO

Chinese hamster ovary

EPO

erythropoietin

Glc

glucose

Man

mannose

GlcNAc

N-acetylglucosamine

Gal

galactose

NANA

N-acetylneuraminic acid

FUT8

α-1,6 fucosyltransferase

GMD

GDP-mannose 4,6-dehydratase

FX

GDP-keto-6-deoxymannose 3,5-epimerase/4-reductase

GFT

GDP-fucose transporter

siRNA

short interfering RNA

GnTIII

β-1,4-N-acetylglucosaminyltransferase

ManII

α-mannosidase II