Ranking the Susceptibility of Disulfide Bonds in Human IgG1

May 21, 2010 - However, a complete ranking of the susceptibility of disulfide bonds in IgG1 molecules is lacking. A method including reduction, differ...
1 downloads 0 Views 2MB Size
Download PDF
Anal. Chem. 2010, 82, 5219–5226

Ranking the Susceptibility of Disulfide Bonds in Human IgG1 Antibodies by Reduction, Differential Alkylation, and LC-MS Analysis Hongcheng Liu,* Chris Chumsae, Georgeen Gaza-Bulseco, Karen Hurkmans, and Czeslaw H. Radziejewski

Downloaded via UNIV OF WINNIPEG on July 10, 2018 at 18:40:05 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Process Sciences Department, 100 Research Drive, Abbott Bioresearch Center, Worcester, Massachusetts 01605 One of the basic structural features of human IgG1 is the arrangement of the disulfide bond structure, 4 inter chain disulfide bonds in the hinge region and 12 intra chain disulfide bonds associated with twelve individual domains. Disulfide bond structure is critical for the structure, stability, and biological functions of IgG molecules. It has been known that inter chain disulfide bonds are more susceptible to reduction than intra chain disulfide bonds. However, a complete ranking of the susceptibility of disulfide bonds in IgG1 molecules is lacking. A method including reduction, differential alkylation, and liquid chromatography-mass spectrometry (LC-MS) analysis was developed and employed to investigate the complete ranking order of the susceptibility of disulfide bonds in two recombinant monoclonal antibodies. The results confirmed that inter chain disulfide bonds were more susceptible than intra chain disulfide bonds. In addition, it was observed that the disulfide bonds between the light chain and heavy chain were more susceptible than disulfide bonds between the two heavy chains. The upper disulfide bond of the two inter heavy chain disulfide bonds was more susceptible than the lower one. Furthermore, disulfide bonds in the CH2 domain were the most susceptible to reduction. Disulfide bonds in VL, CL, VH, and CH1 domains had similar and moderate susceptibility, while disulfide bonds in the CH3 domain were the least susceptible to reduction. Interestingly, a difference between IgG1K and IgG1λ was also observed. The difference in the susceptibility of inter light heavy chain disulfide bonds and inter heavy chain disulfide bonds was smaller in IgG1K than in IgG1λ. The intra chain disulfide bonds in the Fab region of IgG1K were also less susceptible than disulfide bonds in the Fab region of IgG1λ. IgG1 is the most abundant subclass of IgG in human serum and has also been the common choice for recombinant monoclonal therapeutics. Knowledge of the structural features, various posttranslational modifications, and the major degradation pathways has been advanced tremendously and is well documented in the literature. One of the basic structural features of a typical IgG1 is the arrangement of disulfide bonds. A typical IgG1 antibody * To whom correspondence should be addressed. Phone: +1 508 849 2591. Fax: +1 508 793 4885. E-mail: [email protected]. 10.1021/ac100575n  2010 American Chemical Society Published on Web 05/21/2010

Figure 1. Diagram showing the major structural features of an IgG1 antibody. The serine (S) residue is only present in IgG1λ.

contains two identical light chains and two identical heavy chains. It contains a total of 16 disulfide bonds including 4 inter chain disulfide bonds in the hinge region and 12 intra chain disulfide bonds associated with 12 individual domains (Figure 1). Several interesting structural features of antibodies with regard to cysteine residues and disulfide bond have been reported recently. First, incomplete disulfide bonds are common, at least, in IgG1 antibodies.1,2 Second, antibodies exist as an ensemble of multiple structural isoforms because of different disulfide bond structures. It has been well understood that the inter heavy chain disulfide bond structures of IgG4 are in equilibrium with intra chain disulfide bonds.3,4 Such an equilibrium can result in the formation of bispecific antibodies even between recombinant therapeutic antibodies and endogeneous antibodies.5 Recent studies have shown multiple forms of IgG2 antibodies because of different disulfide bond structures in addition to the classical (1) Chumsae, C.; Gaza-Bulseco, G.; Liu, H. Anal. Chem. 2009, 81, 6449–6457. (2) Xiang, T.; Chumsae, C.; Liu, H. Anal. Chem. 2009, 81, 8101–8108. (3) Schuurman, J.; Perdok, G. J.; Gorter, A. D.; Aalberse, R. C. Mol. Immunol. 2001, 38, 1–8. (4) Bloom, J. W.; Madanat, M. S.; Marriott, D.; Wong, T.; Chan, S. Y. Protein Sci. 1997, 6, 407–415. (5) Labrijn, A. F.; Buijsse, A. O.; van den Bremer, E. T.; Verwilligen, A. Y.; Bleeker, W. K.; Thorpe, S. J.; Killestein, J.; Polman, C. H.; Aalberse, R. C.; Schuurman, J.; van de Winkel, J. G.; Parren, P. W. Nat. Biotechnol. 2009, 27, 767–771.

Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

5219

disulfide bond linkage.6-8 Third, trisulfide bonds have also been detected in all classes of human IgG.9,10 One of the serious consequence of the dynamic nature of disulfide bond structures is the formation of covalent aggregates of recombinant monoclonal antibodies.11,12 A fundamental question that needs to be addressed is what is the relative stability of disulfide bond structures of IgG molecules? It has been well-known that inter chain disulfide bonds are more susceptible to reduction than intra chain disulfide bonds. However, the complete ranking order of disulfide bond susceptibility including inter chain and intra chain disulfide bonds is lacking. Disulfide bond structure has a significant impact on antibody structure, stability, and biological function. Incomplete formation of disulfide bonds resulted in reduced thermal stability.13 Studies using isolated CH3 domains of human IgG1 antibodies suggested that intra chain disulfide bonds are critical for the domain stability.14,15 Disulfide bonds in other domains likely have the same stabilizing effect because of the similarity of the domain structures and disulfide bond locations. Disulfide bonds also have a significant effect on antibody biological functions. For example, incomplete formation of an intra chain disulfide bond in the variable domain of an IgG1 molecule decreased its potency significantly.16 Site-directed mutagenesis and partial reduction studies indicated that disulfide bonds can affect the effector functions significantly. Site-directed mutagenesis studies indicated that the lack of the inter heavy chain disulfide bonds resulted in reduced complement dependent cytotoxicity (CDC) and antibodydependent cellular cytotoxicity (ADCC) of a chimeric mousehuman IgG.17 In addition, the arrangement of disulfide bonds was also important. A complete loss of ADCC and significant reduction of CDC was observed if the inter light chain heavy chain disulfide bond structure of an IgG1 molecule was altered to that of an IgG4 molecule.18 Partial reduction of disulfide bonds of rabbit IgG and human IgG resulted in a significant reduction in their binding affinities to the first component of complement (C1q) and (6) Wypych, J.; Li, M.; Guo, A.; Zhang, Z.; Martinez, T.; Allen, M. J.; Fodor, S.; Kelner, D. N.; Flynn, G. C.; Liu, Y. D.; Bondarenko, P. V.; Ricci, M. S.; Dillon, T. M.; Balland, A. J. Biol. Chem. 2008, 283, 16194–16205. (7) Martinez, T.; Guo, A.; Allen, M. J.; Han, M.; Pace, D.; Jones, J.; Gillespie, R.; Ketchem, R. R.; Zhang, Y.; Balland, A. Biochemistry 2008, 47, 7496– 7508. (8) Zhang, B.; Harder, A. G.; Connelly, H. M.; Maheu, L. L.; Cockrill, S. L. Anal. Chem. 2010, 82, 1090–1099. (9) Pristatsky, P.; Cohen, S. L.; Krantz, D.; Acevedo, J.; Ionescu, R.; Vlasak, J. Anal. Chem. 2009, 81, 6148–6155. (10) Gu, S.; Wen, D.; Weinreb, P. H.; Sun, Y.; Zhang, L.; Foley, S. F.; Kshirsagar, R.; Evans, D.; Mi, S.; Meier, W.; Pepinsky, R. B. Anal. Biochem. 2010, 400, 89–98. (11) Van Buren, N.; Rehder, D.; Gadgil, H.; Matsumura, M.; Jacob, J. J. Pharm. Sci. 2009, 98, 3013–3030. (12) Brych, S. R.; Gokarn, Y. R.; Hultgen, H.; Stevenson, R. J.; Rajan, R.; Matsumura, M. J. Pharm. Sci. 2010, 99, 764–781. (13) Lacy, E. R.; Baker, M.; Brigham-Burke, M. Anal. Biochem. 2008, 382, 66– 68. (14) Thies, M. J.; Talamo, F.; Mayer, M.; Bell, S.; Ruoppolo, M.; Marino, G.; Buchner, J. J. Mol. Biol. 2002, 319, 1267–1277. (15) McAuley, A.; Jacob, J.; Kolvenbach, C. G.; Westland, K.; Lee, H. J.; Brych, S. R.; Rehder, D.; Kleemann, G. R.; Brems, D. N.; Matsumura, M. Protein Sci. 2008, 17, 95–106. (16) Harris, R. J. Dev. Biol. (Basel) 2005, 122, 117–127. (17) Gillies, S. D.; Wesolowski, J. S. Hum. Antibodies Hybridomas 1990, 1, 47– 54. (18) Dorai, H.; Wesolowski, J. S.; Gillies, S. D. Mol. Immunol. 1992, 29, 1487– 1491.

5220

Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

CDC.19-22 However, the role of inter chain and intra chain disulfide bonds in the reduction of effector function was not completely understood based on partial reduction experiments. Some studies showed that the decreased complement activity was due to the reduction of inter heavy chain disulfide bonds,19,22 while others showed that reduction of inter chain disulfide bonds may not be critical for the loss of complement activity.19,21,23 One of the major reasons for the uncertainty is the lack of understanding of the susceptibility of different disulfide bonds under various conditions. Relative susceptibility of different disulfide bonds in IgG has been a subject of many studies. It has been reported that inter chain disulfide bonds of human IgG and rabbit IgG were likely reduced before intra chain disulfide bonds under native conditions.24 However, there is no general agreement on the relative susceptibility of different inter chain disulfide bonds. One study suggested that the four inter chain disulfide bonds in human IgG1κ were close to random reduction.25 In contrast, it has been reported that the first disulfide bond of the two inter heavy chain disulfide bonds was reactive with 5,5′-dithio (2,2′-dinitro)benzoate (DTNB),26,27 suggesting that the first disulfide bond was more exposed than the second disulfide bond making it more susceptible. The observation that disulfide-linked human IgG-albumin complex formed through cysteine residues in the hinge region suggested that disulfide bonds in the hinge region are exposed and reactive toward either reduction or disulfide/free sulfhydryl exchange.28 Studies of different classes of human IgG revealed that reduction intermediates composed of two heavy chains and one light chain (HHL), two heavy chains (HH), and one heavy chain and one light chain (HL) linked by disulfide bonds, were commonly observed for IgG1, IgG2, and IgG3, while one heavy chain and one light chain (HL) was almost the only intermediate of IgG4.29 The appearance of the intermediate composed of two heavy chains and one light chain (HHL) was much faster than the appearance of the intermediate composed of one heavy chain and one light chain (HL),30 because of either higher susceptibility of the disulfide bonds between the light chain and the heavy chain or the fact that reduction of only one disulfide bond results in the formation of HHL intermediate, while reduction of two disulfide bonds was necessary to form the HL intermediate. The major intermediates of IgG1κ after partial reduction were two heavy chains (HH), and one heavy chain and one light chain (HL), while two heavy chains (HH) was the major intermediate for IgG1λ suggesting similar susceptibility of all three inter chain disulfide bonds in IgG1κ and higher susceptibility of the inter light chain (19) (20) (21) (22) (23) (24) (25) (26) (27)

(28) (29) (30)

Press, E. M. Biochem. J. 1975, 149, 285–288. Wright, J. K. Biochem. Biophys. Res. Commun. 1978, 83, 1284–1290. Johnson, B. A.; Hoffmann, L. G. Mol. Immunol. 1981, 18, 181–188. Isenman, D. E.; Dorrington, K. J.; Painter, R. H. J. Immunol. 1975, 114, 1726–1729. Johnson, B. A.; Hoffmann, L. G. Mol. Immunol. 1984, 21, 77–87. Cecil, R.; Stevenson, G. T. Biochem. J. 1965, 97, 569–572. Sears, D. W.; Mohrer, J.; Beychok, S. Biochemistry 1977, 16, 2031–2035. Schauenstein, E.; Dachs, F.; Reiter, M.; Gombotz, H.; List, W. Int. Arch. Allergy Immunol. 1986, 80, 174–179. Schauenstein, E.; Schauenstein, K.; Dachs, F.; Reiter, M.; Leitsberger, A.; Weblacher, M.; Maninger, K.; Horejsi, H.; Steinschifter, W.; Hirschmann, C.; Felsner, P. Biochem. Mol. Biol. Int. 1996, 40, 433–446. Mohammad, S. F.; Sharma, N.; Woodward, S. C. Biochim. Biophys. Acta 1983, 749, 47–51. Virella, G.; Parkhouse, R. M. Immunochemistry 1973, 10, 213–217. Hong, J.; Lee, A.; Han, H.; Kim, J. Anal. Biochem. 2009, 384, 368–370.

and heavy chain disulfide bond than the two inter heavy chain disulfide bonds in IgG1λ.31 Susceptibility of the disulfide bonds between the light chain and heavy chain can be compared to the susceptibility of the two inter heavy chain disulfide bonds based on the appearance and disappearance of different intermediates monitored by gel electrophoresis. However, monitoring reduction intermediates by gel electrophoresis cannot provide any information regarding the relative susceptibility of the two inter heavy chain disulfide bonds and intra chain disulfide bonds. A new method is therefore required to obtain a complete ranking order of the disulfide bond susceptibility in IgG molecules. A method including partial reduction, alkylation with 12Ciodoacetic acid, complete reduction followed by alkylation with 13 C-iodoacetic acid, and liquid chromatography-mass spectrometry (LC-MS) analysis was developed in the current study. A similar differential labeling strategy was employed previously to determine the levels of free sulfhydryl and domain-level stability of recombinant monoclonal antibodies.2,32 This method allowed an accurate calculation of the percentage of the cysteine residues that was modified by either 12C-iodoacetic acid or 13C-iodoacetic acid based on LC-MS analysis. The percentage of modification of each cysteine residue by 12Ciodoacetic acid correlated directly with the susceptibility of the disulfide bonds to reduction. A complete ranking order of the disulfide bonds of two recombinant monoclonal antibodies was obtained. Furthermore, a difference in the susceptibility of disulfide bonds in the Fab regions of IgG1κ and IgG1λ was observed for the first time. This method could be easily employed to study the relative susceptibility of disulfide bonds in other proteins. MATERIALS AND METHODS Materials. The recombinant monoclonal antibodies were produced by transfected Chinese hamster ovary (CHO) cell lines and purified at Abbott Bioresearch center (Worcester, MA). Human IgG1κ and human IgG1λ antibodies were purchased from SouthernBiotech (Birmingham, AL). Iodoacetic acid (referred to as 12C-iodoacetic acid) and dithiothreitol (DTT) were purchased from Sigma (St. Louis, MO). Iodoacetic acid with a 13C (referred to 13C-iodoacetic acid, 99% enrichment) was purchased from Cambridge Isotope Laboratories (Andover, MA). Acetonitrile, trifluoroacetic acid (TFA), and guanidine hydrochloride were purchased from J. T. Baker (Phillipsburg, NJ). Formic acid was purchased from EMD (Gibbstown, NJ). Trypsin was purchased from Worthington (Lakewood, NJ). Lys-C was purchased from Roche (Indianapolis, IN). Reduction and Alkylation under Native Condition. Aliquots of the recombinant monoclonal antibodies at 1 mg/mL in 100 mM Tris, pH 8.0 were partially reduced by DTT at concentrations of 0, 0.1, 0.2, 0.5, or 5 mM at room temperature for 10 min. The samples were then alkylated by the addition of 12C-iodoacetic acid to a final concentration of 25 mM and incubated at 37 °C for 30 min. Excess reagents were removed by buffer exchange of the samples into 100 mM Tris, pH 8 using NAP-10 columns (GE healthcare, Piscataway, NJ). Then, 250 µL of each sample (31) Montano, R. F.; Morrison, S. L. J. Immunol. 2002, 168, 224–231. (32) Liu, H.; Chumsae, C.; Gaza-Bulseco, G.; Goedken, E. R. Anal. Biochem. 2010, 400, 244–250.

was mixed with 750 µL of 8 M guanidine hydrochloride in 100 mM Tris, pH 8.0, reduced with 10 mM DTT at 37 °C for 30 min and followed by alkylation with 25 mM 13C-iodoacetic acid at 37 °C for 30 min. The samples were buffer exchanged into 10 mM Tris, pH 8.0. Lys-C was added to each sample to a final Lys-C: antibody ratio of 1:20 (w:w). Digestion was carried out at 37 °C for 4 h. The resulting peptides were analyzed by LC-MS. Reduction and Alkylation in the Presence of Guanidine Hydrochloride. Intra chain disulfide bonds are not susceptible under native conditions, as will be discussed later; thus the antibodies were reduced in the presence of various concentrations of guanidine hydrochloride to expose disulfide bonds. The antibody samples at 1 mg/mL in 100 mM Tris, pH 8.0 with 0, 1, 1.2, 1.4, 1.6, 1.8, 2, 2.2, 2.5, 2.6, 2.7, 2.8, 3, 3.5, 4, 5, and 6 M guanidine hydrochloride were incubated at 25 °C for 18 h. The samples were then reduced with 5 mM DTT at room temperature for 10 min, which was followed by the addition of 12C-iodoacetic acid to a final concentration of 25 mM and incubation at 37 °C for 30 min. The samples were buffer exchanged to 100 mM Tris, pH 8.0 using NAP-10 columns to remove excess 12Ciodoacetic acid. Aliquots of each sample were reduced with 10 mM DTT in the presence of 6 M guanidine hydrochloride at 37 °C for 30 min and then alkylated using 25 mM 13C-iodoacetic acid at 37 °C for 30 min. After buffer exchange into 10 mM Tris, pH 8.0, the samples were digested using trypsin at a 1:20 trypsin/antibody ratio and digestion was allowed to proceed at 37 °C for 4 h. The resulting peptides were analyzed by LC-MS. Co-Incubation of Different Antibodies. Experiments were also carried out by co-incubation of different antibodies. The two recombinant monoclonal antibodies or the two human myeloma proteins (0.5 mg/mL of each) were incubated in 100 mM Tris, pH 8.0, in the presence of 2.2 M guanidine hydrochloride at 25 °C for 18 h. The samples were then reduced, alkylated, and digested by following the procedure as described in the previous section. Tryptic peptides were analyzed by LC-MS. LC-MS. LC-MS analysis of the samples was performed using an Agilent 1200 HPLC (Agilent, Santa Clara, CA), a Proto 200 C18 column (250 mm ×1 mm i.d., 5 µm particle size, Higgins Analytical, Inc. Mountain View, CA) and an Agilent 6510 Q-TOF mass spectrometer (Agilent). Approximately 10 µg of each sample was loaded at 98% mobile phase A (0.02% TFA, 0.08% FA in Milli-Q water) and 2% mobile phase B (0.02% TFA, 0.08% FA in acetonitrile). Peptides were eluted from the column using a gradient of increasing mobile phase B from 2 to 55% in 90 min. IonSpray voltage was set at 4200 V. Source temperature was set at 75 °C and m/z was scanned from 250 to 2000. RESULTS AND DISCUSSION Principle of the Method. The domain structures and the locations of inter chain and intra chain disulfide bonds of a typical IgG1 antibody are shown in Figure 1. The cysteine residue of the light chain that connects to the heavy chain is located at the C-terminus of the light chain in IgG1κ. However, the cysteine residue that forms the inter light chain and heavy chain disulfide bond is followed by a serine residue in IgG1λ. Relative susceptibility of disulfide bonds of antibodies was determined by partial reduction, differential alkylation, and Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

5221

LC-MS. Partial reduction was employed to probe the relative susceptibilities of different disulfide bonds. If there was a difference in the susceptibilities among the different disulfide bonds, they would have been reduced to different degrees. On the other hand, if different disulfide bonds were equally susceptible, they would have been reduced to the same degree. The degree of partial reduction of each disulfide bond was determined by differential alkylation and LC-MS analysis. Free sulfhydryl groups from the partial reduction were alkylated using 12C-iodoacetic acid. After removal of the excess reagents, the remaining disulfide bonds were completely reduced and alkylated using 13 C-iodoacetic acid. Alkylation of cysteine residues as a result of partial reduction by 12C-iodoacetic acid was determined by LC-MS analysis of cysteine containing peptides. Susceptibility of Inter Chain Disulfide Bonds. Relative susceptibility of inter chain disulfide bonds was assessed by partial reduction under native conditions. The samples were digested with Lys-C after reduction and alkylation. Mass spectra of the peptides containing the cysteine residue in the light chain and the cysteine residue in the heavy chain that form the inter light chain and heavy chain disulfide bond of the IgG1κ antibody are shown in Figure 2. Alkylation of this peptide by 12C-iodoacetic acid should result in a calculated monoisotopic molecular weight (MH+) of 870.3 Da. Alkylation of this peptide by 13C-iodoacetic acid should result in a molecular weight of 871.3 Da. Therefore, m/z 871.3 (Figure 2, 0 mM) corresponded to the peptide with cysteine residue alkylated by 13C-iodoacetic acid. Trace level (1.6%) of the peak with m/z 870.3 was also observed in this mass spectrum. The first source of this peak was due to the presence of 12C-iodoacetic acid in this 99% enriched 13C-iodoacetic acid reagent. This peak was consistently detected when the same antibody was completely reduced and alkylated using only 13Ciodoacetic acid (see Supporting Information. Figure 1). The second source of this peak was due to the presence of low level of free sulfhydryl in the IgG molecules.1,2 The intensity of the peak with m/z 870.3 Da increased with the increase of DTT concentrations as shown in Figure 2 (0.1 to 5 mM). This data suggested that higher concentrations of DTT led to higher levels of reduction of the disulfide bonds between the light chain and the heavy chain. A similar trend was observed when the peptide containing the cysteine residue that is linked to the light chain was analyzed (see Supporting Information, Figure 2). The percentage of reduced cysteine was calculated using the two series of isotopic peak distributions resulting from alkylation by either 12C-iodoacetic acid or 13C-iodoacetic acid. Mass spectra of the light chain peptide were used to illustrate the calculation procedure. The peak with m/z 870.3 represented alkylation of the peptide by 12C-iodoacetic acid subsequent to partial reduction. The peak with m/z 871.3 Da represented alkylation of the same peptide with 13C-iodoacetic acid, and it overlapped with the second peak of the m/z 870.3 isotopic series. Subtracting the overlap from the peak of m/z 871.3 should result in the intensity that represented modification of this peptide by 13Ciodoacetic acid only. The percentage of the peptide that was alkylated by 12C-iodoacetic acid was therefore equal to the intensity of the peak of m/z 870.3 divided by the sum of peak intensities of m/z 870.3 and 871.3 after subtraction of the overlap. The overlap was calculated by multiplying the peak 5222

Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

Figure 2. Mass spectra of the peptide containing the light chain cysteine residue that connects to the heavy chain. The IgG1κ antibody was partially reduced with 0, 0.1, 0.2, 0.5, and 5 mM DTT as labeled in the figure.

intensity at m/z 870.3 with the ratio of m/z 872.3 over the m/z 871.3 (Figure 2, 0 mM). The ratio of m/z 872.3 over 871.3 should be close to the natural isotopic distribution as alkylation of this peptide with 13C-iodoacetic acid only introduced 1 carbon-13 atom to the peptide, which has 32 carbon atoms. The ratio of m/z 871.3 over m/z 870.3, in Figure 2 5 mM, can also be used if complete alkylation of this peptide by 12C-iodoacetic acid was ensured. The percentage of alkylation of the cysteine residue in the heavy chain peptide was also calculated using the same procedure. The percentage of alkylation by 12C-iodoacetic acid calculated from the two peptides were very similar (data not shown), suggesting that the two cysteine residues maintained similar levels of exposure after partial reduction of the disulfide bond. The percentage of reduction of this disulfide bond was thus presented as the average calculated from the two peptides. As shown in Table 1, higher percentage of the disulfide bond between the light chain and heavy chain was reduced with increased concentrations of DTT. In addition, disulfide bonds between the light chain and heavy chain of IgG1κ were less susceptible than disulfide bonds of the same locations in IgG1λ.

Table 1. Percent Reduction of the Inter Chain Disulfide Bondsa IgG1κ

IgG1λ

DTT (mM)

LC-HC

upper HC-HC

lower HC-HC

LC-HC

upper HC-HC

lower HC-HC

0 0.1 0.2 0.5 5.0

1.5 22.2 38.0 61.5 94.2

2.9 9.0 17.6 39.9 97.6

2.2 5.5 11.7 33.4 93.8

2.6 41.0 64.9 86.3 98.1

2.8 7.9 13.6 38.9 99.9

2.6 4.5 9.5 30.4 87.4

a The data is an average of triplicate experiments with CV% less than 10%.

Figure 3. MS/MS spectrum of the heavy chain hinge region peptide. The spectrum was acquired using Lys-C digested sample of the IgG1κ antibody that was incubated without DTT during the partial reduction step. The amino acid sequences and molecular weights of the major fragment ions are also shown.

Mass spectra of the peptide containing the two cysteine residues that connect the two heavy chains are complicated (see Supporting Information, Figure 3). The peak with m/z 949.8 corresponded to alkylation of the first or the second cysteine residues by 12C-iodoacetic acid, and thus the percentage of alkylation of each cysteine residue by 12C-iodoacetic acid cannot be calculated using the full scan spectra. Instead, the MS/MS spectra were used to calculate the percentage of reduction of each disulfide bond in this peptide. A typical MS/MS spectrum of this peptide that was acquired using the sample that was fully reduced and alkylated by 13C-iodoacetic acid is shown in Figure 3. The sequence of the peptide along with the B ions and Y ions are also shown in Figure 3. B4 and Y22, which are two of the major fragment ions, containing the first cysteine and the second cysteine residues respectively, were used for calculation. Mass spectra of B4 and Y22 ions from the samples that were reduced with different amounts of DTT are shown in Figure 4. The peak with m/z 502.2 corresponded to the fragment with the first cysteine modified by 13C-iodoacetic acid. The peak with m/z 501.2 corresponded to the same fragment ion but alkylated by 12 C-iodoacetic acid, and its intensity increased in the samples that were reduced with higher concentrations of DTT. The peak with m/z 1174.1 corresponded to the Y22 fragment with the second cysteine alkylated by 13C-iodoacetic acid. The peak with m/z 1173.6 corresponded to the Y22 fragment ion that was

alkylated by 12C-iodoacetic acid. By following the same calculation procedure as described previously, the percentage of the cysteine residues that were alkylated by 12C-iodoacetic acid was calculated, which represented the percentage of reduction of each disulfide bond with different amounts of DTT. The data is shown in Table 1. The percentage of reduction increased with increasing amounts of DTT concentration, as expected. Interestingly, the upper disulfide bond was more susceptible to reduction than the lower disulfide bond. Several conclusions can be drawn by comparison of the susceptibilities of the disulfide bonds in the hinge regions. First, inter chain disulfide bonds between light chain and heavy chain are more susceptible than disulfide bonds between the two heavy chains in both IgGκ and IgG1λ. Second, susceptibility difference between the two inter light chain heavy chain disulfide bonds and the two inter heavy chain disulfide bonds was smaller in IgGκ than in IgG1λ. This observation was in agreement with a previous study,31 where a half-IgG molecule (HL) was a major reduction intermediate for IgGκ, while disulfide bond linked two heavy chains (HH) was the major intermediate for IgG1λ. Third, the upper inter heavy chain disulfide bonds in IgGκ and IgG1λ shared similar susceptibility, and the same was true for the lower inter heavy chain disulfide bond. Lastly, the first disulfide bond of the two inter heavy chain disulfide bonds was more susceptible than the second disulfide bond. The lower susceptibility of the second disulfide bond can be due to lower degree of exposure to solvents. It was also possible that reduction of the first disulfide bond triggered a conformational change in the hinge region, which resulted in greater exposure of the second disulfide bond. It is worthwhile to mention that none of the intra chain disulfides were reduced under the native condition. Susceptibility of Intra Chain Disulfide Bonds. As discussed in the previous section, intra chain disulfide bonds were not accessible to reduction under native conditions. This result was expected because intra chain disulfide bonds are buried between the two layers of β-sheet structures.33 To determine the relative susceptibility of the intra chain disulfide bonds, the recombinant monoclonal antibodies were incubated with increasing concentrations of guanidine hydrochloride. After partial reduction and alkylation with 12C-iodoacetic acid, the samples were completely reduced, alkylated with 13C-iodoacetic acid, and digested with trypsin. Peptides containing cysteine residues that are involved in intra chain disulfide bonds were analyzed. The tryptic peptide that contained the second cysteine residue in the CH1 domain was not included in the analysis because it had a molecular weight greater than 6 kDa, and the m/z corresponding to the monoisotopic molecular weight was too low for an accurate calculation. The peptide containing the second cysteine residue in the CH2 domain was also not included in the analysis because it has only two amino acids and was not retained by the C18 column. A representative set of mass spectra of the tryptic peptide containing the first cysteine residue in the CL domain is shown in Figure 5. The peak with m/z 900.4 corresponded to the peptide with the cysteine residue alkylated by 13C-iodoacetic acid. The peak with m/z 899.9 corresponded to the peptide with the cysteine residue alkylated by 12C-iodoacetic acid. The peak (33) Padlan, E. A. Mol. Immunol. 1994, 31, 169–217.

Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

5223

Figure 4. Mass spectra of B4 and Y22 ions from the high region peptide containing the two cysteine residues that connect the two heavy chains in the IgG1κ antibody. The antibody was partially reduced with DTT concentrations of 0, 0.1, 0.2, 0.5, and 5 mM.

intensity of the m/z 899.9 increased with the increase of guanidine hydrochloride concentrations suggesting a higher level of the peptide with the cysteine residue modified by 12Ciodoacetic acid. The percentage of alkylation of each cysteine residue by 12C-iodoacetic acid was calculated using the isotopic peak distribution as described in the previous sections. The percentage of alkylation of different peptides containing cysteine residues that are involved in the same disulfide bonds were similar except the peptides containing the two cysteine residues in the CH3 domain (data not shown). The similarity in the levels of alkylation by 12C-iodoacetic acid suggested that the exposure levels of the two cysteine residues were similar after partial reduction. Furthermore, similar levels of alkylation of cysteine residues that are involved in the same disulfide bond suggested that the primary sequence difference around the cysteine residues had minimal impact on the accessibility of the cysteine residues. The levels of alkylation of the two cysteine residues in the CH3 domain by 12C-iodoacetic acid were slightly different. The second cysteine residue in the CH3 domain was modified to only approximately 90% even after denaturation with 6 M guanidine hydrochloride. The reason for the incomplete alkylation of this cysteine residue was unknown, and it was not due to the lack of reagents as complete 5224

Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

alkylation of other peptides were observed in the samples that were denatured with higher amounts of guanidine hydrochloride. It was possible that there was a fast local conformational change around this cysteine residue after partial reduction, which decreased its exposure to solvents. Despite the slight difference in the percentage of alkylation of the two cysteine residues in the CH3 domain, the percentage of alkylation of cysteine residues involving the same disulfide bonds was averaged to represent the percentage of reduction of the corresponding disulfide bonds. The percentage of reduction of each intra chain disulfide bond is shown in Figure 6. Guanidine hydrochloride concentrations that resulted in 50% of each cysteine residue that was alkylated by 12 C-iodoacetic acid was calculated using a two-state unfolding model,34 which were used as an indicator of the susceptibility of the disulfide bonds (Table 2). The level of reduction for each disulfide bond in IgG1κ and IgG1λ increased with the increase in guanidine hydrochloride concentrations, suggesting increased level of exposure of disulfide bonds. Disulfide bonds in the CH2 domain were the most susceptible to reduction, while disulfide bonds in the CH3 domain were the least susceptible to reduction (34) Bolen, D. W.; Santoro, M. M. Biochemistry 1988, 27, 8069–8074.

Table 2. Guanidine Hydrochloride Concentrations Where 50% of Disulfide Bond Reduction Was Observeda domain

IgG1κ

IgG1λ

VL CL VH CH1 CH2 CH3

2.6 2.6 2.6 2.6 1.4 3.4

2.3 2.2 2.2 2.2 1.4 3.3

a The data is an average of triplicate experiments with CV% less than 10%.

Figure 5. Representative mass spectra of the tryptic peptide containing the first cysteine residue in the CL domain of the IgG1κ antibody after incubation with 0, 2.2, 2.5, 2.6, and 2.8 M guanidine hydrochloride as labeled.

Figure 6. Plots showing the percentage of reduction of intra chain disulfide bonds by partial reduction in the presence of various concentrations of guanidine hydrochloride.

in both IgG1κ and IgG1λ. Disulfide bonds in other domains showed moderate level of susceptibility in both IgG1κ and IgG1λ.

As expected, susceptibility of disulfide bonds in the CH2 domain and CH3 domain of the IgG1κ was comparable to the susceptibility of disulfide bonds in the same domains of the IgG1λ antibody. However, disulfide bonds in VL, CL, VH, and CH1 domains in the IgG1κ antibody were less susceptible than disulfide bonds in these respective domains in the IgG1λ antibody. The two antibodies shared the same sequences in CH2 and CH3 domains, while the sequences in other domains were different. The similarities and differences in the disulfide bond susceptibility between the two antibodies may reflect the difference in their folding, which is generally determined by the primary sequence. Effect of the Type of Light Chains on Intra Chain Disulfide Bond Susceptibility. Significant differences in the susceptibilities of disulfide bonds in the Fab region (VL, CL, VH, and CH1) were observed between the recombinant IgG1κ and the recombinant IgG1λ antibodies. To further confirm this observation, the two recombinant monoclonal antibodies were co-incubated in the presence of 2.2 M guanidine hydrochloride and then analyzed by following the same procedure. Molecular weights of the peptides containing cysteine residues in VH, VL, and CL domains were different between the two recombinant monoclonal antibodies and thus included in this analysis. In agreement with the previous observation, an average of 6.4% of the disulfide bonds in VH, VL, and CL was reduced in the recombinant IgG1κ antibody, while 37.4% of the disulfide bonds in these domains were reduced in the IgG1λ antibody. Two wild type antibodies from myeloma patients, one IgG1κ and one IgG1λ, were also analyzed by following the co-incubation procedure. Only peptides containing the cysteine residues in the CL domain were analyzed because the amino acid sequences in the variable domains were unknown, and the cysteine containing peptides from IgG1κ and IgG1λ in other domains had the same conserved amino acid sequences. A similar difference between the percentages of reduction of the disulfide bond was observed. Therefore, the difference in the susceptibility of the disulfide bonds in the Fab regions between IgG1κ and IgG1λ was probably a universal structural difference between the two types of IgG1 antibodies. CONCLUSIONS A method that can accurately rank the order of susceptibility of disulfide bonds was proposed and employed to study the susceptibility of disulfide bonds of IgG1 molecules. There were similarities and differences between IgG1κ and IgG1λ molecules in regard to disulfide bond susceptibility. The similarities are as follows: (1) inter chain disulfide bonds were more susceptible to Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

5225

reduction than intra chain disulfide bonds. Inter chain disulfide bonds were reduced under native condition, while intra chain disulfide bonds were only reduced in the presence of guanidine hydrochloride; (2) the inter light chain and heavy chain disulfide bonds were more susceptible to reduction than the two inter heavy chain disulfide bonds; (3) the upper disulfide bond of the two inter heavy chain disulfide bonds was more susceptible than the lower one; and (4) disulfide bonds in the CH2 domain were the most susceptible disulfide bonds, while disulfide bonds in the CH3 domain were least susceptible disulfide bonds. Disulfide bonds in other domains had similar and medium susceptibility. Two major differences were observed between IgG1κ and IgG1λ. First, the difference between the susceptibility of inter light chain and heavy chain disulfide bonds and the susceptibility of the two inter

5226

Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

heavy chain disulfide bonds was smaller in IgG1κ compared to that in IgG1λ. Second, for the first time, it was observed that disulfide bonds in the Fab regions of IgG1κ were less susceptible to reduction than disulfide bonds in the Fab regions of IgG1λ. This was true for recombinant as well as wild type IgG1 molecules. SUPPORTING INFORMATION AVAILABLE Additional information as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review March 3, 2010. Accepted May 8, 2010. AC100575N