Articles Anal. Chem. 1995, 67, 3638-3642
Analysis of High=MassBiomolecules Using Electrostatic Fields and MatrixlAssisted Laser Desorption/lonization in a Fourier Transform Mass Spectrometer Jie Yao, Michael Dey, Salvador J. Pastor, and Charles L. Wilkins*
Department of Chemistty, University of Califomia, Riverside, Califomia 9252 1
A new deceleratingtechnique that places dc potentials on the orthogonal excitation and receiver plates as well as the rear trapping plate (conductance limit) of the source cell of a dual cubic cell has been applied to the standard matrix-assistedlaser desorptionhonization Fourier transform mass spectrometrytechnique. When this five-plate trapping method is applied, high-mass ions with large translational kinetic energies can be trapped efficiently and detected. Using this approach, low-resolution spectra of carbonic anhydrase (MW = 29 000),egg albumin (MW = 45 000), and bovine albumin (MW = 66 000) have been obtained. Because the new decelerating method requires no modification to the existing cell, it is also possible to obtain high-resolutionspectra for compounds with masses of ca. 14 000 Da and lower. Utilizing the five-platetrapping method, a bovine insulin spectrumwith a resolving power of 20 000 was obtained. It is not yet possible to obtain higher resolution for the higher mass proteins. The reasons for this difllculty are currently being investigated. A current limitation of metrix-assisted laser desorption/ ionization Fourier transform mass spectrometry (MALDI-FTMS) measurements is the inherent dficulty in trapping high-mass species. Due to the MALDI desorption process, molecules and ions are ejected from the matrix surface with kinetic energies that scale directly with their mass.' As a consequence, ions of higher mlz and accordingly higher kinetic energies are inefficiently trapped if standard low-voltageFITvlS trapping potentials are used. A number of researchers have devoted considerable effort to examining kinetic energy distributions and trapping efficiencies for ions with higher translational e n e r g i e ~ . ~The - ~ trapping problem was partially solved by introduction of a gated trapping (1) Beavis, R. C.; Chait, B. T. Chem. Phys. Lett. 1991,181, 479-484.
(2) Castoro, J. A.; Koster. C.; Wilkins, C. Rapid Commun. Mass Spectrom. 1992, 6,239-241. (3) Castoro, J . A.; Wilkins, C. Anal. Chem. 1993,65,2621-2627. (4) Nikolaev. E. N.; Mordehai, A. V.; Frankevich, V. E., Jr. Rapid Commun. Mass Spectrom. 1991,5,260-262. (5) Hettich, R. L.; Buchanan, M. V. Int. J. Mass Spectrom. Ion Processes 1991, I l l , 365-380.
3638 Analytical Chemistry, Vol. 67, No. 20, October 15, 1995
protocol in which a 9-V deceleration potential is applied to the rear trapping plate of a cubic cell to slow ions entering the cell. Using this method, Castoro and co-worker~~,~ successfully analyzed singly charged biomolecules with molecular masses as great as that of carbonic anhydrase (MW = 29 0o0). However, forhigher mass ions it was necessary to increase the trapping voltage and to average multiple spectra in order to improve the signal-to-noise ratio, because very few ions are trapped. Nikolaev and co-workers4utilized a method to trap low-mass ions scattered from a surface with kinetic energies up to 500 eV by simply biasing the sample relative to all the plates of the cell. Hettich and Buchananj varied the potential difference between the sample probe and the front trapping plate. These workers found that use of retarding fields lower than 10 V did not significantly affect trapping efficiencies. These studies suggested that it might be profitable to examine the use of higher decelerating potentials than had previously been used to trap MALDI-generated ions. Until recently, there was slow progress in addressing the problem of achieving high-mass analysis for singly charged ions by Fourier transform mass spectrometry, in spite of the fact that it is capable of providing extremely high resolving power and mass accuracy in the low part-per-million (gpm) range.6-g The majority of recent high-mass work has focused on the use of electrospray ionization @SI) to produce multiple charge states, allowing detection in a mlz range more suited to FTMS. These efforts have resulted in ultra-high-resolution mass spectral measurem e n t ~ . ' ~ -McLafferty '~ and co-workers'j reported isotopic resolu(6) Guan, S.; Marshall, A. G. Rapid Commun. Mass Spectrom. 1993,7,857860.
(7) Pastor, S. J.; Castoro, J. A; Wilkins, C. L. Anal. Chem. 1995,67,379-384. (8) McIver, R. T.; Li, Y.; Hunter, R. L. Rapid Commun. Mass Spectrom. 1994, 8. 237-241. (9)McIver, R. T.; Li, Y.; Hunter, R L. Int. J. Mass Spectrom. Ion Processes 1994, 132, Ll-L7. (10) Fenn, J. B.; Mann, M.; Meng, C. IC;Wong, S. F.; Whithouse, C. M. Science 1989,246,64-71. (11) Kebarle, P.; Tang, L. Anal. Chem. 1993,65,972A-986A. (12) Senko, M. W.; Beu, S. C.; McLafferty, F. W. Anal. Chem. 1994,66,415417. (13) Senko, M. W.; Speir, J. P.; McLafferty. F. W. Anal. Chem. 1994,66,28012808. (14) Winger, B. E.; Hofstadler, S. A.; Bruce, J. E.; Udseth, H. R.; Smith, R D.J. Am. SOC.Mass Spectrom. 1993,4,566-577.
0003-2700/95/0367-3638$9.00/0 0 1995 American Chemical Society
tion of ions derived from porcine albumin (MW 66 OOO). Hofstadler and Laude16 were able to detect ions corresponding to a dimer of bovine albumin (mlz 132 kDa). Recently, Smith and co-workers17detected individual T4 DNA ions (mass 110 MDa), demonstrating the capability of ESI-FTMS in the 108-Da mass range and representing the highest mass yet determinedby mass spectrometry. Although electrospray ionization has excellent mass range capabilities, ionization efficiency can be suppressed by solution reagents, such as detergents and salts, which are commonly used in bio~hemistry.~8J~ MAL,DI, the topic of this paper, represents an alternative and complementary approach to electrospray ionization of high-mass species. Generally, there are two types of MALDI ion generation sources. An external ion source creates ions outside the magnetic field, and the ions are subsequently guided into the cell by utilizing an ion guide or lenses. The primary advantage of an external ion source is that it facilitates generation of a collimated beam during the ion transfer process. On the other hand, ions generated directly within the magnetic field (internal ion source) are available for immediate analysis. Internal sources avoid possible problems that may arise during ion transmission from an external ion source, where molecules can experience mass discrimination due to flight-time effects, as well as increased metastable decay. The main drawback of internal MALDI ion generation is the large spatial distribution of ions of the same mlz ratio, making ion coherence for excitationldetection less than optimal. There is continued research interest in h d i n g ways to “axialize”or cool the ions withii the cell to a better collimated starting position. Such an approach combines the simplicity of internal ion generation with the advantage of a collimated beam characteristic of an external source. The theoretical upper mass limit for thermal ions in a Fourier transform mass spectrometer is high and has been approximated to be at least 300 OOO Da in a 7-T FTMS.~OI~~ In a recent paper, Solouki, Gillig, and demonstrated the feasibilityof highmass FTMS analysis. In their research, they positioned a copper wire ion guide parallel to the magnetic field lines within a cylindrical cell and reported success in detecting singly charged biomolecule ions with masses as great as 157 kDa (transferrin dimer). Use of the copper wire guide poses some problems because it alters the electric field experienced by the ions. Positive ions attracted to a negatively biased ion guide must be repelled by a positive voltage before subsequent dipolar excitation. As a consequence of the changing electric field and corresponding shifts in frequency, accurate mass calibration is difficult. Reported here is a possible approach to further extend the detectable mass range of singly charged ions using MALDI-ETMS, employing an electrostatic field deceleration method, where decelerating voltages up to 75 V are applied not only to the rear trapping plate of the source region of a dual cubic cell but also to the excitation and receiver plates of the cubic cell. This “five (15) Speir, P. J.; Senko,M. W.; Little, D. P.; Loo, J. A; McLafferty, F.M.J. Mass Spectrom. 1995,30, 39-42. (16)Hofstadler, S.A;Laude, D. A, Jr. Anal. Chem. 1992,64, 569-572. (17)Chen, R;Cheng, X.; Mitchell, D. W.; Hofstadler, S. A; Wu,Q.;Rockwood, A L.; Sherman, M. G.; Smith, R D. Anal. Chem. 1995,67, 1159-1163. (18)Kay, I.; Mallet, A. I. Rapid Commun. Mass Spectrom. 1993, 7, 744-746. (19)Schindler, P.A;Van Dorsselaer, A Anal. Eiochem. 1993,213, 256-263. (20) Gross, M. L.; Rempel, D. L. Science 1984,226,261-268. (21)Wood, T.D.; Schweikhard, L;Marshall, A G.Ana1. Chem. 1992,64,1461-
1469. (22)Solouki, T.;Gillig, IC J.; Russell, D. H. Anal. Chem. 1994,66,1583-1587.
”
Figure 1. Five-plate trapping field lines calculated by MacSimion (Version 2.03, Montech Pty Ltd.) for 50 V on the rear trapping plate and 20 V on both the excitation (yz-plane) and receiver (xz-plane) plates. The magnetic field lies along the z-axis.
plate” trapping method shows improved trapping efficiency during the ionization event. Simulations (MacSimion, Version 2.03, Montech Pty Ltd., Monash University, Victoria, Australia) suggest that application of voltages to five of the six plates of the FTMS cell forms a “cuplike”electrostatic field in which ions can be more efficiently trapped. Simulated ion calculations show a much improved potential surface which should be more conducive to better trapping because off-axis ions are less likely to be ejected (neutralized on cell plates). Figure 1 is a simulation of the field map that results if 50 V is applied to the rear trapping plate and 20 V to both the excitation and receiver plates. The fiveplate trapping approach does not require cell modification. Instead, dc pulses and excitation pulses are separated in time and controlled by existing FTMS s o h a r e . This setup allows maximum flexibility and control for an FTMS experiment without impairing the ability to do lower mass high-resolution work. EXPERIMENTAL SECTION Instrumentation. Experiments were carried out using a Waters-Extrel (Madison, WI) ETMS2OOO Fourier transform mass spectrometer operating at 6.2 T and equipped with an automatic solids probe. The source and analyzer cells of the cubic dual cell are separated by a conductance limit with a 2-mm-diameter hole. A Lambda Physik EMG201-MSC excimer laser (Lambda Physik, Gottingen, Germany) operating at a wavelength of 308 nm with a pulse width of 28 ns was used. The laser beam enters the mass spectrometer through a fused-silicawindow and is focused to a 500-pm-diameter beam spot upon a sample probe tip using a 12.5 cm focal length fused-silicalens. Laser energies were attenuated to between 20 and 30 mJ/pulse. The vacuum chamber is pumped by two diffusion pumps (Edwards Diffstak, Model 160) with maximum pumping speeds of 700 W s . Spectra reported here were obtained using source cell detection. Fiveplate trapping is achieved by using a modified high-power excitation amplifier (Waters-Extrel) that can switch between a dipolar excitation configurationand a dc-only configurationto the transmitter plates of the source cell of a dual cubic cell. Direct current voltages on Analytical Chemistry, Vol. 67, No. 20,October 15, 7995
3639
v0it.p t
v0lt.w
I
Excite Plates
-
Receive Plates
R d v a Plate
Figure 2. Diagram of connections for five-platetrapping. Voltage 1 is the dc voltage applied to the rear trapping plate. Voltage 2 is
the dc voltage applied to both the excitation and receiver plates. the receiver plates and rear trapping plate (conductance limit of a dual cell) are supplied through voltagesensitive switches by two external power supplies power Designs Inc., Modes l O S r q New York). Additional electronics protect the preampliiier and direct the signal to and from the cell controller, where the sequence can be controlled with existing Odyssey software (Version 3.0, WatemExkl FTMS). The instrument has been desmied in detail previously?a M a c S i o n (MacSiou, Version 2.03, Montech FTy Ltd.) was used to calculate the potential surfaces for fiveplate deceleration experiments. Experimental Sequence. A diagram of the fiveplate deceleration connections to the cell appears in Fire 2. The decelerating voltage, up to 75 V, is applied not only to the rear trapping plate but also to the excitation and receiver plates of a cubic FTMS cell. The deceleration dc pulse and the excitation pulse on the excitation plates are simply separated in time. To deal with the high kinetic energies of MALDI ions, which increase as ion mass increases, larger dc potentials are applied to the rear trapping plate (plate opposite to sample probe) and the excitation/receiver plates to facilitate efficient trapping. To trap positive ions, the front trapping plate is set to 0.0 V, the rear trapping plate is set between 30 and 75 V, and the excitation/receiver plates are set between 2 and 75 V prior to firing the laser. These decelerating potentials are maintained for 100-300 ps. Subsequently, the potentials of the front and the rear plates are adjusted to between 0.5 and 3 V, suitable for FTMS observations, while the dc voltages on the excitation and receiver plates are switched off. The ability to trap “hot“ions is accomplished by the increased electrostatic potentials during the desorption/ionization event Once trapped, the ions lose their excess energy either through collisional cooling with background neutrals in the source cell (5 x Torr is typical pressure) or through infrared emission. In addition, trapped ions establish an equilibrium with the different ion motional modes. As a consequence, translational energy is redirected from the axi is, permitting detection at much lower trapping potentials. For the mass spectral observations reported here, ions were excited by a chirp excitation with a general range of 50-100 KHz and a sweep rate of 230 Hz/ps. All detection was done in direct mode. The experimental sequence shown in Fwre 3 was employed using a data acquisition rate of 1 MHz and collecting 32K data points. Sample Preparation. The matrix used for all experiments was 2,Mhydroxybenzoicacid (DKB; Flub Chemical Co., Buchs, Switzerland). PFructose was used as a co-matrix because it has been demonstrated to cool biomolecules during the MALDI 3640
Analytical Chemistty, Vol. 67,No. 20, October 15, 1995
0 4 t
I
.... Flow
:
PmbeTip Mounted on Electric Motor
Aerosol spray sample preparation. Air flow tubing is made from a stainless steel syringe and the sample aspiration tubing is made from a fused-silica capillary. Figure’4.
process.2”3 Bovine insulin, carbonic anhydrase, egg albumin, and bovine albumin were purchased from S i a Chemical Co. (St Louis, MO). A 50 mM DHB matrix was prepared by dissolving 0.410 of DHB in 50 mL of methanol. A N p L aliquot of triiluoroacetic acid (IF& was added to give an overall 0.1%TFA (v/v) concentration to enhance ionization efficiency. A 50 mM Dfructose stock solution was prepared by dissolving 0.450 g in 50 mL of methanol. Bovine insulin (1 mg)was dissolved in 5 mL of methanol. Solutions of carbonic anhydrase, egg albumin, and bovine albumin were made by dissolving 1 mg of protein in 2 mL of water. The bovine insulin, DHB, and fructose solutions were combined to form a solution with an analyte/matrix/co-matrix molar ratio of 15000:2500. Similarly, carbonic anhydrase was prepared for analysis by using solutions with an analyte/matrix/ co-matrix molar ratio of 18ooO:4ooo. For egg albumin and bovine albumin, an analyte/maWmmatrix molar ratio of k10 ~ 5 o o o was used. For analysis, each mixture was aerosprayed onto a rotating stainless steel probe tip using the arrangement diagrammed in F~gure4. As the biomolecule mass increases, aerosol spraying becomes more critical. The air flow rate should be very low to obtain a homogeneous crystal layer on the probe tip. Following deposition, samples were inwoduced into the vacuum
(a)Kbster, C.; Castom, J. A,Wilkins. C. LJ.An. Chm.Soc. 1992,114.7.5727574.
n
20000.0
Figure 5. MALDI-FT mass spectrum of carbonic anhydrase (MW = 29 000). Five-plate trapping voltages were 30 V on the rear trapping plate and 10 V on the excitation and receiver plates. For spectral data acquisition, a trapping voltage of 2 V was applied to both the front and rear trapping plates.
40000.0
60000.0
60000.0
m/z Figure 6. MALDI-FT mass spectrum of egg albumin (MW = 45 000). Five-plate trapping voltages were 50 V on the rear trap plate and 35 V on the excitation and receiver plates. For spectral data acquisition, a trapping voltage of 3 V was applied to both the front and rear trapping plates.
system and the system allowed to pump to a pressure of 5 x Torr before analysis. RESULTS AND DISCUSSION High-MassAnalysis. As mentioned earlier, Russell and coworkerszzextended the analytical mass range by placing an ion guide in a cubic cell. Here, the mass range was extended using an unmodified standard FTMS cubic cell. By replacing the single plate deceleration event of a MALDI-FTMS experiment with the fiveplate trapping method, trapping efficiency is improved, as seen in the carbonic anhydrase spectrum in Figure 5. To obtain the spectrum in F i i r e 5, it was necessary to average four time-domain scans taken at sequential probe positions. This procedure was necessary due to presence of the D-fructose co-matrix. Although the D-fructose provides collisional cooling of the protein,2sZ3its presence appears to limit the number of shots available per spot. The averaging of 4 scans is a signi6cant reduction from the averaging of 10-30 scans needed to produce similar signal-tonoise ratios for this compound’s spectrum in previous studies.3 The best spectrum was obtained for deceleration conditions of 10 V on the excitation and receiver plates and 30 V on the rear trapping plate. For spectral data acquisition, ions were detected under standard 2-V trapping conditions. Figure 6 shows the spectrum of egg albumin with a molecular ion of m / z = 45 000. To decelerate these higher mass species, the rear trapping plate (conductance limit) was held at 50 V and the side plates held at 35 V. It appears that the trapping of high-mass ions is simply an electrostatic potential matching pr0blem.2~3~ The ultimate solution may be to use the highest decelerating voltages that one has available. One frequently overlooked detail is the role of the laser’s interaction with the matrix and initial ion velocities. Most matrix studies have focused on testing families of compounds for their efficiency of the ionization of analytes and determination of the optimum mixture ratios. The researchers in these studies have made use of the various types of lasers including nitrogen (24) Hanson, D. H.; Kerley, E. L;Castro, M. E.; Russell, D. H. Anal. Chem. 1989, 61, 2040-2046. (25) Solouki, T.; Russell, D. H. Proc. Natl. Acud. Sci. USA. 1992,89, 57015704.
..
I
5715.00
5720.00
5725.00
m/z
-
5730.00
5735.00
5740.00
Figure 7. MALDI-FT mass spectrum of bovine insulin (MW =
5734). Five-plate trapping voltages were 9.5V on the rear trap plate and 4 V on the excitation and receiver plates. For spectral acquisition, a trapping voltage of 0.6 V was applied to the front and rear trapping plates. Resolving power is 20 000.
lasers (337.1 nm) , Nd-YAG (357 nm, tripled harmonic), and dye lasers (355 nm using 2,2’”-dimethyLpquaterphenyl (BMQ)). However, there is some evidence that the choice of the laser wavelength may play a significant role in establishment of the initial ion velocities,26-z8previously assumed to be 760.1000 m/s for all cases. Because kinetic energy scales directly with mass, it seems probable that lowering the initial desorption velocity would greatly increase the detectableupper mass range of current systems. High-Resolution Analysis. As mentioned earlier, the fiveplate trapping method allows MALDI-generatedions to be trapped efficiency and guides ions close to the center of the cell. Figure 7 shows a bovine insulin five-platetrap MALDI-FT mass spectrum with an average resolving power of 20 OOO and an isotope peak distribution that is in good agreement with the theoretical distribution. This spectrum was obtained by careful adjustment (26) Ens,W.; Mao, Y.; Mayer, F.; Standing, K G. Rapid Commun.Mass Spectrom. 1991,5,117-123. (27) Spengler, B.; Cotter, R. J. Anal. Chem. 1990,62,793-796. (28)Pan, Y.;Cotter, R. J. Org. Mass Spectrom. 1992,27, 3-8.
Analytical Chemistry, Vol. 67,No. 20,October 15, 1995
3641
20000 0
40000 0
60000 0
80000 0
m/z
Figure 8. MALDI-FT mass spectrum of bovine albumin (MW = 66 000). Five-plate trapping voltages were 50 V on the rear trapping plate and 35 V on the excitation and receiver plates. For spectral data acquisition, a trapping voltage of 3 V was applied to both the front and rear trapping plates.
of the trapping voltage on both the excitation and receiver plates (4 V) and also on the rear trapping plate (9.5 V). A trapping voltage of 0.6 V was used on the front and rear trapping plates during normal excitation/receive events to trap an optimum number of ions allowing measurement of this high-resolution spectrum. The principal advantage of the five-plate trapping method over standard FTMS experiments employing gated trapping is that higher mass ions can be trapped and detected because the upper kinetic energy trapping limit for ions has been extended from 9.529to 75 eV. Figure 8 is an extremely lowresolution MALDI-FT mass spectrum of bovine albumin (MW = 66 O00). To decelerate these ions, the rear trapping plate was held at 50 V while the excitation and receiver plates were held at 35 V. Clearly, the mass resolving power shown in this spectrum (29) Dey, M.; Castoro, J. A; Wilkins, C. L. Anal. Chem. 1995,67, 1575-1579. (30) Covey, T.; Douglas, D. J. /. Am. SOC.Mass Spectrom. 1993,4, 616-623. 590-594. (31) Rempel, D. L.; Gross, M. L. /.Am. Sac. Mass Spectrom. 1992,3, (32) Holliman, C. L.; Rempel, D. L.; Gross, M. L. .] Am. SOC.Mass Spectrom. 1992,3,460-463. (33) Jacoby, C. B.; Holliman, C. L.; Rempel, D. L.; Gross, M. L.]. Am. SOC.Mass Spectrom. 1993,4, 186-189. (34) Solouki, T.; Russell, D. H. Appl. Spectrosc. 1993,47, 211-217. (35) Guan, S.; Wahl, M. C.; Wood, T. D.; Marshall, A G. Anal. Chem. 1993,65, 1753-1757. (36) Guan, S.;Wahl, M. C.; Marshall, A. G. j . Chem. Phys. 1994.100, 61376140. (37) Huang, Y.; Pasa-Tolic, L.; Guan, S.; Marshall, A. G. Anal. Chem. 1994,66, 4385-4389.
3642 Analytical Chemistry, Vol. 67, No. 20, October 15, 1995
is unimpressive. However, it does demonstrate the efficacy of the five-trap approach for capturing at least some of these massive ions. In that sense, these results are encouraging, giving evidence that the detectable mass range in FTMS can be extended by this new method. Obviously, it remains to achieve high-resolution spectra for high-mass molecules. Some of the probable causes of the extremely poor mass resolution include the following: (1) the large collisional cross section of the high-mass ions, which can cause loss of coherence during the detection event as a result of collisions with neutrals;30 (2) possible enhancement of metastable decay of the molecular ions, caused by internal excitation during the MALDI process; and (3) difficulties in exciting highmass ions into coherence. Several groups have developed potential means of addressing some of these problems. Rempel and devised a radio frequency-only mode FTMS experiment to manipulate the broadly distributed ion cloud to improve detection of ions initially located away from the center of the trap. Solouki and RusselP4 demonstrated a method to dampen the translational energies in a “waiting room” prior to the excitation and detection events. Marshall and c o - ~ o r k e r s ~ ~ - ~ ~ have shown promising results based upon use of a quadrupolar excitation pulse and buffer gas to relax ions to the center of the FTMS cell. We are currently investigating the combination of quadrupolar excitation with the five-plate deceleration method described here. CONCLUSION
It appears from the present results that five-plate deceleration shows much promise for MALDI-FTMSof high-mass compounds. Because high-voltage deceleration occurs only during the MALDI step, the rest of the mass spectral measurement can proceed as usual. Moreover, this protocol does not involve modification to the existing cell. Based upon the present successes in analyzing singly charged high mass species, future prospects are promising. Because high-mass ions can be trapped, it will be particularly interesting to study the effects of collision-induced or surfaceinduced dissociation upon these large molecules. ACKNOWLEDGMENT
We thank Dr. John Castoro and Paul Jones for their advice and technical support. This research was supported by the National Institutes of Health (Grant GM-44606) and the National Science Foundation (Grant CHE-92-01277). Received 1995.m
for review May 2, 1995. Accepted July 17,
AC950419L Abstract published in Advance ACS Abstracts, September 1, 1995.
