Article pubs.acs.org/ac
Characterization of the Degradation Products of a Color-Changed Monoclonal Antibody: Tryptophan-Derived Chromophores Yiming Li, Alla Polozova,† Flaviu Gruia, and Jinhua Feng* MedImmune, Analytical Biotechnology, One MedImmune Way, Gaithersburg, Maryland 20878, United States S Supporting Information *
ABSTRACT: We describe the characterization of degradation products responsible for color change in near UV−visible lightirradiated and heat-stressed monoclonal antibody (mAb) drug product in liquid formulation. The treated samples were characterized using reversed-phase HPLC and size-exclusion HPLC with absorption spectroscopy. Both methods showed color change was due to chromophores formed on the mAb but not associated with the formulation excipients in both lightirradiated and heat-stressed mAb samples. These chromophores were further located by a new peptide mapping methodology with a combination of mass spectrometry and absorption spectroscopy. Mass spectrometry identified the major tryptophan oxidation products as kynurenine (Kyn), N-formylkynurenine (NFK), and hydroxytryptophan (OH-Trp). The absorption spectra showed that each of the tryptophan oxidation products exhibited a distinct absorption band above 280 nm shifted to the longer wavelengths in the order of OH-Trp < NFK < Kyn. The Kyn-containing peptide was detected by absorption at 420 nm. No new absorption bands were observed for either methionine or histidine oxidation products. This confirmed that tryptophan oxidation products, but not methionine and histidine oxidation products, were responsible for the color change. It is worth noting that a new oxidation product with the loss of hydrogen (2 Da mass decrease) for Trp-107 of the heavy chain was identified in the heat-stressed mAb sample. This oxidized tryptophan residue exhibited a distinct absorption band at the maximum absorbance wavelength 335 nm, which is responsible for the color change to yellow. This study showed that the new peptide mapping methodology with a combination of mass spectrometry and absorption spectroscopy is useful to identify tryptophan oxidation products as chromophores responsible for color change in stressed mAb drug product.
S
observed in other stability studies. The color change can be caused by the chemical modifications of mAbs, degradation products of formulation excipients, or degradation products formed through the reaction of mAbs and formulation excipients. While some examples of the impact of photoirradiation on color change of mAb solution have been previously reported,6 the structural information and relationships of the PTMs with the color change is relatively limited. However, such information is valuable for understanding the cause of the color change and therefore the potential impact of the color change on safety and efficacy of mAb therapeutics. It will also help us to design short-term and long-term storage conditions to prevent modifications causing color change and biological activity loss before product goes to patient. Color change is a common degradation process that can occur naturally, such as age related yellowing of the human eye lens.7−9 The degradation products in lens proteins have been extensively studied to understand cataract formation and to
ince the mid-1990s, monoclonal antibodies (mAbs) have become an important new drug class in human therapeutics. At present, approximately 30 therapeutic mAbs are marketed in the United States and Europe for a variety of clinical indications, including oncology, chronic inflammatory diseases, transplantation, infectious disease and cardiovascular medicine.1,2 The majority of mAbs approved and in clinical trials bear some form of post-translational modifications (PTMs), which can affect the consistency of product purity and quality and may alter biologically functional properties relevant to their therapeutic applications.3−5 With this rapid growth, PTMs have been extensively characterized with the aid of increasingly sophisticated analytical methods to gain a better understanding of the effect of PTMs on biological activity and to ensure a consistent product purity and quality. Color is a basic physical appearance parameter of therapeutic monoclonal antibody drug products which is visible to human eye. It can be a sensitive measure as relatively small changes lead to visible difference in color. It is one of the important parameters that is usually determined for the drug product release to ensure the product purity and quality, and also monitored in stability and comparability studies. Color change is often reported in photostability studies, and it is sometimes © 2014 American Chemical Society
Received: December 27, 2013 Accepted: June 17, 2014 Published: June 17, 2014 6850
dx.doi.org/10.1021/ac404218t | Anal. Chem. 2014, 86, 6850−6857
Analytical Chemistry
Article
develop strategies to prevent it.7,8,10,11 In the early 1970s, it was found that tryptophan oxidation products are responsible for the color change. However, most of those studies, performed on experimentally treated lens proteins, did not provide conclusive evidence regarding the naturally occurring color change. Mass spectrometry has become instrumental since the early 1990s in identification of the degradation products of lens proteins.10 However, the modifications (or PTMs) in lens proteins have not been correlated with changes in their spectroscopic properties or change in color at either protein or digested peptide level. The cause of the human lens yellowing is still elusive due to the lack of direct identification of molecular changes in lens proteins responsible for the color change. Yellowing of wool and silk in the sun is also well-known.12,13 It was shown that tryptophan- and tyrosine-derived oxidation products were present in photoyellowed wool proteins.14 However, the specific connection between these oxidation products and color change has not been fully elucidated. In this study, we used absorption spectroscopy to show the color change was associated with intact mAb, reduced subunits, and then specific digested peptides. These specific peptides were characterized by both absorption spectroscopy and mass spectrometry in a single peptide mapping LC-MS to directly connect tryptophan oxidation products to color changes of the mAb solutions.
eluted proteins were monitored by UV detection at 280 and 330 nm with a TUV detector. UV absorption spectra were acquired from 220 to 500 nm for the eluted proteins using the Agilent 1200 diode-array detector. Size Exclusion HPLC. The mAb samples were injected onto a size exclusion column (G3000 SWXL, 5 μm, 300 Å, 7.8 × 300 mm, TosoHaas) equipped on an Agilent HPLC system (HP1200, Santa Clara, CA) with a diode-array detector. An isocratic elution using mobile phase of 100 mM sodium phosphate, 100 mM sodium sulfate, 0.05% sodium azide, pH 6.8 at 1 mL/min was used for the chromatographic separation. The eluted protein was monitored by UV absorbance at 280 and 340 nm. Peptide Mapping. The tryptic digestion was performed by mixing mAb (100 μg in 20 μL) with 40 μL of 8 M guanidineHCl, 130 mM Tris, and 1 mM EDTA, pH 7.6, 2 μL of 500 mM DTT, and incubated at 37 °C for 30 min. A 5 μL volume of 500 mM iodoacetamide was added and the mixture was incubated at an ambient temperature for 30 min in the dark. The reduced and alkylated protein was transferred into 100 μL of 2 M urea and 100 mM Tris (pH 7.6) using a 10 000 MWCO Amicon filter device. The buffer-exchanged protein (100 μL) was digested with 5 μL of 1 mg/mL trypsin in 1 mM HCl, and incubated at 37 °C for 2 h. A second aliquot of trypsin (2 μL) was added to the mixture and the incubation was continued at 37 °C for an additional 1 h. The digestion was quenched by the addition of 2 μL of TFA. The online LC-MS/MS analysis of the digested peptides was performed using an ultrahigh pressure UPLC system (Waters Acquity UPLC) and a ThermoElectron LTQ Velos ion-trap or LTQ Orbitrap mass spectrometer (Thermo Electron, San Jose, CA). The HPLC system was equipped with a reversed phase column (BEH300, C18 1.7 μm 2.1 × 150 mm) connected to an Agilent 1200 DAD detector with micro flow cell, followed by mass spectrometry. The mobile phase A was 0.02% TFA in HPLC water and the mobile phase B was 0.02% TFA in acetonitrile. The peptides were eluted using a linear gradient of 0−5% mobile phase B over 3 min, 5−20% over 30 min, and 20−35% over 38.5 min at a flow rate of 200 μL/min. The eluted peptides were monitored by UV detection at the multiple wavelengths of 220, 290, 310, 330, 355, 390, and 420 nm, and by a LTQ Velos ion-trap mass spectrometer in positive ion mode. UV−visible spectra were collected on the Agilent DAD detector from 190 to 500 nm. For the analysis of tryptic peptides, the mass spectrometer was operated in datadependent “triple play” mode with dynamic exclusion enabled. The MS data were searched against the known protein sequences for confirmation and for post-translational modifications. The peak areas were calculated from the extracted ion chromatograms of full scan MS data. The relative percent of the unmodified peptide was determined by the peak area of the unmodified peptide versus the total peak area of the modified and unmodified species for each peptide.
■
MATERIALS AND METHODS Materials. The monoclonal antibody is humanized IgG protein and produced in-house by MedImmune. Water, acetonitrile, trifluoroacetic acid (TFA), water containing 0.1% TFA, acetonitrile containing 0.1% TFA (LC-MS grade), dithiothreitol (DTT), iodoacetamide, guanidine HCl, urea, tris (hydroxymethyl) aminomethane, sodium phosphate, and sodium sulfate were purchased from VWR. Trypsin was purchased from Promega. Preparation of Near UV−visible Light-Irradiated and Heat-Stressed mAb. mAb at a concentration of 145 mg/mL in 50 mM sodium acetate and 85 mM sodium chloride, 0.01% (w/v) polysorbate 80, pH 5.5 was exposed to a near UV− visible light source (300−700 nm) with the light intensity (illuminance) of 9500 lx in a photostability chamber (Model PST54SD, Powers Scientific, Pipersville, PA) at 25 °C for up to 14 days. The mAb samples were placed at marked positions, where the light intensity was measured. Two aliquots (2 × 900 μL) were pulled out at various time points (initial, 1−14 days) for testing. On the basis of the time spent in the photostability chamber, the light exposure for each time point sample was calculated. The same mAb at a concentration of 50 mg/mL in the same formulation buffer as the near UV−visible light-irradiated mAb described above was incubated at 25 °C for 12 months and stored at 2−8 °C until analysis. Reversed Phase HPLC. The mAb was reduced by 20 mM DTT in 40 mM tris buffer, pH 8.0. Both reduced and nonreduced mAb samples were analyzed by reversed phase HPLC (RP-HPLC) using an ultrahigh pressure UPLC system (Waters Acquity UPLC) equipped with a polymer column (PLRP-S, 8 μm, 4000 Å, 1.0 × 50 mm column, Michrom Bioresources) and TUV detector and Agilent 1200 diode-array detector. The column temperature was maintained at 70 ± 1 °C. The protein sample was separated by a linear gradient with a flow rate of 200 μL/min. The mobile phase A was 0.1% TFA in water and mobile phase B was 0.1% TFA in acetonitrile. The
■
RESULTS AND DISCUSSION Color Change of Near UV−visible Light-Irradiated and Heat-Stressed mAb. To evaluate photodegradation of the mAb under near UV−visible light-irradiation similar to conditions encountered under daylight or room light exposure, mAb solution samples were exposed to intense near UV− visible light (300−700 nm) at a light intensity (illuminance) of 9500 lx in a photostability chamber at 25 °C for predetermined time intervals (see Material and Methods for details). After the 6851
dx.doi.org/10.1021/ac404218t | Anal. Chem. 2014, 86, 6850−6857
Analytical Chemistry
Article
the 330 nm chromatogram while it decreased in the 280 nm chromatogram. No other species was detected by absorbance at 330 nm wavelength in the entire chromatogram. The mAb peak in the RP-HPLC chromatograms became broader with increasing light exposure time (Figure 2), possibly because a complex oxidized mixture was formed through the near UV− visible light exposure. Oxidized species are in general more hydrophilic and elute earlier in RP-HPLC than the nonoxidized species.15,16 To further locate yellow chromophores, the light exposed mAb was reduced and analyzed by RP-HPLC. Figure 2 shows that both light chain and heavy chain peaks were associated with the yellow chromophores. The UV−visible spectra collected for the mAb peak in RP-HPLC showed a gradual growth of shoulder at >300 nm with light exposure time (Supporting Information Figure S-1). This absorbance increase is similar to the UV−visible spectra collected for the original near UV−visible light-irradiated mAb samples (Supporting Information Figure S-2). This shows that the yellow discoloration of the light exposed samples was due to chromophore formation on the mAb by photodegradation. The peak area ratio of 330 nm versus 280 nm of the heavy chain is greater than that of the light chain (data not included), indicating that the yellow chromophores were more prevalent on the heavy chain than the light chain, as discussed in a later section. The light exposed mAb was also analyzed by size exclusion HPLC (SEC) with diode-array detector to assess the levels of aggregates and fragments and their associated yellow chromophore. The SEC results showed that progressively increasing aggregates formed under the near UV−visible light exposure, while fragmentation was not significant. Both aggregates and monomers were associated with yellow
initial 1−2 days of exposure to the intense near UV−visible light, the mAb solution was yellow by visual inspection. The extent of the discoloration increased with light exposure time. The solution was intense yellow after 4 days of exposure and brown after 14 days (Figure 1). The heat-stressed mAb was slightly yellow after incubation at 25 °C for 12 months (picture not taken).
Figure 1. Color change of the mAb samples exposed under near UV− visible light for 0 days (left), 4 days (middle), and 14 days (right).
Yellow Chromophore Formed on Near UV−visible Light-Irradiated mAb. Analysis of the photostressed mAb samples by reversed phase HPLC (RP-HPLC) showed that yellow chromophore was associated with the mAb itself, but not with formulation excipients. Figure 2 shows the chromatographic elution profiles monitored at 280 and 330 nm for the photostressed samples at different time points. When the light exposure time increased, the mAb peak intensity increased in
Figure 2. RP-HPLC chromatographic profiles of the mAb samples after near UV−visible light-irradiated for 0 days (red), 4 days (blue), 7 days (green), and 14 days (black). (A) 280 nm and (B) 330 nm profiles of nonreduced mAb. (C) 280 nm and (D) 330 nm profiles of reduced mAb. LC: light chain. HC: heavy chain. 6852
dx.doi.org/10.1021/ac404218t | Anal. Chem. 2014, 86, 6850−6857
Analytical Chemistry
Article
characteristic absorption and fluorescence spectra were reported in the past.18−20 The chemical structures of these Trp oxidation products were described by Simat et al.18 and Reubsaet et al.,21 and are shown in Figure 4 to help us to
chromophores. The peak intensity for the aggregates including dimers, in both 280 and 340 nm chromatograms, increased with light exposure time (Figure 3). These results show that the
Figure 4. Tryptophan oxidation pathways. Oia is oxindolylalanine, DiOia is dioxindolylalanine, Kyn is kynurenine, and NFK is Nformylkynurenine. C* is chiral center (two species/anomers).
understand tryptophan oxidation pathways. OH-Trp is formed by the addition of one oxygen atom and NFK by the addition of two oxygen atoms followed by the cleavage of indole ring. Kyn can result from the deformylation of NFK with the loss of the carbonyl group. The carbonyl group can readily leave from the solution as gas, which drives the reaction toward the formation of Kyn. Kyn is further oxidized to give rise to 3-OHKyn. The molecular weight change of OH-Trp, NFK, Kyn, and 3-OH-Kyn relative to tryptophan is +16, +32, +4, +20 Da, respectively, which allows the identification of these oxidation products by mass spectrometry. Even though the tryptophan oxidation products were often identified by the characteristic electronic absorption spectra in early days,19,20 the distinct absorption properties of the oxidized tryptophan species have not been often utilized in recent peptide mapping analysis since mass spectrometry is by itself a powerful tool capable of detection and identification of these oxidation products.4−6,14 However, the spectroscopic differences between the unmodified tryptophan and the tryptophan oxidation products may be a useful tool for quick identification of the various tryptophan oxidation products in situations where multiple oxidation products with low abundance are present in complicated peptide mapping. In this study, the eluted peptides were monitored at multiple wavelengths to detect the tryptophan oxidation products. The oxidation products of three tryptophan residues, Trp-50 and Trp-104 on the heavy chain and Trp-90 on the light chain, were detected by 355 nm wavelength in the near UV−visible lightexposed mAb (Figure 5A). Only peaks corresponding to peptides with oxidized Trp products were present in chromatogram acquired at 355 nm. The oxidation products of both Trp50 on the heavy chain and Trp-90 on the light chain were initially identified as low abundance peaks in the chromatogram collected at 355 nm, and then confirmed by the mass spectrometry data. The tryptophan oxidation products detected at 355 nm for these three tryptophan residues were confirmed as Kyn with the mass increase of +4 Da by the mass spectrometric data. In addition to Kyn, other tryptophan oxidation products with the mass increase of +16 Da (OH-Trp) and +32 Da (NFK) were identified for these three tryptophan residues by the mass spectrometry (Supporting Information Figures S-2−S-5). It is worth noting that each tryptophan oxidation product has a distinct absorption property. The UV absorption spectra of the peptides containing Trp-104 and its
Figure 3. SEC chromatographic profiles of the mAb samples exposed under near UV−visible light for 0 days (red), 4 days (blue), 7 days (green), and 14 days (black): (A) 280 and (B) 340 nm.
aggregates are formed under the near UV−visible light exposure and they contain the yellow chromophore. The absorbance of the monomer peak at 340 nm increased with light exposure time, while its absorbance at 280 nm decreased (Figure 3). It showed that the yellow chromophore formed on the mAb monomer increased with increasing light exposure time. The peak area ratio of 340 nm versus 280 nm for aggregates was greater compared to monomers (data not included), pointing to the yellow chromophore being more enriched in the aggregates. In addition, the monomer peak shifted to an earlier retention time with the longer light exposure, while no change in molecular weight was detected by MALS (data not included). The peak shifting with no apparent molecular weight change was observed for the dimer peak as well. This peak shifting can be fully explained with a partial unfolding caused by the photo-oxidation.17 Site Identification of Chromophores by Tryptic Peptide Mapping. To characterize the yellow chromophore formed on the mAb, the photoirradiated mAb was analyzed by tryptic peptide mapping. The light exposed mAb samples were reduced with DTT, alkylated with iodoacetamide, digested by trypsin, and followed by LC-MS/MS analysis with online mass spectrometry to identify the modification sites, and with diodearray detector to monitor the spectroscopic properties of the modified peptides. In addition, the online tandem mass spectrometry was used to identify modification sites based on the fragment ions in the MS/MS spectra. Various tryptophan oxidation products, including Nformylkynurenine (NFK), kynurenine (Kyn), hydroxytryptophan (OH-Trp), and 3-hydroxykynurenine (3-OH-Kyn) with 6853
dx.doi.org/10.1021/ac404218t | Anal. Chem. 2014, 86, 6850−6857
Analytical Chemistry
Article
Apparently, the presence of Kyn species results in a more intense and darker yellow color than NFK species because of its absorption above 400 nm in visible light wavelength range (Figure 6A). It is likely that the OH-Trp contributes to the color change to a lesser extent as its absorption band exhibits a smaller red shift compared to Kyn and NFK. Tryptophan oxidation leads to the formation of multiple oxidation products. Besides the abundant Trp-104 oxidation products discussed above, several other Trp-104 oxidation products with a lower abundance were detected by mass spectrometry, but not by spectroscopic absorbance due to its lower sensitivity. Seven distinct oxidation products were identified for Trp-104, including Kyn, NFK, and OH-Trp discussed above. Three oxidized peptides with a 16 Da mass increase were identified as containing either OH-Trp or oxindolylalanine (Oia).18,21 It is likely that the oxidized peptide detected by mass spectrometry (data not included), but not by absorbance at >300 nm, is Oia because Oia does not have an indole ring and electron-donating groups conjugated to its benzene ring and it is not expected to exhibit absorbance in this range. The other two oxidized peptides with a 16 Da mass increase were detected at 310 nm wavelength and are likely to be OH-Trp (Figure 6B). The oxidized peptide with a 32 Da mass increase eluting at retention time of 12.0 min was identified as NFK based on its absorption properties as shown in Figure 6A, while the other two peptides eluting at different retention times (9.15 and 9.55 min) were identified as dioxindolylalanine (DiOia) isomers containing a chiral center (data not included). However, 3-OH-Kyn (+20 Da) was not detected in this study. These data demonstrate the high complexity of tryptophan oxidation products, showing that identification and quantitation of tryptophan oxidation is a very challenging task. The less abundant oxidation products were also detected for the other two tryptophan residues: Trp-50 on the heavy chain and Trp-90 on the light chain. In this study, all the detected oxidation products were taken into the account to calculate the level of oxidation as described below. In addition, our study showed that the peptide containing NFK-104 was not stable at room temperature (data not included), which contributed to the difficulty in the accurate quantitation of tryptophan oxidation products. It indicates that tryptophan oxidation levels potentially could be underestimated due to the loss of the oxidized species during sample preparation. Even though peptide mapping with mass spectrometry is a powerful method for the characterization of the tryptophan oxidation, it is tedious, labor intensive, and complicated because of the presence of multiple tryptophan oxidation products. Because of this, RP-HPLC of intact protein coupled with additional absorbance detection at >300 nm could prove advantageous to peptide mapping for detection and quantitation of the colorchange related tryptophan oxidation on proteins. In addition to tryptophan oxidation, other major observed oxidation products of photoirradiation were oxidized Met residues.3,6 The oxidized Met residues include Met-430, Met254, and Met-360 on the heavy chain (Supporting Information Figures S-6−S-9). However, no new absorption bands were detected for the peptides containing oxidized Met compared to the parent peptides containing the nonoxidized Met (data not included). It was also found that His-312 on the heavy chain was oxidized with a mass increment of 32 Da (Supporting Information Figures S-10 and S-11), while 2-oxo-histidine (+16 Da), as often seen histidine oxidation product,23,24 was below the detection limit of MS/MS. But, it was important that no
Figure 5. Overlay of peptide mapping profiles at 355 nm for (A) nontreated mAb (red) and 14-day near UV−visible light-irradiated mAb (black), and (B) nontreated mAb (red) and heat-stressed mAb (black), and (C) UV absorption spectrum of the oxidized peptide eluted at 72.5 min for the heat-stressed mAb.
oxidation products, Kyn (+4 Da), NFK (+32 Da), and OH-Trp (+16 Da) (10.8 min), were different from each other (Figure 6A). The baseline separation of these peptides made it possible
Figure 6. (A) UV absorption spectra of the peptides containing nonoxidized Trp-104 (black) and oxidized Trp-104: Kyn (+4 Da) (green), NFK (+32 Da) (red), and OH-Trp (+16 Da) (blue) from the 14-day light-irradiated mAb and (B) peptide elution profile covering these peptides at 310 nm wavelength.
to obtain the UV−visible spectrum for each individual peptide (Figure 6B). Structures corresponding to these oxidation products are shown in Figure 4. The UV−visible spectrum of the peptide containing the oxidation product OH-Trp of Trp104 (+16 Da) eluting at 12.5 min could not be obtained because this peptide coeluted with another abundant peptide. However, it was identified as another OH-Trp (+16 Da) by mass spectrometry. The UV absorption spectra of the Kyncontaining and NFK-containing peptides shown in Figure 6A were similar to the published UV−visible spectra of the Kyn and NFK, respectively.22 It is known that NFK is slightly yellow while Kyn is yellow.13 Therefore, this collective evidence indicates that the Trp oxidation products Kyn and NFK are the chromophores formed in the near UV−visible light-irradiated mAb, and they contribute to the color change due to their intrinsic strong absorbance at longer than 280 nm wavelength. 6854
dx.doi.org/10.1021/ac404218t | Anal. Chem. 2014, 86, 6850−6857
Analytical Chemistry
Article
The effect of oxidation at these Trp, Met, and His residues on the biological activity was investigated by combining with other techniques, such as cell-based biological activity and high order structure analysis. We found that the biological activity was affected by the oxidation at these Trp residues, which will be discussed in a subsequent publication. As discussed above, the oxidation products of tryptophan were the chromophores formed on the photoirradiated mAb. In order to better understand the color change of the photoirradiated mAb samples changed from colorless to yellow and then to brown, we examined the distribution of the most abundant oxidization products of these three oxidized tryptophan residues. The relative level of oxidation varied for each oxidation product. In addition, the distribution of these three major oxidation products, OH-Trp, NFK, and Kyn, changed during the 14-day study. It is interesting that Kyn increased to the highest extent with the prolonged exposure up to 14 days, and this trend was similar in all three affected tryptophan residues (data not included). Since the Kyncontaining peptide exhibits absorption above 400 nm in visible light wavelength range (Figure 6A), change in color from slightly yellow to brown can be attributed to the significant increase in levels of Kyn, rather than NFK. The time-dependent accumulation of these tryptophan oxidation products correlates with color change of the mAb samples from colorless to yellow and then to brown after 14-day light-irradiation. Chromophores Formed in Heat-Stressed mAb. To understand the origin of yellow color developed in a heatstressed mAb incubated at 25 °C for 12 months, it was analyzed by reduced and nonreduced RP-HPLC. The only peak observed in the nonreduced RP-HPLC chromatogram at 330 nm was the intact mAb peak. Similarly, only peaks corresponding to light chain and heavy chain were observed in the reduced RP-HPLC chromatogram collected at 330 nm. The UV−visible spectrum of the intact mAb eluted from nonreduced RP-HPLC showed a shoulder band centered at 340 nm, which was similar, but not identical to that of 14-day light-irradiated mAb (Supporting Information Figure S-1). In addition, the absorbance in the wavelength range of 380 to 460 nm did not increase, which is distinct from that of the lightirradiated mAb. No other peaks were detected by both nonreduced and reduced RP-HPLC at 330 nm. These findings showed that in the heat-stressed mAb the chromophore was associated with the mAb itself, but not with the formulation excipients. The heat-stressed mAb was analyzed by tryptic peptide mapping with online tandem mass spectrometry to identify the modifications responsible for the color change. The UV profile of the peptide mapping of the heat-stressed mAb sample (Figure 5B) showed two peptides that exhibited distinct absorption at 355 nm. The peptide eluted at 44 min contained the oxidation product Kyn of Trp-90 on the light chain. The oxidation products of Trp-90 on the light chain identified in the heat-stressed mAb were the same as those in the 14-day lightirradiated samples. However, the distribution of those oxidation products was not the same between the heat-stressed and photoirradiated samples. The 14-day light-irradiated sample showed 7.4% Kyn and 7.6% OH-Trp/Oia (+16 Da), while 2.9% Kyn and 14.6% OH-Trp/Oia (+16 Da) were present in the heat-stressed sample based on mass spectrometric data. The mass of the peptide eluted at 73 min did not match with often reported tryptophan oxidation products, such as Kyn, NFK, OH-Trp, and 3-OH-Kyn with a mass increase of +4 Da, +32
new absorption bands were detected for the peptides containing oxidized His-312 compared to the parent peptide containing nonoxidized His-312 (data not included). It further confirmed that the color change of the mAb under the near UV−visible light exposure was not due to the oxidation products of Met and His residues. In this study, except for the peptide containing Kyn-104, no other peptide was detected with absorbance at 420 nm (data not included). This shows that Kyn contributed to the absorbance increase in the wavelength range of 380 to 460 nm for the near UV−visible light-irradiated mAb (Supporting Information Figure S-1) because of its absorption above 400 nm. Time-Dependent Accumulation of Chromophores in Near UV−visible Light-Irradiated mAb. The degree of oxidation at each residue was determined by selective ion trace analysis of the oxidized peptides and corresponding parent peptide. Seven residues, including three Trps, three Mets, and one His were oxidized to different extents with increasing light exposure time. Figure 7 shows a plot of the % unmodified
Figure 7. Plot of the % unmodified peptide against the light-irradiation time: Met-430 (blue diamond), Met-254 (orange triangle), Met-360 (pink square ), Trp-104 (black circle), His-312 (green diamond), Trp90 (blue circle), and Trp-50 (orange square).
peptide versus the photoirradiation time in the 14-day photodegradation study. The degradation rate at various sites ranked in the following order: Met-254, Met-430 > Met-360, Trp-104 > Trp-90, His-312, Trp-50. These results show that Met residues, which have no absorbance above 260 nm, were oxidized at a much faster rate than Trp residues, which have absorbance above 300 nm.22 This fact indicates that oxidation of Met, Trp, and His is more dependent on their accessibility to reactive oxygen species or hydroxyl radicals generated from light-irradiation in solvent and their redox potentials than direct light absorption by themselves.25 The Met oxidation products were relatively simple. Met sulfoxide was the major Met oxidation product, while Met sulfone was present at a low level or not detected. One out of four tryptophans on the light chain and two out of 11 tryptophans on the heavy chain were partially oxidized (Figure 7). Among these three oxidized Trp residues, Trp-104 on the heavy chain showed the fastest oxidation rate. Its oxidation rate was about three times as fast as other two tryptophan residues: Trp-90 on the light chain and Trp-50 on the heavy chain. Therefore, the relative percent of the tryptophan oxidation on the heavy chain was higher than that of the light chain. These results confirmed that the chromophore was more enriched on the heavy chain than the light chain as shown by RP-HPLC. 6855
dx.doi.org/10.1021/ac404218t | Anal. Chem. 2014, 86, 6850−6857
Analytical Chemistry
Article
Figure 8. (A) MS/MS spectrum and fragment assignment of the oxidized peptide eluting at 72.5 min of the heat-stressed mAb and (B) oxidation pathway of Trp mediated by heating and proton.
twice the absorbance as the unmodified Trp. This also indicates that the heat-stressed mAb may contain less tryptophan oxidation than the 14-day near UV−visible light-exposed mAb even though these two samples exhibited a similar chromophore level based on the RP-HPLC results (data not included). Both methionine and histidine residues exhibited a very low level of oxidation. Trp-104 and Trp-50 exhibited a very low level of oxidation too, such as ≤0.5% Kyn. These results demonstrate that the chromophores formed on the heatstressed mAb are the oxidation products derived from two tryptophan residues: Trp-107 and Trp-90. Kyn was the only degradation product exhibiting absorbance above 400 nm and detected by the absorption at 420 nm for the 14-day light-irradiated mAb. Mass spectrometric data showed 37.8% Kyn in the 14-day light-irradiated mAb, but only 3.7% Kyn in the heat-stressed mAb for three oxidized tryptophan residues: Trp-104, Trp-90, and Trp-50. This difference can explain the absorbance increase in the wavelength range of 380−460 nm for the 14-day light-irradiated mAb and the lack of increase for the heat-stressed mAb, which further demonstrates that Kyn contributed to the brown color of the 14-day light-irradiated mAb.
Da, +16 Da, and +20 Da, respectively. It was noted that the mass of this peptide matched the tryptic peptide with the amino acid sequence of DSSSSW(104)ARW(107)FFDLWGR with a 2 Da mass decrease. The mass spectrum of the fragments generated from the doubly charged ion of this modified peptide further showed that the 2 Da mass decrease occurred at Trp107 (Figure 8A). The nomenclature used for fragment ions was described by Roepstorff and Fohlman.26 The masses of the Nterminal fragment b series ions, b9 to b15, and the C-terminal fragment y series ions, y″8 to y″15, of the modified peptides were 2 Da lower than the calculated masses for those fragment ions containing Trp-107. The masses of the N-terminal fragment b series ions, b6 and b8, and the C-terminal fragment y series ions, y3″ to y7″, of the modified peptides were consistent with the calculated masses for those fragment ions not containing Trp107. These data prove that the oxidized Trp-107 with the 2 Da mass decrease was formed by the loss of hydrogen through a new oxidation pathway (Figure 8B), which has not been reported before. This modification was present in the heatstressed mAb, but not in the near UV−visible light-irradiated mAb. The photo generated hydroxyl radicals or singlet oxygen would not be present in the heat-stressed mAb sample. Therefore, it is less likely that this 2 Da mass loss modification was induced by photo generated hydroxyl radicals or singlet oxygen rather than by heating and proton. The absorption spectrum of this tryptic peptide showed one absorption band at the maximum absorbance wavelength 280 nm because of the two unmodified tryptophan residues and another strong absorption band at the maximum absorbance wavelength 335 nm because of the modified Trp-107 (Figure 5C). The new double bond, formed through the loss of hydrogen, conjugated to the double bond connected to the nitrogen with a free electron pair (Figure 8B) could reduce the energy required to excite electrons, resulting in the new absorption band at 335 nm. It is worth noting that the intensity at the 335 nm absorption band corresponding to the oxidized Trp-107 was almost the same as the intensity at the 280 nm absorption band corresponding to two unmodified Trp residues: Trp-104 and Trp-112. This indicates that the oxidized Trp-107 exhibited
■
SUMMARY In this study, we found that the color change of a near UV− visible light-irradiated mAb was caused by tryptophan oxidation as a primary modification of the mAb. The reduced RP-HPLC results showed that both light chain and heavy chain were involved in the color change, but the chromophores were more enriched on the heavy chain. The SEC results showed that chromophores were present in both aggregate and monomer fractions, but their proportion in aggregates was higher. Seven residues including three tryptophans, three methionines and one histidine were oxidized as identified by peptide mapping. The UV absorption spectra demonstrated that the color change of the mAb was due to the tryptophan oxidation products Kyn, NFK, and OH-Trp, and not the oxidation products from methionine and histidine residues. The oxidation products of tryptophan were complex. Seven oxidation products were 6856
dx.doi.org/10.1021/ac404218t | Anal. Chem. 2014, 86, 6850−6857
Analytical Chemistry
Article
Richard Lew, Hung-Yu Lin, and Christopher Barton for contributions to the peptide mapping.
identified by mass spectrometry for each of three Trp residues (Trp-50 and Trp-104 of the heavy chain and Trp-90 of the light chain), including Kyn, NFK, OH-Trp, Oia, and DiOia. The UV absorption spectra of the abundant Trp oxidation products showed that the absorption band above 280 nm shifted to the longer wavelength in the following order of OH-Trp (+16 Da) < NFK (+32 Da) < Kyn (+4 Da). The semiquantitative analysis further showed that Trp-104 exhibited the fastest oxidation rate among the three oxidized tryptophan residues, resulting in more chromophore on the heavy chain than the light chain as detected by RP-HPLC. The distribution of the tryptophan oxidation products showed that the yellow color observed for the near UV−visible light-irradiated samples in the early part of the time course can be attributed to high abundance of NFK and OH-Trp, and the brown color after 14-days of lightirradiation can be attributed to high level of Kyn, as Kyn has the greatest red shift and exhibits absorption above 400 nm that NFK does not have. Therefore, a protein solution may change color during storage from colorless to yellow to brown because different oxidation forms exhibit different colors. Additionally, we found that the color change observed for the heat-stressed mAb sample was also due to the tryptophan oxidation products, including a new tryptophan oxidation product formed through the loss of hydrogen, which has not been reported before. This finding indicates that a new tryptophan oxidation pathway could also occur in certain environments and should be considered. The difference in Kyn content can explain the absorbance difference in the wavelength range of 380 to 460 nm between the 14-day light-irradiated mAb and heat-stressed mAb because of Kyn absorbance above 400 nm. This observation further supports that Kyn contributes to the brown color of the 14-day light-irradiated mAb. The results from both near UV−visible light-exposed and heatstressed mAb showed that the color change was attributed to chemical modifications of the mAb itself. This study offers a valuable insight into the color-change pathways triggered by near UV−visible light and heat stress in mAb drug product. It also provides a new analytical methodology using absorption spectroscopy in RP-HPLC and SEC to detect color change associated with protein and using a combination of mass spectroscopy and absorption spectroscopy in peptide mapping to directly connect tryptophan oxidation products to color change of the stressed mAb drug product.
■
■
(1) Buss, N. A.; Henderson, S. J.; McFarlane, M.; Shenton, J. M.; Haan, L. d. Curr. Opin. Pharmacol. 2012, 12, 615−622. (2) Carter, P. J. Nat. Rev. Immunol 2006, 6, 343−357. (3) Wei, Z.; Feng, J.; Lin, H. Y.; Mullapudi, S.; Bishop, E.; Tous, G. I.; Casas-Finet, J.; Hakki, F.; Strouse, R.; Schenerman, M. A. Anal. Chem. 2007, 79, 2797−2805. (4) Rehder, D. S.; Chelius, D.; McAuley, A.; Dillon, T. M.; Xiao, G.; Crouse-Zeineddini, J.; Vardanyan, L.; Perico, N.; Mukku, V.; Brems, D. N.; Matsumura, M.; Bondarenko, P. V. Biochemistry 2008, 47, 2518− 2530. (5) Banks, D. D.; Gadgil, H. S.; Pipes, G. D.; Bondarenko, P. V.; Hobbs, V.; Scavezze, J. L.; Kim, J.; Jiang, X. R.; Mukku, V.; Dillon, T. M. J. Pharm. Sci. 2008, 97, 764−779. (6) Qi, P.; Volkin, D. B.; Zhao, H.; Nedved, M. L.; Hughes, R.; Bass, R.; Yi, S. C.; Panek, M. E.; Wang, D.; Dalmonte, P.; Bond, M. D. J. Pharm. Sci. 2009, 98, 3117−3130. (7) Pirie, A. Biochem. J. 1971, 125, 203−208. (8) Lerman, S.; Borkman, R. Ophthalmic Res. 1976, 8, 335−353. (9) Hood, B. D.; Garner, B.; Truscott, R. J. W. J. Biol. Chem. 1999, 274, 32547−32550. (10) Finley, E. L.; Dillon, J.; Crouch, R. K.; Schey, K. L. Protein Sci. 1998, 7, 2391−2397. (11) Pirie, A. Biochem. J. 1972, 128, 1365−1367. (12) Lennox, F. G.; Rowlands, R. J. Photochem. Photobiol. 1969, 9, 359−367. (13) Schäfer, K.; Goddinger, D.; Höcker, H. J. Soc. Dyers Colour. 1997, 133, 350−355. (14) Dyer, J. M.; Bringans, S. D.; Bryson, W. G. Photochem. Photobiol. Sci. 2006, 5, 698−706. (15) Yang, J.; Wang, S.; Liu, J.; Raghani, A. J. Chromatrogr. A 2007, 1156, 174−182. (16) Ji, J. A.; Zhang, B.; Cheng, W.; Wang, Y. J. J. Pharm. Sci. 2009, 98, 4485−4500. (17) Mason, B. D.; Schöneich, C.; Kerwin, B. A. Mol. Pharmaceutics 2012, 9, 774−790. (18) Simat, T.; Meyer, K.; Steinhart, H. J. Chromatogr. A 1994, 661, 93−99. (19) Kuroda, M.; Sakiyama, F.; Narita, K. J. Biochem. (Tokyo) 1975, 78, 641−651. (20) Previero, A.; Coletti-Previero, M. A.; Jolles, P. J. Mol. Biol. 1967, 24, 261−268. (21) Reubsaet, J. L. E.; Beijnen, J. H.; Bult, A.; Maanen, R. J.; Marchal, J.A. D.; Underberg, W. J. M. J. Pharm. Biomed. Anal. 1998, 17, 955−978. (22) Dreaden, T. M.; Chen, J.; Rexroth, S.; Barry, B. A. J. Biol. Chem. 2011, 286, 22632−22641. (23) Zhao, F.; Ghezzo-Schöneich, E.; Aced, G. I.; Hong, J.; Milby, T.; Schöneich, C. J. Biol. Chem. 1997, 272, 9019−9029. (24) Błaszak, M.; Jankowska, E.; Kowalik-Jankowsk, T. J. Inorg. Biochem. 2013, 129, 62−70. (25) Schöneich, C. Biochim. Biophys. Acta 2005, 1703, 111−119. (26) Roepstorff, P.; Fohlman, J. Biomed. Mass Spectrom. 1984, 11, 601.
ASSOCIATED CONTENT
S Supporting Information *
Additional information was noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.
■
REFERENCES
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Phone: 1-301-398-4267. Present Address †
Alla Polozova: Amgen, Analytical Sciences, 40 Technology Way, West Greenwich, RI 02817 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We thank Jose Casas-Finet and Christopher Barton for assistance with preparation of this manuscript. We thank 6857
dx.doi.org/10.1021/ac404218t | Anal. Chem. 2014, 86, 6850−6857
