Techniques for the Optimization of Proteomic Strategies Based on

Anal. Chem. 2000, 72, 2684-2689

Techniques for the Optimization of Proteomic Strategies Based on Head Column Stacking Capillary Electrophoresis Steven Locke and Daniel Figeys*

National Research CouncilsCanada, 1411 Oxford Street, Halifax, Nova Scotia, Canada B3M 3R8

Proteomics is the large-scale study of the proteins related to a genome. Presently, proteomic procedures have relied on mass spectrometry as a tool of choice to perform analysis of proteins. Optimization and understanding of the different steps involved in proteomics using mass spectrometry is expensive and time-consuming and, for this reason, have been typically paid insufficient attention. However, optimization becomes a critical issue as we try to analyze ever shrinking amounts of proteins. We present here the development of a technique that allows the rapid, sensitive, semiquantitative, and automated optimization of the processes involved in proteomics. Furthermore, it allows the rapid testing of new methodologies without having to rely on expensive mass spectrometric techniques. The technique, based on head column stacking capillary zone electrophoresis, allows the concentration, separation, and analysis of protein digests at concentrations from high picomoles to subfemtomoles per microliter and sample volumes from a few microliters to a few hundred microliters produced by proteomic processes. Furthermore, the incorporation of UV detection in the system allows the tracking of the relative changes in peptide levels observed during optimization. In addition, all the buffers and solvents used in this technique are compatible with its future coupling to electrospray ionization mass spectrometry. The potential of this technique for the analysis of low-abundance proteins is demonstrated using peptide standards and tryptic digests of standard proteins. Moreover, we exemplify the application of this technique in proteomic prototyping for the rapid and automated study of the procedure of enzymatic digestion of proteins.

The identification and detailed analysis of minute amounts of proteins is becoming increasingly important to the understanding of biological processes. Typically, these studies are performed either through conventional biochemical approaches1 involving a limited number of relevant proteins or through large-scale screening procedures such as proteomics.2-4 * Corresponding author: MDS-Ocata, 600 University Ave., Suite 1075, Toronto, ON, M5G 1X5 Canada; (e-mail) [email protected]. (1) Gooley, A. A.; Ou, K.; Russell, J.; Wilkins, M. R.; Sanchez, J. C.; Hochstrasser, D. F.; Williams, K. L. Electrophoresis 1997, 18, 1068-72.

2684 Analytical Chemistry, Vol. 72, No. 13, July 1, 2000

2D-gel electrophoretic separation of proteins5 is one of the founding techniques of today’s proteomic field. Over the years, other techniques have been appended to 2D-gel electrophoresis, enhancing the quantitative and qualitative results. Presently, in a typical proteomic study, a cell lysate is generated and the proteins present in the mixture are concentrated, separated, and visualized by 2D-gel electrophoresis5 with silver staining techniques.6 Selected proteins are subjected to in-gel proteolytic digestion7 and identified either by MALDI-TOF mass spectrometry8-10 or by electrospray ionization mass spectrometry7,11-17 combined with protein/DNA database searching.18-22 Proteomic technologies still have some serious limitations that need to be addressed. One of the serious limitations of proteomics that has been recently illustrated is the apparent failure of current (2) Wilkins, M. R.; Sanchez, J. C.; Gooley, A. A.; Appel, R. D.; Humphery, S.-I.; Hochstrasser, D. F.; Williams, K. L. Biotechnol. Genet. Eng. Rev. 1996, 13, 19-50. (3) Wilkins, M. R.; Pasquali, C.; Appel, R. D.; Ou, K.; Golaz, O.; Sanchez, J. C.; Yan, J. X.; Gooley, A. A.; Hughes, G.; HumpherySmith, I.; Williams, K. L.; Hochstrasser, D. F. Bio Technology 1996, 14, 61-65. (4) Haynes, P. A.; Gygi, S. P.; Figeys, D.; Aebersold, R. Electrophoresis 1998, 19, 1862-1871. (5) O’Farrell, P. H. J. Biol. Chem. 1975, 250, 4007-4021. (6) Blum, H.; Beier, H.; Gross, H. J. Electrophoresis 1987, 8, 93-99. (7) Wilm, M.; Shevchenko, A.; Houthaeve, T.; Breit, S.; Schweigerer, L.; Fotsis, T.; Mann, M. Nature 1996, 379, 466-9. (8) Shevchenko, A.; Wilm, M.; Vorm, O.; Mann, M. Anal. Chem. 1996, 68, 850-8. (9) Patterson, S. D.; Thomas, D.; Bradshaw, R. A. Electrophoresis 1996, 17, 877-891. (10) Patterson, S. D.; Aebersold, R. Electrophoresis 1995, 16, 1791-1814. (11) Figeys, D.; Ducret, A.; Oostveen, I. v.; Aebersold, R. Anal. Chem. 1996, 68, 1822-28. (12) Figeys, D.; Ducret, A.; Yates, J. R., III; Aebersold, R. Nat. Biotechnol. 1996, 14, 1579-1583. (13) Yates, J. R., III In Cell Biology: A Laboratory Handbook; Academic Press: San Diego, 1998; Vol. 4, pp 529-538. (14) Yates, J. R., III; McCormack, A. L.; Link, A. J.; Schieltz, D.; Eng, J.; Hays, L. Analyst 1996, 121, R65-R76. (15) Wilm, M.; Mann, M. Anal. Chem. 1996, 68, 1-8. (16) Chervet, J. P.; Ursem, M.; Salzmann, J. B. Anal. Chem. 1996, 68, 15071512. (17) McCormack, A. L.; Schieltz, D. M.; Goode, B.; Yang, S.; Barnes, G.; Drubin, D.; Yates, J. R. Anal. Chem. 1997, 69, 767-776. (18) Eng, J.; McCormack, A. L.; Yates, J. R., III. J. Am. Soc. Mass Spectrom. 1994, 5, 976-89. (19) Yates, J. R., III; Eng, J. K.; McCormack, A. L.; Schieltz, D. Anal. Chem. 1995, 67, 1426-36. (20) Mann, M.; Wilm, M. Anal. Chem. 1994, 66, 4390-9. (21) Taylor, J. A.; Johnson, R. S. Rapid Commun. Mass Spectrom. 1997, 11, 1067-1075. (22) Wise, M. J.; Littlejohn, T. G.; Humphery, S.-I. Electrophoresis 1997, 18, 1399-409. 10.1021/ac0003293 CCC: $19.00 Published 2000 Am. Chem. Soc. Published on Web 06/03/2000

methodologies to routinely and systematically analyze lowabundance proteins.4,23 Many hypothetical explanations and potential remedies for this problem have been suggested. However, it is clear that the underlying problem is the lack of techniques that can be used for the optimization and understanding of proteomic processes. Only recently have the mass spectrometry techniques become available for the analysis of low levels of protein samples. Typically, the methodologies that were used for large amounts of proteins were assumed to be applicable to lowabundance proteins, or optimizations were performed to only a limited extent due to the cost and time related to mass spectrometry analysis. To our knowledge there is currently no system available that can be used for the off-line optimization of proteomic processes. Furthermore, novel ideas and hypotheses to improve proteomic schemes are difficult to test due to the current reliance on mass spectrometric analysis. Here we present a novel approach that allows the rapid, inexpensive, and automated analysis of low femtomole to attomole levels of protein digests for the optimization of proteomic processes. The approach consists of head column stacking capillary electrophoresis with UV detection for the off-line analysis of small volumes and low concentrations of biological samples. We demonstrate that this technique is a rapid, robust, and automated alternative for the analysis of protein digests at the low femtomole to subfemtomole per microliter level and can handle large sample throughput. We also demonstrate that this technique can significantly concentrate the analytes contained in a large sample volume, thus making it a method of choice for the analysis of proteolytic digests. The ease of utilization, sensitivity, and differential quantitation obtained with this technique when coupled to UV detection make it an excellent tool for the optimization and testing of protocols for protein analysis. Furthermore, the approach was developed with the intention for its future coupling to ESI-MS for the analysis of protein digests. EXPERIMENTAL SECTION Material and Supplies. The standard peptides, proteins, ammonium bicarbonate, and formic acid were obtained from Sigma (St Louis, MO). Sequencing grade modified trypsin was obtained from Promega (Madison, WI). Acetic acid, 2-propanol, and acetonitrile were purchased from Caledon (Georgetown, ON). Morpholine, trifluoroacetic acid, and sodium hydroxide were purchased from Aldrich. The ZipTip pipet tips were obtained from Millipore. All the water used in the experiments was doubledistilled and deionized using a MilliQ system (Millipore, Bedford, MA). The fused-silica capillary was procured from Polymicro Technologies (Phoenix, AZ). Preparation of Peptide Standard Solutions. All the standard peptides, GGR, GGYR, TSK, and PPGFSPFR were dissolved in 0.1% HOAc, 50% ACN (v/v) at 1-2 mg/mL and diluted to the desired concentration in the same solvent. These samples were not subjected to ZipTip cleanup. BSA and Apomyoglobin Tryptic Digests. Solution Digests. The samples for the study of the effect of the substrate concentration on the kinetics of digestion were prepared as follows. The samples with varying concentrations of apomyoglobin and a (23) Gygi, S.; Rochon, Y.; Franza, B.; Aebersold, R. Mol. Cell Biol. 1999, 19, 1720-1730.

constant concentration of trypsin (500 fmol/µL) in 10 mM ammonium bicarbonate, pH 8.0, were made by serial dilution of a standard 10 pmol/µL apomyoglobin solution containing trypsin at 500 fmol/µL with another solution containing only trypsin at 500 fmol/µL. The digestion of all the solutions was performed at 37 °C overnight. Aliquots (10 µL) of each digest were lyophilized in a Savant Speed Vac and then heated at 75 °C for 90 min. The samples were resuspended in 10 µL of 0.1% HOAc, 50% ACN (v/ v) containing 1 pmol/µL each of saxitoxin, histidine, and arginine as internal standards. The samples for the study of the effect of trypsin concentration on the kinetics of digestion were prepared as follows. The samples with a constant concentration of BSA and varying concentrations of trypsin were made by diluting a solution that contained trypsin at 500 fmol/µL and BSA at 10 pmol/µL with a solution that only contained BSA at 10 pmol/µL. All the digestions were conducted at 37 °C for 18 h. A 10-µL aliquot of each digest was desalted by the ZipTip cleanup method. ZipTip Cleanup of Digests. After addition of 2 µL of 10% TFA (v/v) to the 10-µL volume, samples were subjected to ZipTip cleanup. Every ZipTip was preconditioned twice with 10 µL of 0.1% TFA in 50% ACN and 10 µL of 0.1% TFA in H2O. The sample was loaded on the tip by pipetting the solution 20 times in and out of the tip. The tip was also rinsed twice with different aliquots of 0.1% TFA in H2O and eluted with 7 volumes of 5 µL of 0.1% HOAc in 2-PrOH/H2O/ACN (25/25/50). Prior to analysis by CE, eluted peptides were lyophilized and reconstituted with 10 µL of 0.1% HOAc, 50% ACN. Capillary Electrophoresis. The concentration and separation of the samples were conducted using a Beckman P/ACE 2100 instrument with a 77-cm-long (70 cm to the detector) uncoated fused-silica capillary (50 µm i.d. × 360 µm o.d.). The background electrolyte was 50 mM morpholine with pH adjusted to 3.00 by the addition of formic acid. The field-amplified sample stacking (head column stacking) technique was used for the concentration and the separation of analytes. In this procedure, the capillary was rinsed with 1 M NaOH, H2O, and electrolyte prior to every sample loading. This was followed by the pressure injection on the capillary of a plug of water (0.8 min at 0.5 psi). The sample was loaded at the head of the column typically by applying 5 kV across the capillary for 5 min. With both electrode reservoirs filled with background electrolyte, the separation of peptides was performed by applying +30 kV across the capillary. The UV absorbance of analytes was monitored at 200 nm. RESULTS AND DISCUSSION The literature on protein analysis by mass spectrometry and large-scale proteomic studies reveals an astonishing and disappointing lack of identification of low-abundance proteins by conventional 2D-gel electrophoresis of whole cell lysates and mass spectrometry analysis of protein digests. In fact, most successes have been obtained not in the frame of large-scale proteomics studies, but in the context of conventional biochemistry studies coupled with affinity enrichment of the selected proteins. Different explanations have been proposed for this problem: unfavorable enzymatic kinetics during the digestion of proteins present at the low-femtomole level and lower; rapid turnover of these low-level proteins in the cells, which means that the amount obtained from Analytical Chemistry, Vol. 72, No. 13, July 1, 2000

2685

typical cell lysate preparation is too low to be visualized by 2Dgel electrophoresis and silver staining; and sample loss through the numerous steps involved in the preparation of protein samples for mass spectrometry analysis. This situation strikingly illustrates the predominant problem in proteomics, viz., the lack of reliable information on the different processes involved. Capillary electrophoresis has been often proposed as a viable and easy-to-use alternative for the analysis of low levels of proteins. Impressive mass limits of detection have been reported for the analysis of peptide mixtures by electrospray ionization mass spectrometry.24-26 However, capillary electrophoresis requires the injection of a small volume (nanoliter range) of sample from a highly concentrated sample to maintain its separation efficiency. Often biological samples, provided as mixtures of dilute proteins or peptides that are contained in 10-100 µL of solution, are incompatible with CE analysis. We have developed a modified capillary electrophoresis technique based on head column stacking that can concentrate dilute levels of protein digests from large sample volumes and can also provide high resolution, rapid separation, and sensitive detection of the proteolytic digest components. Furthermore, it is worth noting that all solvents and buffers utilized in this technique are compatible with the sensitive analysis of protein digests by microelectrospray, making it a potentially powerful technique when coupled to electrospray mass spectrometry. Injection/Concentration Process. In this modified CE technique, peptides contained in a large volume of sample can be effectively stacked at the head of a CE capillary prior to separation. This is achieved by preceding the sample injection with a plug of water that creates a nonuniform electric field across the capillary during the injection of the peptide mixture. A highvoltage drop occurs across the small plug of low-conductivity water while a relatively smaller voltage drop occurs across the rest of the capillary filled with the higher conductivity background electrolyte. Furthermore, at low pH the electroosmosis is limited, which means that the plug of water stays at the injection end. Therefore, the peptides that enter the capillary rapidly move to the front of the water plug. At that point they encounter a lower electric field and slow, a process that literally stacks the peptides at the front of the water plug/electrolytes interface. The amount of sample injected on the column will be related to the current during the electrokinetic injection and the peptide transference number. The transference number is the chargecarrying capacity of the peptide and is related to its concentration and relative mobility. In this case, the relevant quantity is the sum of the [concentration × mobility] products of all the positive ions present in the sample. Our experiments are designed such that the buffer present in the sample is always the same. One of the objectives of this technique is to follow the changes in individual peptide levels produced by proteolytic digestion of proteins as changes are introduced in a proteomic platform. Therefore, we are not seeking absolute quantitation but differential quantitation (24) Figeys, D.; Ducret, A.; Aebersold, R. J. Chromatogr., A 1997, 763, 295306. (25) Smith, R. D.; Udseth, H. R.; Wahl, J. H.; Goodlett, D. R.; Hofstadler, S. A. Methods Enzymol. 1996, 271 (High-Resolution Separation And Analysis Of Biological Macromolecules, Pt. B). (26) Valaskovic, G. A.; Kelleher, N. L.; McLafferty, F. W. Science 1996, 273, 1199-1202.

2686 Analytical Chemistry, Vol. 72, No. 13, July 1, 2000

Figure 1. Electropherogram obtained for the analysis of a peptide mixture using the modified CZE-UV technique. Injection at +5 kV for 5 min was performed from a 200-µL solution of four peptides at 1 fmol/µL each. All other conditions are described in the Experimental Section. Peptide 3 was used to establish the limit of detection (Figure 2).

of the individual peptides. For a constant background electrolyte, such differential quantitation is not affected by the injection method. Limit of Detection. The development of novel techniques requires the demonstration of a limit of detection compatible with the analysis of low-level proteins. Capillary electrophoresis is well known to have a poor concentration limit of detection. Therefore, it is of utmost importance to demonstrate that the modified CE technique developed here is compatible with the analysis of low sample concentrations. We selected a set of four peptides that mimic peptides typically obtained by proteolytic digestion of proteins to establish the limits of detection of the CE-UV system. The concentration limit of detection was evaluated using two sets of constant-volume aliquots of sample at different concentrations, injected/concentrated and analyzed on the system as already described. A stock solution of the four peptides was made and the concentrations of the peptides were calculated based on the dry weights of peptides utilized to prepare the solution. Different concentrations of the mixture were prepared by serial dilution of the stock solution. The head column stacking capillary electrophoresis-UV detection system was utilized to analyze the different solutions and to calculate limits of detection. Figure 1 shows the electropherogram obtained for the injection at +5 kV for 5 min from 200 µL of the peptide mixture at 1 fmol/µL. Peptide 3 was selected to evaluate the limit of detection. The peaks for the selected peptide were integrated and the S/N ratio was calculated as the peak area divided by 3 times the background standard deviation. Figure 2 shows the change in S/N ratio versus concentration for 10- and 200-µL aliquots. It is clear that linearity is fairly well conserved from low-femtomoles to midfemtomoles per microliter. Furthermore, due to the depletion of sample phenomenon the change in slope of the curves is dependent on the volume of sample available. From these traces, a limit of quantitation (conveniently defined as 10 times the background standard deviation) was calculated to be 6 amol/µL for 200 µL of sample and 220 amol for 10 µL of sample. Furthermore, a mass limit of detection of 1-2 fmol was calculated from both curves. We noticed in the electropherograms that the head column stacking technique also concentrated contaminants. We have

Figure 2. Change in S/N ratio versus the concentration of peptide for 10 (9) and 200 µL ([) of sample. The S/N ratio was measured for peptide 3 in Figure 1 by integrating the area under the peak and dividing it by 3 times the background standard deviation. All other conditions are described in the Experimental Section.

Figure 4. Electropherograms obtained for the analysis of (A) R-casein tryptic digest and (B) β-casein tryptic digest both at 10 pmol/ µL. The injections were performed for 30 s at +5 kV. All other conditions as described in the Experimental Section.

Figure 3. Depletion of peptides by repetitive injection from respectively 10- (9), 25- (2), and 200-µL ([) volumes of peptide mixture at 25 fmol/µL. The measured S/N ratio for the labeled peptide in Figure 1, as a function of the injection number, is reported. Subsequent injections and separations by the modified CE technique were performed from the same volume aliquots of peptide mixture. All other conditions are described in the Experimental Section.

established that some trace amounts of contaminants that absorb in the UV were present in the doubly distilled deionized water used for sample preparation. Sample Depletion. Electrokinetic injection is a current-limited phenomenon. The amount of sample injected when the electroosmosis is limited is related to the transference number of the analytes. Hence, during injection, the positive analytes and buffer components leave the sample vials while the volume in the vial stays approximately the same due to the limited electroosmotic pumping. For long injection times, it is possible to significantly deplete the peptides from a sample solution. To test this hypothesis, three sets of different volume aliquots of the peptide mixture at constant concentration were subsequently concentrated and analyzed on the CE system. Figure 3 shows the S/N ratio obtained for the subsequent injection of samples from, respectively, 200-, 25-, and 10-µL aliquots of samples. It is clear for all sample volumes that the intensity significantly decreases for the subsequent injections. It is also evident that the depletion of sample is related to the volume of the sample. For the 10-µL aliquot, ∼60% of the sample was injected in the first injection and the sample was depleted by 90% after two injections. Furthermore, the depletion pattern from the 25-µL aliquot sample was similar. Even higher percentages of the sample can be injected by increasing the injection time. This clearly illustrates that the

head column stacking is an efficient way of injecting/concentrating dilute levels of peptides from a relatively large volume of sample. Typically, the processing of proteins isolated from 1D- or 2D-gel electrophoresis generates a volume of proteolytic digest in the range of 10-25 µL. This is well within the depletion range demonstrated by this study, and larger volumes can always be reduced by concentration before analysis. Protein Solution Digests. The simplicity of the standard peptide mixtures was required to establish the intrinsic limit of detection of this system and to study the phenomena associated with this technique. However, it is important to demonstrate that this novel approach is also compatible with the analysis of more complex samples generated by proteolytic digestion of proteins. Stock solutions of R-casein (R-cas), and β-casein (β-cas) were digested with trypsin overnight at 37 °C. Aliquots of protein digests were prepared and analyzed as described above. Figure 4 shows the separation obtained for R-cas and for β-cas at 10 pmol/µL. The traces are noticeably different and correspond to the numbers of peptides expected for the tryptic digestion of these proteins. These results demonstrate that the technique is compatible with complex protein digests. Furthermore, they also demonstrate that the dynamic range of the technique can easily accommodate highto low-concentration samples by adjusting the injection time appropriately. Enzyme Kinetics. Many publications have reported the analysis of low levels of standard protein digests by coupling separation techniques to electrospray ionization mass spectrometers.12,17,26,27 However, these limits of detection were established by digesting a concentrated stock solution of proteins (i.e., favorable kinetic conditions) which was then brought by serial dilutions to the femtomole to subfemtomole level. It became rapidly apparent that the limit of detection obtained by serial (27) Figeys, D.; Aebersold, R. Electrophoresis 1997, 18, 360-368.

Analytical Chemistry, Vol. 72, No. 13, July 1, 2000

2687

dilution of a concentrated stock solution of a standard digest was not transferable to the levels that could be analyzed for real ingel proteolytic digests. Unfavorable enzyme kinetics for the direct digestion of low-femtomole or subfemtomole levels of proteins was proposed to explain this discrepancy. We propose to utilize the automated off-line head column stacking CE-UV system to test this assumption. Two main variables are likely to affect the kinetics of digestion of a protein; the first variable is the concentration of the substrate (protein) and the other is the concentration of the enzyme, in this case trypsin. Therefore, we performed two sets of experiments, each holding one of these factors constant while the concentration of the other was changed. The objective of these two experiments was to discover any unfavorable kinetic properties present during the digestion of low-femtomole quantities of protein. It was important to discern the kinetic effect from other effects that can occur during the digestion of gel-separated proteins, such as limited diffusion of the enzyme during the gel digestion or chemical interferences. Hence, the digestion process was replicated in solution. In the first set of experiments, digestions of different concentrations of proteins were performed using a constant level of enzyme, and the resulting peptide mixtures concentrated and analyzed on the CE system. Standardized apomyoglobin was used as the substrate and digested at different concentrations using 500 fmol/µL trypsin. We are aware that there is currently a lack of protein reference standards that fully reflect the wide variety of protein biochemistry. Apomyoglobin and BSA (see below) are reasonable standards that have been used in many studies allowing the comparison of results. We found that it was not possible to use our stacking technique with a high level of ammonium bicarbonate present in the sample. Therefore, after digestion, the samples were lyophilized and heated to remove the volatile salt. We found that heating of the sample at 75 °C for 60-90 min was enough to reduce the salt content to a level (