If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Lec13 cells, a variant Chinese hamster ovary cell line, were used to produce human IgG1 that were deficient in fucose attached to the Asn297-linked carbohydrate but were otherwise similar to that found in IgG1 produced in normal Chinese hamster ovary cell lines and from human serum. Lack of fucose on the IgG1 had no effect on binding to human FcγRI, C1q, or the neonatal Fc receptor. Although no change in affinity was found for the His131 polymorphic form of human FcγRIIA, a slight improvement in binding was evident for FcγRIIB and the Arg131 FcγRIIA polymorphic form. In contrast, binding of the fucose-deficient IgG1 to human FcγRIIIA was improved up to 50-fold. Antibody-dependent cellular cytotoxicity assays using purified peripheral blood monocytes or natural killer cells from several donors showed enhanced cytotoxicity, especially evident at lower antibody concentrations. When combined with an IgG1 Fc protein variant that exhibited enhanced antibody-dependent cellular cytotoxicity, the lack of fucose was synergistic.
neonatal Fc receptor
antibody-dependent cellular cytotoxicity
Chinese hamster ovary
enzyme-linked immunosorbent assay
natural killer cell
peripheral blood monocytic cell
The nature and importance of the conserved, Asn297-linked carbohydrate in influencing immunoglobulin G effector functions has long been recognized (
). For human IgG the core oligosaccharide normally consists of Asn297-Nδ2- GlcNAc(fucose)-GlcNAc-mannose-(mannose-GlcNAc)2(Fig. 1). Variation among individual IgG includes attachment of galactose and/or galactose-sialic acid at one or both of the terminal GlcNAc and/or attachment of a third GlcNAc arm (bisecting GlcNAc). The presence or absence of terminal galatose or sialic acid residues affects IgG function (
). Some of these (e.g. mAbs that bind to a receptor or soluble ligand and thereby block ligand-receptor interaction) may function without utilizing antibody effector mechanisms. Abrogating immune system recruitment for these mAbs can be achieved by altering IgG residues in the lower hinge region (
), using IgG2 or IgG4 subclasses, which are relatively less efficient in effector function, or using F(ab) or F(ab′)2 fragments (although these may have undesirable clearance rates). Other mAbs may need to recruit the immune system to kill the target cell (
). In those circumstances where recruitment of immune effector cells is desirable for therapeutic mAb efficacy, engineering the IgG Fc portion to improve effector function (via improved binding to IgG receptors and/or complement) could be a valuable enhancement to antibody therapeutics. Currently, methods that improve the immune system recruitment include bispecific antibodies in which one arm of the antibody binds to an IgG receptor (
) was utilized to express human IgG1. This Chinese hamster ovary (CHO) cell line is deficient in its ability to add fucose but provided IgG with oligosaccharide that was otherwise similar to that found in normal CHO cell lines and from human serum (
). The resultant IgGs were used to evaluate the effect of fucosylated carbohydrate on antibody effector functions, including binding to human FcγR, human C1q, human FcRn, and ADCC using human effector cells.
) with the modification of a puromycin site fused 5′ to the dihydrofolate reductase (DHFR) site to form a fusion gene. Puromycin is used as a selective marker in DHFR(+) cells such as the Lec13 cells and the DHFR site was retained for methotrexate amplification of the stable cell line.
Transfection and Culturing of Lec13 and Wild-type CHO Cells
The CHO cell line Pro-Lec13.6a (Lec13), was obtained from Professor Pamela Stanley of Albert Einstein College of Medicine of Yeshiva University. Parental lines were Pro− (proline auxotroph) and GAT− (glycine, adenosine, thymidine auxotroph). The CHO-DP12 cell line used for wild-type antibodies is a derivative of the CHO-K1 cell line (ATCC number CCL-61) which is DHFR-deficient and has a reduced requirement for insulin. Cell lines were transfected with cDNA using the Superfect method (Qiagen, Valencia, CA). Selection of stably transfected Lec13 cells expressing mAbs was performed using puromycin dihydrochloride (Calbiochem) at 10 μg/ml in growth medium containing minimal essential medium-α medium with l-glutamine, ribonucleosides, and deoxyribonucleosides (Invitrogen), supplemented with 10% inactivated fetal bovine serum (Invitrogen), 10 mm HEPES, and 1× penicillin/streptomycin (Invitrogen). Selection of wild-type CHO-DP12 cells was performed in growth medium containing Ham's F-12 without GHT (glycine, hypoxanthine, thymidine), low glucose Dulbecco's modified Eagle's medium without glycine with NaHCO3, supplemented with 5% dialyzed fetal bovine serum (Invitrogen), 10 mm HEPES, 2 mml-glutamine, and penicillin/streptomycin. Colonies formed within 2–3 weeks and were pooled for expansion. The pools were seeded initially at 3 × 106 cells/10-cm plate for small batch protein expression. The cells were converted to serum-free medium once they grew to 90–95% confluence, and after 3–5 days cell supernatants were collected and quantified for mAb expression using both an anti-IgG-Fc and an anti-intact IgG ELISA.
Protein Expression and Analysis
For each batch of mAb, Lec13 and wild-type CHO cells were seeded at ∼8 × 106 cells/15-cm plate 1 day prior to converting to PSO4 (
). Cells remained in serum-free production medium for 3–5 days. Supernatants were collected and clarified by centrifugation in 150-ml conical tubes to remove cells and debris. The protease inhibitors phenylmethylsulfonyl fluoride and aprotinin (Sigma) were added, and the supernatants were concentrated 5-fold on stirred cells using MWCO30 filters (Amicon, Beverly, MA) prior to immediate purification using protein A chromatography (Amersham Biosciences). All proteins were buffer-exchanged into phosphate-buffered saline using Centriprep-30 concentrators (Amicon) and analyzed by SDS-polyacrylamide gel electrophoresis. Protein concentrations were determined using A280 and verified using amino acid composition analysis. On average, the Lec13 cells generated 10 μg of mAb per 15-cm plate; expression in control CHO cells for all antibodies was 4–5 times higher than in the Lec13 cells. Antibodies generated from CHO-DP12 grown on plates will be denoted as CHO-P.
CHO-DP12 cells were also grown in spinner flasks. Cells were seeded at 6 × 105 cells/ml and grown at 37 °C for 2 days. On the third day, the temperature was shifted to 33 °C, and the cells were allowed to grow until viability dropped to 70% due to the pH dropping to ∼6.5. Antibodies derived from CHO-DP12 cells grown in spinner flasks will be denoted as CHO-S.
Matrix-assisted laser desorption/ionization time-of-flight mass spectral analysis was used to characterize the Asn297-linked oligosaccharides on the mAbs following a previously described protocol (
Immunofluorescence staining of purified NK cells from an FcγRIIIA(Phe158/Phe158) donor was performed using Hu4D5 variants. NK cells were purified by negative selection and gated with phycoerythrin-conjugated anti-CD56 and fluorescein isothiocyanate-conjugated anti-CD16 (Pharmingen, San Diego, CA). Cells were adjusted to 2 × 106/ml and incubated with 2 μg/ml Hu4D5 variants for 30 min in ice-cold phosphate-buffered saline containing 0.1% bovine serum albumin, 0.01% sodium azide. Cells were then washed, and antibody binding was detected by incubating with fluorescein isothiocyanate-conjugated F(ab′)2 goat anti-human IgG for 30 min.
) expressed in stably transfected Lec13 cells consistently had about 10% fucosylated carbohydrate (TableI). In contrast, Hu4D5 mAbs from stably transfected CHO-DP12 cells cultured in spinner flasks (CHO-S) or plates (CHO-P) showed about 98% fucosylated carbohydrate. The variation in galactosylation for the Lec13 mAbs was similar to that found for CHO-S mAbs (as opposed to CHO-P) (Table I); hence, CHO-S mAbs were used as reference in the binding assays.
Hu4D5, humanized anti-p185HER2 IgG1 (40); HuE27, humanized anti-IgE IgG1 (41); CHO-S, IgG expressed by CHO cells in spinner flasks; CHO-P, IgG expressed by CHO cells on 15-cm plates; Lec, IgG expressed by Lec13 cells on 15-cm plates; HEK, IgG expressed by human embryonic kidney 293 cells on 15-cm plates; AAA, S298A/E333A/K334A IgG1 variant (12). For Lec-X and Lec-AAA-X, X represents independently expressed lots of IgG.
Values are percentage of total oligosaccharide. Glycoforms were determined by matrix-assisted laser desorption/ionization time-of-flight mass spectral analysis following a previously described protocol (12, 44). Man, incompletely processed, hypermannosylated oligosaccharides containing 5–8 mannose units (19). Gal0, Gal1, Gal2 are oligosaccharides with no (agalactosyl), one (monogalactosyl), or two (digalactosyl) galactose residues covalently linked to the terminal GlcNAc residues (see Fig. 1). Mean ± S.D. is provided for six independently expressed lots of Hu4D5-Lec13 (A–E), three independently expressed lots of Hu4D5 Lec13-AAA (A–C), and three independently expressed lots of HuE27-HEK293. None of the mAbs contained measurable triantennary oligosaccharide (bisecting GlcNAc), and sialylated oligosaccharide was less than 6% of total oligosaccharide.
); CHO-S, IgG expressed by CHO cells in spinner flasks; CHO-P, IgG expressed by CHO cells on 15-cm plates; Lec, IgG expressed by Lec13 cells on 15-cm plates; HEK, IgG expressed by human embryonic kidney 293 cells on 15-cm plates; AAA, S298A/E333A/K334A IgG1 variant (
). For Lec-X and Lec-AAA-X, X represents independently expressed lots of IgG.
1-b Values are percentage of total oligosaccharide. Glycoforms were determined by matrix-assisted laser desorption/ionization time-of-flight mass spectral analysis following a previously described protocol (
). Gal0, Gal1, Gal2 are oligosaccharides with no (agalactosyl), one (monogalactosyl), or two (digalactosyl) galactose residues covalently linked to the terminal GlcNAc residues (see Fig. 1). Mean ± S.D. is provided for six independently expressed lots of Hu4D5-Lec13 (A–E), three independently expressed lots of Hu4D5 Lec13-AAA (A–C), and three independently expressed lots of HuE27-HEK293. None of the mAbs contained measurable triantennary oligosaccharide (bisecting GlcNAc), and sialylated oligosaccharide was less than 6% of total oligosaccharide.
Monomeric Hu4D5-Lec13 IgG1 bound to human FcγRI equivalent to binding of Hu4D5-CHO-S (Table II). The EC50 values (∼0.1 μg/ml) correspond to an affinity constant of 0.7 nm, at the lower end of the range of previously determined values (
). Binding of Hu4D5-Lec13 dimers to the FcγRIIA(Arg131) polymorphic form and to FcγRIIB exhibited only a slight improvement compared with Hu4D5-CHO-S (Fig.2, A and B). Lack of fucose did not affect binding to the FcγRIIA(His131) polymorphic form (Fig. 2C). Note that in contrast to FcγRI and FcγRIIIA, binding of the antibodies to FcγRIIA and FcγRIIB did not reach a plateau at the highest concentration used (10 μg/ml) (Fig. 2, A and B); hence, EC50 values could not be calculated. At concentrations above 10 μg/ml, aggregation of mAb was problematic.
Table IIBinding of antibodies to human FcγRI and FcγRIIIA
Hu4D5, humanized anti-p185HER2 IgG1 (40); HuE27, humanized anti-IgE IgG1 (41); CHO-S, IgG expressed by CHO cells in spinner flasks; CHO-P, IgG expressed by CHO cells on 15-cm plates; Lec13, IgG expressed by Lec13 cells on 15-cm plates; HEK293, IgG expressed by human embryonic kidney 293 cells on 15-cm plates; AAA, S298A/E333A/K334A IgG1 variant (12).
); CHO-S, IgG expressed by CHO cells in spinner flasks; CHO-P, IgG expressed by CHO cells on 15-cm plates; Lec13, IgG expressed by Lec13 cells on 15-cm plates; HEK293, IgG expressed by human embryonic kidney 293 cells on 15-cm plates; AAA, S298A/E333A/K334A IgG1 variant (
Human FcγRIIIA has naturally occurring allotypes at position 48 (Leu, His, or Arg) and at position 158 (Val or Phe); the FcγRIIIA(Val158) allotype interacts with human IgG better than the FcγRIIIA(Phe158) allotype (
). Both the Phe158 and Val158 polymorphic forms of FcγRIIIA exhibited significantly improved binding to IgG1 that lacked fucose. Binding of dimeric Hu4D5-Lec13 to FcγRIIIA(Phe158) showed at least a 42-fold improvement (Fig. 3A; Table II), and binding to FcγRIIIA(Val158) showed a 19-fold improvement over Hu4D5-CHO-S (Fig. 3B; Table II). HuE27-Lec13 dimers also exhibited a 50-fold enhancement in binding to both polymorphic forms of FcγRIIIA compared with HuE27 CHO-P (Fig.4A; Table II). The binding of HuE27 expressed in HEK293 cells was intermediate to HuE27-Lec13 and CHO-P (Fig. 4A; Table II) and correlates to the percentage of HuE27 mAb without fucose: CHO-P, 2%; HEK293, 16%; Lec13, 78% (Table I). The EC50 values for Hu4D5-Lec13 and HuE27-Lec13 (as dimers) were 0.24 and 0.19 μg/ml, respectively, for FcγRIIIA(Phe158) and 0.07 and 0.05 μg/ml, respectively, for FcγRIIIA(Val158) (Table II), showing that the binding of the Lec13 antibodies was independent of antibody specificity.
In a previous study of the effect of protein sequence variants of human IgG1 on binding to human FcγR, an S298A/E333A/K334A-IgG1 variant showed improved binding to FcγRIIIA as well as more potent cytotoxicity in ADCC assays (
). Expressing Hu4D5 S298A/E333A/K334A-IgG1 in Lec13 cells added fucose deficiency to this protein-sequence variant (Table I). When assayed as dimers, the Hu4D5 S298A/E333A/K334A-IgG1 with fucose exhibited 36- and 12-fold enhancement in binding to FcγRIIIA(Phe158) and FcγRIIIA(Val158), respectively. The same variant without fucose showed even more improvement in binding to FcγRIIIA(Phe158) and FcγRIIIA(Val158) of 63- and 33-fold, respectively (Fig. 3, A and B; Table II), suggesting that the protein and carbohydrate alterations are synergistic.
In addition to dimers, complexes consisting of three anti-IgE HuE27 and three IgE were used to test binding to FcγRIIIA; these complexes are trimeric in anti-IgE HuE27 (
). Binding to FcγRIIIA by the trimeric complexes was stronger than for the corresponding dimeric complexes (Fig. 4, A and B; Table II), suggesting an appreciable avidity component. The avidity resulted in a smaller improvement in binding to FcγRIIIA(Phe158) for trimeric HuE27-Lec13, only 13-fold, compared with 53-fold for dimeric HuE27. In a previous study, the improvement in binding of trimeric HuE27 S298A/E333A/K334A-IgG1 variant to FcγRIIIA(Phe158) and FcγRIIIA(Val158) was 2- and 1.1-fold, respectively; whereas the improvement in binding was small, the effect on ADCC was significant (
). In the current study, the trimeric complex of HuE27 S298A/E333A/K334A-IgG1 showed improved binding to both FcγRIIIA polymorphic forms in line with the values from the previous study (Fig.4, B and C; Table II).
The improved binding of fucose-deficient IgG1 to FcγRIIIA was confirmed with FcγRIIIA(Phe158) full-length α-chain co-expressed with γ-chain on CHO cells (Fig.5; Table II). As for the α-chain fusion protein alone in ELISA format, fucose deficiency was synergistic with the S298A/E333A/K334A-IgG1 variant.
Binding of the native and fucose-deficient IgG1 to murine FcγRII and FcγRIII was also evaluated. As trimeric complexes, the fucose-deficient HuE27 showed slightly reduced binding to both receptors compared with fucosylated HuE27 (data not shown). The major histocompatibility complex class I-like neonatal Fc receptor (FcRn) is structurally unrelated to the FcγR (
). SK-BR-3 cells were opsonized with varying Hu4D5 concentrations, and PBMCs from three FcγRIIIA(Val158/Phe158) donors and three FcγRIIIA(Phe158/Phe158) donors were used as effector cells at a 30:1 effector/target ratio; representative ADCC assays are shown in Fig. 6, Aand B. For all donors studied, the fucose-deficient IgG1 exhibited significant improvement in ADCC compared with IgG1 with fucose, especially at the lower antibody concentrations (Fig. 6).
To verify that the combination of fucose deficiency and protein sequence variant exhibited a synergistic effect, as seen in the ELISA and cell binding assays (Figs. Figure 3, Figure 4, Figure 5), immunofluorescence staining and ADCC assays were performed comparing native Hu4D5 and S298A/E333A/K334A-IgG1 variant with and without fucose against NK cells from two FcγRIIIA(Phe158/Phe158) donors. The intensity of staining showed the pattern: Hu4D5 S298A/E333A/K334A-IgG1 Lec13 > Hu4D5 Lec13 > Hu4D5 CHO-S (Fig.7A). For the ADCC assays, SK-BR-3 cells were opsonized with 1 ng/ml mAb, and NK cells were used as effector cells. Both donors showed similar results; the fucose-deficient S298A/E333A/K334A-IgG1 was more potent than either fucose-deficient Hu4D5 or fucosylated S298A/E333A/K334A-IgG1 (Fig. 7B) and correlated with the immunofluorescence staining.
Although the presence of carbohydrate is necessary for binding to FcγRI (
), the equivalent binding of IgG1 regardless of differences in fucose content (Lec13 versus CHO-S) or galactose content (CHO-P versus CHO-S) suggests that IgG1 binding to FcγRI is insensitive to the presence of these moieties on the carbohydrate. The effect of galactosylation on binding of IgG to human FcγRI has been previously studied (
The small improvement in binding of fucose-deficient IgG1 to both FcγRIIA(Arg131) and FcγRIIB, each having arginine at position 131, versus no effect on FcγRIIA(His131) suggests that the fucose may either directly interact with the FcγR residue at position 131 or alter the IgG1 conformation so as to effect a subtle, negative influence on binding when arginine is present at FcγR position 131. As with FcγRI, the IgG1 galactose content did not seem to affect binding to FcγRII.
In contrast to the FcγRII class of receptors, lack of fucose effected a significant improvement in binding of IgG1 to FcγRIIIA. The absence of fucose not only increased binding of native IgG1 to FcγRIIIA but also augmented binding of the S298A/E333A/K334A-IgG1 variant. For both protein and carbohydrate variants of HuE27 (i.e.S298A/E333A/K334A-IgG1 and Lec13-derived), the improvement in binding of the trimeric complex was much smaller than that observed for the dimeric complex. For example, binding to FcγRIIIA(Phe158) by HuE27-Lec13 (21% fucosylated) showed a 53-fold improvement compared with HuE27-CHO-P (98% fucosylated) when in dimeric form but only a 13-fold improvement when in the trimeric form (Table II). This may result from the avidity effect evident when comparing binding of the dimeric and trimeric complexes (Fig. 4, A and B) and suggests that as the size of the immune complex is increased, the fucose effect diminishes.
Comparison of the carbohydrate found on native IgG1 for the Hu4D5 Lec13-produced and CHO-S-produced antibodies showed no appreciable differences in the extent of galactosylation, and hence the difference in binding can be attributed solely to the presence/absence of fucose. However, for the S298A/E333A/K334A IgG1 variant, the Lec13-produced and HEK293-produced antibodies also showed variation in galactosylation. While this complicates the interpretation of the data, the combination of protein sequence variation and lack of fucose was synergistic in binding to recombinant FcγRIIIA, native FcγRIIIA expressed on transfected CHO cells or on purified NK cells, and in the ADCC activity of FcγRIIIA-expressing NK and PBMC cells.
A previous study of protein sequence variants of human IgG1 found that alanine (and other) substitutions at some Fc positions could reduce or improve binding to FcγR as well as enhance ADCC (
). For example, of the three alanine substitutions S298A/E333A/K334A used in this study, only Ser298 is at the Fc/FcγRIIIA interface in the crystal structures. Likewise, in the co-crystal structures, neither of the fucose residues on the two Fc heavy chains interact with FcγRIIIA. Inspection of crystal structures of human and rodent Fc or IgG shows that the fucose can adopt varying conformations and exhibits high B-factors (
), suggesting a high degree of mobility. Although the presence or absence of fucose definitely affects the interaction of an IgG with FcγRIIIA (and, to a lesser extent, FcγRIIA(Arg131) and FcγRIIB), the current study cannot discern whether this is due to direct interaction of the fucose with receptor or to a conformational influence on the IgG itself.
Improved binding to FcγRIIIA translated into improved ADCC in vitro. Notably, for all donors the enhancement in cytotoxicity was more apparent as the concentration of antibody was reduced. This may reflect the larger improvement in binding seen for dimers compared with that for trimers (i.e. fucose-deficient IgG1 may require fewer mAbs on the surface of the target cell in order to effect binding/activation of an effector cell). If this is the case, it suggests that for therapeutic antibodies that utilize ADCC, nonfucosylated antibody could conceivably be given at lower doses to effect an equivalent cell kill as higher doses of fucosylated antibody.
In addition to the human FcγR, the influence of fucose on binding to C1q and FcRn were also investigated. Previous studies have shown that differences in IgG glycosylation can affect C1q binding, although the results of this study show that the presence/absence of fucose had no discernible effect. Likewise, the presence or absence of fucose did not affect IgG1 binding to FcRn. This is not surprising, since aglycosylated IgG1 binds this receptor similarly to glycosylated IgG1 (
) tested the ADCC function of monoclonal murine IgG purified from hybridomas treated with glycosidase inhibitors that acted at different stages in the carbohydrate processing pathway. Treatment with castanospermine, which inhibits removal of glucose residues from the nascent oligosaccharide (
), generated IgG that showed enhanced ADCC mediated by NK cells, which express only FcγRIIIA, but not by other types of effector cells such as monocytes. Lectin-binding analysis suggested that the castanospermine-treated IgG lacked fucose (
), not routinely found on IgG secreted from nontreated cells or from human serum. In the current study, Lec13 cells were utilized to produce fucose-deficient IgG1. Other than the lack of fucose, the oligosaccharide on the IgG1 was similar to that found in normal CHO cell lines and from human serum (
) and did not suffer from the hypermannosylation or hyperglucosylation resulting from the use of castanospermine. The improvement in FcγRIII binding and ADCC by the fucose-deficient IgG1 suggests that it is possible to generate monoclonal antibody with otherwise normal oligosaccharide that would have enhanced effector function.
Normal CHO and HEK293 cells add fucose to IgG oligosaccharide to a high degree (80–98% in this study). IgGs from sera are also highly fucosylated (
). Unfortunately, the Lec13 cell line is not robust enough to consider as a production cell line, because the expression levels of the anti-HER2 Hu4D5 and anti-IgE HuE27 antibodies in this cell line were low compared with the standard CHO cells used. Treatment of fucosylated IgG with fucosidases (