Anal. Chem. 1994,66, 2077-2083
High-Accuracy Molecular Mass Determination for Peptides and Proteins by Fourier Transform Mass Spectrometry Yunthi Li and Robert T. McIver, Jr.' Department of Chemisby, Unlversliy of California, Iwlne, Callfornia 927 17 Richard L. Hunter IonSpec Corporation, 18009 Skypark Circle, Suite F, Iwlne, California 92714 A new calibration method has been developed for Fourier trrnsform mass spectrometry (FI'MS) that is accurate to better than 0.001%(10 ppm) for peptides and proteins up to 5700 Da. The custom-designed FlMS instrument used for this work has a matrix-assisted laser desorption/ionization (MALDI) source located outside of the magnetic field in a differentiallypumped chamber, and ions are injected through the fringing fields of the magnet into the Fl'MS analyzer cell by a long quadrupole ion guide. The mass spectrometer is calibrated with four model compounds ([Arfivasopressin, melittin, bovine insulin B-chain, and bovine insulin) of kwm molecular mass. The set of "dion resonance frequencies ( t ) for these compounds are fit to a three-term calibration equation of the form f = A(z/m) B(V) C(V)(m/z), where m / z is the mass-to-charge ratio of a calibrant peak, V is the trapping.voltage, and A, B, and C are calibration coefficientsthat depend on the magnetic field strength and the dimensions of the analyzer cell. The same set of calibration coefficients can be used for many weeks because the magnet and the electronics of the FTMS instrument are very stable. This method is useful because unknowns can be run separately without the need to add an internal calibration compouad in with the sample.
+
+
Matrix-assisted laser desorption/ionization (MALDI) mass spectrometryhas becomea widely used method for determining the molecular masses of peptides and proteins.*d In the technique of Karas and Hillenkamp, analyte molecules are embedded in a solid matrix, such as nicotinicacid,that strongly absorbs ultraviolet (UV) light. The sample is placed on the tip of a direct probe and inserted into the source of a mass spectrometer. Upon irradiation with a pulse from a UV laser, large numbers of intact protonated molecular ions of the analyte are produced in a burst and detected by time-of-flight (TOF) mass spectrometry. This method has been used successfully with proteins having masses up to more than 100 kDa with detection sensitivities down to femtomole levels. In MALDI-TOF mass spectrometry, ions produced by the laser pulse are accelerated out of the source region and allowed (1) Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. Inr. 1.Muss Specerrom. Ion Processes 1987, 78, 53.
(2) Kanr,M.; Hillenkamp, F. Anal. Chem. IWuI,60, 2291. (3) Kam, M.;Bahr, U.; Hillenkamp, F. Inr. J. Mass Specrrom. Ion Processes 1989, 92, 231. (4) Hillcnkamp, F.; Kans, M.; Bcavis, R. C.; Chait, B. T. Anal. Chem. 1991,63, 1193A. (5) Biemann, K. Annu. Rev. Biochem. 1992, 30, 917. (6) Chait, 9. T.; Kent, S.B. H. Science 1992, 257. 1885.
oo03-2700/94/0368-2017~Q4.5QlQ (0 1994 American Chemical Soclety
to drift down a long tube before striking a channel electron multiplier. In principle, by measuring an ion's flight time ( t ) and knowing the accelerating voltage (V)and the length of the mass-to-charge ratio (m/q in kg/C) can be the tube (4, determined from
Substituting typical values ( V = 20 000 V and d = 1 m) with the proper SIunits gives the result that an ion with a measured flight time of 50.00 ps has a mass of 9648 Da (assuming unit charge). In practice, this equation is inadequate for accurate mass measurements, and it is better to calibrate the TOF mass spectrometer with samples of known molecular mass. In 1990, Beavis and Chait showed that protein molecular masses could be determined by TOF with an accuracy of approximately 0.01% by using a single internal calibrant.' The calibration equation they used is (m/z)'I2= Ar
+B
(2)
where A and Bare calibration constants, and m/z is the masstwharge ratio in unitsof Da/charge ( z = q/e). The calibration coefficients are determined uniquely by measuring the flight times ( t ) for two calibrant peaks of known m/z. Beavis and Chait added a calibrant protein to a sample to be analyzed and used the two ions corresponding to the protonated protein (M H)+ and the doubly protonated protein (M 2H)*+ to determine A and B for their instrument. Use of this procedure is contingent on the observation of clearly resolved protonated molecular ions in the mass spectrumof the calibrant protein. Mass measurement errors can result if unresolved Na+ adduct peaks or fragments such as (M H - NH3)+ skew the peak for the protonated molecular ion. To achieve higher mass resolution and more accurate mass measurement, several groups have investigated using other types of mass spectrometers with MALDI. Biemann and coworkers demonstrated the feasibility of performing MALDI with a magnetic mass spectrometer.*v9 Because of the pulsed nature of the ion signal, an integrating focal-plane detector was used to record mass spectra. Using cluster ions of CsI
+
+
+
(7)&vis, R. C.; Chait, 9. T. Anal. Chem. 1990.62, 1836. (8) Hill, J. A.; Annan, R. S.;Biemann, K. Rupid Commun. MassSpcrrom. 1991, 5, 395. (9) Annan, R. S.;Kochling, H. J.; Hill, J. A.; Bicmann, K. Rupid Commun. Mass Specerrom. 1992,6, 298.
AnaljlHcel ChedstYy, Vd. 66,No. 13, Ju& 1, 1994 2077
SOURCE
/--
SUPERCONDUCTING MAGNET ION GUIDE
PUMP 1 PUMP 2 PUMP3 Flgute 1. Block diagram of a Fourier transform mass spectrometer with an external MALDI ion source and quadrupole ion guide for in/ecting ions into the analyzer cell.
as a calibrant, they achieved a mass assignment error of only +1.3 u (0.015%) for the [M + H]+ peaks of the protein ubiquitin. With a focal-plane diode array detector, the resolution and mass measurement accuracy are limited by the dispersion setting,which under the conditionsused for ubiquitin was 0.5 mass unit per pixel.9 Glish and co-workers recently reported MALDI with a quadrupole ion trap mass spectrometer (ITMS).l0 The sample probe and the laser beam were brought through holes in the ring electrode so that the ions could expand directly into the trap cavity. By using two calibrant peptides of known mass (bradykinin and melittin), they achieved a mass accuracy of 0.05% for the [M + H]+ ion of neurotensin at m/z 1673.9. Problems in mass calibration occurred in the ion trap when there was excessive ion density, resulting in a peak shift to higher m/z values. One limitation encountered with the ITMS was an unanticipated drop in sensitivity and mass resolution for ions above m/z 3000. Several laboratories have been involved in adapting MALDI to Fourier transform mass spectrometry (FTMS). The first MALDI-FTMS experiments were reported by Hettich and Buchanan in 1991.*1-13 They demonstrated reduced fragmentation for small peptides and oligonucleotides, but the sensitivity and mass resolution were lower than anticipated. MALDI-FTMS experiments by Solouki and Russell in 1992 used an external "waiting room" device mounted on the end of the direct probe for collisional relaxation of the ions prior to their transfer to the analyzer cell.14 Promising results were reported in 1992 by Castoro and Wilkins, who used carefully timed pulses to trap the ions in the FTMS analyzer cell and a sugar co-matrix to minimize metastable d e ~ a y . ' ~ ' 'With this method, mass resolution in excess of 100 000 was achieved for small peptides and 27 300 for bovine insulin. Recently, we reported the highest resolution ever achieved for MALDI mass spectrometry.18919 For [Arg8]-vasopressin at m/z 1084.4,the mass resolution (M/hM1/2) was 1 100 000, and for bovine insulin at m/z 5730.6, the resolution was 90 000. Our work was done with an external ion source FTMS instrument in which the MALDI source is outside of the magnetic field in a separately pumped chamber. Ions are injected through the fringing fields of the magnet into the FTMS analyzer cell by a long quadrupole ion guide operated in the RF-only mode. The purpose of this paper is to describe a new calibration procedure for the FTMS instrument that can be used for high-accuracymolecular mass determinations. A three-term 2078
Ana&ticalChemistry, Vol. 66, No. 13, Juty 1, 1994
calibration equation is derived that accounts for the effect of trapping voltage on the ion resonance frequencies. The new calibration procedure provides better than 0.001% (10 ppm) mass measurement accuracy for analytes up to a mass of 5700 Da. Once calibrated, the mass spectrometer is stable for several weeks because the superconducting magnet used with the FTMS instrument has a drift of less than 0.1 ppm/week. This external calibration method is useful because unknowns can be run separately without the need to mix in an internal calibration compound. EXPERIMENTAL SECTION The external ion source Fourier transform mass spectrometer used in this work was built at the University of California, Irvine, and has been described previously. 18-25 The layout of the spectrometer is shown in Figure 1. Analyte samples embedded in a solid matrix of 2,5-dihydroxybenzoic acid (DHB)26are deposited on the tip of a direct insertion probe and irradiated with a laser pulse. Ions are extracted from the source region, accelerated to 40 eV, and transported by a RF-only quadrupole ion guide to a FTMS analyzer cell that is centered inside the bore of a powerful solenoidalmagnet. The purpose of the quadrupole ion guide is to focus the ions and enable them to be efficientlyinjected through the fringing fields of the magnet, which are particularly severe at the end of the magnet bore t ~ b e . The ~ ~laser * ~used ~ in our experiments is a Lambda Physik Model EMG 201 MSC excimer laser (IO) Chambers, D. M.; Goeringer, D. E.; McLuckey, S. A.; Glish, G. L. Anal. Chem. 1993.65, 14. (1 1) Hettich, R. L.; Buchanan, M. V. J. Am. Soc. Muss Specfrom. 1991, 2. 22. (12)Hettich, R. L.; Buchanan. M. V. J. Am. Soc. Muss Spectrom. 1991,2,402. (13) Hettich, R.L.; Buchanan, M. V. Inf.J. MussSpecfrom. Ion Processes 1991, 111, 365. (14)Solouki, T.;Russell, D. H. Proc. Nufl. Acud. Sci. U.S.A. 1992.89, 5701. (15) Castoro, J. A.; Koster, C.; Wilkins, C. L. Rupid Commun. Mass Specfrom. 1992,6, 239. (16)Koster, C.;Castoro, J. A.; Wilkins, C. LJ. Am. Chem.Soc. 1992,114, 7572. (17)Castoro, J. A.; Wilkins, C. L. Anal. Chem. 1993.65, 2621. (18)McIver, R.T.,Jr.; Li, Y.; Hunter, R. L. Inr. J. Mass Specrrom. Ion Processes 1994, 132, L1. (19)McIver, R. T.,Jr.; Li, Y.; Hunter, R. L. Rupid Commun. Mass Specfrom. 1993, 8,237. (20)McIver, R. T.,Jr.; Hunter, R. L.; Bowers, W. D. Inr. J. MussSpecrrom.Ion Processes 1985.64, 67. (21)Lebrilla, C.B.; Amster, I. J.; McIver, R. T.,Jr. Inf. J. Muss Specfrom. Ion Processes 1989.87, R7. (22)Lebrilla, C. B.; Wang, D. T.-S.; Mizoguchi, T.J.; McIver, R. T., Jr. J. Am. Chem. Soc. 1989.11 1,8593. (23)McIver, R. T.,Jr. Inf. J. Mass Specfrom. Ion Processes 1990, 98,35. (24)Lebrilla, C. B.;Wang, D. T.-S.;Hunter, R.L.; McIver, R.T.,Jr. Anal. Chem. 1990,62,878.
(25)McIver, R. T.,Jr. US.Patent 4,535,235,Aug 13, 1985. (26)Strupat, K.; Karas, M.; Hillenkamp, F. Inf.J. Muss Specrrom.Ion Processes 1991, Ill, 89.
USER PUSES
c--
DIGITAL PUS€ GENERATOR
DIA CONVERTER PROQRAMUER
t
t
-
ARBITRARY WAVEFORM GENERATOR
FREQUENCY SYNTHESIZER
t * COWUTER NTERFACE
SYNTHESIZER PROQRAUYR FILTERS
0
I
COMPUTER
El I
1
Flgurr 2. Main components of the Omega 586 data system for instrument control and data analysis.
operated with xenon fluoride at 35 1 nm. The FTMS analyzer cell is an elongated trapped ion cell (inside dimensions 5 X 5 X 7.6 cm) with stainless steel plates.27 The magnet is a 6.5 T superconducting magnet with a 15-cm-diameter room temperature bore made by Oxford Instruments. The magnetic field strength is very stable, with a drift rate of less than 0.1 ppm/week. There are three Air Products APD-6 cryopumps on the vacuum system to maintain a base pressure in the analyzer region below 10-9 Torr, even when the pressure in the ion source region is as high as lo-’ Torr. All operations of the mass spectrometer are controlled by an Omega 586 data system designed and built by IonSpec Corp. The main parts of the data system are shown in Figure 2. MALDI experiments are initiated when one of the User Pulses from the Digital Pulse Generator triggers open a valve (General Valve Corp. Model 90) for 2 ms to admit a pulse of argon gas to the analyzer cell region. Pulses from the digital pulse generator can be set to a resolution of f l ps. At the same time, the power supply for the quadrupole ion guide is pulsed on so that a 10&500-V, 1-MHz signal is applied differentially to the two pairs of rods. Initially, the trapping plates of the FTMS analyzer cell are set at 0 V on the front trapping plate and 10 V on the rear trapping plate. These potentials are controlled by a 12-bit digital-to-analog (D/A) converter unit. After a delay time of 50 ms, the laser is triggered by a different user pulse, and the ions move rapidly out of the source region and down the quadrupole ion guide to the analyzer cell. After a delay time of 0.5-0.8 ms, when most of the ions for a particular sample are at the center of the cell, the potential on the front trapping plate is pulsed rapidly to 10.0 V. This is a well-established method for trapping ions and was used in the first external ion source (27) Hunter, R. L.; Sherman, M.G.; McIver, R. T., Jr. Inr. 1.Mars Specrrom. Ion Processes 1983, 50, 259.
FTMS experimentsover 8 years ago.= Typically,ions covering a mass range of several hundred daltons can be trapped at a particular setting for the trapping plate pulse. As the ions move about the analyzer cell, they collidewith the argon buffer gas and lose their excessive kinetic energy. After waiting 10-1 5 s for the argon to pump out of the chamber, the D/A unit ramps down the trapping voltages to a low value prior to detection. To acquire a mass spectrum, the trapped ions are accelerated by either an impulse ~ i g n a l ~ ~or- a~radiofrequency O (RF) ~ h i r p . Figure ~ ~ , ~2 shows ~ there are two RFsignal generators, a frequency synthesizer and an arbitrary waveform generator. Following acceleration, the image current signals induced on the receiver plates by the coherent cyclotron motion of the ions are amplified, filtered, and sent to a 16-MHz analogto-digital converter with 9-bit resolution. Although all the experiments reported in this paper were performed in this way by using the direct mode of signal acquisition, Figure 2 also shows an alternate signal path with a mixer so that mixer mode, or heterodyne, acquisitions can performed. The digitized signals are stored initially in a large (1024-K word) static RAM buffer memory. At the end of the pulse sequence, the data are transferred to a host computer where a 66-MFLOP floating point array processor performs the Fourier transform calculations. A Hanning apodization window and one zero fill are used in the FFT calculations to improve the accuracy of the peak centroid calculations. The array processor is very fast, and a new mass spectrum is displayed on the color graphics monitor after only a few seconds. (28) McIver, R. T., Jr.; Hunter, R. L.;Bayht, G. Anal. Chem. 1989, 61, 489. (29) McIvcr, R. T., Jr.;Hunter, R. L.; Baykut,G. Rev. Scf. Instrum. 1989,60,400. (30) McIver, R. T., Jr.; Baykut, G.; Hunter, R. L. Inr. J. Mars Specrrom. Ion Processes 1989.89, 343. (31) Comisarow, M.B.; Marshall, A. G. Chem. Phys. Lcrr. 1974, 26,489. (32) Marshall, A. G. Acc. Chem. Res. 1985, 18, 316.
The analyte solutions were prepared by dissolving approximately 0.2 mg of peptide in 1 mL of methanol. The matrix solution was prepared by adding 10 mg of 23dihydroxybenzoicacid (DHB) to 1 mL of ethanolF6 Melittin, [Arg*]-vasopressin,bovine insulin B-chain, and bovine insulin were obtained from Sigma and were used without further purification. DHB was obtained from Aldrich. For bovine insulin, D-fructose was added as a co-matrix with DHB to suppress metastable decay.16J7 Equal 0.5-pL volumes of the analyte and matrix solutions were delivered to a stainless steel sample probe and allowed to evaporate to dryness prior to insertion into the vacuum lock of the ion source.
(4)
where oca is the measured resonance frequency (s-l) and K = 2aqV/muz is a term that accounts for the shift due the electrostatic trapping potential V. The other parameters in K are a geometrical factor a for the elongated analyzer cell and the spacing u between the plates. Under normal operating conditions, the approximation K
