Microfluidic Electrocapture-Assisted Mass Spectrometry of Membrane

Anal. Chem. 2008, 80, 7116–7120

Microfluidic Electrocapture-Assisted Mass Spectrometry of Membrane-Associated Polypeptides Mohammadreza Shariatgorji,† Juan Astorga-Wells,‡ Hans Jornvall,‡ and Leopold L. Ilag*,† Department of Analytical Chemistry, Stockholm University, SE-106 91 Stockholm, Sweden, and Department of Medical Biochemistry and Biophysics, Chemistry I, Karolinska Institute, SE-171 77 Stockholm, Sweden To isolate membrane-associated proteins, which play diverse structural, catalytic, and regulatory roles in cells, they are often initially solubilized in detergents. Although detergents are essential for purifying membrane proteins, they tend to interfere strongly with subsequent analyses. A microfluidic method is presented here that surmounts this problem, allowing well-resolved mass spectra of test membrane-associated polypeptides, and their complexes with ions and detergents, to be acquired. As a front-end module it allows access to other advanced mass spectrometric strategies to be utilized toward defining biomolecular interactions. This opens up a new avenue for studying complexation and analysis of membrane proteins of general importance. Membrane proteins are essential components of the proteome and important drug targets.1 Therefore, numerous attempts have been made to analyze them, with varying degrees of success. They are difficult to study because their yields tend to be low, they are often difficult to solubilize, they have strong tendencies to aggregate,2 and the detergents used to solubilize them can strongly interfere with their subsequent analysis. The undesirable influence of detergents on protein signals obtained from ESI-MS can be reduced sufficiently to acquire informative spectra from many samples by simply decreasing the detergent concentration below a protein-specific threshold level. Fluorinated surfactants such as perfluorooctanoic acid and perfluorooctanesulfonic acid, which have high volatility and weak intermolecular interactions due to their low surface energy, have thus been used as electrospray compatible surfactants.3 Low electrospray signal suppression has been achieved for small molecules, soluble protein, and peptides after surfactant evaporation.3 Thus finding other quick, effective,and more versatile ways to separate proteins from interfering detergents would clearly be highly beneficial for analyses of membrane-associated polypeptides. Extraction of the * To whom correspondence should be addressed. Phone: +46 8 16 24 35. Fax: +46 8 15 63 91. E-mail: [email protected]. † Stockholm University. ‡ Karolinska Institute. (1) Fang, Y.; Frutos, A. G.; Lahiri, J. J. Am. Chem. Soc. 2002, 124, 2394–2395. (2) Gutmann, D. A. P.; Mizohata, E.; Newstead, S.; Ferrandon, S.; Henderson, P. J. F.; Van Veen, H. W.; Byrne, B. Protein Sci. 2007, 16, 1422–1428. (3) Ishihama, Y.; Katayama, H.; Asakawa, N. Anal. Biochem. 2000, 287, 45– 54.

7116

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

Figure 1. Schematic diagram of the electrocapture system used for separating detergent from soluble and membrane-associated polypeptides.

protein into an appropriate organic solvent,4 dialysis,5 ion exchange chromatography, gel chromatography, and hydrophobic adsorption6 are the main techniques often employed for detergent removal from protein samples. Microfluidic electrocapture (Figure 1) is a cleanup and separation technique that has been recently developed for preconcentrating samples,7 separating analytes of interest from interferents,8 and facilitating microfluidic reactions.9 An electrocapture cell consists of a thin PEEK tube (e.g., with a 125 µM diameter), interrupted by two small junctions of cation exchange Nafion membrane. These junctions are immersed in two separate buffer reservoirs. By application of an appropriate voltage, an electric field gradient is generated between the anodic and cathodic zones counter to the direction of the hydrodynamic flow. Charged molecules are then concentrated in a sharp zone in which the hydrodynamic and electric field forces are balanced. Therefore, it is possible for nonionic compounds to be swept away from the system while slowly migrating ionic compounds are captured. Here, we demonstrate the potential utility of electrocapture for separating nonionic detergents from the membraneassociated polypeptides bacteriorhodopsin and gramicidin prior to ESI-MS analysis. (4) Barnidge, D. R.; Dratz, E. A.; Jesaitis, A. J.; Sunner, J. Anal. Biochem. 1999, 269, 1–9. (5) Ohlendieck, K. Methods in Molecular Biology, Protein Purification, 2nd ed.; Humana Press: Totowa, NJ, 2004; Vol. 244, pp 295-300. (6) Von Jagow, G.; Link, T. A.; Schagger, H. Membrane Protein Purification and Crystallization, 2nd ed.; Academic Press: Amsterdam, The Netherlands, 2003; pp 1-18. (7) Astorga-Wells, J.; Swerdlow, H. Anal. Chem. 2003, 75, 5207–5212. (8) Astorga-Wells, J.; Jo ¨rnvall, H.; Bergman, T. Anal. Chem. 2003, 75, 5213– 5219. (9) Astorga-Wells, J.; Bergman, T.; Jo¨rnvall, H. Anal. Chem. 2004, 76, 2425– 2429. 10.1021/ac800877k CCC: $40.75  2008 American Chemical Society Published on Web 08/12/2008

EXPERIMENTAL SECTION Reagents and Chemicals. Methanol was purchased from BDH (Poole, England) and acetonitrile, gramicidin, R-lactalbumin, N-octyl β-D glucopyranoside (OG), ammonia, and formic acid from Sigma. Stock solutions of 6% N-octyl β-D glucopyranoside (OG) in water, 2 µg/µL R-lactalbumin in water, and 10 µg/µL of gramicidin in methanol were prepared. Water was purified by a Millipore purification system with a resistivity of 18 MΩ cm. Bacteriorhodopsin was supplied as a gift from Professor Oesterhelt’s group (Max Planck Institute for Biochemistry, Martinsried, Germany) through Dr. Huseyin Besir, EMBL, Germany. Electrocapture System and Operation. The electrocapture cell was fabricated as previously reported.8,9 Briefly, two small openings 2 cm apart were created in a piece of PEEK tubing (127 µm i.d., 512 µm o.d.; Upchurch, Oak Harbor, WA) using a scalpel. The holes were covered with 330 µm i.d., 610 µm o.d. cationselective membrane of poly(tetrafluoroethylene-sulfonate) tubing (Permapure Inc., Toms River, NY). The covered openings were placed into two separate 500 µL Eppendorf tubes (Hamburg, Germany). The electrocapture instrument (Biomotif AB, Danderyd, Sweden) used for all of the experiments consists of platinum electrodes placed on top of a cell holder, a microinjector, PC-controlled syringe pump, and power supply. Electrocapture software controls and records flow rate, voltage, and current. The instrument was operated with the anode located at the upstream electric junction of the cell. Platinum electrodes were placed in the electrode chambers filled with appropriate buffer solutions. Hydrodynamic flow was provided by the syringe pump equipped with a 250 µL gastight syringe and filled with suitable buffer. An ISCO CV4 capillary UV-vis detector was connected to the electrocapture system and recorded UV electrocapturegrams. Different concentrations of polypeptides and OG were selected to evaluate the feasibility of the technique to handle different concentrations. The pH of the sample solutions and washing buffers were selected to be above the pI of the polypeptides with the aim of trapping negatively charged polypeptides since the electrocapture system works slightly better for trapping anions. Three model peptide/protein samples containing OG were prepared with the following concentrations: 0.1 µg/µL R-lactalbumin and 0.5% OG dissolved in 10 mM pH 7 ammonium formate with 10% acetonitrile; 0.48 µg/µL gramicidin and 1.4% OG dissolved in 100 mM pH 10 ammonium formate; and 0.1 µg/µL bacteriorhodopsin, 0.4% OG dissolved in 20 mM pH 10 ammonium formate with 30% methanol. The same concentrations of gramicidin and OG were also prepared in 100 mM of Tris-HCl and 10 mM ammonium formate to evaluate the effect of the buffer compositions on the capturing process. The same buffers used in sample preparation were employed as washing buffer during electrocapture except for the gramicidin samples the buffers of which were 10 mM ammonium formate and 10 mM Tris-HCl both buffered at pH 10. The samples were injected using a 3 µL loop while the washing buffer was passing through the system, partially sweeping out the detergent. A capturing voltage of 100 V/cm was applied with a buffer flow rate of 0.25 µL/min. The capturing time was 30 min for cleaning up of R-lactalbumin and gramicidin and 60 min for bacteriorhodopsin. The UV absorbance at 218 nm and the voltage/current profile of

the system were continuously recorded during the experiment. The captured fractions were collected after turning off the voltage and were introduced into the mass spectrometer. To obtain stable, intense signals both electrocaptured and crude samples of gramicidin and bacteriorhodopsin were acidified using formic acid solution. All mass spectra were acquired using a MicromassWaters Q-Tof 1 with a 3.6 GHz TDC card instrument in positive ion mode. All collected fractions were electrosprayed using a metal-coated borosilicate capillary emitter (Proxeon ES387). The mass spectrometer settings were as follows: capillary voltage 1.3 kV, cone voltage 35 V, extractor 4 V, source block temperature 120 °C, and collision energy 5 V. The experiments were done in triplicate to confirm the reproducibility RESULTS AND DISCUSSION Dodecyl maltoside (DDM) and N-octyl β-D glucopyranoside (OG) as nonionic detergents have higher protein-specific threshold levels as compared to some commonly used detergents such as SDS thus making them electrospray compatible detergents.10 However a cleanup technique is often necessary to decrease the concentration of detergents below the threshold to prevent masking and suppressing the signals from the analyte of interest. Between DDM and OG, the latter is slightly more challenging, so it was chosen as the test detergent in attempts to establish the present technique. The first test protein preparation was a simple solution of the soluble protein R-lactalbumin in the presence of OG, to benchmark starting conditions for separating intact proteins from detergents. In these experiments, R-lactalbumin was negatively charged since the running buffer has a pH above the protein’s isoelectric point. Under this polarity, flow rate, and voltage, the protein was captured in the electric field since the electrophoretic velocity (toward the upstream anode electrode) has the same magnitude but opposite direction than velocity of the flow. On the other hand, the uncharged detergent was swept out from the system by the flow stream since it is electrophoretically neutral. Figure 2a shows the electrocapturegram and current/voltage profile from this experiment. Both electrocaptured fractions and crude samples were introduced to the nanoelectrospray mass spectrometer using the same instrument settings. Figure 2b shows a mass spectrum of the crude sample in which the protein peaks are not discernible due to masking by peaks from detergent clusters and another spectrum in which the protein peaks are clearly visible obtained after separating the protein from the detergent by electrocapture (Figure 2c). For a more challenging application, the capacity of the electrocapture system to remove nonionic detergent from gramicidin prior to ESI-MS was tested using the previously described setup. Gramicidin A is a linear hydrophobic polypeptide with no readily ionizable groups (formyl-L-X-Gly-L-Ala-D-Leu-L-Ala-D-Val-LVal-D-Val-L-Trp-D-Leu-L-Y-D-Leu-L-Trp-D-Leu-L-Trp-ethanolamine) that is synthesized by Bacillus brevis and as a dimer forms selective channels for small monovalent cations with an internal diameter of 3.7 Å.11 Comparison of parts a and b of Figure 3 show that detergent removal by electrocapture is a remarkably effective cleanup (10) Loo, P.; Dales, N.; Andrews, P. C. Protein Sci. 1994, 3, 1975–1983. (11) Seoh, S.; Busath, D. Biophys. J. 1993, 64, 1017–1028.


7117

Figure 3. Separation of detergent from the membrane-associated polypeptide. ESI mass spectrum of OG solubilized gramicidin in ammonium formate. (b) ESI spectrum of the sample after cleaning up by electrocapture. Asterisks show some of the major peaks of OG clusters. The model structure of gramicidin was produced based on the protein databank coordinates 1GRM using Rasmol.18

Figure 2. Separation of detergent from a soluble protein.(a) Electrocapturegram of R-lactalbumin dissolved with OG recorded at 218 nm. Inset shows current and capture/release potential profiles of the cleanup process. (b) ESI mass spectrum of a crude protein-detergent sample. Asterisks show some of the major peaks of OG clusters. (c) ESI mass spectrum of the protein-detergent sample after clean up by electrocapture. Charge states of the protein are indicated above the peaks. The model structure of the R-lactalbumin was produced based on the protein data bank coordinates 1HFZ using Rasmol.18 7118 Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

technique, for this test system. Despite using a relatively high concentration of gramicidin, it is possible to gain better sensitivity by increasing the cleanup time and injection volume. However, the volume injected is limited because of the relatively low flow rate of the system. We have tried a sample with 0.01 µg/µL gramicidin (48 times less concentrated) and the same concentration of OG that was used before. By injecting 5 µL of the sample and cleanup for 2 h, we got discernible signal for gramicidin. Indeed, given the lack of easily ionizable groups in the peptide, it is surprising that gramicidin was captured at all. Thus, although its capture (Figure 4a) indicates that the technique has a highly promising general capacity to clean up hydrophobic proteins and peptides, it also raises interesting questions regarding the mechanisms involved. Hypothetically, the apparent behavior of gramicidin as an ionic compound under electrocapture conditions can be explained either by the formation of positively charged gramicidin-cation complexes that are trapped by the high electric

Figure 4. (a) Electrocapturegram of OG solubilized gramicidin dissolved in 100 mM ammonium formate, recorded at 218 nm. Inset shows the current and capture/release potential profiles of the cleanup process. The current increment seen in the inset which differs from Figures 2a and 5a is due to the higher concentration of buffer (100 mM) in which the polypeptide was dissolved in compared to the washing buffer (10 mM). The ESI-TOF mass spectrum of the collected fraction is shown in Figure 3b. (b) Electrocapturegram of the same solution (as in part a) dissolved in 10 mM ammonium formate. (c) Electrocapturegram of the same solution (as in part a) dissolved in 100 mM Tris-HCl. Peaks corresponding to noncaptured gramicidin are labeled with asterisks.

field gradient or the capture of the complexes as anions due to the presence of electroneutralizing formate anions around them. However, the protein solutions used prior to the ESI-MS did not contain significant levels of cations like Na+ or K+ capable of forming charged complexes. Therefore, ammonium ions probably

complexed with the gramicidin. In fact, the observed enhancement of capturing by increasing the concentration of ammonium formate supports the idea that complexation aids the charging of this hydrophobic peptide. By comparison of parts a and b of Figure 4, it can be seen that the peak at about 10 min, which corresponds to noncaptured gramicidin, decreased with increased concentration of ammonium formate buffer while the captured gramicidin dramatically increased. To further validate this hypothesis, we used Tris-HCl buffer noting that Tris is a relatively large cation which is not likely to penetrate through the gramicidin channel, thus precluding the formation of the essential complex. As expected, no capture occurred even under a higher capturing voltage of about 400 V (Figure 4c). Comparison of parts a and c of Figure 4 shows that the amount of noncaptured gramicidin appearing as a peak around 10 min increased significantly when Tris-HCl was used as a buffer. Additional support to this hypothesis is the fact that gramicidin can act as an ion channel for ammonium ions and that two ammonium ions may be present inside its channel simultaneously.11 This illustrates a way whereby neutral (hydrophobic) domains typical of membrane proteins can be made compatible with this method, provided suitable complexation occurs. It also indicates that the technique has potential utility in studies of phenomena such as interactions of peptides/proteins with metals and other small molecules. The crucial test for any separation method such as the one presented here is whether or not it is capable of separating multipass integral membrane proteins from detergents, since they have strong interactions with the membranes they are embedded in. Therefore, we turned our attention to bacteriorhodopsin (BR), which is present at high concentrations in the purple membrane of Halobacterium salinarium.12 It is also extremely stable4 and a member of a highly important protein family, the 7-transmembrane receptors.13 These features make it an ideal model for attempts to develop methods for preparing integral membrane proteins prior to MS analysis. Since BR is a very hydrophobic protein, the cleanup conditions were adjusted to include use of a washing buffer with an organic phase and longer washing times than in the previous experiments. Bacteriorhodopsin was efficiently captured (Figure 5a), although subsequent MS analysis indicated that it was not separated as cleanly as the previously tested polypeptides. However, significant cleanup was still achieved (Figure 5b,c). The spectrum obtained, though complex, provides rich information. Peaks generated from BR alone can be clearly distinguished, allowing direct identification solely by mass measurement, and others corresponding to BR associated with one, two, or three bound detergent molecules (Figure 5c). It should be noted that the lipid to protein molar ratio in the purple membrane is low (about 4:1),4 favoring the maintenance of essential interactions that prevent analytes from precipitating in solution The results indicate that the approach presented here has considerable potential utility for studies of transmembrane proteins such as BR and other 7-transmembrane receptors, which are related to the largest class of drug targets: G-protein coupled receptors (GPCRs). There was no known GPCR (12) Oesterhelt, D.; Stoeckenius, W. Methods Enzymol. 1974, 31, 667–678. (13) Jacoby, E.; Bouhelal, R.; Gerspacher, M.; Seuwen, K. ChemMedChem. 2006, 1, 761–782.


7119

structure before the rhodopsin structure was determined in 2000, and the second structure has only recently been published, after several years of work involving protein engineering.14 Clearly, complementary tools would greatly facilitate attempts to elucidate important aspects of the structure, functions, interactions, and evolution of membrane proteins such as these. Mass spectrometry can make significant contributions to such efforts. Electrocapture-MS affords a means to separate and analyze the proteins directly, without engineering to improve their analytical properties, and hence a means to probe them in more natural states. ESI-MS has been shown as a useful tool for interrogating drug-bound EmrE: an integral membrane protein, which functions as a drug resistance pump.15 However this required an MS instrument specifically designed for tandem MS at high m/z where the micelle-protein complexes tend to fall. Direct analysis by ESIMS of other membrane proteins like the K+ channel KcsA16 and Apolipoprotein CII17 reconstituted in phospholipids vesicles has also been performed, but without lipid removal, the detection of proteins is not as straightforward as the current protocol allows. Electrocapture is compatible to any MS platform providing versatility to the technique, which with further refinement certainly has the potential to allow analysis even of membrane protein complexes. CONCLUSION Electrocapture provides a useful and efficient way for separating membrane proteins from detergents sufficiently and thoroughly to obtain informative mass spectra of the proteins, especially in cases where limited amounts of analytes are available since there are no losses due to unspecific surface interactions. The technique has wide potential utility for preparing samples of membrane proteins, which remain very challenging systems for structural analysis. More importantly, as a front-end module it can be coupled to any other mass spectrometric configuration (e.g., systems with different types of analyzers), allowing the full range of current and future mass spectrometric techniques to be exploited. Its successful application to gramicidin and BR indicates that it offers exciting potential for future studies of host-guest complexes, more challenging membrane proteins (e.g., GPCRs), and noncovalent macromolecular assemblies. ACKNOWLEDGMENT This work was supported by the Stockholm University Proteomics Facility, the Swedish Research Council, and the European Commission. We thank Dr. Gunnar Thorsen for helpful discussions and for critically reading the manuscript. Received for review April 30, 2008. Accepted July 4, 2008. AC800877K Figure 5. Separation of detergent from a membrane-associated protein. (a) Electrocapturegram of OG solubilized bacteriorhodopsin in ammonium formate. Inset shows capture/release potential profiles of the process. (b) ESI-TOF mass spectrum of crude sample. OG clusters are shown under brackets. (c) ESI-TOF mass spectrum of the sample after cleaning up by electrocapture. The calculated mass is 26 899 Da while the measured mass is 26 902 Da. The model structure of the bacteriorhodopsin was produced based on the protein data bank coordinates 1brx using Rasmol.18 7120


(14) Cherezov, V.; Rosenbaum, D. M.; Hanson, M. A.; Rasmussen, S. G.; Thian, F. S.; Kobilka, T. S.; Choi, H. J.; Kuhn, P.; Weis, W. I.; Kobilka, B. K.; Stevens, R. C. Science 2007, 318, 1258–1265. (15) Ilag, L. L.; Ubarretxena-Belandia, I.; Tate, C. G.; Robinson, C. V. J. Am. Chem. Soc. 2004, 126, 14362–14363. (16) Demmers, J. A. A.; Van Dalen, A.; De Kruijff, B.; Heck, A. J. R.; Killian, J. A. FEBS Lett. 2003, 541, 28–32. (17) Hanson, C. L.; Ilag, L. L.; Malo, J.; Hatters, D.; Howlett, G.; Robinson, C. V. Biophys. J. 2003, 85, 3802–3812. (18) Sayle, R. A.; Milner-White, E. J. Trends Biochem. Sci. 1995, 20, 374–376.

Microfluidic Electrocapture-Assisted Mass Spectrometry of Membrane

Recommend Documents