1461
Electrophoresis 2014, 35, 1461–1468
Junzi Cao1 Wen Sun2 Feifei Gong1 Wanhui Liu1,2 1 School
of Pharmacy, Yantai University, Yantai, P. R. China Key Laboratory of Long-acting and Targeting Drug Delivery System, Luye Pharma Group Ltd, Yantai, P. R. China
2 State
Received September 28, 2013 Revised February 19, 2014 Accepted February 24, 2014
Research Article
Charge profiling and stability testing of biosimilar by capillary isoelectric focusing CIEF was developed for the rapid analysis of charge heterogeneity of trastuzumab biosimilar using commercially available fluorocarbon-coated capillary. The CIEF master mix was composed of 0.30% w/v methyl cellulose, 2.3 M urea, 56.8 mM L-arginine, 1.52 mM iminodiacetic acid, 4.5% v/v carrier ampholytes (broad-range pI 3–10 and narrow-range pI 8–10.5 with ratio of 3:1), and 0.45% v/v 10.0, 9.5, 7.0, 5.5, 4.1 pI markers. To get a robust method to analyze charge heterogeneity, some separation parameters, including focusing time and separation temperature, were investigated and optimized. The optimized method gave good precision in estimated pI values of charge variants with RSDs of not more than 0.16% intraday analysis (n = 6) and ⬍ 0.18% interday analysis (n = 9). In addition, the applications of this method including purity, stability, lot consistency, peptide N-glycosidase F digest, and C-terminal lysine variants characterization were also investigated. Keywords: Charge heterogenity / CIEF / Monoclonal antibody / Trastuzumab biosimilar DOI 10.1002/elps.201300471
1 Introduction MAbs have drawn significant attention in the last few decades to treat inflammatory, oncologic, and autoimmune diseases [1–3]. Trastuzumab is developed by Roche (Indianapolis, IN, USA) for treatment of breast cancer patients who show an overexpression of HER2 [4]. Compared with trastuzumab, the biosimilar has only two different amino acids, which vary in heavy chain at the sites of 359 and 361, aspartic acid, and leucine instead of glutamic acid and methionine. The effects of trastuzumab and trastuzumab biosimilar on growth of Bt474 cells were evaluated using the method of AlamarBlue [5, 6]. Fluorescence was detected at an excitation of 560 nm and emission at 590 nm using a Microplate Reader (TECAN, Switzerland). The cellular activity of biosimilar was equivalent to that of trastuzumab. The biosimilar is promising to be a candidate biological drug. Analysis or characterization of mAbs is a challenging task, because they are complex compounds with heterogeneity in size and charge. Charge heterogeneity can result from multiple sources such as deamidation, formation of N-terminal pyroglutamate, aggregation, isomerization, sialylated glycans, antibody fragmentation, and glycation at the lysine (Lys) residues [7–9]. In some cases, such changes affect binding, biological activity, patient safety, and shelf life [10].
Correspondence: Dr. Wanhui Liu, School of Pharmacy, Yantai University, Yantai, P. R. China E-mail: [email protected] Fax: +86-535-3808293
Abbreviations: CA, carrier ampholyte; CpB, carboxypeptidase B; iCIEF, imaged CIEF; MC, methyl cellulose; MT, migration time; PNGase F, peptide N-glycosidase F C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Therefore, analyzing charge heterogeneity of mAbs is critical for characterizing and monitoring quality attributes of antibodies. Characterizing charge variants in the biopharmaceutical field relies on tools, such as IEC [11], IEF gel electrophoresis [12], IPGs [13] and CE including CZE [14], CIEF [15], and imaged CIEF (iCIEF) [16]. Among these methods, IEF, IPGs, CIEF, and iCIEF separate proteins or antibodies are based on pIs. CIEF and iCIEF have the advantage over traditional gel IEF. First, the use of slab gels is known to suffer from lack of investigation in the course of validation, as well as accuracy and reproducibility for quantitative measurement of the separated charged components. Second, CIEF has proven its ability to combine the resolving power of IEF with advantages of CE quantitation and automation [17–19]. Compared with CIEF, IPGs are widely used in 2D electrophoresis, but are labor-intensive and difficult to quantification. Given the advantages of CIEF, separations based on pIs have been widely used in mAbs analysis for identification, purity assessment, degradation checking, microheterogeneity analysis, and clinical diagnosis [9, 15, 20]. Besides, because of the extremely high peak capacity of CIEF, it is selected as the more preferred method for the analysis of mAbs. CIEF has two general modes of performance, that is, onestep and two-step CIEF, where focusing and mobilization occur simultaneously or sequentially [21–23]. The latter was used in this study. To get a much higher reproducibility and reduced chance of accidental protein adsorption to the silica wall, coated capillaries, such as fluorocarbon [10], DB-1 or DB17 [17], polyacrylamide (neutral) [24], PVA [25], and cellulous derivants [26] coated capillaries, are used in CIEF or iCIEF. Fluorocarbon-coated capillary, which is widely used in iCIEF [10, 27–29], was selected in our study and a high separation efficiency was obtained. www.electrophoresis-journal.com
1462
J. Cao et al.
Apart from coated capillaries, the key factors such as sample preparations and separation conditions are also related to separation efficiency in CIEF. Therefore, to establish a robust method, sample compositions, such as urea, methyl cellulose (MC), L-arginine (L-Arg), carrier ampholytes (CAs), and separation conditions including focusing time, separation temperature, were studied. A robust CIEF method for biosimilar was established based on the standard separation previously published by Mack S. et al [24]. Precision studies were conducted to demonstrate that the closeness of results could be obtained over a short time under the same operating conditions. Finally, CIEF applications, including the stability of mAb, mAb digested by enzymes, purity, and lot-to-lot consistency, were also performed.
2 Materials and methods
Electrophoresis 2014, 35, 1461–1468
centrifuge (Thermo, Fresco 21, Rockford, IL, USA) speed of 13 500 rpm/min, under the temperature of 4°C. After removing salts, the final volume of the deglycosylation solution was about 70 L and the concentration of mAb was 2.75 mg/mL which was determined by micro-ultraviolet spectrophotometer (Thermo, nanodrop 2000).
2.2.2 Trastuzumab biosimilar digested by CpB MAb (200 g) was mixed with 1 U (activity units) CpB in 1.0 mM phosphate, 10 mM NaCl solution, pH 8.0, and incubated at 37°C for 2 h. As a reference sample, water was added instead of the enzyme. After incubation, remove the salt at the same conditions as PNGase F’s. The concentration of mAb was 2.75 mg/mL.
2.1 Materials and instrumentation 2.2.3 CIEF master mix preparation Experiments were performed on Agilent 7100 CE (external pressure 2–12 bar) equipment with DAD and a 280 nm filter, and the bandwidth was 4 nm. Data were collected and analyzed using Chemstation (Agilent, version 2009). The temperature setting was 20°C. Fluorocarbon-coated capillary was obtained from Agilent Technology (The total length is 33 cm, 24.5 cm effective length, 50 m id). Pharmalyte pH 3–10 CAs (PN 17–0456–01, 0.36 meq/mL) and pH 8–10.5 CAs (PN 17–0455–01, 0.36 meq/mL) were acquired from GE Healthcare (Piscataway, NJ, US) and pI markers 10.0, 9.5, 7.0, 5.5, and 4.1 (PN A58481) were obtained from Beckman Coulter (Brea, CA, US). Urea (PN U6504), carboxypeptidase B (CpB, PN C9548), Triton X-100 (PN T8532), sodium chloride (NaCl, PN S7653), MC (PN M0262), acetic acid (purity ࣙ99.7%, PN 45727), and iminodiacetic acid (PN 220000) were supplied by Sigma-Aldrich (St. Louis, MO, US). Peptide N-glycosidase F (PNGase F, PN 11365193001) was purchased from Roche. Sodium hydroxide (purity ࣙ95.0%, PN 567530) and L-Arg (purity ࣙ99.0%, PN 1820) were from Merck Millipore (Darmstadt, Germany). Phosphoric acid (purity ࣙ85%, PN E582) was brought from Amresco (Solon, IA, US). Milli-Q water was made in-house. Trastuzumab biosimilar was provided by pharmaceutical laboratory of Shandong Luye Pharma Co. 2.2 Sample preparation 2.2.1 Deglycosylation of N-linked oligosaccharides from trastuzumab biosimilar Deglycosylation of trastuzumab biosimilar was conducted as follows. MAb (200 g) was mixed with 5 U (activity units) PNGase F in 1.0 mM phosphate, 10 mM NaCl, 0.5% v/v Triton X-100 solution, pH 7.5 and incubated at 37°C for 24 h. As a reference sample, water was added instead of the enzyme. After incubation, a spin filter (5 kD cut-off, Sartorius) was used to remove salts. The retentate was washed three times with 70 L of Milli-Q water for 20 min each time at the C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
To prepare CIEF samples, the following reagents were mixed in a centrifuge tube: 200 L of 3 M urea containing 0.4% w/v MC, 9.0 L of pharmalyte 3–10 CAs, 3.0 L of pharmalyte 8–10.5 CAs, 20.0 L of cathodic stabilizer (750 mM Arg), and 2.0 L of anodic stabilizer (200 mM iminodiacetic acid).
2.2.4 CIEF sample preparation A 97.5 L of master was mixed with 10 L of protein with concentration of 2.75 mg/mL and 2.5 L of pI markers mixture, which contained 0.5 L of each pI marker. The final sample concentration was 0.25 mg/mL.
2.3 Method The fluorocarbon-coated capillary was successively rinsed with water for 5 min, 350 mM acetic acid for 2 min, and 0.4% w/v MC solution for 3 min at a pressure of 3.5 bar at the beginning of each day. To decrease protein precipitate, before each injection the capillary was flushed with 4.3 M urea for 3 min, water for 2 min, at a pressure of 3.5 bar, respectively. Focusing was performed at 25 kV in normal polarity for 12.5 min. A 200 mM phosphoric acid and 300 mM sodium hydroxide was used as anode and cathode solution, respectively. After focusing, the cathode solution was changed with 350 mM acetic acid, and the charge variants were mobilized to cathode at 30 kV in normal polarity. The sample was injected for 2 min at a pressure of 1 bar. A shutdown method was run at the end of each day to rinse the capillary for 5 min at a pressure of 5 bar with 4.3 M urea, followed by water for 10 min at the same pressure. The DAD lamp was switched off and both of the capillary ends were immersed in Milli-Q water. And the capillary was stored at room temperature overnight. www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1461–1468
1463
3 Results and discussion 3.1 Sample composition 3.1.1 Urea Protein precipitation causes a lack in precision in migration time (MT), peak area (thus affecting quantitation), capillary clogging, unstable current, slow mobilization, and other undesirable effects [30, 31]. Therefore, the additives, such as urea [24], Triton X-100 [32], or glycerol [33], are used to improve the solubility of the focused mAbs and prevent precipitation before they move past the detection window. In this research, urea was studied and the optimal concentration ultimately depended on the solubility of the mAb. The CIEF profiles of trastuzumab biosimilar were evaluated at different urea concentrations. As shown in Fig. 1, in the environment lacking of urea, the separation profile for biosimilar was poorly defined, presumably because of its low solubility. It was sufficient to increase the solubility of the protein sample and prevent aggregate formation by breaking hydrogen–bond interactions at the concentration of 2.3 M (The subdenaturing concentration is not more than 4 M [34]). The separation efficiency, especially to basic variants, dramatically degraded at the concentration of 4.6 M, possibly due to protein denaturation. Besides, higher concentrations most likely unfolded the protein and thus affecting differences in charge due to conformational shifts. Figure 1 showed that the MT extended with concentration increasing because separation medium viscosity increased. Therefore, the optimal concentration of urea was determined to be 2.3 M. 3.1.2 Methyl cellulose MC was used not only as a viscosity-enhancing agent to influence longitudinal diffusion, but also as a neutral hy-
Figure 2. Effect of MC concentration on separation; (A) 0.15% w/v; (B) 0.30% w/v; (C) 0.60% w/v. For other sample compositions and experimental conditions, see Fig. 1.
drophilic polymer used to dynamically coat the capillary walls to minimize analyte–wall interaction and helped suppress EOF. When the concentration of MC increased from 0.15 to 0.60% w/v, the MT increased (Fig. 2), because MC could enhance the viscosity of separation medium. At the concentration of 0.30% w/v, the separation efficiency and resolution was greatly improved, especially for basic variants. However, increasing the concentration of MC from 0.30 to 0.60% w/v, the resolution for basic charge variants degraded. Furthermore, high concentration of MC not only easily led to capillary clogging, which in turn caused poor separation, but also made pipetting more difficult, as a result, reducing the precision of the measurements. Based on the experimental results, the optimized concentration of MC was found to be 0.30% w/v, which could produce adequate separation efficiency and precision.
3.1.3 Arginine
Figure 1. Effect of urea concentration on separation; (A) 0 M; (B) 2.3 M; (C) 4.6 M. Sample concentration: 0.25 mg/mL, containing 0.3% w/v MC, 4.5% v/v pI 3–10 CAs, 37.9 mM Arg, 1.52 mM iminodiacetic acid. A 25 kV focusing for 15 min, 30 kV separation at 25°C. For other conditions see Section 2.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Arg (pI 10.7) was cathodic stabilizer, which was used to prevent the loss of the ampholytes into the electrode solutions. The amount of cathodic stabilizer within the sample needed to be optimized to ensure alkaline compounds are not focused behind the detector. In addition, the amount of cathodic stabilizer probably affected resolution. The concentrations at 0, 37.9, 56.8, 75.8, and 89.1 mM were investigated. An abnormal profile was observed without Arg, because the mAb peaks were detected during focusing and the pI markers of 10.0 and 9.5 did not appear in the electropherogram (Fig. 3). MT from 0 to 5 min was concerned. There were no peaks in trace A, and peak I and peak II appeared in trace B with addition of pI markers, and there was no additional peaks in trace C except that the height of peak I and II increased with increasing concentration of markers. This demonstrated that peak I and peak II were cathodic peaks stemming from pI markers. www.electrophoresis-journal.com
1464
J. Cao et al.
Figure 3. Effect of Arg on pI markers of 10.0 and 9.5; (A) mAb without pI markers and Arg; (B) 0.19% v/v pI markers (10.0, 9.5, 7.0, 5.5, 4.1) without mAb and Arg; (C) 0.25 mg/mL mAb and 0.58% v/v pI markers (10.0, 9.5, 7.0, 5.5, 4.1) without Arg; (D) 0.25 mg/mL mAb and 0.45% v/v pI markers (10.0, 9.5, 7.0, 5.5, 4.1) with 37.9 mM Arg.
Electrophoresis 2014, 35, 1461–1468
Figure 5. Effect of CAs compositions on separation; (A) 4.5% v/v pI 3–10 CAs; (B) 4.5% v/v CAs, pI 3–10 and pI 8–10.5 with ratio of 3:1; (C) 4.5% v/v CAs, pI 3–10 and pI 8–10.5 with ratio of 2:2; (D) 4.5% v/v CAs, pI 3–10 and pI 8–10.5 with ratio of 1:3. For other sample compositions and experimental conditions, see Fig. 1.
3.1.4 Ampholyte
Figure 4. Effect of Arg concentration on separation; (A) 37.9 mM; (B) 56.8 mM; (C) 75.8 mM; (D) 89.1 mM. For other sample compositions and experimental conditions, see Fig. 1.
The pI markers of 10.0 and 9.5 were detected in presence of 37.9 mM Arg (Fig. 3, trace D), because Arg fulfilled a blocking function occupying the region between the detection window and the capillary outlet. Moreover, the signal heights especially to mAb and cathodic peaks increased compared with that without Arg (Fig. 3, trace C), probably because Arg could compress the pH gradient. Figure 4 indicated that with the concentration of Arg increasing, the number of peaks obviously increased and maximized at the concentration of 56.8 mM. A 75.8 mM or higher Arg was an excessive amount of cathodic stabilizer, because it not only delayed the detection of mAb, but also led to a loss in resolution. The optimum concentration of Arg was determined to be 56.8 mM. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Ampholytes were crucial to the success of CIEF since ampholytes established the pH gradient that affected resolution of the protein in the capillary. So, the proper selection of CAs and their concentrations was found to be essential. The concentration of CAs was chosen to be 4.5% after referring to the CIEF application guide analysis kit (Beckman, A78788AE). The mixture of the wide-range ampholytes (3–10) and narrowrange (8–10.5) was used to overcome the limitation of using single ampholytes (3–10) in detecting charge heterogeneity of basic mAbs [35,36]. As shown in Fig. 5, by increasing the percentage of narrow-range ampholytes (8–10.5), resolution was markedly enhanced when MT prolonged. However, increasing the concentration of narrow-range pharmalyte (8–10.5) to 2.25% or higher caused split peaks for pI marker of 9.5, which implied that a longer focusing time was needed. But increasing focusing time would increase the risk of precipitation. The best compromise between MT and separation efficiency was obtained with 3:1 v/v mixtures of broad and narrow pH range CAs. 3.2 Separation condition 3.2.1 Focusing time Because the pH gradient was homogeneous at the start of the separation, the focusing time must be sufficient to allow the cathodic and anodic peaks for each component to merge. The best focusing time was directly related to the separation efficiency. An excess of focusing time would waste time, easily cause protein precipitate, and shorten the lifespan of coated capillaries. Furthermore, insufficient focusing time would lead to partial detection or split peaks. Focusing time that varied from 10 to 17.5 min was investigated to explore the www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1461–1468
1465
A
Figure 6. Effect of focusing time on separation; (A) 10 min; (B) 12.5 min; (C) 15 min; (D) 17.5 min. For sample composition and other experimental conditions, see Section 2 and Fig. 1, respectively.
B
Figure 8. Specificity of trastuzumab biosimilar; (A) Specificity; (B) Enlargement of trace mAb and markers from 25 to 29 min. For sample compositions and experimental conditions, see Section 2. Figure 7. Effect of separation temperature on separation; (A) 15°C; (B) 20°C; (C) 25°C; (D) 30°C. For sample composition and other experimental conditions, see Section 2 and Fig. 1, respectively.
proper value. Figure 6 showed that the number of charge variants maximized at focusing at 12.5 min or longer. A total of 12.5 min were enough for this separation, and no split peaks were detected in the electropherograms. Therefore, the focusing time at 12.5 min was selected as the prime choice. 3.2.2 Temperature Separation temperature had a significant effect on the resolution, because it influenced not only solubility of sample, but also Joule heat, and both of them were related to separation efficiency. Temperatures from 15 to 30°C with 5°C as a step were investigated. Figure 7 indicated that MT decreased and resolution degraded with increasing temperature. It is more likely that higher temperatures reduced viscosity, enhanced longitudinal diffusion, and thus peak broadening. It seemed C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
that 15°C was the best choice. However, the MTs of mAb shifted more greatly (variation range: 0.45 vs. 0.04 min) at 15°C than 20°C within three injections. Therefore, 20°C was chosen as the ideal temperature.
3.3 Method validation 3.3.1 Specificity The purpose of specificity test was to discriminate between the target peaks and other interference peaks from pI markers or master mix. Bulk materials of trastuzumab biosimilar, the concentrated protein solutions that are prior to formulation and final manufacturing were studied here. All injections were performed using the improved CIEF method. As shown in Fig. 8A, there was no interference to analyze the target mAb in the master mix analysis. And addition of the pI markers also did not bring in any other interference peaks. Charge variants of trastuzumab biosimilar were shown in Fig. 8B. www.electrophoresis-journal.com
1466
J. Cao et al.
Electrophoresis 2014, 35, 1461–1468 Table 3. Stability under different temperatures (n = 3)
Table 1. pI precisions test of the CIEF method for mAb
Peak 1 Peak 2 Peak 3 Peak 4 Peak 5 Peak 6 Peak 7 Peak 8 Intraday precisiona) Mean 8.64 8.56 RSD (%) 0.084 0.061
8.46 0.12
8.25 0.12
8.18 0.14
8.10 0.16
8.03 0.14
7.94 0.15
Interday precisionb) Mean 8.64 8.55 RSD (%) 0.12 0.097
8.45 0.16
8.25 0.14
8.18 0.14
8.10 0.17
8.03 0.18
7.94 0.16
a) One sample preparation and six replicate injections in one day. b) One sample preparation and three injections each in three separate days. Table 2. Area percent (%) precisions test of the CIEF method for mAb
Basic
Main
Acidic
Intraday precisiona) Mean RSD (%)
3.940 2.83
56.01 1.17
40.05 1.69
Interday precisionb) Mean RSD (%)
3.960 3.01
55.64 1.45
40.40 1.94
Area percent (%) Sample
Basic
Main
Acidic
pI
Control 37°C 4 h 37°C 24 h 4°C 15 days 4°C 30 days
4.040 4.060 4.240 3.850 3.540
56.75 55.05 54.31 54.21 50.97
39.21 40.89 41.45 41.94 45.49
8.47 8.46 8.47 8.44 8.43
precision for their low content. Compared with pI precisions, area% precisions were more variable. This effect is mainly attributed to instability and progressive degradation of the coating with subsequent enhanced protein adsorption which will continuously impair precision (e.g. pI, peak, area and so on) [37, 38].
3.4 Applications 3.4.1 Stability
a) One sample preparation and six replicate injections in one day. b) One sample preparation and three injections each in three separate days.
3.3.2 Precision To investigate the precision of the optimized CIEF method, a precision study was performed using trastuzumab biosimilar as a model. Intraday precision was evaluated by performing a single sample preparation and six replicate injections in one day. Interday precision was evaluated by performing one sample preparation that was immediately stored in the 4°C refrigerator after three injections every day, and three injections per day for three consecutive days. Sample preparations and separation conditions for intraday and interday precision were the same as specificity. CIEF method yielded a pH gradient over a wide pI range in the capillary. A typical calibration plot of MTs versus pI values of the markers, and a linear calibration curve was obtained for pI markers ranging from 10.0 to 4.1 (R ⬎ 0.99). The pIs of the mAb isoforms were estimated from the calibration curve according to their MTs. As shown in Table 1, the intra- and interday precision for pI values of charge variants had good precisions with RSD of not more than 0.18%. CIEF was applied to purity assessment of mAbs by area normalization method. The percent isoform group composition values were calculated by dividing the sum of the peak areas for each isoform group by the total mAb peak area and multiplying by 100. Table 2 showed that intraday and interday precision for main component and acidic variants had good precision with RSD ⬍ 2.0%, while basic variants had poor C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The stability study referred to the standardized protocol established in house. The stability of therapeutic antibodies during downstream processing and storage was important for functionality and quality. The purpose of stability test was to provide evidence on how a protein product changes with time under the influence of different factors such as light, pH, and temperature. The increase of temperature might cause changes in antibody conformations, which might be reflected in pIs, peak shape, and absorbance of the antibody [39]. In this study, the stability of bulk materials of trastuzumab biosimilar was determined at 4°C for 15 and 30 days, and an increased temperature of 37°C for 4 and 24 h. As can be seen from Table 3, at 37°C within 24 h, the pI of main peak and content of charge variants were relatively stable. And at 4°C in 30 days, the pI of main peak was slightly decreased and the acidic charge variants increased by 6.3%, while basic and main components decreased by 0.5 and 5.8%, respectively. The observed change in area percent or pIs after incubation at different conditions was probably due to deamidation or aggregation.
3.4.2 Lots consistency Lot-to-lot consistency was evaluated by this optimum method and there were no differences in the CIEF preparations of each individual lot. Three lots of bulk material of trastuzumab biosimilar were studied and three preparations for each lot were injected, and the data were listed in Table 4. The RSDs for area percent of charge variants varied more extensively than pI of main component (1.28–7.03% vs. 0.323%). The charge variants among different lots were closely related to the process of manufacturing, such as cell culture, purification, formulation, and storage. For instance, manufacturing www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1461–1468
1467
Table 4. Lots consistency
A
Area percent (%)
Lot 1
Lot 2
Lot 3
Sample 1 Sample 2 Sample 3 RSD (%) Sample 1 Sample 2 Sample 3 RSD (%) Sample 1 Sample 2 Sample 3 RSD (%)
RSD (%) (n = 9)
Basic
Main
Acidic
pI
3.810 3.850 3.760 1.19 3.990 3.960 4.080 1.56 4.440 4.530 4.360 1.91 7.03
56.04 55.21 55.05 0.959 56.02 55.33 56.03 0.719 57.42 56.32 56.05 1.28 1.28
40.15 40.94 41.19 1.33 39.99 40.71 39.89 1.11 38.14 39.15 39.59 1.91 2.37
8.41 8.46 8.46 0.342 8.44 8.41 8.39 0.299 8.46 8.46 8.45 0.0683 0.323
B
Table 5. Biosimilar treatment with PNGase F
Area percent (%)
Without PNGase F PNGase F
Basic
Main
Acidic
pI
4.440 7.520
55.09 53.45
40.47 39.03
8.47 8.14
in glucose-rich culture media where glucose or lactose could react with the primary amine of a Lys residue easily generated acidic variants. Other mechanisms for generating acidic variants were deamidation and sialyation. Formation of the basic variants could result from the presence of C-terminal Lys or glycine amidation, succinimide formation, amino acid oxidation, or removal of sialic acid, which introduced additional positive or removal of negative charges and caused an increase in pI values [40].
3.4.3 Digest by enzyme PNGase F could selectively remove N-linked carbohydrates from glycosylated proteins. Neutral oligosaccharides do not bring in any charge variants on a protein unless sialic acid is present [41]. If the mAb contains sialic acid, deglycosylation would cause the loss of the negatively charged variants and the deglycosylated variants should show higher pI values than the original one. However, it was surprising that after treating mAb with PNGase F, all the peaks shifted toward a more acidic region (Table 5 and Fig. 9A), which was mainly induced by the hydrolyzation of asparagine (pI = 5.41) linked with N-glycans into more acidic aspartic acid (pI = 2.77) [42]. From Fig. 9A, the number of basic charge variants increased while acidic variants decreased, and the content of basic variants also obviously increased after deglycosylation (see Table 5). This was probably caused by the loss of the negatively charged variants, and the deglycosylated variants showed higher pIs. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 9. CIEF profiles of trastuzumab biosimilar digest by enzyme; (A) Digest with PNGase F, (B) Digest with CpB. For sample compositions and experimental conditions, see Section 2.
These results indicated that trastuzumab biosimilar should have sialic acid containing N-glycans in its structure. The optimum CIEF method could distinguish two charge profiles between intact and deglycosylated preparations even if they had slightly difference (⌬ pI = 0.33). CpB could selectively cleave Lys and Arg residues at the C-terminal end of proteins and peptides. As shown in Fig. 9B, when mAb was treated with CpB, the two basic peaks faded away, which significantly illustrated that two basic peaks observed in CIEF analysis originated from incomplete posttranslation cleavage of C-terminal Lys.
4 Concluding remarks CIEF has become one of the important methods to analyze charge heterogeneity of mAbs. In this study, the sample compositions had significant influence on separation, especially urea, CAs, and Arg. Focusing time made little impact on separation while capillary temperature had great effect on it. The results of method validation implied that this optimum method of analyzing trastuzumab biosimilar charge variants www.electrophoresis-journal.com
1468
J. Cao et al.
showed good precision with RSD ⬍ 3.0%. The optimal CIEF method was applied to evaluate mAbs’ charge heterogeneity, stability, determination of lot consistency, and purity assessment, which demonstrated that this method was a promising tool for the routine analysis of therapeutic mAbs. We would like to thank Analytical Research members at Shandong Luye Laboratories for their support.
Electrophoresis 2014, 35, 1461–1468
[18] Tang, S., Nesta, D. P., Maneri, L. R., Anumula, K. R., J. Pharm. Biomed. Anal. 1999, 19, 569–583. [19] Shen, Y., Berger, S. J., Andersen, G. A., Smith, R. D., Anal. Chem. 2000, 72, 2154–2159. [20] Quan, C., Alcala, E., Petkovska, I., Matthews, D., CanovaDavis, E., Taticek, R., Ma, S., Anal. Biochem. 2008, 373, 179–191. [21] Shimura, K., Electrophoresis 2009, 30, 11–28. [22] Righetti, P. G., Electrophoresis 2006, 27, 923–938.
The authors have declared no conflict of interest.
[23] Rodriguez-Diaz, R., Zhu, M., Wehr, T., J. Chromatogr. A 1997, 772, 145–160. [24] Mack, S., Cruzado-Park, I., Chapman, J., Ratnayake, C., Vigh, G., Electrophoresis 2009, 30, 4049–4058.
5 References [1] Ratanatharathorn, V., Pavletic, S., Uberti, J. P., Cancer Treat. Rev. 2009, 35, 653–661. [2] Singh, J. A., Christensen, R., Wells, G. A., SuarezAlmazor, M. E., Buchbinder, R., Lopez-Olivo, M. A., Ghogomu, E. T., Tugwell, P., CMAJ 2009, 181, 787–796. [3] Sautto, G., Mancini, N., Gorini, G., Clementi, M., Burioni, R., Biomed. Res. Int. 2013, 2013, 1–21. [4] Ghosh, R., Narasanna, A., Wang, S. E., Liu, S., ´ Chakrabarty, A., Balko, J. M., Gonzalez-Angulo, A. M., Mills, G. B., Penuel, E., Winslow, J., Sperinde, J., Dua, R., Pidaparthi, S., Mukherjee, A., Leitzel, K., Kostler, W. J., Lipton, A., Bates, M., Arteaga, C. L., Cancer Res. 2011, 71, 1871–1882. [5] Tanner, M., Kapanen, A. I., Junttila, T., Raheem, O., Grenman, S., Elo, J., Elenius, K., Isola, J., Mol. Cancer Ther. 2004, 3, 1585–1592. ¨ [6] Baroket, M., Tanner, M., Koninki, K., Isola, J., Breast Cancer Res. 2011, 13, R46.
[25] Lin, J., Tan, Q., Wang, S., J. Sep. Sci. 2011, 34, 1696–1702. [26] Poitevin, M., Morin, A., Busnel, J. M., Descroix, S., Hennion, M. C., Peltre, G., J. Chromatogr. A 2007, 1155, 230–236. [27] Anderson, C. L., Wang, Y., Rustandi, R. R., Electrophoresis 2012, 33, 1538–1544. [28] Lin, J., Lazar, A. C., Methods Mol. Biol. 2013, 1045, 295–302. [29] Salas-Solano, O., Kennel, B., Park, S. S., Roby, K., Sosic, Z., Boumajny, B., Free, S., Reed-Bogan, A., Michels, D., McElroy, W., Bonasia, P., Hong, M., He, X., Ruesch, M., Moffatt, F., Kiessig, S., Nunnally, B., J. Sep. Sci. 2012, 35, 3124–3129. [30] Rodriguez-Diaz, R., Wehr, T., Zhu, M., Electrophoresis 1997, 18, 2134–2144. [31] Conti, M., Galassi, M., Bossi, A., Righetti, P. G., J. Chromatogr. A 1997, 757, 237–245. [32] Zhu, M., Rodrigue, R., Wehr, T., J. Chromatogr. 1991, 599, 479–488.
[7] Seo, J., Lee, K. J., J. Biochem. Mol. Biol. 2004, 37, 35–44.
[33] Wu, J., Li, S. C., Watson, A., J. Chromatogr. A 1998, 817, 163–171.
´ [8] Beck, A., Sanglier-Cianferani, S., Van Dorsselaer, A., Anal. Chem. 2012, 84, 4637–4646.
[34] Bhuyan, A. K., Biochem. 2002, 41, 13386–13394.
[9] Dick, L. W. Jr., Qiu, D., Mahon, D., Adamo, M., Cheng, K. C., Biotechnol. Bioeng. 2008, 100, 1132–1143. [10] Michels, D. A., Salas-Solano, O., Felten, C., Bioprocess Int. 2011, 9, 48–54. [11] Rea, J. C., Wang, Y. J., Moreno, T. G., Parikh, R., Lou, Y., Farnan, D., Innovations in Biotechnology, InTech Press, 2012, pp. 439–464. [12] Wenisch, E., Reiter, S., Hinger, S., Steindl, F., Tauer, C., Jungbauer, A., Katinger, H., Righetti, P. G., Electrophoresis 1990, 11, 966–969.
[35] Kronsteiner, B., Malissa, H. Jr., Stutz, H., Electrophoresis 2007, 28, 2241–2251. [36] Cifuentes, A., Moreno-Arribas, M. V., de Frutos M, D´ıezMasa, J. C., J. Chromatogr A. 1999, 830, 453–463. ¨ [37] Graf, M., Watzig, 2959–2964.
H.,
Electrophoresis
2004,
25,
[38] Bonn, R., Rampal, S., Rae, T., Fishpaugh, J., Electrophoresis 2013, 34, 825–832. [39] Dai, H. J., Krull, I. S., J. Chromatogr. A 1998, 807, 121–128.
´ F., Electrophoresis 2003, 24, 3908–3916. [15] Kilar,
[40] Khawli, L. A., Goswami, S., Hutchinson, R., Kwong, Z. W., Yang, J., Wang, X., Yao, Z., Sreedhara, A., Cano, T., Tesar, D., Nijem, I., Allison, D. E., Wong, P. Y., Kao, Y. H., Quan, C., Joshi, A., Harris, R. J., Motchnik, P., MAbs 2010, 2, 613–624.
[16] Michels, D. A., Salas-Solano, O., Felten, C., Bioprocess Int. 2011, 9, 48–54.
[41] Li, N., Kessler, K., Bass, L., Zeng, D., J. Pharm. Biomed. Anal. 2007, 43, 963–972.
[17] Maeda, E., Urakami, K., Shimura, K., Kinoshita, M., Kakehi, K., J. Chromatogr. A 2010, 1217, 7164–7171.
[42] Santora, L. C., Krull, I. S., Grant, K., Anal. Biochem. 1999, 275, 98–108.
[13] Righetti, P. G., Bossi, A., Anal. Biochem. 1997, 247, 1–10. [14] He, Y., Isele, C., Hou, W., Rueschetal, M., J. Sep. Sci. 2011, 34, 548–555.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.electrophoresis-journal.com
Electrophoresis 2014, 35, 1461–1468
Junzi Cao1 Wen Sun2 Feifei Gong1 Wanhui Liu1,2 1 School
of Pharmacy, Yantai University, Yantai, P. R. China Key Laboratory of Long-acting and Targeting Drug Delivery System, Luye Pharma Group Ltd, Yantai, P. R. China
2 State
Received September 28, 2013 Revised February 19, 2014 Accepted February 24, 2014
Research Article
Charge profiling and stability testing of biosimilar by capillary isoelectric focusing CIEF was developed for the rapid analysis of charge heterogeneity of trastuzumab biosimilar using commercially available fluorocarbon-coated capillary. The CIEF master mix was composed of 0.30% w/v methyl cellulose, 2.3 M urea, 56.8 mM L-arginine, 1.52 mM iminodiacetic acid, 4.5% v/v carrier ampholytes (broad-range pI 3–10 and narrow-range pI 8–10.5 with ratio of 3:1), and 0.45% v/v 10.0, 9.5, 7.0, 5.5, 4.1 pI markers. To get a robust method to analyze charge heterogeneity, some separation parameters, including focusing time and separation temperature, were investigated and optimized. The optimized method gave good precision in estimated pI values of charge variants with RSDs of not more than 0.16% intraday analysis (n = 6) and ⬍ 0.18% interday analysis (n = 9). In addition, the applications of this method including purity, stability, lot consistency, peptide N-glycosidase F digest, and C-terminal lysine variants characterization were also investigated. Keywords: Charge heterogenity / CIEF / Monoclonal antibody / Trastuzumab biosimilar DOI 10.1002/elps.201300471
1 Introduction MAbs have drawn significant attention in the last few decades to treat inflammatory, oncologic, and autoimmune diseases [1–3]. Trastuzumab is developed by Roche (Indianapolis, IN, USA) for treatment of breast cancer patients who show an overexpression of HER2 [4]. Compared with trastuzumab, the biosimilar has only two different amino acids, which vary in heavy chain at the sites of 359 and 361, aspartic acid, and leucine instead of glutamic acid and methionine. The effects of trastuzumab and trastuzumab biosimilar on growth of Bt474 cells were evaluated using the method of AlamarBlue [5, 6]. Fluorescence was detected at an excitation of 560 nm and emission at 590 nm using a Microplate Reader (TECAN, Switzerland). The cellular activity of biosimilar was equivalent to that of trastuzumab. The biosimilar is promising to be a candidate biological drug. Analysis or characterization of mAbs is a challenging task, because they are complex compounds with heterogeneity in size and charge. Charge heterogeneity can result from multiple sources such as deamidation, formation of N-terminal pyroglutamate, aggregation, isomerization, sialylated glycans, antibody fragmentation, and glycation at the lysine (Lys) residues [7–9]. In some cases, such changes affect binding, biological activity, patient safety, and shelf life [10].
Correspondence: Dr. Wanhui Liu, School of Pharmacy, Yantai University, Yantai, P. R. China E-mail: [email protected] Fax: +86-535-3808293
Abbreviations: CA, carrier ampholyte; CpB, carboxypeptidase B; iCIEF, imaged CIEF; MC, methyl cellulose; MT, migration time; PNGase F, peptide N-glycosidase F C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Therefore, analyzing charge heterogeneity of mAbs is critical for characterizing and monitoring quality attributes of antibodies. Characterizing charge variants in the biopharmaceutical field relies on tools, such as IEC [11], IEF gel electrophoresis [12], IPGs [13] and CE including CZE [14], CIEF [15], and imaged CIEF (iCIEF) [16]. Among these methods, IEF, IPGs, CIEF, and iCIEF separate proteins or antibodies are based on pIs. CIEF and iCIEF have the advantage over traditional gel IEF. First, the use of slab gels is known to suffer from lack of investigation in the course of validation, as well as accuracy and reproducibility for quantitative measurement of the separated charged components. Second, CIEF has proven its ability to combine the resolving power of IEF with advantages of CE quantitation and automation [17–19]. Compared with CIEF, IPGs are widely used in 2D electrophoresis, but are labor-intensive and difficult to quantification. Given the advantages of CIEF, separations based on pIs have been widely used in mAbs analysis for identification, purity assessment, degradation checking, microheterogeneity analysis, and clinical diagnosis [9, 15, 20]. Besides, because of the extremely high peak capacity of CIEF, it is selected as the more preferred method for the analysis of mAbs. CIEF has two general modes of performance, that is, onestep and two-step CIEF, where focusing and mobilization occur simultaneously or sequentially [21–23]. The latter was used in this study. To get a much higher reproducibility and reduced chance of accidental protein adsorption to the silica wall, coated capillaries, such as fluorocarbon [10], DB-1 or DB17 [17], polyacrylamide (neutral) [24], PVA [25], and cellulous derivants [26] coated capillaries, are used in CIEF or iCIEF. Fluorocarbon-coated capillary, which is widely used in iCIEF [10, 27–29], was selected in our study and a high separation efficiency was obtained. www.electrophoresis-journal.com
1462
J. Cao et al.
Apart from coated capillaries, the key factors such as sample preparations and separation conditions are also related to separation efficiency in CIEF. Therefore, to establish a robust method, sample compositions, such as urea, methyl cellulose (MC), L-arginine (L-Arg), carrier ampholytes (CAs), and separation conditions including focusing time, separation temperature, were studied. A robust CIEF method for biosimilar was established based on the standard separation previously published by Mack S. et al [24]. Precision studies were conducted to demonstrate that the closeness of results could be obtained over a short time under the same operating conditions. Finally, CIEF applications, including the stability of mAb, mAb digested by enzymes, purity, and lot-to-lot consistency, were also performed.
2 Materials and methods
Electrophoresis 2014, 35, 1461–1468
centrifuge (Thermo, Fresco 21, Rockford, IL, USA) speed of 13 500 rpm/min, under the temperature of 4°C. After removing salts, the final volume of the deglycosylation solution was about 70 L and the concentration of mAb was 2.75 mg/mL which was determined by micro-ultraviolet spectrophotometer (Thermo, nanodrop 2000).
2.2.2 Trastuzumab biosimilar digested by CpB MAb (200 g) was mixed with 1 U (activity units) CpB in 1.0 mM phosphate, 10 mM NaCl solution, pH 8.0, and incubated at 37°C for 2 h. As a reference sample, water was added instead of the enzyme. After incubation, remove the salt at the same conditions as PNGase F’s. The concentration of mAb was 2.75 mg/mL.
2.1 Materials and instrumentation 2.2.3 CIEF master mix preparation Experiments were performed on Agilent 7100 CE (external pressure 2–12 bar) equipment with DAD and a 280 nm filter, and the bandwidth was 4 nm. Data were collected and analyzed using Chemstation (Agilent, version 2009). The temperature setting was 20°C. Fluorocarbon-coated capillary was obtained from Agilent Technology (The total length is 33 cm, 24.5 cm effective length, 50 m id). Pharmalyte pH 3–10 CAs (PN 17–0456–01, 0.36 meq/mL) and pH 8–10.5 CAs (PN 17–0455–01, 0.36 meq/mL) were acquired from GE Healthcare (Piscataway, NJ, US) and pI markers 10.0, 9.5, 7.0, 5.5, and 4.1 (PN A58481) were obtained from Beckman Coulter (Brea, CA, US). Urea (PN U6504), carboxypeptidase B (CpB, PN C9548), Triton X-100 (PN T8532), sodium chloride (NaCl, PN S7653), MC (PN M0262), acetic acid (purity ࣙ99.7%, PN 45727), and iminodiacetic acid (PN 220000) were supplied by Sigma-Aldrich (St. Louis, MO, US). Peptide N-glycosidase F (PNGase F, PN 11365193001) was purchased from Roche. Sodium hydroxide (purity ࣙ95.0%, PN 567530) and L-Arg (purity ࣙ99.0%, PN 1820) were from Merck Millipore (Darmstadt, Germany). Phosphoric acid (purity ࣙ85%, PN E582) was brought from Amresco (Solon, IA, US). Milli-Q water was made in-house. Trastuzumab biosimilar was provided by pharmaceutical laboratory of Shandong Luye Pharma Co. 2.2 Sample preparation 2.2.1 Deglycosylation of N-linked oligosaccharides from trastuzumab biosimilar Deglycosylation of trastuzumab biosimilar was conducted as follows. MAb (200 g) was mixed with 5 U (activity units) PNGase F in 1.0 mM phosphate, 10 mM NaCl, 0.5% v/v Triton X-100 solution, pH 7.5 and incubated at 37°C for 24 h. As a reference sample, water was added instead of the enzyme. After incubation, a spin filter (5 kD cut-off, Sartorius) was used to remove salts. The retentate was washed three times with 70 L of Milli-Q water for 20 min each time at the C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
To prepare CIEF samples, the following reagents were mixed in a centrifuge tube: 200 L of 3 M urea containing 0.4% w/v MC, 9.0 L of pharmalyte 3–10 CAs, 3.0 L of pharmalyte 8–10.5 CAs, 20.0 L of cathodic stabilizer (750 mM Arg), and 2.0 L of anodic stabilizer (200 mM iminodiacetic acid).
2.2.4 CIEF sample preparation A 97.5 L of master was mixed with 10 L of protein with concentration of 2.75 mg/mL and 2.5 L of pI markers mixture, which contained 0.5 L of each pI marker. The final sample concentration was 0.25 mg/mL.
2.3 Method The fluorocarbon-coated capillary was successively rinsed with water for 5 min, 350 mM acetic acid for 2 min, and 0.4% w/v MC solution for 3 min at a pressure of 3.5 bar at the beginning of each day. To decrease protein precipitate, before each injection the capillary was flushed with 4.3 M urea for 3 min, water for 2 min, at a pressure of 3.5 bar, respectively. Focusing was performed at 25 kV in normal polarity for 12.5 min. A 200 mM phosphoric acid and 300 mM sodium hydroxide was used as anode and cathode solution, respectively. After focusing, the cathode solution was changed with 350 mM acetic acid, and the charge variants were mobilized to cathode at 30 kV in normal polarity. The sample was injected for 2 min at a pressure of 1 bar. A shutdown method was run at the end of each day to rinse the capillary for 5 min at a pressure of 5 bar with 4.3 M urea, followed by water for 10 min at the same pressure. The DAD lamp was switched off and both of the capillary ends were immersed in Milli-Q water. And the capillary was stored at room temperature overnight. www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1461–1468
1463
3 Results and discussion 3.1 Sample composition 3.1.1 Urea Protein precipitation causes a lack in precision in migration time (MT), peak area (thus affecting quantitation), capillary clogging, unstable current, slow mobilization, and other undesirable effects [30, 31]. Therefore, the additives, such as urea [24], Triton X-100 [32], or glycerol [33], are used to improve the solubility of the focused mAbs and prevent precipitation before they move past the detection window. In this research, urea was studied and the optimal concentration ultimately depended on the solubility of the mAb. The CIEF profiles of trastuzumab biosimilar were evaluated at different urea concentrations. As shown in Fig. 1, in the environment lacking of urea, the separation profile for biosimilar was poorly defined, presumably because of its low solubility. It was sufficient to increase the solubility of the protein sample and prevent aggregate formation by breaking hydrogen–bond interactions at the concentration of 2.3 M (The subdenaturing concentration is not more than 4 M [34]). The separation efficiency, especially to basic variants, dramatically degraded at the concentration of 4.6 M, possibly due to protein denaturation. Besides, higher concentrations most likely unfolded the protein and thus affecting differences in charge due to conformational shifts. Figure 1 showed that the MT extended with concentration increasing because separation medium viscosity increased. Therefore, the optimal concentration of urea was determined to be 2.3 M. 3.1.2 Methyl cellulose MC was used not only as a viscosity-enhancing agent to influence longitudinal diffusion, but also as a neutral hy-
Figure 2. Effect of MC concentration on separation; (A) 0.15% w/v; (B) 0.30% w/v; (C) 0.60% w/v. For other sample compositions and experimental conditions, see Fig. 1.
drophilic polymer used to dynamically coat the capillary walls to minimize analyte–wall interaction and helped suppress EOF. When the concentration of MC increased from 0.15 to 0.60% w/v, the MT increased (Fig. 2), because MC could enhance the viscosity of separation medium. At the concentration of 0.30% w/v, the separation efficiency and resolution was greatly improved, especially for basic variants. However, increasing the concentration of MC from 0.30 to 0.60% w/v, the resolution for basic charge variants degraded. Furthermore, high concentration of MC not only easily led to capillary clogging, which in turn caused poor separation, but also made pipetting more difficult, as a result, reducing the precision of the measurements. Based on the experimental results, the optimized concentration of MC was found to be 0.30% w/v, which could produce adequate separation efficiency and precision.
3.1.3 Arginine
Figure 1. Effect of urea concentration on separation; (A) 0 M; (B) 2.3 M; (C) 4.6 M. Sample concentration: 0.25 mg/mL, containing 0.3% w/v MC, 4.5% v/v pI 3–10 CAs, 37.9 mM Arg, 1.52 mM iminodiacetic acid. A 25 kV focusing for 15 min, 30 kV separation at 25°C. For other conditions see Section 2.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Arg (pI 10.7) was cathodic stabilizer, which was used to prevent the loss of the ampholytes into the electrode solutions. The amount of cathodic stabilizer within the sample needed to be optimized to ensure alkaline compounds are not focused behind the detector. In addition, the amount of cathodic stabilizer probably affected resolution. The concentrations at 0, 37.9, 56.8, 75.8, and 89.1 mM were investigated. An abnormal profile was observed without Arg, because the mAb peaks were detected during focusing and the pI markers of 10.0 and 9.5 did not appear in the electropherogram (Fig. 3). MT from 0 to 5 min was concerned. There were no peaks in trace A, and peak I and peak II appeared in trace B with addition of pI markers, and there was no additional peaks in trace C except that the height of peak I and II increased with increasing concentration of markers. This demonstrated that peak I and peak II were cathodic peaks stemming from pI markers. www.electrophoresis-journal.com
1464
J. Cao et al.
Figure 3. Effect of Arg on pI markers of 10.0 and 9.5; (A) mAb without pI markers and Arg; (B) 0.19% v/v pI markers (10.0, 9.5, 7.0, 5.5, 4.1) without mAb and Arg; (C) 0.25 mg/mL mAb and 0.58% v/v pI markers (10.0, 9.5, 7.0, 5.5, 4.1) without Arg; (D) 0.25 mg/mL mAb and 0.45% v/v pI markers (10.0, 9.5, 7.0, 5.5, 4.1) with 37.9 mM Arg.
Electrophoresis 2014, 35, 1461–1468
Figure 5. Effect of CAs compositions on separation; (A) 4.5% v/v pI 3–10 CAs; (B) 4.5% v/v CAs, pI 3–10 and pI 8–10.5 with ratio of 3:1; (C) 4.5% v/v CAs, pI 3–10 and pI 8–10.5 with ratio of 2:2; (D) 4.5% v/v CAs, pI 3–10 and pI 8–10.5 with ratio of 1:3. For other sample compositions and experimental conditions, see Fig. 1.
3.1.4 Ampholyte
Figure 4. Effect of Arg concentration on separation; (A) 37.9 mM; (B) 56.8 mM; (C) 75.8 mM; (D) 89.1 mM. For other sample compositions and experimental conditions, see Fig. 1.
The pI markers of 10.0 and 9.5 were detected in presence of 37.9 mM Arg (Fig. 3, trace D), because Arg fulfilled a blocking function occupying the region between the detection window and the capillary outlet. Moreover, the signal heights especially to mAb and cathodic peaks increased compared with that without Arg (Fig. 3, trace C), probably because Arg could compress the pH gradient. Figure 4 indicated that with the concentration of Arg increasing, the number of peaks obviously increased and maximized at the concentration of 56.8 mM. A 75.8 mM or higher Arg was an excessive amount of cathodic stabilizer, because it not only delayed the detection of mAb, but also led to a loss in resolution. The optimum concentration of Arg was determined to be 56.8 mM. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Ampholytes were crucial to the success of CIEF since ampholytes established the pH gradient that affected resolution of the protein in the capillary. So, the proper selection of CAs and their concentrations was found to be essential. The concentration of CAs was chosen to be 4.5% after referring to the CIEF application guide analysis kit (Beckman, A78788AE). The mixture of the wide-range ampholytes (3–10) and narrowrange (8–10.5) was used to overcome the limitation of using single ampholytes (3–10) in detecting charge heterogeneity of basic mAbs [35,36]. As shown in Fig. 5, by increasing the percentage of narrow-range ampholytes (8–10.5), resolution was markedly enhanced when MT prolonged. However, increasing the concentration of narrow-range pharmalyte (8–10.5) to 2.25% or higher caused split peaks for pI marker of 9.5, which implied that a longer focusing time was needed. But increasing focusing time would increase the risk of precipitation. The best compromise between MT and separation efficiency was obtained with 3:1 v/v mixtures of broad and narrow pH range CAs. 3.2 Separation condition 3.2.1 Focusing time Because the pH gradient was homogeneous at the start of the separation, the focusing time must be sufficient to allow the cathodic and anodic peaks for each component to merge. The best focusing time was directly related to the separation efficiency. An excess of focusing time would waste time, easily cause protein precipitate, and shorten the lifespan of coated capillaries. Furthermore, insufficient focusing time would lead to partial detection or split peaks. Focusing time that varied from 10 to 17.5 min was investigated to explore the www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1461–1468
1465
A
Figure 6. Effect of focusing time on separation; (A) 10 min; (B) 12.5 min; (C) 15 min; (D) 17.5 min. For sample composition and other experimental conditions, see Section 2 and Fig. 1, respectively.
B
Figure 8. Specificity of trastuzumab biosimilar; (A) Specificity; (B) Enlargement of trace mAb and markers from 25 to 29 min. For sample compositions and experimental conditions, see Section 2. Figure 7. Effect of separation temperature on separation; (A) 15°C; (B) 20°C; (C) 25°C; (D) 30°C. For sample composition and other experimental conditions, see Section 2 and Fig. 1, respectively.
proper value. Figure 6 showed that the number of charge variants maximized at focusing at 12.5 min or longer. A total of 12.5 min were enough for this separation, and no split peaks were detected in the electropherograms. Therefore, the focusing time at 12.5 min was selected as the prime choice. 3.2.2 Temperature Separation temperature had a significant effect on the resolution, because it influenced not only solubility of sample, but also Joule heat, and both of them were related to separation efficiency. Temperatures from 15 to 30°C with 5°C as a step were investigated. Figure 7 indicated that MT decreased and resolution degraded with increasing temperature. It is more likely that higher temperatures reduced viscosity, enhanced longitudinal diffusion, and thus peak broadening. It seemed C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
that 15°C was the best choice. However, the MTs of mAb shifted more greatly (variation range: 0.45 vs. 0.04 min) at 15°C than 20°C within three injections. Therefore, 20°C was chosen as the ideal temperature.
3.3 Method validation 3.3.1 Specificity The purpose of specificity test was to discriminate between the target peaks and other interference peaks from pI markers or master mix. Bulk materials of trastuzumab biosimilar, the concentrated protein solutions that are prior to formulation and final manufacturing were studied here. All injections were performed using the improved CIEF method. As shown in Fig. 8A, there was no interference to analyze the target mAb in the master mix analysis. And addition of the pI markers also did not bring in any other interference peaks. Charge variants of trastuzumab biosimilar were shown in Fig. 8B. www.electrophoresis-journal.com
1466
J. Cao et al.
Electrophoresis 2014, 35, 1461–1468 Table 3. Stability under different temperatures (n = 3)
Table 1. pI precisions test of the CIEF method for mAb
Peak 1 Peak 2 Peak 3 Peak 4 Peak 5 Peak 6 Peak 7 Peak 8 Intraday precisiona) Mean 8.64 8.56 RSD (%) 0.084 0.061
8.46 0.12
8.25 0.12
8.18 0.14
8.10 0.16
8.03 0.14
7.94 0.15
Interday precisionb) Mean 8.64 8.55 RSD (%) 0.12 0.097
8.45 0.16
8.25 0.14
8.18 0.14
8.10 0.17
8.03 0.18
7.94 0.16
a) One sample preparation and six replicate injections in one day. b) One sample preparation and three injections each in three separate days. Table 2. Area percent (%) precisions test of the CIEF method for mAb
Basic
Main
Acidic
Intraday precisiona) Mean RSD (%)
3.940 2.83
56.01 1.17
40.05 1.69
Interday precisionb) Mean RSD (%)
3.960 3.01
55.64 1.45
40.40 1.94
Area percent (%) Sample
Basic
Main
Acidic
pI
Control 37°C 4 h 37°C 24 h 4°C 15 days 4°C 30 days
4.040 4.060 4.240 3.850 3.540
56.75 55.05 54.31 54.21 50.97
39.21 40.89 41.45 41.94 45.49
8.47 8.46 8.47 8.44 8.43
precision for their low content. Compared with pI precisions, area% precisions were more variable. This effect is mainly attributed to instability and progressive degradation of the coating with subsequent enhanced protein adsorption which will continuously impair precision (e.g. pI, peak, area and so on) [37, 38].
3.4 Applications 3.4.1 Stability
a) One sample preparation and six replicate injections in one day. b) One sample preparation and three injections each in three separate days.
3.3.2 Precision To investigate the precision of the optimized CIEF method, a precision study was performed using trastuzumab biosimilar as a model. Intraday precision was evaluated by performing a single sample preparation and six replicate injections in one day. Interday precision was evaluated by performing one sample preparation that was immediately stored in the 4°C refrigerator after three injections every day, and three injections per day for three consecutive days. Sample preparations and separation conditions for intraday and interday precision were the same as specificity. CIEF method yielded a pH gradient over a wide pI range in the capillary. A typical calibration plot of MTs versus pI values of the markers, and a linear calibration curve was obtained for pI markers ranging from 10.0 to 4.1 (R ⬎ 0.99). The pIs of the mAb isoforms were estimated from the calibration curve according to their MTs. As shown in Table 1, the intra- and interday precision for pI values of charge variants had good precisions with RSD of not more than 0.18%. CIEF was applied to purity assessment of mAbs by area normalization method. The percent isoform group composition values were calculated by dividing the sum of the peak areas for each isoform group by the total mAb peak area and multiplying by 100. Table 2 showed that intraday and interday precision for main component and acidic variants had good precision with RSD ⬍ 2.0%, while basic variants had poor C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The stability study referred to the standardized protocol established in house. The stability of therapeutic antibodies during downstream processing and storage was important for functionality and quality. The purpose of stability test was to provide evidence on how a protein product changes with time under the influence of different factors such as light, pH, and temperature. The increase of temperature might cause changes in antibody conformations, which might be reflected in pIs, peak shape, and absorbance of the antibody [39]. In this study, the stability of bulk materials of trastuzumab biosimilar was determined at 4°C for 15 and 30 days, and an increased temperature of 37°C for 4 and 24 h. As can be seen from Table 3, at 37°C within 24 h, the pI of main peak and content of charge variants were relatively stable. And at 4°C in 30 days, the pI of main peak was slightly decreased and the acidic charge variants increased by 6.3%, while basic and main components decreased by 0.5 and 5.8%, respectively. The observed change in area percent or pIs after incubation at different conditions was probably due to deamidation or aggregation.
3.4.2 Lots consistency Lot-to-lot consistency was evaluated by this optimum method and there were no differences in the CIEF preparations of each individual lot. Three lots of bulk material of trastuzumab biosimilar were studied and three preparations for each lot were injected, and the data were listed in Table 4. The RSDs for area percent of charge variants varied more extensively than pI of main component (1.28–7.03% vs. 0.323%). The charge variants among different lots were closely related to the process of manufacturing, such as cell culture, purification, formulation, and storage. For instance, manufacturing www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 1461–1468
1467
Table 4. Lots consistency
A
Area percent (%)
Lot 1
Lot 2
Lot 3
Sample 1 Sample 2 Sample 3 RSD (%) Sample 1 Sample 2 Sample 3 RSD (%) Sample 1 Sample 2 Sample 3 RSD (%)
RSD (%) (n = 9)
Basic
Main
Acidic
pI
3.810 3.850 3.760 1.19 3.990 3.960 4.080 1.56 4.440 4.530 4.360 1.91 7.03
56.04 55.21 55.05 0.959 56.02 55.33 56.03 0.719 57.42 56.32 56.05 1.28 1.28
40.15 40.94 41.19 1.33 39.99 40.71 39.89 1.11 38.14 39.15 39.59 1.91 2.37
8.41 8.46 8.46 0.342 8.44 8.41 8.39 0.299 8.46 8.46 8.45 0.0683 0.323
B
Table 5. Biosimilar treatment with PNGase F
Area percent (%)
Without PNGase F PNGase F
Basic
Main
Acidic
pI
4.440 7.520
55.09 53.45
40.47 39.03
8.47 8.14
in glucose-rich culture media where glucose or lactose could react with the primary amine of a Lys residue easily generated acidic variants. Other mechanisms for generating acidic variants were deamidation and sialyation. Formation of the basic variants could result from the presence of C-terminal Lys or glycine amidation, succinimide formation, amino acid oxidation, or removal of sialic acid, which introduced additional positive or removal of negative charges and caused an increase in pI values [40].
3.4.3 Digest by enzyme PNGase F could selectively remove N-linked carbohydrates from glycosylated proteins. Neutral oligosaccharides do not bring in any charge variants on a protein unless sialic acid is present [41]. If the mAb contains sialic acid, deglycosylation would cause the loss of the negatively charged variants and the deglycosylated variants should show higher pI values than the original one. However, it was surprising that after treating mAb with PNGase F, all the peaks shifted toward a more acidic region (Table 5 and Fig. 9A), which was mainly induced by the hydrolyzation of asparagine (pI = 5.41) linked with N-glycans into more acidic aspartic acid (pI = 2.77) [42]. From Fig. 9A, the number of basic charge variants increased while acidic variants decreased, and the content of basic variants also obviously increased after deglycosylation (see Table 5). This was probably caused by the loss of the negatively charged variants, and the deglycosylated variants showed higher pIs. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 9. CIEF profiles of trastuzumab biosimilar digest by enzyme; (A) Digest with PNGase F, (B) Digest with CpB. For sample compositions and experimental conditions, see Section 2.
These results indicated that trastuzumab biosimilar should have sialic acid containing N-glycans in its structure. The optimum CIEF method could distinguish two charge profiles between intact and deglycosylated preparations even if they had slightly difference (⌬ pI = 0.33). CpB could selectively cleave Lys and Arg residues at the C-terminal end of proteins and peptides. As shown in Fig. 9B, when mAb was treated with CpB, the two basic peaks faded away, which significantly illustrated that two basic peaks observed in CIEF analysis originated from incomplete posttranslation cleavage of C-terminal Lys.
4 Concluding remarks CIEF has become one of the important methods to analyze charge heterogeneity of mAbs. In this study, the sample compositions had significant influence on separation, especially urea, CAs, and Arg. Focusing time made little impact on separation while capillary temperature had great effect on it. The results of method validation implied that this optimum method of analyzing trastuzumab biosimilar charge variants www.electrophoresis-journal.com
1468
J. Cao et al.
showed good precision with RSD ⬍ 3.0%. The optimal CIEF method was applied to evaluate mAbs’ charge heterogeneity, stability, determination of lot consistency, and purity assessment, which demonstrated that this method was a promising tool for the routine analysis of therapeutic mAbs. We would like to thank Analytical Research members at Shandong Luye Laboratories for their support.
Electrophoresis 2014, 35, 1461–1468
[18] Tang, S., Nesta, D. P., Maneri, L. R., Anumula, K. R., J. Pharm. Biomed. Anal. 1999, 19, 569–583. [19] Shen, Y., Berger, S. J., Andersen, G. A., Smith, R. D., Anal. Chem. 2000, 72, 2154–2159. [20] Quan, C., Alcala, E., Petkovska, I., Matthews, D., CanovaDavis, E., Taticek, R., Ma, S., Anal. Biochem. 2008, 373, 179–191. [21] Shimura, K., Electrophoresis 2009, 30, 11–28. [22] Righetti, P. G., Electrophoresis 2006, 27, 923–938.
The authors have declared no conflict of interest.
[23] Rodriguez-Diaz, R., Zhu, M., Wehr, T., J. Chromatogr. A 1997, 772, 145–160. [24] Mack, S., Cruzado-Park, I., Chapman, J., Ratnayake, C., Vigh, G., Electrophoresis 2009, 30, 4049–4058.
5 References [1] Ratanatharathorn, V., Pavletic, S., Uberti, J. P., Cancer Treat. Rev. 2009, 35, 653–661. [2] Singh, J. A., Christensen, R., Wells, G. A., SuarezAlmazor, M. E., Buchbinder, R., Lopez-Olivo, M. A., Ghogomu, E. T., Tugwell, P., CMAJ 2009, 181, 787–796. [3] Sautto, G., Mancini, N., Gorini, G., Clementi, M., Burioni, R., Biomed. Res. Int. 2013, 2013, 1–21. [4] Ghosh, R., Narasanna, A., Wang, S. E., Liu, S., ´ Chakrabarty, A., Balko, J. M., Gonzalez-Angulo, A. M., Mills, G. B., Penuel, E., Winslow, J., Sperinde, J., Dua, R., Pidaparthi, S., Mukherjee, A., Leitzel, K., Kostler, W. J., Lipton, A., Bates, M., Arteaga, C. L., Cancer Res. 2011, 71, 1871–1882. [5] Tanner, M., Kapanen, A. I., Junttila, T., Raheem, O., Grenman, S., Elo, J., Elenius, K., Isola, J., Mol. Cancer Ther. 2004, 3, 1585–1592. ¨ [6] Baroket, M., Tanner, M., Koninki, K., Isola, J., Breast Cancer Res. 2011, 13, R46.
[25] Lin, J., Tan, Q., Wang, S., J. Sep. Sci. 2011, 34, 1696–1702. [26] Poitevin, M., Morin, A., Busnel, J. M., Descroix, S., Hennion, M. C., Peltre, G., J. Chromatogr. A 2007, 1155, 230–236. [27] Anderson, C. L., Wang, Y., Rustandi, R. R., Electrophoresis 2012, 33, 1538–1544. [28] Lin, J., Lazar, A. C., Methods Mol. Biol. 2013, 1045, 295–302. [29] Salas-Solano, O., Kennel, B., Park, S. S., Roby, K., Sosic, Z., Boumajny, B., Free, S., Reed-Bogan, A., Michels, D., McElroy, W., Bonasia, P., Hong, M., He, X., Ruesch, M., Moffatt, F., Kiessig, S., Nunnally, B., J. Sep. Sci. 2012, 35, 3124–3129. [30] Rodriguez-Diaz, R., Wehr, T., Zhu, M., Electrophoresis 1997, 18, 2134–2144. [31] Conti, M., Galassi, M., Bossi, A., Righetti, P. G., J. Chromatogr. A 1997, 757, 237–245. [32] Zhu, M., Rodrigue, R., Wehr, T., J. Chromatogr. 1991, 599, 479–488.
[7] Seo, J., Lee, K. J., J. Biochem. Mol. Biol. 2004, 37, 35–44.
[33] Wu, J., Li, S. C., Watson, A., J. Chromatogr. A 1998, 817, 163–171.
´ [8] Beck, A., Sanglier-Cianferani, S., Van Dorsselaer, A., Anal. Chem. 2012, 84, 4637–4646.
[34] Bhuyan, A. K., Biochem. 2002, 41, 13386–13394.
[9] Dick, L. W. Jr., Qiu, D., Mahon, D., Adamo, M., Cheng, K. C., Biotechnol. Bioeng. 2008, 100, 1132–1143. [10] Michels, D. A., Salas-Solano, O., Felten, C., Bioprocess Int. 2011, 9, 48–54. [11] Rea, J. C., Wang, Y. J., Moreno, T. G., Parikh, R., Lou, Y., Farnan, D., Innovations in Biotechnology, InTech Press, 2012, pp. 439–464. [12] Wenisch, E., Reiter, S., Hinger, S., Steindl, F., Tauer, C., Jungbauer, A., Katinger, H., Righetti, P. G., Electrophoresis 1990, 11, 966–969.
[35] Kronsteiner, B., Malissa, H. Jr., Stutz, H., Electrophoresis 2007, 28, 2241–2251. [36] Cifuentes, A., Moreno-Arribas, M. V., de Frutos M, D´ıezMasa, J. C., J. Chromatogr A. 1999, 830, 453–463. ¨ [37] Graf, M., Watzig, 2959–2964.
H.,
Electrophoresis
2004,
25,
[38] Bonn, R., Rampal, S., Rae, T., Fishpaugh, J., Electrophoresis 2013, 34, 825–832. [39] Dai, H. J., Krull, I. S., J. Chromatogr. A 1998, 807, 121–128.
´ F., Electrophoresis 2003, 24, 3908–3916. [15] Kilar,
[40] Khawli, L. A., Goswami, S., Hutchinson, R., Kwong, Z. W., Yang, J., Wang, X., Yao, Z., Sreedhara, A., Cano, T., Tesar, D., Nijem, I., Allison, D. E., Wong, P. Y., Kao, Y. H., Quan, C., Joshi, A., Harris, R. J., Motchnik, P., MAbs 2010, 2, 613–624.
[16] Michels, D. A., Salas-Solano, O., Felten, C., Bioprocess Int. 2011, 9, 48–54.
[41] Li, N., Kessler, K., Bass, L., Zeng, D., J. Pharm. Biomed. Anal. 2007, 43, 963–972.
[17] Maeda, E., Urakami, K., Shimura, K., Kinoshita, M., Kakehi, K., J. Chromatogr. A 2010, 1217, 7164–7171.
[42] Santora, L. C., Krull, I. S., Grant, K., Anal. Biochem. 1999, 275, 98–108.
[13] Righetti, P. G., Bossi, A., Anal. Biochem. 1997, 247, 1–10. [14] He, Y., Isele, C., Hou, W., Rueschetal, M., J. Sep. Sci. 2011, 34, 548–555.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.electrophoresis-journal.com
