Article Cite This: Mol. Pharmaceutics XXXX, XXX, XXX−XXX
Photodisruption of the Structurally Conserved Cys-Cys-Trp Triads Leads to Reduction-Resistant Scrambled Intrachain Disulfides in an IgG1 Monoclonal Antibody Aaron T. Wecksler,† Jian Yin,‡ Paula Lee Tao,† Bruce Kabakoff,‡ Alavattam Sreedhara,*,§ and Galahad Deperalta*,† †
Protein Analytical Chemistry Department, Genentech Inc., 1 DNA Way, South San Francisco, California 94080, United States Early Stage Pharmaceutical Development, Genentech Inc., 1 DNA Way, South San Francisco, California 94080, United States § Late Stage Pharmaceutical Development, Genentech Inc., 1 DNA Way, South San Francisco, California 94080, United States ‡
S Supporting Information *
ABSTRACT: Photostability conditions as prescribed by ICH guidelines induced highly reduction-resistant scrambled disulfides that contribute to the population of apparent nonreducible aggregates in an IgG1 mAb. Photoinduced cross-linked species were isolated under reducing conditions using an organic phase size exclusion chromatography (OP-SEC) method, followed by O18labeling tryptic mapping to identify cross-linked peptides. Disulfide scrambling was observed within the IgG1 structurally conserved-intrachain cysteinecysteine-tryptophan triads (Cys-Cys-Trp), and correlated with Trp-tokynurenine (Kyn) photodegradation within these triads. We hypothesize that intrachain disulfides protect the proximal Trp within the Cys-Cys-Trp triads from photodegradation by enabling dissipation of Trp-absorbed UV energy via electron transfer to the disulfide bond. Finally, we propose three distinct mechanisms of photochemical degradation of monoclonal antibodies mediated by Trp residues. KEYWORDS: monoclonal antibody, Cys-Cys-Trp triad, disulfide scrambling, photoinduced oxidation, kynurenine
■
of light.11 The ACWOP pathway is attenuated by the addition of the singlet oxygen quencher NaN3, leading to the hypothesis that singlet oxygen generated from surface exposed Trp residues is in part responsible for the observed photoinduced oxidation of mAbs. In addition, light exposure to antibodies has been shown to affect the structurally conserved disulfide bonds by generating thioethers,12 thiyl radicals,13 and free thiols.14 The latter study is one of the few that have investigated the connection between Trp residues and disulfides in antibodies, by demonstrating that UV exposure to an antibody scFv fragment leads to aggregation and free thiol formation within disulfide-Trp (Cys-Cys-Trp) triads. Monoclonal IgG1 have two distinct types of structurally conserved disulfides bonds, denoted as interchain and intrachain disulfides.15 Each intrachain disulfide is a part of a CysCys-Trp triad, which are structurally conserved for all IgG subtypes.16 The Cys-Cys-Trp triads are hypothesized to have evolved as a light protective moiety for proteins by harboring the ability to dissipate UV energy absorbed by the Trp residue
INTRODUCTION Therapeutic monoclonal antibodies (mAbs) are exposed to different light conditions throughout manufacturing, purification, characterization, and storage.1 The photostability of monoclonal antibodies is routinely investigated as part of stress analysis for new drug substance (DS), as outlined by the International Conference on Harmonisation (ICH) document Q1B.2 Although the relevance of the ICH guideline’s light exposure to the production of biotherapeutics has recently been questioned,3 understanding the photosusceptibility under conditions of excessive light exposure is important for maintaining product quality during manufacturing and administration. Photo-oxidation of tryptophan (Trp) is one of the most common results from light exposure and routinely monitored in light-stressed stability studies.4,5 Studies performed to understand the effects of light exposure to mAbs have shown various product quality changes including increased aggregation,6 color change,5 changes in charge variants,7 oxidation of Trp, and methionine (Met) and histidine (His)8,9 and His-His covalent cross-linking.10 We have previously shown that surface exposed Trp residues generate reactive oxygen species (ROS), and proposed that this phenomenon was a result of antibodycatalyzed water oxidation pathway (ACWOP) in the presence © XXXX American Chemical Society
Received: Revised: Accepted: Published: A
December 13, 2017 January 31, 2018 March 4, 2018 March 5, 2018 DOI: 10.1021/acs.molpharmaceut.7b01128 Mol. Pharmaceutics XXXX, XXX, XXX−XXX
Molecular Pharmaceutics
■
via electron transfer to the proximal disulfide bond.17 The intracellular redox environment then enables re-formation of the intact disulfide. However, the structural/biological role of the Cys-Cys-Trp triads in IgG molecules has yet to be explored. We had previously reported that an IgG1 mAb (previously reported as mAb2) was uniquely sensitive to light, especially when exposed to intense light conditions.3 In particular, we reported that mAb2 displays significant photoinduced discoloration, site-specific oxidation, and aggregation, the latter of which was of particular interest due to recent findings demonstrating that photoinduced aggregation of an IgG1 elicited the highest immunogenic response in a transgenic mouse model compared to other stress-induced aggregates.18 Thus, we were interested in understanding the nature of the photoinduced aggregates. In this report, we further characterized the photoinduced aggregate of mAb2 after exposure to about 540 W h/m2 of UV light and around 1.2 × 106 lx h in the visible region per the ICH guideline. Utilizing capillary electrophoresis−sodium dodecyl sulfate (CE-SDS) under reducing conditions, we observed putative nonreducible crosslinked aggregates. We then utilized an organic phase−size exclusion chromatography−mass spectrometry (OP-SEC-MS) method (under reducing/denaturing conditions) which identified putative heavy chain−heavy chain (HC-HC ∼ 100 kDa) and heavy chain−light chain (HC-LC ∼ 75 kDa) cross-linked species. Further enrichment of the cross-linked species by the OPSEC method enabled O18-labeled tryptic mapping and the subsequent identification of mispaired (scrambled) disulfide dipeptides. We identified scrambled disulfide dipeptides from nearly every intrachain disulfide bond pair, leading us to hypothesize that photodisruption of the intrachain Cys-CysTrp triads leads to covalent cross-linking. The observed scrambled dipeptides were maintained after sample reduction for enrichment using the OP-SEC method, and a second reduction step was performed for tryptic mapping. This indicates that photoinduced scrambling within the disulfides associated with the Cys-Cys-Trp triads results in highly reduction-resistant species. Finally, we observed hyperoxidation of Trp to kynurenine (Kyn) for Trp residues within Cys-CysTrp triads, but this could only be observed after excessive light treatment and the enrichment of the photoinduced cross-linked species. We have previously proposed the two distinct roles of surface exposed tryptophans in the antibody mediated water oxidation pathways especially after photoexcitation. From the data presented in this manuscript we propose a mechanism in which Trp absorbs light energy and reduces disulfides in the conserved Cys-Cys-Trp triads. The disulfides, however, seem to protect the proximal Trp from oxidation similar to that proposed for the Cys-Cys-Trp moieties found in other proteins. In addition, these data indicate that excessive UV light exposure leads to photodisruption of the triads, disulfide bond scrambling, and Kyn formation of the associated Trp residue. This results in reduction-resistant intermolecular cross-linked species that likely contribute to the overall photoinduced aggregation. This work also provides insight on the photoprotective role of the IgG1 structurally conserved Cys-Cys-Trp triads. Together these three distinct mechanisms indicate the various roles Trp residues play in photosensitivity of proteins, especially monoclonal antibodies.
Article
MATERIALS AND METHODS
Materials. The therapeutic IgG1 mAb2 in this study was produced in Chinese Hamster Ovary (CHO) cells, purified, and formulated to bulk drug substance (DS). All chemicals were purchased from Sigma-Aldrich. Light Stress Sample Preparation. Samples were prepared using an Atlas Suntest CPS+ Xenon Test Instrument (Chicago, IL) using an irradiance level = 250 W/m2, total UV dose = 538 W h/m2, total visible dose = 1,320,000 lx h/m2 using Option 1 using xenon lamps. Native Size Exclusion Chromatography (SEC). Native SEC was performed on an Agilent 1200 HPLC system using a Tosoh-Bioscience SEC TSKgel G3000SWxl (7.8 × 300 mm, 5 μm) column. Isocratic runtime was 30 min at 0.5 mL/min using the mobile phase buffer (0.2 M K2HPO4, 0.25 M KCl, pH 6.2) at ambient temperature. 50 μg of mAb2 diluted in mobile phase buffer was injected for each analysis and monitored at 280 nm. Data was analyzed using Chromeleon Software package (Dionex). Organic Phase−Size Exclusion Chromatography− Mass Spectroscopy (OP-SEC-MS). OP-SEC was performed on an Agilent 1200 HPLC system using two Tosoh-Bioscience SEC TSKgel SuperSW3000 (2 × 300 mm, 4 μm) columns linked in series. Isocratic runtime was 45 min at 0.25 mL/min using an organic mobile phase buffer (60% acetonitrile (ACN), 39.9% H2O, 0.1% TFA) at a column temperature of 70 °C. All samples subjected to OP-SEC analysis were prepared at 1 mg/ mL concentrations in 1 mL of 20 mM Tris (pH 7.5) buffer and reduced with 10 mM dithiothreitol (DTT) at 70 °C for 15 min, followed by the addition of 20 μL of 0.1% TFA. 25 μg of reduced mAb2 was injected for each analysis and monitored at 280 nm. Data was analyzed using Chromeleon Software package (Dionex). In-line MS analysis was performed using a QTOF (quadrupole time-of-flight) Premier mass spectrometer (Waters, Milford, MA, USA) operated in positive electrospray ionization mode and coupled in-line to the HPLC system for the OP-SEC-MS analysis. Instrument control and data analysis were performed using Waters MassLynx software (version 4.1). Deconvolution of multiply charged ions was performed with the MaxEnt 1 software provided with MassLynx. RP-HPLC Oxidation Assay. The detection and quantitation of protein oxidation was performed using a RP-HPLC assay. Samples were prepared at 1 mg/mL concentrations in 50 mM Tris, pH 8.0, and digested with FabRICATOR (IdeS) (Genovis) (50 unit per 100 μg of antibody) for 4 h at 37 °C. Digested samples were then reduced and denatured (10 mM DTT, 8 M guanidine, 50 mM Tris, pH 8.0) for 30 min at 37 °C. Reduced digests (50 μg) were separated using an Agilent 1200 HPLC with Thermo Fisher Scientific BioBasic Phenyl Column (2.1 × 150 mm, 5 μm, 300 Å). Peptide elution was performed using a gradient from 68% solvent A (H2O, 0.1% TFA) to 55% solvent B (ACN, 0.1%TFA) over 19 min at a flow rate of 0.3 mL/min with a column temperature of 50 °C. Assessment of Oxidation and Reactive Oxygen Species (ROS). Samples were exposed to the indicated ICH light exposure after the addition of 0.1−100 mM sodium azide (NaN3). O18-Labeling Tryptic Peptide Mapping and LC-MS/MS Analysis. Samples (500 μg, 6.7 μM total protein) were diluted into denaturing buffer (6 M guanidine, 360 mM Tris, 2 mM EDTA, pH 8.6) and reduced by incubation at 45 °C for 10 min in the presence of 10 mM DTT. S-carboxymethylation was B
DOI: 10.1021/acs.molpharmaceut.7b01128 Mol. Pharmaceutics XXXX, XXX, XXX−XXX
Article
Molecular Pharmaceutics performed by incubating samples at 45 °C for 10 min in the presence of 20 mM sodium iodoacetate, then quenched at room temperature with 40 mM DTT. Samples were then desalted by loading onto NAP-5 columns (GE Healthcare) and eluted with 800 μL of trypsin digest buffer (20 mM Tris, pH 8.0). For O18-labeling, the samples were split in half after the NAP-5 desalting procedure, and dried under vacuum. The samples were then reconstituted in either LCMS grade H2O16 or H2O18 (99.7% atom purity, Sigma-Alrich) and digested with 3% trypsin (w/w, Roche recombinant, reconstituted in appropriate isotopic water) at 37 °C for 3.5 h. TFA was added to a final concentration of 0.3% to stop the digest. Tryptic peptides were separated using an Agilent 1200 HPLC with Phenomenex Jupiter C-18 column (2 × 250 mm, 5 μm, 300 Å). Peptide elution was performed using a gradient from 100% solvent A (H2O, 0.1% TFA) to 45% solvent B (ACN, 0.1%TFA) over 215 min at a flow rate of 0.25 mL/min. Mass spectrometric analysis was performed with a Thermo Fisher Scientific Orbitrap Elite operating in the positive ion mode. A full MS1 scan in positive mode (60,000 resolution from 400− 2000 m/z) was performed using the Fourier transform (FT) analyzer with Data Dependent Acquisition (DDA) MS2. High collision-induced dissociation (HCD) was performed on the top 5 most intense product ions using the Fourier transform (FT) analyzer. Data analysis was performed as previously described to identify the putative cross-linked dipeptides.19 Homology Modeling and Solvent Accessible Surface Area Calculations. The homology model used for visualization of a generic mAb structure was previously generated,20 and the solvent accessible surface area calculations were performed using VADAR.21
To further investigate, we developed an organic phase-size exclusion chromatography (OP-SEC) method for detection of nonreducible cross-linked species (Figure 1). This method is similar to traditional size exclusion chromatography, but is performed in an acetonitrile organic phase for compatibility with online mass spectrometry.22 To observe the presence of putative nonreducible cross-linked species, samples are analyzed after addition of dithiothreitol (DTT) and incubated at denaturing temperature (70 °C). Using this approach, the chromatograms of mAb2 samples exposed to ICH light revealed species that migrated with a retention time between the expected retention times of the intact protein (150 kDa) and reduced heavy chain (HC) (50 kDa) of mAb2 (Figure S1C). By coupling this technique with online mass spectrometry analysis, we observed species with molecular weights of ∼100 kDa and ∼75 kDa. These species were presumed to be nonreducible, covalent cross-linking of heavy chain−heavy chain (HC-HC) and heavy chain−light chain (HC-LC), respectively (Figure S2). To determine if the formation of the putative nonreducible cross-linked species was linked to ROS generation, we exposed mAb2 to light in the presence of NaN3, a singlet oxygen quencher, and observed a dose-dependent attenuation of all the observed product quality changes, including the cross-linked species (Figure 2, Figure S1D). Therefore, the photoinduced cross-linking was in part due to a singlet oxygen-driven process that was likely linked to Trp oxidation as previously observed.11 Identification of Reduction-Resistant Scrambled Disulfide Dipeptides. To investigate the chemical nature of the cross-linked species, we first isolated/enriched the cross-linked species using the OP-SEC method (Figure S3), and then utilized the previously described O18-labeling tryptic digestion method to identify putative cross-linked dipeptides.10,19 The O18-labeling method takes advantage of the fact that the reaction mechanism of trypsin incorporates two oxygen molecules from the buffer into the newly formed tryptic peptides. Linear peptides with a single C-terminus will have two oxygen molecules incorporated, whereas dipeptides with two Ctermini will have four oxygen molecules incorporated. Thus, by comparing the digest performed in H2O16 vs H2O18, linear peptides and dipeptides will have +4 Da mass shifts and +8 Da mass shifts, respectively. Since retention times are not affected, searching for difference in mass shifts between overlapping peaks enables rapid detection of putative dipeptides. Using this methodology, we initially uncovered two dipeptides, which were surprisingly determined to be scrambled disulfides between intrachain disulfides and the hinge region (Figure 3 and Figure S4, Table S1 and Table S2). Following these initial dipeptide identifications, we searched the mass spectrometry data for all possible combinations of disulfide dipeptides and found 13 different scrambled disulfide crosslinked peptides from the intrachain disulfides participating in Cys-Cys-Trp triads (Table 1). An example of a disulfide dipeptide from the scrambling of two intrachain Cys-Cys-Trp triads is shown in Figure 4. In addition, we observed disulfide scrambling at the interchain disulfides not associated with triads (data not shown), as previously observed from thermal stress exposure.23 Remarkably, these identified disulfide dipeptides survived an initial reduction step with DTT prior to the OP-SEC enrichment, and a second DTT reduction step during the tryptic digestion procedure, indicating that they are significantly “reduction-resistant”. Similar observations were made when we
■
RESULTS AND DISCUSSION Detection of Photoinduced Cross-Linking in mAb2. We have previously observed that mAb2 was unusually susceptible to light exposure during routine photostability studies.3 Further analytical characterization detailed in this report demonstrates a linear correlation between increases in protein oxidation and protein aggregation, and the formation of apparent nonreducible cross-linked species using CE-SDS under reducing conditions (R-CE-SDS) (Figures 1 and S1).
Figure 1. The effects of ICH light exposure to mAb2. Linear increase in changes in product quality with respect to ICH light exposure. HMWS = high molecular weight species (aggregates). C
DOI: 10.1021/acs.molpharmaceut.7b01128 Mol. Pharmaceutics XXXX, XXX, XXX−XXX
Article
Molecular Pharmaceutics
Figure 2. The protective effects of NaN3 on the ICH light exposure to mAb2. Dose-dependent suppression of the photoinduced effects with NaN3 addition.
Figure 3. Identification of hinge-to-Fc intrachain scramble disulfide dipeptide. Comparison of the O16-labeled and O18-labeled HCD MS/MS spectrum of the triply charged precursor ions 1297.3420 m/z of the trypsin generated cross-linked peptides HC20−HC36. The deconvoluted parent masses are separated by 8 Da as expected for cross-linked peptides.
used TCEP instead of DTT as the reducing agent (data not shown). The reduction of these disulfides could only be
accomplished upon addition of DTT after complete tryptic digestion (Figure S5 and Figure S6), further supporting their D
DOI: 10.1021/acs.molpharmaceut.7b01128 Mol. Pharmaceutics XXXX, XXX, XXX−XXX
Article
Molecular Pharmaceutics
(protein subunit) state. The identified disulfide dipeptides are connected by intrachain Cys residues from all of the different regions of mAb2 (HC Fab, Fc region, LC Fab) (Table 1). In addition, Figure 3 and Figure S4 demonstrate that the hinge region is susceptible to scrambling with disulfides associated with triads, indicating that this region is susceptible to photoinduced cross-linking, which is consistent with the previously observed His−His cross-linking at the hinge region.10 Due to the significant spatial distances of these disulfide dipeptides on the same mAb (>50 Å), these data suggest the formation of intermolecular cross-linking (two different mAbs linked together), which likely contributes to the observed photoinduced aggregation. Photodisruption of the Cys-Cys-Trp Triads Leads to Kyn Formation. It has been well established that not only are Trp residues in mAbs highly susceptible to reactive oxygen species (reviewed in ref 24) but also Trp residues themselves can initiate ROS generation via ACWOP.11 Hyperoxidation of Trp can lead to the degradation product cascade from singly oxidized tryptophan to N-formylkynurenine (NFK), and eventually kynurenine as final oxidative product.5 In mAbs, Trp residues are found in three types of structural environments: solvent exposed, buried, and Cys-Cys-Trp triads. To our knowledge, kynurenine formation in mAbs has only been
Table 1. Cys-Cys-Trp Scrambled Disulfide Dipeptides Identified Using O18-Labeling in the Nonreducible CrossLinked Fraction Isolated from OP-SEC cross-linked cysteines
cross-linked region
C22−C150 C22−C373 C23−C327 C96−C150 C96−C235 C150−C226 C150−C267 C150−C327 C150−C373 C195−C373 C235−C327 C235−C373 C327−C373
HC Fab−HC Fab HC Fab−Fc LC Fab−Fc HC Fab−HC Fab HC fab−hinge HC Fab−HC Fab HC Fab−Fc HC Fab−Fc HC Fab−Fc HC Fab−Fc hinge−Fc hinge−Fc Fc−Fc
m/za 1146.87 1093.52 449.22 846.41 1354.33 857.41 1115.22 756.38 1089.58 974.17 1012.52 1297.34 676.36
(z (z (z (z (z (z (z (z (z (z (z (z (z
= = = = = = = = = = = = =
3) 3) 2) 2) 3) 3) 3) 2) 2) 3) 3) 3) 2)
a All peptides exhibited +8 Da incorporation in O18-labeled tryptic mapping. All disulfide dipeptides were reduced using DTT post trypsin digestion.
identification as scrambled disulfide dipeptides that are resistant to reduction in the native (intact protein) or denatured
Figure 4. Identification of Fc-to-Fab HC intrachain scramble disulfide dipeptide. Comparison of the O16-labeled and O18-labeled HCD MS/MS spectrum of the triply charged precursor ions 1090.0839 m/z of the trypsin generated cross-linked peptides HC20−HC14. The deconvoluted parent masses are separated by 8 Da as expected for cross-linked peptides. E
DOI: 10.1021/acs.molpharmaceut.7b01128 Mol. Pharmaceutics XXXX, XXX, XXX−XXX
Article
Molecular Pharmaceutics
Table 2. Comparison of the Site-Specific Photo-Oxidation of Tryptophan of the Unfractionated vs Isolated Cross-Linked Fraction isolated cross-linksa
unfractionated mAb2 peptide
tryptophan
region
HC5 HC11 HC26 HC41 LC8 average HC3 HC15 HC23 HC37 LC4 LC14 average
47,d 53 108, 113d 319 423e 94
VH CH1 CH2 CH3 CL
36 164 283 387 35 149
VH CH1 CH2 CH3 VL CL
solvent accessible surface area (Å2) 0.7, 60.1 100.6, 7.3 16.6 36.0 18.5 37.7 0.5 0.1 1.9 0.1 0.3 4.2 1.6
Cys-Cys triad na na na na na C22−C96 C150−C206 C267−C327 C373−C431 C23−C88 C135−C195
c
% Kynb
% total Trp oxidative species
% Kyn
% total Trp oxidative species
2.2 7.7 0.2 0.3 18.6 5.8
