Anal. Chem. 1999, 71, 2390-2397
Capillary Electrophoresis Sodium Dodecyl Sulfate Nongel Sieving Analysis of a Therapeutic Recombinant Monoclonal Antibody: A Biotechnology Perspective Glenn Hunt and Wassim Nashabeh*
Department of Quality Control Biochemistry, Genentech, Inc., 1 DNA Way, South San Francisco, California 94080
With the increasing interest in the therapeutic use of recombinant monoclonal antibodies (rMAbs), a generic analytical approach for the analysis of size-based rMAb variants is desired. Such a method using capillary electrophoresis (CE) with laser-induced fluorescence detection is described. The assay was developed as a replacement for silver-stained SDS-PAGE and was validated according to the guidelines of the International Committee on Harmonization for use in routine lot release testing of a rMAb pharmaceutical. In this assay, the rMAb solution is first derivatized with a neutral fluorophore, e.g., 5-carboxytetramethylrhodamine succinimidyl ester. The labeled sample is then incubated with SDS, and the SDSprotein complexes are then separated by CE using a hydrophilic polymer as a sieving matrix. The precolumn labeling conditions described in this study allowed the detection of rMAb at a low-nanomolar concentration (9 ng/mL), with no apparent loss in resolution or changes to the distribution of rMAb analyte species, when compared to an unlabeled sample. In addition, the traditional practice of heating proteins at elevated temperatures in the presence of SDS to facilitate SDS-protein binding resulted in the generation of significant levels of rMAb fragmentation, and alternative conditions to minimize this artifact are discussed. Illustrations of the uses of this assay in monitoring consistency of bulk manufacture of a protein pharmaceutical, and in providing a size-based separation of product-related variants, as well as nonproduct impurities are shown. In brief, the assay described in this paper demonstrated comparable resolution and sensitivity to silver-stained SDS-PAGE but offered the advantages of enhanced precision and robustness, speed, ease of use, and on-line detection. The manufacture and quality control of protein products from recombinant DNA technology have required the development of state-of-the-art analytical methodologies for the elucidation of their structure and assurance of their homogeneity, purity, and molecular stability.1 This has been accomplished with an array of analytical methodologies that are complementary in the evaluation * Corresponding author: (e-mail)
[email protected]. (1) Garnick, R. L.; Soli, N. J.; Papa, P. A. Anal. Chem. 1988, 60, 2546-2557.
2390 Analytical Chemistry, Vol. 71, No. 13, July 1, 1999
of the physicochemical attributes of proteins in terms of charge, hydrophobicity, or size. Among the various size-based separation methods, size-exclusion chromatography (SEC) provides accurate and reliable separations of proteins in their aggregate or monomeric forms but is generally limited in its ability to separate lower molecular weight fragments. On the other hand, sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) typically provides superior separations of a denatured protein in its aggregate, monomeric, and fragment forms in the molecular mass range of 10-200 kDa, and as such has been used for over 25 years as the primary method of choice for size-based protein separations.2,3 In SDS-PAGE, advantage is taken of the fact that proteins tend to bind a relatively constant amount of SDS on a weight basis4 (∼1.4 g of SDS/1 g of protein), resulting in SDSprotein complexes with practically identical free solution mobilities.5 In a sieving medium, the mobility of the complex is thus proportional to the log of the effective molecular radius and, hence, the log molecular weight of the polypeptide backbone chain.6 Besides being a tool to estimate the apparent molecular weight of proteins, SDS-PAGE is used to confirm the consistency of manufacture of biologics. Depending upon the intended use of SDS-PAGE analyses, visualization of the separated proteins is generally accomplished by staining with either Coomassie Brilliant Blue7 or the more sensitive silver stain8 dyes. While Coomassie Brilliant Blue is used for semiquantitative evaluation of the relative intensity of the major protein bands, its limit of detection (∼100fold less sensitive than silver stain)9 is considered inadequate for low-level impurity detection. Use of the silver stain method significantly improves the concentration limit of detection (cLOD) to 10 ng/mL as determined by analysis of bovine serum albumin (2) Weber, K.; Osborn, M. J. Biol. Chem. 1969, 244, 4406-4412. (3) Andrews, A. T. Electrophoresis: Theory, Techniques and Biomedical and Clinical Application, 2nd ed.; Clarendon Press: Oxford, 1986. (4) Reynolds, J. A.; Tanford, C.J. Biol. Chem. 1970, 245, 5161-5165. (5) See, Y. P.; Jackowski, G. In Protein Structure: A Practical Approach; Creighton, T. E., Ed.; IRL Press: Oxford, 1989; pp 1-21. (6) Shapiro, A. L.; Vinuela, E.; Maizel. J. V. Biochem. Biophys. Res. Commun. 1967, 28, 815-820. (7) Chrambach, A.; Reisfeld, R. A.; Wyckoff, M.; Zaccari, J. Anal. Biochem. 1967, 20, 150-15. (8) Oakley, B. R.; Kirsch, D. R.; Morris, N. R. Anal. Biochem. 1980, 105, 361363. (9) Jorgenson, J. W. Anal. Chem. 1986, 58, 743A. 10.1021/ac981209m CCC: $18.00
© 1999 American Chemical Society Published on Web 05/22/1999
(BSA) as a sensitivity marker.10 Despite these advantages, SDSPAGE suffers from several limitations. It is labor intensive (gel preparation, separation, staining, destaining), requires use of toxic reagents, and there is high intra- and intergel effective mobility and staining variability. Over the past few years, there has been increasing activity in evaluating capillary electrophoresis (CE) as an automated and instrumental approach to classical electrophoresis.11-13 The initial work on CE separations with sieving matrixes were performed by Hjerten14 and Karger, who reported the use of cross-linked polyacrylamide gel-filled capillaries.15,16 Later work focused on the introduction of replaceable linear (non-cross-linked) polymer networks that were shown to be appropriate in resolving SDSprotein complexes on the basis of their size.17-27 For further details on the principles and applications of this technique, refer to two recent review articles by Guttman28 and Takagi.29 Though the transition from gel to polymer networks is evident in recent literature, the term “capillary gel electrophoresis” is still widely used. Other terms such as “SDS-polymer solution”, “polymer network”, or “nongel sieving” have also been used. In the absence of a consensus on the naming of this technique, we are using the term capillary electrophoresis sodium dodecyl sulfate nongel sieving (CE-SDS-NGS), to denote CE separations of SDS-protein complexes in the presence of a replaceable sieving matrix. Despite the extensive documentation on the efficiency and ease of use of CE-SDS-NGS, the technique has found limited applications in the biopharmaceutical industry,30-32 as SDS-PAGE remains the method of choice. While the difficulties associated with the introduction of any new methodology into a highly regulated environment undoubtedly plays a role, the lack of sensitivity of CE-SDS-NGS using absorption detection is a major (10) Merril, C. R.; Switzer, R. C.; Van Keuren, M. L. Proc. Natl. Acad. Sci. U.S.A. 1979, 76, 4335-4339. (11) Jorgenson, J. W.; Lukacs, K. D. Science 1983, 222, 266-272. (12) Karger, B. L.; Cohen, A. S.; Guttman, A. J. Chromatogr. 1989, 492, 585613. (13) Landers, J. P., Ed. CRC Handbook of Capillary Electrophoresis: Principles, Methods and Applications; CRC Press: Boca Raton, 1993. (14) Hjerten, S. In Electrophoresis ’83; Hirai, H., Ed.; Walter de Gruyter Press: New York, 1984; pp 71-79. (15) Cohen, A. S.; Karger, B. L. J. Chromatogr. 1987, 397, 409-417. (16) Karger, B. L.; Paulus, A.; Cohen, A. S. Chromatographia 1987, 24, 15-22. (17) Zhu, M.; Hanson, D. L.; Burd, S.; Garrison, F. J. Chromatogr. 1989, 480, 311-319. (18) Widhalm, A.; Schwer, C.; Blaas, D.; Kenndler, E. J. Chromatogr. 1991, 549, 446-451. (19) Wu, D.; Regnier, F. E. J. Chromatogr. 1992, 608, 349-356. (20) Ganzler, K.; Greve, K. S.; Cohen, A. S.; Karger, B. L.; Guttman, A.; Cooke, N. C. Anal. Chem. 1992, 64, 2665-2671. (21) Nakatami, M.; Shibukawa, A.; Nakagawa, T.; Biol. Pharm. Bull. 1993, 16, 1185-1188. (22) Werner, W.; Demorest, D.; Wiktorowicz, J. E. Electrophoresis 1993, 14, 759-763. (23) Karim, M. R., Janson, J. C.; Takagi, T. Electrophoresis 1994, 15, 15311534. (24) Nakatami, M.; Shibukawa, A.; Nakagawa, T. J. Chromatogr., A 1994, 672, 213-218. (25) Guttman, A.; Nolan, J.; Cooke, N. J. Chromatogr. 1993, 632, 171-175. (26) Guttman, A.; Nolan, J. Anal. Biochem. 1994, 221, 285-289. (27) Alfonso, E. S.; Conti, M.; Gelfi, C.; Righetti, P. G. J. Chromatogr., A 1995, 689, 85-96. (28) Guttman, A. Electrophoresis 1996, 17, 1333-1341. (29) Takagi, T. Electrophoresis 1997, 18, 2239-2242. (30) Liu, J.; Abid, S.; Lee, M. S. Anal. Biochem. 1995, 229, 221-228. (31) Bennedek, K.; Thiede, S. J. Chromatogr., A 1994, 676, 209-217. (32) Hunt, G.; Moorhouse, K. G.; Chen, A. B. J. Chromatogr., A 1996, 744, 295301.
contributing factor to its limited use. One means of enhancing detection sensitivity is the use of laser-induced fluorescence (LIF) detection, a common practice in DNA analysis. This requires precolumn derivatization of the analyte of interest with a fluorophore. Earlier work in this area by Gump and Monnig33 using fluorescamine, naphthalene-2,3-dicarboxyaldehyde, and o-phthaldialdehyde as tagging reagents resulted in enhanced absorption and fluorescence detection. Most importantly, this study showed that derivatization of standard analytes resulted in only a small reduction in separation efficiency and had minimal impact on the accuracy of the molecular mass determinations. These findings were also confirmed by Wise et al.34 using 4-fluoro-7-nitrobenzofurazan as a labeling reagent. In this paper, we show that CE-SDS-NGS using precolumn labeling and LIF detection can be developed and validated for routine analysis of rDNA protein products in the biopharmaceutical industry as a replacement for SDS-PAGE visualized with silver stain. Illustrations of the potential uses of CE-SDS-NGS are demonstrated using humanized recombinant IgG1-monoclonal antibodies, a class of proteins that is by far the most abundant among the various recombinant biopharmaceuticals that have recently received regulatory approval or are in late-stage clinical testing. The use of 5-carboxytetramethylrhodamine succinimidyl ester (5-TAMRA.SE), as a labeling reagent, is shown to substantially enhance the detection sensitivity while maintaining a genuine representation of the analyte species. Optimum temperature conditions for generation of SDS-protein complexes resulting in minimal sample degradation are described. Other aspects of methods development are discussed, as well as the utility of the assay as it relates to detection of mass variants for product purity and lot-to-lot consistency. EXPERIMENTAL SECTION Apparatus. CE-SDS-NGS was performed in fused-silica capillaries (Polymicro Technologies, Phoenix, AZ) of 50 µm i.d. and 375 µm o.d. with a total length of 24 cm and an effective length of 19.4 cm. All analyses were performed using a Bio-Rad BioFocus 3000 capillary electrophoresis system equipped with a LIF detector (Bio-Rad Laboratories, Inc., Hercules, CA). The LIF detector uses a 3.5-mW argon ion laser with an excitation wavelength of 488 nm and an emission band-pass filter of 560 ( 20 nm. UV detection was conducted at 220 nm. Reagents. Recombinant humanized monoclonal antibodies (rMAbs) were produced in transfected Chinese hamster ovary cells (Genentech Inc., So. San Francisco, CA). Recombinant peptide-N-glycosidase F (PNGase F) from Flavobacterium meningosepticum was obtained from Oxford Glycosystems (Bedford, MA). All chemicals used were of analytical reagent grade. SDS sample buffer, SDS running buffer, and 2-mercaptoethanol were purchased from Bio-Rad Laboratories, Inc. 5-Carboxytetramethylrhodamine succinimidyl ester was obtained from Molecular Probes (Eugene, OR). Dithiothreitol (DTT), dimethyl sulfoxide (DMSO), and sodium bicarbonate were purchased from Sigma (St. Louis, MO). NAP-5 columns were purchased from Pharmacia (Piscataway, NJ). (33) Gump, E. L.; Monnig, C. A. J. Chromatogr., A 1995, 715, 167-177. (34) Wise, E. T.; Navjot, S.; Hogan, B. L. J. Chromatogr. A 1996, 746, 109121.
Analytical Chemistry, Vol. 71, No. 13, July 1, 1999
2391
Preparation of SDS-rMAb-Labeled Conjugates. rMAb samples (2.5 mg) were buffer exchanged into 800 µL of 0.1 M sodium bicarbonate, pH 8.3, using a NAP-5 column. Ten microliters of 5-TAMRA.SE (1.4 mg/mL) dissolved in DMSO was then added to 190 µL of rMAb solution and the resultant mixture incubated for 2 h at 30 °C. After incubation, 190 µL of the antibody-dye conjugate was loaded onto a second NAP-5 column and collected in 700 µL of 0.1 M sodium bicarbonate, pH 8.3. Nonreduced SDS-rMAb conjugates were then prepared by mixing equal volumes (100 µL) of the rMAb-dye conjugate and the CE-SDS sample buffer. For reduced SDS-rMAb complexes, 80 µL of antibody-dye conjugate was mixed with 100 µL of CESDS sample buffer and 20 µL of neat 2-mercaptoethanol. Both reduced and nonreduced samples were incubated at 90 °C in a heating block for 3 min, unless otherwise noted in the Results and Discussion section. CE Analysis. Prior to injection, the capillary was rinsed with 0.1 M NaOH, 0.1 M HCl, and SDS running buffer, for 60, 60, and 120 s, respectively. A preinjection water plug (2 s at ∼100 psi) was introduced to prevent injection into the sieving matrix and reduce the risk of differential migration during injection. Samples were injected electrophoretically for 15 s at 10 kV. The CE analysis was conducted in the negative polarity mode (-15 kV, -625 V/cm). Typical current obtained was ∼20 µA. Capillary and sample temperatures were maintained at 20 °C. SDS-PAGE. SDS-PAGE analyses were performed using a FisherBiotech electrophoresis system, model FB500 (Fisher Scientific, Pittsburgh, PA). The acrylamide gels (135 mm × 135 mm × 1 mm) were poured in-house and consisted of 4% stacking gel and a 5-20% gradient resolving gel. Electrophoresis proceeded at 20 mA for ∼3 h. The protein bands were visualized with an Oakley silver stain.8 Nonreduced samples were prepared by adding 126 µL of 8% SDS (w/v) to 500 µL of rMAb (1 mg/mL). Reduced samples were prepared by adding 113 µL of 8% SDS (w/ v) and 13 µL of 2 M DTT to 500 µL of rMAb (1 mg/mL). All samples were heated for 3 min at 90 °C. Aliquots containing 12 and 6 µL of reduced and nonreduced samples, respectively, were analyzed. RESULTS AND DISCUSSION CE-SDS-NGS with UV Detection. A comparison of the results obtained from analysis of a rMAb by both CE-SDS-NGS with UV detection at 220 nm and silver-stained SDS-PAGE is shown in Figure 1A. For samples run under nonreducing conditions, there was generally good agreement between the band profile on the slab gel and the peaks in the CE electropherogram, except for bands 3, 4, and the aggregate, which were undetected in the CE analysis. On the other hand, CE-SDS-NGS run under reducing conditions provides good resolution between the predominant light- and heavy-chain components, but lacks the sensitivity for monitoring the minor species in this sample. Attempts to increase absorption sensitivity through combinations of larger inner diameter capillaries and/or larger injection plugs resulted in limited success at the expense of loss in resolution, mainly around the monomer peak. Under optimum conditions, the cLOD of the monomer peak was ∼500 ng/mL, below the values obtained for silver-stained SDS-PAGE (10 ng/mL) but comparable to what is routinely achieved with Coomassie Blue 2392 Analytical Chemistry, Vol. 71, No. 13, July 1, 1999
Figure 1. (A) CE-SDS-NGS separations of nonreduced and reduced preparations of a therapeutic rMAb. Insets show silverstained SDS-PAGE traces of the same sample preparations: M, monomer; A, aggregate. Peak numbers are identified in Results and Discussion section. Conditions: instrument, Bio-Rad BioFocus 3000 capillary electrophoresis system; buffer, Bio-Rad SDS running buffer; capillary, untreated fused silica, 50 µm i.d. and 375 µm o.d.; effective length, 19.4 cm; injection, 15 s at - 417 V/cm; applied electric field, - 625 V/cm; temperature, 20 °C for capillary and sample compartment; detection, UV at 220 nm. (B) CE-SDS-NGS separations of nonreduced and reduced preparations of a 5-TAMRA.SE-labeled rMAb. All conditions are as in (A), except detection was performed with laser-induced fluorescence using a 3.5-mW argon ion laser, 488nm excitation, 560 ( 20 nm emission. Insets show silver-stained SDS-PAGE traces of unlabeled sample preparations.
staining. CE-SDS-NGS using UV detection, although limited in sensitivity, is a useful replacement of SDS-PAGE visualized with Coomassie Blue for the quantitative evaluation of the major molecular mass variants within a protein preparation. CE-SDS-NGS with LIF Detection. The abundance of the lysine residues in most proteins provides ideal reactive sites for precolumn labeling because their aliphatic -amines are reasonably good nucleophiles above pH 8.0,35 and thus can be reacted with (35) Brinkley, M. Bioconjugate Chem. 1992, 3, 2-13.
Figure 2. Structure of the amine-reactive fluorophore 5-TAMRA.SE and a typical reaction scheme with an aliphatic primary amine.
Figure 3. CE-SDS-NGS separations of nonreduced preparations of 5-TAMRA.SE-labeled rMAb using different SDS-rMAb incubation times at 90 °C. All other conditions are as in Figure 1B.
succinimidyl esters with high selectivity in the pH range 8-9, resulting in the formation of stable amide bonds (see Figure 2).36 Among the various fluorophores that can be readily excited by the argon ion laser (readily available in commercial CE instruments), we selected the 5-isomer succinimidyl ester derivative of carboxytetramethylrhodamine (5-TAMRA.SE; see Figure 2 for structure). This fluorophore has intrinsic photostability and superior spectral properties; in particular, its insensitivity to pH changes between 4 and 10, which is an important advantage over fluoresceins for CE applications.36 To enhance the reliability of the labeling step, particularly for routine analysis in a quality control setting, we developed a general procedure that is applicable for several proteins. It consisted of the following main steps: (a) The protein sample is buffer exchanged into a sodium bicarbonate labeling reaction buffer using a Sephadex G25 column. This removes any amine-containing components in the formulation buffer that would compete with the labeling of the protein; (b) the protein undergoes derivatization with the fluorophore (5-TAMRA.SE) for 0.5-2 h at 30 °C, using an optimum dye-to-protein molar ratio; (c) the protein-fluorophore conjugate is passed through a second Sephadex G25 column both to remove excess (unreacted) dye that may comigrate with earlier eluting peaks in the CE profile and to effectively quench the labeling reaction. This step is particularly important because extended incubation of the excess dye with the sample resulted in protein aggregation and subsequent changes in the electrophoretic profile, notably peak broadening, even when stored at 2-8 °C; and finally (d) the sample is incubated with or without a reducing agent and the SDS-containing buffer prior to CE analysis. Results obtained from analysis of a rMAb under both nonreducing and reducing conditions using CE-SDS-NGS with precolumn labeling are shown in Figure 1B along with comparisons to silver-stained SDS-PAGE analysis of the same sample. The improved sensitivity is clearly illustrated by the presence of the minor peaks, which were previously undetected with UV detection. It is important to note that the separation between the rMAb species is not significantly affected by the addition of the fluorophore, as evident in the identical electrophoretic profiles in
Figure 1A and B (nonlabeled vs labeled samples). With the exception of peak 1, which was detected in the blank run, no dyerelated artifact peaks, multiple peak formation from single analytes, or loss of resolution was observed as a result of the heterogeneous nature of the labeling reaction. In addition, while the CE analysis conditions defined in the Experimental Section were used, which were optimized for resolution rather than maximal sensitivity, the cLOD for the monomer peak was determined to be 9.4 ng/mL, based on analyte serial dilution and a signal-to-noise ratio greater than or equal to 2.5, consistent with the guidelines of the International Conference on Harmonization.37 This compares well with silver-stained SDS-PAGE and is a 140-fold increase over CESDS-NGS using UV detection otherwise run under the same conditions. These data clearly demonstrate the potential of CESDS-NGS with precolumn labeling and LIF detection for the analysis of a recombinant protein therapeutic. CE-SDS-NGS Peak Assignment. With the exception of peak 3, the peaks observed in the nonreduced CE profile were attributed to mass variants of a rMAb preparation. The identity of peaks 2 and 4 was inferred from the reduced CE-SDS-NGS analysis that showed the expected light and heavy chains, respectively. Partial reduction of the protein using 2-mercaptoethylamine, which was expected to reduce the protein selectively along the disulfides in the hinge region, resulted in increased levels of peak 5, thus deemed to be a half antibody molecule (one light and one heavy chain). Supporting data were also obtained using a time course of 90 °C heating for a sample reduced with 2-mercaptoethanol (see Figure 4). Other peaks in the CE profile were tentatively assigned on the basis of a comparison with the bands in SDS-PAGE, whose identities were obtained through sequential Edman degradation. The band in the gel corresponding to peak 6 showed mainly the heavy-chain N-terminal sequence with an apparent molecular weight consistent with a heavy-chain dimer structure. The band corresponding to peak 7 showed heavyand light-chain N-termini at a 2:1 ratio, consistent with an antibody lacking one light chain; the nature of the doublet in peak 7 is unknown. Accuracy and Precision. Two important criteria for the success of an assay in monitoring consistency and purity are
(36) Haugland, R. P. Handbook of Fluorescent Probes and Research Chemicals; Molecular Probes, Inc., Eugene, OR, 1996; pp 8-12.
(37) International Conference on Harmonization: Guideline on the validation of analytical procedures: methodology. Fed. Regist. 1997, 62 (96), 27464-7.
Analytical Chemistry, Vol. 71, No. 13, July 1, 1999
2393
acceptable accuracy and precision. First, demonstrating good recovery of the rMAb from the analytical capillary is particularly important when bare fused-silica capillaries are used in CE methods since, historically, loss of proteins through nonspecific capillary wall adsorption has been reported.38,39 In contrast to HPLC, the use of small volumes in CE render recovery measurements challenging. Earlier validation studies at Genentech of capillary isoelectric focusing (cIEF) methods using coated capillaries relied on the use of enzyme-linked immunosorbent assays (ELISA),40,41 or radiolabeling42 to measure the concentration of the sample collected from the CE capillary. For this study, recovery of the protein from the capillary was determined using 125I-labeled rMAb, which gave a CE-SDS-NGS profile identical to that of a nonradiolabeled sample. A baseline value or control was obtained by performing the assay through the injection step, followed by a high-pressure rinse to expel the entire capillary contents into a collection vial. Following this, the assay was performed to completion as outlined in the Experimental Section. After the last peak had migrated past the detector, the voltage was disconnected and the capillary content was purged into a collection vial. The counts per minute of both samples were determined using a γ-counter. Comparison of the control and experimental samples resulted in a mean recovery of 95% (n ) 3). It should be noted that such recovery studies are often not done for SDS-PAGE where selective loss of material may occur in the stacking gel. The precision of the assay for nonreduced samples was demonstrated by the evaluation of six independent sample preparations on a single day (repeatability) and by the analysis of independent sample preparations on three separate days by two different analysts (intermediate precision). The relative standard deviation (RSD) values for the migration time were e0.9%. The RSD values for peak area percent of the main peak and the minor peaks in the profile were e0.6 and e12.6%, respectively. The higher variability observed with the minor peaks is primarily related to the SDS heating step (see below). These results demonstrate that the use of uncoated fused-silica capillaries in combination with a sieving matrix can provide adequate precision and analyte recovery. Effects of Labeling on rMAb Aggregation. During optimization of the dye-to-protein ratio (D/P), a direct relationship between increasing D/P and protein aggregation was observed. The aggregate peak area increased from 0.5 to 1.7% upon increasing the dye-to-protein molar ratio from 8 to 32. This relationship is believed to be because rhodamine-rMAb conjugates are more prone to aggregation than the unlabeled sample.36 However, no changes were observed in the relative distributions of the monomer and fragment species upon varying the D/P. In addition, within the range of D/P investigated, there was no observed correlation between increasing D/P and enhanced sensitivity. Thus, a D/P of 8 was selected for this study in order to minimize sample aggregation. This resulted in the addition of an average (38) Jorgenson, J. W.; Lukacs, K. D. Anal. Chem. 1981, 53, 1298-1302. (39) El Rassi, Z.; Nashabeh, W. In Capillary Electrophoresis Technology; Guzman, N. A., Ed.; Marcel Dekker Press: New York, 1993; pp 383-434. (40) Moorhouse, K. G.; Rickel, C. A.; Chen, A. B. Electrophoresis 1996, 17, 423430. (41) Thorne, J. M.; Goetzinger, W. K.; Chen, A. B.; Moorhouse, K. G.; Karger, B. L. J. Chromatogr., A 1996, 744, 155-165. (42) Hunt, G.; Hotaling, T.; Chen, A. B. J. Chromatogr., A 1998, 800, 355-367.
2394 Analytical Chemistry, Vol. 71, No. 13, July 1, 1999
Figure 4. CE-SDS-NGS separations of reduced preparations of 5-TAMRA.SE-labeled rMAb using different SDS-rMAb incubation times at 90 °C. All other conditions are as in Figure 1B.
of 2-3 dye molecules per protein molecule, as determined spectrophotometrically. These labeling conditions yielded an aggregate peak area percent of ∼0.5% as compared to 0.3% for an unlabeled sample as determined by native SEC. The impact of the labeling on aggregation is expected to vary from protein to protein, and the use of an alternative analytical method (i.e., SEC) could be helpful in determining the optimum labeling conditions. Effects of SDS-rMAb Heating on Analyte Distribution. To accelerate SDS-protein binding, protein-SDS solutions are traditionally heated at 90 °C. Exposure of proteins to such high temperatures may result in artifacts in the form of thermally induced fragmentation and/or aggregation and thus significantly alter the true representation of the sample. To examine the initial distribution of each species present in the sample prior to heating, eight nonreduced and reduced preparations of a single labeled sample were heated at 90 °C at 0.5, 1, 2, 3, 4, 6, 8, and 10 min, and analyzed by CE-SDS-NGS (Figures 3 and 4). It is seen in Figure 3 that the rMAb fragment peaks’ distribution in the nonreduced samples gradually increased over the heating time course, from an average of 4% at 1 min to 21% at the 10-min time point, with the fragments being ∼6% at the target heating time (3-4 min). A substantial increase in peaks 7 and 6 (each resulting from the sequential loss of one light-chain) and a corresponding increase in the light chain peak (peak 2) were observed. On the other hand, the effect of heating is initially less pronounced on the disulfide bonds in the hinge area because both peaks 4 (heavy chain) and 5 (half antibody) were virtually not present in the sample during the first 2 min of heating, but then increased substantially with exposure times up to 10 min (see Figure 3). These results indicate that the thermal degradation of the protein occurred initially through intrachain disulfide bond rupture between the light and heavy chains followed by the disulfide bonds holding the two heavy chains. This is not unexpected since the release of a light chain would only necessitate the rupture of a
single disulfide bond, in contrast with the heavy chains that are held together by two disulfide bonds in the hinge region. The shoulders or splits in the monomer peak observed at the 0.5- and 1-min time points (see Figure 3) were indicative of incomplete SDS binding and/or unfolding of the protein. As for the reduced sample preparations, complete rupture of all intrachain disulfide bonds resulted within 3 min of heating as evidenced by the disappearance of the intermediate half-antibody signal, resulting in the formation of the predominant light- and heavy-chain species (see Figure 4). Further exposure of the reduced samples to high temperature resulted in only minor changes in the electrophoretic profile, mainly the appearance of a diffuse peak between the light and heavy chains. To alleviate the adverse effects of heating at high temperatures, particularly for nonreduced samples, we investigated SDS-protein binding using three additional temperatures (25, 37, and 60 °C). With the exception of the 25 °C time points, the data for peaks 2, 5, monomer, and aggregate species are presented in Figure 5 as a plot of absolute area vs heating time. Over the first 10 min of SDS-sample heating at 37, 60, and 90 °C, the light-chain signal (peak 2) increased by factors of ∼2, 2.3, and 6.8, respectively (see Figure 5a). Similar trends were observed for peaks 6 and 7 (data not shown). On the other hand, peak 5 (half-rMAb) exhibited little change over 20 min of heating at 37 °C (
