Mass Spectrometry in the Development of Drugs from Traditional Medicines

Transcription

1 Mass Spectrometry in the Development of Drugs from Traditional Medicines Fred W. McLafSerty, Michael W. Senko Baker Laboratory, Cornell University, Ithaca, New York, USA Key Words. Mass spectrometry Tandem mass spectrometry Electrospray ionization Fourier-transform mass spectrometry Protein sequencing Nucleotide sequencing Abstract. The mass spectrometer ionizes molecules to separate and weigh the resulting molecular ions and their dissociation products. These product masses indicate directly the sequence of constituents of the original molecule, such as amino acids in proteins or bases in nucleotides. Because single ions can be detected, even subfemtomole sensitivities are possible. In tandem mass spectrometry (MSMS), molecular ions from a mixture can be separated to isolate ions of a specific component, whose further dissociation can then give structural information on that component. Electrospray ionization and other recently developed methods make possible the ionization of biomolecules even larger than 100 kda. The Fourier-transfer mass spectrometer has unusual capabilities for measuring ions over a 100 kda mass range simultaneously at unit resolution for MSMS applications utilizing subfemtomole sample quantities. Introduction Traditional medicines developed from plants and other natural sources over millennia of trial and error represent a treasure house of medical information that has been little exploited for new drug development until the last generation. For new drug development, the molecular structure of the drug s active component(s) is of critical value not only for defining safe and effective human dosages but also for developing analogous synthetic drugs of superior efficacy and/or safety. Mass spectrometry has been a key tool for Correspondence: Dr. Fred W. McLafferty, Department of Chemistry, Baker Laboratory, Cornell University, Ithaca, NY , USA. Received November 29, 1993; accepted for publication November 29, OAlphaMed Press /94/$5.OOlO characterizing smaller active molecules [ 11, but in the last decade great progress has been made in applying this to molecules in the kda range. This has resulted from dramatic developments in new techniques for large molecule ionization and for tandem mass spectrometry [ Basic Advantages of Mass Spectrometry For decades, mass spectrometry has been a key structural and analytical method. The mass spectrometer is both a weighing and a separating device; after a molecule is ionized, the mass spectrometer separates the ions according to their massto-charge ratio (dz), with single ion detection possible for the separated ions. This results in unusual capabilities in sensitivity, molecular weight accuracy, molecular structure information, speed and specificity, especially when connected to powerful separation methods such as gas and liquid chromatography (GCMS and LCMS), capillary electrophoresis [22,23] and even to the mass spectrometer itself (MSMS) [26,27]. As an example of a molecule considered large for MS structural characterization in 1959 [28], the electron ionization (EI) mass spectrum of 10 g of the methyl ester of mycocerosic acid showed m/z 494 as the peak of highest mass. Commercial 1959 instruments had mass accuracies of one part in a thousand, so that this molecular weight was known to be 494 Da, not 493 or 495 Da. Further, such ionization with energetic electrons (70 ev) also forms molecular ions with sufficient energy to dissociate before mass analysis; in this case these product ions gave significant peaks such as m/z 59, 87, 101, 129, 143, 171, 185 and 199. The great advantage of such linear molecules is that backbone bond cleavage STEM CELLS 1994;12:68-73

2 McLafferty/Senko 69 can generate structurally significant products; a linear alkyl chain of -CH2- groups should give peaks spaced 14 Da apart (C =12, H = 1 Da). As found here, methyl esters usually give a d z 59 peak, corresponding to CH30CO+ ( Da). Now the peak spacings of the significant peaks of 28 (87-59), 14,28,14,28,14 and 14 Da indicate correctly the chain positions of methyl (-CHCH3-) substitution: CH30-CO-CHCH3-CH2- CHCH3-CH*-CHCH3-CH2-CH2-C2lb3. (Actually the most abundant peak in this spectrum is at dz 88, arising from a well understood rearrangement of one hydrogen involving the cleavage of the same C2-C3 bond that produces m/z 87 and is much more definitive for showing the 2-methyl substitution [l].) This accurate and rigid structure determination required only a microgram of material, an amount of sample that would be challenging for many structural techniques even today. Structural interpretation of the mass spectra of cyclic molecules is not necessarily this straightforward, but literally thousands of research publications have described the gas phase chemistry of unimolecular ion dissociation [ 13 so that at least partial structural information can be obtained from the EI mass spectra of most molecules. The Self-Training Interpretive and Retrieval System (STIRS) [29], one of the first successful Artificial Intelligence programs, is a valuable aid to the interpreter for mass spectra of molecules that have never been characterized. However, for known compounds, such mass spectra can contain literally hundreds of peaks at any of 500 or even 1000 m/z values and are thus highly specific characteristics of the molecular structure; the Probability Based Matching (PBM) computer program can identify or greatly restrict the possible identities of an unknown mass spectrum by matching it against the current file of 225,000 different EI reference mass spectra. With the modern dedicated computer of an MS instrument system, this search can be accomplished in a few seconds, greatly speeding the investigation of natural product mixtures [30]. GUMS, LUMS The modern mass spectrometer can measure a complete mass spectrum in much less than one second, faster than the separation time for the most commonly used methods such as gas and liquid chromatography [3 11. Interfaces that allow the direct coupling of GC and LC with MS are now surprisingly reliable and routine in operation, so that a natural product mixture from a traditional medicine can be separated and mass spectra obtained from literally hundreds of components in less than an hour. For many years GCMS and LC/MS have been routine tools for characterizing commercially important mixtures of natural products such as flavors, perfumes, insect pheromones and drug metabolites, as well as by-products of the chemical and petroleum industry. Reports of the successful use of GCMS and LCMS in identifying active components in traditional medicines are increasing rapidly. MS/MS [25, 261 The mass spectrometer is also a separation device. If a soft ionization technique such as chemical ionization (CI) is used, ionization of a mixture can be made to produce mainly molecular ions, so that the resulting mass spectrum shows peaks for each of the mixture components. However, if the separated molecular ions corresponding to one component are then dissociated, these fragment ions can then provide structural information concerning that component, just as the hard ionization EI mass spectra gives structural information on a pure compound. This is especially valuable for the targeted compound analysis for specific compounds in complex mixtures such as plant extracts, in that several or even many compounds in the mixture could have the same molecular weight as that of the targeted compound. For this, the reference dissociation spectrum of the compound s molecular ions can be measured to see which of the fragment peaks, or combinations of fragment peaks, are uniquely representative of the compound; remember that there are hundreds of possible mass positions for these peaks. This also can greatly increase the sensitivity for such a targeted analysis by reducing the chemical noise from interfering mixture components. The triple quadruple mass spectrometer is the most widely used instrument for this, with the first quadruple separating the mixture of molecular ions, the second quadruple serving as a collision chamber to dissociate the ions and focus them into the third quadruple, where the fragment ion mass analysis takes place. More sophisticated and expensive instruments that can provide high resolution separation of both

3 I0 the original ion mixture and the dissociated product ions are the tandem double-focusing and Fourier-transform mass spectrometers. New Capabilities for Large Molecules For many years the only method of making the gas phase ions necessary for mass spectral separation was to vaporize the molecule and then ionize it. Nearly twenty years ago Macfarlane showed that the 100 MeV particles from 252Cf fission would desorb and ionize larger molecules [2]. Fast atom bombardment (FAB) of larger molecules in a nonvolatile liquid matrix can achieve this routinely and simply for molecules as large as 10 kda, while matrix-assisted laser desorption ionization (MALDI) can produce molecular ions as large as several hundred kda [4]. Although the time-of-flight mass spectrometer [7] utilized for MALDI has a resolving power of only -500, this method s mass accuracy of 0.01 % to 0.1 % is broadly applicable for routine operation, even for complex mixtures. Electrospray ionization (ESI), originally proposed by Dole, has developed recently into a powerful technique based on seminal discoveries of John Fenn [3]. In a crude sense this is like a sneeze that projects large virus molecules into the gas phase to be inhaled by an unsuspecting friend. In practice a solution of the high molecular weight sample is sprayed from a capillary through several kilovolts of potential, forming electrostatically charged droplets that shrink by evaporation and break up by charge repulsion. Eventually the droplet is so small that the electrostatic charges are lost by expelling these from the droplet on a large molecule; positively charged droplets usually expel molecules with many protons attached, while negatively charged droplets expel molecules from which many protons are missing. The resulting molecularly charged ions contain a charge for every 500 to -2,000 Da, and thus are separated in the mass spectrometer as d z 500 to 2,000. This means that most common types of mass spectrometers can be used for the mass analysis, even for high mass values, as compared with the MALDI method that requires a time-of-flight instrument for its high mass ions with one or a few charges, Most electrospray experiments utilize aqueous solutions or those containing polar solvents, and suitable spray conditions often must be found Mass Spectrometry in Drug Development by trial and error, a disadvantage in comparison with the MALDI technique. A serious problem for other ionization methods is that increasingly large singlycharged ions become increasingly difficult to dissociate by adding energy to them, such as by collisional activation [8]. Thus, although both mass analyzers of commercial tandem double-focusing mass spectrometers can separate ions larger than d. 10,000, collisionally activated dissociation of singly charged ions larger than -2,500 Da is usually very inefficient. For ESI however, its multiply charged ions in the usual dz range appear to have no upper limit to their dissociation by collisional activation. Smith, Loo, and coworkers [ have reported extensive MSIMS data for the dissociation of different forms of albumin, 66 kda, giving fragment ions whose d. values could be correlated with known sequence data. Recently, Feng has reported the dissociation of 150 kda immunoglobulin ions [21]. Not only does MSIMS with MALDI generated ions have this disadvantage for dissociating large singly charged ions, but also special techniques for performing MSIMS on the time-of-flight instrument have been investigated only recently [7]. ESVFourier-Transform Ion-Cyclotron-Resonance Mass Spectrometer (FTMS) at Cornell [ An instrument that appears to have attributes uniquely suited to MSIMS with ESI for large molecules is the FTMS [5,6, 101. In a sufficiently high magnetic field, ions with a sufficiently low kinetic energy will follow circular cyclotron orbits. By giving stationary trapped ions a radio frequency (RF) push of appropriate magnitude, all ions of the same d z value will orbit together at a frequency only dependent on this value and on the magnetic field strength; ions of lower kinetic energy will orbit in a smaller circle, but with exactly the same frequency. This results in orders of magnitude more resolving power for FTMS than for magnetic sector instruments, in which the deflection angle depends on the ion kinetic energy as well as the magnetic field strength. A second advantage is that ion detection is nondestructive. Parallel plates above and below the ion orbit are connected in a detection circuit, so that positive ions approaching one of the plates cause an electron flow toward the plate, which is then reversed as the ions approach the other plate; a

4 McLaffertyKenko 71 sine-wave signal is produced whose frequency is dependent on the d z value and whose amplitude is dependent on the number of charges in the ion packet. FTMS also exhibits Fellget s advantage or the multi-channel advantage, in that a mixture of ions of many d z values produces a multitude of frequencies simultaneously, but the Fourier-transform of these frequencies yields the masses of all the ions in the frequency range; with ESI, a mass range of > 100 kda can be measured in less than one second. Of course this is a special advantage for very large unknown molecules, as scanning instruments would have to consume the sample while scanning across all these mass values to find the ions produced. When large ions lose energy by collision with the comparatively light molecules of the background gas, the latter are scattered from the cell, but most of the large ions wind back down to the center of the cell where they can be measured again or are available for reaction for MSIMS experiments. Specific ions can be selected from the mixture, such as by stored waveform inverse Fourier-transform (SWIFT) [6] kinetic excitation of just those ions which are to be expelled from the cell or to be excited for collisional dissociation. The product ions can then be mass analyzed in the same ion cell, and this process can even be repeated for MS experiments. Simultaneous measurements of MS/MS spectra are possible with a Hadamard transform technique [32]. With the ESI/FTMS instrument, the unique mass spectrometry advantages described above for small molecules can now be extended to those even as large as albumin (67 ma). Illustrating sensitivity, 5 x 10-l6 mols (500 attomoles) of ubiquitin (8.6 kda) introduced into the FTMS gives a complete mass spectrum of molecular ions of different charge states with a resolving power of lo5 [15]. This is sufficient to resolve the isotopic peaks; with the 1.1 % natural abundance of I3C in I2C, the most abundant isotopic peak is that containing five 13C atoms, consistent with -400 carbon atoms in ubiquitin; the peak containing only I2C atoms is of only 4% abundance. Computer deconvolution techniques can combine all ions that are separated by the exact values predicted for unit charge changes, so that the resulting mass spectrum of the combined isotopic peaks shows these with abundances close to those predicted theoretically for these isotopic combinations. Thus the 5 x 10-l6 mols of sample produced an ESI mass spectrum with a signal-to-noise (S/N) ratio of 30, despite the fact that the spectrum actually recorded all ions produced over a mass range of more than 100 kda. By heterodyne recording over a much smaller d z range, the resolving power can be increased substantially with 2 x lo6 achieved for ubiquitin [15]. High resolving power should also mean high mass measuring accuracy, especially with internal standards. For example, to determine the mass of a molecular ion of myoglobin (17 kda), the addition of cytochrome c (12.3 kda) as an internal standard allows the 13+ molecular ion peak of the latter to be compared directly with a nearly overlapping isotopic peak of the 18+ ion from myoglobin, measuring this with ppm accuracy. The largest molecule for which we have achieved unit resolution is albumin, 67 kda. The capability of FTMS to record simultaneously many mass values is shown also by the ESI mass spectra of polyethylene glycols (PEG). For example, the mass spectrum of PEG 20,000 (30 co-added scans, resolving power 50,000) shows peaks representing (M + nna) + for M = HO(CHzCHzO),H for n values of 460 to 520; each n value is shown in 5 to 15 charge states, and each of these is represented by 7 to 20 isotopic peaks, so that there are approximately 5,000 measurable peaks in this spectrum. The current data system can acquire a 2 M point frequency transient in one second for Fourier-transform. Sequence Information [8, 9, 12, 18-21] After mass analysis of the FTMS trapped ESI ions, these can be dissociated by adding energy from collisions with an added gas or a surface, or from laser photons. In recent studies, collision energy was added to molecular ions of carbonic anhydrase (B form) entering the instrument, so that the resulting mass spectrum contained more than 100 ions whose d z values could be measured accurately and whose z values could be determined from the isotopic peak spacing. The measured mass of the molecular ion is 29, Da, and several combinations of fragment ions were found whose masses sum to this value. For example, 21, and 7, Da are such a pair (Z = Da), while the larger of these corresponds to the 15, and 6, Da fragments (C = 21, Da), and

5 72 Mass Spectrometry in Drug Development the smaller of these corresponds to the 4, and 10, fragments (C = 15, Da). As found by Loo and Smith [20] these abundant fragment ions correspond to cleavage on the N-terminal side of proline, and less abundant fragment ions arise from cleavages at N-terminal adjacent amino acids; 80% of the fragment ions in this spectrum correspond to proline-directed cleavages. Thus peaks at 7,764.64, 7, and 8, Da show not only that the 7, Da fragment is C-terminal, but also that its N-terminal adjacent amino acids are tyr ( Da; 7, , = Da), thr ( Da), and rrp ( Da). Isolation of the 7, Da fragment ion in the cell by SWIFT [6] followed by its collisional dissociation yielded ten more assignable fragment ions that must also be near the C-terminus. A continuation of such MS" experiments could, at least in theory, give the complete sequence. However, the above information is obviously valuable for checking the validity of a sequence on quality control of genetically-engineered proteins. The corresponding dissociation mass spectrum of the A form of carbonic anhydrase showed the molecular and many fragment ions to be Da less than those of the B form, with the other fragment ions of the same mass, consistent with the location of the Gln + Arg substitution. Future Very encouraging preliminary results on nucleotide sequencing have also been reported. We find that the same EWFTMS high-resolution mass spectra (but working in negative ion mode) can provide complete sequence information for oligonucleotides such as 5'-d(ATGC- TACGT)-3', with partial information for a 25-mer (8 kda). A unit resolution molecular ion spectrum has been obtained for the 76-mer, phenylalanine-specific trna from Brewer's yeast (26 kda). An immediate application of such information for the human genome sequencing project would be for sequence verification to reduce the error rate of current electrophoresis methodologies. With careful optimization this could take advantage of the high sensitivity (subfemtomole) and speed (one second per spectrum) of obtaining highly specific information. However, the capability to provide highly accurate (0.1 to Da) mass and sequence data on important biomolecules should immediately provide critical new information for basic research and pharmaceutical development, including the identification and isolation of the active molecules of traditional medicines. Acknowledgments Numerous coworkers, some of whom are listed in the references, made key contributions to these studies. Generous financial support was provided by the National Institutes of Health (grant GM 16609), with partial instrumentation support by the National Science Foundation. References 1 McLafferty FW, Turecek F. Interpretation of Mass Spectra. Mill Valley, CA: University Science Press, 1993;l Macfarlane RD, Torgerson DF. Californium-252 plasma desorption mass spectrometry. Science 1976;191: Fenn JB, Mann M, Meng CK, Wong SF, Whitehouse CM. Electrospray ionization-principles and practice. Mass Spectrom Rev 1990;9: Hillenkamp F, Karas M, Beavis RC, Chait BT. Matrix-assisted laser desorptionhonization mass spectrometry of biopolymers. Anal Chem 1991;63:1193A-l203A, 5 Comisarow MB, Marshall AG. Fourier transform ion cyclotron resonance spectroscopy. Chem Phys Lett 1974;25: Alber GM, Marshall AG, Hill NC, Schweikhard L, Ricca TL. Ultrahigh-resolution Fourier transform ion cyclotron resonance mass spectrometer. Rev Sci Instrum 1993;64: Cotter RJ. Time-of-flight mass spectrometry for the structural analysis of biological molecules. Anal Chem 1992;65:1027A-l039A. 8 Biemann K. Mass spectrometry of peptides and proteins. Annu Rev Biochem 1992;61: Cody RB, Amster IJ, McLafferty FW. Peptide mixture sequencing by tandem Fourier-transform mass spectrometry. Proc Natl Acad Sci USA 1985;82: Buchanan MV, Hettich RL. Fourier transform mass spectrometry of high mass biomolecules. Anal Chem 1993;65:245A-259A.

6 McLaffertylSenko Castor0 JA, Wilkins CL. Ultrahigh resolution matrix-assisted laser desorption. Ionization of small proteins by Fourier transform mass spectrometry. Anal Chem 1993;65: Chait BT, Wang R, Beavis RC, Kent SBH. Protein ladder sequencing. Science 1993;262: Feng R, Konishi Y. Analysis of antibodies and other large glycoproteins in the mass range of 150, ,000 Da by electrospray ionization mass spectrometry. Anal Chem 1992;64: Ganem B, Li Y-T, Henion JD. Detection of noncovalent receptor-ligand complexes by mass spectrometry. J Am Chem SOC 1991;113: Beu SC, Senko MW, Quinn JP, Wampler FM 111, McLafferty FW. Fourier-transform electrospray instrumentation for tandem high-resolution mass spectrometry of large molecules. J Am SOC Mass Spectrom 1993;4: Henry KD, Williams ER, Wang BH, McLafferty FW, Shabanowitz J, Hunt DF. Fourier-transform mass spectrometry of large molecules by electrospray ionization. Proc Natl Acad Sci USA 1989;85: Henry KD, Quinn JP, McLafferty FW. High-resolution electrospray mass spectra of large molecules. J Am Chem SOC 1991;113: Loo JA, Edmonds CG, Smith RD. Tandem mass spectrometry of very large molecules: serum albumin sequence information from multiply charged ions formed by electrospray ionization. Anal Chem 1991;63: Light-Wahl KJ, Loo JA, Edmonds CG, Smith RD, Witkowska HE, Shackleton CHL, Wu CSC. Collisionally activated dissociation and tandem mass spectrometry of intact hemoglobin P-chain variant prot -ins with electrospray ionization. Biol Mass Spec,;om 1993;22: Feng R, Konishi Y. Collisionally activated dissociation of multiply charged 150 kda antibody ions. Anal Chem 1993;65: Smith RD, Wahl JH, Godlett DR, Hofstadler SA. Capillary electrophoresis/mass spectrometry. Anal Chem 1993;65:574A-584A. 23 Hofstadler SA, Wahl JH, Bruce JE, Smith RD. On-line capillary electrophoresis with Fourier transform ion cyclotron resonance mass spectrometry. J Am Chem SOC 1993;115: Li Y-T, Hsieh Y-L, Henion JD, Senko MW, McLafferty FW, Ganem B. Mass spectrometric studies on noncovalent dimers of leucine zipper peptides. J Am Chem SOC 1993;115: Williams ER, McLafferty FW. High resolution and tandem Fourier-transform mass spectrometry with Californium-252 plasma desorption. J Am SOC Mass Spectrum 1990;1: McLafferty FW, ed. Tandem Mass Spectrometry. New York: Wiley, 1983;l Busch KL, Glish GL, McLuckey SA. Mass Spectrometry/Mass Spectrometry. New York: VCH Publishers, 1988;l Ryhage R, Stenhagen E. Mass spectrometry of long-chain esters. In: McLafferty FW, ed. Mass Spectrometry of Organic Ions. New York: Academic Press, 1963: Kwok K-S, Venkataraghavan R, McLafferty FW. Computer aided interpretation of mass spectra 111. A self-training interpretive and retrieval system. J Am Chem SOC 1973;95: McLafferty FW, Loh SY, Stauffer DB. Computer identification of mass spectra. In: Meuzelaar HCL, ed. Computer Enhanced Analytical Spectroscopy. Plenum: New York 199O;II: McLafferty FW, Michnowicz J. State-of-the-art GCMS. Chemtech 1992;26: Loo JA, Edmonds CG, Smith RD. Tandem mass spectrometry of very large molecules Williams ER, Loh SY, McLafferty FW, Cody Dissociation of multiply charged proline-con- RB. Hadamard transform measurement of tantaining proteins from electrospray ionization. dem Fourier-transform mass spectra. Anal Chem Anal Chem 1993;65: ;62: