Article pubs.acs.org/ac
Low-Flow Sheathless Capillary Electrophoresis−Mass Spectrometry for Sensitive Glycoform Profiling of Intact Pharmaceutical Proteins Rob Haselberg,* Gerhardus J. de Jong, and Govert W. Somsen Biomolecular Analysis, Utrecht University, Universiteitsweg 99, 3584 CG Utrecht, The Netherlands S Supporting Information *
ABSTRACT: Capillary electrophoresis coupled to time-offlight mass spectrometry (CE−TOF-MS) via a porous tip sheathless electrospray ionization (ESI) interface was studied for the characterization of pharmaceutical glycoproteins. To achieve optimal glycoform separation, background electrolytes of low pH were used in conjunction with a capillary with a neutral coating exhibiting near-zero electroosmotic flow. Crucial interfacing parameters, like ESI voltage and ESI tipto-end plate distance, were optimized for very low flow rates (∼5 nL/min) in order to attain maximum sensitivity and stable performance. Under optimal conditions, the sheathless CE−MS interface provided significantly increased ionization efficiencies for intact proteins and decreased ionization suppression leading to detection limits in the picomolar-range. Analysis of a sample of recombinant human interferon-β allowed the assignment of at least 18 glycoforms, plus a variety of deamidation, succinimide, and oxidation products, representing a considerable improvement over sheath-liquid CE−MS. The sheathless CE−MS system also proved highly suitable for the glycoprofiling of recombinant human erythropoietin, revealing 74 glycoforms in a 60-min run. In addition, oxidation and acetylation products were detected, overall resulting in assignment of more than 250 different isoforms. Semiquantitative glycoprofiles could be derived for both pharmaceutical proteins, with estimated glycoform concentrations analyzed ranging from 0.35 to 950 nM. These profiles may be very useful for quality control of biopharmaceuticals and their biosimilars.
B
of intact glycoforms may be a difficult task. CE coupled with electrospray ionization (ESI) MS potentially provides the high separation efficiency and mass-selective detection that can be very useful for characterization of glycosylated biopharmaceuticals.8−10 Protein modifications, like glycosylation, may involve changes of net charge and size of a protein and, thus, of protein electrophoretic mobility. CE has the intrinsic capacity to produce narrow peaks and is very useful for the separation of a variety of protein modifications in a single run. MS detection of high mass accuracy and resolution, such as provided by time-offlight (TOF) instruments, yields information on the molecular weight of separated protein species. The feasibility and usefulness of CE−MS for intact proteins has been widely demonstrated.8−10 Still, there lies a distinct challenge in the improvement of protein analyte responses in CE−MS in order to meet relevant biopharmaceutical queries. CE is most commonly coupled to MS using sheath liquid ESI interfacing.11,12 Unfortunately, because of dilution of the capillary effluent by the sheath liquid, the detection sensitivity is inherently compromised. Recently, Beckman Coulter has developed a prototype sheathless CE−MS interface based on a porous tip design proposed by Moini.13 A nanoESI emitter was created by etching 3−4 cm of the distal end of a 30-μm i.d.
iopharmaceuticals take an increasingly important position in drug development and production.1 As for any drug for human use, biopharmaceuticals have to meet stringent quality requirements. However, because of their macromolecular nature, the full characterization of biopharmaceuticals poses a great challenge.2,3 Moreover, multistep biotechnological production processes may show variability, introducing product diversity that can affect activity. Protein heterogeneity is an important quality attribute that has to be assessed. Over 40% of approved biopharmaceuticals are glycoproteins,4 making glycosylation an essential source of heterogeneity. The type and degree of glycosylation depends on the host organism, production cell line, and/or culture conditions.4 This means that different productions of the same biopharmaceutical might show variable glycosylation patterns. When the molecular weights of the glycosylated and nonglycosylated protein significantly differ, protein glycosylation can be confirmed by gel electrophoresis.5 Glycan heterogeneity can be determined by glycopeptide mapping after protein digesting and/or by glycan analysis after their release using liquid chromatographic (LC) or capillary electrophoretic (CE) techniques.6,7 The continuous improvement of mass spectrometric (MS) and separation technologies, and the aspiration to simplify sample pretreatment, has led to a trend toward direct analysis of intact proteins. Moreover, in pharmaceutical quality control, analytical methodologies allowing characterization of the intact (i.e., unmodified) product are essential. Nevertheless, resolution and assignment © 2013 American Chemical Society
Received: November 7, 2012 Accepted: January 16, 2013 Published: January 16, 2013 2289
dx.doi.org/10.1021/ac303158f | Anal. Chem. 2013, 85, 2289−2296
Analytical Chemistry
Article
060016A) from Biogen Idec (Cambridge, MA) and recombinant human erythropoietin (rhEPO, NeoRecormon 30 000 IU, lot no. H0002H01) from Roche (Mannheim, Germany) were obtained as prefilled syringes. Sample pretreatment was performed as described in the Supporting Information. CE System. CE experiments were carried out using a PA800 plus capillary electrophoresis system from Beckman Coulter Inc. (Brea, CA). Unless stated otherwise, the separation voltage was 30 kV and the capillary temperature was 20 °C. During CE analysis, a pressure of 0.5 psi was applied at the capillary inlet. Capillaries (total length, 100 cm; inner diameter, 30 μm; outer diameter, 150 μm) equipped with a porous tip (length, 3−4 cm) and a neutral bilayer coating were supplied by Beckman Coulter. The hydrophilic capillary coating with a layer of polyacrylamide is a proprietary product currently in development at Beckman Coulter. Prior to use, the capillary was rinsed with deionized water for 30 min at 50 psi and stored overnight filled with water. After use, the capillary was rinsed for 10 min with deionized water and air at 50 psi, respectively, and subsequently stored at +4 °C. Before each run, the capillary was flushed with fresh BGE for 3 min at 50 psi. BGEs of acetic acid were prepared by diluting the appropriate amount glacial acetic acid to 30 mL with deionized water and adjusting the pH with 0.1 M ammonium hydroxide. The sample was injected for 10 s at 5 psi equaling 1% of the capillary volume. CE−MS Interface. The porous section of the capillary is inserted into a housing comprising a stainless steel cylinder with a retractable outer head, as described in detail previously.23,24 Shortly, the housing is positioned in a bracket mounted on a custom stage (designed by Beckman Coulter) to fit the mass spectrometer. This stage includes an XYZ-platform that allows the positioning of the porous tip in front of the mass spectrometer inlet. Typically, the tip was placed on-axis to the entrance of the MS instrument at a distance of about 1−2 mm. Before performing any CE−ESI-MS experiment, the stainless steel cylinder was filled with the BGE and grounded to close the electrical circuit of the CE. Mass Spectrometry. CE−MS experiments were performed using a microTOF orthogonal-accelerated time-of-flight (TOF) mass spectrometer (Bruker Daltonics, Bremen, Germany) equipped with a nanospray end plate and gas diverter to allow nanoESI. The mass accuracy was 0.998). LODs (S/N 3) for the proteins were in the low-nanogram/milliliter range, which corresponds with injected concentrations of 200−650 pM (Table 1). Although the absolute protein signals obtained at 5 nL/min were somewhat lower than measured at 135 nL/ min (see also Figures 1 and 2), the baseline noise at very low flow rates was reduced by a factor of 2−4 as compared to sheathless CE−MS of the same test proteins using a positively charged capillary coating23 leading to improved signal-to-noise ratios (factor 1.7−3.0) and, thus, lower LODs. To the best of our knowledge, these are the most favorable LODs reported for CE−MS of intact proteins so far. Glycoform Profiling of Intact Pharmaceutical Proteins. Recombinant human interferon-β-1a (rhIFN-β) is a 23kDa glycoprotein (Asn80) used for treatment of multiple sclerosis. Variation in glycosylation of rhIFN-β may affect its biological activity.37 Establishment of glycoform heterogeneity of produced batches of rhIFN-β is of importance for quality control. Sheathless CE−MS of rhIFN-β (45 μg/mL) using a BGE of 50 mM acetic acid (pH 3.0) and a capillary with a neutral coating yielded a distinct pattern of partially resolved bands (Figure 3A). Deconvoluted mass spectra were constructed for 0.2-min time intervals throughout the entire peak profile. The upper trace of Figure 3B shows the deconvoluted mass spectrum obtained in the apex of the peak at 30.1 min. The
mass with highest intensity (22 375.4 Da) nicely corresponds to the molecular mass of rhIFN-β with a fucosylated disialylated biantennary glycan structure comprising four hexose and five N-acetylhexosamine units. This glycoform has been described as the most abundant present in rhIFN-β.37 In the same mass spectrum, two relatively minor bands with respective mass differences of −17 Da and +16 Da in comparison with the main band are observed. The latter species (22 391.1 Da) can be ascribed to an oxidation product of the major glycoform. No protein oxidation was observed for the test proteins under the applied ESI conditions. Therefore, we assume that oxidation products were part of the rhIFN-β sample and not the result of the ESI process. The compound with a mass of 22 358.6 Da could be a succinimide intermediate which is formed during the process of protein deamidation38,39 and has been reported to be part of the analyzed rhIFN-β.40 Indeed, the deamidated form of the major glycoform (22 376.4 Da) was observed in high abundance migrating at 30.5 min (Figure 4A,B, lower trace). Deamidation gives rise to a protein mass increase of 1 Da but also to an increase of overall negative charge, resulting in a significant increase of electrophoretic mobility and thus CE separation from the parent glycoform. Deliberate deamidation is a means to increase the half-life and potency of rhIFN-β.41 No multiply deamidated rhIFN-β was observed which is probably due to the fact that Asn25 is most prone to deamidation.40 The mass spectrum of the deamidated glycoform also revealed oxidized species (Figure 3B, lower trace). Both the succimide formation and oxidation do not cause a charge change in the molecule, and a relatively small mass change only, explaining the comigration of these products with the parent and/or deamidated glycoform. 2293
dx.doi.org/10.1021/ac303158f | Anal. Chem. 2013, 85, 2289−2296
Analytical Chemistry
Article
Table 2. Migration Time, Molecular Mass, N-Glycan Composition and Structure, and Relative Peak Area for rhIFN-β Glycoforms As Observed with Sheathless CE−MS Employing a Neutrally-Coated Capillarya
acid residues and, to a lesser extent, difference in number of HexHexNAc residues. Negatively charged sialic acid residues reduce the overall protein electrophoretic mobility. Neutral sugar units add to the mass of the protein only, giving rise to much smaller changes in electrophoretic mobility.27,42−44 Assuming equal ESI-MS response for all glycoforms and taking injected protein concentration and calculated relative peak areas into account, the glycoform concentrations analyzed were estimated to lie between 0.53 and 946 nM. This demonstrates the excellent sensitivity of the sheathless CE−MS setup for intact glycoproteins. These results represent a considerable improvement over previously reported MS-based analyses of rhIFN-β. Using direct ESI-MS, Orru et al.37 were able to detect six glycoforms. With sheath-liquid CE−MS,45 10 glycoforms could be assigned, while the concentration of protein injected was 10 times higher. The significant gain in glycoform coverage obtained with the sheathless CE−MS system performance is due to both the improved glycoform resolution employing capillaries with neutral coating and the high nanoESI sensitivities provided by the porous tip interface efficiency. The performance of the sheathless CE−MS system for glycoform profiling was further tested by the analysis of recombinant human erythropoietin (rhEPO). EPO is an essential glycoprotein hormone for red cell production, and rhEPO is used for the treatment of anemia resulting from chronic kidney disease. The protein has three N-glycosylation sites (Asn24, Asn38, and Asn83) and one O-glycosylation site (Ser126). The total glycan content can make up about 40% of the protein’s molecular weight and introduces a high variation in protein heterogeneity,46 making glycoform profiling quite a challenge. The usefulness of CE−MS employing capillaries with a neutral coating for the characterization of intact EPO has been demonstrated using sheath-liquid interfacing. Injection of high concentrations (>2 mg/mL) of EPO were needed to allow minor glycoforms to be detected.27,28,30,47,48 A rhEPO sample (epoetin-β) of 200 μg/mL was prepared and analyzed using a BGE of 2.0 M acetic acid (pH 2.1) and a neutrally coated capillary. Sheathless CE−MS analysis showed a wide pattern of sharp peaks (Figure 4A). The EPO glycoforms are resolved over a 20-min time window exhibiting a high signal-to-noise ratio. Presentation of the obtained electropherogram as a contour plot (Figure 4C1) nicely depicts the glycoform resolution obtained by combining high-efficiency CE with high-resolution mass spectrometry. Notably, the current system appears to provide a more efficient separation of EPO glycoforms than previous CE−MS studies employing other neutral coatings.27,28,30,47,48 Taking into account that the injection volume in our study was similar compared to the other studies, we assume that the improved separation is the result of the coating, more efficiently suppressing the EOF and/ or preventing protein adsorption. Figure 4B shows the deconvoluted mass spectrum obtained in the apex of the peak migrating at 38.0 min, revealing a glycoform with a mass of 29 597 Da. Minor spectral bands with masses of +16 Da and +42 Da with respect to the main mass indicated the presence of oxidation and acetylation products, which has been reported before for epoetin-β.28,47 Low intensity bands with a mass of −365 Da and +365 Da with respect to the main mass were observed, indicating modest comigration of EPO glycoforms comprising one HexHexNAc residue less or more, respectively. Good quality deconvoluted mass spectra were obtained throughout the entire peak profile (steps of 0.2 min), and the masses of the main protein bands were listed (Table 3). In
a
Symbols: green circle/yellow circle, hexose (mannose/galactose); red triangle, fucose; blue square, N-acetylhexosamine; purple diamond, sialic acid.
From the deconvoluted mass spectra obtained throughout the CE−MS profile of rhIFN-β, up to 18 masses of different glycoforms could be extracted. The majority of detected glycoform masses of rhIFN-β mutually differed either 291 or 365 Da, which correlates to a sialic acid (SiA) or a hexose Nacetylhexosamine (HexHexNac) residue, respectively. Three glycoforms with an additional hexose unit (+162 Da) were observed. Unglycosylated rhIFN-β (20 027 Da) was not detected throughout the entire pattern. Taking the glycan structure of the major glycoform as a starting point, the overall glycan compositions of the detected glycoforms could be derived based on their mass (Table 2). For most glycoforms, their deamidation, succinimide, and/or oxidation products were observed as well. Overall, more than 80 different isoforms of rhIFN-β could be distinguished in one CE−MS run. From EIEs constructed for the different glycoforms (see the Supporting Information), migration times and peak areas were derived (Table 2). Figure 3C depicts EIEs for 10 selected glycoforms including their deamidated forms. Glycoform resolution clearly is based on difference in number of sialic 2294
dx.doi.org/10.1021/ac303158f | Anal. Chem. 2013, 85, 2289−2296
Analytical Chemistry
Article
Table 3. Glycan Composition and Molecular Mass of the rhEPO Glycoforms Observed with Sheathless CE−MS Employing a Neutrally-Coated Capillarya
a
The relative intensity of the different glycoforms is indicated by the color intensity (log scale).
runs due to clogging. The second capillary was used for the remaining 109 runs without any problems No ion source contamination and related effects were observed despite the onaxis alignment of the interface. This is most probably due to the extremely low amounts of BGE and sample entering the ion source. During analysis of biopharmaceuticals, the CE−MS system functioned steadily. Migration time and peak area RSDs (n = 10) for the main glycoform of rhIFN-β were 1.9% and 11%, respectively. For repeated rhEPO analysis (n = 6) RSDs were 2.4% and 13%, respectively. Overall, we conclude that the sheathless CE−MS system has good potential for characterization of therapeutic proteins, exhibiting a sufficiently stable performance.
total, 74 distinct glycoforms were detected that differed in number of SiA or HexHexNAc residues. Notably, more than 10 times less protein was injected as compared to earlier reports, but still a similar number of glycoforms was found in the present study. Almost each glycoform was accompanied by oxidized and/or acetylated variants resulting in more than 250 isoforms detected. It is not possible to derive an exact glycan composition solely from the measured intact glycoforms’ masses. We took reported masses of main glycoforms of EPO as a starting point28 and subsequently assigned all detected masses (Table 3). Taking the observed glycan composition into consideration, a closer look at the contour map (Figure 4C) now provides better insight into the glycoform separation obtained. Focusing on the 14+ charge state (Figure 4C2), the glycoforms containing the same number of SiA residues present themselves as clear diagonal rows. A difference of one SiA residue causes a migration time shift of about 2 min. Within a group of sialoforms, differences in neutral HexHexNAc residues contribute to further glycoform separation (Figure 4C3), although to a lesser extent (about 0.5 min per HexHexNAc). Assuming equal detection response for all glycoforms, relative intensities of the glycoforms were derived from the recorded mass spectra and depicted in shades of blue in Table 3. The relative intensities span more than 3 orders of magnitude from the most abundant glycan composition (SiA 1 3 Hex 2 2 HexNAc 1 9 Fuc 3 ) to the least abundant (SiA14Hex32HexNAc29Fuc3) detected. Next to glycoform assignment, this quantitative information could serve as discrimination of rhEPO batches and biosimilars. On the basis of the relative intensities of the detected glycoforms and the injected rhEPO concentration (200 μg/mL), injected glycoform concentration were estimated to range from 0.35 to 630 nM, again illustrating the superior sensitivity of the sheathless CE−MS system. The analytical stability of the low-flow sheathless CE−MS system for intact protein analysis was quite satisfactory. The entire study presented here has been carried out on two porous tip capillaries. The first capillary had to be discarded after 66
■
CONCLUSIONS This paper demonstrates the usefulness of sheathless CE−MS using a capillary with a porous tip ESI emitter and a neutral coating for the analysis of intact proteins. Optimization of the sprayer tip position and ESI voltage resulted in highly favorable mass sensitivity at low-nL/min flow rates. The applied nanoESI conditions also provided reduction of ionization suppression. LODs for test proteins were in the picomolar range. The sheathless CE−MS system demonstrated to be highly suitable for the analysis of complex mixtures of glycosylated proteins. The efficient separation enabled characterization of tens to hundreds of isoforms of pharmaceutical proteins in one single CE−MS run. This kind of information is difficult to obtain with another analytical platform in such a short time (
