Improving Spot Homogeneity by Using Polymer Substrates in Matrix

May 5, 2001 - Chemistry is quantum computing's killer app. In an IBM lab in Yorktown Heights, N.Y., a circuit board sits at the bottom of a large tub ...
0 downloads 13 Views 101KB Size
Anal. Chem. 2001, 73, 2617-2624

Improving Spot Homogeneity by Using Polymer Substrates in Matrix-Assisted Laser Desorption/ Ionization Mass Spectrometry of Oligonucleotides Yongseong Kim,†,‡ Gregory B. Hurst,*,‡ Mitchel J. Doktycz,§ and Michelle V. Buchanan‡,§

Chemical and Analytical Sciences Division and Life Sciences Division, Oak Ridge National Laboratory, Oak Ridge Tennessee 37831-6365

We describe a method for improving the homogeneity of MALDI samples prepared for analysis of small, singlestranded oligonucleotides using the widely used DNA matrix system, 3-hydroxypicolinic acid/picolinic acid/ ammonium citrate. This matrix system typically produces large crystals around the rim of the dried sample and requires tedious searching of this rim with the laser. However, when a substrate is prepared using both Nafion and a hydrophilic, high-molecular-weight polymer, such as linear polyacrylamide, linear poly(ethylene oxide), or methyl cellulose, oligonucleotide-doped matrix crystals tend to be smaller and more uniformly distributed across the entire spot, thus decreasing the time that is required for locating a usable signal. In addition to MALDI characterization of the spatial distribution of “sweet spots,” fluorescence microscopy allows for imaging dye-labeled DNA in dried MALDI spots. The mechanism of enhanced uniformity may involve increased viscosity in the MALDI sample droplet due to partial solubilization of the substrate by the MALDI sample solvent as well as partitioning of the matrix or DNA between the solvent and the undissolved portion of the polymer substrate. The use of matrix-assisted laser desorption/ionization timeof-flight mass spectrometry (MALDI-TOFMS)1,2 for size analysis of oligodeoxyribonucleic acids (DNA) has been demonstrated3-9 for a variety of applications ranging from the analysis of synthetic oligonucleotides,10-17 hybridized or polymerase-extended normal * Corresponding author. Phone: (865) 574-7469. Fax: (865) 576-8559. E-mail: [email protected]. ‡ Chemical and Analytical Sciences Division. § Life Sciences Division. † Present address: Division of Chemistry and Chemical Engineering, Kyungnam University, Masan, Kyungnam, South Korea 631-701. (1) Hillenkamp, F.; Karas, M.; Beavis, R. C.; Chait, B. T. Anal. Chem. 1991, 63, 1193A-1203A. (2) Gross, J.; Strupat, K. Trends Anal. Chem. 1998, 17, 470-484. (3) Fitzgerald, M. C.; Smith, L. M. Annu. Rev. Biophys. Biomol. Struct. 1995, 24, 117-140. (4) Limbach, P. A. Mass Spectrom. Rev. 1996, 15, 297-336. (5) Murray, K. K. J. Mass Spectrom. 1996, 31, 1203-1215. (6) Guo, B. Anal. Chem. 1999, 71, 333R-337R. (7) Nordhoff, E.; Kirpekar, F.; Roepstorff, P. Mass Spectrom. Rev. 1996, 15, 67-138. (8) Monforte, J. A.; Becker, C. H. Nature Med. 1997, 3, 360-362. (9) Lubman, D. M.; Bai, J.; Liu, Y.-H.; Srinivasan, J. R.; Zhu, Y.; Siemieniak, D.; Venta, P. J. In Mass Spectrometry of Biological Materials. Larsen, B. S., McEwen, C. N., Eds.; Marcel Dekker: New York, 1998; pp 405-434. 10.1021/ac001392v CCC: $20.00 Published on Web 05/05/2001

© 2001 American Chemical Society

or modified oligonucleotide probes,18-23 and short-sequence ladders24-29 to restriction enzyme digestion fragments30-33 and polymerase chain reaction (PCR) products34-40 as large as 2000 bases.41 Although capillary and microchip electrophoresis systems show great potential for high-throughput analysis of DNA se(10) Spengler, B.; Pan, Y.; Cotter, R. J.; Kan, L. Rapid Commun. Mass Spectrom. 1990, 4, 99-102. (11) Parr, G. R.; Fitzgerald, M. C.; Smith, L. M. Rapid Commun. Mass Spectrom. 1992, 6, 369-372. (12) Tang, K.; Allman, S. L.; Jones, R. B.; Chen, C. H.; Araghi, S. Rapid Commun. Mass Spectrom. 1993, 7, 435-439. (13) Nordhoff, E.; Kirpekar, F.; Karas, M.; Cramer, R.; Hahner, S.; Hillenkamp, F.; Kristiansen, K.; Roepstorff, P.; Lezius, A. Nucl. Acids Res. 1994, 22, 2460-2465. (14) Dai, Y.; Whittal, R. M.; Li, L.; Weinberger, S. R. Rapid Commun. Mass Spectrom. 1996, 10, 1792-1796. (15) Shaler, T. A.; Wickham, J. N.; Sannes, K. A.; Wu, K. J.; Becker, C. H. Anal. Chem. 1996, 68, 576-579. (16) Zhu, Y.; He, L.; Srinivasan, J. R.; Lubman, D. M. Rapid Commun. Mass Spectrom. 1997, 11, 987-992. (17) Little, D. P.; Cornish, T. J.; O’Donnell, M. J.; Braun, A.; Cotter, R. J.; Ko ¨ster, H. Anal. Chem. 1997, 69, 4540-4546. (18) Little, D. P.; Braun, A.; O’Donnell, M. J.; Ko ¨ster, H. Nature Med. 1997, 3, 1413-1416. (19) Haff, L.; Smirnov, I. P. Nucleic Acids Res. 1997, 25, 3749-3750. (20) Fei, Z.; Ono, T.; Smith, L. M. Nucleic Acids Res. 1998, 26, 2827-2828. (21) Li, J.; Butler, J. M.; Tan, Y.; Lin, H.; Royer, S.; Ohler, L.; Shaler, T. A.; Hunter, J. M.; Pollart, D. J.; Monforte, J. A.; Becker, C. H. Electrophoresis 1999, 20, 1258-1265. (22) Jiang-Baucom, P.; Girard, J. E.; Butler, J.; Belgrader, P. Anal. Chem. 1997, 69, 4894-4898. (23) Ross, P. L.; Lee, K.; Belgrader, P. Anal. Chem. 1997, 69, 4197-4202. (24) Mouradian, S.; Rank, D. R.; Smith, L. M. Rapid Commun. Mass Spectrom. 1996, 10, 1475-1478. (25) Shaler, T. A.; Tan, Y.; Wickham, J. N.; Wu, K. J.; Becker, C. H. Rapid Commun. Mass Spectrom. 1995, 9, 942-947. (26) Fu, D.-J.; Tang, K.; Braun, A.; Reuter, D.; Darnhofer-Demar, B.; Little, D. P.; O’Donnell, M. J.; Cantor, C. R.; Ko¨ster, H. Nature Biotech. 1998, 16, 381-384. (27) Smirnov, I. P.; Roskey, M. T.; Juhasz, P.; Takach, E. J.; Martin, S. A.; Haff, L. Anal. Biochem. 1996, 238, 19-25. (28) Nordhoff, E.; Karas, M.; Cramer, R.; Hahner, S.; Hillenkamp, F.; Kirpekar, F.; Lezius, A.; Muth, J.; Meier, C.; Engels, J. W. J. Mass Spectrom. 1995, 30, 99-112. (29) Polo, L. M.; McCarley, T. D.; Limbach, P. A. Anal. Chem. 1997, 69, 9, 1107-1112. (30) Tang, K.; Allman, S. L.; Chen, C. H.; Chang, L. Y.; Schell, M. Rapid Commun. Mass Spectrom. 1994, 8, 183-186. (31) Liu, Y.; Bai, J.; Liang, X.; Lubman, D. M.; Venta, P. J. Anal. Chem. 1995, 67, 3482-3490. (32) Bai, J.; Liu, Y.; Lubman, D. M.; Siemieniak, D. Rapid Commun. Mass Spectrom. 1994, 8, 687-691. (33) Kirpekar, F.; Berkenkamp, S.; Hillenkamp, F. Anal. Chem. 1999, 71, 23342339. (34) Hurst, G. B.; Weaver, K.; Doktycz, M. J.; Buchanan, M. V.; Costello, A. M.; Lidstrom, M. E. Anal. Chem. 1998, 70, 2693-2698.

Analytical Chemistry, Vol. 73, No. 11, June 1, 2001 2617

quencing samples, restriction fragments, and variable number tandem repeats (VNTR),42-47 MALDI-TOFMS offers a different set of potential advantages for these applications.8 Because the mass-to-charge ratio (m/z) is measured directly in mass spectrometry, size determination is more accurate than by electrophoretic measurement. With single-spectrum measurement times 200 units was counted from the DNA/matrix droplet dried on A, bare gold plate; B, LPA/Nafion two-layer substrate.

Deegan et al. have reported dense, ring-like deposits around the perimeter of an evaporating liquid drop containing dissolved or dispersed solid.82 The model for this phenomenon is that as the liquid evaporates, outward flow is necessary to maintain a constant droplet diameter due to “pinning” of the outside edge of the droplet to the substrate.82 Due to this flow, the solute concentration is enriched at the outside edge of the sample as evaporation proceeds. This effect may be in part84 responsible for the familiar deposition of the matrix crystals on the rim of MALDI spots that are deposited on bare metal sample plates. In our system, we believe that both polymer and Nafion could be partially dissolved in the water/acetonitrile solvent when a sample drop containing DNA and matrix is deposited on the two-layer substrate. Although this redissolution is to be avoided for singlecomponent nitrocellulose 31 or Nafion 40 substrates, it appears to be beneficial for our two-layer substrate system. Because of the increased viscosity in a droplet containing a high-molecular-weight polymer, transport of DNA and matrix molecules toward the perimeter of the sample droplet could be slowed. This more viscous medium may allow time for more extensive nucleation of DNA-doped matrix crystals throughout the drying MALDI droplet, before the solutes are transported to the periphery of the droplet by evaporation-driven flow. Nucleation and cocrystallization of DNA and matrix molecules, therefore, would become more likely (84) Amado, F. M. L.; Domingues, P.; Santana-Marques, M. G.; Ferrer-Correia, A. J.; Tomer, K. B. Rapid Commun. Mass Spectrom. 1997, 11, 1347-1352.

Analytical Chemistry, Vol. 73, No. 11, June 1, 2001

2621

Figure 3. Fluorescence microscopy images of Nile red-labeled 2-µm polystyrene beads with 5-s CCD exposure time: A, on glass substrate 2 min after 1 µL of bead solution is loaded; B, on glass substrate 5 min after loading; C, on LPA substrate (1 µL of LPA solution with 1.0% concentration was dried) 2 min after loading; D, on LPA substrate 5 min after loading. The width of each image corresponds to 850 µm.

at any location inside the drop as a result of reduced transport of those molecules in a more viscous solution, resulting in more spatially uniform deposition of the crystals. We tested the idea of increased viscosity due to a partially dissolved polymer substrate in drying MALDI droplets by measuring the velocity of fluorescent beads moving toward the periphery as the solvent evaporates from the droplet.82,83 Figure 3, obtained with a fluorescence microscope, shows traces of migrating fluorescently labeled polystyrene beads during a 5-s CCD exposure time. The larger spots are due to the immobile beads that are over-exposed in the photographs, leading to saturation of CCD pixels and corresponding dark centers. The thin lines are the traces of the beads that moved in suspension from the inside toward the outside of the MALDI droplet (top to bottom in Figure 3) while the solvent was evaporating. The length of the line is proportional to the average velocity of the bead during the exposure. The motion of the bead toward the edge of the spot is slower on a polymer substrate (Figure 3C,D), as compared to that on a glass substrate (no polymer layer applied, Figure 3A,B). For example, 5 min after depositing a droplet, the average velocity of the beads on a glass substrate was 27 µm/s (Figure 3B), but the average velocity of the beads on an LPA substrate was 12 µm/s (Figure 3D). We attempted to prepare two-layer substrates on the glass slides for imaging the bead movement, but were unable to prevent the Nafion droplet from rolling off of the polymer substrate onto the glass slide before drying. To determine the extent to which an increase in the viscosity of the droplet causes the beads to migrate more slowly, we prepared a series of aqueous bead suspensions with LPA concentrations (and, therefore, varying viscosities) ranging from 0.13 to 1.0% LPA, and recorded the bead traces as the droplets dried. In this experiment, no polymer substrate was applied to the glass surface; instead, the polymer was dissolved in the bead suspension. The initial concentration of LPA is known and homogeneous throughout the droplet. We observed decreasing bead velocities as initial LPA concentration and droplet viscosity increased. The resulting data set can be described by the relation L ) 26.8C-0.65, 2622

Analytical Chemistry, Vol. 73, No. 11, June 1, 2001

where L is the length of the bead trace (µm) and C is the initial LPA concentration (%) in the droplet. Using this calibration, it is possible to estimate (ignoring concentration gradients) the concentration of LPA dissolved from a dried LPA substrate layer into a subsequently applied MALDI sample droplet. When 1 µL of 1% LPA was dried first to form a substrate, and 1 µL of bead suspension was then applied on top, a 58-µm-long trace was observed over a 5-s CCD exposure. From the expression given above that relates trace length to initial LPA concentration, the concentration of LPA redissolved in this droplet is estimated to be ∼0.3%. Because this concentration lies between 0 (no LPA dissolution) and 1% (complete dissolution of the LPA substrate by the bead suspension droplet), it is consistent with partial dissolution of the polymer layer into the MALDI sample droplet during the drying process. While the experiments described above indicate that the LPA substrate dissolves and increases the viscosity in the drying MALDI sample droplet, additional effects are necessary to account fully for the improved MALDI performance observed with twolayer polymer/Nafion substrates. We prepared 20-mer/matrix solutions containing LPA (final LPA concentration, 0.10 to 1.0%), thereby producing MALDI spots containing LPA, but not as a dried substrate. MALDI spots prepared in this way appeared visually homogeneous; however, either no MALDI signal due to the 20-mer or broad peaks were obtained for all LPA concentrations tested (data not shown). One obvious difference between the LPA viscosity measurement experiments described above and the experiments in which successful MALDI signals were obtained (vide infra) is the presence of Nafion in the latter case. Nafion is a perfluorinated cation exchanger with sulfonic acid groups at the active sites. Sodium ions or other salts in the MALDI sample could be sequestered by these active sites, giving more tolerance to the effect of salts on MALDI spectra when Nafion is present.40 One further potential contribution of the polymer substrate to the uniform deposition of DNA and matrix could be that transport of DNA and matrix molecules toward the rim of the spot may be retarded as a result of processes similar to chromatographic partitioning of those molecules between the solution and the surface of the undissolved portion of the polymer/Nafion layer.40 This partitioning may allow more time for DNA and matrix molecules to nucleate before reaching the periphery of the drop and could also have some effect on the incorporation of impurities in DNA-doped matrix crystals.85 Performance of Substrates for MALDI-MS. Figure 4 shows the mass spectra of the HEX-labeled 20-mer obtained with some of the substrates that were shown in Figure 1. The multiple peaks in the MALDI spectrum shown in Figure 4A, obtained using a bare gold substrate, are apparently due to the presence of a mixture in the 20-mer sample. The calculated molecular mass (isotopically averaged) of the HEX-labeled 20-mer is 6901 Da; the measured m/z for the peak marked with an asterisk in Figure 4 is 6899. MALDI spectra (not shown) of unlabeled oligonucleotides in this size range exhibited only a single peak. With the LPA/ Nafion two-layer system or the mixed LPA/Nafion one-layer substrate (Figure 4C,D), the MALDI spectrum was similar to that shown in Figure 4A obtained on a bare gold plate. When only one of the two components was used alone as the substrate, (85) Cohen, S. L.; Chait, B. T. Anal. Chem. 1996, 68, 31-37.

Figure 4. MALDI spectra of HEX-labeled DNA 20-mer. Spectra were obtained from DNA/matrix crystals prepared on the substrates described below. Vertical scale was normalized to spectrum shown in A. A, bare gold plate; B, Nafion only; C, LPA/Nafion two-layer; D, mixed LPA/Nafion; E, LPA only. The spectra were smoothed using a 19-point Savitsky-Golay filter. Accelerating voltage, -25 kV; grid voltage, 92.5%; guide wire voltage, 0.30%; extraction delay, 200 ns; low-mass gate, 1000 Da.

spectra with lower resolution or adduct peaks were obtained (Figure 4B,E). In addition to LPA, polymers having a broad range of chemical and physical properties were investigated for preparation of homogeneous MALDI spots, as summarized in Table 1. In combination with a Nafion layer applied on top, substrates prepared using hydrophilic polymers such as LPA, PEO, PAA, and MC showed DNA MALDI spectra of reasonable quality. In contrast, no appreciable 20-mer signal was obtained with polymer substrates PEI and PDA. PEI contains multiple amine groups that can be positively charged and may strongly interact via ionic interactions with the phosphate groups of the DNA backbone. The behavior of the hydrophobic polymer substrate, PDA, was quite in contrast to solid hydrophobic films such as Parafilm or Teflon that have been reported as good substrates for analysis of DNA by MALDI.68,69 The reason could be unfavorable interaction between the PDA layer and DNA and matrix molecules as a result of different hydrophilicity, producing low-quality DNA-doped matrix crystals. For the polymer substrates that yielded favorable MALDI results, a range of polymer concentrations was also tested (Table 1). For LPA, PEO, and MC substrates, homogeneous DNA/matrix spots were obtained using substrates prepared from 0.5-5.0% (w/ w) initial polymer concentration in water. At higher initial polymer concentrations, no DNA/matrix crystal formation was observed, but at initial polymer concentrations below 0.5%, the pattern of DNA/matrix crystal formation resembled that obtained on a bare gold plate shown in Figure 1A. It has been pointed out that imaging of fluorescent analytes in MALDI samples is susceptible to artifacts because of scattered light from refraction or internal reflections in the matrix crystals.75,77 Furthermore, the fluorescence results described here do not distinguish between DNA that has been incorporated into matrix crystals and should, thus, yield favorable MALDI results

and DNA that has not been incorporated into matrix crystals. Therefore, while the fluorescence data described above provide useful insights into the homogeneity of HEX-labeled DNA distribution in MALDI samples, a more practical test for the utility of various substrates is MALDI profiling by acquisition of mass spectra from various locations on the sample spot. Such an interrogation of spot homogeneity using MALDI is summarized in Table 2. Generally, the spots prepared on a bare gold plate or polymer substrates have a diameter of 1-2 mm, and the focused laser has a diameter of ∼100 µm. Therefore, investigation of ∼10 locations along the diameter of a MALDI spot would be necessary to profile the distribution of sweet spots. For spots prepared on a bare gold plate, 12 locations equally spaced around the rim of the spot were interrogated. For spots prepared on polymer substrates, >14 locations were investigated (at least 6 on the vertical diameter, at least 6 on the horizontal diameter, and 1 near the center of one or more quadrants), and several different DNA/ matrix spots on each type of substrate were examined. The success rate (%) was calculated for each distinct sample preparation (spot) by counting the number of locations within that spot that generated a characteristic MALDI signal of the HEX-labeled 20-mer with S/N > 4 and dividing by the total number of interrogated locations. The uncertainties listed in Table 2 for the success rates are the standard deviations of the success rates measured for n different sample spots, with n ranging from 5 to 12. For spots prepared on the bare gold plate, the average success rate was ∼50%, but the average success rate was >80% for the spots prepared on LPA- and PEO-based two-layer substrates. The standard deviation of the success rate for spots prepared on the bare gold plate was ∼16%, but the standard deviation was < 3% for substrates prepared from Nafion atop LPA, PEO, or MC. Low resolution and extensive adducts in the DNA MALDI spectra were obtained from the spots prepared on the acidic polymer, poly(acrylic acid) (PAA), and poor or no 20-mer signal was observed in MALDI spectra obtained using PDA or PEI substrates. Both the success rate and its variability are important figures of merit for an automated MALDI laser search strategy. The success rate is a measure of reproducibility in signal obtained from different locations within a spot, and the standard deviation is a “betweenspot” variation measurement. A higher success rate for a given substrate/matrix combination means that an automation strategy can rely on illuminating fewer locations per spot with the laser to generate a useful spectrum, and reduces the number of spectra that must be subjected to postrun evaluation. The effect of polymer molecular weight and concentration on success rate for obtaining MALDI signal was tested with a PEO/ Nafion two-layer substrate (Table 3). When the PEO molecular weight was >1 000 000 Da or the initial concentration was >1%, the success rate fell to 4) was quite reproducible from spot to spot for some substrate compositions (see Tables 2 and 3), there was nonetheless a rather wide variation in the actual Analytical Chemistry, Vol. 73, No. 11, June 1, 2001

2623

Table 2. Success Rate for Obtaining MALDI Signal from DNA/Matrix Spots on Various Substratesa substrate bare gold plate

LPA

PEO

MC

PDA

PAA

PEI

52 ( 16 n ) 6b

700 000-1 000 000 1.0 84 ( 3 n ) 12

1 000 000 1.0 87 ( 2 n ) 12

86 000 2.0 71 ( 3 n)8

NA 10.0 12 ( 4 n)5

450 000 2.0 -c n)5

1 000 000 2.0 -c n)5

polymer molecular weight, Da initial polymer concentration, % (w/w in water) success rated, %

a Nafion was added on top of the dried polymer layer. b n is the number of the spots investigated. c -, no MALDI signal was obtained from the spot prepared on this substrate. d Ratio of locations within a spot yielding S/N > 4 for MALDI signal of HEX 20-mer to the total number of interrogated locations for that spot.

Table 3. Effect of Polymer Molecular Weight and Concentration on the Success Ratea for Obtaining MALDI Signal from DNA/Matrix Spot Dried on PEO/ Nafion Substrates PEO molecular weight, Da 8 000 000 1 000 000 1 000 000 1 000 000 init PEO concn, % 0.5 (w/w in water) success ratea, % 49 ( 7 n)6

0.5

1.0

2.0

300 000 2.0

83 ( 3 84 ( 3 25 ( 8 64 ( 7 n ) 12 n ) 12 n ) 12 n)8

a Ratio of locations within a spot yielding S/N > 4 for MALDI signal of HEX 20-mer to the total number of interrogated locations for that spot.

value of the signal-to-noise ratio. For instance, S/N ranging from 4 to >100 was observed from the spot prepared on the LPA/ Nafion two-layer substrate shown in Figure 1B. This means that even though the HEX-labeled DNA was deposited very uniformly, as demonstrated by fluorescence microscopy, the factors that combine to yield sweet spots (e.g., DNA-to-matrix ratio or the distribution of signal-quenching impurities) were apparently not constant throughout the spot. However, the ability to obtain DNA signals at S/N > 4 more uniformly across a MALDI spot still offers the potential of improved throughput for the analysis of small oligonucleotides. CONCLUSIONS Two-layer polymer/Nafion substrates were developed for homogeneous spot preparation in MALDI-TOFMS of small oligonucleotides. The homogeneity of DNA/matrix spots was investigated using fluorescence microscopy with a HEX-labeled DNA 20-mer and by acquisition of MALDI mass spectra from various locations on the spots. Homogeneous spot formation was attributed to increased matrix nucleation in the interior of the drying droplet due to increased viscosity caused by partial dissolution of the polymer substrate. In comparison to typical scratched metal MALDI sample plates, the more uniform surface offered by a polymer substrate may also contribute to homogeneous spot formation, as could a partitioning of DNA or matrix (or both) between the solution and the undissolved portion of the polymer substrate. The Nafion component of the substrate may serve an ion-exchange function in reducing multiple cation adduction in

2624

Analytical Chemistry, Vol. 73, No. 11, June 1, 2001

MALDI spectra. We believe that our approach can increase throughput in the MALDI-TOFMS analysis of small oligonucleotides because of a decreased need to search for sweet spots. An increasing number of applications for MALDI-TOFMS analysis of such small oligonucleotides, including quality control for synthetic oligonucleotides and genotyping by primer extension, should benefit from a more uniform dried sample that facilitates automated analysis. Although sample amount is less of an issue in the former case, it will be necessary for some applications to investigate the sensitivity limits of the current technique for lower levels of DNA. As has been noted for single-component nitrocellulose and Nafion substrates,9 the preparation of these two-layer substrates is still dependent on the skill of the experimenter, and will require further work to determine more reproducible methods for convenient fabrication of the substrates for routine work. The use of a two-layer substrate increases the opportunity to incorporate multiple functions (e.g., sample homogeneity and cation adduct reduction) into the MALDI sample plate without the need for specialized sample preparation or low-volume pipetting equipment. ACKNOWLEDGMENT The authors thank Dr. Phillip F. Britt, ORNL, for providing sublimed HPA and Dr. Robert D. Deegan, University of Texas, for providing a preprint of ref 83. Y.K. acknowledges support through an appointment to the Postdoctoral Research Associates Program at the Oak Ridge National Laboratory, administered by the Oak Ridge Institute for Science and Education. This research was supported in full under Grant No. 55108, Environmental Management Science Program, Office of Science and Technology, Office of Environmental Management, United States Department of Energy (DOE); however, any opinions, findings, conclusions, or recommendations expressed herein are those of the authors and do not necessarily reflect the views of DOE. Oak Ridge National Laboratory is managed and operated by UT-Battelle, LLC, for the U.S. Department of Energy under contract DE-AC0500OR22725. Received for review November 28, 2000. Accepted March 26, 2001. AC001392V