Elsevier

Current Opinion in Biotechnology

Volume 30, December 2014, Pages 128-139
Current Opinion in Biotechnology

Aglycosylated full-length IgG antibodies: steps toward next-generation immunotherapeutics

https://doi.org/10.1016/j.copbio.2014.06.013Get rights and content

Highlights

  • Aglycosylated IgGs are nearly identical to glycosylated IgGs in terms of antigen binding.

  • Significant progress has been made to enhance production of aglycosylated IgGs.

  • Engineered aglycosylated Fc domains would be useful in therapeutic applications.

Albeit the removal of Asn297 glycans of IgG perturbs the overall conformation and flexibility of the IgG CH2 domain, resulting in the loss of Fc–ligand interactions and therapeutically critical immune effector functions, aglycosylated full-length IgG antibodies are nearly identical to the glycosylated counterparts in terms of antigen binding, stability at physiological or low temperature conditions, pharmacokinetics, and biodistribution. To bypass the drawbacks of glycosylated antibodies that include glycan heterogeneity and requirement of high capital investment for biomanufacturing, aglycosylated antibodies have been developed and several are under clinical trials. Comprehensive cellular and bioprocess engineering has enabled to produce highly complex aglycosylated IgGs in a simple bacterial cultivation with comparable production level as that of mammalian cells. Moreover, extensive engineering of aglycosylated Fc has converted the aglycosylated IgG antibodies into a new class of effector functional human immunotherapeutics.

Introduction

Humans are exposed to numerous pathogens causing serious acute or chronic diseases. To bind and neutralize the pathogenic antigens, human plasma B cells secrete IgG antibodies, complex tetrameric soluble glycoproteins (∼150 kDa) composed of two heavy chains and two light chains. The IgG antibody molecule, which is composed of the antigen-binding fragment, Fab, and the crystallizable fragment, Fc, for immune effector cell recruitment, bridges humoral and cellular immunity (Figure 1a). In the natural immune system, the IgG antibodies are produced from multiple different B-cell lineages and are therefore polyclonal. Since 1975, when Georges Köhler and Cesar Milstein opened the door for the generation of monoclonal antibodies via hybridoma technology, monoclonal antibodies, which are specific to a single target antigenic epitope, have been extensively harnessed as highly potent therapeutic agents. Owing to their outstanding high selectivity, low toxicity, and prolonged serum catabolic half-life (approximately 3 weeks), monoclonal antibodies such as trastuzumab (anti-Her2: Herceptin®) or bevacizumab (anti-VEGF: Avastin®) have shown significant clinical outcomes and have been considered to be one of the most effective therapeutics for treating cancer, inflammation, infection, and autoimmune diseases [1].

Currently, more than 30 monoclonal antibodies have been commercialized with the approval from the United States Food and Drug Administration (US FDA) and European Medicines Agency (EMEA) for human therapy (http://www.antibodysociety.org/news/approved_mabs.php/), and more than 240 therapeutic monoclonal antibodies are under clinical trials [2]. The market for monoclonal antibodies has been expanding, and these antibodies have been replacing conventional, nonspecific, chemically synthesized small molecule drugs. Monoclonal antibodies were expected to constitute approximately 40% of the market ($54 billion) of all biopharmaceutical products in 2013 [1]. Among the top 10 sales blockbuster biopharmaceutical drugs in 2010, six (i.e. infliximab, bevacizumab, rituximab, adalimumab, trastuzumab, and ranibizumab) were monoclonal antibody drugs [3].

Natural human IgG is a glycosylated protein appending complex biantennary glycans at Asn297 residues located within the canonical N-linked glycosylation motif (Asn-X-Ser/Thr) of Fc. The presence of a glycan at Asn297 is indispensable for the recognition of Fc-binding ligands (FcγRs and C1q) and for the activation of a variety of therapeutically critical immune effector functions such as antibody-dependent cell-mediated cytotoxicity (ADCC), antibody-dependent cell-mediated phagocytosis (ADCP), and complement-dependent cytotoxicity (CDC), consequently allowing the clearance of aberrant cells such as tumor cells or infected cells. Therefore, from the beginning of monoclonal antibody therapeutics era, the biopharmaceutical industry has focused on producing glycosylated antibodies containing similar glycan composition with native human IgGs and improving the production yield of highly valuable IgGs in mammalian expression host strains. Notably, extensive cellular engineering and bioprocess optimization efforts from the past few decades have enabled to reach a production level of approximately 10 g/L of glycosylated IgG in mammalian cells [4]. However, antibodies expressed in mammalian cells show highly variable glycoform profiles depending on the cell lines, cell culture conditions, and downstream processing procedures [5]. Due to glycan heterogeneity, which causes high variation of biological activity and stability of the final products, biomanufacturing processes for monoclonal antibody production require highly complicated, costly, and challenging downstream steps for purification, glycan mapping, and quality control. Additionally, mammalian cell cultures usually require more time-consuming cell line development, longer manufacturing time, and higher capital investment compared to microbial cell cultures.

To address the drawbacks of glycosylated antibodies produced in mammalian cell cultures, aglycosylated antibodies have been expressed in eukaryotic hosts (mammalian cells, plant cells, or yeasts) by introducing a mutation at the N-linked glycosylation site for aglycosylation or in prokaryotic hosts. Because the glycosylation status does not significantly affect the pH-dependent FcRn binding [6], which is critical for prolonged serum circulating half-life of IgG antibodies (∼3 weeks), the use of aglycosylated wild-type full-length IgG antibodies is a great choice for a range of applications such as receptor blocking and targeted delivery [7] not requiring to activate Fc-binding ligands, while possessing the beneficial prolonged serum half-life of full-length IgG format relative to antibody fragments. Moreover, sometimes, to reduce undesired inflammation and cytotoxicity, it is preferable to inhibit effector functions by abrogating the binding of FcγRs or C1q. For these applications in which effector functions are not desirable or mandatory, several aglycosylated antibodies are currently under investigation for clinical efficacy and safety before commercialization.

Recent in-depth analyses have revealed unique performances of aglycosylated full-length IgG antibodies distinguishable from conventional glycosylated antibodies; the optimization of genetic parts, expression host strains, and bioprocesses have increased the yield of aglycosylated full-length IgG antibodies in simple bacterial cultures up to the level comparable to that of mammalian cell culture. Moreover, although wild-type aglycosylated full-length IgGs cannot recruit and activate immune effector cells for the clearance of problematic cells such as tumor cells, recent extensive aglycosylated antibody Fc engineering efforts allowed to isolate aglycosylated Fc variants displaying restored, or even improved, therapeutic effector functions compared to the conventional glycosylated antibodies.

In this review, we overview the recent efforts to understand the impact of glycosylation on IgG antibodies and primarily highlight the recent advances in the field of the aglycosylated full-length IgG antibodies, with an emphasis on the production and the engineering of Fc for the new class of future immunotherapeutics.

In human IgGs, a complex biantennary glycan containing two highly dynamic branches [8] is attached at each Asn297 residue of the homodimeric Fc domain. The N-linked oligosaccharide, which is composed of a core heptasaccharide, GlcNAc4Man3, and decorating carbohydrate residues such as fucose, bisecting GlcNAc, galactose, and sialic acid, interacts with the internal amino acid residues of the antibody CH2 domain and allows inner hydrophobic amino acid residues not to be exposed to the surface while filling the space of the two CH2 domains. In human serum, circulating IgG glycoforms are highly variable depending on disease, pregnancy, aging, and inflammation status [9, 10]. In clinical use, therapeutic efficacy is largely determined by the glycosylation patterns of the administered IgG molecules. Many reports indicate that the absence of fucose, the presence of bisecting GlcNAc, the cleavage of terminal sialic acid in the N-linked glycan promote IgG molecules to bind to FcγRIIIa of NK cells and enhance ADCC. Therefore, extensive resources have been invested to improve the performance of therapeutic antibodies by glycoengineering approaches in mammalian and non-mammalian hosts [11, 12].

In particular, for the defucosylation of IgGs leading to improved FcγRIIIa binding, the development of cell lines not expressing the fucosylating enzyme, α-1,6 fucosyltransferase (FUT8), has been the most widely adopted glycoengineering technique in the biopharmaceutical industry [13]. In the comparison of two X-ray crystallographic structures (i.e. defucosylated IgG-FcγRIIIa complex and fucosylated IgG-FcγRIIIa complex), a strong interaction was observed between the glycan of defucosylated IgG and the glycan of FcγRIIIa, which was not identified in the structure of fucosylated IgG-FcγRIIIa complex [14]. To circumvent reduced efficacy, costly purification, and complex glycan mapping of glycosylated antibodies produced in mammalian cell culture, Huang et al. removed the N-linked glycan of commercial rituximab by treatment with the endoglycosidase (EndoS) derived from Streptococcus pyogenes, and then, they re-glycosylated with engineered glycosynthase (EndoS-D233A or D233Q) and synthetic glycan oxazoline to obtain a homogenous non-fucosylated product. The homogeneous defucosylated rituximab, which resulted from glycan remodeling, exhibited improved binding affinity to FcγRIIIa-V and FcγRIIIa-F by 20 fold and 9 fold, respectively [15]. In an earlier study, it was reported that α2,6-sialylation of intravenous immunoglobulin (IVIG) enhanced anti-inflammation activity [16, 17]. By using a chemoenzymatic glycoengineering technique for the re-glycosylation of IgG with sialoglycan oxazoline, human IVIG could be transformed to fully sialylated IVIG displaying higher anti-inflammatory activity [15].

The removal of the glycan at Asn297 significantly reduces the binding affinity of the antibody Fc to the high-affinity FcγRI by over two orders of magnitude [6] and leads to ablation or significant decrease of critical FcγRI-mediated immune cell activation and effector functions [6•, 18]. Additionally, the interaction of the aglycosylated Fc with low-affinity FcγRs (FcγRIIa, FcγIIb, and FcγIIIa) expressed on the surface of leukocytes and with serum complement C1q are almost completely abrogated, resulting in the loss of ADCC, ADCP, and CDC effector functions [6•, 19••, 20•, 21, 22].

The X-ray crystal structure of the human aglycosylated IgG Fc domain expressed in Escherichia coli has been recently solved and compared with human glycosylated IgG Fc produced in mammalian cells (Figure 1b) [23••]. Based on the crystal structure (PDB: 3S7G), the aglycosylated Fc domain showed shorter distance between the two P329 residues located at the top of each CH2 region of the homodimeric Fc compared to the glycosylated form. Furthermore, the C′/E loop containing the N-linked glycosylation site, N297, was highly disordered by the removal of the glycans. Notably, the distance between the CH2 domains of the aglycosylated homodimeric Fc expressed in bacteria was largely different from the deglycosylated Fc expressed in mammalian cells; the bacterial Fc was subsequently treated with the PNGase-F enzyme, which changes Asn297 into Asp297 during the glycan hydrolysis step. Surprisingly, by small-angle X-ray scattering (SAXS), larger radius of gyration (Rg) was observed in the aglycosylated Fc than in the glycosylated Fc, suggesting that aglycosylated Fc has more flexible conformation of the CH2–CH3 interface region than the glycosylated counterpart and that the crystal packing may result in ‘closed upper CH2 region’ of aglycosylated Fc in the X-ray crystal structure [23••].

Hristodorov et al. introduced the mutation N297A and expressed six aglycosylated IgG antibodies in mammalian cells for comparison with their glycosylated counterparts. They observed that the aglycosylated IgGs were nearly identical with the glycosylated ones in terms of antigen-binding affinity, solubility, and heterogeneity [24••]. Even though aglycosylated IgGs are reported to display reduced CH2-domain thermostability and low pH-induced aggregation propensity [25, 26], they showed very similar stability to glycosylated counterparts at 4°C for four weeks and at physiological temperature (37°C) for three weeks without apparent aggregation or fragmentation. Moreover, in vivo serum half-life in rats was not significantly affected by the absence of the N-linked glycan (62 hours for the glycosylated form and 64 hours for the aglycosylated one) [24••]. In a separate study, deglycosylated human IgG1 was prepared by treatment with PNGase-F [27]. The deglycosylated antibody showed very similar secondary structure with its glycosylated form in the analysis by Fourier transform infrared spectroscopy. Additionally, no significant discrepancy of tertiary structure around the Trp residues between glycosylated and deglycosylated IgGs was observed using fluorescence spectroscopy. Although the Tm of the CH2 region for deglycosylated IgG was 6–8°C lower than glycosylated IgG, the thermal stabilities of the Fab and CH3 regions were similar, irrespective of the presence of glycans [27].

The physicochemical and biological properties of the hemi-glycosylated IgGs, which are generated by asymmetrical assembly of glycosylated half-IgGs and aglycosylated half-IgGs, have been recently elucidated. Ha et al. purified the hemi-glycosylated IgGs from the mixture containing fully glycosylated, hemi-glycosylated, and aglycosylated IgGs using cationic exchange chromatography. Both fully glycosylated and hemi-glycosylated IgGs showed the same Tm for the CH3 and Fab domains, suggesting no disturbance of thermostability. Thermostability of the hemi-glycosylated IgG CH2 domain was marginally lower than that of the glycosylated form (Tm = 70°C and 71°C, respectively). The loss of one of the dual glycans in a conventional homodimeric Fc domain reduced the binding affinity to all kinds of FcγRs (FcγRI, FcγRIIa-131R, FcγRIIa-131H, FcγRIIb, FcγRIIIa-158V, FcγRIIIa-158F, and FcγRIIIb) by a few fold [28], which was a milder decrease than that observed after full aglycosylation [6•, 20•]. The hemi-glycosylation lowered the C1q-binding affinity of approximately 20% and the ADCC activity of approximately 3.5 fold [28]. In a separate study, Shatz et al. assembled E. coli-derived aglycosylated half-IgG and CHO cell-derived glycosylated half-IgG to generate hemi-glycosylated IgG antibodies by employing the knobs-into-holes technique, which has been used to produce bispecific IgG antibody formats by introducing a bulky amino-acid mutation in one chain and a small amino-acid mutations in the other chain of Fc for the heterodimeric packing of IgG heavy chains. They found that the hemi-glycosylated IgG with a defucosylated glycan exhibited approximately 2-fold higher NK cell-mediated ADCC compared to the hemi-glycosylated IgG containing a fucosylated glycan [29].

A new biological function of a glycan-hydrolyzed IgG as a potential anti-arthritis therapeutic agent has been reported. In autoimmune disorders, the formation of immune complex (IC: aggregate of antigens and antigen-specific antibodies) is a critical step for the progression of the diseases. Streptococcus pyogenes, a Gram-positive pathogenic bacterium, releases endo-β-N-acetylglucosaminidase (EndoS) to evade from the immune surveillance by digesting a glycosidic bond between the first GlcNAc and the second GlcNAc residue of the IgG N297 glycan. In a previous study, the administration of the IgGs prepared by EndoS hydrolysis relieved the symptoms of arthritis in mice [30]. Nandakumar et al. found that EndoS-hydrolyzed IgGs blocked the progression of arthritis, regardless of the target epitope specificities, even though there was no significant change in the antigen-binding affinity and complement activation by EndoS hydrolysis. Based on the high correlation between the size reduction of ICs by administration of EndoS-hydrolyzed IgG and the suppression of inflammation, they concluded that EndoS-hydrolyzed IgGs disturbed the Fc-Fc interaction during IC formation and induced enhanced anti-inflammatory effects [31]. It is not clear which epitope of Fc is important for the Fc-Fc interaction at this stage, and the concept needs to be validated in human systems. However, the agent that either reduces the Fc-Fc packing or modifies surface-exposed deglycosylated Fc residues may enhance the disruption of IC deposition on human cartilage and would be potentially helpful to block the destruction of self-tissues/organs mediated by autoimmune disorders.

So far, based on the accumulated evidences and clinical data on several aglycosylated antibodies, no immunogenicity issues of aglycosylated antibodies have been reported (Table 1) [32]. Tolerx (MA, USA) has produced TRX-series humanized aglycosylated IgG1 antibodies (TRX1, TRX4, and TRX518) by incorporating the N297A mutation [33, 34, 35], for which phase I clinical trial was completed. TRX1 (anti-CD4) is a monoclonal IgG1κ aglycosylated antibody developed to reduce the immunogenicity for foreign antigens and treat autoimmune diseases by blocking CD4-mediated functions [33]. TRX4 (otelixizumab; anti-CD3) was designed to treat autoimmune diseases and type 1 diabetes [34]. Results from in vitro and in vivo studies indicate that the antibody blocks T cells efficiently, and its phase III clinical evaluation is currently taking place [35]. TRX518 (anti-GITR), a humanized aglycosylated IgG1, specifically binds to glucocorticoid-induced tumor necrosis factor receptor (GITR) expressed on T cells and subsequently activates tumor antigen-specific effector T cells. Phase I clinical tests of this anti-GITR antibody are ongoing [36, 37].

Alder Biopharmaceuticals (WA, USA) developed therapeutic antibody candidates for clinical use (ALD518). ALD518 (BMS-945429) is an aglycosylated humanized monoclonal antibody designed to block interleukine-6 (IL-6). This aglycosylated antibody was produced in Pichia pastoris by N297A mutation for aglycosylation, and its phase II clinical evaluation has been completed [38, 39].

Genentech (CA, USA) developed onartuzumab (anti-MET; OA-5D5), a monovalent monoclonal aglycosylated full-length antibody for the treatment of lung cancer. To generate a ‘one-arm’ monovalent format preventing MET dimerization, knobs-into-holes mutations were introduced. Onartuzumab is the first therapeutic full-length aglycosylated antibody produced in E. coli and is under clinical trials. The aglycosylated antibody was designed to block Met signaling of tumor cells by antagonistically binding to the extracellular domain of Met and inhibiting hepatocyte growth factor (HGF)-mediated activation [40]. Currently, onartuzumab is under phase II and III clinical trials for the treatment of MET-positive cancers in the US and United Kingdom, respectively.

Genzyme (MA, USA) developed an aglycosylated antibody known as GMA161, which is the humanized version of the murine antibody 3G8 targeting the human low-affinity FcγRIII (CD16). GMA161 is a potential therapeutic candidate to block CD16 for the treatment of autoimmune diseases. GMA161 has a single mutation of Asn297 to Glu (N297Q) to remove glycosylation and effector functions. Preclinical studies of GMA161 have shown the potential utility of GMA161 as a therapeutic mAb candidate for autoimmune disorders [41]. Moreover, a safety study was conducted in patients with idiopathic thrombocytopenic purpura (ITP) (Table 1).

Since 1986, when the first monoclonal antibody (OKT-3: anti-CD3) was approved from the US FDA for human therapy, the successful marketing of commercialized antibodies and the effective demonstration of clinical outcomes by targeted immunotherapy have expanded the clinical demand for monoclonal therapeutic antibodies (∼1,000 kg/year) [42]. Due to the critical roles of glycosylation in therapeutic effector functions and potential immunogenicity concerns, the expression host for the production of US FDA-approved IgG antibodies requiring effector functions has been restricted to CHO cells or mouse myeloma cell lines (NS0 or SP2/0) [43]. Despite the substantial progresses in mammalian cell culture systems that have been made since the advent of the monoclonal antibody era, alternative expression systems allowing less expensive production without issues of glycan heterogeneity have been highly desirable. As an example, the glycoengineered P. pastoris strain, developed by GlycoFi and acquired by Merck, produced human-like glycosylated IgG in up to 1.4 g/L yield [44], which is a superior volumetric productivity compared to mammalian cells, even though N-linked glycan compositions are still slightly different from native human IgGs and occasionally additional heterogeneous O-linked glycans are appended [2].

Compared to other expression systems, fast growth, fully characterized genetics, facile scale-up, and simple quality control without glycan heterogeneity issues are key advantages of the E. coli expression system in the standpoint of biomanufacturing. Since the first approval of insulin analog (Humulin®) in 1982, about 30% of over 150 protein pharmaceutics approved from the US FDA or EMEA are produced in E. coli [45]. For the production of antibody fragments, E. coli have been employed as an ideal host, as it was already shown in the two cases of approved Fab fragments, Lucentis® (ranibizumab: anti-VEGF-A for the treatment of macular degeneration) and Cimza®(certolizumab pegol: pegylated anti-TNF-α for the treatment of Crohn's disease and rheumatoid arthritis) [45]. However, it has been highly challenging to produce in E. coli full-length IgG antibodies, which are tetrameric complex proteins (∼150 kDa) containing 16 disulfide bonds and exhibiting prolonged serum half-life compared to antibody fragments. In 1984, Cabilly et al. at Genentech firstly reported on the expression of full-length aglycosylated IgG antibody recognizing carcinoembryonic antigen (CEA) in E. coli. Transformants harboring both plasmids for the expression of the IgG heavy chain and the IgG light chain were cultured. After solubilization and reduction, cell lysates were reconstituted in a refolding buffer containing reduced (GSH) and oxidized (GSSG) forms of glutathione. However, only low CEA-binding activity was observed for the assembled aglycosylated full-length IgGs [46].

It took 18 years to obtain fully antigen binding-capable aglycosylated full-length IgG antibodies in the E. coli periplasmic fraction. Yansura and coworkers, at the same company, used a monocistronic expression plasmid containing two separate PhoA promoters, StII signal sequences, and optimized translation initiation regions (TIRs) for the balanced expression of heavy and light chains in the periplasmic space of E. coli. In a 10-L fermenter, they were able to produce anti-TF (tissue factor) aglycosylated IgG antibodies up to ∼150 mg/L in 72 hours of cultivation. The purified aglycosylated antibodies showed similar binding affinity to FcRn, but no binding to C1q and FcγRI due to the absence of N-linked glycan, as expected [47]. Dramatic improvement in full-length IgG production was achieved by the co-expression of disulfide oxidoreductase (DsbA) and disulfide bond isomerase (DsbC). In high-cell density cultivation, the yield of IgG production was increased up to over 1 g/L [48].

To enhance full-length IgG production, Makino et al. optimized signal peptide sequence, host strain, TIR sequence, and type of chaperone for co-expression. After construction of a chemically mutagenized E. coli library for the analysis of IgG expression at a single cell level and high throughput screening, they employed the periplasmic expression with cytometric screening (PECS) system, which was developed for the screening of antibodies specific to small-molecule antigens in an earlier work [49]. Multiple rounds of PECS enabled to isolate a mutant E. coli strain (TM4) displaying 3-fold increased expression of cardiac digoxigenin-specific 26-10 IgG antibody, and the optimization of TIR of the second cistron allowed balanced translational levels for both the heavy and the light chains, leading to a few-fold enhanced IgG assembly. Although the beneficial effect of DsbA co-expression for enhanced full-length IgG expression in E. coli was not synergistic with using the isolated TM4 E. coli mutant strain and the optimized TIR sequence, they could produce three kinds of full-length IgG antibodies with an increased yield (up to ∼4 mg/L) in a shake-flask culture [50].

Instead of using Sec-dependent PelB signal peptides for the expression of both heavy and light chains in a dicistronic expression system [6], Jeong and coworkers used a combination of DsbA (SRP-dependent) signal peptide and PelB signal peptide for the expression of the light and heavy chains, respectively. Co-expression of a disulfide bond isomerase (DsbA) and a signal recognition particle component protein (Ffh) could improve the yield of the assembled full-length IgG in E. coli XL1-Blue [51]. In a separate study on the E. coli strain MG1655, harboring a dicistronic plasmid for the expression of both IgG heavy and light chains with PelB leader peptides, they introduced a modified 5′ untranslated region (5′ UTR) sequence that was predicted to have less propensity to form an mRNA secondary structure. By co-expression of DsbC, the yield of the full-length IgG was improved up to 362 mg/L [52]. The yield was still lower than the previously reported yield at Genentech (1,050 mg/L) [48]. However, considering the short cultivation time (22 hours) versus 70–80 hours [48], the productivity was higher (16.5 mg/L/hour) than the other reported full-length IgG production systems using E. coli (∼14 mg/L/hour) [48], mammalian cells (∼10 mg/L/hour) [4], or engineered yeasts (∼8 mg/L/hour) [53].

A refolding method to produce full-length IgG in E. coli inclusion bodies was revisited [54]. By performing consecutive steps of expression of the heavy and the light chains in E. coli inclusion bodies, solubilization, denaturation, and refolding at optimized conditions, Benhar and coworkers could produce functional fluorescent full-length aglycosylated IgG antibody fused with superfolder GFP or two tandem-repeat superfolder GFPs at each polypeptide of the IgG molecule. The refolded fluorescent aglycosylated IgG antibodies showed similar antigen-binding properties with the parental glycosylated antibodies, and their specific binding to the antigens on the surface of target cells could be analyzed by flow cytometry and confocal microscopy without additional labeling with fluorescent secondary antibodies [55]. In addition to in vivo protein expression systems, aglycosylated full-length IgGs were produced in the E. coli in vitro protein expression system, which enables more rapid production and is more likely to be adaptable to high-throughput screening. By optimization of the TIR region and co-expression of DsbC or PDI, the yield of full-length aglycosylated IgG could be increased up to 400 mg/L [56].

All the monoclonal antibodies prepared in nature are monospecific, which limits their utility to recognize only single-type antigens, even though they have two antigen-binding arms. In many diseases, multiple pathological factors are involved, and they crosstalk to each other for the progress of a disease or symptoms. For example, most cancers are caused by multifactorial oncogenes and the progression of tumors is affected by multiple signaling events. Therefore, for improved efficacy and better clinical outcomes, it is highly desired to either activate or inhibit multiple critical pathological targets simultaneously. Spiess et al. used co-culture of two different E. coli strains expressing each half of the full-length IgG. One half antibody contained a knob mutation (T366W) and the other half had the hole mutations (T366S, L368A, and Y407V) for the efficient formation of the knobs-into-holes mutations. After cell lysis, the assembled bispecific full-length aglycosylated IgG1 antibodies could be obtained. By using this simple bispecific antibody-generation method, they could produce 28 different bispecific antibodies, without light chain mispairing issues [57••]. The same technology was employed to allow E. coli to produce aglycosylated IgG4 bispecific antibodies allowing binding to both IL-4 and IL-13 simultaneously and exhibiting pharmacokinetic profiles and biodistributions in cynomolgus monkeys comparable to the aglycosylated IgG1 bispecific antibody [58]. These are excellent examples demonstrating the utility of the E. coli expression system, which can be expanded to produce highly potent dual antigen-targeting aglycosylated bispecific antibodies as well as monospecific aglycosylated antibodies.

As alternative expression systems to address glycan heterogeneity issues and broaden the availability of antibody-expression systems, aglycosylated IgG antibodies were also expressed in transgenic tobacco plants [59] or green algal chloroplasts [60]. Notably, the aglycosylated nimotuzumab (anti-EGFR), which was produced in transgenic plants and contained the N297Q mutation for preventing N-linked glycosylation, exhibited EGFR binding, blocking of EGFR-mediated signaling, pharmacokinetics, and biodistribution in rats equivalent to those of the glycosylated nimotuzumab (Table 2) [59].

The antibody Fc regions, which connect the humoral and cellular immune responses, are essential for an array of therapeutic effector functions [61]. The correlation between FcγR-binding affinity and clinical efficacy has been well documented [62, 63, 64]. In the last several years, significant progress has been made in engineering glycosylated antibody Fc domains for enhanced therapeutic effector functions [65, 66, 67, 68, 69]. Recently, outcomes from more extensive Fc engineering have been reported. Anti-CD276 Fc variant, which includes multiple mutations in the Fc domain (L235V/F243L/R292P/Y300L/P396L), exhibited improved ADCC results and reduced the size of tumor in a renal-cell carcinoma xenograft mouse model [68]. Anti-Her2 Fc variant, which contains the same Fc mutations of Anti-CD276 Fc variant, induced significantly improved ADCC effects even in medium or low Her2-expressing cancer cells [69]. An Fc engineering technique comprising multiple rounds of ribosome display and screening was used to isolate an Fc mutant (F243L/T393A/H433P) exhibiting improved FcγRIIIa-binding efficiency and ADCC [70].

In parallel with engineering glycosylated Fc, engineering aglycosylated Fc has been also successfully performed. Although wild-type aglycosylated IgGs show almost no engagement of Fcγ receptors and elicit no effector functions, recent protein engineering efforts allowed aglycosylated Fc to restore or even improve its binding to FcγRs compared to glycosylated IgG Fc. Therefore, the use of aglycosylated antibodies is not restricted to the only cases that effector functions are not essential. For the engagement of FcγRIIa, Sazinsky et al. constructed a library composed of three sub-libraries-randomized amino acids (296–299, 297–299, and 297–300) in the C′/E loop containing the N-linked glycan site Asn297. Display of full-length IgGs by capturing secreted fluorescein-specific IgGs on the surface of FITC-decorated yeast, followed by flow cytometry screening, enabled the isolation of aglycosylated IgG Fc variants exhibiting comparable binding to activating FcγRIIa and inhibitory FcγRIIb with glycosylated IgG Fc. The aglycosylated IgG Fc variant (S298G/T299A) prepared in HEK293 cells could induce FcγRIIa-mediated platelet clearance in a human FcγRIIa transgenic mouse model (Figure 2a) [20].

Glycosylated human IgG binds to all kinds of human FcγRs. In contrast, selective FcγR engagement is especially critical for high efficacy of the anti-tumor therapeutic antibodies, which should selectively trigger the activation of immune leukocytes (NK cells, macrophages, and dendritic cells) by binding to activating FcγRs (FcγRI, FcγRIIa, and FcγRIIIa) instead of triggering inhibitory FcγRIIb-mediated signaling. The use of a bacterial display system for capturing aglycosylated homodimeric Fc fragments in E. coli spheroplasts and high-throughput FACS screening of an error-prone PCR aglycosylated Fc library enabled the isolation of the aglycosylated Fc variant conferring selective high binding affinity to FcγRI without exhibiting significant binding to other FcγRs such as FcγRIIa, FcγRIIb, and FcγRIIIa, in sharp contrast to glycosylated Fc domains. The aglycosylated trastuzumab-Fc5, which contained two isolated mutations (E382V/M428I) and was produced in E. coli, induced potent dendritic cells (DCs)-mediated effector function, which was not observed on using glycosylated IgGs produced in mammalian cells (Figure 2b) [6]. Georgiou and coworkers proposed that the two mutations identified in the CH3 region (E382V/M428I), which are located distant from the known FcγRI-binding sites of Fc (B/C, C′/E, and F/G loops of CH2 region), might stabilize the conformation of the flexible upper CH2 region of aglycosylated Fc [6•, 71]. This hypothesis was tested by SAXS experiments, with which the radii of gyration (Rg) of three samples (wild-type aglycosylated Fc, aglycosylated-Fc5 (E382V/M428I), and glycosylated Fc. Aglycosyalted-Fc5) were obtained and an intermediate radius of gyration between wild-type aglycosylated Fc and glycosylated Fc was measured, suggesting the change of conformation of CH2 by introducing two identified mutations (E382V/M428I) [23••]. Recently, by further evolution of the upper CH2 region, we isolated an Fc variant exhibiting higher affinity to FcγRI than the previously isolated Fc5 mutant (E382V/M428I) without interfering with the pH-dependent FcRn binding, which is essential for prolonged serum half-life of IgG molecules. Similar to the Fc5 variant, the isolated mutant showed specific binding to FcγRI, with no significant binding to other FcγRs (FcγRIIa, FcγRIIb, and FcγRIIIa) [72].

Beyond restoring the FcγR-binding affinity for aglycosylated IgG Fc to the comparable level with glycosylated IgG Fc, even much higher FcγR-binding affinity and selectivity could be achieved [19••]. ADCP (antibody dependent cell-mediated phagocytosis) is one of the main effector functions for the clearance of aberrant cells. FcγRIIa on the surface of macrophages binds tumor cell-antibody immune complexes, which are generated by antibody opsonization, activating macrophages for the phagocytosis of antibody-decorated tumor cell immune complexes. Because macrophages express inhibitory FcγRIIb, which downregulates effector functions as well as activates FcγRIIa on their surface, selective binding to activating FcγRIIa over FcγRIIb is critical for improved ADCP activity. However, engineering the Fc domain for selective FcγRIIa engagement is very difficult due to the high sequence identity (96%) between the two FcγRs (FcγRIIa and FcγRIIb). To isolate Fc variants exhibiting higher affinity and selectivity to activating FcγRIIa over inhibitory FcγRIIb, Jung et al. developed a covalently anchored full-length IgG display system for tethering the N-terminal region of one of the light chains in a full-length IgG onto the E. coli inner membrane. By anchoring the antigen-binding region onto the inner membrane, soluble Fc-binding ligands are more accessible to the Fc region of the immobilized full-length IgG. Bacterial display of aglycosylated IgG Fc variants, competitive binding of fluorescent dye-labeled FcγRIIa and nonfluorescent FcγRIIb, and flow cytometry-mediated high-throughput library screening allowed the isolation of an aglycosylated IgG Fc variant, Fc1004, containing five mutations (S298G/T299A/N390D/E382V/M428L) (Figure 2c). Compared to clinical-grade trastuzumab (Herceptin®), aglycosylated trastuzumab-Fc1004 (AglycoT-Fc1004) exhibited over 160-fold superior FcγRIIa-R131-binding affinity and 25-fold improved selectivity to FcγRIIa-131R over FcγRIIb. When using Her2-expressing tumor cells, SkOv3 (medium level Her2) or MDA-MB475 (low level Her2), AglycoT-Fc1004 showed markedly enhanced ADCP activity [19••] compared to either clinical-grade glycosylated trastuzumab (Herceptin) or glycosylated trastuzumab-G236A containing the previously identified mutation for the selective engagement of FcγRIIa over FcγRIIb [73].

Section snippets

Conclusions

Aglycosylated antibodies are almost identical to their glycosylated counterparts in terms of binding to target antigens, pharmacokinetics, and biodistributions. Currently, several aglycosylated antibodies are under clinical trials and no immunogenicity issues have been reported. The absence of the N-linked glycan at Asn297 does not affect the structure of Fab domains [8] and is not critical for the stability of IgG antibodies under physiological conditions [24••]. Low pH-induced aggregation,

Conflicts of interest

The authors declare no conflicts of interest regarding the preparation and submission of this manuscript.

References and recommended reading

Papers of particular interest, published within the period of review, have been highlighted as:

  • • of special interest

  • •• of outstanding interest

Acknowledgements

Research on aglycosylated antibody engineering has been supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (2013R1A1A1004576), and a grant from the National R&D Program for Cancer Control, Ministry of Health and Welfare, Republic of Korea (1420160). Man-Seok Ju was supported by BK21 Plus from the Ministry of Education of Korea. We thank Dr. Tae Hyun Kang for his valuable comments on

References (78)

  • D. Loo et al.

    Development of an Fc-enhanced anti-B7-H3 monoclonal antibody with potent antitumor activity

    Clin Cancer Res

    (2012)
  • S.T. Jung et al.

    Bypassing glycosylation: engineering aglycosylated full-length IgG antibodies for human therapy

    Curr Opin Biotechnol

    (2011)
  • N. Tokuriki et al.

    Protein dynamism and evolvability

    Science

    (2009)
  • Particia Rao DR, Winsor-Hines D: Inducing tolerance in primates. US patent...
  • D. Hristodorov et al.

    With or without sugar? (A)glycosylation of therapeutic antibodies

    Mol Biotechnol

    (2013)
  • D.S. Dimitrov

    Therapeutic proteins

    Methods Mol Biol

    (2012)
  • F. Li et al.

    Cell culture processes for monoclonal antibody production

    MAbs

    (2010)
  • T.S. Raju

    Assessing Fc glycan heterogeneity of therapeutic recombinant monoclonal antibodies using NP-HPLC

  • S.T. Jung et al.

    Aglycosylated IgG variants expressed in bacteria that selectively bind FcγRI potentiate tumor cell killing by monocyte-dendritic cells

    Proc Nat Acad Sci U S A

    (2010)
  • A.F. Labrijn et al.

    When binding is enough: nonactivating antibody formats

    Curr Opin Immunol

    (2008)
  • A. Lux et al.

    Impact of differential glycosylation on IgG activity

  • R.M. Anthony et al.

    Novel roles for the IgG Fc glycan

    Ann N Y Acad Sci

    (2012)
  • S.T. Jung

    Tailoring immunoglobulin Fc for highly potent and serum-stable therapeutic antibodies

    Biotechnol Bioproc Eng

    (2013)
  • A. Loos et al.

    IgG-Fc glycoengineering in non-mammalian expression hosts

    Arch Biochem Biophys

    (2012)
  • C. Ferrara et al.

    Unique carbohydrate–carbohydrate interactions are required for high affinity binding between FcγRIII and antibodies lacking core fucose

    Proc Nat Acad Sci U S A

    (2011)
  • W. Huang et al.

    Chemoenzymatic glycoengineering of intact IgG antibodies for gain of functions

    J Am Chem Soc

    (2012)
  • R.M. Anthony et al.

    Recapitulation of IVIG anti-inflammatory activity with a recombinant IgG Fc

    Science

    (2008)
  • R.M. Anthony et al.

    Identification of a receptor required for the anti-inflammatory activity of IVIG

    Proc Nat Acad Sci U S A

    (2008)
  • T.C. Nesspor et al.

    Avidity confers FcγR binding and immune effector function to aglycosylated immunoglobulin G1

    J Mol Recognit

    (2012)
  • S.T. Jung et al.

    Effective phagocytosis of low Her2 tumor cell lines with engineered, aglycosylated IgG displaying high FcγRIIa affinity and selectivity

    ACS Chem Biol

    (2013)
  • S.L. Sazinsky et al.

    Aglycosylated immunoglobulin G1 variants productively engage activating Fc receptors

    Proc Natl Acad Sci U S A

    (2008)
  • R. Jefferis

    Glycosylation of recombinant antibody therapeutics

    Biotechnol Progr

    (2005)
  • R. Jefferis

    Glycosylation as a strategy to improve antibody-based therapeutics

    Nat Rev Drug Discov

    (2009)
  • M.J. Borrok et al.

    Revisiting the role of glycosylation in the structure of human IgG Fc

    ACS Chem Biol

    (2012)
  • D. Hristodorov et al.

    Generation and comparative characterization of glycosylated and aglycosylated human IgG1 antibodies

    Mol Biotechnol

    (2013)
  • M.A. Alsenaidy et al.

    High-throughput biophysical analysis and data visualization of conformational stability of an IgG1 monoclonal antibody after deglycosylation

    J Pharm Sci

    (2013)
  • R.F. Latypov et al.

    Elucidation of acid-induced unfolding and aggregation of human immunoglobulin IgG1 and IgG2 Fc

    J Biol Chem

    (2012)
  • K. Zheng et al.

    The impact of glycosylation on monoclonal antibody conformation and stability

    MAbs

    (2011)
  • S. Ha et al.

    Isolation and characterization of IgG1 with asymmetrical Fc glycosylation

    Glycobiology

    (2011)
  • Cited by (49)

    • Production of antibodies in SHuffle Escherichia coli strains

      2021, Methods in Enzymology
      Citation Excerpt :

      Given the importance of these effector functions in the development of therapeutic mAbs, and the potential for increased immunogenicity of aglycosylated antibodies, the use of these molecules as therapeutics was initially questioned (Ju & Jung, 2014; Jung et al., 2011). However, multiple aglycosylated antibodies have now been evaluated in clinical trials, reducing concerns regarding their potential immunogenicity (Ju & Jung, 2014; Jung et al., 2011). Further, amino acid mutations to the Fc region of IgG have been identified which rescue the ability to recruit specific Fc γ receptors (FcγR), allowing certain receptors and effector functions to be targeted in a surgical manner for desired in vivo outcomes (Chen et al., 2017; Jo, Kwon, Lee, Lee, & Jung, 2018; Jung et al., 2010, 2013; Sazinsky et al., 2008; Yoon et al., 2019).

    • Considerations for the Design of Antibody-Based Therapeutics

      2020, Journal of Pharmaceutical Sciences
    View all citing articles on Scopus
    View full text