Efficient Electrophoretic Method to Remove Neutral Additives from

Mar 11, 2011 - Pei-Jing Pai, Stephanie M. Cologna, William K. Russell, Gyula Vigh, and David H. Russell*. Department of Chemistry, Texas A&M Universit...
0 downloads 0 Views 2MB Size
Download PDF
TECHNICAL NOTE pubs.acs.org/ac

Efficient Electrophoretic Method to Remove Neutral Additives from Protein Solutions Followed by Mass Spectrometry Analysis Pei-Jing Pai, Stephanie M. Cologna, William K. Russell, Gyula Vigh, and David H. Russell* Department of Chemistry, Texas A&M University, College Station, Texas 77843, United States

bS Supporting Information ABSTRACT: A mass spectrometry (MS)-compatible, isoelectric point-based separation method for removal of neutral additives from protein solutions is described. The separation is performed by electrophoretic migration and trapping using a device referred to as membrane separated wells for isoelectric focusing and trapping (MSWIFT). Electrophoretic separation in the MSWIFT device is fast; the entire process can be carried out in a matter of minutes, and it does not require further sample cleanup prior to MS analysis. Proof-of-concept experiments in which neutral additives (e.g., Triton X-100, Tween 20, poly(ethylene glycol)) are removed from protein solutions using the MSWIFT device followed by MS analysis are described. Coupling the MSWIFT separation with ion mobility MS provides additional separation via the gas phase and assists in achieving higher quality ESI mass spectra when small amounts of additives remain in solution.

P

rotein solubility is critical for studies of solution-phase proteins, protein mixtures, and protein complexes. Oftentimes, detergents are the common additives used to improve protein solubility and stability, especially for hydrophobic membrane proteins.1,2 Neutral polymers such as poly(ethylene glycol) (PEG) are frequently used as an additive to assist in crystal formation for crystallographic studies,2 4 and are also widely used in protein separation and purification.5,6 Mass spectrometry (MS) is now an essential technique used in protein studies, including the analysis of protein crystals to confirm that the protein has not been altered during the crystallization process;7,8 however, detergents and other additives are not typically compatible with MS. Although matrix-assisted laser desorption/ionization (MALDI) is more tolerant to detergents and polymer additives compared to electrospray ionization (ESI),7 in both cases signals corresponding to detergent or polymer ions dominate the mass spectrum or result in protein ion suppression effects even in protein solutions containing small amounts of detergents or polymers.9 Although methods for removing additives from protein solutions date back more than 30 years, a universally adopted approach is still lacking.10 12 Commercial devices that rely on techniques such as dialysis, solid-phase extraction, and membrane filtration are available for the removal of SDS, Tween, Triton-X, and other detergents,13,14 but these methods are timeconsuming, costly, and still problematic when handling protein samples with a high concentration of additives. Electrophoretic methods, i.e., capillary electrophoresis can be used to remove nonionic and zwitterionic detergents,10,11,15 but inherent challenges related to this method include (i) the low sample volume loading and (ii) high protein concentrations required to obtained sufficient ultraviolet (UV) or MS data. r 2011 American Chemical Society

Here, we describe a versatile, high-throughput method to remove neutral additives from protein solutions and protein crystals. This method, utilizes an isoelectric point (pI)-based separation device referred to as membrane separated wells for isoelectric focusing and trapping (MSWIFT), which separates and traps the protein on the basis of pI.16,17 The protein migrates under the influence of an electric field, whereas the neutral additives remain in the well in which the sample is initially placed. Following isoelectric separation and trapping the “purified” protein can be analyzed without additional cleanup steps prior to MS analysis. A major advantage of using the MSWIFT device is the ability to simultaneously remove detergents as well as prefractionate proteins based on their isoelectric point. This is not possible with commercial detergent removal devices which require additional sample preparation steps and potential sample loss. It is important to note that while the MSWIFT device provides an excellent platform to perform these experiments; this additive removal method can be performed using an array of commercially available isoelectric focusing (IEF) devices such as the Zoom IEF Fractionator (Invitrogen), the OFFGEL fractionator (Agilent), or the microRotofor (Bio-Rad). Using the MSWIFT device is advantageous over the commercial devices because (i) membranes are synthesized in-house and can be tuned to a specific experiment, (ii) proteins are separated in the isoelectric trapping mode and therefore carrier ampholytes are not used, (iii) fast separations occur by using high field strengths owing to the design of the device, and (iv) an array of solvent Received: November 11, 2010 Accepted: February 25, 2011 Published: March 11, 2011 2814

dx.doi.org/10.1021/ac1029743 | Anal. Chem. 2011, 83, 2814–2818

Analytical Chemistry systems are compatible with the membranes which is important for analyzing more hydrophobic samples. Experiments performed in our laboratory have tested the compatibility of poly(vinyl alcohol) (PVA) membranes used in the MSWIFT device in a variety of organic solvents (data not published). In addition, IET separations have been reported using mixed methanol aqueous solutions.18 This is an important feature to point out the potential of using the MSWIFT device for hydrophobic protein separations and cleanup while reducing and/or eliminating protein precipitation. Results from proof-of-concept experiments are included for both proteins in solution and protein crystals that are redissolved and analyzed by ESI-MS. In cases where the protein additive interaction is strong, we show that ion mobility MS analysis provides an additional dimension of separation for elimination of chemical background from the additive.

’ EXPERIMENTAL SECTION All chemicals were purchased from Sigma (St. Louis, MO) and used as received unless noted otherwise. HPLC grade methanol was obtained from EMD Chemicals Inc. (Gibbstown, NJ). All solutions were prepared using purified 18 MΩ water (Barnstead International, Dubuque, IA) The MSWIFT device has been described in previous reports.16,17 Briefly, MSWIFT of the neutral detergent and polymer removal experiments were carried out where the MSWIFT device was assembled to contain three separation wells and the volume in each of the wells ranged from 200 to 220 μL. Buffering membranes, which were synthesized in-house, separate each well,19 21 and the pH of the buffering membranes were pH 3.8, pH 4.8, pH 5.2, and pH 9.8. Ampholytic buffers were used as pH biasers to maintain the pH and charge of each compartment as previously described.22 The anode and cathode solution were 3 mM methanesulfonic acid and 3 mM sodium hydroxide, respectively. Solutions of myoglobin (100 μg, pI 7.2) containing either the neutral detergent (1% Tween 20 or 1% Triton X-100) or the neutral polymer (2% or 20% PEG 3000) were prepared to test the protein additive separation. The protein solutions were incubated at 4 °C overnight prior to MSWIFT separation. Solutions of nickel superoxide dismutase Y9F (NiSOD Y9F; pI 6.65) were prepared by dissolving crystals in a solution containing 8.0 M urea, 2.0 M thiourea, and 50 mM ammonium bicarbonate buffer (pH 7.8). The MSWIFT separation of the neutral additivecontaining protein solutions (myoglobin and protein crystal) were performed by loading the sample solution in an acidic well (pH 3.8 pH 4.8) followed by application of an electric field (500 V for 5 min and then 1000 V ∼5 10 min). Following separation, the resulting solutions were mixed in a 1:1 volume ratio with an electrospray (ESI) solution (50% methanol with 0.1% formic acid) and then directly analyzed using a SYNAPT G2 HDMS mass spectrometer (Waters Corp., Milford, MA) equipped with a nanoESI source without any further cleanup step. Data analysis and extraction were performed using the MassLynx and DriftScope software packages (Waters Corp., Milford, MA). ’ RESULTS AND DISCUSSION MSWIFT is a solution-phase electrophoresis device which provides high-throughput separations based on the isoelectric point (pI) of analytes.16,17,19,23 Our previous report emphasized prefractionation of protein digests and large scale proteomics profiling experiments to improve peptide detection and protein

TECHNICAL NOTE

Scheme 1. Experimental Design Used to Remove Neutral Additives from Protein Solutions Using the MSWIFT Device: (A) Sample Loading and (B) the Completion of Electrophoresis and Trapping

identification.17 The current study focuses on another important problem in the area of proteomics and more generally related to protein science, viz., isoelectric point fractionation of proteins and removal of neutral additives such as detergents or polymers. The experimental design for the removal of neutral additives from protein solutions using the MSWIFT device is shown in Scheme 1. Protein solutions containing neutral additives are loaded in a well where the protein pI does not fall between the pH values of the buffering membranes (Scheme 1A). Upon application of an electric field, the protein migrates into a well where the pI value is bracketed by the pH values of the buffering membranes. Because the net charge of the neutral additive is zero, the electrophoretic mobility is zero and it does not migrate under the electric field. Proteins can be stripped of the additive by migration according to their charge and trapped in a well where the pH values of the buffering membranes bracket the protein pI (Scheme 1B). It is important to note that for this experiment to work as designed, the additive must dissociate from the protein and the best results are obtained for protein additive systems having relatively weak binding forces. The utility of using the MSWIFT device for neutral additive removal from protein solutions prior to MS analysis is illustrated by several proof-of-concept experiments. Myoglobin is a mass spectrometry friendly protein, i.e., the protein is easily ionized by both MALDI and ESI and the resulting ions are stable on the time scale of the mass analysis measurement. In spite of this fact, ionization of myoglobin is adversely affected by the presence of additives such as detergents and polymers. This statement is illustrated by the data shown in Figure 1, where the effects of two common neutral detergents and one neutral polymer are examined: Tween20, Triton X-100, and poly(ethylene glycol) (PEG). The ESI-MS spectra of myoglobin (pI 7.2) solutions containing 2815

dx.doi.org/10.1021/ac1029743 |Anal. Chem. 2011, 83, 2814–2818

Analytical Chemistry

TECHNICAL NOTE

Figure 1. ESI mass spectra obtained from a solution of myoglobin (pI 7.2) containing (A) 1% Tween 20, (C) 1% Triton X-100, (E) 2% PEG, and (G) 20% PEG prior to cleanup and after additive removal by MSWIFT: (B) 1% Tween 20 removal, (D) 1% Triton X-100 removal, (F) 2% PEG removal, and (H) 20% PEG removal. Protein charge states are denoted on the spectra from the purified protein.

1% Tween 20 (Figure 1A), 1% Triton X-100 (Figure 1C), 2% PEG (Figure 1E), and 20% PEG (Figure 1G) are dominated by detergent or polymer signals. Note that myoglobin signals are

not observed in any of the spectra. MSWIFT separation of myoglobin from neutral additive containing solutions was performed by loading samples in an acidic well (pH 3.8 pH 4.8). 2816

dx.doi.org/10.1021/ac1029743 |Anal. Chem. 2011, 83, 2814–2818

Analytical Chemistry

TECHNICAL NOTE

Figure 2. (A) Electrospray mass spectrum of nickel superoxide dismutase Y9F mutant (NiSOD Y9F, pI 6.65, MW 13 185) acquired from the third fraction of MSWIFT (pH 5.2 9.8). Charge states corresponding to the protein NiSOD Y9F are provided. (B) Two-dimensional plot of arrival time distribution (ATD) versus m/z for the ESI-IM-MS analysis of MSWIFT fraction 3 (pH 5.2 9.8) of NiSOD Y9F. The circle indicates the trend line corresponding to NiSOD Y9F protein signals. (C) Electrospray mass spectrum of fraction 3 (pH 5.2 9.8) selected from the ion mobility profile. (D) The deconvoluted ESI mass spectrum.

The electric field was applied to the device for ∼10 min. Under these conditions, myoglobin migrated cathodically until it was trapped in a well where the pH values of the buffering membranes (pH 5.2 and pH 9.8) bracket the pI of myoglobin. The ESI-MS spectra obtained from aliquots of this well are rich in myoglobin signals (Figure 1B,D,F,H). The low abundance signals of Triton X-100 and PEG observed in Figure 1D,F,H indicate that trace amount of additives migrated together with protein under an electric field. Although some additives were still present in the protein solution, the majority of the detergent is removed by MSWIFT and the protein signals are clearly detected in the MS spectrum. We have compared the MSWIFT device with a commercial device for detergent removal using myoglobin solutions containing 1% Tween 20 and 20% PEG. Following removal, the ESI-MS spectra of each preparation was acquired; these spectra are dominated by signals from Tween 20 and PEG, and signals corresponding to myoglobin are not observed (see the Supporting Information Figures S-1A,B). These results clearly illustrate the utility of the MSWIFT device for removal of neutral additives from the protein solutions more efficiently than the commercial device and can be used for higher concentrations of additives. The separation of myoglobin clearly illustrates the applicability and advantage of using MSWIFT for isolation and purification of intact proteins which contain neutral additives. An important application of this experiment is using MS analysis to confirm molecular weight determination of proteins following X-ray crystallographic analysis. To illustrate this, protein crystals of nickel superoxide dismutase Y9F mutant (NiSOD Y9F, pI 6.65) were

grown in crystallization screening media consisting of 15% methoxypolyethylene glycol (MPEG) 5000, 50 mM HEPES, 50 mM CaCl2, and 5% methanol.24 Urea and thiourea were used to facilitate dissolving of the protein crystals and to break up the possible intermolecular hydrogen bonding complexes of proteins and MPEG.15 Following MSWIFT separation (∼15 min), NiSOD Y9F migrated from the acidic loading well (pH 3.8 4.8) to the well bracketing the pI of NiSOD Y9F (pH 5.2 9.8). The resulting solution was analyzed using MALDI-MS (data not shown) and nano-ESI-MS without further sample cleanup steps (see Figure 2A). The ion signals derived from MPEG are significantly attenuated following MSWIFT separation, but the protein signals are superimposed on a strong background of detergent signals. Ion mobility-mass spectrometry (IM-MS) has been shown to be highly effective for distinguishing analyte signals from chemical background25,26 (see Figure 2). IM-MS provides two-dimensional (2D) gas phase separations (microsecond to millisecond time scale) on the basis of both analyte ion shape and mass-tocharge (m/z) ratio (see Figure 2B). That is, different chemical species appear on distinct trend lines in the 2D plot of arrival time distribution (ATD) versus m/z26,27 owing to a correlation of ion size with mass to charge. As shown in Figure 2B, the 2D plot of the NiSOD Y9F protein from the MSWIFT fraction (pH 5.2 9.8) shows several different trend lines. The protein signals were extracted from the 2D plot by selecting the corresponding trend line (circled in Figure 2B) to yield the spectrum shown in Figure 2C. The ESI spectrum of NiSOD Y9F is clean and the measured mass (13 201 Da, shown in Figure 2D) is in good 2817

dx.doi.org/10.1021/ac1029743 |Anal. Chem. 2011, 83, 2814–2818

Analytical Chemistry agreement with the expected mass of oxidized NiSOD Y9F (13 201 Da). For comparison, the ESI-IM-MS spectrum was acquired for the dissolved NiSOD Y9F crystals without MSWIFT separation. As shown in Figure S-2A in the Supporting Information, the raw ESI spectrum of NiSOD Y9F is dominated by polymer type signals. The 2D plot of arrival time distribution versus m/z is also provided (Supporting Information, Figure S-2B). Although the mass spectrum is dominated by signals from the neutral additive, in the 2D plot a trend line corresponding to the protein signals is clearly visible (see the Supporting Information, Figure S-2C). The spectrum quality is not as good as that obtained from MSWIFT coupling with IM-MS analysis. These results underscore the utility of using MSWIFT to obtain the unambiguous molecular weight measurement of proteins from protein crystals. Quantitative MSWIFT studies of protein separations have been performed. On the basis of experiments carried out on standard proteins, the estimated protein recoveries ranged from >95% down to ∼50% depending on the protein. Theoretically, there is no maximum sample capacity of the MSWIFT device since it does not require binding to a substrate (i.e., chromatography). Protein mixtures on the milligram scale (e.g., 10 mg) and tryptic digests of standard protein mixtures and cell lysates at the low microgram level (10 50 μg) have been separated by the MSWIFT device.

’ CONCLUSIONS MSWIFT is an MS-compatible electrophoretic device that provides high-throughput fractionation of ampholytic components based on pI prior to downstream analysis using MS without the need for sample cleanup. Removal of several neutral additives by fractionation of proteins according to pI using the MSWIFT device is shown in which the purified protein is isolated followed by direct MALDI and ESI-MS analysis. Several different neutral additives at a range of concentrations have been removed allowing for intact protein analysis. Proteins that have been crystallized in the presence of MPEG can be redissolved and further analyzed using this method. The MSWIFT neutral additive removal method can be coupled offline with ion mobility mass spectrometry, which provides an additional dimension of separation, which is particularly useful when a small amount of additive is present. ’ ASSOCIATED CONTENT

bS

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

TECHNICAL NOTE

’ REFERENCES (1) Man, P.; Novak, P.; Cebecauer, M.; Horvath, O.; Fiserova, A.; Havlicek, V.; Bezouska, K. Proteomics 2005, 5, 113–122. (2) Newby, Z. E. R.; O’Connell, J. D.; Gruswitz, F.; Hays, F. A.; Harries, W. E. C.; Harwood, I. M.; Ho, J. D.; Lee, J. K.; Savage, D. F.; Miercke, L. J. W.; Stroud, R. M. Nat. Protoc. 2009, 4, 619–637. (3) Nettleship, J. E.; Walter, T. S.; Aplin, R.; Stammers, D. K.; Owens, R. J. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2005, 61, 643–645. (4) Patel, S.; Cudney, B.; McPherson, A. Biochem. Biophys. Res. Commun. 1995, 207, 819–828. (5) Gagnon, P.; Godfrey, B.; Ladd, D. J. Chromatogr., A 1996, 743, 51–55. (6) Larive, C. K.; Lunte, S. M.; Zhong, M.; Perkins, M. D.; Wilson, G. S.; Gokulrangan, G.; Williams, T.; Afroz, F.; Schoneich, C.; Derrick, T. S.; Middaugh, C. R.; Bogdanowich-Knipp, S. Anal. Chem. 1999, 71, 389R–423R. (7) Funk, J.; Li, X. P.; Franz, T. Rapid Commun. Mass Spectrom. 2005, 19, 2986–2988. (8) Rundlett, K. L.; Armstrong, D. W. Anal. Chem. 1996, 68, 3493–3497. (9) Loo, R. R. O.; Dales, N.; Andrews, P. C. Protein Sci. 1994, 3, 1975–1983. (10) Dresdner, G.; Ciddresdner, H. Anal. Biochem. 1977, 78, 171–181. (11) Stralfors, P.; Tornqvist, H.; Jergil, B.; Belfrage, P. Biochem. Biophys. Acta 1978, 533, 90–97. (12) Bhakdi, S. J. Biochem. Biophys. Methods 1980, 2, 79–90. (13) Zhao, C.; O’Connor, P. B. Anal. Biochem. 2007, 365, 283–285. (14) Rigaud, J. L.; Levy, D.; Mosser, G.; Lambert, O. Eur. Biophys. J. Biophys. Lett. 1998, 27, 305–319. (15) Shen, J.; Buko, A. Anal. Biochem. 2002, 311, 80–83. (16) Lim, P.; North, R.; Vigh, G. Electrophoresis 2007, 28, 1851–1859. (17) Cologna, S. M.; Russell, W. K.; Lim, P. J.; Vigh, G.; Russell, D. H. J. Am. Soc. Mass Spectrom. 2010, 21, 1612–1619. (18) Shave, E.; Vigh, G. J. Chromatogr., A 2004, 1036, 3–6. (19) Fleisher-Craver, H. C.; Vigh, G. Electrophoresis 2008, 29, 4247–4256. (20) Lalwani, S.; Shave, E.; Vigh, G. Electrophoresis 2004, 25, 3323–3330. (21) Lalwani, S.; Shave, E.; Fleisher, H. C.; Nzeadibe, K.; Busby, M. B.; Vigh, G. Electrophoresis 2004, 25, 2128–2138. (22) Shave, E.; Vigh, G. Electrophoresis 2004, 25, 381–387. (23) Shave, E.; Vigh, G. Electrophoresis 2007, 28, 587–594. (24) Barondeau, D. P.; Kassmann, C. J.; Bruns, C. K.; Tainer, J. A.; Getzoff, E. D. Biochemistry 2004, 43, 8038–8047. (25) Fernandez-Lima, F. A.; Becker, C.; McKenna, A. M.; Rodgers, R. P.; Marshall, A. G.; Russell, D. H. Anal. Chem. 2009, 81, 9941–9947. (26) McLean, J. A.; Ruotolo, B. T.; Gillig, K. J.; Russell, D. H. Int. J. Mass Spectrom. 2005, 240, 301–315. (27) Ruotolo, B. T.; Gillig, K. J.; Woods, A. S.; Egan, T. F.; Ugarov, M. V.; Schultz, J. A.; Russell, D. H. Anal. Chem. 2004, 76, 6727–6733.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Phone: (979) 845-3345.

’ ACKNOWLEDGMENT P.-J.P. and S.M.C. contributed equally to this work. The nickel superoxide dismutase crystals were provided as a generous gift from Mr. James Vranish and Dr. David Barondeau, Texas A&M University. This work has been supported by the Department of Energy Grant BES-DE-FG-04ER-15520 and the National Science Foundation Grant DBI-0821700. 2818

dx.doi.org/10.1021/ac1029743 |Anal. Chem. 2011, 83, 2814–2818