Capillary Zone Electrophoresis Method for a Highly Glycosylated and

Dec 18, 2014 - A purity method based on capillary zone electrophoresis (CZE) has been developed for the separation of isoforms of a highly glycosylate...
0 downloads 14 Views 1MB Size
Article pubs.acs.org/ac

Capillary Zone Electrophoresis Method for a Highly Glycosylated and Sialylated Recombinant Protein: Development, Characterization and Application for Process Development Le Zhang,*,† Ken Lawson,† Bernice Yeung,†,‡ and Jette Wypych† †

Analytical Sciences of Drug Substance Department, Amgen, Inc., Thousand Oaks, California 91320, United States S Supporting Information *

ABSTRACT: A purity method based on capillary zone electrophoresis (CZE) has been developed for the separation of isoforms of a highly glycosylated protein. The separation was found to be driven by the number of sialic acids attached to each isoform. The method has been characterized using orthogonal assays and shown to have excellent specificity, precision and accuracy. We have demonstrated the CZE method is a useful in-process assay to support cell culture and purification development of this glycoprotein. Compared to isoelectric focusing (IEF), the CZE method provides more quantitative results and higher sample throughput with excellent accuracy, qualities that are required for process development. In addition, the CZE method has been applied in the stability testing of purified glycoprotein samples.

I

bare fused silica capillary was used with buffer modifiers such as 1,4-diaminobutane,17,21 SDS (at below the critical micelle concentration level)16 and polyamines such as putrescine13,14,16,18,20,22 to help control electro-osmotic flow and to inhibit binding of the protein to the capillary inner wall. In addition, capillary regeneration through extensive rinsing with water, sodium hydroxide and hydrochloric acid between sample injections may be required to improve method robustness and reproducibility.21 The glycoprotein discussed here is a recombinant protein expressed in Chinese Hamster Ovary cells. The protein backbone has a molecular weight of approximately 18 kDa, and it contains complex carbohydrates of both N- and O-linked types. A total of seven N-linked and one O-linked glycosylation sites exist within the structure of this glycoprotein, which means that the carbohydrate moiety may comprise >60% of the total molecular weight for this molecule. When fully glycosylated and sialylated, the glycoprotein exhibits an extremely low isoelectric point (pI) of ∼2.0, which is lower than most reported pI values of proteins in the literature and, as such, presents challenges for analysis of the most charged glycoforms by conventional IEF or cIEF. In this study, a CZE method utilizing a neutral coated capillary and a simple buffer system is developed for the separation of glycoforms of a glycoprotein. The analysis time per sample is under 40 min, including capillary conditioning time. Due to concerns over in vivo bioactivity,23,24 only

n the past decade, rapid progress has been made in the development of therapeutic proteins using recombinant DNA technology.1 Glycoproteins represent an important class of therapeutic proteins. Glycosylation has been shown to be critical in the clearance, molecular recognition processes including antigenicity and biological activity of therapeutic glycoproteins.2−5 Therefore, during the development of a manufacturing process for therapeutic glycoproteins, it is important to monitor variability of the glycoforms, which consist of a complex mixture of carbohydrate moieties. In general, there are several strategies to study glycoform heterogeneity;6−8 one strategy is to analyze intact glycoproteins directly, the second strategy involves enzymatic digestion of the protein and subsequent analysis of the glycopeptides and the third strategy is to release the carbohydrate from the protein core structure followed by fluorescent labeling and chromatographic separation. Isoelectric focusing (IEF) has been traditionally used for intact glycoprotein isoform analysis.9,10 However, the method is time-consuming and often lacks quantitative information. Use of capillary isoelectric focusing (cIEF) overcomes these limitations.10,11 But cIEF method development and optimization require expertise,12 and its utility may be dependent on the commercial ampholytes available. For example, for proteins with an isoelectric point (pI) at the extreme acidic or basic ends, it may be difficult to develop a suitable method. Another alternative to IEF for glycoprotein isoform analysis is free solution capillary zone electrophoresis (CZE). Recently, CZE has been reported for analysis of various glycoproteins, including α1-acid glycoprotein,13,14 antithrombin,15 ovalbumin,16 fetuin10 and human erythropoietin.17−22 In most cases, a © 2014 American Chemical Society

Received: July 19, 2014 Accepted: December 6, 2014 Published: December 18, 2014 470

dx.doi.org/10.1021/ac504187v | Anal. Chem. 2015, 87, 470−476

Analytical Chemistry

Article

N-Glycan Mapping. The N-glycans were released by PNGaseF, labeled with 2-AB and then analyzed by high pH anion exchange (HPAE) chromatography with fluorescent detection.25 A Dionex DX-600 HPLC equipped with a quaternary pump, a thermostated autosampler, a column compartment and a RF2000 fluorescence detector was used for the analysis. The gradient used composed of water, sodium acetate and sodium hydroxide, and the column used was a CarboPac PA1 anion exchange column from Dionex (Sunnyvale, CA). Digestion with PNGase F took place at 37 °C for up to 3 h, and the labeling with 2-AB occurred at 37 °C for overnight. Peak identification was determined by a combination of matrix assisted laser desorption/ionizationmass spectrometry (MALDI-MS) analysis and exoglycosidase treatment of the collected peaks from high performance anionexchange (HPAE) chromatography. Sialic Acid Quantification. The quantification of sialic acids is based on a colorimetric method of the reaction of sialic acid with ninhydrin.26 Briefly, a standard curve of sialic acid was created to cover the estimated levels of sialic acid in the glycoprotein samples. The glycoprotein purification intermediate samples were mixed with acetic acid and ninhydrin in glass tubes and heated in a boiling water bath for 20 min. After chilling on ice, the content of each tube was transferred to a microcuvette and the absorbance was recorded at 470 nm using a Beckman DU-640 spectrophotometer. The slope and intercept from the standard curve were applied when the sialic acid content was calculated (in moles) in each sample. Based on the molar content of the glycoprotein sample, the results were reported as moles of sialic acid per moles of the glycoprotein. O-Glycan Characterization. The O-glycan structure of the glycoprotein was characterized using peptide mapping followed by confirmation through the use of sialidase and O-glycosidase digestions. The glycoprotein was first digested with endoproteinase Lys-C and the resulting peptides were separated on a reversed-phase C5 column and detected by both UV and mass spectrometry using an Agilent 1100 HPLC system coupled with Thermo Scientific LTQ mass spectrometer. The initial Lys-C digest was incubated with PNGase F to remove N-glycans and then subjected to sialidase and subsequent O-glycosidase treatment. The resulting peptides after each enzyme treatment were all analyzed by HPLC/mass spectrometry.

glycoforms that represent full glycosylation site occupancy are desired in the final product profile of the glycoprotein. Therefore, the CZE method has become a key in-process assay for the development of the cell culture and purification process for this molecule, as this technique provides quantitative measurement of the isoform distribution with fast speed, high precision and good accuracy. Details on the method development, characterization and performance, as well as examples of its application to support the process development, will be described in this paper.



EXPERIMENTAL SECTION Materials and Reagents. The glycoproteins, Epogen and Aranesp were all manufactured at Amgen, Inc. (Thousand Oaks, CA). β-Alanine, tetramethylethylenediamine (TEMED), urea, sialic acid standard (N-acetylneuraminic acid), ninhydrin, 2-aminobenzamide (2-AB) and peptide-N-glycosidase F (PNGaseF) were obtained from Sigma (St. Louis, MO). Sialidase and O-glycosidase were obtained from QA-Bio (Palm Desert, CA). Endoproteinase Lys-C was obtained from Roche Diagnostics (Penzberg, Germany). Glacial acetic acid was from Mallinckrodt (St Louis, MO). Tween-20 was from BioRad (Hercules, CA). Poly(vinyl alcohol) (PVA) coated capillaries were purchased from Agilent Technologies (Waldbronn, Germany). 0.1% Orange G was obtained from Beckman Coulter (Fullerton, CA). Sepacryl-30 solution was purchased from Midwest Scientific (St Louis, MO). Servalyte 2-4 was from Serva (Heidelberg, Germany). the Mini-cell gel box and IEF running buffers were obtained from Invitrogen (Grand Island, NY). Gel Code Blue Stain was from Pierce (Rockford, IL). CZE Analysis. CZE was performed on a Beckman MDQ or PA-800 system equipped with a UV detector (Beckman, Fullerton, CA). The capillary was thermostated at 15 °C, and detection was performed at 214 nm. A PVA coated capillary (50 μm id × 50 cm) was used with a constant voltage of −30 kV for all analyses. Before each sample injection, the capillary was rinsed with 10 mM H3PO4, high performance liquid chromatography (HPLC) grade water and then the running buffer of 25 mM β-alanine/acetic acid, 0.01% Tween-20 (v/v), pH 3.3. Prior to analysis, purified samples were mixed with internal standard Orange G at a 9:1 (v/v) ratio, and were typically injected at 0.2 psi*20 s (for a 1 mg/mL concentration). Due to relative high salt content, purification intermediate samples were usually buffer exchanged and concentrated to approximately 1 mg/mL using Microcon YM-30 centrifugal filtration device (Millipore, Billerica, MA) prior to mixing with internal standard. IEF Analysis. All IEF gels were manually prepared as follows: An acrylamide solution (12.0 g urea, 8 mL Sepacryl-30, 2 mL Servalyte 2-4, filled to 40 mL total volume with water) was mixed with 1.3 mL of a 0.75% (w/v) ammonium persulfate solution and 20 μL of TEMED. The mixture was immediately poured into empty plastic cassettes using a syringe. After inserting the combs for sample lane preparation, the gels were allowed to polymerize overnight at room temperature. The IEF gels were stored at 2 to 8 °C and were usually used within 3 days of preparation. After sample loading, the IEF gel was run at cold temperature with a voltage ramp of 100 to 500 V over the course of 4.5 h. After analysis was complete, gels were destained with 40% methanol/10% acetic acid, and then stained with Gel Code Blue Stain for visualization.



RESULTS AND DISCUSSION Method Development. In CE separation of proteins, the internal coating of the capillary is of primary importance since protein molecules tend to adsorb onto the negatively charged surface of the bare fused silica walls.27 This unwanted interaction between the protein and the silanol groups in the capillary occurs at pH above ∼4 (pKa of silanol), which may lead to irreproducible resolution, long migration time and band broadening during the separation. Therefore, at the beginning of CZE method development, neutral PVA coated capillaries from Agilent were considered. An additional benefit of using a PVA coated capillary is the lack of electroosmotic flow (EOF) due to the neutral coating.28 Under these conditions, the migration of separated protein species is driven by their own inherent electrophoretic mobility. Without interference from EOF, high resolution may be obtained, especially for species that are closely related, such as the isoforms in the glycoprotein. As mentioned earlier, the glycoprotein has an extremely low pI of ∼2, and consequently, the CZE separation of isoforms 471

dx.doi.org/10.1021/ac504187v | Anal. Chem. 2015, 87, 470−476

Analytical Chemistry

Article

would be best achieved with a buffer pH at 3.0−3.5, which is about 1.0−1.5 pH units above the pI value, and using negative polarity for the separation. Under these conditions, the product isoforms would retain negative charges. To achieve the desired pH range, it is important to select a buffer with a pKa value near the desired pH, as this will ensure best buffering capacity. At the same time, a buffer with low conductivity would be desirable, as it would generate low running current, thereby minimizing the Joule heating effect that can be detrimental to the resolution during CZE separation. In addition, a low running current would allow for the use of high voltage, another factor that would lead to higher resolution in the separation.29 An assessment of commonly used CE buffers shows β-alanine (pKa = 3.5) to be the best candidate.30 With β-alanine as the starting buffer, initial experiments were conducted in an Agilent PVA capillary (50 μm × 50 cm) to determine the best buffer composition for the CZE method, including counterion in the buffer (Figure 1A), most robust pH for the buffer (Figure 1B), appropriate buffer concentration (Figure 1C) and finally the need for buffer additives such as Tween-20. As shown in Figure 1A, it was evident that acetic acid would be the choice of counterion, because neither citric acid nor hydrochloric acid appeared to resolve the product isoforms of the glycoprotein when buffer was titrated to the pH of 3.5. Thus, further development using acetic acid as a counterion was performed. From Figure 1B, it was clear that buffer pH from 3.1 to 3.5 did not have a significant impact on the resolution; therefore the middle pH at pH 3.3 was chosen as the pH of the running buffer to ensure a reasonable working range. It should be noted that the concentration of β-alanine was lowered from 100 mM in Figure 1A to 25 mM in Figure 1B to decrease running current and potential Joule heating. Later it was demonstrated that concentrations from 10 to 50 mM βalanine did not have a significant impact on the resolution (Figure 1C); meanwhile, at 25 mM β-alanine, the current was maintained at relatively low level of less than 15 μA; therefore, the middle concentration at 25 mM was chosen as the buffer concentration. Evaluation of buffer additive indicated the baseline of CZE electropherograms was much more variable when Tween-20 was absent (Supporting Information, Figure S1). The relative amounts of individual isoforms were most consistent when adding 0.005, 0.01 and 0.02% (v/v) of Tween20 to the running buffer (Supporting Information, Table S-1). The middle value of 0.01% (v/v) of Tween-20 was chosen to ensure good reproducibility. In summary, the optimal CZE running buffer was determined to be 25 mM β-Alanine/acetic acid, 0.01% (v/v) Tween-20, pH 3.3. Using optimal buffer conditions, the next set of experiments was performed to determine the best instrumental parameters, including separation voltage, separation temperature and capillary type. In commercially available CE instruments, such as the Beckman CE, the maximum voltage supplied is 30 kV. Using the optimized running buffer, we were able to apply −30 kV for the CZE analysis to maximize peak resolution while maintaining relatively low current (90% of total). To characterize the O-glycan structure, the glycoprotein was digested with endoproteinase Lys-C and the peptide mixtures were further treated with endoglycosidases, which resulted in shifts in mass as well as retention time for the peptide containing the O-glycosylation site (Supporting Information, Figure S-3). It was found the Oglycans exist as either mono- or disialyated form for the purified glycoprotein sample. As mentioned earlier, up to eight glycosylation sites (7 N-linked glycosylation and 1 O-linked glycosylation sites) are possible; therefore, when fully occupied and sialylated, these complex glycans may carry a maximum of 30 SA per molecule. To ensure full bioactivity, the desired product profile would contain the top five most sialylated isoforms, representing species with full glycosylation site occupancy. Shown in Figure 4 is the CZE electropherogram

stored dry at room temperature. Therefore, the Agilent PVA capillary was chosen for the CZE method. Method Characterization. Correlation of CZE and IEF Data. Because the amount of protein being injected onto the CZE capillary is typically on the order of hundreds of picograms, it is not practical to collect the individual protein isoform peaks from the capillary for further characterization. However, characterization of the CZE method may be accomplished by injection of individual isoforms isolated or enriched by other means, for instance, IEF. Figure 2 shows an

Figure 2. IEF analysis of the glycoprotein. Lanes 1−4 show a partially purified sample that contain 10 bands, and lanes 5 and 6 show a crude sample that contain all isoforms of the protein as harvested from cell culture.

example of an IEF gel containing a partially purified sample and a crude sample of the glycoprotein. In the IEF gel (Figure 2), the protein load was not optimized for best resolution. The gel was overloaded to obtain sufficient material for subsequent characterization by CZE. To characterize the isoforms in the product region, the area where bands #1−10 migrated was excised and cut into 10 strips of equal widths, and a total of 4 IEF lanes, each with 10 gel slices, were obtained. Corresponding gel bands were pooled and incubated in water overnight on a shaker at 4−6 °C. An aliquot of each incubated gel band, collected after further concentration, was then mixed with the internal standard Orange G and analyzed by CZE. The CZE results of these gel bands are shown in Figure 3. Although none

Figure 4. CZE analysis of a purification intermediate sample of the glycoprotein, with each peak label indicating the number of sialic acids attached to that isoform.

for a purification intermediate sample with a typical isoform distribution. The peak assignment is mostly based on the results from the Correlation of CZE and IEF Data and the Nglycan as well as O-glycan characterization. Peaks 2 through 6 represent the five most sialylated isoforms and their tentative identities are indicated in the parentheses as 30 to 26SA, suggesting 30 to 26 sialic acids are attached to the corresponding isoforms. Peaks 7 through 9 have tentatively 25 to 23 sialic acids attached. Peak 1 is tentatively assigned as

Figure 3. CZE analysis of excised IEF gel bands collected from the partially purified sample of the glycoprotein (bands 1−10). The bottom red trace is an analysis of a direct injection of the same partially purified material. All traces have been aligned by time according to the internal standard Orange G (marked with an asterisk). 473

dx.doi.org/10.1021/ac504187v | Anal. Chem. 2015, 87, 470−476

Analytical Chemistry

Article

Figure 5 shows that the CZE method is specific for our glycoprotein; it can be used to differentiate from the other

31SA, indicating the presence of 31 sialic acids (or equivalent) and this will be discussed later. To further evaluate peak assignment, three purification intermediate samples A, B and C enriched with different levels of isoforms were analyzed by CZE and the purity of each isoform for these samples are determined (Supporting Information, Table S-3). From the purity results, a weighted mean of SA can be calculated using the following equation, where the number of SA of each peak is based on peak assignment shown in Figure 4. weighted mean =

sum of (purity of each peak × number of SA of each peak) sum of purity

Meanwhile, the same samples were analyzed by a spectrophotometric assay via reaction with ninhydrin to determine their sialic acid contents. Each sample was analyzed in duplicate at two different protein levels, resulting in a total of four analyses per sample. Table 1 shows the sialic acid results by the

Figure 5. Analysis of the glycoprotein, Epogen and Aranesp by CZE under the final optimized conditions, demonstrating the specificity of the CZE method. The same amount was injected for all three samples for this analysis.

Table 1. Comparison of CZE Method with the Ninhydrin Assay for Sialic Acid Quantification of Three Purification Intermediate Samples (A, B and C) sialic acid results by ninhydrin assay

proteins, as the method parameters are dependent on the pI and sialic acid content of the molecule. The pI of Epogen is ∼4.4 whereas the pI of Aranesp is ∼3. The average number of sialic acids per molecule is the highest in our glycoprotein; therefore it migrates much earlier than the other two proteins. For intermediate precision, data from a purified sample of our glycoprotein in over 100 separate CZE assays was evaluated. The % peak area of the three most abundant isoforms, peaks 2 to 4, as well as the total % product isoforms (peaks 1 to 6), were trended from these assays. During this period, six different lots of the PVA capillary, three instruments (including both Beckman MDQ and PA-800) and three analysts were the variables that contributed to the data set. The results in Table 2 demonstrate excellent intermediate precision

sialic acid results by CZE

sample

average SA/molecule

% RSD

number of SA weighted mean

difference (%)

A B C

28.6 26.0 25.1

2.3 1.9 4.1

29.1 27.4 25.3

−1.6 −5.1 0.7

ninhydrin assay as compared to that of CZE. The results from the two orthogonal methods are consistent, which indicates that the assignment of the CZE peaks very likely represents the number of sialic acids labeled for each isoform peak, as shown in Figure 4. Given the structure of the glycoprotein, a sialic acid content of more than 30 per molecule is less likely. Therefore, Peak 1 in the CZE profile (Figure 4) would have been contributed by another negative charge in addition to being fully sialylated. In order to characterize Peak 1, a purification intermediate sample enriched with 14.1% of Peak 1 was analyzed by N-glycan mapping and peptide mapping (results not shown). The results from N-glycan mapping did not indicate presence of sulfated glycans, which could potentially add an additional negative charge.31 Reduced peptide mapping also did not reveal any additional post-translational modifications (e.g., sulfation, deamidation, etc.) that could potentially contribute a negative charge to the molecule. It was possible that, at only a 14.1% level enrichment, neither of these analyses offered the sensitivity needed to detect the modifications. Moreover, it is possible that multiple species contributed to the additional negative charge on Peak 1, which would make it extremely difficult to detect by N-glycan and peptide mapping, as the detectability would now be further decreased among the multiple species. Method Performance. Using the final optimized parameters, the performance of the developed CZE method was evaluated by examining the following characteristics: specificity, intermediate precision and accuracy. For specificity, two other highly glycosylated and sialylated proteins, Epogen and Aranesp, were analyzed by the CZE method. The results in

Table 2. Intermediate Precision Results of the CZE Method for the Purified Glycoprotein Sample from over 100 Assays peak name

peak ID

% purity (average)

SD

% RSD

Peak 2 Peak 3 Peak 4 sum of peaks 1−6

30SA 29SA 28SA product isoforms

36.6 42.3 15.9 99.8

0.5 0.5 0.3 0.2

1.5 1.1 1.8 0.2

of the CZE method with relative standard deviation (RSD) values from 1.1 to 1.8% for the individual peaks 2 to 4, and a RSD of 0.2% for the % product isoforms (sum of peaks 1 to 6). Accuracy of the CZE method was examined through spike recovery experiments. A purified sample and a purification intermediate sample containing high levels of nonproduct isoforms were blended at a 1:1 (v/v) ratio. Table 3 shows the results of this experiment, which indicate very similar values between the theoretical and experimental % peak area, thus demonstrating the CZE to have excellent accuracy. The average accuracy of all peaks from the spike recovery experiments is 93.4%. Application of CZE to Process Development. Besides the purified samples, the CZE method developed for this glycoprotein has been applied to cell culture samples starting 474

dx.doi.org/10.1021/ac504187v | Anal. Chem. 2015, 87, 470−476

Analytical Chemistry

Article

Table 3. Results from Spike Recovery Experiments for the CZE Method sample

% Peak 1

% Peak 2

% Peak 3

% Peak 4

% Peak 5

% Peak 6

% Peak 7

% Peak 8

% Peak 9

purified sample purification intermediate sample theoretical peak % experimental peak % % difference % accuracy % accuracy (average for all peaks)

1.1 0.5 0.8 0.7 −8.9 91.1

36.3 10.5 23.4 24.4 4.5 95.5

43.3 18.7 31.0 32.0 3.1 96.9

16.5 16.6 16.5 16.7 1.0 99.0

2.2 14.9 8.6 8.2 −4.6 95.4 93.4

0.4 18.4 9.4 8.7 −7.5 92.5

0.2 14.9 7.6 6.8 −9.4 90.6

0.0 5.1 2.5 2.2 −12.0 88.0

0.0 0.6 0.3 0.3 −8.8 91.2

The experimental data shown here represent the average results obtained from triplicate analysis of each of the samples. The shaded section of the table indicates isoforms that are not considered to be part of the final product profile. Refer to Figure 4 for peak identities.

with the clone selection process. Prior to analysis, the cell culture samples were concentrated and buffer exchanged against water. Once prepared, the samples were mixed with the internal standard and injected for analysis. In Figure 6, an

Figure 7. Progression of isoform selection as analyzed by CZE. % Product isoform purity for each sample: Column 1 Pool (86.4%), Column 2 Pool (95.3%), Column 3 Pool (99.6%), Final Purified Sample (99.6%). The vertical line denotes cutoff between product and nonproduct isoforms.

Figure 6. An example electropherogram showing the CZE results from cell culture samples derived from four different clones.

Along with other product purity and step yield information, the CZE results have provided guidance in determining purification parameters and column pooling criteria in the purification process. Stability Testing. The CZE method has been shown to have stability indicating power for purified product of the glycoprotein. As indicated in Figure 8, when the protein was exposed to 25 or 37 °C for an extended period of time, a shift of the peak distribution toward the less sialylated isoforms was observed, indicating a loss of glycosylation or sialylation for these samples.

example demonstrates how the CZE method was applied in the clone selection process in early development. At this point in process development, a large number of parental cell lines were under consideration, and materials were produced under different cell culture conditions. Using CZE for product purity assessment and a HPLC method for titer analysis, both the quality and expression level of these clones were determined using a minimal volume of each sample. Moreover, compared to conventional IEF that was historically used during clone selection, the sample throughput by CZE was higher (same day vs more than 3 days for analysis of ∼10 samples); and the results obtained with CZE were both more quantitative and more reproducible. In the example shown in Figure 6, even by visual assessment, Clone 1 clearly shows the highest relative proportion of most sialylated isoforms. After the top clone had been selected, the CZE method continued to play an important role in the development of the cell culture process, where different additives and feed strategies were evaluated and demonstrated to have an impact on the production of the desired product isoforms. The CZE method was also successfully applied to the optimization of the purification process for this glycoprotein. As noted earlier, the desired product profile should contain the most sialylated isoforms (Peaks 1 to 6, as shown in Figure 4), and CZE was the ideal method for providing this information in a high throughput manner. An example is shown in Figure 7, where an improved selection of product isoforms is demonstrated with the progression of the purification steps.

Figure 8. CZE analysis of purified glycoprotein sample at T0 (bottom trace), T = 10 month at 25 °C (middle trace) and T = 10 month at 37 °C (top trace). 475

dx.doi.org/10.1021/ac504187v | Anal. Chem. 2015, 87, 470−476

Analytical Chemistry



Article

(14) Puerta, A.; Diez-Masa, J. C.; Martin-Alvarez, P. J.; MartinVentura, J. L.; Barbas, C.; Tunon, J.; Egido, J.; de Frutos, M. Analyst 2011, 136, 816−822. (15) Kremser, L.; Bruckner, A.; Heger, A.; Grunert, T.; Buchacher, A.; Josic, D.; Allmaier, G.; Rizzi, A. Electrophoresis 2003, 24, 4282− 4290. (16) Legaz, M. E.; Pedrosa, M. M. J. Chromatogr. A 1996, 719, 159− 170. (17) Watson, E.; Yao, F. Anal. Biochem. 1993, 210, 389−393. (18) De Frutos, M.; Cifuentes, A.; Diez-Masa, J. C. Electrophoresis 2003, 24, 678−680. (19) Sanz-Nebot, V.; Benavente, F.; Vallverdu, A.; Guzman, N. A.; Barbosa, J. Anal. Chem. 2003, 75, 5220−5229. (20) Sanz-Nebot, V.; Benavente, F.; Gimenez, E.; Barbosa, J. Electrophoresis 2005, 26, 1451−1456. (21) Zhang, J.; Chakraborty, U.; Villalobos, A. P.; Brown, J. M.; Foley, J. P. J. Pharm. Biomed. Anal. 2009, 50, 538−543. (22) Brinks, V.; Hawe, A.; Basmeleh, A. H.; Joachin-Rodriguez, L.; Haselberg, R.; Somsen, G. W.; Jiskoot, W.; Schellekens, H. Pharm. Res. 2011, 28, 386−393. (23) Delorme, E.; Lorenzini, T.; Giffin, J.; Martin, F.; Jacobsen, F.; Boone, T.; Elliott, S. Biochemistry 1992, 31, 9871−9876. (24) Elliott, S.; Chang, D.; Delorme, E.; Eris, T.; Lorenzini, T. J. Biol. Chem. 2004, 279, 16854−16862. (25) Kotani, N.; Takasaki, S. Anal. Biochem. 1998, 264, 66−73. (26) Yao, K.; Ubuka, T. Acta Med. Okayama. 1987, 41, 237−241. (27) Bushey, M. M.; Jorgenson, J. W. J. Chromatogr. A 1989, 480, 301−310. (28) Steiner, F.; Hassel, M. Electrophoresis 2003, 24, 399−407. (29) Issaq, H. J.; Atamna, I. Z.; Muschik, G. M.; Janini, G. M. Chromatographia 1991, 32, 155−161. (30) Pospichal, J.; Gebauer, P.; Bocek, P. Chem. Rev. 1989, 89, 419− 430. (31) Hermentin, P.; Witzel, R.; Kanzy, E.-J.; Diderrich, G.; Hoffmann, D.; Metzner, H.; Vorlop, J.; Haupt, H. Glycobiology 1996, 6, 217−230.

CONCLUSION A CZE method for the product isoform separation of a highly glycosylated and sialylated protein has been developed and characterized in this study. The method has been shown to have excellent specificity, precision and accuracy. The peaks separated by CZE have been characterized using orthogonal methods, and demonstrated to be likely correlate with isoforms containing different number of sialic acids within the protein. The CZE method has been widely applied as an in-process purity method during the process development of this glycoprotein, providing a much needed purity information in a high throughput manner. Finally, the method has also been used for stability testing of the purified glycoprotein sample.



ASSOCIATED CONTENT

* Supporting Information S

Effect of Tween-20 on the CZE separation, N-glycan mapping results of the glycoprotein, O-glycan characterization of the glycoprotein by Lys-C peptide map, and determination of weighted mean of SA by CZE. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Le Zhang. Address: One Amgen Center Drive, Mail Stop 30W-3-A, Thousand Oaks, CA 91320. Tel: (805) 447-0615. Fax: (805) 375-8251. E-mail: [email protected]. Present Address ‡

Analytical Development, Shire, 300 Shire Way, Lexington, MA 02421 Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS The authors thank Izydor Apostol for comments and discussions. REFERENCES

(1) Salas-Solano, O.; Tomlinson, B.; Du, S.; Parker, M.; Strahan, A.; Ma, S. Anal. Chem. 2006, 78, 6583−6594. (2) Dubé, S.; Fisher, J. W.; Powell, J. S. J. Biol. Chem. 1988, 263, 17516−17521. (3) Baenziger, J. U.; Kumar, S.; Brodbeck, R. M.; Smith, P. L.; Beranek, M. C. Proc. Natl. Acad. Sci. U. S. A. 1992, 89, 334−338. (4) Hu, A.; Cattaneo, R.; Schwartz, S.; Norrby, E. J. Gen. Virol. 1994, 75, 1043−1052. (5) Rudd, P. M.; Elliott, T.; Cresswell, P.; Wilson, I. A.; Dwek, R. A. Science 2001, 291, 2370−2376. (6) Mechref, Y.; Novotny, M. V. Chem. Rev. 2002, 102, 321−370. (7) Wuhrer, M.; Deelder, A. M.; Hokke, C. H. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2005, 825, 124−133. (8) Balaguer, E.; Neusüss, C. Anal. Chem. 2006, 78, 5384−5393. (9) Cifuentes, A.; Moreno-Arribas, M. a. V.; de Frutos, M.; DíezMasa, J. C. Journal of Chromatography A 1999, 830, 453−463. (10) Kinoshita, M.; Murakami, E.; Oda, Y.; Funakubo, T.; Kawakami, D.; Kakehi, K.; Kawasaki, N.; Morimoto, K.; Hayakawa, T. Journal of Chromatography A 2000, 866, 261−271. (11) Moorhouse, K. G.; Rickel, C. A.; Chen, A. B. Electrophoresis 1996, 17, 423−430. (12) Chang, W. W.; Hobson, C.; Bomberger, D. C.; Schneider, L. V. Electrophoresis 2005, 26, 2179−2186. (13) Ongay, S.; Martín-Á lvarez, P. J.; Neusüß, C.; de Frutos, M. Electrophoresis 2010, 31, 3314−3325. 476

dx.doi.org/10.1021/ac504187v | Anal. Chem. 2015, 87, 470−476