Sequence-Specific Formation of d-Amino Acids in a Monoclonal

Article pubs.acs.org/molecularpharmaceutics

Sequence-Specific Formation of D‑Amino Acids in a Monoclonal Antibody during Light Exposure Olivier Mozziconacci and Christian Schöneich* Department of Pharmaceutical Chemistry, University of Kansas, 2095 Constant Avenue, Lawrence, Kansas 66047, United States S Supporting Information *

ABSTRACT: The photoirradiation of a monoclonal antibody 1 (mAb1) at λ = 254 nm and λmax = 305 nm resulted in the sequence-specific generation of D-Val, D-Tyr, and potentially DAla and D-Arg, in the heavy chain sequence [95−101] YCARVVY. D-Amino acid formation is most likely the product of reversible intermediary carbon-centered radical formation at the αC-positions of the respective amino acids (αC• radicals) through the action of Cys thiyl radicals (CysS•). The latter can be generated photochemically either through direct homolysis of cystine or through photoinduced electron transfer from Trp and/or Tyr residues. The potential of mAb1 sequences to undergo epimerization was first evaluated through covalent H/D exchange during photoirradiation in D2O, and proteolytic peptides exhibiting deuterium incorporation were monitored by HPLC−MS/MS analysis. Subsequently, mAb1 was photoirradiated in H2O, and peptides, for which deuterium incorporation in D2O had been documented, were purified by HPLC and subjected to hydrolysis and amino acid analysis. Importantly, not all peptide sequences which incorporated deuterium during photoirradiation in D2O also exhibited photoinduced D-amino acid formation. For example, the heavy chain sequence [12−18] VQPGGSL showed significant deuterium incorporation during photoirradiation in D2O, but no photoinduced formation of D-amino acids was detected. Instead this sequence contained ca. 22% D-Val in both a photoirradiated and a control sample. This observation could indicate that D-Val may have been generated either during production and/or storage or during sample preparation. While sample preparation did not lead to the formation of D-Val or other D-amino acids in the control sample for the heavy chain sequence [95−101] YCARVVY, we may have to consider that during hydrolysis N-terminal residues (such as in VQPGGSL) may be more prone to epimerization. We conclude that the photoinduced, radical-dependent formation of D-amino acids requires not only the intermediary formation of a αC• radical but also sufficient flexibility of the protein domain to allow both pro-chiral faces of the αC• radical to accept a hydrogen atom. KEYWORDS: IgG1, photodegradation, D-amino acids, tryptophan, thiyl radicals B light (λ = 280−315 nm).11−15 More importantly, several Trp oxidation products display absorption maxima at λ > 300 nm. For example, N-formylkynurenin (NFK) has an absorption maximum with λmax = 318 nm, and is an efficient photosensitizer of amino acid oxidation via singlet oxygen generation.12,13,16,17 Approximately 3% of the sunlight reaching the surface of the earth account for UV-B.18−20 In addition, indoor fluorescent light sources emit small but measurable levels of UV-B light.21−23 Therefore, the potential for photodegradation of a pharmaceutical protein exists, but its extent will depend on the length of light exposure, the light intensity, and the nature of the protein and container. Mechanistic studies have provided strong evidence for lightinduced electron transfer from Trp to protein disulfide bonds, ultimately resulting in the formation of Cys and Cys thiyl radicals, CysS• (reactions 1 and 2).24−29 A pair of CysS• can also be generated through the direct photolysis of cystine (reactions 3).26−30 Our recent studies on the reactivity of CysS•

1. INTRODUCTION Over recent years pharmaceutical scientists have become increasingly aware of the pronounced sensitivity of pharmaceutical proteins to light. Light exposure may occur during production, purification, fill and finish, storage, and patient administration and/or handling.1,2 The consequences of light exposure will vary from protein to protein and depend on additional factors such as formulation conditions. To this end, the light exposure of various protein pharmaceuticals and formulations has been reported to induce discoloration, aggregation, and chemical degradation such as oxidation, covalent cross-linking, and fragmentation.3,4 However, our knowledge on potential chemical degradation pathways during the photodegradation of protein pharmaceuticals is currently at best limited, corroborated by an ever increasing number of covalent modifications of proteins reported in the pharmaceutical and biochemical literature on photolytic and oxidative stress.3,5−10 The major chromophores in proteins are the aromatic amino acids and cystine, which display absorption maxima with λmax < 300 nm. However, specifically the absorption spectrum of tryptophan (Trp) extends to wavelengths >300 nm, rendering Trp the amino acid most sensitive to photodegradation by UV© 2014 American Chemical Society

Received: Revised: Accepted: Published: 4291

July 25, 2014 September 30, 2014 October 4, 2014 October 4, 2014 dx.doi.org/10.1021/mp500508w | Mol. Pharmaceutics 2014, 11, 4291−4297

Molecular Pharmaceutics

Article

m2 and 6.4 W/m2, respectively. After 30 min of irradiation, the samples irradiated at λmax = 254 nm and λmax = 305 nm received a dose of light of 0.6 and 1.35 J, respectively. The power of lights and doses were calculated by means of a radiometer JX11 (Jelight Company Inc., Irvine, CA). 3.2. Removal of D2O. The control and UV-irradiated samples, prepared in either H2O or D2O, were purified through an Amicon membrane with a cutoff of 10 kDa (Millipore, Billerica, MA). Each sample was placed in an Amicon Ultra tube of which the membrane was equilibrated with ammonium bicarbonate buffer (NH4HCO3, 50 mM, pH 8.0). After purification, for each sample, a volume of ∼20 μL was recovered and diluted with 480 μL of ammonium bicarbonate buffer. These samples were purified a second time and reconstituted in a final volume of 500 μL of ammonium bicarbonate buffer. 3.3. Digestion. After purification, the controls and irradiated samples were heated to 75 °C in the presence of 4 mM BMS for 30 min to reduce disulfide bonds. The reduced cysteine residues were alkylated at 45 °C with 8 mM IAA. After 2 h of incubation at 45 °C, the samples were purified through an Amicon membrane according to the protocol described above. For each sample, the purified fraction was reconstituted in 500 μL of ammonium bicarbonate buffer (50 mM, pH 8.0). The samples were sonicated for 1 min prior to the addition of 5 μg of thermolysin. The samples were then incubated at 75 °C for 1 h. After 1 h, 10 μg of trypsin was added to each sample and incubation was continued at 37 °C for 2 h. The digestion was completed by addition of 2.5 μg of Glu-C per sample and maintenance of the incubation at 37 °C overnight. The digestion was stopped with the addition of 50 μL of formic acid (10% v:v) to each sample. We used three different enzymes to obtain the shortest proteolytic peptides in order to simplify the MS/MS analysis of the proteolytic peptides which covalently incorporated a deuterium atom. 3.4. High Performance Liquid Chromatography and Mass Spectrometry Analysis. The following protocol was used to detect the covalent incorporation of deuterium into the proteolytic peptides. Note that there is no possibility for back exchange from C−D bonds during sample processing and analysis. A volume of 3 μL of the digests was injected onto a reversedphase capillary C18 column (Vydac, 250 × 0.5 mm, 5 μm) which was attached to a capillary set of pumps (CapLC), delivering a mixture of solvents A (99% H2O, 1% acetonitrile (ACN), 0.1% formic acid (FA), v:v:v), and B (80% ACN, 10% H2O, 10% isopropanol (iPr), 0.1% FA, v:v:v) at a flow rate of 20 μL/min. The capillary column was attached to a Micromass Q-TOF Premier (Waters Corp., Milford, MA) mass spectrometer. The mixture of solvents A and B was delivered according to the following gradient program: between 0 and 10 min the mixture of solvents consisted of 99% of solvent A and 1% of solvent B. After 10 min, the content of solvent B was linearly increased to 40% within 100 min. 3.5. High Performance Liquid Chromatography and UV−Vis Analysis. The following protocol was used to purify the proteolytic fragments, for which mass spectrometry analysis revealed covalent incorporation of deuteron atoms. A volume of 200 μL of the digests was injected onto a reversed-phase C18 column (Vydac, 250 × 4.6 mm, 5 μm) which was attached to a set of Shimadzu pumps (LC-10AS), delivering a mixture of solvents A (100% H2O, 0.1% trifluoroacetic acid (TFA), v:v) and B (80% acetonitrile

radicals in peptides and proteins, including IgG1 and IgG2, have shown a pronounced tendency of CysS• to reversibly and selectively abstract hydrogen atoms from adjacent amino acids, from both the αC−H and side chain C−H bonds (equilibrium 4).3,31 In a model peptide containing the sequence -Ala-CysAla-, such hydrogen transfer reactions resulted in the conversion of L-Ala into D-Ala, i.e., epimerization of the peptide sequence.28 This result prompted us to look into the possibility of photochemical, CysS•-dependent epimerization in a monoclonal antibody (mAb), referred to throughout this manuscript as mAb1. We will report here that the exposure of mAb1 to light results in highly selective epimerization of specific peptide sequences, including the conversion of L-Tyr into D-Tyr and L-Val into D-Val. Epimerization of these amino acids has not been observed during accelerated stability testing of IgG1, IgG2, and IgG4 at elevated temperatures and alkaline pH, where predominantly epimerization at Ser and at one Cys residue in the hinge region were detected.32,33 Hence, the photolytic epimerization targets a set of amino acids significantly differently from that generated through pHdependent epimerization. Trp + hv → Trp* → Trp + e−

(1)

CysSSCys + e− → CysSSCys•− ⇌ CysS• + CysS−

(2)

CysSSCys + hv → 2CysS•

(3)

CysS• + C−H ⇌ CysSH + C•

(4)

2. MATERIAL mAb1 (14 mg/mL) was provided by a pharmaceutical company in phosphate buffer. mAb1 was dispensed in 200 μL aliquots and stored at −80 °C. Sodium acetate (NaAc), sodium phosphate monobasic (NaH2PO4), potassium phosphate dibasic (K2HPO4), o-phthalaldehyde (OPA), N-acetyl-L-cysteine (NAC), methanol (MeOH), bis(2-mercaptoethyl) sulfone (BMS), and iodoacetamide (IAA) were supplied by Sigma-Aldrich (St. Louis, MO) at the highest purity grade. The nonionic polyoxyethylene detergent Brij-35 (30% w:v) was supplied by Pierce (Thermo Scientific, Rockford, IL). Deuterium oxide (D2O) was supplied by Cambridge Isotopes Laboratories, Inc. (Tewksbury, MA). Thermolysin, Glu-C, and trypsin were purchased at Promega (Madison, WI). 3. EXPERIMENTAL METHODS 3.1. Irradiation. 200 μL mAb1 aliquots (30 mg/mL) were diluted with 800 μL of either H2O or D2O containing phosphate buffer (110 mM, pH/pD = 7.4, final protein concentration: 6 mg/mL or 43 μM). Each sample was split into two, of which 500 μL was used as control sample and 500 μL was UV irradiated. The samples were purged with Ar prior to irradiation and were UV irradiated for 30 min at λmax = 254 nm in quartz tubes or at λmax = 305 nm in pyrex tubes. (The spectral distribution corresponds to approximately 90% of photons within the wavelengths 280 and 320 nm, with λmax = 305 nm. The Pyrex wavelength cutoff is λ ∼ 290 nm.) The irradiation was performed by means of a UV irradiator (Rayonet, The Southern New England Ultraviolet Company, Branford, CN) equipped with four UV lamps (either RMR2537Å or RPR3000, The Southern New England Ultraviolet Company, Branford, CN). The lamps emitting at λmax = 254 nm and λmax = 305 nm deliver a power of approximately 3.5 W/ 4292

dx.doi.org/10.1021/mp500508w | Mol. Pharmaceutics 2014, 11, 4291−4297


Article

T, 150 × 4.6 mm, 3 μm), which was attached to a set of Shimadzu pumps (LC-10AS), delivering a mixture of solvents A (100% sodium acetate, 25 mM, pH 6.5) and B (100% MeOH) at a flow rate of 0.7 mL/min. The C18 column was maintained at 35 °C. The mixture of solvents A and B was delivered according to the following gradient program: between 0 and 15 min the mixture of solvents consisted of 100% solvent A. After 15 min, the content of solvent B was increased to 10% within 30 min. After 30 min, the content of solvent B was increased to 50% within 110 min. Subsequently, the column was washed for 5 min with 50% solvent B, and reequilibrated for 45 min with 100% solvent A prior to the next injection. The chromatograms were recorded with a Shimadzu fluorescence detector (RF10Axl, λex = 340 nm, λem = 455 nm).

(ACN), 20% H2O, 0.1% TFA, v:v:v) at a flow rate of 1 mL/ min. The C18 column was maintained at room temperature. The mixture of solvents A and B was delivered according to the following gradient program: between 0 and 10 min the mixture of solvents consisted of 99% solvent A and 1% solvent B. After 10 min, the solvent B was linearly increased to 50% within 100 min. The chromatograms were recorded with a Shimadzu UV detector (SPD-10AV) set at λ = 280 nm. Each peak was collected using a Shimadzu fraction collector (FRC-10A). Each fraction was concentrated under vacuum until a final volume of approximately 500 μL using a CentriVap concentrator (Labconco, Kansas City, MO). 3.6. Control of the Purity of the Collected Fractions. The following protocol was designed to quickly control the purity of each fraction collected during the fractionation of the peptide digests. Ten microliters of each fraction was injected onto a reversed-phase capillary C18 guard-column (LCPacking, 10 × 0.5 mm, 5 μm), which was attached to a capillary set of pumps (CapLC), delivering a mixture of solvents A (99% H2O, 1% acetonitrile (ACN), 0.1% formic acid (FA), v:v:v) and B (80% ACN, 10% H2O, 10% isopropanol (iPr), 0.1% FA, v:v:v) at a flow rate of 20 μL/min. The guard column was attached to a Micromass Q-TOF Premier (Waters Corp., Milford, MA) mass spectrometer. The mixture of solvents A and B was delivered according to the following gradient program: between 0 and 1 min the mixture of solvents consisted of 10% solvent A and 90% solvent B. After 1 min, the content of solvent B was increased to 80% within 10 min. 3.7. Amino Acid Analysis. The collected fractions containing the proteolytic peptides were vacuum-dried and reconstituted in 200 μL of 6 N HCl (Pierce, Thermo Scientific, Rockford, IL). 100 μL of the 6 N HCl solution was placed under vacuum for 30 s and then maintained under an N2 atmosphere for 2 min. The N2-saturation−vacuum cycle was repeated five times. Then, the solution was placed at 130 °C for ca. 45 min. The hydrolyzed sample was vacuum-dried and reconstituted in 200 μL of sodium acetate (25 mM, pH 6.5). 3.8. High Performance Liquid Chromatography Analysis of L- and D-Amino Acids. 200 μL samples of the hydrolyzed peptide solutions prepared in sodium acetate 25 mM, pH 6.5, were placed in an autosampler (Shimadzu, SIL10A) at 5 °C in the dark. Three different stock solutions were prepared in order to derivatize the amino acids with OPA and NAC. Solution S1 consisted of 9,967 μL of potassium phosphate buffer, K2HPO4 (190 mM, pH 9.5), and 33 μL of Brij-35 solution (30% w:v). Solution S2 consisted of 5.5 mg of OPA, and 12 mg of NAC in 265 μL of methanol (MeOH) and 4,735 μL of potassium phosphate buffer, K2HPO4 (490 mM, pH 9.5). Solution S3 consisted of sodium phosphate buffer, NaH2PO4 (200 mM, pH 4.5). Solutions S1, S2, and S3 were placed in the autosampler at 5 °C in the dark. The solutions were mixed together using the autosampler to ensure reproducibility of the derivatization reaction. The solutions were mixed as follows: 50 μL of each of S1, S2, and the hydrolyzed peptide solution were mixed together. During 1 min, the final solution was mixed by soaking and dispensing 100 μL of the solution using the syringe of the autosampler. After mixing, the solution was left for 1 min at 5 °C in the dark. Then, 150 μL of solution S3 was added to the solution. Again, the final solution was mixed for 1 min by soaking and dispensing 100 μL of the solution using the syringe of the autosampler. Then, 25 μL of the final solution was injected onto a reversed-phase capillary C18 column (Supelcosil LC-18-

4. RESULTS 4.1. Strategy. Our strategy to identify D-amino acid containing sequences in photoirradiated mAb1 followed the methodology that we had applied to characterize C−H bonds sensitive to light-induced covalent H/D exchange,3,27−29,34,35 and which Huang et al.32 and Amano et al.33 had applied to detect pH-dependent epimerization in several IgG variants. In initial experiments, mAb1 was dissolved in D2O and lightinduced incorporation of deuterium into the peptide map of mAb1 monitored by HPLC−MS and HPLC−MS/MS analysis. Covalent deuterium incorporation into a peptide sequence would indicate the sensitivity of such a sequence to carboncentered radical formation according to equilibrium 4, and potential epimerization at the αC-position. We then monitored the retention time of deuterium-incorporating peptides on an analytical column, so that these peptides could be fractionated and collected for amino acid analysis after photolysis in H2O. Peptides of interest were collected and subjected to acid hydrolysis, and, after derivatization with N-acetyl-L-cysteine and OPA, the derivatives of L- and D-amino acids were separated by reversed-phase chromatography, coupled to fluorescence detection. In our photolysis experiments, we used two different light sources. For proof-of-concept, we exposed IgG1 to high intensity 254 nm light as these experimental conditions would allow us to optimize the detection of D-amino acids. In a second set of experiments, we exposed IgG1 to UV-B and UV-A of wavelengths >295 nm, representative for light exposure that mAb1 formulations may experience under cool white (fluorescent) light and/or sunlight. 4.2. Covalent Incorporation of Deuterium during Photoirradiation of IgG1 in D2O. The LC−MS analyses (protocol described in section 3.4) of the proteolytic peptides obtained after photoirradiation (λ = 254 nm) of IgG1 in H2O and D2O revealed that eight ions with mass-to-charge ratios (m/z) of 802.3, 657.3, 889.3, 1047.4, 984.8, 1102.4, 723.4, and 893.4 covalently incorporated deuterium atoms after photoirradiation in D2O (Supporting Information, Figures S1−S8). The complete procedure to analyze the isotopic distributions of the peptide ions recovered from H2O and D2O is given elsewhere.35 In brief, we observed that, for a given peptide ion, generated after the proteolytic digest of photoirradiated mAb1 in D2O (panels A, red curves, Figures S1−S8 in the Supporting Information), the isotopic distributions are moved toward higher masses, although, for the same peptide ion, the nonirradiated sample prepared in D2O (panels B, red curves, Figures S1−S8 in the Supporting Information) did not show any variation of its isotopic distribution. 4293



Article

The structures of the six other peptides ions with m/z 802.3, 1047.4, 984.8, 1102.4, 723.4, and 893.4 could not be identified by MS/MS analysis, due to a poor fragmentation. We did not analyze these peptides for potential D-amino acid incorporation, as we were not able to localize these peptides to specific sequences, suggesting that they are photoproducts. Based on UV detection and peak intensity during HPLC analysis, these photoproducts appear to be generated in an equal or higher amount than the peptides that we have selected for amino acid analysis. For example, the peptide that eluted at 77 min (Figure 2, last peak) is one of the major photoproducts. Further investigations about the structures of these photoproducts are underway.

Peptide mapping permitted the identification of the sequence of the peptide ions with m/z 657.3 and 889.3. MS/MS analysis of the ions with m/z of 657.3 and 889.3 are presented in the Supporting Information (Figures S9 and S10). The positions of each of these two peptides within the sequence of mAb1 are highlighted in Figure 1.

Figure 2. HPLC chromatogram recorded at λ = 280 nm of proteolytic digest of mAb1. The chromatographic peaks are collected. Fractions 29 and 45 contain the ions with m/z 657.3 and 889.3, respectively. The purity of these fractions is assessed by mass spectrometry analysis (see Supporting Information, Figure S11).

Figure 1. Localization of the proteolytic peptides showing covalent deuterium incorporation after UV irradiation of mAb1. The peptide VQPGGSL is located in the variable domain of the heavy chain. The residues YCAR of the peptide YCARVVY are located in the variable domain of the heavy chain. The residues VVY of the peptide YCARVVY are located in the third complementary determining region of the heavy chain (CDR-H3).

After UV irradiation of mAb1 at λmax = 305 nm, only the peptide YuARVVY (m/z 889.3) was collected for amino acid analysis. 4.3. Fractionation and Amino Acid Analysis of Peptides with Potential for Light-Induced Epimerization. In the following, the label “u” in the sequence of the peptide YuARVVY refers to sulfenic acid of the original cysteine residue. 4.3.1. Fractionation of the Peptides VQPGGSL and YuARVVY. The proteolytic peptides were fractionated according to the protocol described in section 3.5 (Figure 2). The fractions were collected and reanalyzed by mass spectrometry to ensure the purity of each fraction (protocol described in section 3.6). The mass spectrometry analysis of fractions 29 and 45 (Figure 2) reveals the presence of pure peptide ions with m/ z 657.3 (VQPGGSL) and 889.3 (YuARVVY), respectively (Figure S11 in the Supporting Information). 4.3.2. Amino Acid Analysis of the Peptides VQPGGSL and YuARVVY. 4.3.2.1. Amino Acid Analysis of VQPGGSL after Photoirradiation of mAb1 at λ = 254 nm. Our amino acid analysis does not permit the detection of proline. The amino acid analysis of the peptide VQPGGSL obtained from the proteolytic digest of the nonirradiated mAb1 shows the following amino acids (Figure 3, blue curve): (i) L-Glu, which is the result of the hydrolysis of L-Gln, (ii) Gly, (iii) LVal, (iv) D-Val, and (iv) L-Leu. The Ser residue was not detected. Ser can undergo β-elimination during hydrolysis.36,37 A comparison of the amino acid analyses of the peptide VQPGGSL obtained before and after irradiation of mAb1 at λmax = 254 nm does not reveal any additional formation of D-

The ion with m/z 657.3 corresponds to the heavy chain sequence [12−18] VQPGGSL. The MS/MS spectrum of the peptide ion with m/z 657.3 shows the formation of the fragments ions y2−y6 and b3−b6, which allow for the identification of the sequence VQPGGSL (Figure S9 in the Supporting Information). The ion with m/z 889.3 corresponds to the heavy chain sequence [95−101] YCARVVY where the cysteine residue is oxidized to sulfenic acid. The MS/MS spectrum of the peptide ion with m/z 889.3 shows the formation of the fragment ions y3−y6, and b2−b6, which allow for the identification of the sequence YuARVVY, where “u” stands for the sulfenic acid of cysteine (CysSOH). The b2, y5, and y6 fragment ions permit, in particular, identification of the oxidation of the cysteine residue into its sulfenic acid (Figure S10 in the Supporting Information). Importantly, sulfenic acids are labile toward further oxidation or disulfide formation with thiols. Derivatization of Cys prior to digestion ensured that no free thiol was available for reaction with the sulfenic acid; however, alkylation of the specific Cys residue in YCARVVY was not observed. Because the peptides were exposed to air for several hours during the different purification steps, the formation of sulfenic acid is likely an oxidation reaction that occurred during air exposure of the nonalkylated cysteine of the peptide YCARVVY. 4294



Article

Figure 3. Amino acid analysis of the proteolytic peptide VQPGGSL obtained before (blue curve) and after (red curve) photoirradiation of mAb1 at λmax = 254 nm. HPLC chromatogram recorded by means of a fluorescence detector (λex = 340 nm, λem = 455 nm).

amino acids (specifically D-Val) as a result of the photoirradiation (Figure 3), suggesting that D-Val in this peptide is present in the starting material. Another possibility is that some D -Val is generated during sample preparation and/or hydrolysis; however, we did not detect D-Val in nonphotoirradiated control samples of another peptide (see below), suggesting that our experimental conditions were sufficiently mild to avoid D-amino acid formation during hydrolysis. From the representative full-scale chromatogram (Figure 3), we estimate that the ratio D-Val:L-Val is approximately 0.22:1.22 in both the control and the irradiated samples. This value is highly reproducible, where the ratio DVal:L-Val = 0.21:1.21 (±1%) over three separate experiments. 4.3.2.2. Amino Acid Analysis of YuARVVY after Photoirradiation of mAb1 at λ = 254 nm. The amino acid analysis of the peptide YuARVVY obtained from the proteolytic digest of the nonirradiated IgG1 reveals the presence of the following amino acids (Figure 4, blue curve, panel A): (i) L-Arg, (ii) L-Ala, (iii) L-Val, and (iv) L-Tyr. The cysteine-sulfenic acid residue (referred to as “u” in the sequence) was not recovered during our analysis. In contrast with the peptide VQPGSSL, the proteolytic peptide YuARVVY obtained from the nonirradiated control IgG1 does not show any trace of D-Val. The amino acid analysis of the peptide YuARVVY obtained from the proteolytic digest of photoirradiated mAb1 reveals the presence of the following amino acids (Figure 4, red curve, panel A): (i) L-Arg (and potentially D-Arg), (ii) L-Ala (and potentially D-Ala), (iii) L-Tyr and D-Tyr, and (iv) L-Val and D-Val. The absence of Damino acids in the nonirradiated sample suggests that D-Arg and/or D-Ala, D-Tyr, and D-Val are generated during photoirradiation of IgG1 at λ = 254 nm (Figure 4, red curve, panel A). We could not obtain sufficient chromatographic resolution to clearly define whether we observe either D-Arg, D-Ala, or other. However, the formation of D-Val and D-Tyr is clearly observed. Three separate IgG1 samples were photoirradiated at λ = 254 nm. The amino acid analysis of the peptide YuARVVY yields ratios D-Tyr:L-Tyr = D-Val:L-Val = 0.33:1.33 (±3%). 4.3.2.3. Amino Acid Analysis of YuARVVY after Irradiation of mAb1 at λmax = 305 nm. The amino acid analysis of the peptide YuARVVY obtained from the proteolytic digest of the nonirradiated mAb1 reveals the presence of the following amino acids (Figure 5, blue curve, panel A): (i) L-Arg, (ii) L-Ala, (iii) L-Val, and (iv) L-Tyr. The cysteine-sulfenic acid residue (noted as “u” in the sequence) was not recovered during our analysis. The amino acid analysis of the peptide YuARVVY obtained from the proteolytic digest of the irradiated mAb1 revealed the presence of the following amino acids (Figure 5,

Figure 4. Amino acid analysis of the proteolytic peptide YuARVVY obtained before (blue curve, A) and after (red curve, A) photoirradiation of mAb1 at λmax = 254 nm. Amino acid standards L-Ala (blue, B) and D-Ala (red, B). Amino acid standards L-Val (blue, C) and D-Val (red, C). Amino acid standards L-Tyr (blue, D) and D-Tyr (red, D). Amino acid standards L-Arg (blue, E) and D-Arg (red, E). The HPLC chromatograms are recorded by means of a fluorescence detector (λex = 340 nm, λem = 455 nm).

Figure 5. Amino acid analysis of the proteolytic peptide YuARVVY obtained before (blue curve) and after (red curve) photoirradiation of mAb1 at λmax = 305 nm. The HPLC chromatograms are recorded by means of a fluorescence detector (λex = 340 nm, λem = 455 nm).

red curve, panel A): (i) L-Arg (and potentially D-Arg), (ii) L-Ala (and potentially D-Ala), (iii) L-Tyr and D-Tyr, and (iv) L-Val and D-Val. The absence of D-amino acids in the nonirradiated sample suggests that D-Arg and/or D-Ala, D-Tyr, and D-Val are generated during photoirradiation of mAb1 at λmax = 305 nm (Figure 5, red curve, panel A). 4295



Article

Quantitative analysis reveals that photoirradiation at λmax = 305 nm yields the ratios D-Val:L-Val = 0.38:1.38 and D-Tyr:LTyr = 0.25:1.25, respectively.

S3 in the Supporting Information). Here, amino acid analysis of YuARVVY revealed that photoirradiation of IgG1 at λ < 280 nm and λ > 300 nm, induces the formation of D-Tyr and D-Val (Figures 4 and 5). Our results suggest that Arg or/and Ala are also epimerized. However, the retention times of D-Arg and DAla are too close to determine which of Arg and/or Ala were epimerized.

5. DISCUSSION The exposure of proteins to UV light (λ < 280 nm or λ > 290 nm) can lead to the formation of CysS• radicals. UV-C light (λ < 280 nm) can be directly absorbed by the cystine disulfide function, leading predominantly to the homolytic cleavage of the disulfide bond and to the formation of a pair of CysS• radicals (reaction 3).29 UV-B light (λ > 290 nm) is absorbed essentially by aromatic residues such as tryptophan (Trp). The absorption of light by Trp generates an excited state, which can ultimately convert into a radical cation by emitting an electron, which will solvate. The solvated electron will react with a disulfide bond to form a disulfide radical anion (SS•−), which ultimately dissociates into CysS• radical and Cys (reactions 1 and 2). The formation of CysS• radicals is not without consequences for proteins. CysS• radicals can intramolecularly abstract hydrogen atoms from adjacent C−H bonds, leading to the formation of carbon-centered radicals (C•), and establish an equilibrium between CysS•/C−H and CysSH/C• (equilibrium 4). Where such intramolecular H atom transfer occurs between CysS• and a C−H bond of a chiral carbon, such as the αC−H bond of all amino acids except Gly, the resulting αC• displays two prochiral faces, where reverse H atom transfer can generate either the L- or D-form of the amino acid. Covalent deuterium incorporation documents that the photoirradiation of mAb1 at λ < 280 nm and λ > 300 nm leads to the intermediary formation of carbon-centered radicals. In fact, the observation of a shift toward higher masses of the isotopic distributions of the peptide ions, which were obtained after the photoirradiation of mAb1 in D2O, demonstrates that, during the establishment of equilibrium 4, deuterium atoms are covalently incorporated into the peptide structures (Figures S1−S8 in the Supporting Information). More details about the use of deuterium incorporation to monitor the intermediary formation of carbon-centered radicals are given elsewhere.35 However, this technique alone does not permit determination of whether and which amino acid(s) are potentially epimerized, since the intramolecular H atom transfer reaction between CysS• and C−H bonds can occur at nonchiral carbon atoms (e.g., the βC−H bond of Ala or the αC−H bond of Gly),28,34,35,38 or may not involve epimerization due to conformational restrictions. Mass spectrometry analysis of the peptide VQPGSSL revealed covalent incorporation of deuterium atoms (Figure S2 in the Supporting Information). However, amino acid analysis did not show any photoinduced epimerization (Figure 3). For this peptide, the presence of small levels of D-Val is observed for both the nonirradiated control and the photoirradiated sample, suggesting that D-Val was either present in the sequence of mAb1 prior to irradiation or generated during sample preparation. Under the same conditions of hydrolysis, the peptide YuARVVY, which also contains Val residues, did not show any trace of D-Val in the nonirradiated control sample (Figure 4). However, we may have to consider that N-terminal amino acids may behave differently during hydrolysis reactions, and further experiments are necessary to reveal the cause for D-amino acids in the control sample of the peptide VQPGSSL. Mass spectrometry analysis of the peptide YuARVVY revealed the covalent incorporation of deuterium atoms (Figure

6. CONCLUSION The combination of mass spectrometry and amino acid analyses allows us to identify epimers of amino acid residues in a complex structure sequence such as that of mAb1. Our results show that either UV-C or UV-B can induce the epimerization of particular amino acid residues in mAb1. Since the epimerization of certain amino acid residues could lead to the loss of potency of IgG, the development of a quick screening technique to identify the presence of D-amino acid residues in immunoglobulins will be developed.

■

ASSOCIATED CONTENT

* Supporting Information S

Isotopic distributions of the ion with m/z 802.3, 657.3, 889.3, 1047.4, 984.8, 1102.4, 723.4, and 893.4; CID of the ion with m/ z 657.3 and 889.3; mass spectrometry analysis of fractions 29 and 45. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Kerwin, B. A.; Remmele, R. L. J. Protect From Light: Photodegradation and Protein Biologics. J. Pharm. Sci. 2007, 96, 1468−1479. (2) Hawe, A.; Wiggenhorn, M.; van de Weert, M.; Garbe, J. H. O.; Mahler, H.-C.; Jiskoot, W. Forced Degradation of Therapeutic Proteins. J. Pharm. Sci. 2012, 101, 895−913. (3) Mozziconacci, O.; Kerwin, B. A.; Schöneich, C. Exposure of a Monoclonal Antibody, IgG1, to UV-Light Leads to Protein Dithiohemiacetal and Thioether Cross-Links: a Role for Thiyl Radicals? Chem. Res. Toxicol. 2010, 23, 1310−1312. (4) 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. Characterization of the Photodegradation of a Human IgG1Monoclonal Antibody Formulated as a High-Concentration Liquid Dosage Form. J. Pharm. Sci. 2009, 98, 3117−3130. (5) Davies, M. J. The Oxidative Environment and Protein Damage. Biochim. Biophys. Acta 2005, 1703, 93−109. (6) Pattison, D. I.; Rahmanto, A. S.; Davies, M. J. Photo-Oxidation of Proteins. Photochem. Photobiol. Sci. 2012, 11, 38−53. (7) Luo, Q.; Joubert, M. K.; Stevenson, R.; Ketchem, R. R.; Narhi, L. O.; Wypych, J. Chemical Modifications in Therapeutic Protein Aggregates Generated Under Different Stress Conditions. J. Biol. Chem. 2011, 286, 25134−25144. (8) Haywood, J.; Mozziconacci, O.; Allegre, K. M.; Kerwin, B. A.; Schöneich, C. Light-Induced Conversion of Trp to Gly and Gly Hydroperoxide in IgG1. Mol. Pharmaceutics 2013, 10, 1146−1150. (9) Manning, M. C.; Chou, D. K.; Murphy, B. M.; Payne, R. W.; Katayama, D. S. Stability of Protein Pharmaceuticals: An Update. Pharm. Res. 2010, 27, 544−575.

4296



Article

(10) Zhang, Q.; Schenauer, M. R.; McCarter, J. D.; Flynn, G. C. IgG1 Thioether Bond Formation in Vivo. J. Biol. Chem. 2013, 288, 16371− 16382. (11) Bent, D. V.; Hayon, E. Excited State Chemistry of Aromatic Amino Acids and Related Peptides. 3. Tryptophan. J. Am. Chem. Soc. 1975, 97, 2612−2619. (12) Creed, D. The Photophysics and Photochemistry of the NearUv Absorbing Amino-Acids 0.1. Tryptophan and Its Simple Derivatives. Photochem. Photobiol. 1984, 39, 537−562. (13) Davies, M. J.; Truscott, R. J. W. Photo-Oxidation of Proteins and Its Role in Cataractogenesis. J. Photochem. Photobiol., B 2001, 63, 114−125. (14) Grossweiner, L. I. Photochemistry of Proteins: A Review. Curr. Eye Res. 1984, 3, 137−144. (15) Laustriat, G.; Hasselmann, C. Photochemistry of Proteins. Photochem. Photobiol. 1975, 22, 295−298. (16) Davies, M. J. Singlet Oxygen-Mediated Damage to Proteins and Its Consequences. Biochem. Biophys. Res. Commun. 2003, 305, 761− 770. (17) Walrant, P.; Santus, R. N-Formyl-Kynurenine, a Tryptophan Photooxidation Product, as a Photodynamic Sensitizer. Photochem. Photobiol. 1974, 19, 411−417. (18) Gueymard, C. A. The Sun’s Total and Spectral Irradiance for Solar Energy Applications and Solar Radiation Models. Sol. Energy 2004, 76 (4), 423−453. (19) Leckner, B. The Spectral Distribution of Solar Radiation at the Earth’s SurfaceElements of a Model. Sol. Energy 1978, 20 (2), 143− 150. (20) Bird, R. E.; Hulstrom, R. L.; Lewis, L. J. Terrestrial Solar Spectral Data Sets. Sol. Energy 1983, 30 (6), 563−573. (21) Harm, W. Biological Determination of the Germicidal Activity of Sunlight. Radiat. Res. 1969, 40 (1), 63−69. (22) Harm, W. Biological Effects of Ultraviolet Radiation; IUPAB Biophysics Series; Cambridge University Press: New York, 1980. (23) Photocatalytic Inactivation of Gram-Positive and GramNegative Bacteria Using Fluorescent Light. J. Photochem. Photobiol., A 2007, 186, 335−341. (24) Correia, M.; Neves-Petersen, M. T.; Parracino, A.; di Gennaro, A. K.; Petersen, S. B. Photophysics, Photochemistry and Energetics of UV Light Induced Disulphide Bridge Disruption in Apo-ALactalbumin. J. Fluoresc. 2012, 22, 323−337. (25) Vanhooren, A.; Devreese, B.; Vanhee, K.; Van Beeumen, J.; Hanssens, I. Photoexcitation of Tryptophan Groups Induces Reduction of Two Disulfide Bonds in Goat A-Lactalbumin. Biochemistry 2002, 41, 11035−11043. (26) Mozziconacci, O.; Williams, T. D.; Schöneich, C. Intramolecular Hydrogen Transfer Reactions of Thiyl Radicals From Glutathione: Formation of Carbon-Centered Radical at Glu, Cys, and Gly. Chem. Res. Toxicol. 2012, 25, 1842−1861. (27) Mozziconacci, O.; Kerwin, B. A.; Schöneich, C. Reversible Hydrogen Transfer Reactions of Cysteine Thiyl Radicals in Peptides: the Conversion of Cysteine Into Dehydroalanine and Alanine, and of Alanine Into Dehydroalanine. J. Phys. Chem. B 2011, 115, 12287− 12305. (28) Mozziconacci, O.; Kerwin, B. A.; Schöneich, C. Reversible Hydrogen Transfer Between Cysteine Thiyl Radical and Glycine and Alanine in Model Peptides: Covalent H/D Exchange, Radical-Radical Reactions, and L- to D-Ala Conversion. J. Phys. Chem. B 2010, 114, 6751−6762. (29) Mozziconacci, O.; Williams, T. D.; Kerwin, B. A.; Schöneich, C. Reversible Intramolecular Hydrogen Transfer Between Protein Cysteine Thiyl Radicals and αC−H Bonds in Insulin: Control of Selectivity by Secondary Structure. J. Phys. Chem. B 2008, 112, 15921− 15932. (30) Zhou, S.; Mozziconacci, O.; Kerwin, B. A.; Schöneich, C. The Photolysis of Disulfide Bonds in IgG1 and IgG2 Leads to Selective Intramolecular Hydrogen Transfer Reactions of Cysteine Thiyl Radicals, Probed by Covalent H/D Exchange and RPLC-MS/MS Analysis. Pharm. Res. 2013, DOI: 10.1007/s11095-012-0968-1.

(31) Zhou, S.; Mozziconacci, O.; Kerwin, B. A.; Schöneich, C. Fluorogenic Tagging Methodology Applied to Characterize Oxidized Tyrosine and Phenylalanine in an Immunoglobulin Monoclonal Antibody. Pharm. Res. 2013, 30, 1311−1327. (32) Huang, L.; Lu, X.; Gough, P. C.; De Felippis, M. R. Identification of Racemization Sites Using Deuterium Labeling and Tandem Mass Spectrometry. Anal. Chem. 2010, 82, 6363−6369. (33) Amano, M.; Hasegawa, J.; Kobayashi, N.; Kishi, N.; Nakazawa, T.; Uchiyama, S.; Fukui, K. Specific Racemization of Heavy-Chain Cysteine-220 in the Hinge Region of Immunoglobulin Gamma 1 as a Possible Cause of Degradation During Storage. Anal. Chem. 2011, 83, 3857−3864. (34) Mozziconacci, O.; Kerwin, B. A.; Schöneich, C. Photolysis of an Intrachain Peptide Disulfide Bond: Primary and Secondary Processes, Formation of H2S, and Hydrogen Transfer Reactions. J. Phys. Chem. B 2010, 114, 3668−3688. (35) Mozziconacci, O.; Sharov, V.; Williams, T. D.; Kerwin, B. A.; Schöneich, C. Peptide Cysteine Thiyl Radicals Abstract Hydrogen Atoms From Surrounding Amino Acids: the Photolysis of a Cystine Containing Model Peptide. J. Phys. Chem. B 2008, 112, 9250−9257. (36) Downs, F.; Pigman, W. Isolation and Characterization of Peptides Produced by the Mild Acid Hydrolysis of Bovine Submaxillary Mucin. Int. J. Protein Res. 1970, 2, 27−36. (37) Allerton, S. E.; Perlmann, G. E. Chemical Characterization of the Phosphoprotein Phosvitin. J. Biol. Chem. 1965, 240, 3892−3898. (38) Nauser, T.; Pelling, J.; Schöneich, C. Thiyl Radical Reaction with Amino Acid Side Chains: Rate Constants for Hydrogen Transfer and Relevance for Posttranslational Protein Modification. Chem. Res. Toxicol. 2004, 17, 1323−1328.

4297


Sequence-Specific Formation of d-Amino Acids in a Monoclonal

Recommend Documents