Cysteine-SILAC Mass Spectrometry Enabling the Identification and

Subscriber access provided by UNIV OF CAMBRIDGE

Article

Cysteine-SILAC mass spectrometry enables the identification and quantitation of scrambled inter-chain disulfide bonds: preservation of native heavy-light chain pairing in bispecific IgGs generated by controlled Fab-arm exchange. Ewald T.J. van den Bremer, Aran F. Labrijn, Ramon van den Boogaard, Patrick Priem, Kai Scheffler, Joost P.M. Melis, Janine Schuurman, Paul Parren, and Rob N de Jong Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.7b02543 • Publication Date (Web): 18 Sep 2017 Downloaded from http://pubs.acs.org on September 20, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Cysteine-SILAC mass spectrometry enables the identification and quantitation of scrambled inter-chain disulfide bonds: preservation of native heavy-light chain pairing in bispecific IgGs generated by controlled Fab-arm exchange.

a

a

a

a

b

Ewald T.J. van den Bremer , Aran F. Labrijn , Ramon van den Boogaard , Patrick Priem , Kai Scheffler , a a ,c* a* Joost P.M. Melis , Janine Schuurman , Paul W.H.I. Parren and Rob N. de Jong a

Genmab, Yalelaan 60, 3584CM, Utrecht, the Netherlands.

b

Thermo Fisher Scientific GmbH, Im Steingrund 4-6, 63303, Dreieich, Germany.

c

Department of Immunohematology and Blood Transfusion, Leiden University Medical Center, Albinusdreef 2, 2333 ZA, Leiden, the Netherlands

*The authors contributed equally address correspondence to: dr. Rob de Jong Email: [email protected]

ABSTRACT: Bispecific antibodies (bsAbs) are one of the most versatile and promising pharmaceutical innovations for countering heterogeneous and refractory disease by virtue of their ability to bind two distinct antigens. One critical quality attribute of bsAb formation requiring investigation is the potential randomization of cognate heavy chain (H)/light (L) chain pairing, which could occur to a varying extent dependent on bsAb format and the production platform. To assess the content of such HL chain-swapped reaction products with high sensitivity, we developed cysteine-SILAC, a method that facilitates the detailed characterization of disulfidebridged peptides by mass-spectrometry. For this analysis, an antibody was metabolically labeled with

13

15

C3, N-cysteine and incorporated into a comprehensive panel of distinct bispecific

molecules by controlled Fab-arm exchange (DuoBody technology). This technology is a postproduction method for the generation of bispecific therapeutic IgGs of which several have progressed into the clinic. Herein, two parental antibodies each containing a single heavy chain domain mutation, are mixed and subjected to controlled reducing conditions during which they exchange heavy-light chain (HL) pairs to form bsAbs. Subsequently reductant is removed and all disulfide bridges are re-oxidized to reform covalent inter- and intrachain bonds. We conducted a multilevel (top-middle-bottom-up) approach focusing on the characterization of both ‘left-arm’ and ‘right-arm’ HL inter-chain disulfide peptides, and observed that native HL pairing was preserved in the whole panel of bsAbs produced by controlled Fab-arm exchange.

1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 21

Human IgG1 antibodies are composed of two identical heavy (H) and two identical light (L) chains with a molecular weight of approximately 50 kDa and 25 kDa, respectively, and contain four inter-chain and twelve intra-chain disulfides (Figure 1A). Two inter-chain disulfides bridge the two H chains and one disulfide connects each of the L chains to a H chain to form a 150 kDa two-armed Y-shaped antibody 1

molecule (Figure 1A) . In a regular IgG1 antibody, each Fab-arm (HL pair) contains identical antigen binding sites and therefore each Ab is monospecific and can bind bivalently. Nowadays, engineered bispecific antibodies (bsAbs) containing two distinct antigen-binding specificities are widely used to 2

investigate and exploit the therapeutic potential of dual-targeting . Human IgG-like bsAb formats have been described in both symmetric and asymmetric variations, which typically differ in the number of different antigen binding sites, H chains, and L chains incorporated into the molecule. Asymmetric IgGlike bsAbs with two different antigen-binding sites can be produced by co-expression chains

1,5-8

, or by post-production recombination approaches

9-11

3,4

of all incorporated

. Depending on the format and production

strategy, different product-related impurities may be formed, including aggregates, fragments, (residual) homodimeric Abs. In addition, the potential occurrence of non-native HL recombination is an oft-asked question. Limitations to the sensitivity of functional methods, such as those probing changes in antigen binding, have thwarted thorough investigations, while peptide isobaricity has precluded analysis by 7

mass . Human IgG4 antibodies are naturally able to recombine into bsAbs by a process called Fab-arm exchange

12,13

. Based on this process we developed controlled Fab-arm exchange (cFAE), which forms

the basis for the DuoBody technology and represents a highly efficient, fully scalable post-production recombination approach to generate bispecific IgG1 (bsIgG1) based on two separately expressed IgG1 mAbs containing single matched point mutations, i.e. F405L or K409R

9,10,14

(Figure 1B). The DuoBody

technology has been validated from discovery to large manufacturing scale production bsAb in clinical development (See Moores et al.

15

10,14

and by several

and clinical trials NCT02609776; NCT02715011,

NCT03145181). When the two parental, homodimeric mAbs are subjected to controlled reducing conditions that disconnect inter-chain disulfides, the matched mutations drive the highly efficient exchange of non-covalently bound HL half-molecules that are subsequently re-oxidized into covalent heterodimeric bsAbs after removal of the reductant. Increased stability of the CH3 domain interaction in the heavy chain heterodimer compared to the homodimer by virtue of the selected single matched mutations strongly drives the equilibrium towards the bispecific heterodimeric reaction product. Theoretically, product-related impurities could include disulfide scrambled products, or HL recombination products, depending on the strength of the non-covalent HL interactions following reduction of the 16

covalent disulfide bonds . Since both H and L chain can contribute amino acids to the antigen binding site, randomization of HL pairing could hypothetically generate bispecific impurities that may have 7

heterogeneous affinity or potentially lack binding properties . An assessment of HL inter-chain disulfide linkages (i.e. HC-Cys-220:LC-Cys-214, Figure 1A) could ensure the integrity of the generated bsAb and enable understanding of this critical quality attribute.


Page 3 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


To quantify the level of potential inter-chain disulfide rearrangements in IgG samples with high sensitivity at minimal added spectral complexity, we developed cysteine-SILAC mass spectrometry. Stable isotope labeling by amino acids in cell culture (SILAC) is a widely applied method in quantitative proteomic and biological research based on the production of proteins using culture media supplemented 17

with an isotopically labeled amino acid, such as arginine, lysine or tyrosine . When light and heavy isotope-labeled cell populations or proteins are mixed, they remain distinguishable by MS and protein abundances are determined from the relative MS signal intensities providing accurate relative 18

quantification . Here, we explored the incorporation of “heavy”

13

15

C3, N-cysteine into antibodies to monitor

rearrangements between cysteine containing peptides after Ab secretion from the host cell (Figure 1B). By metabolically labeling one Ab with heavy cysteine, the formation of disulfide bridges between labeled and non-labeled Ab chains can be monitored also for constant region peptides that would otherwise be chemically equivalent. A parental Ab panel designed to capture naturally occurring Fab domain heterogeneity was recombined into bsAbs by cFAE, and the HL pairing was assessed using peptide level cysteine-SILAC analysis combined with top and middle-up high-resolution mass spectrometry 19

approaches (Figure S1A-C, supporting information) . Although intact mass (top) bsAb measurements could detect global aberrant chain pairings, L chains swapped between Fab-arms would result in isobaric 7

whole bsAbs that cannot be distinguished by mass not even by emergent MS techniques

20,21

. Therefore,

middle-up Fab fragment level assessment was included that allowed the detection of light chain-swapped 22

Fab fragments . Cysteine-SILAC peptide level analysis provided a generic, direct and detailed assessment of both HL and HH disulfide linkages, yielding simultaneous insight in bispecific integrity and yield with high sensitivity. We conclude that cFAE is a robust process that results in the formation of stable therapeutic bsIgG1 molecules in which the native HL pairing is preserved.

EXPERIMENTAL SECTION Parental ‘light’ antibodies were expressed, purified and quality controlled as described in the Supporting Information Experimental Section. Cysteine-SILAC metabolic labeling of ‘heavy’ IgG-CD20-K409R. CHOK1SV cells stably expressing IgG-CD20-K409R

14

derived from human mAb 7D8 directed against CD20

23

were thawed and 13

15

cultured on cysteine-depleted CD-CHO medium (Gibco, no. 1140530). Heavy L-cysteine- C3, N (Sigma, no. 658057) was added and dissolved in the medium to the required concentration of 100 mg/L followed by a sterile filtration with a bottle top filter (Sigma) without pH adjustment. After two passages, three 5

batches were inoculated at 0.2×10 cells/mL. Batches were harvested on day 4, day 5, and day 7, 9,14

centrifuged at 300x g for 10 min, filtered with a bottle top filter, and purified as previously described

.

Generation of bsAb by Controlled Fab-arm Exchange. Equimolar amounts of IgG1-F405L and ‘heavy’ IgG1-K409R-CD20 antibodies were mixed and incubated with 2-mercaptoethylamine (2-MEA;



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 21

Sigma) at a final concentration of 1 mg/mL per antibody. The final concentration of 2-MEA was 75 mM. The mixtures were incubated for at 31 °C for 5h. Subsequently, the 2-MEA was removed by bufferexchanging against PBS. Samples were stored overnight at 4 °C to allow full reoxidation of the disulfide bonds to occur

9,10

. General quality assessment of the purified parental mAbs and generated bsAbs was 9

performed by CE-SDS and HP-SEC as previously described . Bispecific yield was determined using hydrophobic interaction chromatography (HIC) and/or Cation Exchange Chromatography (CEX) as 10

previously described . Residual 2-MEA was assessed via the AccQ-Tag™ (Waters) method in 10

conjunction with fluorescence detection . LC-MS/MS Analysis of Lys-C digests. The digested samples were analyzed by LC-MS/MS, performed on an Thermo Scientific Ultimate 3000 RSLC system coupled online with a Thermo Scientific Q Exactive Plus Hybrid Quadrupole-Orbitrap Mass Spectrometer as described above. Samples were injected (1 µL, 1.2 µg) onto an Acclaim

TM

TM

PepMap

100 C18 column (1.0 mm x 15 cm, 5 µm particle size

with 100 Å pore size) with a temperature maintained at 40 °C. The mobile phase A contained 0.1% formic acid (v/v) in water and mobile phase B was pre-pared in acetonitrile with 0.1% (v/v) formic acid. The nonreduced peptides were eluted at 130 µL/min using 2% B to 30% B in 60 min and subsequently in 10 min from 30% B to 80% B. For MS data acquisition, the MS capillary temperature was maintained at 320°C with the S-lens RF level set to 65. The MS data acquisition was performed using a data dependent Top 5 method comprising scan cycles of full MS scans followed by 5 MS/MS scans. Full MS spectra were 6

acquired at a resolution setting of 70,000 in positive polarity mode with an AGC target value of 3×10 , maximum ion injection time (IT) of 100 ms and a scan range from 300 to 2,000 m/z. Data-dependent fragment ion spectra were acquired at a resolution setting of 17,500 with an AGC target value of 1×10

5

and a maximum ion injection time (IT) of 250 ms. The obtained raw data files were searched manually via targeted analysis of the antibody HH and HL non-reduced Lys-C peptide sequences. To determine the exchange efficacy for the (residual) parental and hemi-labeled HH peptides, the relative abundance was calculated from the three most abundant isotope peaks in each peak distribution. For quantification of the κκ L-swapped peptides the total relative abundance was calculated for 633.25 and 633.75 m/z. For λκ L-swapped peptides the total relative abundance was calculated for the 630.76 (HALBκ) and 633.26 (HBLAλ) m/z and subsequently corrected for background signals.

RESULTS AND DISCUSSION 13

15

Generation of parental ‘heavy’ cysteine- C3, N IgG1-CD20-K409R mAb using SILAC. In theory, the generation of bsAbs by cFAE, involving the transient reduction of HL interchain disulfides, could yield bispecific molecules with L-swapped Fab-arms. This could be tested by isotopic labelling of the cysteines in one parental, which would allow quantification of the level of inter-chain disulfide rearrangements during bsAbs formation by mass spectrometry with high sensitivity at minimal added spectral complexity. Therefore, parental antibody IgG1-CD20-K409R was produced using cell culture


Page 5 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


13

15

media that exclusively contained ‘heavy’ L-cysteine- C3, N as cysteine source (Figure 1B). Cysteine is an essential amino acid for mammalian cells in culture and metabolic conversion was therefore expected 24

to be limited . Nevertheless, metabolic conversion by even marginal extents may result in substantial 25

broadening of the isotope distribution and compromise sensitivity . The theoretical mass difference for each incorporated heavy cysteine is +4.0 Da. To measure the efficiency of ‘heavy’ cysteine incorporation, we analyzed intact deglycosylated IgG1-CD20-K409R by high resolution mass spectrometry. IgG1 comprises 5 and 11 cysteine amino acids in the light chain and heavy chain, respectively (Figure 1A). Hence, a completely isotopically labeled IgG1 structure comprising 32 cysteine residues would theoretically yield a mass increase of 128.0 Da. The obtained intact m/z and 13

15

deconvoluted mass spectra of deglycosylated IgG showed a complete cysteine- C3, N incorporation. The 128.6 Da mass increase is in good agreement with the expected mass difference (Figure S2, supporting information). In summary, cysteine-SILAC yielded proteins that were labeled to saturation (>99.5%) and did not suffer from detectable metabolic conversion. The combination of these two factors minimized the increase in spectral complexity (i.e. 8 unique ‘heavy’ disulfide peptides compared to the unlabeled mAb) and maximized the sensitivity at which native and non-native recombination events could be detected.

Generation of bsAbs with one ‘heavy’ Fab-arm. Besides the generation of IgG1-CD20-K409R, to be used as common ‘heavy’ Fab-arm, a panel of parental mAbs, each containing the matching F405L point mutation, was produced in the presence of natural ‘light’ L-cysteine, to serve as second Fab-arm. Various intrinsic biophysical properties of Fab-arms were represented in the panel, to assess the impact of these attributes on L chain mispairing during cFAE (Table S1 and S2, supporting information). The panel was composed of allotypic, aglycosylated and deglycosylated IgGs, kappa (κ) and lambda (λ) light chain variants and Fab-glycosylated antibodies. Representative antibody binding domains were selected from publicly available antibody panel including two approved mAbs (i.e. trastuzumab and cetuximab). All parental ‘light’ IgG1-F405L batches were deglycosylated prior to analysis and analyzed by high resolution mass spectrometry. The measured masses were in agreement with the theoretical masses based on the introduced primary amino acid sequences (data not shown) excluding mAbs exhibiting Fab glycosylation that require a more advanced deglycosylation approach. BsAbs were generated by cFAE by mixing ‘light’ IgG1-F405L mAbs with the ‘heavy’ IgG1-CD20-K409R parental mAb in the presence of 75 mM 2-MEA reductant in a one-to-one ratio and allowing exchange to progress to completion. The relative abundance of the resulting bsAbs was determined either by cation exchange (CEX) or by hydrophobic interaction chromatography (HIC) as orthogonal methods, depending 10

on the biophysical properties of the bsAb and parental mAb pairs . The exchange efficiency of all generated bsAbs exceeded 90% (Table S3, supporting information). Likely due to complex Fab



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 21

glycosylation, the relative abundance of bsIgG1-CET×CD20 and bsIgG1-CD22×CD20 could not be determined via either of the orthogonal methods.

Intact and middle-up structural assessment of bsAbs with one ‘heavy’ cysteine Fab-arm. First, residual levels of the 2-MEA were determined using the AccQ-Tag™ method in conjunction with fluorescence detection. For all tested bsAbs, the residual 2-MEA concentration had been reduced to less than 4 µM, which permits full re-oxidation by dissolved oxygen of the inter-chain disulfide network. All generated bsAb were subsequently analyzed by intact high resolution mass spectrometry to detect the formation of product-related impurities. All masses obtained for the deglycosylated bsAbs were in close agreement with the calculated theoretical masses (Table S2, supporting information). Furthermore, low fractions of parental homodimers were present supporting the results of the quantitative analyses performed by HIC or CEX. Molecular weight species indicating the formation of other than main bsAb product LAHAHBLB (correct assembly) and/or the theoretical isobaric bsAb LAHBHALB (L-chain swapped assembly) were not detected, as exemplified by intact bsAb IgG1-EGFr×CD20 (Figure 2A, B, left panel; Figure S3, Supporting information). For the Fab glycosylated bsAbs IgG1-CET×CD20 and IgG1CD22×CD20 no definitive intact bsAb spectra could be obtained (Figure 2C, D, left panel) as expected 26

from the extensive glycan-related mass heterogeneity of the parental mAbs . To investigate whether bsAb species comprising the isobaric L-chain swapped LAHBHALB assembly were formed upon cFAE, the bsAb panel was subjected to a middle-up structural 27

characterization, based on specific proteolysis into Fab fragments . Since bsAbs contain two distinctive Fab-arms, generating Fab fragments would assess the abundance of the expected native LAHA and LBHB-derived Fab fragments relative to that of non-native recombined LAHB and LBHA-derived Fabfragments (see simulation in Figure S1, supporting information). Furthermore, this method minimizes sample processing-induced artifacts such as disulfide scrambling. The bsAb panel was subjected to gingipain K (GingisKHAN™) digestions, cleaving human IgG1 between K222 and T223 (i.e ..SCDK/THTCCP..) in the upper hinge. Well-defined Fab fragment spectra were observed for multiple bsAbs except for glycosylated Fab arms and for bsAbs comprising the IgG1-F405L-EGFR arm (data not shown). Hence, for bsAbs containing the IgG1-F405L-EGFR arm the enzyme papain was used, cleaving 28

predominantly between H224 and T225 (i.e. ..SCDKTH/TCPP..) . All obtained digestions were analyzed with a Q Exactive Plus Orbitrap high resolution mass spectrometer without further purification and/or Fc removal. The resulting Fab fragment signals demonstrated the presence of native molecular Fab species, as exemplified by intact bsAb IgG1-EGFr×CD20 (Figure 2A, B, right panel; Figure S3, supporting information). Papain digests exhibited minor additional peaks with a ∆mass of minus 366 Da, indicating an additional cleavage between D221 and K222 (i.e...SCD/KTHTCPP..) (Figure 2A, B; Figure S3, supporting information). The molecular mass expected for non-native mispaired species was calculated for each bsAb Fab fragment and indicated the formation of L chain swapped HBLA or HALB-derived Fab


Page 7 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


fragments after cFAE could not be detected. However, for a more conclusive characterization with improved sensitivity we resorted to a targeted bottom-up approach.

Bottom-up targeting HH and HL control peptides of bsAbs with one ‘heavy’ cysteine Fabarm. SILAC-based experiments are more generally conducted using isotope-labeled forms of arginine and/or lysine, because of subsequent trypsin digestion C-terminal to basic residues in proteins for MS 29

analysis . However, trypsin cleavage failed to yield a detectable non-reduced disulfide-containing HL peptide ((H)SCDK/(L)GEC) (data not shown) and would keep the light chain peptide moiety unlabeled, prohibiting the detection of potential non-native HL recombination events. Instead, Lys-C digestion was found to produce a detectable disulfide-linked HL peptide ((H)SCDK/(L)SFNRGEC) and was therefore used in further bottom-up work in combination with cysteine isotope labeling. Digestion with endoproteinase Lys-C generated two inter-chain disulfide-containing peptides of interest

16,30

, of which the non-reduced HL peptide can probe potential L-swapping, while the non-reduced

HH peptide allowed monitoring of bsAb formation and can serve as an internal control for the formation of hemi-labeled

peptides

(Figure

S1C,

supporting

information).

To

distinguish

cFAE-induced

recombination events from peptide mapping-induced events, we recorded mass spectra for different mixtures of the parental IgG1-EGFR and IgG1-CD20 antibodies. Figure 3 shows the mass spectra representing the interchain HH hinge (left panels) and HL Fab-arm (right panels) disulfide dipeptides for (i) a mixture of unlabeled antibodies (panel A); (ii) a mixture of a labeled IgG1-CD20 and unlabeled IgG1EGFR prior to cFAE (panel B); (iii) a control mixture of labeled IgG1-CD20 with unlabeled IgG1-EGFR lacking the cFAE-enabling mutation exposed to cFAE conditions (panel C); and (iv) a mixture labeled IgG1-CD20 and unlabeled IgG1-EGFR exposed to cFAE conditions (panel D). The observed peaks for the unlabeled (light) and labeled (heavy) parental HH and HL combinations are labeled as appropriate. In cases where single- hemi- or double-labeled peaks are absent, their potential position is indicated by stippled lines for easy reference. The Lys-C HH hinge peptide is composed of two identical heavy chain peptides comprising the amino acid sequence (THTCPPCPAPELLGGPSVFLFPPKPK)2, linked via two inter-chain disulfide bonds (cysteines underlined). The mass difference expected between the unlabeled and doubly-labeled nonreduced HH hinge peptide is 16.0 Da (4 cys). The results obtained for this peptide in the control experiment with the unlabeled mAb mixture showed a typical charge state envelope from 4+ to 9+ with a monoisotopic m/z value of 910.14 [M+6H]

6+

(monoisotopic mass 5454.84 Da) (Figure 3A; left panel). In

contrast, the mAb mixture containing ‘light and ‘heavy’ cysteine-SILAC-labeled mAb clearly showed a ‘SILAC peak pair’ presenting both ‘light’ and ‘heavy’ hinge HH peptides with a monoisotopic m/z value for the heavy peptide of 912.81 [M+6H]

6+

(monoisotopic mass 5470.86 Da). Both non-reduced HH peptides

were in good agreement with the calculated monoisotopic masses 5454.82 Da and 5470.82 Da (Figure 3B, left panel). A similar result was obtained for a FAE-incompatible mAb mixture, containing wild-type ‘light’ IgG1-EGFR that lacks the cFAE-enabling F405L mutation after exposure to cFAE conditions



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 21

(Figure 3C, left panel). Finally, analysis of the cysteine-SILAC-labeled bsAb IgG1-EGFr×CD20 formed by cFAE clearly showed the formation of the hybrid, hemi-labeled HH hinge peptide with a monoisotopic m/z 911,48 [M+6H]

6+

i.e. monoisotopic mass 5462.88 Da (calculated mass 5462,82 Da) (Figure 3D, left

panel). The sequence (H)SCDK/(L)SFNRGEC (letters in parentheses indicate originating chain) represents the non-reduced IgG1κ HL peptide. In the bsAb molecule formed by cFAE using one ‘heavy’ 13

15

parental Ab, one arm was labeled metabolically with ‘heavy’ cysteine- C3, N (+4.0 Da), while the other arm contained natural ‘light’ cysteines. The resulting disulfide peptide spectra of both the Ab mixture and the bsAb formed by cFAE showed “cysteine-SILAC pairs” with m/z 631.25 (light) and 635.25 (heavy) 2+

[M+2H] , which corresponds to monoisotopic masses 1260.5 Da and 1268.5 Da for the unlabeled and labeled non-reduced peptide, respectively (Figure 3B,C,D, right panels) presenting a mass difference of 8.0 Da between the disulfide-containing peptides. The identities of these peptides were confirmed by Higher-energy Collisional Dissociation (HCD) mass spectrometry measurements in which the full peptide sequence was obtained in one single MS/MS spectrum (Figure S4, Supporting information). Nonnative recombined disulfides produced via an L-chain swapping event would result in hemi-labeled species with a mass increase of 4.0 Da. For the non-labeled HL peptide, the two least abundant peaks of the natural isotope distribution overlapped with the theoretical mass of a hemi-labeled L-swapped peptide (HALB/HBLA marked stippled lines). The total abundance of the two isotope peaks in the parental homodimer mixture prior to cFAE (Figure 3B) was determined at 0.4%. The relative abundance these peaks in the cFAE-exposed mixtures of the cFAE incompatible control (Figure 3C) and the IgG1-EGFR×CD20 bsAb (Figure 3D) were determined at 1.0% and 0.6%, respectively. Compared to the non-cFAE-treated mixture, both spectra therefore showed a slight increase in the abundance of these peaks suggesting that the small increase observed was independent of bsAb formation. We therefore inferred that the increase was not caused by HL half-molecule exchange, but more likely induced during the peptide mapping of Abs previously exposed to cFAE conditions. Forced formation of L-swapped peptides detected by cysteine-SILAC. To demonstrate that the cysteine-SILAC method could credibly detect the formation of L-swapped species, we attempted to promote their formation by destabilizing the non-covalent HL interaction. Mild heat stress was administered to bsIgG1-EGFR×CD20 (κκ) at 45°C, 50°C, or 55°C directly after cFAE while retaining the reducing conditions. Subsequently, the reductant was removed by desalting, thus restoring the oxidative conditions that allow reformation of the inter chain disulfide bonds. After incubation at 45°C and 50°C, non-reduced peptide spectra remained similar to those of non-stressed IgG1-EGFR×CD20 (Figure 4A, B). Dissociation and re-association of the L and H in the absence of the inter-chain disulfide would require some degree of unfolding and refolding which is very unlikely under the non-denaturing cFAE conditions provided. Exposure to mild heat-stress (45°C and 50°C) prior to re-oxidation of the disulfide linkages did not cause Fab-arm degradation and/or L-swapping. After exposure to 55°C, the HH hinge peptide


Page 9 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


spectrum displayed increased ion signals representing the original ‘light’ and ‘heavy’ homodimer HH hinge peptides, indicating that heat stress affected the cFAE dynamic equilibrium. In addition, this temperature condition caused a substantial decrease of specifically the ‘heavy’ HL peptide indicating a destabilization of this Fab-arm (i.e. CD20 Fab-arm), but did not promote L-swapping even after increased thermal stress (Figure 4C). These results were consistent with previously obtained (non-reduced) differential scanning calorimetry (DSC) data that indicated a lower Fab domain melting temperature for 9

IgG1-CD20 compared to IgG1-EGFR . Heat-stress studies (10 min, 70°C) by Wang et al. showed that L 16

chain release is the first degradation event for IgG1 under non-reducing conditions . In the absence of inter-chain disulfide linkages, the highest temperature condition (55°C) was able to destabilize the Fab arm to such extent that oxidation was not fully reversible after reductant removal. Unpaired cysteines could thus produce nuclei initiating peptide scrambling, as manifested under alkaline pH. In an attempt to force convincing formation of L-swapped HL peptides and to obtain a positive control we repeated the proteolytic digestions in the absence of iodoacetamide (IAM), which is commonly included to prevent disulfide scrambling under alkaline pH conditions and/or in the presence of free 31

cysteine residues by alkylating free thiol groups . For the hinge peptide, results obtained were comparable to those generated in the presence of the alkylating agent IAM (Figure 4D-F). Under these 2+

conditions however, slightly elevated 633.25 m/z [M+2H]

peptide ion signals could be detected after

sample pretreatment at 45°C and 50°C, indicating that a HL hybrid peptide had been formed (Figure 4D,E). Remarkably, raising the temperature further to 55°C considerably increased the formation of HL ‘scrambled’ peptide at 633.25 m/z, indicating that during the proteolytic digestion a hybrid HL peptide had been formed (Figure 4F), of which the identity was subsequently confirmed by HCD MS/MS experiments (data not shown). We therefore conclude that cysteine-SILAC peptide analysis would be expected to detect cFAE-induced L-swapped species, if present, under condition where method-induced peptide disulfide scrambling is suppressed by the presence of the alkylating agent IAM.

Bottom-up structural assessment of bsAbs and impact of Fab-arm modifications on controlled Fab-arm exchange. Since SILAC might also be used to assess cFAE exchange efficiency, we performed bottom-up quantitative assessment of the HH hinge peptide of a panel of additional bsAbs in comparison to CEX and/or HIC analysis of bispecific Ab formation. The peptide mass spectra predominantly displayed one hinge peptide with a mass increase of 8.0 Da relative to the light cysteine reference peptide, representing a non-reduced hybrid hinge peptide with one ‘light’ and one ‘heavy’ HAHB peptide chain. To quantify, the total relative abundance of the three most abundant isotope peaks of the parental and hemi-labeled HH peptides was determined and showed to be in good agreement with CEX and/or HIC analyses results (n=10) (Table S3, supporting information). To assess whether the absence of HL scrambling during cFAE represents a general finding, we examined the HL peptide spectra for all generated bsAbs. Representative examples are provided for, a bsAb generated from parental mAbs deglycosylated prior to cFAE (bsIgG1-EGFRdeg×CD20deg), a bsAb



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 21

containing a glycosylated κ Fab-arm (bsIgG1-CET×CD20) and a bsAb containing a glycosylated λ Fabarm (bsIgG1-CD22×CD20) (Figure 5A-C). The HL peptide spectra clearly showed the “cysteine-SILAC peak pair” at the expected m/z values with a mass difference of +8.0 Da between the ‘light’ and ‘heavy’ disulfide-containing peptides (Figure 5A and 5B, right panels). No increase of hemi-labeled HL peptide signals was detected for any of the tested bsAbs, consistent with the bsIgG1-EGFR×CD20 analysis (Figure 3D). This indicated that neither Fc domain deglycosylation (bsIgG1-EGFRdeg×CD20deg), nor Fab glycosylation on the κ-Fab-arm (bsIgG1-CET×CD20) induced the formation of light chain swapped bsAb species during controlled Fab-arm exchange. Consistent with these findings, all other tested κκ bsAbs without Fab glycosylation (Table S3, Supporting information) did not show light chain swapped species (Figure S5, Supporting information). For bsIgG1-CD22×CD20, a ‘light’ IgG1λ exchanged with a ‘heavy’ IgG1κ Fab arm produced the non-reduced

IgG1λ

disulfide

linked

HL

peptide

comprising

the

amino

(H)SCDK/(L)TVAPTECS. This peptide was detected at 628.76 m/z [M+2H]

2+

acid

sequence

and the identity was

confirmed by HCD MS/MS experiments (data not shown). Non-reduced Lys-C HL peptides (HALAλ versus HBLBκ) differed in mass based on their amino acid sequence, but it would not be possible to identify Lswapped species without a SILAC label. Although HL peptides HBLBκ and HALAλ did not co-elute chromatographically, complicating quantitative measurement, we reasoned that L-swapped species could be quantitated since no other potentially interfering peptides in the relevant m/z range co-eluted within the chromatographic retention time window. In contrast to κκ bsAbs, for λκ bsAbs, two additional hemilabeled peptides could potentially be formed via light chain swapping, i.e. HBLAλ and HALBκ peptides (Figure 5C). A minor natural isotope peak of the non-labeled HALAλ peptide (628.76 m/z) coincided with the theoretical m/z value of the L-swapped peptide HALBκ (i.e. 630.76 m/z), but m/z signals of the second theoretical HBLAλ swapped ion peak (i.e. 633.26 m/z) were well resolved. The obtained results showed no evidence of the formation of HBLAλ peptide at 633.26 m/z, indicating the absence of L-swapped species in the investigated λκ bsAbs (Figure 5C). Also the peptide spectra recorded for other tested λκ bsAbs without Fab glycosylation failed to show detectable L-swapped species (Figure S5D-F, Supporting information). In summary, the average level of κκ bsAb HL hybrid peptide was determined at (0.81± 0.23)% (n=6) which is very similar to the obtained value (i.e. 1.0%) for the cFAE-control mixture lacking the enabling F405L mutation (Figure 3C, Table S3, supplementary). In addition, for λκ bsAb hybrid species an average level of (1.07 ± 0.2)% (n=4) was determined. The absence of L-swapping suggests that the L and H chains remain non-covalently paired under reducing conditions. Our findings are supported by Hydrogen/Deuterium exchange (HDX) experiments using two fully reduced mAbs (all inter-chains disulfides reduced) and demonstrated no elevated levels of deuterium uptake for peptides situated in the H:L interface. This indicated that the Fab domain was not subjected to any conformational change and/or subsequent light chain dissociation, further supporting 32

that the HL interaction is preserved in the reduced state . This is also consistent with the notion that


Page 11 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


strong non-covalent HL interactions are part of a rigorous assembly control mechanism during production by plasma cells

33

in which the CH1 domain (Figure 1A) is intrinsically unfolded during intracellular

production where it binds the BiP chaperone and requires L binding for chaperone release and antibody 34

secretion .

CONCLUSIONS Cysteine-SILAC facilitated in the direct analysis of disulfide-containing peptides and enabled us to characterize the structural integrity of bsAb generated by post-production recombination. In this work we show unambiguously that native HL pairing is retained when the HL inter-chain disulfides are transiently reduced and underscore the utility of the cFAE process to produce stable bsAbs. We envision that various inter-molecular disulfide recombination events occurring post-secretion from the host-cell could be monitored using cysteine-SILAC, such as recombination events observed in natural IgG4 antibodies under permissive redox conditions

12,13

or for alternative bsAb technologies using

11

similar approaches . Our findings also underscore the stability of Fab-domains under reducing conditions, which holds relevance to the production of Antibody Drug Conjugates (ADCs)

35,36

. More

generally, the formation of covalent inter-molecular disulfide bonds and disulfide scrambling present technical issues during recombinant expression, isolation, manufacturing and storage of proteins such as biotherapeutic agents, receptors, hormones, or enzymes

37-43

. The cysteine-SILAC approach can be

utilized to break the symmetry between homo-oligomeric inter- and intra-molecular interactions allowing individual bonds in heterodimeric interactions to be studied and thereby further the mechanistic understanding of disulfide bond formation and scrambling events.

REFERENCES (1) Davies, D. R.; Padlan, E. A.; Segal, D. M. Annu Rev Biochem 1975, 44, 639-667. (2) Brinkmann, U.; Kontermann, R. E. MAbs 2017, 9, 182-212. (3) Schaefer, W.; Volger, H. R.; Lorenz, S.; Imhof-Jung, S.; Regula, J. T.; Klein, C.; Molhoj, M. MAbs 2016, 8, 49-55. (4) Rajendra, Y.; Peery, R. B.; Hougland, M. D.; Barnard, G. C.; Wu, X.; Fitchett, J. R.; Bacica, M.; Demarest, S. J. Biotechnol Prog 2017, 33, 469-477. (5) Gunasekaran, K.; Pentony, M.; Shen, M.; Garrett, L.; Forte, C.; Woodward, A.; Ng, S. B.; Born, T.; Retter, M.; Manchulenko, K.; Sweet, H.; Foltz, I. N.; Wittekind, M.; Yan, W. J Biol Chem 2010, 285, 19637-19646. (6) Ridgway, J. B.; Presta, L. G.; Carter, P. Protein Eng 1996, 9, 617-621. (7) Schachner, L.; Han, G.; Dillon, M.; Zhou, J.; McCarty, L.; Ellerman, D.; Yin, Y.; Spiess, C.; Lill, J. R.; Carter, P. J.; Sandoval, W. Anal Chem 2016, 88, 12122-12127.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 21

(8) Merchant, A. M.; Zhu, Z.; Yuan, J. Q.; Goddard, A.; Adams, C. W.; Presta, L. G.; Carter, P. Nat Biotechnol 1998, 16, 677-681. (9) Labrijn, A. F.; Meesters, J. I.; de Goeij, B. E.; van den Bremer, E. T.; Neijssen, J.; van Kampen, M. D.; Strumane, K.; Verploegen, S.; Kundu, A.; Gramer, M. J.; van Berkel, P. H.; van de Winkel, J. G.; Schuurman, J.; Parren, P. W. Proc Natl Acad Sci U S A 2013, 110, 5145-5150. (10) Labrijn, A. F.; Meesters, J. I.; Priem, P.; de Jong, R. N.; van den Bremer, E. T.; van Kampen, M. D.; Gerritsen, A. F.; Schuurman, J.; Parren, P. W. Nat Protoc 2014, 9, 2450-2463. (11) Strop, P.; Ho, W. H.; Boustany, L. M.; Abdiche, Y. N.; Lindquist, K. C.; Farias, S. E.; Rickert, M.; Appah, C. T.; Pascua, E.; Radcliffe, T.; Sutton, J.; Chaparro-Riggers, J.; Chen, W.; Casas, M. G.; Chin, S. M.; Wong, O. K.; Liu, S. H.; Vergara, G.; Shelton, D.; Rajpal, A.; Pons, J. J Mol Biol 2012, 420, 204-219. (12) Schuurman, J.; Van Ree, R.; Perdok, G. J.; Van Doorn, H. R.; Tan, K. Y.; Aalberse, R. C. Immunology 1999, 97, 693-698. (13) van der Neut Kolfschoten, M.; Schuurman, J.; Losen, M.; Bleeker, W. K.; Martinez-Martinez, P.; Vermeulen, E.; den Bleker, T. H.; Wiegman, L.; Vink, T.; Aarden, L. A.; De Baets, M. H.; van de Winkel, J. G.; Aalberse, R. C.; Parren, P. W. Science 2007, 317, 1554-1557. (14) Gramer, M. J.; van den Bremer, E. T.; van Kampen, M. D.; Kundu, A.; Kopfmann, P.; Etter, E.; Stinehelfer, D.; Long, J.; Lannom, T.; Noordergraaf, E. H.; Gerritsen, J.; Labrijn, A. F.; Schuurman, J.; van Berkel, P. H.; Parren, P. W. MAbs 2013, 5, 962-973. (15) Moores, S. L.; Chiu, M. L.; Bushey, B. S.; Chevalier, K.; Luistro, L.; Dorn, K.; Brezski, R. J.; Haytko, P.; Kelly, T.; Wu, S. J.; Martin, P. L.; Neijssen, J.; Parren, P. W.; Schuurman, J.; Attar, R. M.; Laquerre, S.; Lorenzi, M. V.; Anderson, G. M. Cancer Res 2016, 76, 3942-3953. (16) Wang, Y.; Lu, Q.; Wu, S. L.; Karger, B. L.; Hancock, W. S. Anal Chem 2011, 83, 3133-3140. (17) Chen, X.; Wei, S.; Ji, Y.; Guo, X.; Yang, F. Proteomics 2015, 15, 3175-3192. (18) Ong, S. E.; Mann, M. Methods Mol Biol 2007, 359, 37-52. (19) Zhang, Z.; Pan, H.; Chen, X. Mass Spectrom Rev 2009, 28, 147-176. (20) Debaene, F.; Wagner-Rousset, E.; Colas, O.; Ayoub, D.; Corvaia, N.; Van Dorsselaer, A.; Beck, A.; Cianferani, S. Anal Chem 2013, 85, 9785-9792. (21) Haberger, M.; Leiss, M.; Heidenreich, A. K.; Pester, O.; Hafenmair, G.; Hook, M.; Bonnington, L.; Wegele, H.; Haindl, M.; Reusch, D.; Bulau, P. MAbs 2016, 8, 331-339. (22) Yin, Y.; Han, G.; Zhou, J.; Dillon, M.; McCarty, L.; Gavino, L.; Ellerman, D.; Spiess, C.; Sandoval, W.; Carter, P. J. MAbs 2016, 8, 1467-1476. (23) Teeling, J. L.; French, R. R.; Cragg, M. S.; van den Brakel, J.; Pluyter, M.; Huang, H.; Chan, C.; Parren, P. W.; Hack, C. E.; Dechant, M.; Valerius, T.; van de Winkel, J. G.; Glennie, M. J. Blood 2004, 104, 1793-1800. (24) Salazar, A.; Keusgen, M.; von Hagen, J. Amino Acids 2016, 48, 1161-1171. (25) Waanders, L. F.; Hanke, S.; Mann, M. J Am Soc Mass Spectrom 2007, 18, 2058-2064. (26) Qian, J.; Liu, T.; Yang, L.; Daus, A.; Crowley, R.; Zhou, Q. Anal Biochem 2007, 364, 8-18.


Page 13 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(27) Ayoub, D.; Jabs, W.; Resemann, A.; Evers, W.; Evans, C.; Main, L.; Baessmann, C.; WagnerRousset, E.; Suckau, D.; Beck, A. MAbs 2013, 5, 699-710. (28) Gadgil, H. S.; Bondarenko, P. V.; Pipes, G. D.; Dillon, T. M.; Banks, D.; Abel, J.; Kleemann, G. R.; Treuheit, M. J. Anal. Biochem. 2006, 355, 165-174. (29) Ahmad, Y.; Lamond, A. I. Trends Cell Biol 2014, 24, 257-264. (30) Zhang, Q.; Flynn, G. C. J Biol Chem 2013, 288, 34325-34335. (31) Sung, W. C.; Chang, C. W.; Huang, S. Y.; Wei, T. Y.; Huang, Y. L.; Lin, Y. H.; Chen, H. M.; Chen, S. F. Biochim Biophys Acta 2016, 1864, 1188-1194. (32) Pan, L. Y.; Salas-Solano, O.; Valliere-Douglass, J. F. Anal Chem 2014, 86, 2657-2664. (33) Feige, M. J.; Groscurth, S.; Marcinowski, M.; Shimizu, Y.; Kessler, H.; Hendershot, L. M.; Buchner, J. Mol Cell 2009, 34, 569-579. (34) Feige, M. J.; Buchner, J. Biochim Biophys Acta 2014, 1844, 2024-2031. (35) Valliere-Douglass, J. F.; Hengel, S. M.; Pan, L. Y. Mol Pharm 2015, 12, 1774-1783. (36) Pan, L. Y.; Salas-Solano, O.; Valliere-Douglass, J. F. Anal Chem 2015, 87, 5669-5676. (37) Vazquez-Rey, M.; Lang, D. A. Biotechnol Bioeng 2011, 108, 1494-1508. (38) Remmele, R. L., Jr.; Callahan, W. J.; Krishnan, S.; Zhou, L.; Bondarenko, P. V.; Nichols, A. C.; Kleemann, G. R.; Pipes, G. D.; Park, S.; Fodor, S.; Kras, E.; Brems, D. N. J Pharm Sci 2006, 95, 126145. (39) Huh, J. H.; White, A. J.; Brych, S. R.; Franey, H.; Matsumura, M. J Pharm Sci 2013, 102, 1701-1711. (40) Iwura, T.; Fukuda, J.; Yamazaki, K.; Kanamaru, S.; Arisaka, F. J Biochem 2014, 155, 63-71. (41) Van Buren, N.; Rehder, D.; Gadgil, H.; Matsumura, M.; Jacob, J. J Pharm Sci 2009, 98, 3013-3030. (42) Trivedi, M. V.; Laurence, J. S.; Siahaan, T. J. Curr Protein Pept Sci 2009, 10, 614-625. (43) Zhang, L.; Chou, C. P.; Moo-Young, M. Biotechnol Adv 2011, 29, 923-929.

ASSOCIATED CONTENT

Supporting Information. Additional information as noted in text.

Author Contributions E.T.J.v.d.B and R.N.d.J designed the research; R.v.d.B, K.S and P.P. carried out the experiments; E.T.J.v.d.B., A.F.L., R.v.d.B., K.S., J.P.M.M., J.S, P.W.H.I.P, and R.N.d.J interpreted the data; E.T.J.v.d.B., A.F.L., K.S., J.P.M.M., J.S, P.W.H.I.P, and R.N.d.J wrote the manuscript.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 21

ACKNOWLEDGMENT We thank E. Noordergraaf-Slootjes and D. Willemsz for expert technical assistance. Furthermore we would like to thank M.J. Gramer for advice on typical cysteine consumption rates.

FIGURES

Figure 1. Schematic overview of IgG1 structure and cysteine-SILAC approach. (A) Typical IgG1 structure composed of two heavy (H) chains consisting of four domains (VH, CH1, CH2 and CH3), and two light (L) chains consisting of two domains (VL and CL), with C and V indicating variable and constant domains respectively. HL chain pairs are indicated either in blue or green. Intra chain disulfide bonds are highlighted in red and inter chain disulfides are indicated in dark purple (HL) and light purple (HH). (B) Schematic overview of cysteine-SILAC approach. One parental mAb was metabolically labeled with heavy cysteines and mixed one-to-one with a normal cysteine mAb under controlled reducing conditions to initiate FAE exchange in order to generate bsAbs with one ‘heavy’ cysteine Fab-arm. The cysteineSILAC methodology is further illustrated in Figure S1C in the Supporting Information.

Figure 2. Middle-up Fab fragment analysis of bispecific IgG do not show L-chain swapped Fab fragment species. Illustrative MS spectra of different intact bispecific antibodies (left panels) and derived Fab-fragments (right panels) for

(A) IgG1-EGFR×CD20; (B) IgG1-EGFRdeg×CD20deg; (C) IgG1-

CET×CD20 and (D) IgG1-CD22×CD20. Filled and unfilled circles indicate SILAC-labeled and non-labeled molecular species, respectively. Peaks marked with both circles represent the bsAb. Half-filled circles in the Fab fragment spectra indicate the position of the expected mass values of calculated potential L chain-swapped Fab masses. Additional spectra are provided in Figure S3 in the Supporting Information.

Figure 3. Bottom-up analysis of non-reduced HL Lys-C peptides in control mixtures defines background levels of cysteine SILAC approach. Mass spectra representing inter-chain HH hinge 6+

[M+6H]

(left panels) and HL Fab-arm [M+2H]

2+

disulfide peptides (right panels). Positions at which

peaks for the unlabeled; hemi labeled and doubly labeled HH hinge peptides (left panels) and HL peptides (right panels) would appear are indicated with dashed lines. The most abundant m/z values are indicated. (A) Spectrum highlighting the non-reduced ‘light’ H-H hinge and accompanied ‘light’ Fab-arm HL peptide . (B) Spectrum of a one-to-one mixture of ‘light’ and ‘heavy’ parental mAbs indicating a nonexchanged HH peptide ‘SILAC-pair’ with a mass difference of 16 Da and the HL peptide ‘SILAC-pair’ with a mass difference of 8 Da. (C) Spectrum of a mixture of IgG1-EGFR-WT (lacking the cFAE-enabling


Page 15 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


F405L mutation) + IgG1-CD20 exposed to cFAE. (D) Spectrum indicating the hybrid hemi-labeled HH hinge peptide peak after exchange to IgG1-EGFR×CD20 and accompanying HL peptides.

Figure 4. Non-reduced Lys-C peptides from temperature stressed IgG allow the detection of HH and HL peptides of interest. Mass spectra representing inter-chain HH hinge [M+6H] 2+

HL Fab-arm [M+2H]

6+

(left panels) and

(right panels) disulfide peptides. Controlled Fab arm exchange was performed by

reduction, application of temperature stress at 45°C (A, D), 50°C (B, E) or 55°C (C, F), and reoxidation, followed by non-reducing bottom-up analysis with (A, B, C) and without Iodoacetamide (IAM) added prior to Lys-C digestion (D, E, F). Positions at which peaks for the unlabeled and doubly labeled HH hinge peptides (left panels) and hemi-labeled HL peptides (right panels) would appear are indicated with dashed lines.

Figure 5. Bottom-up analysis of bsAbs comprising a variety of biophysical molecular properties highlighting the absence of L chain swapping. 6+

[M+6H]

(left panels) and HL Fab-arm [M+2H]

2+

Mass spectra representing inter-chain HH hinge

disulfide peptides (right panels). Bottom-up analysis of

non-reduced Lys-C peptides of bsAb examples generated from (A) parental mAbs deglycosylated prior to cFAE (bsIgG1-EGFrdeg×CD20deg), (B) a bsAb containing a glycosylated κ Fab-arm (bsIgG1CET×CD20), and (C) a bsAb containing a glycosylated λ Fab-arm (bsIgG1-CD22×CD20). Positions at which peaks for the unlabeled and doubly labeled HH hinge peptides (left panels) and hemi-labeled HL peptides (right panels) would appear are indicated with dashed lines. The most abundant m/z values are indicated. Additional examples are provided in Supporting Information Figure S5.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 21

“For TOC only”


Page 17 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. Schematic overview of IgG1 structure and cysteine-SILAC approach. (A) Typical IgG1 structure composed of two heavy (H) chains consisting of four domains (VH, CH1, CH2 and CH3), and two light (L) chains consisting of two domains (VL and CL), with C and V indicating variable and constant domains respectively. HL chain pairs are indicated either in blue or green. Intra chain disulfide bonds are highlighted in red and inter chain disulfides are indicated in dark purple (HL) and light purple (HH). (B) Schematic overview of cysteine-SILAC approach. One parental mAb was metabolically labeled with heavy cysteines and mixed one-to-one with a normal cysteine mAb under controlled reducing conditions to initiate FAE exchange in order to generate bsAbs with one ‘heavy’ cysteine Fab-arm. The cysteine-SILAC methodology is further illustrated in Figure S1C in the Supporting Information. 167x71mm (300 x 300 DPI)

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. Middle-up Fab fragment analysis of bispecific IgG do not show L-chain swapped Fab fragment species. Illustrative MS spectra of different intact bispecific antibodies (left panels) and derived Fab-fragments (right panels) for (A) IgG1-EGFR×CD20; (B) IgG1-EGFRdeg×CD20deg; (C) IgG1CET×CD20 and (D) IgG1-CD22×CD20. Filled and unfilled circles indicate SILAC-labeled and non-labeled molecular species, respectively. Peaks marked with both circles represent the bsAb. Half-filled circles in the Fab fragment spectra indicate the position of the expected mass values of calculated potential L chainswapped Fab masses. Additional spectra are provided in Figure S3 in the Supporting Information. 153x188mm (300 x 300 DPI)


Page 18 of 21

Page 19 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3. Bottom-up analysis of non-reduced HL Lys-C peptides in control mixtures defines background levels of cysteine SILAC approach. Mass spectra representing inter-chain HH hinge [M+6H]6+ (left panels) and HL Fab-arm [M+2H]2+ disulfide peptides (right panels). Positions at which peaks for the unlabeled; hemi labeled and doubly labeled HH hinge peptides (left panels) and HL peptides (right panels) would appear are indicated with dashed lines. The most abundant m/z values are indicated. (A) Spectrum highlighting the non-reduced ‘light’ HH hinge and accompanied ‘light’ Fab-arm HL peptide . (B) Spectrum of a one-to-one mixture of ‘light’ and ‘heavy’ parental mAbs indicating a non-exchanged HH peptide ‘SILAC-pair’ with a mass difference of 16 Da and the HL peptide ‘SILAC-pair’ with a mass difference of 8 Da. (C) Spectrum of a mixture of IgG1-EGFR-WT (lacking the cFAE-enabling F405L mutation) + IgG1CD20 exposed to cFAE. (D) Spectrum indicating the hybrid hemi-labeled HH hinge peptide peak after exchange to IgG1-EGFR×CD20 and accompanying HL peptides. 84x132mm (300 x 300 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment

Page 20 of 21

Page 21 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4. Non-reduced Lys-C peptides from temperature stressed IgG allow the detection of HH and HL peptides of interest. Mass spectra representing inter-chain HH hinge [M+6H]6+ (left panels) and HL Fab-arm [M+2H]2+ (right panels) disulfide peptides. Controlled Fab arm exchange was performed by reduction, application of temperature stress at 45°C (A, D), 50°C (B, E) or 55°C (C, F), and reoxidation, followed by non-reducing bottom-up analysis with (A, B, C) and without Iodoacetamide (IAM) added prior to Lys-C digestion (D, E, F). Positions at which peaks for the unlabeled and doubly labeled HH hinge peptides (left panels) and hemi-labeled HL peptides (right panels) would appear are indicated with dashed lines. 177x98mm (300 x 300 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 5. Bottom-up analysis of bsAbs comprising a variety of biophysical molecular properties highlighting the absence of L chain swapping. Mass spectra representing inter-chain HH hinge [M+6H]6+ (left panels) and HL Fab-arm [M+2H]2+ disulfide peptides (right panels). Bottom-up analysis of non-reduced Lys-C peptides of bsAb examples generated from (A) parental mAbs deglycosylated prior to cFAE (bsIgG1-EGFrdeg×CD20deg), (B) a bsAb containing a glycosylated κ Fab-arm (bsIgG1-CET×CD20), and (C) a bsAb containing a glycosylated λ Fab-arm (bsIgG1-CD22×CD20). Positions at which peaks for the unlabeled and doubly labeled HH hinge peptides (left panels) and hemi-labeled HL peptides (right panels) would appear are indicated with dashed lines. The most abundant m/z values are indicated. Additional examples are provided in Supporting Information Figure S5. 123x124mm (300 x 300 DPI)


Page 22 of 21

Cysteine-SILAC Mass Spectrometry Enabling the Identification and

Recommend Documents