Anal. Chem. 1995, 67, 1477- 1480
Capillary Electrophoresis-Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry for Direct Analysis of Cellular Proteins Steven A. Hofstadler,t Franklin D. Swanek,* David C. Gale,t Andrew 0. Ewing,*i* and Richard D. Smith*lt
Chemical Sciences Department, Pacific Northwest Laboratory, Richland, Washington 99352, and Department of Chemistry, The Pennsylvania State University, University Park, Pennsylvania 16802
The combination of capillary electrophoresis (CE) with electrospray ionization (ESI) mass spectrometry has proven to be broadly applicable to a wide range of biologicallyimportant compounds. When combined with Fourier transform ion cyclotron resonance (FIICR) mass spectrometry, the combined method, in addition to highresolution separations, atrords high-resolution precision mass measurements for analytes separated &om complex mixtures. Direct chemical analysis of single cells has received considerable attention in recent years; the single cell approach provides a major step toward answering important questions in the field of cellular biochemistry. In this work we present preliminary results which demonstrate the feasibility of using the CE-ESI-FIICR combination as a high-performancedetection scheme for the analysis of cellular proteins acquired directly from small populations (Le., 5-10) of intact living cells. The human erythrocytewas chosen as a model system owing to its availability, relatively homogeneous composition, and thorough documentation of contents by previous researchers. In this work we demonstrate the on-line acquisition of high-resolution mass spectra (average resolution 2 45 000 fwhm)of both the a and the /3 chains of hemoglobin acquired from the injection of 10 human erythrocytes (corresponding to 4.5 fmol of hemoglobin). Given the extremely small volume of the human erythrocyte (typically 87 Wcell), the techniques implemented here should also be adaptable to the study of larger mammalian cell systems. As detection schemes for open tubular liquid chromatography
(OTLC)' and capillary electrophoresis (CE)2-4 have become increasingly sensitive, lower concentrations and smaller volumes of analyte have been examined, providing a significant degree of versatility and facilitatingthe study of more challengingbiological problems. Toward this end, chemical analysis of single cells has t Pacific Northwest Laboratory.
The Pennsylvania State University. (1) Oates, M. D.; Cooper, B. R; Jorgenson, J. W. Anal. Chem. 1990,62,15731577. (2) Engelhardt, H.; Beck, W.; Kohr, J.; Schmitt, T. Angeur. Chem., Int. Ed. Engl. 1993, 32, 629-766. (3) Smith, R D.; Udseth, H. R; Loo, J. A; Wright, B. W.; Ross, G. A. Talanfa, 1989,36, 161-169. (4) Smith, R D.; Wahl, J. H.; Goodlett, D. R; Hofstadler, S. A. Anal. Chem. 1993, 65, A574-A584. 0003-2700/95/0367-1477$9.00/0 0 1995 American Chemical Society
received considerable attention in recent year^;^-^ the ability to identify cellular constituentson a cell-tocellbasis provides a major step toward answering important questions in the field of cellular biochemistry. For example, Kristensen et a1.8 used capillary electrophoresis to directly identdy and measure the neurotransmitter dopamine in two vesicular compartments in a single nerve cell of the pond snail Planorbis comeus. Kennedy et al.5obtained amino acid profiles of 17 different amino acids of the land snail Helix aspersa through single cell analysis. The majority of detection schemes utilized to date for single cell analyses are based on electrochemicalgor ~ p t i c a l ' ~detection J~ codgurations. While these detection schemes afford detection limits in the low attomole range or below,'' they often provide little or no information to assist in component identification unless the elution times can be precisely correlated with previously analyzed standards. Additionally, many optically based detection schemes require analyte derivatization, which becomes increasingly difficult and inefficient because of the small volume of an individual ce1l.l' Recent results with on-line capillary electrophoresis (CE) electrospray ionization (ESI) Fourier transform ion cyclotron resonance (FTICR) mass spectr0metry~~-'4have suggested that the CE-ESI-FTICR combination may provide a near ideal approach for microsample analyses owing to the inherent sensitivity of the technique and the enhanced information content available from high-resolution and high-precision mass measurements. Electrospray ionization can provide extremely high sensitivity due to its high ionization efficiency under appropriate solution conditions and constitutes the most effective interface of CE with mass ~pectrometry.~ It has recently been shown that, when ESI is utilized in conjunction with multiple frequency sustained off-resonance irradiation (SORI),15it is possible to obtain Kennedy, R T.; Oats, M. D.; Cooper, B. R; Nickerson, B.; Jorgenson, J. W. Anal. Chem. 1989,61,436-441. Ewing, A. G. J. Neurosci. Methods 1993,48, 1872-1876. McAdoo, D. J.; Coggeshall, R E. J. Neurochem. 1976,26, 163-167. Kristensen, H. K; Lau, Y. L.; Ewing, A. G. J Neurosci. Methods 1994,51, 183-188. Mesaros, J. M.; Ewing, A. G.; Gavin, P. F. Anal. Chem. 1994,66, 527A537A. Hogan, B. L.; Yeung, E. S. Anal. Chem. 1992,64, 2841-2845. Lee, T. T.;Yeung, E. S. Anal. Chem. 1 9 9 2 , 64, 3045-3051. Hofstadler, S. A; Wahl, J. H.; Bruce, J. E.; Smith, R D. 1.Am. Chem. SOC. 1993, 115, 6983-6984. Smith, L. M. Science, 1993,262,530-532. Hofstadler, S. A; Wahl, J. H.; Bakhtiar, R; Anderson, G. A; Bruce, J. E.; Smith, R D. J Am. SOC.Mass Spectrom. 1994,5, 894-899.
Analytical Chemistry, Vol. 67, No. 8, April 15, 1995 1477
complete or partial sequence information from very small amounts of analyte in complex mixtures.14 In this paper we present preliminary results which demonstrate the feasibility of using the CE-ESI-FTICR combination as a high-performance detection scheme for direct analysis of cellular proteins acquired directly from small populations (Le., 5-20) of living cells. The human erythrocyte (red blood cell) has been selected as a model system for direct cellular analysis for several reasons;11J6-18 the chemical contents have been thoroughly documented by previous study,lgand freedom from attachment to other cells or tissue simplifies the isolation and injection process. (In addition, human erythrocytes are readily available at no cost, and a fresh sample is always near at hand.) Because the function of the human erythrocyte is to deliver oxygen to other cells, it is considerably smaller than most mammalian cells. As the erythrocyte matures, it expels its nucleus and mitochondria, and other organelles are Thus, the contents of the erythrocyte are unusually homogeneous. Nearly the entire volume of the cell is filled with hemoglobin, approximately 300 million molecules (=450 amol) per cell,2oa challenging but attainable level for mass spectrometric detection with current instrumentation. Even more challenging is the characterization of the -100 minor componentslgpresent in each erythrocyte. Given the extremely small volume of the human erythrocyte (typically 4 7 &/cell), the techniques implemented here should also be adaptable to the study of the preponderance of larger mammalian cell systems. A number of studies have been undertaken which involve somewhat larger cells which contain several orders of magnitude more material than the average erythrocyte. For example, a recent study by Cooper and co-workers21probed the variation in composition and amount of the catecholamines epinephrine and norepinephrine in individual bovine adrenomedullary cells. On average, each cell contained in excess of 150 fmol of catecholamine, and, due to considerable cell-tocellvariation, some cells contained >290fmol of epinephrine and norepinephrine. Thus, the ability to detect and mass analyze cellular components present even in the femtomole regime is of significant analytical utility. While a number of highly sensitive detection schemes have been employed for the direct analysis of cellular proteins, the results presented here represent the first example of high-performance mass spectrometric detection in conjunction with direct sampling of cellular components. EXPERIMENTAL SECTION Human erythrocytes were obtained from the plasma of a healthy adult male. Blood was drawn from the finger of a volunteer using a sterile lancet (Becton Dickenson, Rutherford, NJ). A small drop of blood was placed on a microscope slide and immediately diluted with 0.5-1 mL of Ca2+-free phosphatebuffered saline solution (PH 7.4) consisting of 150 mM NaCl, 4.2 mM KC1,2.7 mM MgC12,l mM NaH2P04,11.2 mM glucose, and 10 mM HEPES (4(2-hydroxyethyl)-l-piperazineethanesulfonic acid). The injection end of the capillary was attached to a (15) Gauthier, J. W.; Trautman,T. R Jacobson, D. B. Anul. Chim. Acta 1991, 246,211-225. (16) Rosenzweig, 2.; Yeung, E. S. Anul. Chem. 1994, 66, 1771-1776. (17) Xue, Q.;Yeung, E. S.J. Chromutogr. 1994,661,287-295. (18) Lee, T.T.;Yeung, E. S.J. Chmmufogr.1992,595,319-325. (19) Pennell, R B. In The Red Blood Celt Surgenor, D. M., Ed.; Academic Press: New York, 1974. (20) Curtis, H. Biology; Worth Publishers, Inc.: New York, 1983. (21) Cooper, B. R;Jankokwski, J. A Leszczyszyn, D. J.; Wightman, R M.; Jorgenson, J. W. Anul. Chem. 1992,64,691-694.
1478 Analytical Chemisfry, Vol. 67, No. 8, April 15, 1995
, A , Stereo Microscope
r-l
High
Voltage
Suspension
/
-
Micromanipulator
Figure 1. Cell acquisition configuration in which a micromanipulator is used to move the etched terminus of the CE capillary in a buffered cell suspension as observed under the microscope. Cells are injected by electroosmotic flow. Following cell acquisition, the terminus of the capillary is placed in the running buffer, which causes the cell@) to lyse due to osmotic shock.
micromanipulator (Carl Zeiss, Inc., Thornwood, NY) and positioned near a group of erythrocytes within the drop of saline solution on the microscope slide. The injection end of the capillary was prepared by removing approximately 0.5 cm of polyimide and etching with a 40%hydrofluoric acid solution (Aldrich, Milwaukee, WI). The electrospray end of the capillary was kept level with the microscope to prevent gravity-induced laminar flow during the cell injection procedure. An SV8 Stereomicroscope (Carl Zeiss, Inc.) was used to monitor cell injections, with the slide illuminated from the underside by an Intralux 6O00 illuminator (Volpi Manufacturing USA, Auburn, NY). Shown in Figure 1is a schematic of the arrangement utilized for cell acquisition. A small group of cells was observed under the microscope at 256x magnification,and the etched tip of the capillary was manipulated within a few micrometers of the cells. A potential of *-4 kV was applied to the drop of buffered cell suspensionvia a platinum wire connected to a high-voltage power supply (Spellman, Plainview, NY). Intact cells were drawn into the capillary by electroosmotic flow while being observed under the microscope. Total injection times ranged from 5 to 30 s. Following cell injection, the end of the capillary was placed in a vial containing the CE running buffer, 10 mM HOAc (PH 3.4). Upon exposure to the running buffer, the cell membrane is lysed due to osmotic shock, releasing the cellular contents for CE separationll and mass analysis. The small size of even a few erythrocytes prevents signiscant accumulation of inert materials on the capillary wall that would alter electroosmotic flow. The CE capillaries utilized in this study were chemically modified with aminopropylsilane12922.23to prevent protein absorp tion on the inner wall of the capillary and to provide sufficient electroosmotic flow to produce stable operation of the ESI source. The CE running buffer, while not providing ultrahigh separation efficiency for protein mixtures, was chosen on the basis of its compatibilitywith electrospray ionization; the aminopropylsilanecoated capillaries with 10 mM HOAc generate a relatively high electroosmotic flow which is well suited for the generation of a stable sheathless electrospray. Several diflerent capillary diameters (10,20, and 50 pm) were utilized in this study, with 20 pm (22) Bruin, G. L.M.; Huiden, R,Kraak, J. C.; Poppe, H.]. Chromufogr. 1989, 480,339-343. (23) W a c s , K. D.Diss. Absfr. Inf. 1983,44,3766.
found to provide the best overall compromise for cell acquisition, sensitivity, and stable electrospray formation. The capillary length 30 2251 for all experiments was 90 cm. The conductive CE capillary terminus was prepared by removing 1cm of the polyimide coating. The exposed fused silica was etched with a 40% hydrofluoric acid solution to taper the outlet of the analytical capillary. A gold conductive coating (Epoxy Technology,Bderica, MA) was applied 0 2 4 6 8 1 0 1 2 to the tapered CE terminus to provide an effective electrospray. Time (minutes) The conductive tip serves two purposes: it acts as a contact for 100 the application of the electrospray potential and it establishes electricalcontact at the anode of the CE capillary tenninus which defines the CE electric field, typically -200 V/cm. An advantage of the conductive tip interface over a liquid sheath-type ESI 20 .. .. .. . ...-... . interface is that a significantly lower background signal is 0 40 lo ge11erated.2~ The electrospray was produced using a $3.8 kV Time (Minutes) gradient between the CE capillary terminus and the heated 100 desolvation inlet capillary to the first stage of the mass spectrometer vacuum system. An sF6 coaxial sheath gas was utilized to prevent corona discharge and focus the electrospray toward the heated desolvation capillary inlet. The 7 T ESI-FTICR mass spectrometer utilized in these 2 studies has been described in considerable detail e l s e ~ h e r e ; ~ ~ 0 900 1000 1100 1200 1300 thus, only a brief description is given here. Ions are transferred from the modified Analytica Branford, CT) electrospray source d z to the trapped ion cell by two sets of radio frequency Q-only Figure 2. (a) Electrospray ion current entering the mass spectrometer from the injectionand in-column lysing of 20 human erythrocytes quadrupoles. Background pressure in the trapped ion cell is detected from the electrospray ionization current reaching the front maintained at 10-loTorr by a custom cryopumping assembly shutter of the mass spectrometer. Full scale represents 25 pA of consisting of two sets of cryobaffels with radiation shields which electrospray ion current. The first peak is due to salts and buffering are maintained at 77 and 14 K, respectively, by closed cycle agents from the cell suspension media. The broad peak at 7-10 min cryogenic compressors. A typical CEESI-FTICR pulse sequence is due to the eluting cellular constituents. (b) A reconstructed ion electropherogramof the (M + 17H)17+species of the hemoglobin a consists of five events: ion injection/accumulation, ion cooling/ chain (solid line) and the (M + 16H)16+species of the hemoglobin,8 pump down, rf excitation, detection, and data transfer.12 Broadchain (dashed line) demonstrates the broad, poorly resolved peaks band swept excitation over a 500 kHzbandwidth with a 85 Hz/,us observed when relativelylarge (Le., 20) cell populations are sampled. sweep rate was followed by detection of 256K data points at 303 The mass spectrum in (c) was acquired 7.3 min into the run and lrHz (low m/z = 690), resulting in a 836 ms time domain signal. demonstrates the presence of both hemoglobin chains. Trap potentials were maintained at 0.5 V during detection. The initial cell population has a signillcant spatial distribution, and Electronics and all aspects of data acquisition and processing were it is unlikely that all cells lysed simultaneously, leading to band controlled by an Odyssey (Extrel FTMS, Madison, WI) data dispersion greater than that due to the initial spatial distribution station running Odyssey version 2.0 software. of the cells. The noncovalent tetrameric hemoglobin complex RESULTS AND DISCUSSION consisting of two a chains and two B chains (a&) is dissociated Shown in Figure 2a is a total ion current (TIC) trace from the due to the low pH of the 10 mM HOAc CE running buffer. Lee injection of 20 erythrocytes derived from the detection of the and Yeung have reported lower separation efficiencies for comelectrospray ion current signal from the ion beam impinging on ponents extracted from single cells compared to those for the front shutter of the FTTCR instrument.26The initial peak separation of standards obtained from commercial sources.11 observed in the TIC trace is attributed to buffer ions from the While they rule out a variation in lysing time as the cause of the cell injection procedure. While a substantial current was observed zone broadening, it is likely that a transient Cl" preconcentration on the TIC, only a small number of salt and buffers ions were s t e p would signiscantly improve the separation efficiencyfor the observed in the mass spectrum, most likely due to the limited analysis of cell lysates as well, regardless of the zone broadening detection bandwidth employed which precludes observation of mechanism. SigniiGcant improvement in separation performance ions below m/z 690. Approximately 1min after the buffer peak will likely be realized when single cell injections are feasible, as eluted, a broad peak due to the major cell contents was observed overlapping peaks which result from the spatial and temporal from which mass spectra of the a and/or B chains of hemoglobin variations described above are avoided. Preliminary results were acquired. As demonstrated by the reconstructed ion indicate that the resolution of the separation is significantly electropherograms in Figure 2b, the eluting peaks are broad and improved when smaller cell populations are utilized. For example, poorly resolved due to variations in cell lysing and injection times. Figure 3 shows mass spectra in which initiation times for spectral acquisition were chosen on the basis of the magnitude of the ion (24) Wahl, J. H.; Gale, D.C.; Smith, R D.]. Chromutagr. A 1994,659,217222. current impinging on the front shutter of the mass spectrometer. (25) Wmger, B. E.; Hofstadler, S. A; Bruce, J. E.; Udseth, H. R; Smith, R D.]. As described this ion current-mediated acquisition Am. SOC.Mass Spectrom. 1993,4, 566-577.
R
: I,,A,, .
I
.
,
44lhLuL
(26) Wahl, J. H.; Hofstadler, S. A; Smith, R D.Anal. Chem. 1995,67, 462465.
(27) Thompson, T. J.; Foret, F.; Vouros, P.; Karger, B. L. Anal. Chem. 1993, 65,900-906.
Analytical Chemistry, Vol. 67, No. 8, April 15, 1995
1479
(a) 80.0-
60.0-
c 9
1 .o
L
loo 1 1100.0 1200.0 1300.0 1400.0 mh
80.0-
60.0-
40.0~
20.0-
I 1
resulting in an average resolution of only 45 OOO (apodized,fwhm). The data shown in Figure 3a demonstrate a signal-to-noise ratio in excess of 5009, which suggests that the analysis of an individual erythrocyte is well within reach with present instrumentation. It is not unexpected that the only cellular component detected was hemoglobin. The next most abundant protein in the erythrocyte, carbonic anhydrase, is present at only 737 amol/cell, which appears to be below the present detection limits obtained in our initial studies. To date we have been able to detect the hemoglobin from as few as five erythrocytes,which corresponds to an injection of x2.3 fmol. Acquisition of a mass spectrum involves a high-pressure ion injection event of 50-100 ms, a pump down delay of -2 s, and -500 ms for data acquisition. The resulting ionization duty cycle is in the range of 2-5%. The application of significantly longer high-pressure ion injection/accumulation intervals serves only to promote ion loss due to an increased rate of growth of magnetron motion. However, recent advances with quahpolar axializati~n~-~~ (a technique which converts magnetron motion into cyclotron motion which, at high pressure, results in a reaxialized ion ensemble) permit signiiicantly longer lived high-pressure events without increased ion loss. Thus, when the improved ionization duty cycle is combined with quadrupolar excitation during ion injection to accomplish ion we anticipate signiscant improvements in sensitivity.
I
I
““r
mlz
Figure 3. Single scan ESI-FTICR mass spectra of the (a) a chain and (b) /3 chain obtained from 10 human erythrocytes following incolumn lysing and CE separation (see text). Acquisition of the spectra in (a) and (b) were manually triggered on the basis of the ion current impinging on the front shutter of the mass spectrometer. The typical peak width of the eluting analyte was 1-2 min, depending on the number and spatial density of cells injected. The average resolution of the isotope peaks is 70 000 (fwhm, unapodized) and 45 000 (fwhm, apodized) and is data point limited. The /3 chain shown in (b) exhibits a lower signal-to-noise ratio than the a chain, possibly due to lower ionization efficiency of the B chain relative to the a chain (see ref 33).
scheme allows improved temporal overlap of the timevarying analyte ion current with the ion injection event; thus, spectra can be acquired when maximum ion current for each component is reaching the mass spectrometer. These spectra were acquired from the injection of an aggregate of 10 erythrocytes, which corresponds to x4.5 fmol of hemoglobin. As is evident from a comparison of parts a and b of Figure 3, the separation of the a and 9, chains is nearly complete in this case. The resulting mass spectrayield a data point-limitedresolution of 70 OOO (unapodized, fwhm); the spectra shown in Figure 3 have been apodized, (28) Schweikhard, L.;Guan, S. H.; Marshall, A G. Int. I. Mass Spectrom. Ion Processes 1992, 120, 71-83. (29) Guan, S. H.; Xiang, X 2.; Marshall, A G. Int. I. M a s Spectrom. Ion P~~cesses 1993,124, 53-67. (30) Speir, J. P.; Gorman, G. S.; Pitsenberger, C. C.; Turner, C. A; Wang, P.P.; Amster, I. J. Anal. Chem. 1993,65, 1746-1752. (31) Bruce, J. E.; Anderson, G. A; Hofstadler, S. A; Vanorden, S. L.; Sherman, M. S.;Rockwood,A L;Smith, R D. Rapid Commnn. Mass Spechom. 1993, 7,914-919. (32) Kaniansky, D.;Ivanyi, F.; Onuska, F. LAnaZ. Chem. 1994,66,1817-1824. (33) Shackleton, C. H. L;Witkowska, H. E. In Mass Spectrometry in the Characterization of Variant Hemoglobins; Desiderio, D. M., Ed.; Plenum Press: New York, 1994;135-199.
1480 Analytical Chemistry, Vol. 67, No. 8, April 15, 1995
CONCLUSIONS The present work demonstrates the feasibility of CE-ESIFTICR for the analysis of cellular proteins directly from small cell populations. Based upon the present results acquired with 5-20 cells, it is anticipated that these methods can be extended to the analysis of single cells. From these initial experiments, several areas of improvementhave been identified to be incorporated into future studies with the aim of single cell analysis. Key to these will be the incorporation of longer ion injection intervals with broad-band or selective quadrupolar cooling to provide improved sensitivity. Additionally, for experiments with small populations of cells, recent advances with CITP32and transient isotachophore sisZ7suggest that significant gains in sensitivity can be realized by prefocusing the analyte bands prior to separation and mass analysis. Thus, the application of a ClTP preconcentration step is expected to improve the resolution of the separation and assist in the analysis of cellular components which are present at very low levels. ACKNOWLEDGMENT The authors wish to thank Dr. Jon H. Wahl for helpful discussion and CE column preparation, the U.S. Department of Energy, and Laboratory Directed Research and Development of Paciiic Northwest Laboratory for support of this research. Pacific Northwest Laboratory is operated by Battelle Memorial Institute for the U.S. Department of Energy, through Contract No. DEAC0676RLO 1830. F.D.S. and A.G.E. acknowledge the National Science Foundation and the National Institutes of Health for financial support of this research. A.G.E. is a Camille and Henry Dreyfus Teacher-Scholar. Received for review August
IO, 1994.
Accepted February
2, 1995.m AC940795Q e Abstract
published in Advance ACS Abstracts, March 1, 1995.
