J. Biochem. Biophys. Methods 60 (2004) 309 – 317 www.elsevier.com/locate/jbbm
CarboPack PA20: a new monosaccharide separator column with electrochemical detection with disposable gold electrodes Michael Weitzhandler *, Victor Barreto, Christopher Pohl, Petr Jandik, Jun Cheng, Nebojsa Avdalovic Dionex Corporation, 445 Lakeside Drive, Sunnyvale, CA 94086, USA
Abstract This report documents the development of a new monosaccharide separator column (CarboPac PA20, 3150 mm) that allows fast, efficient monosaccharide separations with good spacing. It is based on a new chemistry with a reduced resin particle size (from 10 to 6.5 Am). Faster, more efficient separations of glycoprotein monosaccharides with better spacing were achieved across a range of isocratic NaOH concentrations at lower flow rates. Detection sensitivity was improved, enabling routine low to sub pmol monosaccharide determinations. Glycoprotein monosaccharides eluted in less than 10 min at a flow rate of 0.5 ml/min. Furthermore, when used with an AminoTrap guard column, the protein matrix consisting of amino acids and peptides (released by acid hydrolysis of glycoprotein) did not interfere with monosaccharide analysis. Compared to previous CarboPac columns (CarboPac PA1 and CarboPac PA10), the CarboPac PA20 has improved selectivity with respect to glycoprotein monosaccharides. The improved selectivity results in better separation of glucosamine and galactose, enabling the accurate determination of monosaccharide ratios for undergalactosylated glycoproteins. Finally, disposable gold working electrodes that eliminate the possibility of working electrode recession affecting peak area response were used. D 2004 Elsevier B.V. All rights reserved. Keywords: Glycoprotein; Carbohydrate; Monosaccharide; CarboPac; Amperometric detection; Electrochemical detection; Disposable electrode; High-performance anion exchange chromatography
* Corresponding author. Tel.: +1-408-481-4453; Fax: +1-408-732-2007. E-mail address:
[email protected] (M. Weitzhandler). 0165-022X/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jbbm.2004.01.009
310
M. Weitzhandler et al. / J. Biochem. Biophys. Methods 60 (2004) 309–317
1. Introduction Glycoprotein monosaccharide composition analysis provides information pertaining to the type of oligosaccharides present in a glycoprotein as well as the extent of glycosylation. Identification of galN in a recombinant glycoprotein indicates that O-linked sugar chains may be present. Additionally, identification of the monosaccharides that are present and their relative molar ratios may provide insight into possible oligosaccharide structure(s) that one might expect in a glycoprotein. Direct monosaccharide composition analysis is easily and directly accomplished using high pH anion exchange chromatography with pulsed amperometric detection (HPAE-PAD) [1,2]. Direct electrochemical detection is a sensitive technique that does not exhibit many of the problems associated with carbohydrate analysis techniques that involve labeling [3]. Optimization of a precolumn labeling procedure requires optimization of a multitude of properties with respect to the chosen tag including (a) molecular tag size, (b) hydrophobicity, (c) fluorescent yield, (d) coupling efficiency, (e) difficulty of separation of excess reagent from the product, (f) solubility in solvents used for reductive amination, (g) extinction coefficient in the case of UV detection, and (h) stability and compatibility of the derivatives with downstream characterization. No tag that is optimized with respect to all of these criteria has been identified. The use of HPAE-PAD eliminates (a) the need to select an appropriate tag from the large number of tags available to label monosaccharides, (b) sample preparation associated with labeling, (c) any need for re-N-acetylation of amino sugars, (d) sample cleanup associated with tag usage, and (e) recovery problems encountered when using tags. Other recent approaches to monosaccharide analysis have included mass spectrometry. While mass spectrometry has proven to be a valuable adjunct technique to chromatography for profiling and sizing oligosaccharides, in monosaccharide composition analysis, mass spectrometric techniques suffer from their inability to differentiate between monosaccharides of the same molecular weight. According to Kuster et al. [4], monosaccharides cannot be measured and quantified efficiently by ESI or MALDI-MS. Glycoproteins with low percentages of glycosylation represent a challenge for monosaccharide composition analysis because of matrix interference. For glycoproteins with low levels of glycosylation, an online guard column has been developed to use in conjunction with HPAE-PAD that moves interfering matrix amino acids and peptides out of the monosaccharide elution [2]. This report demonstrates how the AminoTrap in 330 guard cartridge format can be used inline with the CarboPac PA20 column to eliminate matrix interference. Finally, this study examines the use of disposable gold electrodes in HPAE-PAD monosaccharide analysis. Recent advances in electrochemical detection pertaining to carbohydrate analysis document the use of fast wave forms in conjunction with reductive desorption to clean the surface of gold electrodes in pulsed amperometry [5,6]. The importance of using fast waveforms that employ reductive desorption cleaning rather than oxidative cleaning is that it minimizes the time during which gold oxide is formed. It is during the conversion of gold oxide to gold that gold oxide gets slowly sloughed off the gold electrode surface. Thus, minimizing the time during which gold oxide is formed and converted back to gold also minimizes the dissolution and resulting recession of the gold working electrode that previously occurred with oxidative cleaning (in which excessive oxide formation during the electrode cleaning steps at relatively high potentials resulted in
M. Weitzhandler et al. / J. Biochem. Biophys. Methods 60 (2004) 309–317
311
a slow loss of gold from the electrode surface) [7]. Peak area response decreases as working electrode recession increases [8]. By minimizing working electrode recession with the fast quadruple potential waveform [6], peak area reproducibility is improved. The advent of a waveform that eliminates working electrode recession enabled the design and ˚ ) [9]. use of disposable thin gold working electrodes (thin gold layer approximately 3000 A These disposable electrodes are guaranteed to last 1 week for amino acid analysis and 2 weeks for carbohydrate analysis. The shorter lifetime for amino acid analysis is due to the fact that the oxidation step in the amino acid waveform is more aggressive than the oxidation step in the carbohydrate waveform; amino acid oxidation requires a higher oxidative potential compared to carbohydrate oxidation. A detailed description of the fabrication of thin disposable gold electrodes and their use in amino acid analysis was previously published [9]. In this report, disposable gold electrodes were characterized with respect to monosaccharide analysis by examining lifetime, sensitivity, and reproducibility.
2. Materials and methods 2.1. Materials Deionized water was HPLC grade. The NaOH solution used was 50% NaOH (w/w) from Fisher Scientific (Pittsburgh, PA). Polypropylene microcentrifuge tubes (1.5 ml), caps, and O-rings were from Sarstedt (Newton, NC). Autosampler vials: 1232-mm disposable, limited volume sample vials, caps and teflon/silicone septa were from Sun Brokers (Wilmington, NC). Concentrated (13 N) trifluoroacetic acid (TFA), in sealed 1-ml ampoules was from Pierce (Rockford, IL). MonoStandards, a 100 nmol mixture of fucose (Fuc), galactosamine (GalN), glucosamine (GlcN), galactose (Gal), glucose (Glc), and mannose (Man) were from Dionex (Sunnyvale, CA). 2.1.1. Notes on preparation of eluents 2.1.1.1. Eluent 1—HPLC grade water. The use of high-quality water is essential. The water should have as little dissolved carbon dioxide as possible and should be of high resistivity (18 MV). Biological contamination should be absent. Additionally, borate, a water contaminant that can break through water purification cartridges (prior to any other indication of depletion of the cartridge) can be removed by placing a BorateTrap cartridge (Dionex) between the pump and injection valve. Biological contamination is typically the source of unexpected glucose peaks after acid hydrolysis. 2.1.1.2. Eluent 2—0.2 M NaOH. It is extremely important to minimize contamination with carbonate, a divalent anion at pH>12, because it binds strongly to the columns and interferes with carbohydrate chromatography, causing a loss of resolution and efficiency. Commercially available sodium hydroxide pellets are covered with a thin layer of sodium carbonate and should not be used. Rather, a 50% w/w sodium hydroxide solution that is much lower in carbonate is the preferred source for NaOH. Diluting 20.8 ml of a 50% NaOH solution into 2 l of water yields a 0.2 M NaOH solution.
312
M. Weitzhandler et al. / J. Biochem. Biophys. Methods 60 (2004) 309–317
2.2. Methods Glycoprotein samples to be hydrolyzed were diluted to 340 Al with distilled, deionized water, and were placed in 1.5-ml Sarstedt screw cap microfuge tubes. Neat TFA (60 Al) was added to each diluted sample to produce a 2 N TFA solution. The tubes were capped and monosaccharides were released from the glycoproteins by acid hydrolysis at 100 jC for 5 h. Hydrolysates were centrifuged after incubation to unite the condensate with the bulk liquid. The hydrolysates were then vacuum dried on a Speed Vac concentrator (Savant model SVC100) to remove the volatile TFA in the hydrolysate. The dried samples were reconstituted in an appropriate volume of water. A commercially available mix of six monosaccharide standards (Dionex) was dissolved in 1 ml of water to yield a 0.1 mM solution of Fuc, GalN, GlcN, Gal, Glc, and Man. This solution was used for calibration. After the system response had stabilized (determined by injecting the standard solution), four consecutive standard injections were used for autocalibration. Sample injections for all experiments were made with either a Dionex AS50 or a Thermo Spectra-Physics AS3500 autosampler equipped with a 50-Al sample injection loop, 500-Al syringe, and injection valve fitted with a Tefzel rotor seal. A DX500 equipped with a CarboPac PA20 column (3150 mm) with/without an AminoTrap guard column (330 mm) was used, working at a flow rate of 0.5 ml/min. PA20 chromatography was performed at an isocratic concentration of 95% distilled, deionized water (eluent 1) and 5% 200 mM NaOH (eluent 2) for 12 min. A wash step consisting of 100% eluent 2 (200 mM NaOH) is then implemented for 8 min. Finally, the column is reequilibrated to the initial conditions for 12 min, after which the next injection is made. The elapsed time between sample injections was 32 min. The separated monosaccharides were detected using an ED40 or an ED50 electrochemical detector equipped with a disposable gold working electrode, a 2 mil gasket, and Ag/AgCl reference electrode. It is recommended that Ag/AgCl reference electrodes are replaced every 6 months. The waveform used was a quadruple potential pulsed amperometry waveform (E1=+0.1 V from 0 to 0.4 ms; E2= 2.0 V from 0.41 to 0.42 ms; E3=+0.6 V from 0.42 to 0.43 ms, E4= 0.1 V from 0.44 to 0.5 ms). The resulting chromatographic data were integrated and processed using PeakNet data reduction software. Quantification of monosaccharides from hydrolyzed glycoprotein samples was accomplished by using PeakNet 6. Briefly, data from a sequence of chromatographic runs were collected. Chromatographic peaks in every sequence were identified by comparing their retention times with those of the autocalibration chromatograms. Glycoprotein hydrolysate monosaccharide peaks were thus identified, quantified, and organized into spreadsheet format where the data was further processed using Microsoft Excel. 2.3. Equipment The heating block for 12-mm tubes and the heating module for the block (ReactiTherm) were from Pierce. The vacuum centrifuge/concentrator, refrigerated trap and vacuum pump (Speed Vac Centrifuge, model SVC100) were from Savant Instruments (Farmingdale, NY). The chromatograph (Dionex) consisted of a DX-500 system configured for carbohydrate analysis consisting of a gradient pump (GP40 or GP50), pulsed
M. Weitzhandler et al. / J. Biochem. Biophys. Methods 60 (2004) 309–317
313
electrochemical detector (ED40 or ED50), and Dell Pentium computer. The system was controlled by and data collected with Dionex PeakNet software. Sample injection was with either a Dionex AS50 or a Thermo Spectra Physics AS3500 autosampler (Fremont, CA) equipped with a 50-Al sample loop. The Rheodyne injection valve (Cotati, CA) was fitted with a Tefzel rotor seal to withstand the alkalinity of the eluents.
3. Results and discussion The separation of a mixture of six monosaccharides from glycoprotein compositional analysis is shown in Fig. 1. The separation is performed across a range of isocratic NaOH concentrations, demonstrating that the separation of monosaccharides on the PA20 column
Fig. 1. Separation of glycoprotein monosaccharides across a range of isocratic NaOH concentrations. (1) Fucose, (2) galactosamine, (3) glucosamine, (4) galactose, (5) glucose, and (6) mannose. Column: 3150 mm CarboPac PA20, eluent as described in figure, flow rate=0.5 ml/min. Detection using pulsed amperometry, quadruple potential waveform with a disposable gold electrode. Twenty picomoles of monosaccharide standard mix was injected.
314
M. Weitzhandler et al. / J. Biochem. Biophys. Methods 60 (2004) 309–317
is rugged. The improvement in the ruggedness of the separation compared to previous HPAE monosaccharide columns (CarboPac PA1 or more recently the CarboPac PA10) is due in part to the higher peak efficiencies exhibited by the PA20 column as a result of the smaller resin bead diameter (6.5 Am). Additionally, the selectivity of the PA20 phase was intentionally manipulated to exhibit a better separation of glucosamine and galactose than is possible on either the PA1 or PA10 columns. The new selectivity was achieved by optimizing the latex particle size and the degree of cross-linking on the smaller particlesized resin. By optimizing the selectivity with respect to glucosamine and galactose, the PA20 column yields separations that enable accurate quantification of peak areas for the respective sugars, which is particularly relevant for undergalactosylated glycoproteins such as therapeutic monoclonal antibodies [10]. Reproducibility of column performance was evaluated by several hundred consecutive injections of a mixture of six monosaccharide standards (Fig. 2). Previously, reproducibility of the chromatography of 400 consecutive injections of the mixture of six monosaccharides had been documented with a prototype CarboPac PA20 column [11]. Over a longer time period, during which the PA20 column was used intermittently, the column did not exhibit any indication of reduced performance after several thousand injections (data not shown). We then used the CarboPac PA20 and AminoTrap (330 mm format) to profile a sample that was difficult
Fig. 2. Chromatograms from schedule that exhibited 220 consecutive injections of a mixture of six monosaccharides using a disposable gold electrode. Column: 3150 mm CarboPac PA20, eluent was 10 mm NaOH, as described in methods, flow rate=0.5 ml/min. Detection using pulsed amperometry, quadruple potential waveform, with a disposable gold electrode. One hundred picomoles of monosaccharide standard mix was injected.
M. Weitzhandler et al. / J. Biochem. Biophys. Methods 60 (2004) 309–317
315
to analyze; specifically, a monoclonal antibody with less than 3% glycosylation. Fig. 3 compares the CarboPac PA20 chromatography of a monoclonal antibody acid hydrolysate in the absence of an AminoTrap guard cartridge versus in the presence of an AminoTrap guard cartridge. Monosaccharide peaks (fucose, glucosamine, galactose, and mannose) are apparent in the PA20 chromatography in both the presence of the AminoTrap (Fig. 3, top tracing) as well as in the absence of the AminoTrap (Fig. 3, second to bottom tracing). PA20 chromatography of the hydrolysate in the absence of the AminoTrap (Fig. 3, second from bottom tracing) revealed a somewhat noisy, uneven baseline with a minor shoulder peak coelution problem that potentially interferes with integration of the glucosamine peak (Fig. 3; peak 3, second from bottom tracing). Improvement in baseline stability as well as elimination of the glucosamine contaminant ‘‘shoulder’’ is evident when the AminoTrap is used (Fig. 3, top tracing). Additional peaks that eluted between the elution positions of
Fig. 3. Monosaccharide analysis of monoclonal antibody (MAb) hydrolysate. Top two tracings are MAb monosaccharide profile on CarboPac PA20 in the presence of the AminoTrap cartridge (20 Ag hydrolysate injected) and a mixture of six monosaccharide standards (100 pmol), respectively. Column: 3150 mm CarboPac PA20, eluent was 10 mm NaOH, as described in methods, flow rate=0.5 ml/min. Detection using pulsed amperometry, quadruple potential waveform, with a disposable gold electrode. Bottom two tracings, respectively, are the same as above except in the absence of the Aminotrap cartridge. Note the cleaner baseline and absence of contaminant peaks in the top profile (CarboPac PA20+Aminotrap).
316
M. Weitzhandler et al. / J. Biochem. Biophys. Methods 60 (2004) 309–317
fucose and galactosamine were moved out of the analysis when the AminoTrap was used (compare Fig. 3, top tracing with Fig. 3, second to bottom tracing). While a decrease in peak efficiencies of the monosaccharides occurs when the AminoTrap cartridge is used, the monosaccharides are still baseline resolved and the improvement in baseline stability achieved by removal of protein matrix background is significant [2]. An assessment of the sensitivity of the technique for monosaccharide analysis using the combination of the new column and a disposable gold electrode is depicted in Fig. 4, in which 200 fmol of the mixture of six monosaccharides are clearly detected. The narrower diameter column, improved peak efficiencies, and the absence of any electrode recession all contribute to improved sensitivity of the method. An assessment of the durability and reproducibility of the method using the PA20 column with a disposable gold electrode was made by doing a long-term study of several hundred successive injections of monosaccharides. Retention time RSDs for the monosaccharides through 220 consecutive injections were less than 1%. Retention times for the monosaccharides further did not exhibit any evidence of shortened retention times that would be an indication of diminished column capacity over the 220 injection schedule. As shown in Fig. 2, monosaccharide peak area response throughout this 220 injection sequence also exhibited a stable, consistent, and reproducible response that demonstrates the stability of response of the disposable gold working electrode. Similar long sequences of injections of monosaccharides have been performed with disposable gold electrodes from several different manufacturing lots, and all sequences demonstrated similar reproducibility. One caveat in using disposable gold electrodes is that the use of a nonrecessing waveform is required. If a waveform that uses gradually
Fig. 4. Low level separations of a mixture of six monosaccharides (200 fmol). Column: 3150 mm CarboPac PA20, eluent was 8 mM NaOH, flow rate=0.5 ml/min. Detection using pulsed amperometry, quadruple potential waveform, with a disposable gold electrode.
M. Weitzhandler et al. / J. Biochem. Biophys. Methods 60 (2004) 309–317
317
recessing oxidative cleaning is employed, we have found that the gold electrode surface lasts less than 12 h with the peak area response disappearing within 6 h. Examination of the disposable electrode after the twelfth injection of this sequence revealed that the gold surface was no longer present, thereby necessitating the use of the fast, reductive cleaning waveform recommended in this manuscript. The combination of the new CarboPac PA20 column combined with the use of disposable gold electrodes and the nonrecessing quadruple potential waveform is expected to provide the analyst with fast, efficient, and reproducible monosaccharide separations with good spacing compared to the performance of previous columns. Finally, the use of disposable electrodes further eliminates the need for time-consuming polishing and reequilibration of working electrodes.
Acknowledgements The authors would like to thank Robin Higashi for her editorial assistance with this manuscript.
References [1] Hardy MR, Townsend RR, Lee YC. Monosaccharide analysis of glycoconjugates by anion exchange chromatography with pulsed amperometric detection. Anal Biochem 1988;170:54 – 62. [2] Weitzhandler M, Pohl C, Rohrer JS, Narayanan L, Slingsby R, Avdalovic N. Eliminating amino acid and peptide interference in high-performance anion-exchange pulsed amperometric detection glycoprotein monosaccharide analysis. Anal Biochem 1996;241:128 – 34. [3] Anumula KR. High-sensitivity and high-resolution methods for glycoprotein analysis. Anal Biochem 2000;283:17 – 26. [4] Kuster B, Krogh TN, Mortz E, Harvey DJ. Glycosylation analysis of gel-separated proteins. Proteomics 2001;1:350 – 61. [5] Jensen MB, Johnson DC. Fast wave forms for pulsed electrochemical detection of glucose by incorporation of reductive desorption of oxidation products. Anal Chem 1997;69(9):1776 – 81. [6] Rocklin RD, Clarke AP, Weitzhandler M. Improved long-term reproducibility for pulsed amperometric detection of carbohydrates via a new quadruple-potential waveform. Anal Chem 1998;70(8):1496 – 501. [7] Clarke AP, Jandik P, Rocklin RD, Liu Y, Avdalovic N. An integrated amperometry waveform for the direct, sensitive detection of amino acids and amino sugars following anion-exchange chromatography. Anal Chem 1999;71(14):2774 – 81. [8] Rohrer JS, Thayer J, Weitzhandler M, Avdalovic N. Analysis of the N-acetylneuraminic acid and N-glycolylneuraminic acid contents of glycoproteins by high-PH anion-exchange chromatography with pulsed amperometric detection (HPAEC/PAD). Glycobiology 1998;8(1):35 – 43. [9] Cheng J, Jandik J, Avdalovic N. Development and characterization of microfabricated disposable gold working electrodes for high-performance ion chromatography and integrated pulsed amperometric detection. Anal Chem 2003;75:572 – 9. [10] Weitzhandler M, Hardy M, Co MS, Avdalovic N. Analysis of carbohydrates on IgG preparations. J Pharm Sci 1994;83(12):1670 – 5. [11] Heckenberg A, Weitzhandler M, Jandik P, Barreto V. Rapid, efficient separations of glycoprotein monosaccharides using a new anion-exchange separator column and electrochemical detection with disposable gold electrodes. LC-GC Appl Noteb 2002 (June);83:12 – 3.
