Peptide Electroextraction for Direct Coupling of In-Gel Digests with

An electrophoretic method has been developed for the extraction of peptides following in-gel digests of SDSr. PAGE separated proteins. During electroe...
0 downloads 0 Views 234KB Size
Anal. Chem. 2000, 72, 4115-4121

Peptide Electroextraction for Direct Coupling of In-Gel Digests with Capillary LC-MS/MS for Protein Identification and Sequencing Aaron T. Timperman*,† and Ruedi Aebersold‡

Department of Chemistry, West Virginia University, Morgantown, West Virginia 26506-6045, and Department of Molecular Biotechnology, University of Washington, Seattle, Washington 98195-7730

An electrophoretic method has been developed for the extraction of peptides following in-gel digests of SDSPAGE separated proteins. During electroextraction, the peptides are trapped on a strong cation-exchange microcartridge, before analysis by capillary LC-ESI-tandem mass spectrometry. The spectra obtained by tandem mass spectrometry are searched directly against a protein database for identification of the protein from which the peptide originated. By minimizing surface exposure of the peptides during electroextraction, a reduction of the detection limits for protein identification is realized. The performance of the peptide electroextraction was compared directly with the standard extraction method for ingel protein digests, using a standard dilution series of phosphorylase B and carbonic anhydrase, separated by SDS-PAGE. The lowest gel loading in which phosphorylase B was identified using the standard extraction method was 2.5 ng or 25 fmol, and the lowest gel loading in which phosphorylase B was identified using electroextraction was 1.25 ng or 12.5 fmol. The design of the microextraction cartridge allows for direct interfacing with capillary LC, which is crucial for maintaining low detection limits. Furthermore, this method can be used for high-throughput proteomics since it can be easily multiplexed and requires only voltage control and low pressures (∼15 psi) for operation. We believe that peptide electroextraction is a significant advance for identification of proteins separated by one-dimensional or two-dimensional gel electrophoresis, as it can be easily automated and requires less protein than conventional methods. Two-dimensional gel electrophoresis (2DE) has become the most commonly used method for separation of complex protein samples from biological sources prior to identification by mass spectrometry (MS). The popularity of 2DE is primarily due to its high peak capacity and its compatibility with detergents and reducing agents that increase protein solubility. Also, the ability of the gel to trap separated proteins within its matrix has made it † West Virginia University, 217 Clark Hall, Prospect St., Morgantown, WV 26506-6045; (phone) 304-293-3435, x4455; (fax) 304-293-4904; (e-mail) [email protected]. ‡ Univeristy of Washington, 1705 NE Pacific St., Seattle, WA 98195-7730; (phone) 206-685-4235; (e-mail) [email protected].

10.1021/ac000305w CCC: $19.00 Published on Web 07/29/2000

© 2000 American Chemical Society

feasible to excise and manipulate these proteins for subsequent analysis without the need for complex instrumentation. Although this property of the gel has made many simple extraction methods possible, these methods are not ideal. The difficulties are particularly acute for proteomics that requires large-scale identification of proteins from gels. The most interesting proteins that are involved in dynamic processes, such as cellular signaling, are also the most elusive since they are usually present only at low relative abundance or trace amounts. Identification of lowabundance proteins from gels is further complicated by a rapid and nonlinear decrease in signal for faintly silver-stained bands (less than 5 pmol).1 Furthermore, detergents such as SDS help greatly with protein solubility in the gel but must be removed before analysis by MS. Improvements in mass spectrometry have made it a powerful tool for protein identification and sequencing.2-6 Electrospray ionization (ESI) has provided an efficient method for introducing ions into the mass spectrometer from the solution phase, allowing for the interfacing of liquid chromatography (LC) and MS. Capillary or µLC-tandem mass spectrometry (MS/MS) is now routinely used for protein identification of low-abundance samples. The process involves the chromatographic separation and concentration of peptides, the generation of collision-induced dissociation (CID) spectra of selected peptides in a tandem mass spectrometer, and the identification of the protein from which the peptide originated by sequence database searching. The capability of C18 reversed-phase HPLC columns to extract and concentrate peptides from proteolytic digests greatly improves the sensitivity of LC-ESI-MS. The ability to deduce peptide sequence from MS/ MS spectra yields a wealth of information at the amino acid level which is unmatched by MALDI-MS peptide mass fingerprinting experiments. Database searching routines,7,8 such as Sequest,9 can (1) Staudenmann, W.; Hatt, P. D.; Hoving, S.; Lehmann, A.; Kertesz, M.; James, P. Electrophoresis 1998, 19, 901-908. (2) Costello, C. E. Curr. Opin. Biotechnol. 1999, 10, 22-28. (3) Haynes, P. A.; Gygi, S. P.; Figeys, D.; Aebersold, R. Electrophoresis 1998, 19, 1862-1871. (4) Settlage, R. E.; Russo, P. S.; Shabanowitz, J.; Hunt, D. F. J. Microcolumn Sep. 1998, 10, 281-285. (5) Yates, J. R. J. Mass Spectrom. 1998, 33, 1-19. (6) Winston, R. L.; Fitzgerald, M. C. Mass Spectrom. Rev. 1997, 16, 165-179. (7) Yates, J. R. Electrophoresis 1998, 19, 893-900. (8) Fenyo, D.; Qin, J.; Chait, B. T. Electrophoresis 1998, 19, 998-1005. (9) Eng, J.; McCormack, A. L.; Yates, J. R. I. J. Am. Soc. Mass Spectrom. 1994, 5, 976-989.

Analytical Chemistry, Vol. 72, No. 17, September 1, 2000 4115

rapidly identify proteins by matching the MS/MS spectra with predicted spectra of entire protein databases. These factors have made µLC-MS/MS the method of choice for protein identification, and therefore, we have focused on developing an extraction method that is compatible with µLC-MS/MS. However, if MALDI is preferred due to its greater speed, this method can be easily interfaced with MALDI by eluting from the microextraction cartridge directly onto the MALDI target.10 Many methods have been developed to improve the extraction of gel-separated proteins for sequence analysis. In an early paper, DeWald and Pearson developed an automated method for pressurized extraction of gel-separated whole proteins with collection on reversed-phase liquid chromatography columns.11 Both electrophoretic methods12 and anion-exchange cartridges13 have been used for removal of SDS. Electroelution of whole proteins is a well-established technique, but it often suffers from low recovery and is not frequently used for small amounts of protein. FernandezPatron was able to reach the 50-pmol level by electroeluting proteins onto a C18 cartridge.14 Recently, a miniaturized electroelution method for one-step preparation of proteins prior to mass measurement by MALDI-MS was developed by Naylor et al.15 The proteins were eluted through a capillary and collected on a membrane. Mass measurements were made of cytochrome c for gel loadings as low as 900 fmol. However, we avoided protein electroelution, because proteins are generally not very soluble after they have been denatured in the gel. Detergents can be used to solubilize the proteins and remove them from the gel, but proteins often precipitate on the chromatography column after removal of the detergent, which is required prior to introduction into the MS. The most commonly used method for extraction of peptides from in-gel protein digests relies on diffusive or convective mass transport along the concentration gradient between the gel and extraction solution.11,16-18 Because a concentration gradient is required, relatively large volumes of extraction solution are needed for good recovery. The volume of solution used to extract the peptides is usually ∼200 µL, which requires drying down to reduce the sample volume before loading on the capillary LC column. Throughout the process the sample is exposed to many surfaces, such as tubes and pipet tips. Although these surfaces are thought to contain a minimal number of active sites where losses could occur, the multiple sample-handling steps expose the sample to large surface areas that could cause significant losses for lowabundance spots. If sample loss from surface exposure is based on the number of active sites available, then losses for relatively concentrated samples would be negligible, but loss for lowconcentration samples would be a significant percentage of the total available for analysis. To reduce sample handling and (10) Timperman, A. T.; Aebersold, R. 47th ASMS Conference on Mass Spectrometry and Allied Topics, Dallas, TX, June 13-17 1999 Abstr. 1642. (11) DeWald, D. B.; Pearson, J. D. Anal. Biochem. 1989, 180, 340-348. (12) Schuhmacher, M.; Glocker, M. O.; Wunderlin, M.; Przybylski, M. Electrophoresis 1996, 17, 848-854. (13) Vissers, J. P. C.; Chervet, J.-P.; Salzmann, J.-P. J. Mass Spectrom. 1996, 31, 1021-1027. (14) Fernandez-Patron, C. Electrophoresis 1995, 16, 911-920. (15) Clarke, N. J.; Li, F.; Tomlinson, A. J.; Naylor, S. J Am. Soc. Mass Spectrom. 1998, 9, 88-91. (16) Eckerskorn, C.; Grimm, R. Electrophoresis 1996, 17, 899-906. (17) Houthaeve, T.; Gausepohl, H.; Mann, M.; Ashman, K. FEBS Lett. 1995, 376, 91-94. (18) Shevchenko, A.; Wilm, M.; Vorm, O.; Mann, M. Anal. Chem. 1996, 68, 850-858.

4116

Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

automate the extraction, pressurized elution of in-gel protein digests and collection on larger, ∼2 × 18 mm, C18 columns has been used.11,16 This approach has been used to obtain sequence information from low-picomole amounts of protein using off-line nanospray MS/MS. A Cartesian robot has also been developed for automated processing of gel spots and pressurized elution of peptides following in-gel digestion.17 Generally, robotic automation of this standard peptide extraction method is quite complicated, and the speed of sample processing is limited by a slow dry-down step. Our objective was to develop an extraction method that would improve the sensitivity and throughput of protein identification from 2D gels while simplifying the process. We designed a method that interfaced directly with µLC capillary columns to take advantage of their concentration and separation abilities for peptides. To improve the sensitivity, we minimized surface exposure and elution volume. The peptides from an in-gel digest are forced out of the gel using electroextraction and trapped in a small cartridge containing strong cation-exchange beads. We have found no previous reports of peptide electroextraction in the literature, but have used this approach instead of protein electroelution, because peptides have higher solubilities and mobilities in the gel. To improve throughput for protein identification from 2D gels, the system was designed so that a large number of samples can be processed in parallel and could be readily automated. Only the switching of electrical voltages and the application of a low head pressure are required to control the electroextraction. After the peptides are trapped on the small cation-exchange cartridge, this column is connected to the HPLC column. Additionally, the extracted peptides have been stored on the microcartridge for 2 days without deleterious effects, thereby increasing the flexibility of the method. This capability allows many samples to be extracted simultaneously and stored until the peptides can be analyzed by the capillary LC-MS/MS. EXPERIMENTAL SECTION Reagents. The carbonic anhydrase II (bovine, C2522) and phosphorylase B (rabbit, P6635) were electrophoresis standards from Sigma. The Kasil No. 1 and BioChemika grade formamide used to make the frits were from the PQ Corp. (Valley Forge, PA) and Fluka, respectively. The buffers were made with 18-MΩ deionized water from a Milli-Q water purification system. HPLC grade acetonitrile, high-purity acetic acid, high-purity heptafluorobutyric acid (HFBA), and 99% trifluoroacetic acid (TFA) were from Aldrich. The 50 mM ammonium bicarbonate for the in-gel digestions was from Fluka. Lyophilized sequencing grade trypsin was from Promega (Madison, WI). Standard Gels. Dilution series of carbonic anhydrase II and phosphorylase B were prepared to evaluate the performance of the device as a function of the amount loaded on the gel. A 2-fold dilution series, from 320 to 0.6 ng, was loaded on the gel in 10 lanes for each of two identical gels. The SDS polyacrylamide gels were 12% T, 2.6% C, 1 mm thick, and 7 cm in length while the sample wells were 5 mm wide. While in the stacking gel, the proteins were electrophoresed at 100 V. When the proteins reached the running gel, the potential was increased to 160 V. After electrophoresis, the gels were fixed in 50% methanol and 10% acetic acid for 30 min and silver stained according to the protocol of Shevchenko et al.18 Next, the spots were excised from

Figure 1. Schematic of electroextraction device. After in-gel protein digestion, the gel pieces are placed in the 1500-µm-i.d. channel, and the peptides are electrophoresed into the 200-µm-i.d. electroextraction cartridge where they are trapped on strong cation-exchange beads. A small pressure suppresses bubble formation and fills the chamber with buffer. The potential for electrophoresis is supplied by a Pt electrode at +300 V, and the connection to ground is supplied by the buffer vial with grounding electrode at the electroextraction cartridge outlet.

the pair of gels, yielding two identical sets of dilution series for both carbonic anhydrase and phosphorylase B. To help avoid a bias caused by unknown differences in the gels, every other spot within a series was taken from one of the identical gels while the rest of the spots in the series were taken from the other gel. The in-gel digests were preformed according to the methods of Shevchenko et al.18 for both series of gel spots. The standard concentration of 12.5 ng/µL trypsin was used for the digests. The reduction and alkylation steps with dithiotreitol and iodoacetamide were omitted. For the series of gel spots prepared for electroextraction, only 2 µL of ammonium bicarbonate was used to incubate the gel pieces, so the peptides would not diffuse into the incubation solution. For the series of gel spots prepared for standard extraction, each of the spots was covered with 20 µL of solution during the incubation. Fabrication of Extraction Cartridges. The extraction cartridge was made from an 8-10-cm-long section of 200-µm-i.d. fused-silica capillary tubing (Polymicro Technologies). A 2-mmlong frit was formed at one end of the capillary from a porous ceramic material as previously described.19 Briefly, the capillary was rinsed with methanol and dried overnight. The frit was formed from a mixture of 170 µL of Kasil No. 1 solution and 30 µL of formamide. This solution was vortexed for 1 min immediately after mixing and centrifuged at 14 000 rpm for 1 min. Next, the solution was drawn into the end of the capillary by contacting with the solution surface. The frit was then dried overnight at 60° C, rinsed sequentially with 1 M HCl, 1 M ammonium nitrate, Milli-Q water, and ACN for 5 min each. The capillary was then packed with 5-µm Vydac strong cation-exchange material suspended in a 2-propanol slurry using a pressure vessel at 500 psi. The length of the packed cation-exchange bed was 6 mm, and an inlet frit was not made. Care was taken to always apply pressure from the inlet to prevent loss of packing material. Electroextraction Device. A schematic of the electroextraction device is shown in Figure 1. The body of the device was (19) Cortes, H. J.; Pfeiffer, C. D.; Richter, B. E.; Stevens, T. S. HRC&CC, J. High Resolut. Chromatogr. Chromatogr. Commun. 1987, 10, 446-448.

machined from acrylic, and an externally threaded 1500-µm-i.d. union was inserted into the bottom. An internally threaded union was used to attach the extraction cartridge to the bottom of the externally threaded union. An EC-105 power supply for minigels was used to apply +300 V for electroextraction. The voltage was applied to the inlet solution above the gel spots using a Pt electrode sealed in the side of the device. The outlet end of the capillary was placed in a 5-mL vial of buffer, which contained the grounding electrode. A 1-kΩ resistor was placed between the outlet and ground; the voltage drop across this resistor was used to monitor the current. The circuit was completed by also tying the negative outlet of the power supply to ground. Capillary LC Columns and Electrospray Interface. The capillary columns were slurry packed in a manner similar to previously reported methods.20 However, they were packed in pure 2-propanol without any detergent, and the slurry was introduced into the capillary with the high-pressure vessel while the solution was mechanically stirred. The capillaries were packed with 5-µm Magic beads (Michrom BioResources, Inc.) at 1500 psi into 75µm-i.d. columns with bed lengths of 9-10 cm. The electrospray interface was also fabricated in a manner similar to a previously reported method;21 however, the columns used fused-silica frits to retain the packing material and the electrospray tip (New Objectives) was connected to the end of the column using a zero dead volume union. The electrospray voltage was applied at the splitter at the inlet of the capillary column. The flow measured at the electrospray needle was ∼200 nL/min. Electroextraction. For the electroextraction, a 0.05% trifluoroacetic acid/10% acetonitrile solution was used. The extraction cartridge was flushed with this elution buffer using a syringe prior to connecting it with the body of the electroextraction device. The electroextraction device was filled with buffer and the extraction cartridge was connected to the bottom of the device, being careful not to introduce air bubbles. Before loading the gel pieces, the voltage was applied to check for current stability. A stable current between 10 and 20 µA indicated that no air bubbles were in the system. Such bubbles would cause current breakdown and stop the extraction prematurely. Additionally, during the electroextraction, a small head pressure of 15 psi was applied above the inlet solution to inhibit the formation of bubbles. After the in-gel digestion, the gel pieces were directly loaded into the top of the electroextraction device by dumping them into the funnel and pushing them into the channel of the 1500-µm-i.d. union until they were ∼3 mm from the end of the extraction cartridge. Next, the voltage was applied and the peptides were electroextracted at 300 V for 4 h and trapped on the cationexchange material. Upon completion, the extraction cartridge was removed from the electroextraction device and placed in-line with the capillary C18 LC column using an UpChurch Microtight union as shown in Figure 2. Using a high- pressure vessel, 2 µL of cationexchange eluent (1 M ammonium formate, 0.4% acetic acid, 0.005% HFBA, and 5% ACN) was loaded onto the extraction cartridge, to elute the peptides from the cartridge and concentrate them on the LC column. The extraction cartridge was removed from the pressure vessel and connected with the LC pumps while remaining (20) Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1989, 61, 1128-1135. (21) Gatlin, C. L.; Kleemann, G. R.; Hays, L. G.; Link, A. J.; Yates, J. R. Anal. Biochem. 1998, 263, 93-101.

Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

4117

Table 1. Comparison of the Sensitivity of the Electroextraction and Standard Extraction Methods for Equivalent Dilution Series of Phosphorylase B Load on the Gela phosphorylase B mass loaded on gel (ng)

Figure 2. Procedure following electroextraction. After electroextraction, the cartridge is placed in-line with the capillary (75-µm-i.d. ) C18 column. The peptides are eluted directly onto the C18 column by pumping the cation-exchange eluent onto the microextraction cartridge using the high-pressure vessel. Next, the high-pressure vessel is removed and the electroextraction cartridge is attached directly to the HPLC pumps. After rinsing the ammonium formate cationexchange eluent through the microextraction cartridge and C18 column with HPLC A solvent, the peptides are identified by LC-MS/MS with gradient elution.

in-line with the LC column. The transfer of peptides to the LC column was then completed by flushing with 4 µL of A buffer (∼6 min). Standard Extraction. After in-gel digestion, the incubation solution was removed and the peptides were extracted with one 50-µL aliquot of 50 mM ammonium bicarbonate and three 50-µL aliquots of 5% formic acid in 50% acetonitrile for 20 min each.18 The incubation solution and all the fractions were pooled and dried down. Samples were resuspended in 10 µL of the LC A buffer (5% ACN, 0.1% acetic acid, 0.005% HFBA), before injection onto the capillary LC-MS/MS using the high-pressure vessel. LC-MS/MS and Database Matching. Proteins were identified by conventional LC-MS/MS methods. A 5-60% gradient of acetonitrile, with 1% acetic acid and 0.005% HFBA, in 16 min was used for gradient elution of the peptides. A LCQ ion trap mass spectrometer (Finnigan) was used for detection and acquisition of the CID spectra. The CID spectra were obtained in the MS/ MS mode, which was triggered by when the intensity of an ion exceeded a preset threshold. Then Sequest was used to search the database directly with the CID. Only tryptic peptides having a minimum cross correlation score of 2 were considered as a reliable match. RESULTS AND DISCUSSION Method Overview. The electroextraction method presented here links slab gel protein separations and protein identification by µLC-MS/MS. The proteins are separated by one- or twodimensional gel electrophoresis, stained, and excised by standard methods. Following in-gel digestion of the proteins using a slightly modified standard method, the resultant peptides are electrophoresed out of the gel and trapped on a microcartridge using the device shown in Figure 1. After the 4-h electrophoretic extraction, the microcartridge is removed from the electroelution device and is placed in-line with a capillary C18 column as shown in Figure 2. The peptides are transferred from the strong cationexchange material of the microextraction cartridge to the capillary C18 column by injecting 2 µL of elution buffer onto the microcartridge with a pressure vessel. The inlet side of the microextraction cartridge is removed from the pressure vessel and connected to the HPLC pumps. The transfer is completed and the excess salt 4118 Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

10 5 2.5 1.25

no. of peptides matched electroextraction standard extraction 9 6 4 2

11 5 2 0

a All conditions and parameters for the entire analysis, such as the gel separation and capillary LC-MS/MS, were identical; only the extraction method was changed. For a peptide identification to be considered real it must be the best match for the CID spectra, have a Sequest cross-correlation score greater than 2, and must be tryptic. The results suggest that the electroextraction is more sensitive below 10 ng for phosphorylase B. Also the decrease in the number of peptides identified with decreasing concentration is much less rapid for the electroextraction method.

removed by rinsing the microextraction cartridge with 4 µL of HPLC buffer. After rinsing, a typical gradient capillary HPLC separation is performed with on-line peptide analysis by MS/MS. Sequest, a sequence database searching program, correlates the MS/MS spectra with known sequences and determines which protein(s) in the database contain(s) this peptide(s) sequence. Performance. The performance of the electroextraction device was evaluated by comparing it directly with the standard extraction method at different amounts of protein loaded on the gel. A dilution series of carbonic anhydrase II and phosphorylase B standards was applied to a pair of identical polyacrylamide gels to yield two identical sets of separated spots of carbonic anhydrase II and phosphorylase B. After excision and in-gel digestion the resultant peptides from one series of protein spots were extracted by the electroextraction method and the other series was extracted by the standard method, which served as a control. The recovered peptides were then analyzed by µLC-MS/MS and the Sequest database searching program in a similar manner. The extraction method, which yielded the largest number of identified peptides at the lowest protein loading on the gel, was determined to be the most sensitive. The electroextraction method compared favorably with the standard extraction method and gave modest improvements in sensitivity. As shown in Table 1, two peptides were identified with the electroextraction method at the lowest loading of 1.25 ng of phosphorylase B (the CID spectra are shown in Figure 3), while no peptides were identified using the standard extraction method. Identification was not attempted below 1.25 ng, because spots below this level could not be visualized with silver staining. For gel loadings at or below 10 ng of phosphorylase B, more peptides were consistently identified using the electroextraction method. With the standard extraction method, the peptide signal decreased much more rapidly below the 10-ng level than with the electroextraction method. These results are consistent with the hypothesis that the rapid reduction in peptide signal observed with the standard extraction method is due to nonspecific absorption from surface exposure. Similar results were also obtained using carbonic anhydrase, which was compared at 5- and 2.5-ng loadings on the gel. At the 5-ng level, three peptides were identified with electroextraction and only one was identified with the standard

Figure 4. CID spectrum of the peptide detected from a 2.5-ng gel loading of carbonic anhydrase II using the electroelution method. This CID has an odd appearance as there are four pairs of isobaric ions (µm ) 0.01 amu) that cannot be distinguished with the mass spectrometer used here, but a partial sequence can still be read manually. As shown in Table 2, an equivalent run using the same 2.5-ng gel loading and the standard extraction procedure resulted in no detectable peptides.

Figure 3. CID spectra of the two peptides detected from a 1.25-ng loading of phosphorylase B on the gel using the electroextraction method. Even at such a low gel loading, partial sequences of the two peptides, NFNRHLHFTLVKDR shown in (A) and LLSYVDDEAFIRDVAK shown in (B), can be manually read from the CID. As shown in Table 2, no peptides were detected for an equivalent 1.25-ng loading of phosphorylase B on the gel using the standard extraction procedure.

extraction method. At the 2.5-ng level, which was at the silver stain detection limit, one peptide was identified using the electroextraction method (the CID spectrum is shown in Figure 4) while none were identified with the standard extraction method. The sequences of the peptides identified at the lowest levels and their Sequest cross-correlation scores are shown in Table 2. Although improvements in throughput were not characterized directly, large improvements could be realized, as the electroextraction devices were easily multiplexed and could be easily automated. The speed of electroextraction of an individual gel spot is not fast at 4 h, but it is still faster than the standard extraction method, which takes at least 6 h and requires drying the sample down and much more manual manipulation. It was determined that the electric field was responsible for elution by running the same electroextraction procedure with the voltage off. When the voltage was off and only a light head pressure was applied, no peptides were detected. However, the real speed advantage comes from the ability to multiplex these devices and run many simultaneously. Multiplexing is very simple as numerous devices can be run in parallel off of the same power supply as the current consumption is very low at 20 µA. Four devices were run simultaneously, although many more could be run at once. The devices can be easily automated as they require only switching of the voltage and the low-pressure gas. Not only does the gas supply head pressure but it can also be used to fill the device with buffer and flush out air bubbles. A further advantage of the

microextraction cartridge in terms of throughput and ease of use is that once the peptides have been extracted they can be stored on the cartridge before LC-MS analysis. We routinely performed the electroelution 1 or 2 days before LC-MS analysis and stored the microextraction cartridges in the refrigerator without a measurable loss of peptides. Design. During electroextraction, a small force was applied to the peptides in the gel by the electric field and a high force was applied to the peptides in the extraction cartridge. Ideally, one would want the oppositesa high force through the gel to rapidly extract the peptides and a small force through the extraction cartridge to allow the peptides the most opportunity to stick. However, this ideal situation was not allowed by the geometry of the system. The force on the peptides is directly proportional to the magnitude of the electric field, and the magnitude of the electric field is inversely proportional to the cross-sectional area of the channel. Thus, the small force through the gel channel is a result of its large internal diameter, and the large force through the extraction cartridge is a result of its small internal diameter. Standard protein electroblotting procedures are often run in the current-limited mode, and the maximum is stated as current density. A typical current density for protein electroblotting is ∼4 mA/cm2, while the current density is 0.3 mA/cm2 through the gel pieces and 15 mA/cm2 through the extraction cartridge for technique described here. As a result of this configuration, the microelectroextraction requires strong binding of the peptides to the cation-exchange cartridge and is rather slow compared to protein electroelution methods. Clearly, peptides with higher cationic charges have stronger binding to the cation-exchange material, and consequently, a bias for multiply charged cations is observed with the electroextraction method. The peptides recovered and identified with the electroextraction method contained between two and five basic residues, which are cationically charged at this pH. These peptides are rich in histidine, arginine, and lysine as shown Table 2. Since trypsin cuts after lysine and arginine except when followed by proline, the presence of multiple arginine and lysine residues within a peptide is typically the result of missed cleavage(s). In Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

4119

Table 2. Sequences and Corresponding Sequest Cross-Correlation Scores for a Comparison of the Electroelution and Standard Extraction Methods near the Detection Limita

a The peptides identified with the electroextraction method are rather large and contain many basic residues from the presence of histidine and missed tryptic cleavages. This bias for cationic peptides is not surprising since the peptides are trapped on cation-exchange material during the electroextraction. The microelectroextraction method yeilds a ∼2-fold improvement in sensitivity based on the peptides identified by Sequest for both phosphorylase B (PHS2•RABIT) and carbonic anhydrase (CAH2•BOVIN).

the standard protocol, the gel pieces are usually covered with 50 mM ammonium bicarbonate (>20 µL), but for the electroextraction, only 2 µL of 50 mM ammonium bicarbonate was added prior to incubation of the gel pieces with trypsin. The missed cleavages may have been caused by this minor alteration of the in-gel digestion protocol, but this possibility was not investigated. Despite the missed cleavages, the digests were as reproducible as the digests for the standard protocol. The geometry of the system was determined by many practical considerations and limitations. First, a small extraction cartridge is required, because only a small volume of solution can be used to elute the peptides from the extraction cartridge and into the capillary HPLC column. Also, a small cartridge reduces the loss of peptides due to surface exposure. Smaller inner diameter capillaries (50 and 100 µm) were tested for the extraction cartridges, but with the concomitant decrease in the electric field through the gel, no elution of peptides was observed. Second, practical loading considerations made it necessary to hold the gel pieces in a larger tube (1500-µm i.d.) during electroextraction. With this design, it was easy to the load the gel pieces without introducing any air bubbles. Capillaries with the same 200-µm i.d. as the extraction cartridge were originally used to hold the gel pieces during electroextraction in an attempt to increase the magnitude the electric field across the gel. The gel pieces were loaded in the 200-µm-i.d. capillaries by putting them into a syringe and extruding them through the needle and into the capillary. Not only did this process greatly increase surface exposure, but it was tedious and nearly impossible to avoid the introduction of tiny air bubbles which would expand during electroextraction causing current breakdown. 4120

Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

Strong cation-exchange bead material proved to be the material of choice for the electroextraction cartridge for several reasons. First, to obtain the best detection limits, it was important to maintain the concentration and extraction capabilities of C18 capillary columns for peptides prior to MS analysis. Although the 200-µm i.d. of the cartridge is small for electroextraction, it is still much larger than the 50-75-µm capillary HPLC columns used for peptide identification. Electroextraction directly onto a 200µm-i.d. C18 cartridge or column resulted in poorer detection limits caused by sample dilution and diminished chromatographic resolution. Thus, we chose strong cation-exchange material for the electroextraction because peptides can be eluted from the cartridge and concentrated directly on the reversed-phase capillary column. The cation exchanger does not appear to bind the peptides as strongly as the reversed-phase material, but it allows for concentration on the capillary HPLC column, which is critical. Second, it is advantageous to force all of the peptides to move in one direction so they can all be captured on the same trap. When an electrophoretic method is used, such unidirectional movement requires the use of buffers at either very high or very low pH. Working at low pH causes fewer complications, the required pH is not as extreme, and low pH is compatible with positive ion ESIMS. Third, the electroosmotic flow (EOF) generated by the strong cation-exchange material is in the forward direction, which is toward the microextraction cartridge. If the EOF is in the forward direction, this bulk flow of solution will aid the capture of the peptides on the microextraction cartridge, but if it is in the reverse direction, it will make it more difficult for the peptides to enter the microextraction cartridge. At a pH less than ∼4 reversed-phase silica beads have a reverse EOF.

Careful consideration is also required for frit formation in the microextraction cartridge. With the relatively high amount of current and concomitant Joule heating, bubble formation in the frits is problematic. These difficulties with frit formation are well documented in the capillary electrochromatography literature and have been a practical impediment to widespread use of this technique. However, with use of the ∼15 psi head pressure and the Kasil frits, current breakdown caused by buffer outgassing was not a problem. Many methods for sintering frits were tried,22-26 but current breakdown was still problematic and their fabrication in the 200-µm-i.d. capillaries was tedious. Also, care was taken with the design of the electroextraction device to allow release of oxygen and hydrogen gases formed at the electrode surfaces before they would block the channel and cause current breakdown. The strength of peptide binding to the cation-exchange material is directly related to the solution pH, which must be optimized to obtain the best recovery. The pH must be low to impart the maximum positive charge on peptides, but not so low that it protonates the sulfonic acid groups of the cation exchanger. For the electroextraction, a solution of 0.05% TFA at pH ∼2.3 was found to give the best recovery, making it possible to recover peptides with a +2 charge and greater. A 100 mM concentration of acetic acid was used, but peptide recovery was ∼5-fold lower than those observed with 0.05% TFA. The electroextraction was rather slow as it took 4 h to achieve the maximum recovery. At 2 h, ∼40% of the maximum was recovered, and at 6 h, ∼80% of the maximum was recovered. The decrease in yield with longer times indicated that the bound peptides could be stripped off the cationexchange material. The speed of extraction could be increased by applying a higher voltage, but a practical limit was set by buffer outgassing, which causes current breakdown and stops elution. The binding strength of a peptide to the cation-exchange material is determined largely by the overall peptide charge. Modifications that decrease the peptide charge, such as phosphorylation, are recovered at a lower efficiency and exhibit higher detection limits while modified peptides with a more positive charge are recovered at higher efficiency and exhibit lower detection limits. CONCLUSIONS A device has been fabricated for electroextraction of peptides from in-gel digests of proteins separated by 1D or 2D gel electrophoresis. This method minimizes surface exposure and can (22) Bosch, S. E. v. d.; Heemstra, S.; Kraak, J.; Poppe, H. J. Chromatogr., A 1996, 755, 165-177. (23) Asiaie, R.; Huang, X.; Farnan, D.; Horvath, C. J. Chromatogr., A 1998, 806, 251-263. (24) Behenke, B.; Grom, E.; Bayer, E. ?????????????????????? 1995. (25) Boughtflower, R. J.; Underwood, T.; Paterson, C. J. Chromatographia 1995, 40, 329-335. (26) Zimina, T. M.; Smith, R. M.; Myers, P. J. Chromatogr., A 1997, 758, 191197.

be interfaced directly with capillary LC-MS/MS for protein identification. A ∼2-fold improvement in detection limits for the overall protein identification have been realized with this method. Importantly, the observed sensitivity gain increased with decreasing sample size. Furthermore, the method is easily automated as it requires only the switching of electric potentials, and low pressure for buffer loading and bubble suppression. Four extraction cartridges in parallel were used in this study, but many more could be utilized. Theoretically the number that can be run in parallel is limitless, although there is a practical limit imposed by the increasing complexity. Since identification by capillary LCMS/MS is still a serial process, the flexibility of the method is greatly increased by the ability to store the extracted peptides on the microcartridges. Thus, a large number of gels spots could be extracted simultaneously and held on the microcartridges until they can be processed by the MS/MS. The best results were obtained with a two-step process. The peptides were trapped first on cation-exchange material and then concentrated on a C18 column for separation prior to MS/MS analysis. Electroextraction directly onto 50-75-µm C18 columns was complicated by the geometry of the device and its effect on the electric field. As with many extractions, the underlying challenge is reducing the sample volume by ∼2 orders of magnitude with minimal sample loss. Future improvements and optimization could lead to further increases in performance. For instance, further improvements could come from the use of cation-exchange resins optimized with respect to peptide retention and electroosmotic pumping, optimization of the buffer system, and fabrication of cation-exchange cartridges with frits on both the outlet and inlet ends. Reduction in the volume of cation-exchange eluent needed could reduce transfer and rinsing times for the capillary LC columns. If a better separation of peptides was needed for MS/MS analysis, then a step elution from the cation-exchange material onto the C18 could be used for a two-dimensional separation. As the quest for greater sensitivity is constantly reducing the amount of sample needed for MS/MS identification, such a method linking gel separations and the MS/MS analysis with minimal surface exposure and sample loss should become even more important. ACKNOWLEDGMENT We thank the NSF Science and Technology Center for Molecular Biotechnology and the NIH Resource Center for Comprehensive Biology. We also thank Dr. Julian Watts for supplying 32P-labeled protein, which was used in initial experiments for rapid quantitation of peptide recovery. Received for review March 15, 2000. Accepted June 2, 2000. AC000305W

Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

4121