Ionization

---

Anal. Chem. 2000, 72, 4785-4795

Capillary Electrophoresis-Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry Using a Vacuum Deposition Interface Jan Preisler, Ping Hu, Tomas Rejtar, and Barry L. Karger*

Barnett Institute and Department of Chemistry, Northeastern University, Boston, Massachusetts 02115

An improved vacuum deposition interface for coupling capillary electrophoresis with MALDI-TOF MS has been developed. Liquid samples consisting of analyte and matrix were deposited on a moving tape in the evacuated source chamber of a TOF mass spectrometer, enabling 24 h of uninterrupted analysis. The vacuum deposition procedure was compared with the dried-droplet method, and it was found that vacuum deposition generated significantly more reproducible signal intensity, eliminating the need for “sweet spot” searching. A concentration detection limit in the low-nanomolar range has been achieved with a low-attomole amount of sample consumed per spectrum. In addition, ion suppression caused by hydrophobicity differences in the analytes was reduced. To minimize ion suppression further, separation prior to MALDI MS analysis was employed. The performance of capillary electrophoresis (CE)-MALDI-TOF MS using the vacuum deposition interface was evaluated with a peptide mixture injected at low-femtomole levels. All peptides were baseline resolved with separation efficiencies in the range of 250 000-400 000 plates/m (2-3-s band halfwidth), demonstrating the high separation efficiency of the CE-MALDI MS coupling. A fast (∼40 s) CE separation of a mixture of angiotensins was found to reduce significantly ion suppression and enable trace level detection. It was also shown, for the analysis of an enolase digest, that sequence coverage of 65% was obtained using CE separation compared to 52% using step-elution solidphase extraction and 44% in the control experiment using an unseparated mixture.

With the near completion of Human Genome Project, attention is turning toward proteome analysis.1 Peptide mapping, using MALDI MS in combination with genome database searching strategies, has been shown to be a powerful tool for proteome investigation.2 While much emphasis has been placed on the use of MALDI MS-based methods for identifying proteins, there is growing interest in characterizing protein posttranslational modi(1) Williams, K. L. Electrophoresis 1999, 20, 678-688. (2) Chaurand, P.; Luetzenkirchen, F.; Spengler, B. J. Am. Soc. Mass Spectrom. 1999, 10, 91-103. 10.1021/ac0005870 CCC: $19.00 Published on Web 09/15/2000

© 2000 American Chemical Society

fications (PTMs).3 In the latter case, the success of the MALDI MS approach relies upon the ability to detect peptides at the trace level, recover protein digests with high sequence coverage, and obtain MS/MS information.2 MALDI MS is usually considered as a method that can be directly applied to the analysis of protein digest samples. In this protocol, the peptide mixtures are first cleaned from contaminants such as salts and detergents, typically using solid-phase extraction (SPE), and then directly analyzed by MALDI MS. However, the performance of MALDI MS can be severely compromised due to ion suppression effects.4-7 The composition, pH of the samplematrix solution, and rates of crystallization have been shown to affect the ion intensities of peptide components in MALDI mass spectra.4 The hydrophobicity and acid-base properties of amino acids side chains have also been found to be important factors for the observed signal intensity variations in MALDI MS,5 as shown by the dominant intensity of peptides with arginine residues.6 Uneven spatial distribution of peptides on the MALDI sample spot based on hydrophobicity, known as the Marangoni effect, can lead to severe ion suppression as well.7 Since ion suppression compromises performance of the MALDI analysis of peptide mixtures, incorporation of additional procedures prior to MALDI MS analysis is necessary. It has been shown that proper choice of the matrix and solvent system as well as rapid crystal formation reduces signal discrimination for short peptides.4 It can be further expected that ion suppression can be minimized by efficient separation prior to MALDI MS analysis. Separation is especially important for detecting peptides in MALDI MS at the trace level, since even mild ion suppression can lead to the total loss of signal in the background noise.8 Separation is also advantageous for the analysis of peptide mixtures originating from proteins with significantly different molar concentrations, e.g., the digest of the whole cell lysate,9 or (3) Hancock, W.; Apffel, A.; Chakel, J.; Hahnenberger, K.; Choudhary, G.; Traina, J. A.; Pungor, E. Anal. Chem. 1999, 21, 742A-748A. (4) Cohen, S. L.; Chait, B. T. Anal. Chem. 1996, 68, 31-37. (5) Kratzer, R.; Eckerskorn, C.; Karas, M.; Lottspeich, F. Electrophoresis 1998, 19, 1910-1919. (6) Krause, E.; Wenschuh, H.; Jungblut, P. R. Anal. Chem. 1999, 71, 41604165. (7) Amado, F. M. L.; Domingues, P.; Santana-Marques, M. G.; Ferrer-Correia, A. J.; Tomer, K. B. Rapid Commun. Mass Spectrom. 1997, 11, 1347-1352. (8) Zhang, H. Y.; Stoeckli, M.; Andren, P. E.; Caprioli, R. M. J. Mass Spectrom. 1999, 34, 377-383.

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000 4785

even in the digest of a single protein. In these cases, separation can improve the detection limit of high molecular weight peptides by preventing the TOF detector from saturation by highly abundant low molecular weight peptides.10 Another problem in the direct analysis of complex mixtures can be overall lower signalto-noise ratios, since the ion current is shared among many components.11 Furthermore, separation can help to remove contaminants from the sample and simplify the isolation of single analytes for analysis by postsource decay (PSD) or MS/MS, especially when resolution in the first MS dimension is limited. Coupling separation methods to MALDI MS can be accomplished in either the off-line or on-line modes. Off-line collection of an eluent from a separation column on a MALDI target deposited under atmospheric pressure in the form of spots (directly from separation column outlet12-14 or using a piezoelectric dispenser15) or a continuous streak8,16,17 has been demonstrated. In one application, eluents from a CE capillary were directly deposited on a precoated MALDI target.8,16 For stable electrical contact, the target was maintained wet during sample collection. While this approach was useful, loss of separation efficiency was found,16 likely due to analyte diffusion on the wetted membrane. Various interfaces allowing direct on-line coupling of a separation method with MALDI MS have also been developed, such as continuous flow (CF),18-21 aerosol,22 or more recently, rotating ball inlet (ROBIN) interface.23 These approaches have met with limited success, as they are either incompatible with a solid matrix or suffer from high sample consumption, memory effects, and/or low mass resolution. Our laboratory recently introduced a vacuum deposition interface for MALDI-TOF MS, which led to good spot-to-spot reproducibility and high mass sensitivity.24 The liquid sample of analyte and matrix was directly deposited on a rotating quartz wheel in the evacuated source chamber and transported to the repeller where laser desorption occurred. Rapid evaporation of the solvent resulted in formation of a thin, ∼60-µm-wide sample trace, with either an amorphous or a microcrystalline structure.24 The interface allowed rapid analysis of trace sample amounts as well as coupling of microcolumn liquid-phase separation techniques with MALDI MS, as demonstrated using capillary elec(9) Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold, R. Nat. Biotechnol. 1999, 17, 994. (10) Westman A.; Brinkmalm G.; Barofsky D. F. Int. J. Mass Spectrom. Ion Processes 1997, 169/170, 79. (11) Hillenkamp, F. 13th International Symposium on High Performance Capillary Electrophoresis and Related Microscale Techniques; Saarbruecken, Germany, February 20-24, 2000. (12) Stevenson, T. I.; Loo, J. A. LC-GC 1998, 16, 54-58. (13) Bergman, A. C.; Bergman, T. J. Protein Chem. 1997, 16, 421-423. (14) Walker, K. L.; Chiu, R. W.; Monnig, C. A.; Wilkins, C. L. Anal. Chem. 1995, 67, 4197-4204. (15) Miliotis, T.; Kjellstrom, S.; Nilsson, J.; Laurell, T.; Edholm, L. E.; MarkoVarga, G. J. Mass Spectrom. 2000, 35, 369-377. (16) Zhang, H. Y.; Caprioli, R. M. J. Mass Spectrom. 1996, 31, 1039-1046. (17) Biemann, K.; Gagel, J. J. U.S. Patent 4843243, 1989. Biemann, K. U.S. Patent 5770272, 1998. (18) Chang, S. Y.; Yeung, E. S. Anal. Chem. 1997, 69, 2251-2257. (19) Nagra, D.; Li, L. J. Chromatogr., A 1995, 711, 235-245. (20) He, L.; Li, L.; Lubman, D. M. Anal. Chem. 1995, 67, 4127-4132. (21) Zhan, Q.; Gusev, A.; Hercules, D. M. Rapid Commun. Mass Spectrom. 1999, 13, 2278-2283. (22) Murray, K. K. Mass Spectrom. Rev. 1997, 16, 283-299. (23) Ørsnes, H.; Graf, T.; Degn, H.; Murray, K. K. Anal. Chem. 2000, 72, 251254. (24) Preisler J.; Foret, F.; Karger, B. L. Anal. Chem. 1998, 70, 5278-5287.

4786

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

trophoresis. The area available for sample deposition was, however, limited, and the circumference of the wheel had to be cleaned after completion of each turn. In this paper, an improved vacuum deposition interface for coupling capillary electrophoresis with MALDI-TOF MS is described. Samples were deposited on a disposable moving tape, enabling ∼24 h of continuous analysis without interruption. The physical arrangement resembles that of the moving belt interface;25 however, the source chamber was at pressures lower than the vapor pressure of solvent. Detection limits in the low-nanomolar concentration range and mass resolutions up to ∼2000 (fwhm) have been obtained for small peptides under the delayed extraction mode. In addition, the vacuum deposition approach was compared to the commonly used dried-droplet method. It is shown that vacuum deposition produces a more homogeneous sample and reduces ion suppression caused by the difference in hydrophobicity. The interface has been used for coupling of capillary electrophoresis to MALDI-TOF MS and analysis of small peptides with high separation efficiency and attomole detection limits. CE separation prior to MALDI MS analysis has been demonstrated as an effective approach to overcome ion suppression. It is to be noted that the vacuum deposition approach can also be used in the off-line mode with a stand-alone vacuum chamber, followed by analysis on a commercial MALDI MS instrument.26 EXPERIMENTAL SECTION Mass Spectrometer. Figure 1 presents the design of the linear time-of-flight mass spectrometer, many details of which were described previously.24 Briefly, the instrument consisted of a Wiley-McLaren type ion source,27 a 1-m flight tube, and a dual microchannel plate. The distances between the repeller and the first grid and between the first and the second grids were 10.0 and 12.7 mm, respectively. High-voltage pulses for delayed extraction were produced by a voltage pulser (model HTS 151 A, Eurotek, Morganville, NJ); see Figure 1. A constant voltage of 12.0 kV, generated by a 75-MΩ resistor divider from the total voltage of 15.0 kV set by a power supply (model CZE1000R/X2263, Spellman, Hauppauge, NY), was applied on the extraction grid. The repeller, initially at the same potential as the extraction grid, was pulsed to the full voltage (15.0 kV) for ∼10 µs. The delay between the laser and high-voltage pulses was set with a digital delay generator (model 9650A, EG&G, Princeton, NJ), according to the m/z range that was to be focused. The values of the delay time in the text are the sum of the delay time set with the delay generator and 190-ns propagation time of the high-voltage pulser. The rise time of the high-voltage pulse (10-90%) was ∼80 ns. In the continuous extraction mode, the voltage on the repeller was adjusted to 15.0 kV, and both grids were grounded. Pressure in the flight tube was ∼2 × 10-7 (no infusion) and ∼1 × 10-6 Torr (capillary infusion). A 337-nm, 30-Hz nitrogen laser (model VSL337ND-S, Laser Science, Franklin, MA) was used for MALDI. As shown in Figure 1, the laser beam, attenuated by reflection from a wedge prism and a stepped neutral density filter (Edmund Scientific, Barrington, NJ), was directed with a mirror via a quartz lens (focal length 20 cm) on to the sample target at a 60° angle of (25) McFadden, W. H.; Schwartz, H. L.; Evans, S. 1976 J. Chromatogr. 1976, 122, 389-396. (26) Hu, P.; Rejtar, T.; Preisler, J.; Karger, B. L. To be published. (27) Wiley: W. C.; McLaren, I. H. Rev. Sci. Instrum. 1955, 26, 1150-1157.

Figure 1. Schematic diagram of the laboratory-built on-line MALDI-TOF MS. Low-voltage connection (thin solid lines), high-voltage connection (thick solid lines), and laser beam (dash). Components listed in the figure are described in detail in the Experimental Section.

incidence. The comparison experiments using dried-droplet deposition were carried out with a Proflex TOF MS (Bruker Daltonics, Inc., Billerica, MA), which was equipped with a nitrogen laser (337 nm) and delayed extraction circuit. Experimental Control and Data Acquisition. The repetition rate was set with the internal laser trigger, and a program in the DOS mode was used for data acquisition. The “optosync” output signal from the laser triggered the oscilloscope and, if needed, the digital delay generator in order to generate high-voltage pulses for delayed extraction. Vacuum Deposition Interface for On-Line MALDI-TOF MS. Figure 2 presents a diagram of the second-generation interface with a cartridge containing ∼80-m Mylar tape that served as the moving surface for deposition of solutions. The Mylar tape (4.0-mm-wide and 25-µm-thick polyester) was kindly provided by Maxell Corp. (Fair Lawn, NJ) and cleaned with 0.1% TFA in 50% (v/v) methanol solution prior to use. A geared stepper motor (model ABS, Hurst, Princeton, IN), placed outside the chamber, propelled the tape via a shaft (6.35-mm diameter) in a manner similar to that of an audiocassette. The motor typically rotated at 90 steps/s, yielding a tape velocity of 1.0 mm/s; one step of the motor translated into an 11-µm movement of the tape. Compared to the previous design,24 the repeller was modified to accommodate an aluminum tape guide in its center. The guide was attached to the cartridge via a Delrin block with a spring to press the guide against the repeller. This design allowed exchange of the tape in the cartridge without removal of the repeller. Furthermore, the structure ensured the proper alignment of the guide face with the repeller, minimized transfer of vibrations from the motor to the repeller, and electrically insulated the repeller from the cartridge.

Figure 2. Top view of the vacuum deposition interface with the tape cartridge and the liquid junction. The design of the repeller region is also shown. Details can be found in the Experimental Section.

Initial tests of this interface were carried out with the capillary probe described previously.24 The tubular probe, 9.5-mm o.d., 40mm length, was inserted into the source chamber via quick coupling such that the end of the capillary was slightly bent while touching the tape. A mixed solution of analyte and matrix was deposited via a tapered fused-silica capillary, 20-µm i.d., 150-µm o.d., and 12.0-cm length (Polymicro Technologies, Phoenix, AR). The position of the sample trace was determined by scanning the laser beam across the tape using the mirror and monitoring the ion signal. Data acquisition was started ∼130 s after sample introduction, corresponding to the sum of the time required for Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

4787

the sample transport through the infusion capillary (10 s) plus that for the tape transfer between the deposition and the desorption regions (120 s at 1 mm/s). Capillary Electrophoresis. Capillary electrophoresis was performed using 75-µm-i.d. and 375-µm-o.d. fused-silica capillary (length is specified in text) from Polymicro Technologies, covalently coated with poly(vinyl alcohol) (PVA)28 to eliminate electroosmotic flow. In all experiments, the separation and infusion capillaries were connected to a liquid junction29,30 so that the outlet of the CE column and the inlet of the infusion capillary were spaced ∼75 µm apart; see Figure 2. The liquid junction, anodic reservoir and separation capillary were all filled with 10 mM solution of R-cyano-4-hydroxycinnamic acid (RCHCA) in 50% (v/ v) methanol/water, which served as MALDI matrix as well as CE background electrolyte. The sample was injected either hydrodynamically or electrokinetically, and electrophoresis was driven at 500-2000 V/cm by a high-voltage power supply (model PS/ EH30, Glassman, Whitehouse Station, NJ). Chemicals. R-Cyano-4-hydroxycinnamic acid (Sigma Chemical Co., St. Louis, MO) was recrystallized in methanol prior to use. Angiotensins, angiotensinogens, and other peptide standards were all purchased from Sigma and made up as 1 mg/mL stock solutions in water. 1,4-Dithio-DL-threitol (DTT) and iodoacetic acid (IAA) were purchased from Pierce Chemical Co. (Rockford, IL). rabbit enolase was from Boehring Mannheim Corp. (Indianapolis, IN), and HPLC grade methanol and acetonitrile (ACN) were from Fisher Scientific (Fair Lawn, NJ). Deionized water was produced with Alpha-Q system (Millipore Corp., Marlborough, MA). For the direct infusion experiments, the sample solutions were prepared by addition of a small volume of diluted peptide stock solution to the matrix solution. Enzymatic Digestion. Enolase was digested using a standard protocol.31 Briefly, the protein was dissolved and denatured in a solution of 50 mM NH4HCO3 and 8 M urea (pH 8.0), reduced by DTT, alkylated by IAA, and digested overnight with trypsin in a ratio of 1:20 (w/w) at 37 °C. After digestion, excess salt was removed using ZipTipC18 (Millipore Corp., Bedfored, MA) prior to analysis. Peptides were eluted from ZipTipC18 into vials either by single or multistep elution, using different percentages of ACN. Data Evaluation. For the ion suppression study, all MALDI MS operation parameters (laser attenuation, accelerating voltage, etc.) were held constant within one set of measurements. When the laboratory-built TOF MS and the vacuum deposition method were used, 21 average spectra were collected from each sample trace. Each average spectrum was the sum of 50 single-shot spectra. When the Proflex TOF MS and the dried-droplet method were used, spectra were collected from three different targets, with seven evenly spaced spots on each of the positions to ensure representative sampling. At each spot, 50 consecutive spectra were summed to reduce the effect of shot-to-shot variation and improve signal-to-noise ratio. In all cases, the absolute, baseline-corrected, peak intensities of the peptide signals were used for statistical analysis. (28) Karger, B. L.; Goetzinger, W. U.S. Patent 5,840,388, 1998. (29) Lee, E. D.; Muck, W.; Henion, J. D.; Covey, T. R. Biomed., Environ. Mass Spectrom. 1989, 18, 844-850. (30) Foret, F.; Zhou, H. H.; Gangel, E.; Karger, B. L. Electrophoresis 2000, 21, 1363-1371. (31) Stone, K. L.; Williams, K. R. In The Protein Protocols Handbook; Walker, J. M., Ed.; Humana Press: Totowa, NJ., 1996.

4788

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

RESULTS AND DISCUSSION Performance of Vacuum Deposition Interface. Design of the Vacuum Deposition Interface. In the original design of the interface, a sample solution was deposited on a rotating quartz wheel.24 The wheel transported the solutes away from the tip and acted as a heat reservoir, preventing the solvent in the capillary outlet from freezing. However, because laser desorption was not found to remove completely the deposited sample, the surface of the wheel was manually cleaned after each analysis. One possible remedy might be on-line cleaning of the wheel by ablating the sample remnants with a higher energy laser beam after the desorption region. However, a simpler approach was chosen for this work, which was to replace the deposition wheel with a disposable moving surface. Several different kinds of polymer materials, including Kapton (polyimide), Nylon, regenerated cellulose, and Mylar (polyester) were tested as deposition substrate. Mylar was chosen because of its chemical inertness and availability in a tape format. A roll of Mylar tape, 4 mm wide, 80 m long, and 25 µm thick, lasted for ∼1 day of continuous operation at a typical tape velocity 1 mm/s. The period of continuous operation may be easily extended to days by slowing the tape speed and/or using longer tape. The tape could be exchanged in 5-10 min, which included the pump-down time. It was found that, for deposition on the Mylar tape, the content of methanol in water-based solutions needed to be higher than 30% (v/v) in order to prevent freezing of the solution at the capillary tip, while pure aqueous solutions could be deposited on the wheel. This difference between deposition behavior was probably related to the properties of the surface of the wheel and the tape. Because the surface of the tape was smoother and more hydrophobic than that of the quartz wheel, the solution may not transfer from the tip to the tape as efficiently. Hence, vaporization of the remnants of the solution from the tip could cause freezing. To eliminate the chance of freezing, all infused solutions were prepared in 50% (v/v) methanol. Utilization of solvent mixtures with a high content of an organic solvent is common in conventional sample preparation for MALDI MS. Signal Reproducibility. To determine spot-to-spot reproducibility of the signal under the vacuum deposition condition, a mixed 50% (v/v) methanol solution of 1 µM angiotensin I and 10 mM RCHCA matrix was deposited on the tape moving at 1 mm/s for ∼30 s. Then, 1000 single-shot spectra were collected at a laser repetition rate of 10 Hz from the tape advancing at 0.11 mm/s, i.e., 1 spectrum per 11-µm-long segment. The relative standard deviation of peak area of the analyte ion (m/z ) 1296.7) was 15%, a small variation compared to that obtained from a conventional MALDI sample (RSD 20-90%).32 It should be noted that the peptide signal was present in every single-shot spectrum in contrast to most conventional MALDI experiments, thus eliminating the need for searching for a “sweet spot”. The high signal reproducibility could be attributed to (1) fast evaporation, which led to more homogeneous sample incorporation, and (2) full coverage of the laser desorption spot of the deposited trace, which ensured the averaging of the signal from positions across the trace during each shot. (32) Axelsson, J.; Hoberg, A. M.; Waterson, C.; Myatt, P.; Shield, G. L.; Varney, J.; Haddleton, D. M.; Derrick, P. J. Rapid. Commun. Mass Spectrom. 1997, 11, 209-213.

Although the above experiments were carried out under high vacuum (∼1 × 10-6 Torr), it should be emphasized that similar homogeneous sample trace and stable signals were also obtained under rough vacuum (∼1 Torr). On the other hand, atmospheric deposition was found to produce discrete droplets and a discontinuous signal. As long as pressure in the deposition chamber is significantly lower than the vapor pressure of the solvent, the vaporization of solvent will be a fast, nonequilibrium process. Offline vacuum deposition on a standard stainless steel MALDI target positioned on an x-y stage is currently being developed in our laboratory.26 Mass Resolution. To improve the mass resolution, circuitry for delayed extraction was incorporated into the homemade instrument design for MALDI-TOF MS. To investigate proper operation of the instrument in the delayed extraction mode, solutions of 1 µM bovine insulin or angiotensin III and 10 mM RCHCA in 50% (v/v) methanol were deposited on the tape moving at 1 mm/s. Mass resolution values (fwhm) of 1000 and 2000 were obtained for 10-shot average mass spectra of bovine insulin (MW ) 7534.4, delay 710 ns) and angiotensin III (MW ) 931.1, delay 260 ns), respectively. Mass accuracy of 100 ppm was achieved for small peptides using internal calibration. These results were comparable to values reported elsewhere33-36 and could be further improved with a reflectron. It should also be emphasized that, using the same TOF instrument, the values of resolution and mass accuracy obtained for the samples deposited on the tape were similar to those obtained from a conventional metal target; i.e., desorption from the tape did not reduce mass resolution. Limit of Detection. To estimate the detection limit of small peptides, a solution of angiotensin I with 10 mM RCHCA matrix and 25 mM ammonium sulfate (see later) in 50% (v/v) methanol was deposited on the tape moving at 1 mm/s for ∼10 s. The concentration of the peptide in the infused solution varied from 0, 1 × 10-8, 1 × 10-7, to 1 × 10-6 M. The capillary was washed with methanol for ∼30 s between infusion of the sample solutions. Figure 3 shows four 10-shot average mass spectra, which were recorded from a single ∼100-µm-long segment of the trace for all four concentrations using delayed extraction (delay, 260 ns). The peaks corresponding to angiotensin I (m/z ) 1296.5) are marked as [M + H]+. On the basis of the solution flow rate (120 nL/min) and the tape velocity (1 mm/s), the amounts of the peptide consumed to generate the spectra were calculated as 2 (S/N ) 4), 20, and 200 amol for the peptide concentrations 1 × 10-8, 1 × 10-7, and 1 × 10-6 M, respectively. Although MALDI-TOF MS sensitivity varies from peptide to peptide, it can be concluded thatsusing vacuum deposition on the tapesthe detection limits for small peptides are in the low-nanomolar range. The detection limits of the present system can be further lowered using solidphase extraction and/or electrofocusing for sample enrichment, prior to deposition. Compared to the dried-droplet method, vacuum deposition was found to be advantageous for improving MALDI MS sensitivity (33) Colby, S. M.; King, T. B.; Reilly, J. P. Rapid Commun. Mass Spectrom. 1994, 6, 865-868. (34) Whittal, R. M.; Li, L. Anal. Chem. 1995, 67, 1950-1954. (35) Brown, R. S.; Lennon, J. J. Anal. Chem. 1995, 67, 1998-2003. (36) Juhasz, P.; Roskey, M. T.; Smirnov, I. P.; Haff, L. A.; Vestal, M. L.; Martin, S. A. Anal. Chem. 1996, 68, 941-946.

Figure 3. Ten-shot average of MALDI mass spectra of angiotensin I with RCHCA as matrix and ammonium sulfate as additive from a single spot of the trace deposited under vacuum. Concentrations of the peptide in the infused solutions and the amount consumed by laser shots are shown in the graphs. Spectra were acquired on a laboratory-built linear MALDI-TOF mass spectrometer.

for the following reasons: (1) Vacuum deposition yielded a ∼60µm-wide sample trace, and thus the full width of the sample trace could be encompassed with an ∼100-µm-wide laser spot. (2) The trace formed under vacuum deposition conditions was in an amorphous or microcrystalline state, at least 100 times smaller crystals than in the case of the dried-droplet method.24 Smaller crystals have a larger surface area, leading to more useful embedding sites for analyte molecules, thus potentially improving MALDI sensitivity. Vacuum deposition was found to be beneficial for improving sample homogeneity and detection sensitivity, but on the other hand, more alkali metal adducts are generally formed than with the dried-droplet method. Enhanced adduct formation was also reported for other fast-evaporation methods.4,37 In the dried-droplet approach, analyte and common contaminants, such as alkali metals, may accumulate in different regions of the deposited samples. With vacuum deposition, all sample components, including the contaminants, are evenly embedded in matrix. Thus, the contaminants are in close proximity to the analyte in the deposited trace. To obtain spectra with minimized appearance of alkali metal adducts in the vacuum deposition approach, special precautions need to be taken to remove alkali ions from the sample, matrix, and deposition surface. A variety of ammonium salts were tested as matrix additive,38 and it was found that the addition of 25 mM ammonium sulfate effectively eliminated adduct formation (as seen in Figure 3). In summary, the performance of the cartridge interface was similar to that of the original interface,24 with the advantages of easy operation and extended analysis operation time. Influence of the Sample-Matrix Crystallization Rate on Ion Suppression. The rate of matrix crystal growth has been (37) Hensel, R.; King, R. C.; Owens, K. G. Rapid. Commun. Mass Spectrom. 1997, 11, 1785-1793. (38) Pieles, U.; Zurcher, W.; Schar, M.; Moser, H. E. Nucleic Acids Res. 1993, 21, 3191-3196.

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

4789

Table 1. d50a Obtained for Various Binary Peptide Mixtures as a Function of Deposition Method sample deposition methods suppressed peptide peptide T (hydrophilic) Icaria chemotactic peptide (hydrophobic) Icaria chemotactic peptide (hydrophobic) angiotensin I (relatively hydrophobic) acidic peptide basic peptide

suppressing peptide

dried vacuum droplet deposition

Icaria chemotactic peptide 0.5 (hydrophobic) peptide T >100 (hydrophilic) PHE >100 (hydrophobic) angiotensin II 20 (hydrophilic) basic peptide 1 acidic peptide 2

6 >100 >100 40 1 2

a d , molar ratio of suppressing peptide to the suppressed one 50 leading to 50% decrease in the signal intensity of the suppressed peptide.

reported to influence the relative intensity of peptide signals.4 As discussed previously, compared to the dried-droplet method, the crystallization rate during vacuum deposition was much faster, resulting in a more homogeneous sample trace and improved signal reproducibility. To characterize further the vacuum deposition interface and investigate the influence of vacuum deposition on the degree of ion suppression, several samples of model binary peptide mixtures were prepared, and the resulting MALDI mass spectra, using vacuum deposition and the dried-droplet method, were compared. Each peptide pair included peptides with different properties and various concentration ratios. The coefficient d50, described in ref 5, was employed to compare the results for different mixtures, where d50 ) n50/ns. n50 is the molar amount of suppressing peptide that produces a 50% decrease in signal intensity of the suppressed peptide and ns that of the suppressed peptide. The smaller the molar ratio d50, the stronger would be the observed ion suppression. The values of d50 of different binary peptide mixtures are summarized in Table 1. In the following, results obtained for selected peptide systems are discussed in detail. Peptides with Different Hydrophobicities. The signal suppression of a hydrophilic peptide, peptide T (ASTTTNYT), by a hydrophobic peptide, Icaria chemotactic peptide (IVPFLGPLLGLLT-NH2), was investigated. Samples were prepared from mixtures containing a fixed concentration of peptide T (5 µM) and varying concentrations of Icaria chemotactic peptide (0-100 µM) using the drieddroplet and vacuum deposition methods. As shown in Table 1, the presence of the hydrophobic peptide dramatically reduced signal intensity of the hydrophilic species, similar to that reported elsewhere.5 Using the dried-droplet method, a 1:2 molar ratio of Icaria chemotactic peptide to peptide T led to a 50% suppression of the signal of peptide T, while a 4-fold excess of Icaria chemotactic peptide completely quenched the signal of peptide T. Interestingly, using vacuum deposition, the molar ratio of the same peptides needed for 50% suppression and complete quenching was 6 and 20, respectively. An experiment with Icaria chemotactic peptide present at 5 µM concentration and peptide T at variable concentrations was also carried out. It was found that, irrespective of the sample application methods, the signal intensity of the hydrophobic peptide was constant, even with a 100-fold excess of peptide T. 4790

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

Assuming the matrix provides only a limited number of useful sites for desorption, these phenomena can be explained as follows. A more hydrophobic peptide is preferentially incorporated into the crystal structure of the hydrophobic matrix than a hydrophilic analyte.5 In the case of the dried-droplet preparation, slow crystallization allowed incorporation of the hydrophobic species, leading to suppression of the signal of the hydrophilic peptide. In addition, the peptide mixture was nonuniformly distributed according to hydrophobicity (Marangoni effect).7 On the other hand, extremely fast solvent vaporization during the vacuum deposition step incorporated both species into the matrix crystals to form a more homogeneous sample-matrix mixture, and suppression was less dramatic. Even if the distribution of peptides across the trace were not uniform, the resulting signal discrimination would be minimized because the laser irradiated the entire width of the sample trace deposited under vacuum.24 The ion suppression behavior of peptides with similar hydrophobicities was also investigated. Angiotensin I was estimated to be slightly more hydrophobic than that of angiotensin II.39 However, as shown in Table 1, angiotensin II could still suppress the signal of angiotensin I by 50% if the former was added in large excess, i.e., ∼20-fold greater for dried-droplet preparation and ∼40fold greater for vacuum deposition. Here again, due to the fast sample crystallization, suppression was found to be less significant in the vacuum deposition mode. Peptide Mixtures with Significantly Different Acid-Base Properties. A highly acidic peptide (LEEEEEAYGWMDF-NH2) and a highly basic peptide (RKRARKE) were chosen to test the mutual ion suppression between the acidic and basic peptides. In one group of experiments, acidic peptide (5 µM) was mixed with increasing amounts of the basic peptide. As shown in Table 1, a 2-fold excess of the strong basic peptide suppressed the signal of acidic peptide by 50%. Also, as seen in Table 1, very similar suppression profiles were recorded for the vacuum deposition and the dried-droplet methods. This is not surprising because the basic side-chain could be protonated more easily than the acidic peptide and, thus, preferentially ionized.5 Interestingly, when inverse experiments were carried out (i.e., excess of the acidic peptide relative to the basic peptide), it was found that the acidic peptide could suppress the signal of the basic one in both deposition modes. This is an unexpected observation, since the basic peptide is a stronger base than either the acidic peptide or the matrix, and it should thus be preferentially protonated. This suggests that an additional mechanism may take place, such as the formation of ion pairs among oppositely charged peptides that hampers charge retention of either species. CE-MALDI-TOF MS. CE Separation Efficiency. Seven angiotensins, listed in Table 2, were electrokinetically injected into the separation capillary at 50 V/cm for 5 s, and CE-MS analysis was carried out at 500 V/cm, using the liquid junction interface described previously.24 The resulting MS electropherogram of the angiotensin mixture in Figure 4 shows peaks of seven peptides, [M + H]+, accompanied by satelite peaks of alkali metal adducts, [M + Na]+, [M + K]+, etc. High plate counts (250 000-400 000 plates/m) and narrow peak half-widths (2-3 s) were observed; see Table 2. The plate counts suggest little or no extracolumn peak broadening resulting from the incorporation of either the (39) Kyte, J.; Doolittle, R. F. J. Mol. Biol. 1982, 157, 105-132.

Table 2. Efficiency Values Calculated from CE-MALDI MS Analysis of a Model Peptide Mixture analyte

peptide

m/z

t (min)

W1/2 (s)

N (m-1)

1 2 3 4 5 6 7

angiotensin I, goosefish (AIG) angiotensin III, human (AIIIH) angiotensin I, human (AIH) angiotensin I, salmon (AIS) angiotensin II, frag 1-7 (AF17) angiotensin II, human (AIIH) angiotensin II, frag 3-8 (AF38)

1281.5 931.1 1296.5 1258.4 899.0 1046.2 774.9

3.24 3.43 3.63 3.82 3.99 4.05 4.33

2.4 2.4 2.2 2.2 2.4 2.4 2.5

250 000 270 000 360 000 370 000 370 000 380 000 400 000

liquid junction or vacuum deposition interface. The high performance of the interface can be attributed to a number of factors. Inside the liquid junction, the sample zones were hydrodynamically focused into the center of the infusion capillary.30 The laminar flow of analyte and matrix passed through the infusion capillary within a few seconds (∼10 s), and the mixture was deposited on the moving tape under the vacuum of the MALDI source chamber.24 Rapid evaporation of the solvent (water and methanol) resulted in instantaneous formation of a thin, ∼60-µm-wide sample trace, eliminating any possible analyte diffusion. The liquid junction also provided a stable electrical contact for CE separation and acted as a reservoir for the matrix solution. Minimization of Ion Suppression in a Model Peptide Mixture. To verify the benefits of the additional separation step prior to MALDI MS analysis, comparison experiments using model peptide mixtures were carried out. First, the mixture of six angiotensins (1 µM each, dissolved in matrix solution) was deposited using the vacuum interface and analyzed by MALDI MS. The corresponding mass spectra revealed similar intensities of all angiotensin ions (the upper trace in Figure 5a), confirming that no ion suppression occurred. The same mixture was then spiked 50-fold with one of the components (angiotensin II). This mixture was reanalyzed, and it was found that the intensities of all angiotensins present at the lower concentrations were significantly reduced, with some of the signals completely undetectable (the lower trace in Figure 5a) due to the excess of angiotensin II. Finally, the mixture spiked with angiotensin II was separated by CE, and the analytes were deposited using the vacuum interface and analyzed. The comparison between the spectra obtained with (a) vacuum deposition of the unseparated mixture and (b) CEMALDI MS analysis is shown in Figure 5a and b, respectively. It is clear from the MS-electropherogram in Figure 5b that all angiotensins can be easily detected and identified after separation. Multiple peaks marked with asterisks correspond to clusters of matrix and alkali metal ions, indicating that alkali metal ions present in the sample were separated from the peptides. The importance of separation prior to MALDI-TOF MS analysis can be extended to protein mixture analysis. It is well known that many spots on 2-D gels contain more than one protein. Furthermore, the concentrations of the proteins in a given spot are generally different, and some peptides originating from a highly expressed protein may suppress the ion signal of other components. Under such circumstances, a separation step prior to MALDI MS analysis can preserve the signal intensity of the peptides that originated from less expressed protein and thus improve the dynamic range of protein digest analysis. Additionally, separation can isolate highly abundant low molecular weight

peptides, thereby preventing the TOF detector from saturation10 and further improving the dynamic range and sensitivity of MALDI MS analysis. Enolase Digest Analysis. Analysis of a protein digest usually starts with sample desalting and preconcentration by SPE, followed by MALDI MS analysis.40 To investigate the influence of separation prior to MALDI MS analysis on the sequence coverage of a protein digest, single-step and multistep SPE elutions as well as CE separation coupled to MALDI MS analysis were compared. The tryptic digest of rabbit β-enolase (MW ) 47 000), one of the glycolytic pathway enzymes, was chosen as a model complex protein. Digestion was conducted according to a standard protocol.31 First, the single-step SPE method was evaluated. The enolase digest was loaded onto ZipTipC18 and eluted with 50% ACN, as reported elsewhere.41 The eluted sample was mixed with matrix solution and deposited using the dried-droplet method. The resulting mass spectrum is shown in the upper trace in Figure 6a. Eighteen peptide ion masses were obtained, and using the peptide-mass search algorithm MS-Fit, the protein was correctly identified as β-enolase (SwissProt: ENOB•RABIT) with a sequence coverage of 45%. Similar sequence coverage was obtained when the same sample was deposited using the vacuum deposition interface; see Table 3. To increase the sequence coverage, the peptide mixture was fractionated by multistep SPE elution. The enolase digest was loaded onto the ZipTipC18 and eluted consecutively with three ACN/H2O solvent mixtures (15, 20, and 50% ACN).41 The fractions from each elution step were collected and analyzed separately using the dried-droplet method. The mass spectra corresponding to individual fractions are shown in Figure 6a. As expected, shorter peptides (MW ) 500-1200) were preferentially extracted using solvent with the lowest percentage of ACN. For the extraction of higher molecular weight peptides (1200-3500), solvent with a higher ACN percentage was required. The database search was performed using peptide mass values obtained from all SPE fractions. The number of identified peptides increased from 16 to 22, resulting in a sequence coverage improvement from 45 to 52%. Finally, the MALDI MS analysis was performed after CE separation of the digest mixture. The enolase digest was first desalted using ZipTipC18 (50% ACN was used for elution) and then hydrodynamically injected into the separation capillary. The analysis was carried out under the condition specified in Figure (40) Gobom, J.; Nordhoff, E.; Mirgorodskaya, E.; Ekman, R.; Roepstorff, P. J. Mass Spectrom. 1999, 34, 105-116. (41) Hornshaw, M.; Parker, K.; Sheer, D. G. Proceedings of the 46th ASMS Conference on Mass Spectrometry and Allied Topics, Orlando, FL, May 31June 4, 1998; p 672.

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

4791

Figure 4. High-efficiency CE-MALDI MS of 1 µM angiotensin mixtures, as listed in Table 2. (a) MS-electropherogram; (b) selected ion electropherograms. CE conditions: electrokinetic sample injection at 50 V/cm for 5 s, separation at 500 V/cm. PVA-coated capillary 75-µm i.d., 375-µm o.d., and 15-cm length. Electrolyte: 10 mM RCHCA in 50% MeOH. Liquid junction was also filled with CE electrolyte. MS: laboratorybuilt linear MALDI-TOF mass spectrometer. 4792

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

Figure 5. Suppression of MALDI MS signal among angiotensins. (a) MALDI MS analysis of unseparated angiotensin mixture. Upper trace, 1 µM concentration for each angiotensin (m/z and abbreviations see in Table 2; AG 1-14 is angiotensinogen m/z ) 1760.0); lower trace, 50 µM angiotensin II, 1 µM other angiotensins. Samples were deposited under vacuum. (b) CE-MALDI MS for recovering the signal of suppressed peptides. Analytes: 50 µM angiotensin II and 1 µM other angiotensins. Experimental conditions are the same as in Figure 4, except for electrokinetic sample introduction at 100 V/cm for 2 s. CE separation was carried out in 15-cm-long PVA-coated capillary (50-µm i.d., 375-µm o.d.) at 2 kV/cm.

6b, with most of the peptides in the enolase digest zones being baseline separated within 8 min. By minimizing ion suppression, the number of identified peptides increased to 33, which corresponded to 65% sequence coverage. Six out of 39 peptide masses were not identified in the database search. To identify these peptides, MS/MS analysis would be necessary. It should also be mentioned that the current laboratory-built MALDI-TOF MS has lower sensitivity for peptide masses with molecular weight higher than 3000, which can affect the sequence coverage that can be achieved. However, our current result is higher than 55% coverage

reported previously,42 using LC-ESI/IT-MS with mass fingerprinting alone. To improve the sequence coverage, further optimization of the enzymatic digest process would be necessary. CONCLUSIONS An improved vacuum deposition interface, in which liquid samples were deposited on a disposable moving tape, enabling continuous analysis, was constructed and successfully tested. (42) Reid, G. E.; Rasmussen, R. K.; Dorow, D. S.; Simpson, R. J. Electrophoresis 1998, 19, 946-955.

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

4793

Figure 6. MALDI MS spectra of enolase digest. (a) MALDI MS spectra of enolase digest mixture after SPE sample preparation. Upper trace: sample desalted with ZipTipC18, 50% ACN elution. Lower traces: sample fractionated by SPE using 15, 20, and 50% ACN. Both experiments used 200 µM enolase digest samples. After SPE elution, samples were diluted in MALDI matrix (50 mM RCHCA in ACN/0.1% TFA 50/50) before dried-droplet deposition. Final enolase digest concentration, 10 µM. MS: Bruker Proflex MALDI/DE-mass spectrometry, 50-shot average. (b) 3D spectrum of enolase digest obtained by CE-MALDI MS. Sample: 200 µM enolase digest desalted with ZipTipC18, 50% ACN elution, sample diluted in water before CE injection. Same experimental conditions as indicated in Figure 4, except that 20 µM sample was hydrodynamically injected at 5 cm for 10 s (330 fmol). CE separation was carried out in 20-cm-long PVA-coated capillary (75-µm i.d., 375-µm o.d.) at 1000 V/cm.

Vacuum deposition resulted in a narrow sample trace, improved signal reproducibility, and low-nanomolar concentration detection limit. Using this interface design, high-efficiency CE separation could be performed. It should be noted that capillary LC has also been successfully interfaced to MALDI MS using a similar liquid junction (data not shown). CE-MALDI MS was found to be a highly efficient way to minimize ion suppression in peptide mixtures. This method can 4794 Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

improve sequence coverage of protein digests and facilitate the identification of posttranslational modifications. At the same time, CE-MALDI MS enables the detection of the signal of peptides present at low levels and thereby improves the dynamic range for protein mixture analysis. In addition, separation can provide secondary information such as the mobility (CE) or hydrophobicity (LC) of analytes and facilitate ISF, PSD, and MS/MS analysis. Although on-line coupling of CE with MALDI MS was demon-

Table 3. Sequence Coverage of Enolase Digest Obtained by Different MALDI MS Analysis Methods sample preparation

no. of masses observed

no. of masses identified

sequence coverage

unseparated mixture dried droplet vacuum deposition SPE fractionation CE-MALDI MS

18 21 25 39

16 18 22 33

45% 44% 52% 65%

database searched: SwissProt.r03.07.99 peptide mass tolerance: ( 1Da searching tools: MS-Fit

strated in this work, the same principles can be applied to the off-line mode with subsequent analysis using a commercial MALDI MS instrument. The results from this latter work, as well as

multiplex CE-MALDI MS using a capillary array for enhanced sample throughput, will be reported separately. ACKNOWLEDGMENT The authors acknowledge NIH for support of this work (Grant HG02033) and Dr. Frantisek Foret for his ideas and fruitful discussions. Contribution 777 from the Barnett Institute. SUPPORTING INFORMATION AVAILABLE The detailed data obtained from the analysis of enolase digest. This material is available free of charge via the Internet at http:// pubs.acs.org. Received for review May 22, 2000. Accepted August 15, 2000. AC0005870

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

4795