due to degradative processes such as asparagine deamidation or aspartate isomerization. Comprehensive characterization requires elucidation of the sources ...
Mire-Sluis AR (ed): State of the Art Analytical Methods for the Characterization of Biological Products and Assessment of Comparability. Dev BioI (Basel). Basel, Karger, 200S, vol 122, pp 117-] 27.
..................
Heterogeneity of Recombinant Antibodies: linking Structure to Function RJ Harris Analytical Chemistry Department, Genentech, Inc., South San Francisco, CA, USA
Abstract: Structural heterogeneity of recombinant IgG1 antibodies derives from variations in conserved as well as unique structural features. Common sources of heterogeneity include Fc glycosylation, partial heavy chain C-terminal Lys processing, Fc methionine oxidation, hinge-region cleavage, and the glycation of Lys residues. Aspartate residues that are isomerized to iso-aspartate were detected by cation exchange or hydrophobic interaction chromatography for trastuzumab and omalizumab, respectively. Unpaired cysteines were detected in omalizumab using Ellman's reagent, with the thiol-containing Fab resolved using hydrophobic interaction chromatography after papain digestion. Structural variations that cause chromatographic resolution may indicate the presence of a form with reduced potency.
INTRODUCTION Antibodies have conserved as well as unique structural features. The conserved features of IgGl-cIass antibodies contribute to the overall molecular heterogeneity, particularly with respect to the glycosylation of a CH2 domain Asn residue [1],partial carboxypeptidase processing of heavy chain C-terminal Lys residues [2], oxidation of susceptible methionine residues in the Fc region [3],hinge-region fragmentation, and glycation.The unique structural determinants found in the complementarity-detemrining regions confer binding specificity,and may also contribute to the overall heterogeneity due to degradative processes such as asparagine deamidation or aspartate isomerization. Comprehensive characterization requires elucidation of the sources of heterogeneity, with additional studies to determine the potency and/or pharmacokinetic impacts of each modification. Contradictory conclusions may be drawn from published studies for some conserved modifications, particularly glycosylation, requiring new studies with candidate therapeutic antibodies to draw a final conclusion. Several sources of heterogeneity, and methods for their detection are reviewed in this report.
117
METHODS Hinge-region
proteolysis
detection
using SEC
An omalizumab sample was exchanged into 10 mM NaOAc, 140 mM NaCI, 0.05% NaN3, pH 5.2, and its concentration was adjusted to 10 mg/mL. Samples were passed through sterilizing filters into autodaved glass vials for incubations at -20°C, 5°C, 30 C or 40°C for one month. A TSK G3000SWXL column (7.8 x 300 mM;TosoHaas) was equilibrated at ambient temperature with a 100 mM potassium phosphate (pH 6.8) mobile phase, at a flow rate of 0.5 mL/minute using a Hewlett-Packard 1090 HPLC system. 25 flL samples (250 flg) were injected. The effluent was monitored at 280 nm. Glycation detection using TSK-boronate chromatography Using a Hewlett Packard Model 1090 HPLC system, a TSK-boronate (7.5 x 7';, mM) column was equilibrated at 40°C with solvent A (0.1 M N-[2-hydroxyethyl]piperazine-N1-[2-ethane-sulphonic acid] (HEPES) buffer, pH 8.6), operating at a flow rate of ImL/min. To elute bound forms, a linear gradient was started 10 minutes after the injection of 150 flg omalizumab that went from 0% to ]00% solvent B (0.5 M sorbitol in solvent A) in five minutes.
HICafter papain digestion Samples were diluted to 1 mglmL in a solution containing 100 mM Tos, 4 mM EDT A, 1 mM cysteine, pH 7.5. Papain was added at a ratio of 1:100 (w:w) and incubated at 3TC for two hours. A TSK Phenyl 5PW column (7.5 x 75 mM) was equilibrated with 75% mobile phase A containing 2 M ammonium sulphsulphate, 20 mM Tris, pH 7.5 and 25% mobile phase B containing 20 mM Tris, pH 7.5. The column temperature was set at 40°C and the flow rate was 1 mL/min. Digested samples were diluted 1:1 (v:v) after the sample injection. A two-step linear gradient proceeded from 25% to 40% B in three minutes and then from 40% to 80% B in 40 minutes. Mobile phase B was then increased to 100% for two minutes to regenerate the column, followed by re-equilibration at 25% mobile phase B. The effluent was monitored at 214 nm.
Fe GLyeOSYLATION The Fc oligosaccharides found on plasma-derived and recombinant antibodies produced by Chinese hamster ovary cells are primarily biantennary fucosylated structures with 0, 1 or 2 terminal galactose residues, as shown in Figure 1. Oligomannose and truncated forms (e.g.,des-GlcNAc) are also found. These oligosaccharidesare sequestered in the interior of the Fc, where their contribution to the overall Fc structure can inHuence in vitro complement-dependent cytotoxicity (CDC). Higher terminal Gal correlates with increased CDC activity, although the sensitivity of CDC assays to galactosylation diminishes as the C1q concentration increases such that terminal Gal levels may have no in vivo consequence at physiological concentrations. Core fucose inversely correlates with in vitro antibody-dependent cellular cytotoxicity (ADCC) [4,5], but this has not been confirmed by in vivo studies. Galactosylation does not appear to affect ADCC, although it is diminished by the absence of Fc glycans. ADCC is initiated by FcgRIIla binding, which is found in two allelic forms (Valor Phe at position 158). The higher affinity Va1158form correlates with improved cancer survival with rituximab (anti-CD20) therapy, indicating that ADCC is an important contributor to rituximab's efficacy [6]. The clearance of antibodies from the vasculature is modulated by the neonatal receptor, FcRn, which binds to the Fc region, protecting the antibody from catabolism in endosomes [7].lerminal saccharide groups can affect clearance of serum glycoproteins due to interactions with the asialoglycoprotein (galactose) and/or mannose receptors, but such receptor interactions are inhibited by the sequestration of the Fc oligosaccharides. Experiments conducted with omalizumab, a recombinant IgG l/kappa
118
HARRIS
G2
G1 (1-6)
Gal- GIcNAc- Man I
Fuc I Man- GIcNAc- GIcNAc- Asn297 I Gal- GIcNAc- Man
Gal - GIcNAc- Man Fuc I I Man - GIcNAc- GIcNAc-Asn297 I GlcNAc Man ~
GIcNAc- Man I
G1 (1-3)
Fuc I
Man - GIcNAc- GIcNAc Asn297 I Gal - GIcNAc- Man ~
GO
GIcNAc- Man Fuc I I Man - GIcNAc- GIcNAc- Asn297 I GIcNAc- Man
GO-G IcNAc
GIcNAc- Man Fuc I I Man - GIcNAc- GlcNAc- Asn297 I Man
GO-Fuc
GIcNAc- Man I Man - GIcNAc- GIcNAc- Asn297 I GIcNAc- Man
ManS
Man I Man - Man I Man - GlcNAc- GIcNAc~Asn297 I Man - Man
Fig. 1: Fc oligosaccharides present on antibodies produced by CHO cell lines. Abbreviations include: Gal, galactose; Fuc, fucose; Man, mannose; GIcNAc, N-acetylglucosamine. Trace levels of sialic acid may also be present. RECOMBINANTANTIBODIES
119
Table 1: Biological
effects of Fc glycosylation
variability.
CDC
ADCC
Clearance
Non-glycosylated Asn297
defective
defective
no effect
Higher galactosylation
may improve
no effect
no effect
Oligomannose forms
defective
?
no effect
Higher fucosylation
?
lower
no effect
Bisecting G1cNAc
modest increase
no effect
?
anti- IgE antibody, demonstrated that serum clearance m mIce was independent of oligosaccharide type (Harris, in preparation). The effects of Fc glycan heterogeneity are summarized in Table 1. Apart from any anticipated biological consequences, significant Fc glycosylation changes may trigger regulatory concerns, as glycosylation can be influenced by production conditions; therefore,gIycosylationmay be considered a bellwetherfor other (undetected) structural changes. HEAVY CHAIN LyS PROCESSING IgGj heavychains terminate with-Pro-Gly-Lyssequences,withthe C-terminalLys residues susceptible to removal by an unidentified basic carboxypeptidase. Partial removal of terminal Lys residues leads to charge heterogeneity that is detectable by charge-based techniques such as cation exchange chromatography [8].Samples analysed with or without CpB digestion can highlight charge variation due to Lys heterogeneity. Sensitive analysis of heavy chain C-termini is feasible using RP-HPLC after CNBr digestion [2]. The presence or absence of heavy chain Lys residues has no effect on antigen binding, and is not likely to influence Fc effector functions, clearance, or any other biological property.
METHIONINE OXIDATION All therapeutic proteins in aqueous formulations are subject to degradation due to processes such as methionine oxidation, asparagine deamidation, and aspartate isomerization. Oxidation of a methionine residue in CDR-HI of OrthoClone OKT3 led to a potency drop that limited drug product dating [9].For IgGI-class therapeutic antibodies, methionine residues at positions 255 and 430 in the Fc are particularly susceptible to oxidation, consistent with their heightened solvent accessibility [3]. Incubation with tert-butylhydroperoxide (tBHP) can be useful to identify susceptible methionine residues. Oxidation ofFc methionine residues in trastuzumab (anti-pI85her2!neu) and other IgG]-class antibodies can be detected by hydrophobic interaction chromatography after papain cleavage, as shown in Figure 2. tBHP incubation modelled the oxidation observed for accelerated stability samples from an early trastuzumab clinical production lot, as shown in Figure 3. 120
HARRIS
l~ C. E c:
