AC Research
Anal. Chem. 1998, 70, 3721-3727
Accelerated Articles
Nanoflow Solvent Gradient Delivery from a Microfabricated Device for Protein Identifications by Electrospray Ionization Mass Spectrometry Daniel Figeys* and Ruedi Aebersold
Department of Molecular Biotechnology, University of Washington, Box 357730, Seattle, Washington 98195-7730
Microfabrication technology offers the opportunity to construct microfluidic modules which are designed to perform specific, dedicated functions. Here we report the construction of a microfabricated device for the generation and delivery by electroosmotic pumping of solvent gradients at nanoliter per minute flow rates. The device consists of three solvent reservoirs and channels which were etched in glass. Solvent gradients and solvent flows were generated by computer controlled differential electroosmotic pumping of aqueous and organic phase, respectively, from the solvent reservoirs. The device was integrated into an analytical system consisting of the solvent gradient delivery module, a reverse phase microcolumn and an electrospray ionization ion trap mass spectrometer (MS). The system was used for the analysis at high sensitivity of peptides and peptide mixtures generated by proteolytic digestion of proteins. We have measured an absolute limit of detection as low as 1 fmol and a concentration limit of detection at the 100 amol/ µL level. The system was also successfully used for the identification of proteins separated by 1D and 2D gel electrophoresis. This was achieved by gradient frontal analysis of the peptide mixture generated by proteolysis of the respective proteins, and the automated generation and interpretation of collision-induced dissociation spectra. The completion of genomic sequencing projects (e.g., Haemophilus influenzae Rd,1 Saccharomyces cerevisiae,2 and others; see http://www.tigr.org/tdb/mdb/mdb.html) and the widespread availability of these sequences in databases offer new possibilities * Corresponding author: National Research Council Canada Institute for Marine Biosciences, 1411 Oxford St., Halifax, N.S., Canada B3H 3Z1 (phone) (902) 426-0558; (fax) (902) 426-9413; (e-mail)
[email protected]. S0003-2700(98)00502-2 CCC: $15.00 Published on Web 07/30/1998
© 1998 American Chemical Society
to study and understand biological systems by the comprehensive analysis of the genes and proteins which constitute them. In recent years, electrospray ionization (ESI) mass spectrometry (MS) has become a valuable tool for the analysis of complex polypeptide mixtures.3-7 This is in part due to its easy interfacing with high-performance separation techniques such as highperformance liquid chromatography (HPLC)8-10 and capillary electrophoresis (CE).11-13 Furthermore, the combination of ESI and tandem mass spectrometry (MS/MS) allows the generation of collision-induced dissociation spectra (CID) and therefore the deduction of the amino acid sequence of selected peptides. Using an automated HPLC-MS/MS system, we14 have shown the (1) Fleischmann, R.; Adams, M.; White, O.; Clayton, R.; Kirkness, E.; Kerlavage, A.; Bult, C.; Tomb, J.; Dougherty, B.; Merrick, J.; et al. Science 1995, 269, 496-512. (2) Bussey, H.; Storms, R. K.; Ahmed, A.; Albermann, K.; Allen, E.; Ansorge, W.; Araujo, R.; Aparicio, A.; Barrell, B.; Badcock, K.; et al. Nature 1997, 387, 103-105. (3) Affolter, M.; Watts, J. D.; Krebs, D. L.; Aebersold, R. Anal. Biochem. 1994, 15, 74-81. (4) Amankwa, L. N.; Harder, K.; Jirik, F.; Aebersold, R. Protein Sci. 1995, 4, 113-125. (5) Patterson, S. D.; Aebersold, R. Electrophoresis 1995, 16, 1791-1814. (6) Watts, J. D.; Affolter, M.; Krebs, D. L.; Wange, R. L; Samelson, L. E.; Aebersold, R. J. Biol. Chem. 1994, 269, 29520-29529. (7) Watts, J. D.; Affolter, M.; Krebs, D. L.; Wange, R. L.; Samelson, L. E.; Aebersold, R. In Biochemical and Biotechnological Application of Electrospray Ionization Mass Spectrometry; Snyder, A. P., Ed.; ACS Symposium Series 619; American Chemical Society: Washington, DC, 1996; pp 381-407. (8) Yates, J. R., III Cell Biology: A Laboratory Handbook; Academic Press: San Diego, 1998; Vol. 4, pp 529-538. (9) Davis, M. T.; Stahl, D. C.; Hefta, S. A.; Lee, T. D. Anal. Chem. 1995, 67, 4549-4556. (10) Qin, J.; Herring, C. J.; Zhang, X. L. Rapid Commun. Mass Spectrom. 1998, 12, 209-216. (11) Figeys, D.; Ducret, A.; Aebersold, R. J. Chromatogr., A 1997, 763, 295306. (12) Bateman, K. P.; White, R. L.; Thibault, P. Rapid Commun. Mass Spectrom. 1997, 11, 307-315. (13) Smith, R. D.; Udseth, H. R. In Pharmaceutical And Biomedical Applications Of Capillary Electrophoresis; Lunte, S. M., Radzik, D. M., Eds.; Pergamon Press: Tarrytown, NY, 1996; Vol. 2, pp 229-275.
Analytical Chemistry, Vol. 70, No. 18, September 15, 1998 3721
unattended identification of tens of proteins per day. However, such analyses require picomole amounts of proteins. The sensitivity of ESI-MS analysis of peptides has been dramatically enhanced by the introduction of microelectrospray (µESI) and nanoelectrospray (nESI) ion sources,15-17 the coupling on-line of microseparation techniques with MS,15,18-20 and the development of MS instruments of increasing sensitivity. Continuous-flow nanoelectrospray,15,16 solid-phase extraction (SPE)-capillary zone electrophoresis (CZE)-µESI-MS/MS,11,20-24 and micronanoflowHPLC-MS/MS18,25,26 represent high-sensitivity protein and peptide analysis systems which demonstrated sensitivities in the subpicomole range. However, automation of these high-sensitivity analytical systems has been difficult. Analytical systems that offer both a high level of automation and sensitivity have been constructed by the connection on-line of microfluidic modules generated by microfabrication technology and ESI-MS.19,27-30 We have introduced a three-position microfabricated device for sample delivery in a continuous mode to an µESI ion trap MS for protein identification at the low-femtomole sensitivity level.19 We have also described a fully automated nineposition microfabricated sample handling device which was connected on-line with a MS/MS instrument.29 With both systems, we achieved concentration limits of detection below the femtomole per microliter level. Both systems infused unseparated peptide mixtures into the MS/MS system, thus essentially representing an automated implementation of the nanospray approach to protein identification. In this paper, we demonstrate the use of a microfabricated microfluidic module for the generation and delivery of solvent gradients at nanoliter per minute flow rates. The solvent gradients were achieved by directed differential electroosmotic pumping of an aqueous and an organic phase, respectively, present in different reservoirs on the device. Differential electroosmotic pumping was achieved by varying the potential applied to the respective reservoirs. This solvent gradient module was applied, in conjunc(14) Ducret, A.; van Oostveen, I.; Eng, J. K.; Yates, J. R., III; Aebersold, R. Protein Sci. 1998, 7, 706-709. (15) Wilm, M.; Shevchenko, A.; Houthaeve, T.; Breit, S.; Schweigerer, L.; Fotsis, T.; Mann, M. Nature 1996, 379, 466-469. (16) Wilm, M.; Mann, M. Anal. Chem. 1996, 68, 1-8. (17) Smith, R. D.; Udseth, H. R.; Wahl, J. H.; Goodlett, D. R.; Hofstadler, S. A. In Methods in Enzymology; Karger, B. L., Hancock, W. S., Eds.; Academic Press: New York, 1996; Vol. 271, pp 448-486. (18) Chervet, J. P.; Ursem, M.; Salzmann, J. B. Anal. Chem. 1996, 68, 15071512. (19) Figeys, D.; Ning, Y.; Aebersold, R. Anal. Chem. 1997, 69, 3153-3160. (20) Figeys, D.; Ducret, A.; Yates, J. R., III; Aebersold, R. Nature Biotechnol. 1996, 14, 1579-1583. (21) Figeys, D.; Aebersold, R. Electrophoresis 1997, 18, 360-368. (22) Tomlinson, A. J.; Benson, L. M.; Jameson, S.; Johnson, D. H.; Naylor, S. J. Am. Soc. Mass Spectrom. 1997, 8, 15-24. (23) Tomlinson, A. J.; Guzman, A.; Naylor, S. J. Capillary Electrophor. 1995, 2, 247-266. (24) Settlage, R. E.; Russo, P. S.; Shabanowitz, J.; Hunt, D. F. J. Microcolumn Sep. 1998, 10, 281-285. (25) Yates, J. R.; McCormack, A. L.; Link, A. J.; Schieltz, D.; Eng, J.; Hays, L. Analyst 1996, 121, R65-R76. (26) Huczko, E. L.; Bodnar, W. M.; Benjamin, D.; Sakaguchi, K.; Zhu, N. Z.; Shabanowitz, J.; Henderson, R. A.; Appella, E.; Hunt, D. F.; Engelhard, V. H. J. Immunol. 1993, 151, 2572-2587. (27) Xue, Q. F.; Foret, F.; Dunayevskiy, Y. M.; Zavracky, P. M.; McGruer, N. E.; Karger, B. L. Anal. Chem. 1997, 69, 426-430. (28) Ramsey, R. S.; Ramsey, J. M. Anal. Chem. 1997, 69, 1174-1178. (29) Figeys, D.; Gygi, S. P.; McKinnon, G.; Aebersold, R. Anal. Chem., in press. (30) Figeys, D.; Aebersold, R. Electrophoresis 1998, 19, 885-892.
3722 Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
Figure 1. Diagram of the three-position microfluidics module coupled to an ion trap MS via a transfer capillary and a µESI interface. C18 material was incorporated in the µESI interface.
tion with a small C18 cartridge for the separation by frontal analysis of peptides present in complex peptide mixtures. The peptides were then identified by an ESI-ion trap MS connected on-line. We established that as little as a few femtomoles of sample can be analyzed this way. The possibility of coupling the solvent gradient system to electrochromatography columns to perform gradient electrochromatography on-line with µESI-MS/ MS is discussed. EXPERIMENTAL SECTION Chemicals, Materials, and Instrumentation. Sequencing grade modified trypsin (porcine) was from Promega (Madison, WI) and standard horse apomyoglobin from Sigma (St. Louis, MO). Acetic acid, acetonitrile, and methanol were obtained from J. T. Baker (Phillipsburg, NJ). Rapidly hardening epoxy glue was from Devcon Corp. (Danvers, MA) and temperature-cured epoxy (H77T) from Epoxy Technology (Billerica, MA). The fused-silica capillary tubing was from Polymicro Technologies (Phoenix, AZ). Stainless steel tubings were from Small Parts (Miamilakes, FL), and the fingertight fitting was from Upchurch Scientific (Oak Harbor, WA). The 5-µm-diameter 300-Å pore size C18 material was from Phase Separations (Norwalk, CT). Distilled water was deionized (18 MΩ) using a Milli-Q system from Millipore (Bedford, MA). CE high-voltage power supplies were purchased from Spellman (Plainview, NY). The LCQ ion trap MS was a product of Finnigan MAT (San Jose, CA). Construction of the Microfabricated Device-Frontal Analysis System. The microfabricated devices were made by the Alberta Microelectronic Centre (Edmonton, Alta, Canada) as previously described.19 In this present design, channels (30 µm depth × 72-73 µm width) and three reservoirs (1 mm × 1 mm) were etched on a piece of glass (540 µm thick) as indicated in Figure 1. The link between the device and the transfer capillary, which also acted as an electroosmotic pump, was made perpendicular to the plane of the device. This was achieved by inserting a 200 µm i.d. × 350 µm o.d. sheath capillary in the 350-µm hole drilled at the end of the main channel and gluing it in place using heat-curable epoxy. A 250-µm-i.d. and 1.5-mm-o.d. Teflon tube was inserted over the sheath capillary and glued in place. A small section of the Teflon tube had been previously expanded to fit the full length of 350-µm o.d. of the sheath capillary. A fingertight fitting was added to the other end of the Teflon tubing. A transfer capillary was inserted into the Teflon tube and into the sheath
capillary so that its end reached to the etched channel on the microfabricated device. It was then held in place by tightening the fingertight fitting. The 15-cm-long transfer fused-silica capillary (50 µm i.d. × 150 µm o.d.) served as a connection between the device and a liquid junction µESI interface. Bare fused silica or derivatized capillaries with (3-aminopropyl)silane31 were used. The liquid junction was also used as a solid-phase extraction cartridge. The transfer capillary was inserted with a small platinum wire midway through another 250-µm-i.d. Teflon tube and glued in place using 5-min epoxy. A small piece of membrane followed by a 1-3-mm-long bed of C18 resin and another piece of membrane were inserted from the other end of the Teflon tubing. Finally, a µESI needle, made from a 50-µm-i.d. and 150µm-o.d. fused-silica capillary, which was tapered in a flame, was inserted and glued in place. This liquid junction interface was positioned with the µESI needle facing the entrance of the MS. The potential for the ESI process was supplied by the power supply of the ion trap MS and applied through the small platinum wire of the µESI interface. A constant voltage of +1.3 to +1.7 kV was applied for the duration of an experiment. Sample Loading and Generation of Gradients. The samples were pressure loaded on the SPE device in the liquid junction interface. This was done by disconnecting the transfer line from the device and inserting it into a pressurizable microvial as previously described.11,20,21 The samples were generally loaded at 6-9 psi for up to 10 min. This allowed the loading of 10-20 µL of sample. Once the samples were applied, the column was rinsed and equilibrated with 10 mM acetic acid/10% methanol and the transfer capillary was reconnected to the device. The gradients were generated by controlled electroosmotic pumping from two reservoirs filled with different solvents. Reservoir one was filled with 10 mM acetic acid/10% methanol (v/v) (solvent A). Reservoir two, although not used for the experiment, was filled with the same solvent. Reservoir three was filled with 65% acetonitrile (v/v)/10 mM acetic acid (solvent B). Platinum electrodes connected to individual power supplies were inserted into reservoirs one and three, respectively. The gradient was generated by ramping potentials to reservoirs one and three. Care must be taken when dealing with high voltages. In a typical experiment, reservoir one was held to -5.2 kV while the potential of reservoir three was ramped from -4.6 to -5.2 kV over 14 min. Then, the potential of reservoir one was ramped from -5.2 to -4.5 or -4.2 kV over 14 min, while reservoir three was maintained at -5.2 kV. Finally, both reservoirs were kept at these final voltages for the rest of the experiment. The columns were reequilibrated by switching reservoir one back to -5.2 kV and reservoir three to -4.5 kV. Mass Spectrometry. The ion trap MS was used essentially as described21 with the following modifications. The trap was run with automatic gain control for all experiments. In this mode, the system automatically selects the gating parameters to keep the number of ions present in the trap to a constant preset value. In MS mode, the target number of ions was set to 1 × 108, and in CID mode, to 2 × 107. The electron multiplier was set to -1060 V. Three sequential scan ranges were used in the MS mode. They ranged from 400 to 725, from 725 to 1000, and from 1000 to 1850 (31) Figeys, D.; Ducret, A.; Oostveen, I. v.; Aebersold, R. Anal. Chem. 1996, 68, 1822-1828.
amu, respectively. The most intense ion from each scan range was selected with a 3-amu window, and CID experimentss were performed with the energy set at 55%, the maximum trap time set at 1500 ms, and the number of microscans set to 7. All the spectra were recorded in centroid mode. Sequence Database Search Strategy. The MS/MS spectra were searched against protein databases using the Sequest software32,33 to identify the source of the peptides. All the parameters have been described.31 RESULTS AND DISCUSSION Since the sensitivity achieved by ESI-MS is essentially dependent on the analyte concentration, the method of introducing the sample to the MS system is critical. For high-sensitivity peptide analysis, two principally different sample application methods are commonly used. The first is the continuous application of unseparated peptide mixture (nanospray technique).15,16,34 The second is the sequential introduction of spatially separated peptides. This is typically achieved by the coupling on-line of HPLC18,25,26 or CE11,13,20-24,35-38 separation systems with the MS. The nanospray technique has the advantage of extensive signal averaging over a long time for the analysis of selected sample constituents, thus improving the quality of the spectra obtained. Because the MS only generates CID spectra of one analyte at a time, the other analytes present in the mixture go undetected during the analysis of the selected analyte and are therefore wasted. Other disadvantages of the nanospray technique are that no concentration of the analyte is achieved and that matrix effects affect the analysis of the analytes. The second method consists of coupling spatial separation techniques such as HPLC or CE to the MS. Concentration of analytes by up to a few orders of magnitude has been achieved. The separation technique presents analytes sequentially to the MS and essentially eliminates global matrix effects. However, the time to perform the CID experiment is limited to the peak width of the analytes; the experiments are more technically difficult to perform and are only realizable if the MS/MS system used supports automated data-directed CID experiments. In this report, we introduce temporal separation of analytes (without spatial separation) as a third principle for introducing peptides to an µESI or nESI MS system. This was achieved by the generation of solvent gradients on a microfabricated device and the sequential mobilization of peptides immobilized on a SPE cartridge by frontal analysis. The mobilized peptides were then identified by MS/MS in an ion trap MS. The advantages of this technique are the separation and concentration of analytes, elimination of matrix effects, and controllable time of analysis. Generation of Gradients. Microfabricated fluidic devices are well suited for the generation of solvent gradients as multiple (32) Eng, J.; McCormack, A. L.; Yates, J. R., III J. Am. Soc. Mass Spectrom. 1994, 5, 976-989. (33) Yates, J. R., III; Eng, J. K.; McCormack, A. L.; Schieltz, D. Anal. Chem. 1995, 67, 1426-1436. (34) Shevchenko, A.; Wilm, M.; Vorm, O.; Mann, M. Anal. Chem. 1996, 68, 850-858. (35) Valaskovic, G. A.; McLafferty, F. W. J. Am. Soc. Mass Spectrom. 1996, 7, 1270-1272. (36) Tomlinson, A. J.; Benson, L. M.; Jameson, S.; Naylor, S. Electrophoresis 1996, 17, 1801-1807. (37) Naylor, S.; Tomlinson, A. J. Biomed. Chrom. 1996, 10, 325-330. (38) Thibault, P.; Paris, C.; Pleasance, S. Rapid Commun. Mass Spectrom. 1991, 5, 484-490.
Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
3723
Figure 2. Three steps for the generation of an aqueous/organic solvent gradient. (A) Flow from reservoir one toward the MS and back in reservoir three; (B) flow from reservoirs one and three, respectively, toward the MS; (C) flow from reservoir three toward the MS and back to reservoir one. The aqueous phase was 10 mM acetic acid/10% (v/v) methanol. The organic phase was 65% (v/v) acetonitrile/3 mM acetic acid.
reservoirs and flow paths can be present on a single device and can be individually addressed. Electroosmotically generated flows from specific reservoirs can be joined in a controlled fashion to generate solvent gradients. The magnitude and the direction of the electroosmotic flows are controlled by the potentials applied to the reservoirs. Gradients were generated on the system described in Figure 1 by ramping potentials to reservoirs one and three. For these experiments, a bare fused-silica capillary was used as transfer capillary. Reservoir one was filled with 10 mM acetic acid/10% (v/v) methanol and reservoir three was filled with 65% (v/v) acetonitrile/3 mM acetic acid. The difference in potential between the reservoirs and the microelectrosprayer-generated electroosmotic flows toward the microelectrosprayer, and therefore, solvent gradients were generated without the need for a pump. Two highvoltage power supplies were used to furnish the appropriate potential to reservoirs one and three. Each power supply was controlled through a digital to analog converter board (DAC) and a Labview procedure. This procedure controlled the potential (010 V) applied to two analog outputs of the DAC which in turn controlled the high-voltage power supply. Thes procedure ramped the potential applied to the DAC output according to preset initial and final voltages and ramping times. The generated gradients were followed by monitoring in the MS a signal representing the clustering of acetonitrile. The signal increased as the acetonitrile concentration increased. The course of a typical experiment is illustrated in Figure 2. At the start of an experiment, reservoir one was held at -5.2 kV and reservoir three was held at -4.5 kV (Figure 2A). This generated a forward flow from reservoir one to the µESI. It also generated a small secondary flow from reservoir one to reservoir three, ensuring that only the buffer from reservoir one reached the µESI interface. In a second phase (Figure 2B), the potential on reservoir three was ramped from -4.5 to -5.2 kV over 14 min. At the same time, the potential on reservoir one was held constant. During this phase the flow from reservoir one to reservoir three slowly decreased. At a certain potential value, at which the potential on reservoir three was higher than the potential at the junction of 3724 Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
Figure 3. Gradient profile generated using (A) a 12-cm-long uncoated 50-µm-i.d. capillary and (B) a 12-cm-long (3-aminopropyl)silane-coated 50-µm-i.d. capillary. The gradient was generated using the device described in Figure 1 and the procedure described in Figure 2. The gradient was monitored in the MS following a signal generated by acetonitrile clustering.
the flow path from reservoirs one and three, no flow was apparent from reservoir three. At higher potential values an increasing flow from reservoir three toward the µESI interface was generated. The net effect of this step was an increase in the acetonitrile concentration flowing toward the MS. In a third phase (Figure 2C), the potential on reservoir one was slowly increased from -5.2 to -4.2 or -4.4 kV over 14 min and the potential on reservoir three was kept constant at -5.2 kV. During this ramping, the flow from reservoir one slowly decreased. At a certain point, the flow from reservoir one stopped and then reversed into reservoir one from reservoir three. Figure 3 A illustrates the solvent gradient. The gradient is well described with an exponential equation. The gradient started at 28 min and finished at 50 min, thus providing 22 min for a gradient to develop from 0 to 65% acetonitrile. The change in the solvent composition over the course of an experiment was also apparent from the current delivered by the power supply. During the formation of a gradient, the ionic composition of the solution in the channels of the device and in the transfer capillary changed. The current supplied by the µESI power supply was the summation of the current from the microfabricated device and the current for the electrospray ionization process. Changes in ionic composition in the channels of the microfabricated device and in the transfer capillary were
Figure 4. Analysis of 74 fmol of a myoglobin tryptic digest by gradient frontal analysis-MS/MS. The procedure described in Figure 2 was used to generate the gradient and a (3-aminopropyl)silanecoated capillary was used as the transfer line. Myoglobin (10 µL) at a concentration of 7.4 fmol/µL was injected off-line. The trace for each peptide identified is displayed.
therefore reflected by changes in the current provided by the MS high-voltage power supply. In the first phase of the experiment, the current provided by the MS power supply was stable, reflecting a stable ionic composition of aqueous buffer in the channels and in the transfer capillary. In the second phase, the current provided by the µESI power supply started to drop, due to the difference in conductivity between the solutions in reservoirs one and three, indicating an increasing concentration of acetonitrile present in
the channels and transfer capillary. In the third phase, the current continued to drop and eventually reached a plateau reflecting a stable acetonitrile concentration in the channels and transfer capillary. Typically, the shapes of the gradients were similar for all the experiments done with the same capillary. However, the starting point and end point of the gradient were influenced by the size and the compression state of the C18 cartridge. As the size of the cartridge increased, the flow restriction also increased. Similarly, as the compression of the beads increased, the flow restriction increased and the onset of the gradient was delayed. Currently, it is difficult to control these parameters with this type of experiment. However, since the analytes were identified by their m/z ratio or CID spectra, reproducible elution times were not essential. We believe that in an electrochromatography type of experiment in which the column is longer and well packed, these parameters will be better controlled and that therefore the onset of the gradients will be more reproducible. Modification of the Gradient. The onset and the shape of the gradient are expected to depend on the flow rate of the solvent. We therefore tested the effect on the gradient generated by replacing the bare silica transfer capillary with an (aminopropyl)silane-coated transfer capillary. At pH 3.0, the inner surface of this coated capillary is protonated and generates a strong electroosmotic flow toward the µESI. The gradient generated using a coating capillary under conditions otherwise identical to the ones described above is shown in Figure 3B. Because the electroosmotic pumping was stronger, the gradient started earlier and was sharper compared to the bare fused-silica capillary. The gradient
Figure 5. Analysis of a band of yeast proteins separated by 1D gel electrophoresis of yeast total cell lysate and digested with trypsin. The proteins migrating to a band with an apparent molecular mass of 34 kDa were digested with trypsin and the extracted peptide mixture was analyzed by frontal analysis-MS/MS. Six different proteins were identified from the mixture: G3P1 yeast, G3P2 yeast, G3P3 yeast, IPYR yeast, RLAO yeast, and BMH1 yeast. The peptides identified are detailed in Table 1. The position of the peptides for only two of the proteins is illustrated for clarity. The procedure described in Figure 2 was used to generate the gradient, and an uncoated capillary was used as the transfer line. The sample was loaded off-line.
Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
3725
started at 20 min and was finished by 25 min. This indicates that the shape, slope, and direction of the gradient can be controlled by various parameters, including the control of the voltages applied to the reservoirs and surface chemistry of the transfer capillary. Application of Gradients: Spatial Separation of Protein Digests. Standard Protein Digests. The development of the C18 cartridge present in the ion source with the solvent gradient generated on the microfabricated device created a system that was suitable for the concentration and frontal analysis of protein digests by MS. First we established the limit of detection of this system for the analysis of a calibrated protein digest. Subsequently, protein samples of increasing complexity were analyzed. A standardized solution of myoglobin digested with trypsin was used. Figure 4 illustrates the analysis of a 10-µL aliquot of 7.4 fmol/µL (total 74 fmol) digested myoglobin on the gradient system equipped with a derivatized transfer capillary. The sample was pressure loaded (6-9 psi) off-line. The transfer capillary was then reconnected to the microfluidic device, and the gradient was developed. As the peptides eluted from the C18 cartridge, they were detected by the MS. If a specific peptide ion exceeded a predetermined intensity, the instrument automatically switched to MS/MS mode and the resulting CID spectra were recorded. These spectra were used in conjunction with the Sequest program to search a horse protein sequence database to identify the origin of the peptides. The control of the high-voltage power supply and the generation of CID spectra were performed automatically using a Labview procedure and a method developed with the software provided with the MS. The trace of the intensity versus time for each one of the identified peptides was recreated and displayed in Figure 4. The drop in intensity in the profiles is due to the MS continuously switching from MS to MS/MS mode. The secondary peaks are artifacts caused by daughters product ions from other precursor ions that have the same m/z as the identified peptides. Six peptides were identified as being derived from myoglobin. Other peptides from myoglobin were also present. These peptides did not generate CID spectra of good enough quality for unambiguous identification or were small peptides with a 1+ charge. Such spectra were not assigned by the Sequest software. This experiment was repeated at least three times with different solutions of trypsinized myoglobin and similar results were obtained each time. It has to be pointed out that the detected peaks are not chromatographic peaks but represent a frontal analysis performed on a small column. Therefore, the peak widths are a function of the gradient shape and the peptide diffusion coefficients. The signal intensity was well above the background signal. Using the signal from the peptide ion at m/z ) 804.22+ and a noise value calculated as three times the average background signal in a mass window 20 Da below the detected peptide ion, we determined a limit of detection of 1 fmol and a concentration LOD of 100 amol/µL, assuming that 10 µL of sample was applied to the C18 cartridge. Analysis of Proteins Separated by 1D and 2D Gel Electrophoresis. To demonstrate the general applicability of this approach the analysis of unknown proteins separated by standard techniques, such as 1D and 2D gel electrophoresis, was performed on this system. Saccharomyces cerevisiae was selected as the source of the proteins because the complete genome sequence is available, and therefore, the CID spectra of each peptide are 3726 Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
Table 1. Peptides Identified by the Sequest Software during the Analysis of a 34-kDa Band Digest by Gradient Frontal Analysis-MS/MS (see Figure 5) [MH]+ Xcorra 1148.7 1197.6 1470.8 1845.9 2593.3 2403.2 2631.3 2575.3 2207.1 1752.8 1539.7 1400.7 1834.9 1985.1 1378.8 1834.9 1574.8 2517.3 1267.6 1295.7 2131.3 1283.7 1060.6 1834.0 1476.0 1718.0 1358.3 1383.7 3266.1
3.0 2.5 3.8 4.3 3.3 5.7 5.1 4.3 3.5 3.1 2.4 2.4 3.2 3.4 3.0 3.4 3.0 4.9 2.3 3.5 2.6 3.9 2.0 3.3 3.6 4.2 4.1 2.9 5.9
2211.1
5.2
sequence
protein
(R)VVDLIEYVAK (R)DPANLPWGSLK (R)VPTVDVSVVDLTVK (K)VINDAFGIEEGLMTTVH (K)VINDAFGIEEGLMTTVHSMTATQK (K)DPANLPWGSSNVDIAIDSTGVFK (K)PNVEVVALNDPFITNDYAAYMFK (K)VINDAFGIEEGLMTTVHSLTATQK (K)VVITAPSSTAPMFVMGVNEEK (K)LVSWYDNEYGYSTR (S)STAPMFVMGVNEEK (K)EETLNPIIQDTK (K)IPDGKPENQFAFSGEAK (K)LEITKEETLNPIIQDTK (K)VIAIDINDPLAPK (K)GIDLTNVTLPDTPTYSK (K)ENNIFNMVVEIPR (K)AVGDNDPIDVLEIGETIAYTGQVK (R)GFLSDLPDFEK (K)TSFFQALGVPTK (K)SLFVVGVDNVSSQQMHEVR (R)AGAVAPEDIWVR (R)GTIEIVSDVK (K)TASEIATTELPPTHPIR (R)FLEQQNQVLQTK (K)QISNLQQSISDAEQR (K)LNDLEDALQQAK (K)SLNNQFASFIDK (K)DIENQYETQITQIEHEVSSSGQEVQSSAK (K)LGEHNIDVLEGNEQFINAAK
G3P1 G3P1 G3P1, G3P2, G3P3 G3P1, G3P2, G3P3 G3P1, G3P2 G3P3 G3P3 G3P3 G3P2, G3P3 G3P2, G3P3 G3P2, G3P3 IPYR IPYR IPYR IPYR IPYR IPYR IPYR RLA0 RLA0 RLA0 RLA0 RLA0 BMH1, BMH2 K2C1 human K2C1 human K2C1 human K2C1 human K1CI human TRYP pig
a A Xcorr higher than 2.0 indicates a confident match between the generated spectrum and the protein database.
expected to match to the sequence database. We first analyzed tryptic digests of protein bands obtained from a 1D gel electrophoresis separation of total yeast lysate. Figure 5 shows the frontal analysis of a tryptic digest of a sample representing a single band migrating at 34 kDa in the gel. The sample was pressure loaded off-line on the C18 cartridge. The transfer capillary was then reconnected to the microfabricated device, and the gradient was generated. CID spectra of the eluted peptides were generated on the fly. The signal intensity was a few orders of magnitude above the limit of detection. Six different proteins were identified as being present in this band with up to seven peptides identifying a specific protein. The results from the search of the CID spectra against a yeast protein sequence database using the Sequest software are summarized in Table 1. All the peptides identified had a Xcorr higher than 2.0, which indicates a confident match to the sequence database. Only one peptide was matched with proteins BMH1/BMH2. Because of the quality of the database match and the match of the observed and calculated molecular weight of BMH1/BMH2 (29960/30930), we included the proteins in the list of positively identified proteins. It should be pointed out that the acetonitrile cluster signal used to document the gradient in Figure 3 was not visible in Figure 5 because of the way the data were displayed. Furthermore, not all the peaks present in Figure 5 contained peptides. Some peaks were due to other chemicals present in the sample and others were artifacts caused by the continuous switching of the mass spectrometer between MS mode and MS/MS mode. Peptides
Figure 6. Analysis of a spot of yeast protein separated by 2D gel electrophoresis of yeast total cell lysate. The gel piece was digested with trypsin, and the extracted peptide mixture was analyzed by frontal analysis-MS/MS. The procedure described in Figure 2 was used to generate the gradient and a (3-aminopropyl)silane-coated capillary was used as the transfer line. Trypsin peptides are identified with a “T” and keratin peptides with a “K”. The sample was loaded off-line. The peptides identified are detailed in Table 2. Table 2. Peptides Identified by the Sequest Software during the Analysis of a Single Spot from an Yeast 2D Gel by Gradient Frontal Analysis-MS/MS (see Figure 6) [MH]+
Xcorr
sequence
protein
943.0 1886.5 2419.6 630.4 1940.9 2299.2 2284.9 2210.6 1475.4
3.1 3.1 3.5 1.6 2.3 3.7 3.3 4.4 4.8
(K)INEGILQR (K)DTLPLGFTFSYPASQNK (K)GFDIPNVEGHDVVPLLQNEISK (K)SLGIIGA (R)LGEHNIDVLEGNEQFIN (K)IITHPNFNGNTLDNDIM#LIK (K)IITHPNFNGNTLDNDIMLIK (R)LGEHNIDVLEGNEQFINAAK (R)FLEQQNQVLQTK
HXKA HXKA HXKA HXKA TRYP pig TRYP pig TRYP pig TRYP pig K2C1 human
from keratin and trypsin were also observed but were not labeled in Figure 5 and were added to Table 1 for clarity. These results clearly demonstrate that this system can be used to deconvolute a complex mixture of peptides at high sensitivity and in an automated manner. We next analyzed proteins obtained from a 2D gel separation of a total yeast lysate. A representative analysis of the protein in a selected spot is presented in Figure 6. Four peptides were identified as being from HXKA yeast along with one peptide from keratin and four peptides from trypsin. The results from the search of the CID spectra against the yeast protein database are summarized in Table 2. All the peptides identified had a Xcorr higher than 2.0 except for [MH]+ ) 630.4 which was a small peptide observed as a 1+ ion. The amount of protein present in the gel was estimated to be 200-300 fmol. CONCLUSIONS In the experiments described in this paper, we have demonstrated for the first time the use of a microfabricated microfluidics
module for the generation of solvent gradients which sequentially elute peptides absorbed on a SPE cartridge. The system was used for frontal analysis of complex peptide mixtures and the analysis on-line of the separated peptides by MS/MS. We have used the system for the analysis of protein digests, determined limit of detection in the low-femtomole level, and demonstrated the general usefulness of such a system for high-sensitivity protein analysis by identifying yeast proteins separated by 1D and 2D gel electrophoresis. We anticipate the coupling of an electrochromatography column to the microfabricated system to generate a pumpless gradient electrochromatography system. We also foresee that the gradient system will be used to mix nanoliter amounts of different reactants, such as enzyme and substrates, to perform microvolume reactions monitored on-line by MS/MS. The integration of modules with specific tasks on a microfluidic device and the coupling of such devices to a MS will be crucial for the evolution and general applicability of microfabrication technology in the analytical biosciences. ACKNOWLEDGMENT This work was supported by the National Science Foundation Science and Technology Center for Molecular Biotechnology, Oxford GlycoSciences, and Zymogenetic. The 1D gel of yeast was provided by S. Gygi, and the 2D gel of yeast was provided by Y. Rochon and B. Franza, Department of Molecular Biotechnology, University of Washington.
Received for review May 7, 1998. Accepted July 7, 1998. AC980502J Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
3727