Anal. Chem. 1994,66, 355-361
Spatial and Temporal Imaging of Gas-Phase Protein and DNA Produced by Matrix-Assisted Laser Desorption Theodore W. Heise and Edward S. Yeung' Department of Chemistry and Ames LaboratovUSDOE, Iowa State Universi& Ames, Iowa 500 11 The sensitivity of a transient imaging technique based on laserexcited fluorescence is shown to be within the levels required for studying matrix-assistedlaser desorption (MALD) of large biomolecules. Results show that film morphology, particularly film thickness, has a major influence on plume dynamics. Fluorescent labeling of protein and of DNA is used to allow imaging in UV-MALDgeneratedplumes. Evidence that intact molecules are monitored is provided. Dye molecules, proteins, and DNAs were found to behave very differently in the vaporization process. The spatial and temporal distributions are important for the design of secondary ionization schemes to enhance ion yields, for the optimization of ion collection, and for maximizing resolution in time-of-flight mass spectrometers. Matrix-assisted laser desorption (MALD) is a method192 which has shown great utility in mass spectral analysis of large biomolecules. The mild desorption process produces primarily intact molecular ions and is thus particularly useful for molecular weight determination. The MALD technique is not fully understood, but has rather been developed empirically. In order to better understand the dynamics of the process, more information about the fundamental mechanisms is needed. The quality and intensity of mass spectra (MS) are strongly affected by sample morphology. Sample preparation typically consists of mixing an analyte solution with a large excess of a matrix. The prepared sample can take many forms3ranging from rather homogeneous amorphous layers, closely packed microcrystallites, loosely scattered needles, or some combination of these. The variations are dependent on matrix type and sample preparation method. Samples often contain a higher matrix concentration in a ring around the edge with analyte dispersed more uniformly throughout the center? Films are typically rather heterogeneous. MS peak intensity varies widely with sampling positions with stronger signals coming from irradiation of film edges.4 In ref 5 , the authors also reported more concentrated samples (presumably also thicker) giving larger signals. For the UV-absorbing matrices, the role of the substrate appears to be of minor or no importance. This has been demonstrated by MALD fromcrystals of sample distributed on a plastic-coated electron microscope grid.6 (1) Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. In?. J. Mass Specrrom. Ion Processes 1981, 78, 53-68. (2) Bcavis, R. C.; Chait, B. T. Rapid Commun. Mass Spectrom. 1989,3,233237. (3) Karas, M.; Bahr, U.; Giessmann, U. Mass Spectrom. Rev. 1991,10,335-357. (4) Doktycz, S.J.; Savickas, P. J.; Krueger, D. A. Rapid Commun.MassSpecrrom. 1991, 5, 145-148. ( 5 ) Salehpour, M.; Pcrera, I.; Kjellberg,J.; Hedin, A.; Islamian, M. A.; Hikansson, P.; Sundqvist, B. U. R. Rapid Commun. Mass Specrrom. 1989.3, 259-263. QQQ3-27QQl94lQ36~Q355~Q4.~QlQ
0 1994 American Chemical Society
The MALD process is known to be a bulk process rather than a monolayer phenomenon, as would be implied when surface scientists speak of de~orption.~ Vertes and co-workers have set forth a model they term the homogeneous bottleneck? Johnson and Sundqvist9 have compared MALD to fast-ioninduced sputtering incorporating two components they termed a thermal spike and a pressure pulse. Rapid heating seems to be a vital component of any model. For LD, it has been said that rapid heating desorbs molecules before they have time to decompose.1° The nanosecond lasers typically used in UV-MALD can easily produce 101o-lO1l K/s surface heating rates.' Spallation, the ejection of large relatively cold pieces of material, has been indicated as occurring, especially for thicker films." For thinner films, the MALD process appears to be more of an ablation process producing diffuse ejecta with a forward peaked angular distribution. Beavis and Chait, on the other hand, maintain that production of gas-phase ions cannot be the result of removal of large pieces of cold solid material.12 For the more general case of laser desorption in the absence of a matrix, some authors have proposed nonequilibrium processes such as explosive desorption13or a shock wave.14 There is no immediately obvious reason why these processes might not also apply to MALD. More recently, the concept of shock heating of the matrix material has been emphasized.11 The importance of a shock wave type of mechanism seems to be more pronounced in the case of backside desorption geometry.l5 These workers also make reference to the concept of a pressure pulse and relate the process to the shock wave transfer of momentum seen in plasma desorption MS. One can also refer to the MALD process as a nonequilibriumphase transition.1° A significant amount of effort has been directed toward determining ion velocities. There appears to be a significant (6) Vertes, A,; Balazs, L.; Gijbels, R. Rapid Commun. Mass Specrrom. 1990,4, 263-266. (7) Sundqvist, B. U. R. In?. J. Mass Spectrom. Ion Processes 1992, 1181119, 265-281. (8) Vertes, A.; Gijbels, R.; Levine, R. D. Rapid Commun. Mass Specrrom. 1990, 4, 228-233. (9) Johnson, R. E.; Sundqvist, B. U. R. Rapid Commun. Mass Spectrom. 1991, 5 , 574-578.
(10) Vertcs, A. In Methods and Mechanisms for Producing Ions from Large Molecules; Standing, K. G., Ens, W., Eds.; Plenum Press: New York, 1991; pp 275-286. (1 1) Williams, P.; Nelson, R. W. In Methods and Mecham'smsfor Producing Ions from Large Molecules; Standing, K. G., Ens, W., Eds.; Plenum Press: New York, 1991; pp 265-273. (12) Bcavis, R. C.; Chait, B. T. In Merhods and Mechanisms for Producing Ions from Large Molecules; Standing, K. G., EM, W., Eds.; Plenum Press: New York, 1991; pp 227-234. (13) Fain, B.; Lin, S. H. J. Chem. Phys. 1989, 91, 2726-2734. (14) Lindner, B.; Seydel, U. Anal. Chem. 1985,57, 895-899. (1 5) Bitensky, I. S.;Goldenberg, A. M.; Parilis, E. S.In Merhods and Mechanisms for Producing Ions from Large Molecules; Standing, K. G., Ens, W. Eds.; Plenum Reas: New York, 1991; pp 287-291.
Analytlcal Chemism, Vol. 68, No. 3,Febnrety 1, 1994 355
difference between radial and axial velocities. Ens and coworkers have reported radial velocity spreads -20 times less than axial.I6 Reported axial velocities range from as high as 1500 m/s,17 to as low as 20 m/s,18 with most values falling somewhere between these two extremes.’ 1119 Most evidence indicates that axial velocities are independent of analyte molecular weight.20 Sundqvist7 has concluded that the axial velocity of the ions is largely a function of supersonicexpansion of the forward ejected matrix molecules. There is belief that the dominant mechanism is proton transfer,Zl but this is far from proven. There is evidence that a significant portion of desorbed analyte in MALD appears as neutral species.22 It has been shown that ionization efficiency falls off sharply outside a narrow range of laser irradiances. The immediate conclusion is that a secondary ionization scheme would help increase sensitivity for MS applications substantially. Further, optimization of the desorption step itself can be most readily achieved if it is decoupled from the ionization step.23 It is clear that knowledge of the spatial and temporal dynamics of the laser-desorbed plume would be most helpful in designing suitable secondary ionization schemes, especially if the neutral molecules can be monitored. In addition, knowledge of plume dynamics might provide insight into the fundamental processes occurring in MALD and thereby help to elucidate the operative desorption mechanisms. If fragmentation involves destruction of the chromophore, one can even measure the extent of f r a g m e n t a t i ~ n . ~We ~ . ~describe ~ here the use of laser-excited f l u o r e ~ c e n c eto~ provide ~ spatial and temporal maps of large biomolecules in plumes generated by MALD. The overall scheme is similar to that designed for absorption p r o b e ~ , 2but ~ , ~with ~ the high sensitivity needed for MALD events.
EXPERIMENTAL SECTION Laser Desorption. A nitrogen laser (Lambda Physik, EMG 101) was used for sample desorption with backside desorption geometry. It has been shown that laser-generated plumes can significantly attenuate the laser energy delivered to the surface if irradiated from the front.27 It is also thought that backside desorption tends to minimize plasma effects due to interaction of the desorption laser with the vaporized sample. A concave mirror was used to focus the laser beam to a line. Measurements from burn spots on photographic film and on sample films showed the desorption pulse to be focused to 0.5 X 6 mm, or an area of 3 mm2. There were several reasons for using this configuration. First, this focusing arrangement ~
(16) Ens, W.; Mao, F.; Mayer, F.; Standing, K. G.Rapid Commun. Mass Spectrom. 1991,5, 117-123. (17) Pan, Y.; Cotter, R. J. Org. Mass Spectrom. 1992, 27, 3-8. (18) Hansen, S.G. J. Appl. Phys. 1989, 66, 3329-3336. (19) Beavis, R. C.; Chait, B. T. Chem. Phys. Lett. 1991, 181,479484. (20) Zhou, J.; Ens, W.; Standing, K.G.; Verentchikov, A. Rapid Commun. Mass Spectrom. 1992, 6, 671678. (21) Stecnvoorden,R. J. J. M.; Weeding, T. L.; Kistemaker, P. G.; Boon, J. J. In MeihodsandMechanisms for Producinglomfrom Large Molecules;Standing. K. G., Ens, W., Eds.; Plenum Press: New York, 1991; pp 315-323. (22) Van Vaeck, L.;Van Roy, W.; Gijbels, R. Analusis 1993, 21, 53-75. (23) KinseLG. R.; Lindner, J.; Groterneyer,J.;Schlag, E.W. J. Phys. Chem. 1991,
95,7824-7830.
(24) Heisc, T. W.; Yeung, E. S.Anal. Chem. 1992,64, 2175-2179. (2s) Kimbrcll, S.M.; Yeung, E. S.Appl. Spectrosc. 1991, 45, 442-447. (26) Yappert, M. C.; Kimbrell, S. M.; Yeung, E. S.Appl. Opt. 1987, 26, 35363541. (27) Patel, R. S.;Brewster, M. Q. J. Heat Transfer 1990, 112, 170-177.
356 Ana&ticalChemistry, Vol. 66, No. 3, February 1, 1994
I
&am
U
I
0‘2
was found to give the most uniform beam profile. Second, a line was used rather than a point to increase the path length of the probe beam within the plume for the fluorescencestudies. Finally, this produces spatial maps without the need for deconvolution inherent to cylindrically symmetric systems. The laser was operated at low repetition rates (-0.5 Hz) with a shutter manually opened to admit a single laser pulse. Therefore each set of data collected was for a single desorption event. It should be stressed here that, unlike the majority of MALD work, no signal averaging was done. Laser fluence was varied by placing glass plates at 45O angles in the incident beam path. This allowed up to 1 order of magnitude beam attenuation in 15-20% increments. Desorption of samples was carried out at atmospheric pressure without any special enclosures. The pulse energy was monitored by directing a partially reflected portion of the desorption beam to an energy probe (Laser Precision, Utica, NY, Models Rj-7200 and Rjp734). Typical pulse energies ranged from 1 to 4 mJ at the desorption site with pulse widths of -20 ns. Samples were prepared on quartz plates. Quartz absorbs minimally at the desorption wavelength so MALD can be studied independent of substratecontributions. The efficiency of sample desorption was checked by visual inspection of the films using a microscope. Instrumental Arrangement. The probe setup is shown in Figure 1. The laser-generated plume evolved vertically out of the plane of the sample plate and was excited with an argon ion laser (Laser Ionics, Orlando, FL, Model 554A) lasing at 488 nm. Two mirrors were used to direct the probe beam into the first acoustooptic deflector (AOD). The first AOD (Isomet, Springfield, VA, Model 1205-C) scanned the probe beam in the horizontal ( x ) direction. The output from the first AOD was sent through a second AOD (IntraAction, Belwood, IL, Models ADM- 150 and DE-1 50) which scanned the probe beam in the vertical (y) direction. The AODs were driven by asymmetric triangular waveforms at 200 kHz vertically and at 10 kHz horizontally (Wavetek Corp., San Diego, CA, Models 162 and 182A, respectively). The waveform asymmetries produced unidirectional scans with periods of 5 ps vertically and 100 ps horizontally. During the last 5% of each sweep, the beam was reset or returned to its starting position. Thevertical waveform generator was gated with TTL output pulses from the
10 rnrn
cprobe region (focal plane)
a
Y
t,
Torrance, CA) were used to discriminate against scattered probe light. Finally, an interference filter was used to complete wavelength discrimination. A variety of interference filters (IO-nm bandwidth) were used including 520, 550, and 577 nm. The specific filter was selected to yield the maximum fluorescence signal for each sample. The PMT output was terminated at 1 kn and sent to a digital oscilloscope (LeCroy, Chestnut Ridge, NY, Model 9410). The digitizer was operated at 4 MHz with 8 bits of resolution. For each desorption event, 8000data points were recorded. This allowed 20 sequential 100-ps two-dimensional images to be collected. Therefore, each plume was monitored for a total of 2 ms. Data were dumped to an IBM-AT microcomputer through an IEEE-488 interface (Capital Equipment Corp., Burlington, MA, Model PC-488) for subsequent evaluation. Data acquisition was triggered in the single-event mode by the SYNCH OUT signal from the desorption laser and commenced 12 ps after the desorption event. After data manipulation, contour plots of fluorescence intensities can be constructed, as shown in Figure 2b. The AODs efficiency is not necessarily constant over the length of a scan, with the efficiency falling off at the ends. To allow for correction of AOD efficiencies, a portion of the postplume probe beam was reflected to a 1 cm X 1 cm photodiode (Hamamatsu, Middlesex, NJ, Model S1790-01). Normalization of the PMT signal with respect to the reference photodiode signal showed no detectable change from the raw signal with proper adjustment. We therefore assumed that the scanning efficiency was constant over the probe region of these two-dimensional maps. Reagents. Matrices were chosen based on previous utility in UV-MALD and were obtained from Aldrich. All matrices were used as received. Ferulic, caffeic, and gentisic acids were all studied. Ferulic acid gave the best results and was used for all the work described in this paper. The dye was chosen on the basis of its molar absorptivity at 337 and at 488 nm. Specifically it was desired to use a dye with a small absorption coefficient at the desorption wavelength and a large absorption coefficient at the fluorescence excitation wavelength. The dye chosen was coumarin 6 obtained from Eastman. All solutions were prepared in methanol. Bovine serum albumin (BSA) was chosen as a representative protein. It has an intermediate molecular mass (-66 kDa) and is easily within the capability of MALD. It is also readily available and can be purchased with fluorescent labels already attached, e.g., sulforhodamine 101 (TR-BSA), fluorescein isothiocyanate (FITC-BSA), and tetramethylrhodamine isothiocyanate (TRITC-BSA). Unlabeled BSA was used as a blank. All were obtained from Sigma and used as received. Protein solutions were prepared at 1 g/L in 0.1% trifluoroacetic acid (TFA). The matrix solution was 0.2 M ferulic acid in 5050 methanol/water. Working solutions were prepared by mixing 200 pL of protein solution with 2 mL of matrix solution. Aliquots (100 pL) of this mixture were cast directly onto quartz plates and dried under a gentle stream of air at room temperature. The DNA used was degraded free acid from herring sperm obtained from Sigma. The molecular mass was in the 10-30 kDa range. The dye reagent was TOTO-1 from Molecular Probes. TOTO-1 is an intercalating dye which exhibits a
-
Figure 2. (a) Scan pattern of the probe laser at the Image plane and (b) graphical representatlon of the signal derlved from a plume withln the probe region.
horizontal waveform generator. The horizontal waveform generator was itself gated by the SYNCH OUT signal from the N2 laser. The sweep pattern of the probe beam is shown in Figure 2a. In order to map the largest area possible, a long-focallength (200 cm) lens was used to focus the probe beam. It has been documented28that AODs can have significant cylindrical lensing effects when operated at high frequency. The lower frequency (horizontal) AOD has an inconsequential focal length of more than 30 m. However, the higher frequency (vertical) AOD has a focal length of -1.7 m, which is significant in comparison to the 200-cm lens. Because the AOD lensing characteristics are different in each dimension, the optimum focal lengths for the two scanners are not the same. Empirically, the best resolution was obtained with the spherical lens (Ll) positioned 120 cm from the probe region. Resolution was determined by measuring the probe beam intensity transmitted through a grid at the focal point. The photodiode measurements demonstrated a resolution of about 20 X 24 spots in a 16 mm vertical X 10 mm horizontal scan. Naturally, optimization between the resolution and the image size can be changed by proper focusing. An aperture was placed in the beam path to block out the zeroth and second orders from the AODs. A 10 cm diameter X 10 cm focal length spherical lens (L2) was used to collect the fluorescence signal from the entire probe region onto a photomultiplier tube (PMT; Products for Research, Danvers, MA, Model 56 AVP). To minimize scattered radiation, the PMT was placed normal to the probe beam. In addition, a 500-nm cutoff filter and a 488-nm holographic Raman edge filter (Physical Optics Corp.,
-
(28) Lckavich, J. Lusers Appl. 1986, 4, 59-64.
Ana~lcalChemlstty, Vol. 66, No. 3,February 1, 1994
357
1000-fold enhancement of fluorescenceefficiency upon binding with DNA. Reagents were used as received. DNA was dissolved in tris-borate buffer, pH 8.4, at 1 mg/mL. To 1mL of this solution was added 31 pL of stock dye solution. After incubation for 30 min, the mixture was diluted 1:lO with 0.2 M ferulic acid (5050 methanol/water). Aliquots (100 pL) of this sample solution were cast directly onto quartz plates and dried under a gentle stream of air at room temperature. A sample without dye and a sample without DNA were prepared in the same manner for use as controls. Samples were always prepared on 2 X 2 X l/16 in. quartz plates. Spin-coating from alcoholic solutions was found to give the most uniform films. This procedure was capable of producing uniform films up to a maximum thickness of 700 nmol/cm*of the matrix. The spin-coater was home-built and operated at 800 rpm. It produced films which were essentially circular and quite uniform. Casting of films in air without spinning was used to obtain even thicker films, but at the expense of film uniformity. The air-cast films had visibly thicker crusts around the circumference. Film thicknesses were determined from the concentrations and volumes of solution deposited. With this knowledge and the assumption of uniform film distribution, measurement of the area of the film allowed calculation of the film thickness.
-
RESULTS AND DISCUSSION Film thicknesses studied varied over a range from several hundred nanometers to several micrometers thick. These estimates were based on the assumption of uniform matrix distribution and a density estimate of -1 g/mol. Film morphology varied tremendously with sample preparation method, analyte, and matrix. In agreement with most of the literature reports, a thicker ring of sample formed at the edge of cast films. Films formed by spin-coating did not exhibit this edge effect. Although microscopically they were too thin to provide much information, macroscopically they were extremely uniform. Figure 3 depicts part of a 20-frame run for coumarin 6 desorbed from a thin film of ferulic acid matrix. The numbers refer to successive frames 100 ps apart starting 12 ps after the desorption pulse. The axes are marked with I-mm divisions. In the first frame one sees mainly the signal from sample fluorescence excited by the desorption laser. It is not correlated with the probe position or with the probe intensity and can be considered an unwanted signal which gradually decays. If plume fluorescence excited by the probe laser is present, it is not observable above the large background. By the second frame, the initial fluorescence has largely decayed and formation of the plume is seen to begin. The third frame shows motion of the plume away from the sample surface along with increasing intensity. By this time (-250 ps), the velocity of the center of the plume away from the surface has slowed. The plume does continue to grow in intensity throughout all 20 frames. One also observes diffusion of the plume throughout the remaining frames. The plume in Figure 3 was obtained from a relatively thin film of sample. Plume dynamics was found to vary considerably with film thickness. Plumes generated from thicker films are also studied at a fixed laser irradiance. For an intermediate film of 300 nmol/cm2, one immediately notes 358 Ana&tIcaIChemIstfy, Vol. 66,No. 3, February 1, 1994
Flgure 3. Selected sequential Images of a MALO plume. Condklons: 5 fmol of coumarln 6 desorbed; 30 nmol/cm2ferulic acid matrix: 2000fold molar excess.
that, unlike the thin film, the plume is visible in the first frame. This is due in part to the significantly greater amount of material desorbed. Because of the increase in fluorescence signal, the PMT gain could be reduced. The result is an effective increase in discrimination against fluorescence induced by the desorption pulse. The fact that the observed signal increased with increasing film thickness shows that attenuation of the laser beam in the thicker films is not important. The second obvious difference from the thin film is that much more rapid movement of the plume away from the sample surface is evident. By frame 3 the plume center is 9 mm above the surface. This is over twice as far as the thin-film plume traveled in the same time period. Yet another difference is the degree of lateral dispersion. While the thinnest film appeared to disperse horizontally in roughly the same degree as vertically, the thicker film is somewhat more focused vertically. A similarity is that by frame 3 the rapid vertical motion has ceased. As in Figure 3 the plume continues to move away from the surface, but at a reduced rate. The thickest coumarin 6/ferulic acid film is shown in Figure 4. The uncertainty in film thickness is due to the presence of thicker crusts at the circumference of this air-dried film. Here the initial rapid motion is even more pronounced. It does not subside until frame 6 , where the plume center is located 14
1.21
OOl
2
4
6
8 1 0 12 14 1 6 FRAME NUMBER
1's 20
Figure 5. Normalizedfluorescence by frame for backsidedesorption. Conditions: coumarin 6 In ferulic acid matrlx; 30 (*), 300 (+), and 1000 nmol/cm2 (X); FITEBSA (A)in ferulic acid matrix, 1000 nmol/cm2.
-
-
Table 1. CompoOniaf~of Sample Flknr lor Fluorescence Imaglng
sample bg) matrix (mg) mass ratio molar ratio
coumarin 6 FITC-BSA TR-BSA TRITC-BSA BSA DNA-TOTO
Figure 4. Selected sequential images of a MAID plume. Conditlons: -500 fmol of coumarin 6 desorbed; -1000 nmol/cm2 ferulic acid matrix; 2000-foM molar excess.
mm above the surface. Again, the plume appears to show less lateral diffusion than the thin-film case. From visual inspection of the sample slide, backside desorption typically resulted in virtually complete sample removal. It was observed, however, that a second laser shot on the same spot would often generate a fluorescence signal. It is interesting that the vertical motion of the plume then appears the same as the thin film. The horizontal diffusion is not as great, however. It is possible that the edges left from the first shot serve to focus the second plume. It is also possible that this plume arises from sample residue which recondensed from the first desorption event. Other possible explanations are that the slide was not totally cleaned with the first shot or that the plume is generated from the edges of the remaining sample. The edges, being only 0.5 mm apart, would not have generated a resolvable bimodal plume in these plots. The amount removed can be estimated based on the relative fluorescence intensity. By calculating the spatially integrated fluorescence for each frame, one can get a sense of the relative fluorophore populations as a function of time. Figure 5 shows the normalized fluorescence (relative to the largest integrated value in each series) by frame for the threecoumarin 6/ferulic acid films. For the thick film, fluorescence intensity grows to a maximum at -600 pus. This agrees well with the time
0.35 14 40 74 80 10
0.39 3.49 1.94 1.94 1.94 0.35
1100 250 50 26 24 35
2000 86OOo
17000 9ooo
8300 3600
of maximum absorbance signal in previous ~ o r k , *showing ~,~~ that the rates of production of the vaporized materials are comparable. As film thickness decreases, however, the maximum fluorescence signal takes longer to develop. We note that all of these velocities (tens of meters per second) are representative of plumes retarded by collisions in atmosphericpressure gas and are therefore substantially slower than velocities observed in vacuum.'* The shorter time in achieving maximum fluorescence and the greater distance traveled at early times suggest that the fluorophores are desorbed from the thicker films with greater velocities. Dye labeling allowed the collection of images of MALDgenerated protein plumes with good signal-to-noise ratios. Sample morphology of these films was quite heterogeneous, with lots of flakes protruding from the surface. Actual amounts of sample deposited as well as calculated mass and molar ratios are presented in Table 1. Fluorescence intensity from shot to shot was extremely variable. The largest signals tended to come from the thickest portions of the sample, particularly from larger crystallites which were intimately connected to the quartz surface. MALD desorption of unlabeled BSA produced no fluorescence signal above background. Figure 6 shows images obtained from a TRITCBSA sample. These images were collected with a 577-nm interference filter. Other tagged proteins were also studied. To accommodate the different emission spectral profiles, a 550-nm filter was used for TR-BSA and a 520-nm filter for FITC-BSA. There does not appear to be any significant difference between the various labeled proteins. This is not too surprising and bears out the supposition that dye labeling has little influence on the protein desorption. It is interesting to compare the dynamics of the plumes of proteins with those of the dye films at different thicknesses. Although the protein films correspond closely to the thickest dye film in actual thickness and are prepared in a similar manner, the vertical motion of the plume center resembles Analytical Chemistry, Vol. 66, No. 3, February 1, 1994
359
Flgure 6. Selected sequentlei Images of a M A D plume from TRITCBSA In a ferullc acid matrlx (- 1000 nmol/cm*).
Figure 7. Selected sequential Images of a MALD plume from DNATOTO-1 In a ferullc acid matrix (- 1000 nmol/cm*).
that of the thin dye film. The time dependence of the integrated fluorescence for FITC-BSA also seems to be similar to that of the thin dye film, as shown in Figure 5 . In fact, continued production of fluorescent material from the surface creates the illusion that the plume actually moves back toward the surface in the later frames of Figure 6. If one compares the horizontal diffusion, however, the proteins resemble the thick dye films. That the dynamics of protein films and dye films are different is evidence that the dye label did not dissociate from the protein during vaporization. This of course is necessary to provide insight into the behavior of the protein in MALD. Although the signal-to-noise ratio was rather poor, useful images of MALD-generated DNA plumes were obtained. These results are presented in Figure 7. Here the plume appears to move away from the surface much more rapidly than the protein plumes. One explanation is that the DNAdye complex does not remain intact throughout the desorption process, thereby producing fluorophores that are free dyes that would behave like coumarin 6. To test this hypothesis, we investigated the spectral dependence of the observed fluorescence. A PMT signal was observed at 520 nm from the MALD of all samples, including unlabeled DNA, DNAdye, and dye alone in identical matrices and at identical pulsedlaser energies. This is likely due to incomplete discrimination
against scattering of the probe laser by thevaporized material. At 550 and 577 nm no signal was observed from unlabeled DNA. At 550nm signal was seen from the DNA-dyecomplex only. At 577 nm both the dye and the DNA-dye complex gave signals. To help determine whether matrix interactions might affect the spectral characteristics of fluorescence, solution fluorescence spectra of the various combinations were collected. Fluorescence intensities were determined for constant dye concentrations. The results are presented in Figure 8. Maximum fluorescence intensities (a:b:c:d = 5 : 10:100:30) confirmed that there was a fluorescence enhancement upon binding, but under our experimental conditions it was lower than the 3 orders of magnitude claimed by the manufacturer. The presence of matrix doubled the fluorescence signal of dye, but quenched that of the complex. Referring to Figure 8, one can see that the matrix also shifts the emission wavelength of the DNA-dye complex. In the gas phase, it is likely that matrix interactions are minimal. One would thus be comparing spectra c and a at a 1005 intensity ratio. Although it is possible that the DNA-dye complex is not preserved upon desorption, the fact that the DNA-dye plume yields a signal at 550 nm while dye alone does not strongly suggests that the complex remains intact to produce the images
360 Analyticel Chemistry, Vol. 66,No. 3, February 1, 1994
1.2
1.o
2 0.4 0.2 0.0 450.0
I
Q.0
550.0
600.0
650.0
7( 0
WAVELENGTH (nm)
Flgurr 8. Normalized fluorescence spectra (488m excltatlon) of (a) 1010-1, (b) TOTO-l/ferullc acM, (c) DNA-TOTO-1 complex, and (d) DNA-TOTO-Vferullc acid.
in Figure 7. The fact that it might do so should not be surprising with the soft nature of MALD desorption. The question remains as to why DNA and protein behave so differently in the MALD process. The molecular weight ranges of these samples are not widely different. One hint comes from the morphology of these samples. As indicated above, the protein films appear inhomogeneous, with platelike grains interdispersed in the matrix. The DNA films are also inhomogeneous but, because of the lower concentration used, disperse in a finer and more even manner in the matrix. Quantitative assessment of the extent of inhomogeneity however was not possible. So, if the driving force is the matrix molecules moving away from the surface, one would expect the DNA to be carried more effectively along to approach the velocity of the matrix material. Another explanation is that proteins possess relatively well defined tertiary structure while DNAs resemble random coils of strands without intramolecular adhesion due to the uniform charge distribution. When these are carried in a flow stream of matrix molecules, the DNA can relax by stretching out as a long strand. This natural adaptation to the direction of flow can cause the DNA to appear as a small molecule, i.e., with a small cross section for (29) Spengler, B.; Kirsch, D.; Kaufrnann, R.Rapid Commun. Mass Specrrom. 1991, 5, 198-202.
retardation due to collisions with ambient air. The velocity and the spatial distribution of MALD plumes are important parameters in the design of MS experiments to achieve the highest resolution and to enhance the effectiveness of postionization. It is not clear from these results alone whether the observations are related to the substantially different sensitivities in MS for proteins and for DNAs. Although the present study involves plumes in a normal atmosphere, adaptation to measurements in vacuum (e.g., inside a mass spectrometer) should be straightforward. Without collisions with the ambient gas, the velocities are expected to be substantially higher.1620 This implies that the frame rate should be increased or the size of the imaged area should be enlarged. Signal levels will be lower due to dilution, but the background scattering will be minimized in a vacuum. It will be most interesting to correlate the spatial images with mass spectra for the same MALD event. Since the integrated concentration is available for each time segment, one can potentially assess the rate of desorption and the rate of fragmentati~n;~if any. If in the future information about the spectroscopicdifferences between the neutral and the ionic forms become available, one can then determine the fraction of the desorbed analytes that is ionized at each point. Such information will provide insight into whether ionization is directly associated with the desorption step or whether it is a result of subsequent collisions with the matrix gas (e.g., proton transfer). Similar insights can be derived about the mechanism of fragmentation.
ACKNOWLEDGMENT The Ames Laboratory is operated for the U.S. Department of Energy by Iowa State University under Contract W-7405Eng-82. This work was supported by the Director of Energy Research, Office of Basic Energy Sciences, Division of Chemical Sciences and the Office of Health and Environmental Research. Additional support was provided to T.W.H. as a fellowship by Iowa State University, College of Liberal Arts and Sciences. Recehred for review August 9, 1993. Accepted November 10, 1993." -
~
~~~~
* Abstract
published in Aduance ACS Abstracts, December 15, 1993.
Ana~IcaIChernIstry,Vol. 66, No. 3, February 1, 1994
901
