Detection and Identification of the Vibrational Markers for the

Subscriber access provided by UCL Library Services

Article

Detection and Identification of the Vibrational Markers for the Quantification of Methionine Oxidation in Therapeutic Proteins Gurusamy Balakrishnan, Gregory V. Barnett, Sambit R. Kar, and Tapan K. Das Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.8b01238 • Publication Date (Web): 09 May 2018 Downloaded from http://pubs.acs.org on May 10, 2018

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 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 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.

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 29 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

Detection and Identification of the Vibrational Markers for the Quantification of Methionine Oxidation in Therapeutic Proteins

Gurusamy Balakrishnan*†, Gregory V. Barnett*, Sambit R. Kar, and Tapan K. Das

Molecular and Analytical Development, Bristol-Myers Squibb 311 Pennington Rocky Hill Road, Pennington, New Jersey 08534, United States

*These authors contributed equally to this work †Corresponding Author: email: [email protected] Phone: 609-818-3452

ACS Paragon Plus Environment

1

Analytical Chemistry 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 29

Abstract Methionine oxidation is a major degradation pathway in therapeutic proteins which can impact the structure and function of proteins as well as risk to drug product quality. Detecting Met oxidation in proteins by peptide mapping followed by liquid chromatography with mass spectrometry (LC-MS) is the industry standard but is also labor intensive and susceptible to artifacts. In this work, vibrational difference spectroscopy in combination with 18O isotopic shift enabled us to demonstrate the application of Raman and FTIR techniques for the detection and quantification of Met oxidation in various therapeutic proteins, including, mAbs, fusion proteins and antibody drug conjugate. Vibrational markers of Met oxidation products, such as, sulfoxide and sulfone, corresponding to S=O and C-S=O stretching frequencies were unequivocally identified based 18O isotoptic shifts. The intensity of the isolated νC-S Raman band at 702 cm-1 was successfully applied to quantify the average Met oxidation level in multiple proteins. These results are further corroborated by oxidation levels measured by tryptic peptide mapping and thus confirmed Met oxidation level derived from Raman and mass spectrometry are indeed consistent with each other. Thus, we demonstrate the broader application of vibrational spectroscopy to detect the subtle spectral changes associated with various chemical or physical degradation of proteins, including Met oxidation as well as higher order structural changes.

Keywords: Raman, FTIR, methionine, oxidation, monoclonal antibodies, hydrogen peroxide, isotopic shift


2

Page 3 of 29 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


Introduction Proteins and peptides have gained prominence as therapeutic agents in the treatment of a number of cancers and autoimmune diseases due to their high specificity towards the targeted antigen. In particular, monoclonal antibodies (mAbs) are attractive drug candidates due to long in-vivo half-lives, favorable toxicity profiles, as well as amenability to molecular design for targeted binding.1 The complex nature of biomolecules with their unique chemical composition, higher order structure, and various instabilities (chemical, conformational, colloidal, and interfacial), pose several challenges to product manufacturing, storage, and delivery. Methionine (Met) oxidation is a major degradation pathway in therapeutic proteins that can impact product quality and patient’s safety.2-7 Oxidation of Met residues in a protein can occur upon exposure to reactive oxygen species (ROS), such as peroxides or free-radicals.2 Common sources of ROS encountered during biologics manufacturing and storage include alkyl peroxides resulting from the degradation of surfactants, trace level transition metals found in raw materials or storage and processing surfaces, or superoxides generated through prolonged exposure of a protein formulation to UV or visible light.8-9 Oxidation in mAbs occurs readily in two solvent exposed Met residues located in the highly conserved Fc (crystallizable fragment) region. These residues are located near the neonatal Fc receptor (FcRn) binding region, thereby altering drug clearance rates,5, 10-12 and located near the Fcγ receptor (FcγRIIIa) binding region, thereby impacting antibody dependent cytotoxic effector function in IgG1.13-14 Further, the antigen binding regions of mAbs can also have Met residues susceptible to oxidation that may lead to altered biological activity, especially when the oxidized residues are located at or in proximity to the complementary determining regions (CDR).15-16


3


Page 4 of 29

Detecting Met oxidation in proteins by peptide mapping followed by liquid chromatography with mass spectrometry is the industry standard as it highly sensitive to the oxidation levels of individual residues.17-18 Nevertheless, peptide mapping is labor-intensive, involving protein denaturation, disulfide reduction, alkylation, and enzymatic digestion, all of which can lead to method-induced artifacts.19-21 Separation methods such as reverse phase, mixed-mode, hydrophobic interaction, Protein A, and FcRn affinity chromatography have been applied with some success to monitor protein oxidation with UV detection, but all have limitations associated with chromatographic separation.12, 22 Robust and sensitive tools orthogonal to peptide mapping are needed to monitor in-situ oxidation of protein in real-time. Protein characterization using vibrational spectroscopy is a promising orthogonal technique that can provide multiple structural markers in a single experiment and it is a non-destructive solution measurement method with minimal sample preparation, thereby preserving native protein conformations. 23-27 Limited number of studies are available on the detection of Met oxidation in proteins by vibrational spectroscopy and the vibrational band assignments are inconsistent.28-30 Therefore, detailed studies are necessary to establish vibrational markers that can be used to detect and quantify Met oxidation products in protein and peptides. In this work, we have identified both Raman and FTIR bands associated with Met sulfoxide, the predominant Met oxidation product, based on 18O isotopic shifts observed in methionine amino acid. These vibrational assignments in combination with difference spectral technique, enable us to identify Met oxidation markers in proteins from both Raman and FTIR spectroscopic methods. The average Met oxidation level in a number of proteins including mAbs, fusion proteins, and an antibody-drug conjugate are successfully determined based on the intensity of the isolated νC-S Raman band at 702 cm-1. These results are further supported by


4

Page 5 of 29 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


data from tryptic peptide mapping coupled with LC-MS measurements. Thus, this study demonstrates the unique potential of Raman spectroscopy as a simple yet sophisticated orthogonal tool for the detection and quantification of Met oxidation levels in proteins. Materials and Methods Sample preparation Therapeutic proteins used in this work were expressed in Chinese hamster ovary (CHO) cells and purified by standard protein purification processes including protein A and ionexchange chromatographic steps. The antibody drug conjugate was manufactured by thiol clickchemistry to conjugate the small molecule drug to the antibody as described in published literature.31 To test broader applicability of the vibrational spectroscopic technique, a total of nine therapeutic proteins including two IgG4 (mAb1 and mAb2), 3 IgG1 (mAb3-mAb5), two fusion proteins (FP1 and FP2), an antibody drug conjugate (ADC+) and its parent unconjugated mAb (ADC-) were included in this study. All therapeutic proteins were dialyzed into 10 mM sodium phosphate at pH 6.5, filtered using 0.22 µm filter (Millipore, Billerica, MA), and stored at 4°C until further use. Dialysate was collected and used for background subtraction in spectroscopic measurements. DL-Met and hydrogen peroxide (>30% for ultra-analysis) were purchased from Sigma Aldrich. A working stock of hydrogen peroxide was prepared by diluting hydrogen peroxide to 3 % (w/v) using distilled deionized water. In addition, 18O hydrogen peroxide solution (2-3 % (w/v) in water, 90 atom % 18O) and 18O water (97 atom % 18O) were also purchased from Sigma Aldrich. A 50 mM Met solution (free Met solution) was prepared using distilled deionized water (Millipore, Billerica, MA). The appropriate amount of 3 % (w/v) hydrogen peroxide (16O or 18O) was added to the 50 mM Met solution to obtain a final Met:peroxide molar ratio of 1:1. For


5


Page 6 of 29

protein solutions, 4 µL of hydrogen peroxide (3% stock) was added to 96 µL of 50 mg/mL protein solutions to obtain a final protein:peroxide molar ratio of 1:100. “Free Met” and “free Met sulfoxide” terms are used in this article to differentiate from the Met residues and their oxidation products in proteins. Raman Spectroscopy Raman spectroscopic measurements were performed to monitor the oxidation reaction of therapeutic protein solutions or aqueous free Met solution by hydrogen peroxide. Raman spectra were acquired using Helix Raman system (Malvern Instruments, Malvern, UK) equipped with 785 nm excitation wavelength and a Peltier temperature controlled sampling compartment equilibrated at 25°C. 60-80 µL of reaction mixture was loaded into a cuvette (Malvern Instruments) and sealed to minimize sample evaporation during the data collection period of 24 hrs. 48 spectra were continuously collected with each spectral acquisition time of 30 min (60 coadds of 30 sec scans). Background spectra were obtained from the protein dialysis filtrate or water and subtracted from the protein or free Met spectra, respectively, prior to taking the difference spectra. Difference Raman spectra were obtained by the subtraction of the initial spectra from the subsequent time resolved spectra or by the subtraction of the free Met spectra from the free Met sulfoxide spectra. Integrated intensity of the Raman band at 702 cm-1 (νC-S) from the difference spectra was used to quantify the Met oxidation levels in proteins as described in Supporting Information. Data analyses were performed using Helix Analyze software from Malvern Instruments. ATR-FTIR Spectroscopy FTIR spectra were recorded on a Nicolet iZ10 spectrometer (Thermo Scientific) equipped with a diamond ATR (attenuated total reflection) accessory and a DTGS (Deuterated Tri-Glycine


6

Page 7 of 29 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


Sulfate) detector. ATR accessory is made up of diamond crystal operating at a nominal incident angle of 45° with ten reflections (ConcentratIR2, Harrick Scientific Products, Inc.). About 30-50 µL protein sample was deposited in the ATR cell and covered to minimize solvent evaporation and keep it free from dusts. The reaction of free Met and hydrogen peroxide was carried out using excess reaction volume (~100 µL) to minimize evaporation effects during the spectral measurement period of 3 hrs. 60 time-dependent spectra were collected continuously with each spectral accusation time of 3 min (average of 128 scans) for a period of 3 hrs. All spectra were acquired in the single beam mode with a 4-cm-1 resolution with an average of 128 to 512 scans for both samples as well as buffer background. Both spectral acquisition and processing were conducted using Omnic Software supplied with the spectrometer. The absorption spectrum of each solution was obtained from its single beam spectrum using the ATR window reference spectrum. ATR correction and automatic atmospheric suppression were performed for each spectrum. Water spectrum was subtracted from both protein as well as buffer spectra. Final spectra were obtained by subtraction of the buffer spectra from the sample spectra. Results Vibrational spectra of proteins are ‘crowded’ with bands from multiple structural markers, including secondary structure-sensitive amide bands and a number of tertiary structure-sensitive bands from tryptophan, tyrosine and disulfide. Signal strengths of these individual markers are dependent on the relative abundance of the bond contributing to specific marker and their Raman scattering strength or cross section.23-27 A typical monoclonal antibody contains about 1320 amino acid residues and has an average of about 20,000 atoms. It is difficult to identify a particular band from a functional group, for example S=O stretching band for oxidized Met from


7


Page 8 of 29

the ‘crowded’ protein background spectra. Various techniques, including resonance Raman spectroscopy (rRS), isotopic shifts, spectral deconvolution, and difference spectroscopy are employed to resolve the overlapping bands and identify specific vibrational markers. However, due to lack of near-UV or visible absorption, Met residues are not suitable candidate for rRS and therefore detecting Met oxidation in protein is very challenging. Prior studies have reported that when free Met was exposed to gamma-irradiation, multiple Raman bands in the region 650-730 cm-1 were show intensity increase, which were subsequently assigned to νC-S.29, 32 Dong et al. reported bands for νS=O at 1010 cm-1 and νC-S at 702 cm-1 of amyloid plaques by Raman microscopy.33 In a separate study, when a mAb was oxidized by peroxide, FTIR and Raman bands appeared at 1044 cm-1 and were then assigned to νS=O.28, 30 Therefore, prior work has shown multiple vibrational bands may be indicative of Met oxidation, but further work is need to definitely establish bands that are sensitive to Met oxidation. We have identified the Met oxidation sensitive vibrational bands through 18O isotopic shifts in free Met sulfoxide in combination with in-situ time-dependent spectral measurement on protein+H2O2 reaction mixtures followed by difference spectroscopy. We demonstrate that our methodology provides unambiguous vibrational assignment and highly resolved signatures exclusively from Met oxidation induced spectral changes in proteins. Free Methionine Oxidation by Peroxide Oxidation of free Met by hydrogen peroxide (at 25 °C) was monitored by in-situ Raman spectroscopy. The resulting time dependent spectra are shown in Figure 1 with vibrational band assignments adapted from literature.23, 27 Time-dependent Raman spectra of free Met oxidation reaction (Figure 1a) show intensities increase over time for bands at 702, 945, 1010 and 1428 cm-1 due to Met sulfoxide formation and decrease in intensity at 876 cm-1 corresponding to the


8

Page 9 of 29 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


concomitant loss of H2O2 as reaction proceeds. The formation of Met sulfoxide is essentially complete within 3 hours as the spectrum at the completion of the reaction is nearly identical to the Raman spectrum of pure methionine sulfoxide (Figure S1). The time-dependent concentration profiles of individual components (Met, Met sulfoxide, Met sulfone and H2O2) were readily resolved and determined by linear least squares fitting of the time-dependent spectra to the pure spectra of individual components in the reaction mixture (Figure S1). The residuals for least squares show little or no features and therefore confirm a good fit of data to the model. Time dependent concentration profiles follow a single exponential decay (inset in Figure 1a) with a time constant of 0.5± 0.03 hr for the loss of Met, which is consistent with the one-step formation of Met sulfoxide without presence of intermediate species. At longer time (>3 hr), slow formation of sulfone follows linearly with the final concentration reaching about 5% at 10 hrs. Free Met was also reacted with 18O hydrogen peroxide and the resultant Raman spectra of Met sulfoxide containing 18O (blue) and the naturally occurring 16O (red) isotopes and their difference (green) spectrum are shown in Figure 1b. The difference spectrum shows cancelling out of all bands except at 1010 and 945 cm-1, indicating only these two bands are associated with S=O bond of sulfoxide. The observed 18O isotopic shifts of -34 and -12 cm-1 are consistent with the predicted values for νS=O and νC-S=O modes, respectively, based on the ratio of reduced mass calculated for harmonic oscillators.34 Similarly, Raman spectra of free Met sulfone containing 16O (red) and 18O (blue) isotopes and their difference (green) spectrum are shown in Figure 1c. A single band at 1134 cm-1 shows an

18

O shift of -38 cm-1, which is consistent with

the predicted value for pure νS=O mode of sulfone.


9


Page 10 of 29

Two bands at 702 cm-1 and 1010 cm-1 correspond to νC-S and νS=O modes and show strong to moderate increase in intensity over time with increasing oxidation. Some of the weaker bands also showed increase in intensity upon oxidation, such as the νC-S=O at 945 cm-1 and δCH2 at 1428 cm-1. Due to strong overlapping bands from phenylalanine and tryptophan, the νS=O band at 1010 cm-1 is not ideal for the detection or quantification of Met oxidation in proteins. However, the isolated and well-resolved νC-S Raman band at 702 cm-1, as demonstrated below, is found to be a reliable Raman marker for detection as well as quantification of Met oxidation in proteins. Oxidation of Met by hydrogen peroxide at room temperature was also monitored by insitu ATR-FITR spectroscopy, and the resulting time dependent spectra are shown in Figure 2. FTIR spectra of Met (Figure S2 - pure Met spectrum) show strong bands at 1600 and 1408 cm-1, corresponding to the asymmetric and symmetric νCOO¯ stretching modes. Other strong bands observed at 1518 and 1348 cm-1 are due to δNH3+ and δCH3 bending modes.27 A strong band at 1010 cm-1 and a weak band at 945 cm-1 grow over time and these bands are attributed to νS=O and νC-S=O, respectively, consistent with the assignments from Raman spectroscopy as described above. Free Met was also reacted with 18O hydrogen peroxide and the resultant FTIR spectra of Met sulfoxide containing 18O (blue) and 16O (red) isotopes, and their difference (green) spectra are shown in Figure 2b. Similarly, the difference spectrum shows cancelling out of all bands except at 1010 and 945 cm-1, indicating that these two bands are associated with the S=O bond of sulfoxide consistent with Raman spectroscopy. 18O isotopic shifts of -34 and -14 cm-1 for νS=O and νC-S=O modes, respectively, are consistent with those predicted based on the reduced mass ratio calculated for harmonic oscillators.34 The strong FTIR band of νS=O mode at 1010 cm-1 is


10

Page 11 of 29 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


can be used as an FTIR marker for Met oxidation in proteins as described below. Thus, the vibrational frequencies of νS=O and νC-S=O are unambiguously established from both Raman and FTIR spectroscopy. Detailed spectral deconvolution of the Raman and FTIR spectra of Met sulfoxide band at 1010 cm-1 (Figure S3) reveals complex intensity patterns for both O16 (top panel) and O18 (bottom panel) isotopes with 3-4 overlapping bands. Upon isotopic substitution, the νS=O band at 1010 cm-1 (red) and the νC-S=O band at 945 cm-1 (blue) undergo expected frequency downshifts along with the unexpected loss of intensity for former whereas nearly no intensity change for the later. Similar intensity decreases linked to isotopic substitution have been attributed to change in Fermi resonance.35-36 Therefore, it is possible that the intensity decrease observed for νS=O band is due to Fermi resonance between 1010 cm-1 and a combination band, possibly from the bands observed at 470 and 540 cm-1 (Figure 1a). Thus, we suggest that the frequency downshift (1010 to 976 cm-1) associated with 18O substitution disrupts the Fermi resonance resulting in loss of intensity enhancement. Met Oxidation in Proteins Monitored by Raman Spectroscopy The application of Raman spectroscopy for monitoring oxidation of methionine residues in proteins is demonstrated using a series of therapeutic proteins, including IgG1 and IgG4 mAbs (mAb1-mAb5), fusion proteins (FP1, FP2), and antibody drug conjugate (ADC+) as well as its unconjugated mAb (ADC-), to illustrate the broad applicability of the technique. Oxidation of mAb1 (50 mg/mL, ~0.3 mM) by hydrogen peroxide (~30 mM final concentration, 25 °C) was monitored over a period of 24 hr using in-situ Raman spectroscopy as shown in overlays of the full spectra at various time points in Figure 3a. Raman bands of the protein are labeled based on assignments available from the literature.23 With the exception of a subtle increase in intensities


11


Page 12 of 29

of the 680-720 cm-1 region (inset in Figure 3a), no changes in the time-dependent Raman spectra were observed. No change in Amide I band at 1671 cm-1 and Amide III at 1238 cm-1 as well as Trp (1553 cm-1 and 1359/1339 cm-1) and Tyr (856/826 cm-1) bands indicate no significant changes in secondary and tertiary structures. Conformational changes induced by Met oxidation in mAbs have been detected by NMR and hydrogen/deuterium exchange mass (HDX-MS) spectroscopic studies3,15 but these conformational changes are considered to be minor and likely not detectable by vibrational spectroscopy. Selection of an optimal reaction conditions such as peroxide level, temperature and duration is critical to minimize harsh reaction condition leading to unintended gross structural impact on proteins. The small increase in intensity in the region 680-720 cm-1 can be monitored because of the fortuitous absence of any other protein or excipient spectral features in this region. As demonstrated in Figure 1, this regions corresponds to the νC-S band for Met sulfoxide. Raman difference spectra of mAb1 at various time points relative to initial spectrum were then generated to resolve the subtle differences for oxidation and other degradation products as well as higher order structural changes as shown in Figure 3b. Our approach of in-situ Raman spectral measurement followed by the generation of difference spectra eliminates various possible artifacts, including mismatch of the formulation components, concentration or cleanness of the cuvette and cuvette positioning. The free Met sulfoxide difference spectra is also shown (in red) to aid the identification sulfoxide specific markers in the mAb1 difference Raman spectra. Also, no other detectable changes in the difference spectra are observed, confirming the reaction condition is optimal for selective oxidation of Met without detectable changes in the protein higher order structure.


12

Page 13 of 29 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


Finally, the Raman difference spectra (relative to initial time point) of various proteins including multiple mAbs (mAb1 to mAb5), fusion proteins (FP1 and FP2), and antibody drug conjugate (ADC+) and its unconjugated mAb (ADC-) treated with hydrogen peroxide for 24 hr are shown in Figure 3c. The Raman difference spectra of all the proteins are similar to difference spectrum of free (Met (O) - Met), thus confirming the general applicability of this technique for the detection of Met oxidation in proteins. Also to be noted that other spectral changes are not detected in the difference spectra, except a weak band at 1045 cm-1 for ADC and FPs. Preliminary data on cysteine oxidation reveals that the band observed at 1045 cm-1 for ADC and FPs is νS=O band of oxidized cysteine (data not shown). Together, these data clearly demonstrate that Raman spectroscopy can reliably detect Met oxidation of monoclonal antibodies and proteins and peptides in general. The average Met oxidation levels in proteins over time were determined by normalizing the intensity of the νC-S band at 702 cm-1 by the molar intensity of this band in free Met sulfoxide as 100%, measured under the same experimental conditions (see Supporting information for details). Average Met oxidation kinetics in the mAbs are shown in Figure 4a while those of the fusion proteins and the ADC are shown in Figure 4b. Met oxidation time constants (Table S1) for all the mAbs are similar with values ranging from 4.7 to 6.9 hr, while those for the fusion proteins are about 1.8 hr. In-situ Raman difference spectroscopy can detect unique oxidation behaviors of Fc-fusion proteins and minor differences between mAbs. The Met oxidation levels determined by peptide mapping followed by LC-MS (Table S1), show clear correlation with Met oxidation levels obtained by Raman and LC-MS detection (Figure 4c) and thus demonstrates the utility of Raman spectroscopy as a tool for in-situ monitoring and


13


Page 14 of 29

quantitation of Met oxidation in proteins. Based on the initial oxidation level measured at ~ 1 hr for proteins, the average lowest measureable oxidation level is determined to be about 5-10% . Met Oxidation in Proteins Monitored by ATR-FITR Spectroscopy A subset of the proteins in the Raman study above was used to explore the feasibility of ATR-FTIR for monitoring Met oxidation in proteins. ATR-FTIR spectra were measured for oxidized proteins (24 hr treatment by H216O2 or H218O2) and spectra of selected samples are shown in Figure 5. ATR-FTIR spectra of these proteins are dominated by strong bands at 1636 and 1546 cm-1, corresponding to AmI and AmII bands of β-sheet structure. Bands with moderate to weak intensities observed at 1454, 1400 and 1240 cm-1 are attributed to δCH2/δCH3, νCOO¯ and AmIII bands, respectively. Difference spectra of proteins (oxidized - initial) reveals bands associated with oxidized Met residues at 1010 cm-1, corresponding to νS=O band and is consistent with that for free Met oxidation above (cf. Figure 2). Difference spectra of other proteins used in the FTIR study (mAb3, mAb4, FP2) as shown in the inset of Figure 5 also show detection of the νS=O band at 1010 cm-1, indicative of Met oxidation. Additionally, the FTIR spectra of the oxidized form of mAb2 using H218O2 and H216O2 and their difference spectra show isotopic down-shift of 28 cm-1, thus confirm the assignment of 1010 cm-1 band to νS=O band in proteins. In-situ monitoring of Met oxidation in protein by ATIR-FTIR spectroscopy is limited due to short penetration depths of 0.5–2 µm inherent to the ATR sampling technique and limits the reaction kinetics affected the ability to monitor oxidation reaction kinetics of the bulk solution. Spectral measurements over extended periods of time also lead to protein adsorption on the surface of the ATR-window resulting in baseline drift, which hinders proper calculation of difference spectra. Therefore, measurements from separate samples to generate difference


14

Page 15 of 29 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


spectra result in imperfect spectral subtraction (Figure 5). Future work will explore other sampling techniques for the purpose of quantifying oxidation levels by FTIR, possibly including bulk measurements in a cuvette. Discussion Characterization of protein degradation pathways is a critical component of drug development to establish critical quality attributes and define robust control strategies. Met oxidation is one of the most frequent degradation pathway observed in proteins and traditionally characterized by peptide mapping followed by LC-MS. In the present work, we explore the possibility of utilizing vibrational spectroscopy as a tool to monitor and quantify Met oxidation in proteins. Vibrational spectroscopic signals from Met and its oxidation products are inherently weak and are further complicated by overlapping vibrational signals from aromatic side chains, amide bonds, and formulation components. Together, these complexities make it difficult to identify the correct vibrational markers for Met oxidation in proteins. This work definitely identified Raman and FTIR bands associated with Met oxidation and furthermore demonstrated the Raman vC-S band at 702 cm-1 and FTIR vS=O band at 1010 cm-1 as useful markers for detecting and quantifying Met oxidation in proteins. Under oxidative stress, several amino acid residues including Met, Cys, Trp , Tyr and His can oxidize leading to the appearance of new Raman or FTIR bands. Besides oxidation, other degradation pathways, such as aggregation and clipping as well as higher order structural changes, are often triggered simultaneously when proteins are subjected to oxidizing conditions. Therefore, determining the correct assignment of the relevant Met oxidation markers, particularly in monoclonal antibodies and other therapeutic proteins, requires careful


15


Page 16 of 29

optimization of reaction conditions, sampling procedures, spectral measurement techniques, as well as data processing. Identification of oxidation marker in free Met by 18O isotopic shift Isotopic labeling is a powerful tool widely used in vibrational spectroscopy to assign bands to their corresponding vibrational modes based on frequency shifts. In the present work, free Met was oxidized by H216O2 and H218O2 to generate Met sulfoxide containing 16O and 18O isotopes, respectively. When the free Met oxidation reaction with hydrogen peroxide was carried out in 18O water (data not shown), only bands associated with Met sulfoxide with 16O isotope were detected. This experimentally confirms that the oxygen of Met sulfoxide originates from peroxide and supports a one-step Met oxidation mechanism involving the direct nucleophilic attack of Met sulfur on the electrophilic oxygen of the solvated peroxide.2 18

O isotopic data presented in Figures 1 and 2 for free Met oxidation demonstrated that

multiple vibrational bands indeed undergo changes over time, but no bands were observed for νS=O of Met sulfone at 1045 cm-1 as previously reported.28, 30 In fact, Met sulfone was not detected in the oxidized proteins studied here, and only ~ 5% sulfone was observed at the end of free Met oxidation reaction after 10 hr (Figure 1 and Figure S1; νS=O band at 1134 cm-1), suggesting sulfone is not readily formed. Isotopic experiments using 18O hydrogen peroxide revealed frequency shifts that agreed well to those predicted based on the local mode assumption and thus confirmed the identity of each of the vibrational bands. Isotopic shift experiments also revealed that the νS=O band at 1010 cm-1 from both Raman and FTIR spectroscopy may be involved in Fermi resonance with a possible combination of the bands at 540 and 470 cm-1. This Fermi resonance effect increased the intensity of the 1010 cm-1 band especially in the FTIR spectra, providing a unique advantage for monitoring Met oxidation in proteins.


16

Page 17 of 29 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


Quantification of Met oxidation in proteins by Raman spectroscopy With the emergence of Raman and FTIR spectroscopy for protein characterization in recent years, and the continuous improvement in laser and optical technologies, vibrational spectroscopy is expected to be at the forefront of protein biophysical characterization. Many sophisticated vibrational spectroscopic techniques are being developed, such as surface-enhanced Raman spectroscopy, resonance Raman spectroscopy, drop-coat deposition confocal Raman spectroscopy and various coherent nonlinear Raman spectroscopies to enhance spectroscopic signals from proteins.37-39 In the present study we have demonstrated that it is possible to detect and quantify in-situ Met oxidation in proteins using a commercially available instrument based on simple non-resonant Raman spectroscopy. Raman spectra of proteins in the 680-720 cm-1 region that includes the νC-S band of Met sulfoxide at 702 cm-1 is an ideal spectral window to observe Met oxidation due to the absence of other protein or excipient signals. While the νS=O band at 1010 cm-1 is a sensitive marker for Met oxidation, this band overlaps with the strong Raman bands from phenylalanine and tryptophan making it unsuitable for quantification of Met oxidation in proteins. In contrast, vibrational modes of Phe and other aromatic side chains are not detected in FTIR spectroscopy and the νS=O band at 1010 cm-1 is the optimal FTIR marker for Met oxidation. mAbs contain solvent exposed Met residues that are readily oxidized upon exposure to reactive oxygen species (ROS). Currently available techniques to detect Met oxidation are limited to tryptic peptide mapping with LC-MS and, in certain cases, chromatographic separation methods. Vibrational spectroscopy is a relatively fast (~30 min), material-sparing (~50 µL), and non-destructive technique that can monitor protein oxidation in-situ, which is particularly useful for determining the kinetics in forced oxidation studies or in other applications. Peptide mapping


17


Page 18 of 29

was also performed in this study as an orthogonal technique for quantifying Met oxidation. As shown in Table S1 as well as in Fig 4c the average Met oxidation levels quantified by Raman marker at 702 cm-1 is consistent with those by peptide mapping for all the proteins tested. In this context, Raman spectroscopy provides the additional benefit of monitoring changes in higher order structure in the proteins while Met residues undergo oxidation. The structural consequence of chemical modifications is integral to understanding the inherent liabilities of therapeutic proteins so that we can develop appropriate control strategies during development. Vibrational spectroscopy and peptide mapping are complementary analytical tools for characterizing Met oxidation. The former is particularly useful for monitoring oxidation kinetics, while the latter is ideal for identifying the individual residues most susceptible to oxidation. Together, these characterization techniques establish whether Met oxidation is a critical quality attribute and can be used develop robust analytical control strategies. Conclusions Methionine oxidation is a common degradation pathway for therapeutic proteins and may impact drug quality, safety, potency and pharmacokinetics (e.g. by altering FcRn binding). FTIR and Raman spectroscopy were employed in this study to definitively identify vibrational band associated with Met oxidation and then successfully applied to monitor in-situ the oxidation of free Met and Met residues in therapeutic proteins. Vibrational markers of Met oxidation products, such as, sulfoxide and sulfone, corresponding to S=O and C-S=O stretching frequencies were unequivocally identified based 18O isotoptic shifts. Isotopic substitution also led to decreased intensity of the νS=O band at 1010 cm-1, which enhances the intensity and makes this band a useful Met oxidation marker in FTIR measurements. A strong Raman νC-S band at 702 cm-1 was found to be an ideal Met oxidation marker due to the absence of any


18

Page 19 of 29 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


overlapping protein or excipient bands in its immediate proximity in the spectrum. This is the first report to quantify Met oxidation kinetics of multiple therapeutic proteins by in-situ Raman spectroscopy, thereby monitoring the unique oxidation propensity of each protein. Moreover, oxidation levels quantified by Raman spectroscopy are consistent with those by peptide-mapping LC-MS. We have demonstrated the broad applications of Raman and FTIR spectroscopy for characterizing protein and particularly for the detection and quantification of Met oxidation in therapeutic proteins. Supporting Information Available: Experimental description of the tryptic peptide mapping/LC-MS, comparison of Met oxidation by Raman and TPM/LC-MS, Raman and FTIR spectra of free methionine and its oxidation products, description of quantification of methionine oxidation levels in proteins by Raman spectroscopy and spectral deconvolution analysis of anomalous intensity in the 900-1050 cm-1 region. Acknowledgements and Disclosures We thank Drs. Henrik Andersen, Reb Russell, Anthony Leone, Li Tao, Jacob Bongers, Hui Wei, Yunping Huang, and Hangtian Song of Bristol-Myers Squibb Molecular and Analytical Development group for helpful discussions and critical feedback on the manuscript. The authors declare no personal, financial, or non-financial conflict of interest.


19


Page 20 of 29

Figure Captions Figure 1. Oxidation kinetics of free Met by hydrogen peroxide in water monitored by in-situ Raman spectroscopy. (a) Time dependent Raman spectra of free Met reacted with hydrogen peroxide. Vibrational band labeling is based on the assignments available from the literature29, 32. Inset: Relative concentration of reactants and products obtained from the linear least square fit to the experimental data. (see text for detail). Residuals to the least square fit are shown at the bottom. (b) Raman spectra of Met sulfoxide containing 16O (red) and 18O (blue) isotopes and their difference (green) spectra. (c) Raman spectra of Met sulfone containing 16O (red) and 18O (blue) isotopes and their difference (green) spectra. Figure 2. Oxidation kinetics of free Met by hydrogen peroxide in water monitored by in-situ ATR-FTIR spectroscopy. (a) Time dependent FTIR spectra of free Met reacted with hydrogen peroxide reaction. Vibrational band labeling is based on the assignments available from the literature29. Inset: Relative concentration of Met and Met sulfoxide obtained from the linear least square fit to the experimental data. (see text for detail). Residuals to the least square fit are shown at the bottom. (b) FTIR spectra of Met sulfoxide containing 16O (red) and 18O (blue) isotopes and their difference (green) spectra. Figure 3. Oxidation of Met residues in therapeutic proteins monitored by in-situ Raman Spectroscopy (a) Overlay of the time-dependent Raman spectra of mAb1 (50 mg/mL, ~0.33 mM) oxidized by 30 mM hydrogen peroxide at 25°C over a period of 24 hours. Vibrational band labeling is based on the assignments available from the literature29. Inset: Magnification of the C-S stretching band at 702 cm-1 shows growing intensity with time due to conversion of Met residues to Met sulfoxide in proteins. (b) Difference Raman spectra of mAb1 (c) Difference Raman spectra for multiple therapeutic proteins after 24 hrs of peroxide oxidation. Figure 4. Average oxidation levels of Met residues in therapeutic proteins monitored using νC-S Raman band intensity at 702 cm-1. (a) IgG4 (mAb1 and mAb2) and IgG1 (mAb3-mAb5). (b) Fusion proteins (FP1 and FP2) and antibody drug conjugate (ADC+) and its parent mAb without drug conjugation (ADC-). (c) Correlation between average oxidation levels of Met residues in therapeutic proteins obtained from Raman spectroscopy and tryptic-peptide mapping LC-MS detection. Figure 5. Oxidation of Met residues in therapeutic proteins monitored by ATR-FTIR spectroscopy. FTIR spectra of mAb2 before (black) and after oxidation (red) by H216O2 and by


20

Page 21 of 29 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


H218O2 (blue) and their difference spectra and the isotopic difference spectra (gray). Inset: FTIR difference spectra of mAb3 (black), mAb4 (green) and FP2 (magenta) after oxidation by H216O2 relative to their respective unoxidized states. Vibrational band labeling are based on the assignments available from published literature.27


21


Page 22 of 29

References (1)

Ecker, D. M.; Jones, S. D.; Levine, H. L., mAbs 2015, 7, 9-14.

(2)

Shihong Li, C. S., and Ronald T. Borchardt, Biotech. Bioeng. 1995, 48, 490-500.

(3)

Liu, D.; Ren, D.; Huang, H.; Dankberg, J.; Rosenfeld, R.; Cocco, M. J.; Li, L.; Brems, D.

N.; Remmele, R. L., Jr., Biochemistry 2008, 47, 5088-5100. (4)

Ji, J. A.; Zhang, B.; Cheng, W.; Wang, Y. J., J. Pharm. Sci. 2009, 98, 4485-4500.

(5)

Pan, H.; Chen, K.; Chu, L.; Kinderman, F.; Apostol, I.; Huang, G., Protein Sci. 2009, 18,

424-433. (6)

Luo, Q.; Joubert, M. K.; Stevenson, R.; Ketchem, R. R.; Narhi, L. O.; Wypych, J., J. Biol.

Chem. 2011, 286, 25134-25144. (7)

Boll, B.; Bessa, J.; Folzer, E.; Rios Quiroz, A.; Schmidt, R.; Bulau, P.; Finkler, C.;

Mahler, H. C.; Huwyler, J.; Iglesias, A.; Koulov, A. V., Mol. Pharm. 2017, 14, 1292-1299. (8)

Sreedhara, A.; Yin, J.; Joyce, M.; Lau, K.; Wecksler, A. T.; Deperalta, G.; Yi, L.; John

Wang, Y.; Kabakoff, B.; Kishore, R. S., Eur. J. Pharm. Biopharm. 2016, 100, 38-46. (9)

Sreedhara, A.; Lau, K.; Li, C.; Hosken, B.; Macchi, F.; Zhan, D.; Shen, A.; Steinmann,

D.; Schoneich, C.; Lentz, Y., Mol. Pharm. 2013, 10, 278-288. (10)

Gao, X.; Ji, J. A.; Veeravalli, K.; Wang, Y. J.; Zhang, T.; McGreevy, W.; Zheng, K.;

Kelley, R. F.; Laird, M. W.; Liu, J.; Cromwell, M., J. Pharm. Sci. 2015, 104, 368-377. (11)

Wang, W.; Vlasak, J.; Li, Y.; Pristatsky, P.; Fang, Y.; Pittman, T.; Roman, J.; Wang, Y.;

Prueksaritanont, T.; Ionescu, R., Mol. Immunol. 2011, 48, 860-866. (12)

Stracke, J.; Emrich, T.; Rueger, P.; Schlothauer, T.; Kling, L.; Knaupp, A.; Hertenberger,

H.; Wolfert, A.; Spick, C.; Lau, W.; Drabner, G.; Reiff, U.; Koll, H.; Papadimitriou, A., mAbs 2014, 6, 1229-1242.


22

Page 23 of 29 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)

Tada, M.; Ishii-Watabe, A.; Suzuki, T.; Kawasaki, N., PLoS One 2014, 9, e95787.

(14)

Bertolotti-Ciarlet, A.; Wang, W.; Lownes, R.; Pristatsky, P.; Fang, Y.; McKelvey, T.; Li,

Y.; Li, Y.; Drummond, J.; Prueksaritanont, T.; Vlasak, J., Mol. Immunol. 2009, 46, 1878-1882. (15)

Yan, Y.; Wei, H.; Fu, Y.; Jusuf, S.; Zeng, M.; Ludwig, R.; Krystek, S. R., Jr.; Chen, G.;

Tao, L.; Das, T. K., Anal. Chem. 2016, 88, 2041-2050. (16)

Mo, J.; Yan, Q.; So, C. K.; Soden, T.; Lewis, M. J.; Hu, P., Anal. Chem. 2016, 88, 9495-

9502. (17)

Song, H.; Xu, J.; Jin, M.; Huang, C.; Bongers, J.; Bai, H.; Wu, W.; Ludwig, R.; Li, Z.;

Tao, L.; Das, T. K., Pharm. Res. 2016, 33, 932-941. (18)

Li, Y.; Polozova, A.; Gruia, F.; Feng, J., Anal. Chem. 2014, 86, 6850-6857.

(19)

Liu, H.; Ponniah, G.; Neill, A.; Patel, R.; Andrien, B., Anal. Chem. 2013, 85, 11705-

11709. (20)

Chen, M.; Cook, K. D., Anal. Chem. 2007, 79, 2031-2036.

(21)

Cohen, S. L., Anal. Chem. 2006, 78, 4352-4362.

(22)

Pavon, J. A.; Li, X.; Chico, S.; Kishnani, U.; Soundararajan, S.; Cheung, J.; Li, H.;

Richardson, D.; Shameem, M.; Yang, X., J. Chromatogr. A 2016, 1431, 154-165. (23)

Wen, Z. Q., J. Pharm. Sci. 2007, 96, 2861-2878.

(24)

Huang, C. Y.; Balakrishnan, G.; Spiro, T. G., J. Raman Spectros. 2006, 37, 277-282.

(25)

Balakrishnan, G.; Weeks, C. L.; Ibrahim, M.; Soldatova, A. V.; Spiro, T. G., Curr. Opin.

Struct. Biol. 2008, 18, 623-629. (26)

Balakrishnan, G.; Spiro, T. G., Ultraviolet Resonance Raman (UVRR) Spectroscopy

Studies of Structure and Dynamics of Proteins. In Encyclopedia of Biophysics, Roberts, G. C. K., Ed. Springer Berlin Heidelberg: 2013; pp 2697-2707.


23


(27)

Barth, A., Prog. Biophys. Mol. Biol. 2000, 200, 141-173.

(28)

Ravi, J.; Hills, A. E.; Cerasoli, E.; Rakowska, P. D.; Ryadnov, M. G., Eur. Biophys. J.

Page 24 of 29

2011, 40, 339-345. (29)

Torreggiani, A.; Barata-Vallejo, S.; Chatgilialoglu, C., Anal. Bioanal. Chem. 2011, 401,

1231-1239. (30)

Zhou, C.; Qi, W.; Lewis, E. N.; Carpenter, J. F., Anal. Biochem. 2015, 472, 7-20.

(31)

Jain, N.; Smith, S. W.; Ghone, S.; Tomczuk, B., Pharm. Res. 2015, 32, 3526-3540.

(32)

Jurasekova, Z.; Tinti, A.; Torreggiani, A., Anal. Bioanal. Chem. 2011, 400, 2921-2931.

(33)

Dong, J.; Atwood, C. S.; Anderson, V. E.; Siedlak, S. L.; Smith, M. A.; Perry, G.; Carey,

P. R., Biochemistry 2003, 42, 2768-2773. (34)

Wilson, E. B.; Decius, J. C.; Cross, P. C., Molecular vibrations : the theory of infrared

and Raman vibrational spectra. McGraw-Hill: New York, 1955. (35)

Lipkin, J. S.; Song, R.; Fenlon, E. E.; Brewer, S. H., J. Phys. Chem. Lett. 2011, 2011,

1672-1676. (36)

Windisch, C. F., Jr.; Glezakou, V. A.; Martin, P. F.; McGrail, B. P.; Schaef, H. T., Phys.

Chem. Chem. Phys. 2012, 14, 2560-2566. (37)

Laing, S.; Jamieson, L. E.; Faulds, K.; Graham, D., Nat. Rev. Chem. 2017, 1, 0060.

(38)

Yang, S.; Dai, X.; Stogin, B. B.; Wong, T. S., Proc. Natl. Acad. Sci. U S A 2016, 113,

268-273. (39)

Butler, H. J.; Ashton, L.; Bird, B.; Cinque, G.; Curtis, K.; Dorney, J.; Esmonde-White,

K.; Fullwood, N. J.; Gardner, B.; Martin-Hirsch, P. L.; Walsh, M. J.; McAinsh, M. R.; Stone, N.; Martin, F. L., Nat. Protoc. 2016, 11, 664-687.


24

Page 25 of 29 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.


25


Page 26 of 29

Figure 2.


26

Page 27 of 29 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.


27


Page 28 of 29

Figure 4.

Figure 5.


28

Page 29 of 29 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


For Table of Contents Only

Monoclonal Antibody

Raman Spectroscopy

[O]

Methionine Oxidation


29

Detection and Identification of the Vibrational Markers for the

Recommend Documents