High Throughput Profiling of Charge Heterogeneity in Antibodies by

Sep 20, 2011 - High throughput analytics are thus essential nowadays to provide adequate ... were calibrated by the two pI markers using the iCE280 so...
0 downloads 0 Views 3MB Size
Download PDF
ARTICLE pubs.acs.org/ac

High Throughput Profiling of Charge Heterogeneity in Antibodies by Microchip Electrophoresis Hongling Han, Eliza Livingston, and Xiaoyu Chen*,† Integrated Biologics Profiling, Novartis Pharmaceuticals, Cambridge, Massachusetts 02139, United States

bS Supporting Information ABSTRACT: A high throughput microchip capillary zone electrophoresis (CZE) method was developed for the analysis of charge heterogeneity in antibodies. The method utilizes high speed microchip electrophoresis separation and is well-suited for high throughput charge profiling of antibodies during process and formulation development. The method involves derivatization of protein molecules with Cy5 N-hydroxysuccinimide ester (NHS-ester), which does not change the protein charge profile and enables fluorescence detection on a commercial microchip instrument. The sample preparation can be performed in 96-well microtiter plates within 1 h, and each sample analysis takes only 80 s. Protein charge variants with a pI difference of 0.1 can be readily resolved in the 12.5 mm microfluidic channel. Charge profiles similar to those obtained using conventional CZE technology were found for all antibodies tested (pIs in the range of 7.5 9.2). The separation efficiency corresponds to 1.2  104 theoretical plates (1.0 μm plate height). Assay performance is assessed by demonstrating specificity, carryover, linearity, limit of detection, and precision.

P

rotein therapeutics are inherently heterogeneous due to their susceptibility to fragmentation, aggregation, disulfide shuffling, and other post-translational modifications. During bioprocessing and storage, many enzymatic and chemical reactions can occur that lead to charge variants of protein products.1 For example, acidic variants can be caused by deamidation, sialylation, glycation,2 and acid adduct formation,3 whereas basic variants can be caused by noncyclization of N-terminal glutamine,4,5 residual N-terminal signal peptides containing basic residues,6 C-terminal proline amidation,7 incomplete lysine processing,5,8 and succinimide formation from aspartic acid.9 These charge variants may have altered biological properties or they may impact the product shelf life. Charge heterogeneity is thus an important quality attribute of recombinant protein products that needs to be closely controlled and monitored during process development, manufacturing, and storage. The charge variant profile is also a sensitive indicator for both manufacturing process consistency and protein stability. A comprehensive understanding of the relationship between process parameters and product quality attributes is becoming a major focus of the biopharmaceutical industry in the era of qualityby-design. Manufacturing process development and formulation screening can be carried out in large sets of experiments, taking into account also the design-of-experiment (DoE) principles. The number of samples generated from such DoE studies can easily overwhelm an analytical laboratory using traditional methods. High throughput analytics are thus essential nowadays to provide adequate and timely analytical development support.10 In addition, platform assays are desirable in order to reduce method development time and to standardize sample preparation. Ideally, such platform assays are applicable to multiple products r 2011 American Chemical Society

and to a wide range of sample types including crude cell culture samples. Ion exchange chromatography (IEC) and isoelectric focusing (IEF) methods are commonly used to profile charge heterogeneity of protein products. IEC separations typically take 30 60 min and can be achieved using either a salt gradient or a pH gradient. Whereas salt gradient IEC methods are often product specific, pH gradient IEC has been demonstrated to be generically applicable for molecules of the same class such as antibodies.11 Slab gel isoelectric focusing can also be used for charge separation based on differences in protein isoelectric points. This technique, however, is both time-consuming and labor intensive. Isoelectric focusing in capillaries, especially the imaged capillary isoelectric focusing (icIEF) technique, has become more popular in recent years to monitor protein charge heterogeneity.12 15 The separation by icIEF typically takes 5 to 10 min, with a total instrument time per sample of 15 to 20 min. Capillary zone electrophoresis (CZE), with its great simplicity in operation, is an attracting alternative for protein charge profiling.16 In CZE, proteins or protein charge variants are separated based on differences in their electrophoretic mobilities. The capillary is often coated to eliminate the electroosmotic flow (EOF) and prevent nonspecific binding. Each injection typically takes 10 to 15 min of instrument time. Recently, CZE separation of multiple antibodies has been reported using uncoated capillaries,17,18 with shortened instrument times of 5 7 min per sample.18 Received: July 5, 2011 Accepted: September 20, 2011 Published: September 20, 2011 8184

dx.doi.org/10.1021/ac201741w | Anal. Chem. 2011, 83, 8184–8191

Analytical Chemistry Microchip assays are being increasingly used as high throughput alternatives to conventional capillary-based separations. With the availability of commercial instruments, a number of protein product quality attributes can be routinely assessed using microchip assays including size variants and high mannose levels,19 total dimer levels,20 disulfide isomers,21 and glycan distribution.22,23 Although microchip CZE has been demonstrated in academic laboratories on model proteins,24 29 the implementation of such assays in industry analytical laboratories for quality assessment of protein products will require available commercial platforms and improved resolution to separate protein charge variants that differ only slightly in their pI values. We describe here a microchip CZE method to profile protein charge heterogeneity using a commercial microfluidic chip and instrument. After conjugation with a fluorescence dye, protein charge variants with a pI difference of ∼0.1 can be readily separated and detected. This technique can be applied to multiple antibody products. The charge profiles acquired are similar to those obtained using conventional CZE. The application of this technique to stressed antibody samples is also demonstrated.

’ EXPERIMENTAL SECTION Materials. HT DNA 5K chips were purchased from Caliper Life Sciences (Hopkinton, MA). The five monoclonal antibodies tested are commercially marketed products. The pI values of the antibodies were determined by icIEF and ranged from 7.5 to 9.2. Cy5 NHS-ester (CyDye DIGE Fluor Cy5 minimal dye) was purchased from GE Healthcare Biosciences (Piscataway, NJ). Zeba 96-well spin desalting plates were purchased from Thermo Scientific (Rockford, IL). Hydroxypropyl methylcellulose (HPMC) was purchased from Spectrum Chemical (New Brunswick, NJ). 6-Aminocaproic acid (EACA) was purchased from Sigma (St. Louis, MO). Acetic acid and dimethyl sulfoxide (DMSO) were HPLC-reagent grade. Imaged Capillary Isoelectric Focusing. Imaged capillary isoelectric focusing was performed on an iCE280 Analyzer with PrinCE autosampler (Convergent Bioscience, Toronto, Canada). The separation cartridge (Convergent Bioscience) contains a 5 cm long, 100 μm ID  200 μm OD capillary coated with fluorocarbon. Sodium hydroxide (0.1 M) and phosphoric acid (80 mM) were used as the catholyte and anolyte, respectively. The sample mixture contains 0.33 mg/mL protein sample, 0.35% (w/v) methylcellulose, 8% (v/v) Pharmalyte 3 10 (GE Healthcare, Piscataway, NJ), 0.33% (v/v) each of pI markers 5.85 and 9.77 (Convergent Bioscience), and 3 M urea. The focusing was performed at 1500 V for 1 min followed by 3000 V for 9 min. The whole column images were acquired at 280 nm. The pI values of protein peaks were calibrated by the two pI markers using the iCE280 software. Conventional CZE. Conventional CZE analyses were performed on a Beckman Coulter PA800 Enhanced instrument with eCAP coated neutral capillary (total length of 30 cm, effective length of 20 cm, ID of 50 μm). The electrophoresis buffer consists of 20 mM 6-aminocaproic acid (EACA)/acetic acid (pH 4.4) and 0.01% polysorbate 20. Samples (0.5 mg/mL) were kept at 10 °C and injected at 0.5 psi for 4 s. The separations were conducted at 35 °C for 10 min at separation voltage of 14.5 kV. Detection was by UV absorbance at 214 nm. Between injections, the capillary was flushed with electrophoresis buffer at 40 psi for 2 min.

ARTICLE

Figure 1. Schematic layout of the sipper chip used for protein charge profiling. Channels in bold lines are wider and deeper than those in thin lines to facilitate pneumatic flow. The sipper is perpendicular to the chip, the attachment site of which is indicated. The channel connecting well 3 and 6 is the loading channel and that connecting well 5 and 8 is the separation channel. The effective separation length between the injection cross-section and the detection point is 12.5 mm.

Preparation of Cy5 Labeled Proteins. Stock solution of 200 μM Cy5 NHS-ester was prepared by dissolving 5 nmol of dye in 25 μL of DMSO and stored frozen. Prior to application to protein samples, the 200 μM stock solution was diluted to 20 μM using water. Protein samples (125 μg) were buffer exchanged into 50 μL of 37.5 mM Tris (pH 8.0) using Zeba 96-well spin desalting plate. Eight microliters of buffer-exchanged protein samples was then transferred to a 96-well microtiter plate and mixed with 4 μL of 20 μM Cy5 NHS-ester. The resulting solution was then incubated at room temperature in the dark for 30 min. After incubation, 28 μL of water was added to each well followed by desalting using the Zeba 96-well spin desalting plate to remove excess dye. Microchip CZE. High throughput charge separations of labeled protein samples were performed on a LabChip GXII instrument (Caliper Life Sciences) using commercial fused-silica sipper chips (HT DNA 5K chip from Caliper Life Sciences). The chip is maintained at 30 °C throughout the analysis. A schematic diagram of the sipper chip is shown in Figure 1. Details of the chip fabrication and channel layout are described by Chow.30 Unless noted otherwise, all reagent wells of the chip are filled with 120 μL of electrophoresis buffer containing 0.2% HPMC and 0.1% polysorbate 20 in 50 mM sodium acetate pH 6. The electrophoresis buffer is then pressurized into the microfluidic channels and used throughout the analysis without replenishment. Protein samples from the microtiter plate were brought onto the chip through the sipper (∼500 nL), followed by subsequent electrokinetical loading, injection, and separation on chip. The protein sample sipped into the loading channel first fills the cross-section with the separation channel by applying 2000 V across the loading channel. To prevent sample leakage into the separation channel during the loading step, small positive currents (i.e., electric current away from the corresponding well) are applied at both ends of the separation channel to pinch the sample plug at the cross-section. After sample loading, a reverse pinch is applied to expand the sample plug along the separation channel to improve sensitivity. The expanded sample plug is then injected into the separation channel (12.5 mm in effective length and 5 μm in depth) at ∼470 V/cm, followed by separation for 56 s at the same field strength. During injection and separation, small negative currents are applied at both ends of the loading channel to limit sample leakage. At the end of the separation, the next sample is sipped from the plate to prepare for sample loading. Each injection takes 80 s, and analysis of an entire 8185

dx.doi.org/10.1021/ac201741w |Anal. Chem. 2011, 83, 8184–8191

Analytical Chemistry

ARTICLE

Figure 2. Microchip CZE of mAb A derivatized at different D/P ratios. Protein concentration at labeling was the same at 2.5 mg/mL (corresponding to 0.5 mg/mL loading concentration). All other conditions are as described in the Experimental Section.

Figure 3. Microchip CZE of mAb A using electrophoresis buffers containing 0.2% HPMC and 0.1% polysorbate 20 in (a) 100 mM sodium acetate pH 5.5, (b) 50 mM sodium acetate pH 6.0, (c) 50 mM sodium acetate pH 5.5, and (d) 50 mM EACA pH 4.4.

Table 1. Microchip CZE Percent Peak Area of mAb A Labeled at Different D/P Ratios

residues, and (c) has absorption and emission maxima of 648 and 669 nm, respectively, which are compatible with the excitation (630 nm) and detection (700 nm) wavelengths of the commercial microchip instrument used. Dye to Protein Ratio. Microchip CZE electropherograms of mAb A labeled at different dye-to-protein molar ratios (D/P ratios) are shown in Figure 2. As illustrated in Figure 2, signal intensities increase with increasing D/P ratios. Percent peak areas of the electropherograms shown in Figure 2 are listed in Table 1. Within the D/P range evaluated, consistent %acidic, %main, and %basic values were obtained, indicating that the protein charge profile was not affected by conjugation. Because the microchip CZE method is intended for high throughput screening of large numbers of samples, D/P ratio of 0.6 was selected for further evaluation which provides a good balance between sensitivity and cost of analysis. It is, however, demonstrated in Figure 2 and Table 1 that a higher D/P ratio may be employed for better sensitivity without affecting charge profiles. Effect of Labeling Conditions. The labeling conditions were optimized. Tris and bicarbonate buffers were compared as reaction buffers. The results obtained using the two buffer systems (25 mM, pH 8.5) were virtually identical. Although Tris contains an amino group, its reaction with the Cy5 NHS-ester dye appears to be negligible under the labeling conditions used, likely due to steric hindrance effect. Tris buffer is selected because of its better stability upon storage. The labeling reactions were also performed in 25 mM Tris buffers at pH 7.5, 8.0, and 8.5 to examine the pH effect. Whereas results obtained at pH 8.0 and 8.5 are very similar, the signal intensities at pH 7.5 are slightly lower. It is known that NHS-esters react with amines more rapidly at alkaline pH. However, the hydrolysis rate of NHS-esters also increases with increasing pH. It is likely that at pH 8.5 and above, and at the protein and dye concentration used here, hydrolysis of NHS-esters competes at a higher rate with the coupling reaction to primary amines. Labeling reactions were also conducted at room temperature and at 4 °C. The reactions at both

D/P ratio

%acidic

%main

%basic

0.3 0.6

12.5 13.0

66.5 66.1

21.0 20.9

1.2

12.9

66.3

20.8

2.4

13.0

66.6

20.5

96-well plate takes approximately 2.5 h to complete without any manual intervention. Microchip CZE data were exported in AIA format and analyzed by 32Karat software (Beckman Coulter) using time-corrected areas.

’ RESULTS AND DISCUSSION Optimization of Sample Preparation. Selection of Labeling Dye. Fluorescent derivatization of protein molecules followed by

laser-induced fluorescence detection has been successfully demonstrated in capillary electrophoresis sodium dodecyl sulfate methods for detection of proteins after size-based separation in an electrical field.31,32 In the application of microchip CZE, there are a few factors that need to be considered when selecting the labeling reagent: (a) The target reactive group should be commonly present in proteins so that the assay can be used as a platform method. (b) Under the separation conditions, the conjugation should introduce no or minimum change to the charge and the spatial structure of the protein. (c) The fluorescent dye needs to have effective absorption and emission at the excitation and detection wavelengths of the microchip instrument. On the basis of these criteria, Cy5 NHS-ester dye was selected as the derivatization reagent, as it (a) reacts selectively with protein N-terminus and ε-amino groups of lysine residues which are abundant in most proteins, (b) still presents one positive charge after conjugation, replacing the intrinsic positive charge on lysine

8186

dx.doi.org/10.1021/ac201741w |Anal. Chem. 2011, 83, 8184–8191

Analytical Chemistry

ARTICLE

Figure 4. Comparison of microchip CZE (a) and conventional CZE (b) for five marketed antibody therapeutics. The main peak pI values determined by icIEF are indicated. For mAb C, the pI values of the three major peaks from left to right are 7.81, 7.63, and 7.46, respectively.

Table 2. Recovery by Sample Dilution total corrected area (RFU) concna (mg/mL)

theorb

expt

recovery (%)

0.25

337637

303859

90

0.375

506456

491038

97

0.5 0.625

844093

675274 848199

100

0.75

1012911

968831

96

a

The mAb D sample was prepared at loading concentration of 0.75 mg/mL and then serially diluted to concentrations that represent 50 150% of nominal loading concentration. All samples were injected in triplicate. b The theoretical total corrected areas were calculated based on the value from the nominal concentration (0.5 mg/mL).

temperatures were quenched after 30, 60, and 120 min of reaction by addition of 10 mM lysine. There is no noticeable difference observed at different temperatures and reaction times. Thirty minutes of incubation at room temperature was selected because of the shorter reaction time and ease of operation at room temperature. The final optimized labeling conditions are described in the Experimental Section. Optimization of Separation Conditions. Different electrophoresis buffers were evaluated using mAb A as the model protein. The pHs of those electrophoresis buffers are all lower than the pI values of the protein molecules studied here. Under such separation conditions, the protein molecules are positively charged and the direction of their electrophoretic migration is the same as that of EOF. To suppress analyte adsorption, a combination of neutral polymer (0.2% HPMC) and nonionic surfactant (0.1% polysorbate 20) was used as dynamic coating agent in the electrophoresis buffers.33 An overlay of microchip CZE electropherograms of mAb A using different electrophoresis buffers is shown in Figure 3. EACA buffer at pH 4.4, although widely used in conventional CZE, provided little separation of protein charge variants in microchip CZE. Much better separations were

Figure 5. Linearity of minor species. Both mAb E and mAb B were prepared at final concentration of 0.5 mg/mL. Labeled mAb E was then spiked into labeled mAb B at (a) 5% (25 μg/mL), (b) 2% (10 μg/mL), (c) 1% (5 μg/mL), and (d) 0.5% (2.5 μg/mL).

obtained with sodium acetate buffers at higher pHs. Similar to the observations with bare fused silica capillaries,18 improvement in resolution can be achieved by using higher buffer concentration and choosing separation pH values closer to protein pI values. As can be seen in Figure 3, protein migration time increases as the concentration of the sodium acetate buffer (pH5.5) increases from 50 mM to 100 mM. This is consistent with decreased EOF at higher ionic strength. Concomitant increase in resolution can also be noticed. As pH increases from 5.5 to 6.0 (50 mM sodium acetate) and gets closer to protein pI, the net charge on protein decreases, resulting in decreased effective mobility. Although higher EOF is expected at higher pH, the longer migration time indicates that the increase in EOF mobility is relatively small and 8187

dx.doi.org/10.1021/ac201741w |Anal. Chem. 2011, 83, 8184–8191

Analytical Chemistry

ARTICLE

Table 3. Precision of Microchip CZE repeatabilitya (N = 6)

intermediate precisionb (N = 18)

%basic

%main

%acidic

%basic

%main

%acidic

average

1.0

61.3

37.7

1.0

61.7

RSD (%)

10.4

0.5

0.7

9.5

3.0

injection repeatabilityc (N = 32) %basic

%main

%acidic

37.3

0.9

60.6

38.5

5.1

12.9

1.9

3.1

a

Six individual preparations of mAb E were analyzed by one analyst on one day. Each preparation was injected once. b Two analysts analyzed mAb E samples using two different instruments and two different chips on three separate days (six individual preparations each time). Each preparation was injected once. c A total of 96 injections. One sample preparation was injected 32 times, with two blank injections between each two sample injections.

Figure 6. Injection repeatability of mAb E using electrophoresis buffers of 0.2% HPMC and 0.1% polysorbate 20 in (a) 50 mM sodium acetate pH 6 and (b) 65 mM sodium acetate pH 5.5. A total of 96 injections were performed in each experiment with repetitive injections of one mAb E sample followed by two blank injections. Each overlay consists of injection #1 to injection #94 (mAb E injections only). An offset of 0.05 min is applied between adjacent injections for better visualization of the charge profiles.

the apparent electrophoretic mobility of the protein decreases because of the lower effective mobility. Sodium acetate buffer at 50 mM concentration and pH 6.0 was selected for most of the work presented here. However, alternative combinations of buffer concentration and pH value may be used either for optimized separation of proteins with different pI values or for analysis of a larger number of samples where higher buffer capacity is desired. The latter will be discussed in the next section.

Assay Performance. The performance of the microchip CZE method was evaluated in terms of separation efficiency, specificity, carryover, linearity, limit of detection (LOD), precision, and analysis of stressed protein products. Separation Efficiency. Five different commercial antibody therapeutics with pI values ranging from 7.5 to 9.2 were selected to compare the microchip CZE (Figure 4a) with the conventional CZE method (Figure 4b). The charge profile of each Cy5 8188

dx.doi.org/10.1021/ac201741w |Anal. Chem. 2011, 83, 8184–8191

Analytical Chemistry

ARTICLE

Figure 7. Stability monitoring of stressed mAb A (a, b, c) and mAb E (d, e, f) using microchip CZE (a and d) and conventional CZE (b and e). Both mAb A and mAb E were incubated in 0.5 M sodium phosphate pH 8.0 at 40 °C for the time intervals indicated prior to analysis.

NHS-ester-derivatized protein obtained using microchip CZE is very similar to the profile of the corresponding native protein obtained using conventional CZE. The instrument time of microchip CZE analysis is 80 s per sample, much faster than that of the conventional CZE method used (12 min). All sample preparation steps for microchip CZE analysis, including buffer exchange and derivatization, can be performed on microtiter plates within approximately 1 h. For the mAb A example shown in Figure 4a, the pI of the main peak species is 8.85, whereas those of the two basic variants are 8.98 and 9.08, respectively. As can be seen in Figure 4a, these species with pI differences of only ∼0.1 can be readily separated. The theoretical plates were calculated to be 1.2  104 (1.0 μm plate height) for the mAb A main peak, compared to conventional CZE (Figure 4b) with 8.6  104 theoretical plates (2.3 μm plate height). The resolutions between the mAb A main peak and the first acidic peak are 1.0 and 2.1 for microchip CZE and conventional CZE, respectively. Although the field strength in microchip CZE separation (∼470 V/cm) is comparable to that in the conventional CZE method used here (∼480 V/cm), better resolution is observed in conventional CZE because of the much longer separation length (200 mm vs 12.5 mm in microchip CZE). The separation by microchip CZE, however, is much faster and provides sufficient resolution for high throughput protein charge profiling. Specificity and Carryover. The specificity of the method was evaluated for any potential interference from the sample matrix. To also evaluate any potential carryover from sample injection, a reagent blank was injected both before and after a sample injection. As shown in Figure S-1 (Supporting Information), there is no interfering peak from the reagent blank, and there is no detectable carryover.

Linearity and Limit of Detection. The linearity of the method was evaluated by sample dilution. The mAb D sample was labeled at 3.75 mg/mL (corresponding to loading concentration of 0.75 mg/mL) and then diluted to concentrations representing 50 150% of the nominal protein concentration (Table 2). These samples were injected in triplicates. The mean recoveries were calculated by comparing the total corrected areas to that calculated from nominal loading concentration of 0.5 mg/mL (100%). As shown in Table 2, the mean recoveries ranged from 90 to 100%. The means of the corrected peak areas from the recovery study were plotted versus the protein concentration over the range of 50 150% of the target concentration (0.5 mg/mL). The correlation coefficient was >0.99, indicating that the assay is linear between 50% and 150% of the target protein concentration. Linearity of minor species were evaluated by spiking labeled mAb E into mAb B at four different levels (Figure 5). The corrected main peak areas of mAb E were plotted versus percentage of mAb E in mAb B. The correlation coefficient was >0.99, indicating that the assay is linear for minor species in the range of 0.5 5%. The signal-to-noise ratio of mAb E main peak at the 0.5% level is 12, and the %main peak for mAb E is ∼61%, suggesting that the limit of detection (defined as a signalto-noise ratio of 3) of the mAb E main peak is ∼0.08% (or ∼0.4 μg/mL loading concentration). The sensitivity compares favorably to those of typical icIEF (3 μg/mL)13 and conventional CZE (2 μg/mL)17 methods, as well as those reported in literatures for microchip CZE analyses of native (12.5 μg/mL)26 and on-chip labeled (∼1 μg/mL)24,28 proteins. It has been demonstrated that a detection limit of 9 ng/mL can be achieved 8189

dx.doi.org/10.1021/ac201741w |Anal. Chem. 2011, 83, 8184–8191

Analytical Chemistry in microchip CZE analysis of BSA with off-chip labeling.28 In that study, the fluorescence dye is in large molar excess to the protein (∼106 fold), compared with a D/P ratio of 0.6 in the present work. Detection limits of lower than 1 ng/mL has also been reported, for example, in microchip analyses of immunocomplexes.34,35 The labeling dyes used, however, alter the protein charge profile upon conjugation and are therefore not suited for microchip CZE applications. Precision. To evaluate assay repeatability, six individual preparations of mAb E were analyzed. Intermediate precision was assessed by two analysts on three separate days using two different instruments and two different chips. As shown in Table 3, for mAb E containing ∼61% of main peak and ∼1% of basic peak, the RSD value of the %peak area are 0.5% and 10.4%, respectively, for repeatability (N = 6), and 3.0% and 9.5%, respectively, for intermediate precision (N = 18). In conventional CZE, the capillary is typically washed and refilled with fresh electrophoresis buffer between sample injections. This step, although taking additional instrument time, is critical to achieve reproducible results especially for uncoated capillaries.17,18 In the microchip CZE method described here, the microfluidic channels are filled with the electrophoresis buffer only at the beginning of each assay, a further improvement in throughput, as replenishment of electrophoresis buffer is not required between injections. To assess injection repeatability of this method for large number of samples, 96 injections were performed with repetitive injections of one mAb E sample followed by two blanks for 32 cycles. The results of this experiment are summarized in Table 3 and Figure 6a. As indicated in Table 3, good precision is achieved for percent peak areas with RSD values of 1.9% and 12.9% for %main and %basic, respectively (N = 32). However, shifts in peak migration times were noticed after 55 injections (see Figure 6a), with the largest difference between main peak migration times being approximately 6 s. This is believed to be due to the lower buffer capacity for sodium acetate at pH 6. To improve the reproducibility of the migration time, the injection repeatability experiment was repeated using 65 mM sodium acetate at pH 5.5 with the same dynamic coating agents. As shown in Figure 6, the charge profiles obtained using the two electrophoresis buffers are very similar, and consistent migration times can be achieved throughout the entire 96 injections using 65 mM sodium acetate (pH 5.5) in the electrophoresis buffer. In addition, there is no detectable carryover at the end of the 96 injections, as illustrated in Figure S-2 (Supporting Information). Analysis of Stressed Samples. The ability of the method to detect changes in protein charge profile was evaluated by analyzing mAb A and mAb E after forced degradation under basic conditions. Proteins (2.5 mg/mL) were incubated in 0.5 M sodium phosphate (pH 8.0) for 0, 1, 4, and 8 days and stored frozen until analysis. Electropherograms of stressed samples obtained using microchip CZE and conventional CZE are shown in Figure 7. As can be seen in Figure 7, both mAb samples showed an increase in acidic variants after stress, consistent with protein deamidation under basic conditions. Percent main peak determined by both conventional and microchip CZE were plotted versus incubation time (Figure 7c and 7f). As shown in Figure 7c and 7f, the degradation rates determined by the two methods are very similar to each other. These results indicate that the microchip CZE method can be used in stability programs to monitor changes in the charge profiles of protein products.

ARTICLE

’ CONCLUSIONS A high throughput microchip CZE assay has been developed to profile protein charge heterogeneity. Labeling protein samples with the Cy5 NHS-ester maintains the net charge or valence on protein molecules and enables fluorescence detection on a commercial microchip instrument. Antibodies with a pI range of 7.5 9.2 have been tested by this method. Protein charge variants with pI differences of ∼0.1 can be readily resolved. This method demonstrated good specificity, linearity, and precision and could detect charge profile changes in stressed samples. When labeled at a D/P ratio of 0.6, a LOD of ∼0.08% (or ∼0.4 μg/mL) can be achieved. When higher sensitivity is desired, labeling reactions can be conducted at higher D/P ratios of up to 2.4 without changing the protein charge profile. Each microchip CZE analysis takes 80 s, as compared to 7 min or longer for conventional CZE. Sample loading, injection, separation, and data collection are fully automated and do not require operator intervention during the experiment. All sample preparation steps including buffer exchange/desalting can be performed in microtiter plates using multichannel pipettes or liquid handlers, enabling high throughput sample analysis. Multiple plates of samples can be prepared in approximately 1 h, and analysis of each 96-well plate takes approximately 2.5 h. The spin desalting plate used in the first sample preparation step has been demonstrated to efficiently remove interfering species from cell culture media for protein coupling reactions with NHSesters.20 The method described here may thus be applicable to crude cell culture samples. Moreover, higher sensitivity would be possible using microchips with deeper microfluidic channels, and better resolution may be obtained by removing the reversed pinch step or using microchips with longer separation lengths. However, instrument control parameters would need to be adjusted to achieve optimum separation if different types of microchips were to be employed. ’ ASSOCIATED CONTENT

bS

Supporting Information. Additional figures as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Present Addresses †

Merck Research Laboratories, 1011 Morris Ave., MS U13-1-1980, Union, NJ 07083.

’ ACKNOWLEDGMENT We thank John Davis for his contribution to this study, and Michael Vetsch for providing the antibody samples. We thank Bahram Fathollahi of Caliper Life Sciences for helpful discussions. We also thank Kenneth LeClair and Markus Bluemel for critical reviews of the manuscript. ’ REFERENCES (1) Vlasak, J.; Ionescu, R. Curr. Pharm. Biotechnol. 2008, 9, 468–481. (2) Quan, C.; Alcala, E.; Petkovska, I.; Matthews, D.; Canova-Davis, E.; Taticek, R.; Ma, S. Anal. Biochem. 2008, 373, 179–191. 8190

dx.doi.org/10.1021/ac201741w |Anal. Chem. 2011, 83, 8184–8191

Analytical Chemistry

ARTICLE

(3) Liu, Y. H.; Wylie, D.; Zhao, J.; Cure, R.; Cutler, C.; CannonCarlson, S.; Yang, X.; Nagabhushan, T. L.; Pramanik, B. N. Anal. Biochem. 2011, 408, 105–117. (4) Rehder, D. S.; Dillon, T. M.; Pipes, G. D.; Bondarenko, P. V. J. Chromatogr., A 2006, 1102, 164–175. (5) Moorhouse, K. G.; Nashabeh, W.; Deveney, J.; Bjork, N. S.; Mulkerrin, M. G.; Ryskamp, T. J. Pharm. Biomed. Anal. 1997, 16, 593–603. (6) Meert, C. D.; Brady, L. J.; Guo, A.; Balland, A. Anal. Chem. 2010, 82, 3510–3518. (7) Johnson, K. A.; Paisley-Flango, K.; Tangarone, B. S.; Porter, T. J.; Rouse, J. C. Anal. Biochem. 2007, 360, 75–83. (8) Harris, R. J. J. Chromatogr., A 1995, 705, 129–134. (9) Harris, R. J.; Kabakoff, B.; Macchi, F. D.; Shen, F. J.; Kwong, M.; Andya, J. D.; Shire, S. J.; Bjork, N.; Totpal, K.; Chen, A. B. J. Chromatogr., B: Biomed. Sci. Appl. 2001, 752, 233–245. (10) Chen, X. Bioanalysis 2009, 1, 1183–1186. (11) Farnan, D.; Moreno, G. T. Anal. Chem. 2009, 81, 8846–8857. (12) Wu, J.; Li, S. C.; Watson, A. J. Chromatogr., A 1998, 817, 163–171. (13) Li, N.; Kessler, K.; Bass, L.; Zeng, D. J. Pharm. Biomed. Anal. 2007, 43, 963–972. (14) Sosic, Z.; Houde, D.; Blum, A.; Carlage, T.; Lyubarskaya, Y. Electrophoresis 2008, 29, 4368–4376. (15) He, X. Z.; Que, A. H.; Mo, J. J. Electrophoresis 2009, 30, 714–722. (16) Ma, S.; Nashabeh, W. Chromatographia 2001, 53, S75–S89. (17) He, Y.; Lacher, N. A.; Hou, W.; Wang, Q.; Isele, C.; Starkey, J.; Ruesch, M. Anal. Chem. 2010, 82, 3222–3230. (18) He, Y.; Isele, C.; Hou, W.; Ruesch, M. J. Sep. Sci. 2011, 34, 548–555. (19) Chen, X.; Tang, K.; Lee, M.; Flynn, G. C. Electrophoresis 2008, 29, 4993–5002. (20) Chen, X.; Flynn, G. C. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2009, 877, 3012–3018. (21) Lacher, N. A.; Wang, Q.; Roberts, R. K.; Holovics, H. J.; Aykent, S.; Schlittler, M. R.; Thompson, M. R.; Demarest, C. W. Electrophoresis 2010, 31, 448–458. (22) Smejkal, P.; Szekrenyes, A.; Ryvolova, M.; Foret, F.; Guttman, A.; Bek, F.; Macka, M. Electrophoresis 2010, 31, 3783–3786. (23) Primack, J.; Flynn, G. C.; Pan, H. Electrophoresis 2011, 32, 1129–1132. (24) Liu, Y.; Foote, R. S.; Jacobson, S. C.; Ramsey, R. S.; Ramsey, J. M. Anal. Chem. 2000, 72, 4608–4613. (25) Olvecka, E.; Kaniansky, D.; Pollak, B.; Stanislawski, B. Electrophoresis 2004, 25, 3865–3874. (26) Schulze, P.; Ludwig, M.; Kohler, F.; Belder, D. Anal. Chem. 2005, 77, 1325–1329. (27) Mohamadi, M. R.; Mahmoudian, L.; Kaji, N.; Tokeshi, M.; Baba, Y. Electrophoresis 2007, 28, 830–836. (28) Yu, M.; Wang, H. Y.; Woolley, A. T. Electrophoresis 2009, 30, 4230–4236. (29) Tran, N. T.; Ayed, I.; Pallandre, A.; Taverna, M. Electrophoresis 2010, 31, 147–173. (30) Chow, A. W. Methods Mol. Biol. 2006, 339, 129–144. (31) Salas-Solano, O.; Tomlinson, B.; Du, S.; Parker, M.; Strahan, A.; Ma, S. Anal. Chem. 2006, 78, 6583–6594. (32) Michels, D. A.; Brady, L. J.; Guo, A.; Balland, A. Anal. Chem. 2007, 79, 5963–5971. (33) Corradini, D. J. Chromatogr., B: Biomed. Sci. Appl. 1997, 699, 221–256. (34) Schrott, W.; Nebyla, M.; Pribyl, M.; Snita, D. Biomicrofluidics 2011, 5, 14101. (35) Ye, F.; Shi, M.; Huang, Y.; Zhao, S. Clin. Chim. Acta 2010, 411, 1058–1062.

8191

dx.doi.org/10.1021/ac201741w |Anal. Chem. 2011, 83, 8184–8191