Anal. Chem. 1995, 67, 1400-1408
Sequence-Specific Fragmentation Generated by Matrix-Assisted Laser Desorption/lonization in a Quadrupole Ion Trap/Reflectron Time-of-Flight Device Heewon Lee and David M. Lubman* Department of Chemistty, The University of Michigan, Ann A M r , Michigan 48 109
Sequence-specific fragmentation for structural analysis has been generated by activation of ions via matrixassisted laser desorption/ionization (MALDI) in an ion trap storage/reflectron time-of-flightdevice (IT/reTOF). The key to this work is that ion decay can be induced by MALDI activation but requires an extended period of time to occur in large peptides. This extended decay period, which may be in excess of 20 ms, is provided in these experiments using the long storage times of the ion trap device. The ions are stored until decay is complete and then rapidly pulsed into a reflectron TOF for analysis. Since the ions decay within the trap, they ultimateiy appear as stable ions in the reTOF rather than as metastable decay peaks. The ion fragmentationwas found to depend strongly on laser power and the rf voltage placed on the ring electrode of the trap. The fragmentation obtained was shown to be similar to but Merent from that observed in FAB-low-energy CAD. In particular, enhanced fragmentation was obtained in the lower mass range and large species could be more easily fragmented than with FAB-low-energyCAD. The types of fragmentation for several target peptides are discussed. The advent of matrix-assisted laser desorption/ionization (MALDI) has resulted in the ability to detect and identify large
biomolecules based upon molecular weight in a mass spectrometer.1-6 MALDI is generally a soft ionization method where the molecular ion is produced intact allowing such identification of the molecule. This method in itself though has been thought to produce little fragmentation for structural analysis. Although MALDI has been combined with collisionally activated dissociation (CAD) to produce fragmentati~n,~ it becomes increasingly more difficult to do so as the size of the molecule increases.8 Thus, an important aspect of MALDI-MS has become the effort to find methods for generating sequence information on large molecules using this technique. (1) Karas M.; Hillenkamp, F. Anal. Chem. 1988,60, 2299-2301.
(2) Beavis, R. C.; Chait, B. T. Rapid Commun. Mass Spectrom. 1989,3, 436439. (3) Spengler, B.; Cotter, R J. Anal. Chem. 1990,62, 793-796. (4) Hillenkamp, F.: Karas, M.; Beavis, R C.; Chait, B. T.Anal. Chem. 1991, 63, 1193A-1198A (5) Chait, B. T.; Kent, S. B. H. Science 1992,257,1885-1894. (6) Spengler, B.: Kirsch, D.: Kaufmann, R. J. Phys. Chem. 1992,96, 96789684. (7) Cornish, T.; Cotter, R. J. Rapid Commun. Mass Spectrom. 1992,6, 242248. (8) Bunker, D. L.; Wang, F. M. J. Am. Chem. Sot. 1977,99, 7457-7459.
1400 Analytical Chemistry, Vol. 67, No. 8, April 75,7995
Although prompt fragmentation is not observed in the MALDI process on the time-of-flight (T0F)MS time scale, it has been shown that extensive metastable decay does occur as a function of time? Metastable decay may be quite extensive even for large ions and had been reported early on in TOF work for several desorption processes that generate large activated i o n ~ . ' ~ -Chait '~ and FieldloS1lfor example had studied extensive metastable decay that resulted in the plasma desorption of molecules such as bovine insulin. Cotter et a1.I2 used a delayed pulse-out technique for a linear TOFMS to demonstrate that metastable decay was produced in the plasma desorption or liquid SIMS process, but over an extended time period due to the lower predicted rates for unimolecular decomposition of large ions. The main problem with detecting metastable decay in a linear TOFMS is that if metastable decay occurs in the drift region after the ions have reached their halvelocity, the velocity corresponds to that of its parent ion and metastable fragment ions cannot be distinguished. Metastable decay must occur in the acceleration region of the TOFMS to be observed as a broadening of the ion packet with a tailing observed to longer flight times.'* The metastable tailing has been observed readily with small ions, which may fragment within the 1-2 p s that the ions remain within the acceleration region of the TOFMS.15-19In the case of large ions where fragmentation may require a signiscant amount of time to occur, such decay will not be readily observed. However, a reflectron (reTOF) device, which can correct for the difference in energy of fragments as they are generated in a time-of-flight device, has been used to distinguish metastable ions from parent ions.20~21~30 Indeed, quite extensive work has been performed on studying pathways in metastable decay of small fragments ions (9) Spengler, B.; Kirsch, D.; Kaufmann, R; Jaeger, E. Rapid Commun. Mass Spectrom. 1992,6, 105-108. (10) Chait, B. T.; Field, F. H. Int. I. Mass Specfrom.Ion Processes 1985,65, 169180. (11) Chait, B. T.; Field, F. H. Int. J. Mass Spectrom. Ion Phys. 1981,41, 17-29. (12) Demirev, P.; Olthoff, J. IC; Fenselau, C.; Cotter, R. J. Anal. Chem. 1987, 59, 1951-1954. (13) Ngoka, L.; Lebrilla, C. B.J. Am. Sot. Mass Spectrom. 1993,4, 210-215. (14) Syage, J. A: Wessel, J. E. Appl. Spectrosc. Reo. 1988,24, 1-79. (15) Kuhlewind, H.; Neusser, H. J.; Schlag, E. W.J. Chem. Phys. 1985,82,54525456. (16) Baer, T. Ado. Chem. Phys. 1986,64, 111-202. (17) Proch, R.; Rider, D. M.; &re. R. N. Chem. Phys. Lett. 1981,81, 430-434. (18) Durant, J. L.; Rider, D. M.; Anderson, S. L.; Zare, R N.J. Chem. Phys. 1984, 80, 1817-1825. (19) Lifshitz, C. Mass Spectrom. Reo. 1982,1, 309-348. (20) Boesl, U.; Neusser, H. J.; Weinkauf, R.; Schlag, E. W. J. Phys. Chem. 1982, 86, 4857-4863. (21) Spengler, B.; Kirsch, D.; Kaufmann, R Rapid Commun. Mass Spectrom. 1991,5, 198-202. 0003-2700/95/0367-1400$9.00/0 0 1995 American Chemical Society
generated by REMPI using reflectron energy a n a l y ~ i s . 2Stand~'~~ ing et al.24,25 first demonstrated the use of metastable decay of large ions as a method for obtaining sequence information of peptides using Cs+ ion desorption and reflectron energy analysis of the ions produced from metastable decay during the time of flight of the ions in the field-free region. In recent work, Spengler and KaufmannZ6demonstrated that the activation of ions produced in the MALDI process results in abundant long-lived metastable ions which can be observed using a reflectron energy analyzer. Fragment ions from metastable decay were mass analyzed by adjusting the ion reflectron in stages according to the kinetic energies of the ions. This process was shown to provide unique sequence information based upon the metastable decay according to which the structure of several peptides could be assigned. This procedure though must be done in stages since the reflectron energy analyzer can only correct for a given range of energies at any particular set of voltages. Thus, the reflectron must be tuned over several such energy ranges to obtain a full spectrum, requiring a rather lengthy procedure. A key point that resulted from the work of Kaufmann6was that this postsource decay had a distinctive unimolecular component where the decay in various molecules exhibited rate constants that implied times of tens of milliseconds. An obvious limitation of studying MALDI-activated decay in a reTOFMS is the limited flight time, which is often < 100 ps, so that decay may be far from complete on the time scale of the experiment. In the case of sector instruments or quadrupoles, the flight time is much shorter yet, so that fragmentation resulting from large ions activated by MALDI will not have sufficient time to be observed on these instruments. The use of ion traps (ITS) or ion cyclotron resonance (ICR) methods with storage times in the hundred of milliseconds might present opportunities for taking advantage of the full spectrum of long-lived metastable d e c a ~ . ' ~ ' ~ ~ In recent work by Fountain et alSz8it has been shown that a quadrupole ion trap/reTOFMS device can be used to study the decay of large ions following MALDI. It is shown that, by using the ability to store ions in the ion trap for variable amounts of time followed by analysis in the reTOFMS, fragmentation occurs over long periods of time, even tens of milliseconds in some cases. These fragments are stored and allowed to decay in the trap and are thus detected as stable ions in the reTOFMS, so that a complete spectrum can be obtained in 20 ms or less without scanning the energy analyzer. The ion decay was found to depend strongly on the laser power and on the radio frequency (I$ voltage applied to the ring electrode during the MALDI Similar results have also recently been observed by Chait et al.,29where (22) Kuhlewind, H.; Neusser, H. J.; Schlag, E. W. Int. J. Mass Spectrom. Ion Phys. 1983,51, 255-265. (23) Fountain, S. T.; Lubman,D. M. Anal. Chem. 1993,65, 1257-1266. (24) Tang, X.; Ens, W.; Standing, K. G.; Westmore, J. B. Anal. Chem. 1988,60, 1791-1799. (25) Tang, X.; Ens, W.; Standing, K G.; Mayer, F.; Westmore, J. B. Rapid Commun. Mass Spectrom. 1989,3, 443-448. (26) Kaufmann, R; Kirsch, D.; Spengler, B. Int. J. Mass Spectrom. Ion Processes 1994,131,355-385. (27) Cooks, R G.; Kaiser, R. E., Jr. Acc. Chem. Res. 1990,23,213-219. (28) Fountain, S. T.;Lee, H.; Lubman, D. M. Rapid Commun. Mass Spectrom. 1994,8,407-416. (29) Qin, J.; Steenvoorden, R. J. J. M.; Chait, B. T. 42nd ASMS Conference on Mass Spectrometry and Allied Topics, Chicago, May 1994; poster. (30) Huberty, M. C.; Vath, J. E.; Yu, W.: Martin, S. A Anal. Chem. 1993,65, 2791-2800.
MALDI was performed external to an ion trap and the ions drifted into the trap followed by a rapid rampup of the rf ring electrode voltage to 15 kV. Extensive fragmentation of MALDI-activated ions was observed in the ion trap that depended strongly on the rf voltage. In this work, we demonstrate the ability to generate sequencespecific fragmentation of ions activated by MALDI in an IT/reTOF device. The laser power and rf voltage have been optimized in this study to produce extensive fragmentation that reveals the structure of the target peptides. It is shown that such information can be obtained rapidly and that often a complete set of b and y series ions are observed which correspond to the cleavage of the peptide backbone. In addition, the storage time is set to be sufficiently long so that a complete decay spectrum is obtained on the time scale of the experiment. It is shown that characteristic b, y, and a series sequence ions are often produced and that the fragmentation often resembles that obtained by FAJ3-CAD at low energy, but is also distinctly different. The MALDI activation of ions in the IT/reTOF also provides extensive fragmentation results on larger ions than can normally be obtained by low-energy CAD methods, and the fragmentation of low-mass ions produced in the MALDI process is much more intense than that observed in lowenergy CAD. EXPERIMENTAL SECTION
The ion trapheflectron time-of-flight instrument used in this work and its applications to MALDI has been discussed in detail elsewhereSBA differentially pumped reflectron time-of-flight mass spectrometer (Model D850, R M. Jordan Co., Grass Valley, CA) was interfaced to a quadrupole ion trap storage device (Model C1251, R M. Jordan Co.), which replaced the conventional reTOF ion acceleration region. The ring electrode of the trap was crossdrilled so that a Macor sample probe tip entered the trap on one side and a laser beam entered the trap from the opposite side. The laser beam was collimated to a 1.1 mm2 spot on the probe using a converging/diverging lens system to a power density that was typically between 3 x lo6-1 x lo7 W/cm2. MALDI was performed using the third harmonic (355 nm) of a Nd:YAG laser. The resulting ions were stored in the QIT for a time sufficient to allow the full ion decay spectrum to be observed, i.e., at least 10 ms before the ions were extracted into the reTOF drift tube. The trapping time was never longer than 99 ms so that at the present 10 Hz repetition rate of the laser only the ions that resulted from the storage of one laser pulse were observed. A dc extraction pulse of +400 V with a 2 ps duration was applied to the back endcap electrode of the ion trap for ion ejection from the trap into the reTOFMS where the ions are mass analyzed. The detector was a 40 mm triple microchannel plate (MCP) detector with a gain of 107-108. The data were acquired using a LeCroy 9400A digital oscilloscope and processed using a Gateway 486 DX PC. The TOF spectra were calibrated by use of an internal standard, and the x axis was converted to a m/z scale using a PC spreadsheet. During the extraction event, the back end cap is pulsed to +400 V dc, while the !irst MCP detector stage is held at -3700 V. Thus, only 4100 V of total acceleration is applied to the ions. This value is relatively low compared to the 20 kV ion acceleration used in the work of Kaufmanne6The flight tube liner is held at -1650 V so that the ions are accelerated to only 2050 V in the flight tube. This relatively low acceleration voltage, as compared to the experiments of Kaufmann,6 limits the extent of postsource Analytical Chemistry, Vol. 67,No. 8,April 15, 7995
1401
I
a4
v2 b2
1
b4
I
+ z
1 f, f
Y
0
100
‘f
200
300
400
500
600
m/z
Figure 1. 1. MALDI-activated fragmentation spectrum of Leu-enkephalin with 10 ms trapping time, 1000 Vp-p rf level, and laser power density of 3.5 x lo6 W/cm2. An asterisk indicates mass difference of 17 u, Le., loss of ammonia. All the spectra shown represent an average of 500 single shots.
fragmentation via collision with the background gas in the flight tube. The rf power supply (Model D-1203, R M. Jordan Co.) provided a 0-4000 Vp-p, 1.0 MHz waveform applied to the ring electrode. The rf voltage could be varied to change the degree of fragmentation achieved. A helium buffer gas was introduced into the ion trap with a typical pressure of W4-W3Torr. A buffer gas is customarily used in ion traps in order to enhance storage via collisional relaxation to the center of the trap. The use of a buffer gas in the trap is especially important in MALDI experiments where the ions are produced with relatively high translational energy and must be collisionally cooled to be stored effectively. In previous work, the use of a buffer gas has been shown to be essential for enhancing the resolution of the IT/reTOF. However, as recently shown,28the presence of a buffer gas is important for production of ion fragmentation following MALDI in the trap. In this process the highly activated MALDI ions collide with the background gas in the trap and are further activated, resulting in long-lived decay in the trap. The pressure range cited above has been found to be optimal for these experiments. Although Nz and Ar buffer gas can be used, He has been found to provide extensive fragmentationZ8while the production of metastable decay during the extraction process is avoided. All peptides used in this study were purchased from Sigma Chemical Co. (St. Louis, MO) and were used without further purification. The peptides were dissolved in deionized water at a concentration of 1 mg/mL. The matrix used was a nearly saturated 2,5dihydroxybenzoic acid (DHB) solution prepared in 1:l (v/v) mixture of HZO/EtOH solvent with 0.1%trifluoroacetic acid. The peptide solution and the matrix were mixed at a 1:l volume ratio, and 3 pL of the mixture was placed on the Macor probe tip and air-dried. The total amount of sample placed on the probe corresponds to 3 nmol, which is a large amount of material for MALDI experiments. This amount of sample was 1402 Analytical Chemistry, Vol. 67,No. 8,Apriil 75, 7995
used typically to enhance the total fragment signal observed for analysis; however, only a very small amount of sample placed on the probe was actually consumed. RESULTS AND DISCUSSION
In this work we have optimized the experimental parameters of the MALDI process in the ITheTOF to obtain sequence-specifc fragmentation of peptides that could be used for structural analysis. As shown in previous work,28the parameters that affect the fragmentation in this experiment include the trapping time, trapping rf field, laser intensity, and pressure and type of buffer gas used. The use of the time variable is particularly significant since an extended time may be required for complete unimolecular decay to occur in these large molecules. In these experiments, the ions are stored in the trap for a sufficient time such that the decay process is complete. The rf trapping voltage was set at a level to produce the optimal fragmentation and mass trapping range. For most cases the maximum rfvoltage available of -4000 Vp-pwas found to enhance fragmentation. However, for smaller molecules such as Leu-enkephalin, a lower rf voltage was used in order to avoid loss of lower mass fragments due to the low-mass cutoff of the trap. The fragmentation was also very sensitive to the laser power, which might be expected since the MALDI process provides the initial activation of ions in the trap. The laser power was adjusted to produce extensive fragmentation for each particular sample. The buffer gas used was He at -1 x Torr, which provided fragmentation of these peptides in the mass range between 500 and 1400 u. As shown in previous the highly activated MALDI-produced ions are accelerated by the rf field into the background gas, where they are further activated and fragment. Thus, the extent of fragmentation achieved is influenced by these factors. Leucine Enkephalin. In Figure 1 is shown a MALDIactivated fragmentation spectrum of Leuenkephalin (r”-Gly-Gly-
Phe-Leu) in the IT/reTOF. Although high resolution has been demonstrated in the MALDI process using the IT/reTOF, the experimental parameters have been adjusted for producing an intense fragmentation pattern at the expense of resolution. The typical resolution for precursor ions is -250, while that for the product ions is 200. The [M HIf ion and its Na and K adducts are observed but are not the most intense peaks in the spectrum. In this case, a rather modest laser power density (3.5 x lo6 W/cm2> was required to achieve fragmentation. The spectrum shows a complete sequence of b, series ions and a,, series ions (except a3), which dominate the spectrum in intensity. A complete set of y, ions of lower intensity is also observed. In addition, several internal fragments of the peptides are produced with considerable intensity including the peaks at mlz 262, 234, 205, and 177. Gaskell et al.31reported the CID spectrum of Leuenkephalin on a hybrid BEqQ instrument and assigned the ions of m/z 262 and 205 to internal B-type fragments ((b4y4)3 and @4y3)2, respectively) incorporating the Gly-Gly-Phe and Gly-Phe residue, respectively. The ions at m / z 234 and 177 were attributed to internal A-type fragments ((a4y4)3 and (&y3)2) due to further loss of CO. Other major peaks include m / z 120,which is probably the immonium ion of phenylalanine. Also, m/z 107 is the fragment of the tyrosine moiety while that at mlz 91 corresponds to that of phenylalanine. The fragmentation mass spectrum of Leu-enkephalin has been studied by several other methods which can be compared to this work. For example, Alexander and Boyd32reported collisioninduced dissociation spectra of Leuenkephalin using FAB in a tandem hybrid mass spectrometer (BEqQ). In the low-mass range, their spectrum showed a smaller number of fragments and the fragments observed were of considerably lower intensity relative to the MALDI-activated spectrum. In the FAB-CAD work it was observed that the intensity ratio of Y2/b3 fragment ions was sensitive to the degree of internal energy of the precursor ions depending on He collision gas pressure with E m = 8 keV using QMIKES, though no calibration of the intensity ratio vs internal energy was made.32 However, in the MALDI experiments of the present work, the range of the ratio we could obtain by changing the laser intensity, rf voltage, or trapping time was very limited. The transition of the ratio reported by Alexander and Boyd, where b3 becomes greater than y2 as the He collision gas pressure increases, was not observed. The ratio of y2/b3 obtained in our study was typically -0.7, and it corresponded to the spectrum under the highest collision gas pressure used in their experiment, which is 2.6 x mbar (correspondingto 1.95 x Torr) of He. In a more recent paper by the same the ratio of aJb4 was used as an index of internal energy content, since this ratio provided a wider range of dependence on the experimental conditions. In their work, the ratio 1397/1425 (i.e., a4/b4) varied from 0.2 to 2.1, depending on the center-of-mass collision energy of 0-11 eV (Aror Xe) in an rf-only collision cell q of a BEqQ hybrid instrument. In our experiment, this ratio yields 0.8-3.4 as a function of laser power density and trapping time (see Figure 2). The ratio obtained has a tendency to increase as the laser intensity becomes higher and as the longer trapping time is employed. The
3.5
.
3 1
I 8
+
(31) Gaskell, S. J.; Reilly, M. H.; Porter, C. J. Rapid Commun. Mass Spectrom. 1988,2, 142-145. (32) Alexander, A. J.; Boyd, R IC Int. J. Mass Spectrom. Ion Processes 1989,90, 211-240. (33) Thibault, P.; Alexander, A. J.; Boyd, R IC;Tomer, IC B. J. Am. SOC.Mass Spectrom. 1993,4, 845-854.
n
0.5 -.
higher laser intensity is believed to impart more energy to the molecule, and the deposited internal energy is used to fragment the ions. This suggests that the internal energy accessible in the process is at least comparable to and may be considerably greater than that of the lowenergy collision-induced dissociation procedure. Alternatively, it may also mean that the trapping time in the trap provides a sufficient time window for the reaction sequence [M HI+ b4 a4 products to develop such a degree of fragmentation. Thus, the fragmentation pathway from b4 to a4 is more likely to proceed as the longer trapping time is employed, as can be seen in Figure 2. This fragmentation process is highly favored in this molecule, as reported by Gaskell et al.34 A direct comparison of the fragmentation obtained for Leuenkephalin by various methods is shown in Table 1. The fragmentation produced in MALDI-activated decay is similar to but different from that observed in FAB-CAD, for example. MALDI-activatedfragmentation resulted in a significantly greater number of low-mass fragments (~2 employing high-energy CAD in the literature resulted in abundant a and d ions almost exclusively,without ammonia loss satellite ions and internal fragments. However, a low-energy CAD spectrum42 revealed peaks due to ammonia loss and some internal fragmentation without any d series ions. In this respect, the MALDIactivated spectrum of substance P is rather similar to that of lowenergy CAD. Spengler et al?6,43reported MALDI-postsource decay (PSD) spectra of substance P, and the spectra showed far
more extensive fragmentation of the molecule including a, b, and y series ions and one d ion ( d d . The nontrivial spectral differences between MALDI-PSD and our work, though the same ionization technique (MALDI) was employed, may result from the large difference in acceleration voltage where greater than 10 kV was used in the work of S~englel.2~1~~ as compared to 2050 V in our work. The high-voltage acceleration may cause high-energy collisions with residual gas molecules that result in this more extensive fragmentation. In order to investigate the Arg effect, a MALDI-activated spectrum of substance P fragment 2-11 (des-Arg’wbstance P, Analytical Chemistty, Vol. 67, No. 8, April 75, 7995
1407
M+H
I
I,
0
1
b2
ll
v4 b5
I
I
1
200
400
600
I
800
I
I I
1000
1200
mlz
Figure 7. 7.MALDI-activated fragmentation spectrum of substance P fragment 2-11 with 10 ms trapping time, 2000 Vp-p rf level, and laser power density of 3.5 x lo6 W/cm2.
Pro-Lys-Pro-Gln-Gln-Phe-Phe-Gly-Leu-Met) was obtained (Figure 7). The low- and high-energy CAD spectra have been reported by other g r ~ u p s ~for ~ athe ~ des-Arg'-substance P. The only difference between substance P and substance P fragment 2-11 is that the latter does not contain the arginine residue at the N-terminus of the former molecule. The effect of absence of the arginine on the spectra is significant, as shown by comparing Figures 6 and 7. In Figure 7, abundant b series peaks are observed while a series ions are very weak in contrast to the case of substance P. The b series ions are complete except for bq, b3, and bl. Quite prominent C-terminal-specific ions (yg, y8, y4, y3) are also obtained. Abundant internal fragment ions are formed, and most of them are derived from y8 cleavage (including (b9y8)7, (b8y8)6, (b7y8)5, PQQF, and PQQ), which require bond breaking of Lys2-Pro3.The presence of these f r a g m e n t ~can ~ , be ~ ~explained by the proline e f f e ~ t , 3 ~as , * in ~ , the ~ ~ case of bradykinin. CONCLUSIONS
MALDI-activated fragmentation generally provides a rather extensive series of a-and btype ions that can be used for structural analysis. This method was demonstrated for several peptides between m / z 500 and 1400. The fragmentation pattern obtained was similar to that observed in FAB-CAD, but unique fragments were observed which are not observed in FAE-CAD with any intensity. In addition, much more intense fragmentation was observed in the lower m / z range. The ion decay for these large ions was found to require an extended period of time for completion, ranging from hundreds of microseconds for smaller
1408 Analytical Chemistty, Vol. 67, No. 8, April 15, 1995
peptides to many milliseconds for larger species. The ion storage capability of the ion trap provides the required time for complete decay in the trap followed by analysis using the reTOF. Since the decay occurs in the trap, the ions are detected in one complete spectrum as stable ions and the procedure that uses scanning of the reflectron to detect the entire range of metastable ions is not required. In addition, the fragmentation has been found to be strongly dependent on the laser intensity, the rf voltage on the ring electrode of the trap, and the storage time. These parameters have been optimized in this study for producing the most diverse fragmentation for structural analysis. The use of the ion trap as a front end storage source for the reTOF also has the potential to provide a means of performing MUMS in these experiments in order to obtain further structural information on the fragments generated by MALDI. A goal of future work will be to use the selective storage and excitation modes of the trap to obtain further fragmentation by CID of selected fragments in the trap with analysis by the reTOF. ACKNOWLEDGMENT
This work was supported by the National Science Foundation under Grant BIR-9223677 and by University of Michigan Jacobson and Sokol Fellowships to H.L. Received for review September 15, 1994. Accepted January 30, 1995.@ AC940921Q @Abstractpublished in Advance ACS Abstracts, March 1, 1995.
