Anal. Chem. 2006, 78, 827-834
Automated Acoustic Matrix Deposition for MALDI Sample Preparation Hans-Rudolf Aerni, Dale S. Cornett, and Richard M. Caprioli*
Mass Spectrometry Research Center, Departments of Chemistry and Biochemistry, Vanderbilt University, 465 21st Avenue South, Medical Research Building 3, Room 9160, Nashville, Tennessee 37232-8575
Novel high-throughput sample preparation strategies for MALDI imaging mass spectrometry (IMS) and profiling are presented. An acoustic reagent multispotter was developed to provide improved reproducibility for depositing matrix onto a sample surface, for example, such as a tissue section. The unique design of the acoustic droplet ejector and its optimization for depositing matrix solution are discussed. Since it does not contain a capillary or nozzle for fluid ejection, issues with clogging of these orifices are avoided. Automated matrix deposition provides better control of conditions affecting protein extraction and matrix crystallization with the ability to deposit matrix accurately onto small surface features. For tissue sections, matrix spots of 180-200 µm in diameter were obtained and a procedure is described for generating coordinate files readable by a mass spectrometer to permit automated profile acquisition. Mass spectral quality and reproducibility was found to be better than that obtained with manual pipet spotting. The instrument can also deposit matrix spots in a dense array pattern so that, after analysis in a mass spectrometer, two-dimensional ion images may be constructed. Example ion images from a mouse brain are presented. Matrix-assisted laser desorption/ionization (MALDI) mass spectrometry may be used as a tool for obtaining molecular patterns in biological samples such as tissue sections. Sample preparation involves applying matrix to one or more regions of the tissue either as an ordered array of spots or a coating of the entire surface. Mass spectra are then acquired from discrete coordinates on the sample or by rastering the laser across the sample. Mass spectral quality is influenced by the molecular constituents in the region being analyzed as well as the skill and experience of the analyst who prepares and analyzes the samples. Components such as lipids, carbohydrates, and salts can promote adduct formation, ion suppression, or poor crystallization, each of which can affect the quality of the mass spectra. This effect may be even more pronounced for larger sections where cellular heterogeneity across different regions can influence the crystallization and desorption processes. * Corresponding author: (e-mail)
[email protected]; (fax) (615) 3438372; (tel) (615) 322-4336. 10.1021/ac051534r CCC: $33.50 Published on Web 12/21/2005
© 2006 American Chemical Society
A number of variations of pipetting, spraying, and immersion have been described for applying matrix to tissues.1 Spray techniques are most common and involve electrospray or pneumatic spray of matrix solution directly onto the tissue section. Immersion techniques most often involve dipping the mounted section directly into a matrix solution bath but, more broadly defined, can also include droplets of several microliters spread over a large area.2 Immersion in matrix solution is favorable for extracting proteins, but potentially allows proteins to migrate across the tissue section or to be dissolved into the immersion solvent. The small size of the aerosol droplets formed by either of the spray techniques effectively limits protein migration to the area wetted by the microdroplet, but the uncontrolled size and placement of the spray is problematic for protein extraction and migration. While there are many methods for depositing matrix on samples, we have found individual droplet placement to be much more reproducible than spray coatings and also more amenable to automation. The droplet size can be varied, and the optimal size depends on the analytical information desired and the type of tissue. Smaller, closely spaced droplets are best for highresolution images, but are harder to generate and harder to evenly sample with the desorption laser. Larger droplets spread on surfaces, increase the possibility of compound translocation within the matrix spot, and are better suited to “profiling” rather than “imaging”. A number of devices are available for generating and accurately depositing droplets onto a surface. Contact deposition methods typically require a pin or capillary to facilitate liquid transfer through physical contact with the sample. Reproducibility associated with spotting small amounts of liquid using pins or microcapillaries can be low.3 Moreover, contact between the fluid and the substrate can introduce cross-contamination or may require washing of the capillary between depositions, which reduces the duty cycle. Noncontact printing devices such as piezoelectric, thermal ink jet, solenoid valves, and pulsed field ejectors do not have these limitations. Uniform microdroplets of liquids are dispensed from small capillaries by applying pressure pulses to eject droplets with a low velocity onto the tissue. Cross-contamination is not a concern with these devices since there is no contact (1) Schwartz, S. A.; Reyzer, M. L.; Caprioli, R. M. J. Mass Spectrom. 2003, 38, 699-708. (2) Stoeckli, M.; Staab, D.; Staufenbiel, M.; Wiederhold, K.-H.; Signor, L. Anal. Biochem. 2002, 311, 33-39. (3) Rose, D. Drug Discovery Today 1999, 4, 411-419.
Analytical Chemistry, Vol. 78, No. 3, February 1, 2006 827
between capillary and the tissue. Noncontact microdispensers are widely used for office printing and industrial manufacturing and, more recently, in scientific applications including MALDI sample preparation.4 In the latter case, both sample and matrix solutions have been printed for proteomic5-8 and genomic9 sample preparations. Protein profiles from single cells have been obtained from microspots of matrix deposited onto single cells that were either deposited or grown on a MALDI target.10 A second advantage of noncontact microdispensing is the ease at which ejection parameters can be varied to fine-tune the matrix/analyte ratio. However, one significant limitation of capillary- or nozzle-driven microdispensers is that the combination of high matrix concentration and solvent volatility frequently can cause partial or complete clogging of the capillary which, in turn, produces erratic ejection of the droplets. An approach that minimizes these problems is the use of technologies that generate a monodisperse stream of microdroplets in such a manner that they can be precisely deposited onto the tissue surface. Small droplets effectively limit protein migration to the area covered by the droplets. Multiple droplets can be successively deposited onto the same location at varying ejection frequencies to provide some control of protein extraction from the tissue. A third benefit to such an approach is that it greatly improves reproducibility of the coating process and minimizes variation among users. Acoustic drop ejection is a novel technology capable of generating microdroplets without need for a capillary or nozzle.11 The technique utilizes short pulses of focused acoustic energy directed onto a liquid surface from beneath an open reservoir. Depending on the properties of the focused acoustic transducer, the applied radio frequency (rf) energy, and the properties of the liquid in the reservoir, uniform droplets in the range of 0.1 pL to more than 10 µL can be ejected from the liquid surface at a relatively high rate. In this report, we describe a new automated device that incorporates acoustic drop ejection technology for matrix deposition for MALDI analysis. The nozzle-free design of the ejector readily permits use of traditional matrix solutions, namely, concentrated matrix solutions in organic solvents, without concern for droplet misdirection or stoppage due to clogging. Operational characteristics of the device are examined, and its reliability and effectiveness is demonstrated for preparing tissue samples for both MALDI profiling and imaging analyses. (4) Lee, E. R. Microdrop Generation; CRC Press: New York, 2002. (5) Bogan, M. J.; Agnes, G. R. Rapid Commun. Mass Spectrom. 2004, 18, 26732681. (6) Ericson, C.; Phung, Q. T.; Horn, D. M.; Peters, E. C.; Fitchett, J. R.; Ficarro, S. B.; Salomon, A. R.; Brill, L. M.; Brock, A. Anal. Chem. 2003, 75, 23092315. (7) Wallman, L.; Ekstro¨m, S.; Marko-Vaga, G.; Laurell, T.; Nilsson, J.; Electrophoresis 2004, 25, 3778-3787. (8) Sloane, A. J.; Duff, J. L.; Wilson, N. L.; Gandhi, P. S.; Hill, C. J.; Hopwood, F. G.; Smith, P. E.; Thomas, M. L.; Cole, R. A.; Packer, N. H.; Breen, E. J.; Cooley, P. W.; Wallace, D. B.; Williams, K. L.; Gooley, A. A. Mol. Cell. Proteomics 2002, 1, 490-499. (9) Little, D. P.; Cornish, T. J.; O’Donnell, M. J.; Braun, A.; Cotter, R. J.; Ko ¨ster, H. Anal. Chem. 1997, 69, 4540-4546. (10) Bergquist, J. J. Chomatogr. Suppl. I 1999, 49, S41-S48. (11) Elrod, S. A.; Hadimioglu, B.; Khuri-Yakub, B. T.; Rawson, E. G.; Richley, E.; Quate, C. F.; Mansour, N. N.; Lundgren, T. S. J. Appl. Phys. 1989, 65, 3441-3447.
828 Analytical Chemistry, Vol. 78, No. 3, February 1, 2006
Figure 1. Schematic of the acoustic reagent multispotter with translational stage, sample and drop imaging systems and the acoustic ejector for matrix dispensing.
EXPERIMENTAL SECTION Material. HPLC grade acetonitrile, trifluoroacetic acid (TFA), and reagent grade ethanol were purchased from Fisher Scientific (Pittsburgh, PA). Protein standards, fluorescein, and TRIS buffer were from Sigma-Aldrich (St. Louis, MO). Sinapinic acid (3,5dimethoxy-4-hydroxycinnamic acid, 99% purity) was purchased from Fluka (Buchs, Switzerland) and used without further purification. A standard matrix solution consisted of 25 mg/mL sinapinic acid in 1:1 acetonitrile/0.2% TFA (aq). Sinapinic acid seed crystals were prepared by grinding sinapinic acid in a ceramic mortar to a fine powder of 0.3-3-µm diameter. Acoustic Reagent Multispotter (ARM). A schematic representation of the ARM, custom-made at Labcyte Inc. (Sunnyvale, CA), is illustrated in Figure 1. Essential parts include an acoustic ejector, translational stage, and video telescopes integrated under software control. The acoustic ejector consists of a spot-focused piezoelectric transducer (30 MHz, Panametric, Waltham, MA) that is driven by short bursts of rf energy from a waveform generator (Pragmatics, Santa Clara, CA) and amplified in a 150-W broadband rf amplifier (Amplifier Research, Souderton, PA). A reservoir for matrix solution is fabricated by thermally forming cyclic olefin copolymer film into a flat bottom vessel of ∼1.5-mL capacity. The transducer and reservoir are coupled acousticly through a column of water. A linear stage driven by a stepper motor raises and lowers the transducer respective to the reservoir for optimal focusing of the acoustic energy onto the underside of the liquid surface. A motorized x-y stage (Prior Scientific, Rockland, MA) translates the MALDI plates a distance of 4 mm above the ejector. The lateral precision of the stage is 3 µm. The MALDI plate is held in one of two positions with samples facing downward. When positioned in the inspection position, a color CCD telescope captures digital images of the tissue section and displays them in the control software. Locations for depositing matrix are selected by clicking the desired pixel in the captured plate image using the mouse (for depositing a single spot) or by defining a large area for depositing rectangular arrays of matrix spots. Spot coordinates and printing parameters are saved into a file and the target is transferred to the second, or printing position, which is located just above the acoustic ejector. A second imaging system for viewing droplet ejection incorporates an adjustable stroboscopic white light source (LED) for back-light illumination and a model KV long distance video microscope (Infinity Photo-
Optical, Boulder, CO) equipped with a 1/2-in. black and white highsensitivity (0.000 15 lux) CCD camera, which is connected to the computer using a 30-Hz frame capture card. Ejector tuning requires finding the minimum rf energy needed to produce stable droplet formation. The tuning process and droplet ejection are illustrated by the sequence of images shown in Figure 2. The reservoir is filled with matrix solution, and the focal point of the transducer is brought close to the surface of the solution by adjusting the relative distance between reservoir and transducer. As the focal point of the transducer approaches the liquid surface, the acoustic energy causes a visible distortion (Figure 2A). Subsequent increases in the amplitude of the rf increases the acoustic pressure wave to the point that the surface deformation elongates into a cone-shaped structure. Threshold energy is reached when the rf amplitude is sufficient to cause the point of the liquid cone to break away as a spherical droplet (Figure 2B). A final slight increase in rf amplitude causes vertical ejection of the droplet toward the sample (Figures 2C and D), suspended 4 mm above the liquid surface. A transparent lid (Figure 1) is then placed on the reservoir to minimize solvent evaporation, which not only changes the concentration of matrix but lowers the level of the liquid surface affecting tuning. The lid has a 400-µm-diameter hole in the center for droplets pass through to reach the sample surface. The relatively large reservoir capacity allows for prolonged periods of operation with minimal effect from small evaporative losses. Ejection of at least 50 000 droplets is possible before retuning is needed. Tissue Preparation. Brain and liver from adult CD1 mice and Sprague-Dawley rats were dissected from the animals, snap frozen in liquid nitrogen, and stored at -80 °C until use. Sectioning of the tissue was carried out in a cryostat using a procedure described earlier.1 Thin (12 µm) tissue sections were thaw mounted onto cold indium tin oxide (ITO)-coated glass slides12
or gold clad stainless steel MALDI targets. The targets were immediately transferred to a vacuum desiccator and allowed to equilibrate to room temperature (10 min) followed by a series of ethanol/water washes to fix the tissue while simultaneously removing MALDI contaminants such as physiological salts and other soluble low molecular weight compounds. Washing consists of gently agitating the plates in two successive 30-s baths of 25 mL of 70:30% v/v ethanol/water followed by a 15-s wash in 25 mL of ethanol. Excess solvent on the tissue is removed with a gentle blow of compressed dry air. Residual solvent trapped in the tissue is pumped away by placing the sample for 10 min in a vacuum desiccator. Tissue samples were then spotted with the matrix or stored in a closed container on the bench for up to 6 h without noticeable degradation of the MALDI spectra. Unless otherwise noted, tissue sections were first coated with a thin layer of seed matrix to enhance crystallization of the matrix solution during spotting of the matrix.13 A layer of ground sinapinic acid, typically 2-3 mm thick, was deposited onto the tissue. A painter’s brush was used to distribute the powder over the complete surface of the sample taking care to avoid direct contact of the brush with the tissue. Excess material was removed using a gentle blow of compressed air in a laboratory hood. Observation of the tissue under a microscope indicated a high-density dispersion of 0.3-3-µm-sized particles across the section. Magnified images from the spotted tissues were obtained on an Olympus BX 50 (Olympus America Inc., Melville, NY) microscope equipped with a Q-Imaging 3.3 megapixel digital camera. Image-Pro-Plus Software from Media Cybernetics (Silver Spring, MD) was used for image analysis. Mass Spectrometry and Data Processing. MALDI TOF mass spectrometry was performed using a Voyager-DE STR (Applied Biosystems, Framingham, MA) in linear positive mode using 25-kV acceleration. Delayed extraction conditions of 91% for the grid and 450 ns for the delay time provided an optimal mass resolution of ∼1000 (fwhh) for m/z ∼15 000. Spectra were recorded at 20 Hz using a nitrogen laser (λ ) 337 nm), which was focused to a spherical spot of 100-µm diameter using an adjustable iris placed between the laser and the focusing lens of the mass spectrometer. Signals between 2 and 70 kDa were recorded with 4-ns time bins. External calibration of the instrument was performed using a mixture of porcine insulin [(M + H)+ ) 5778.6 Da], horse heart cytochrome c [(M + H)+ ) 12 361.2 Da] and horse ampomyoglobin [(M + H)+ ) 16 952.5 Da] deposited onto the plate immediately adjacent to the section with a mechanical pipet. Spectra were acquired manually or under instrument control using the Sequence Editor module of the Voyager Control software package. To facilitate automated spectral acquisition, custom plate files based on the matrix spot locations were generated. Absolute stage coordinates for the center of a number of registration spots were transferred from the mass spectrometer software into a spreadsheet where a bilinear fit was performed to interpolate the coordinates of all spots in the array. The resulting list of interpolated spot coordinates was exported into a plate file format and submitted to Voyager Control for automated acquisition. For each matrix spot, a total of 200-400 individual spectra
(12) Chaurand, P.; Schwartz, S. A.; Billheimer, D.; Xu, B. J.; Crecelius, A.; Caprioli, R. M. Anal. Chem. 2004, 76, 1145-1155.
(13) Aerni, H.-R.; Erskine, A. R.; Reyzer, M. L.; Lee, D.; Cornett, D. S.; Caprioli, R. M. Proc. Am. Soc. Mass Spectrom. 51st Ann. Conf. Mass Spectrom. Allied Top., Montreal, Canada, 2003.
Figure 2. Photomicrographs obtained from the surface of the reservoir during tuning of the acoustic ejector. Acoustic radiation focused from beneath the matrix solution in the reservoir induces a surface disturbance (A). The focal spot and amplitude of the acoustic wave produced by the transducer is adjusted to the point where a droplet is formed at the top of the liquid column (B) but not ejected (threshold energy). Adjusting of the energy above the threshold value ejects (C) a droplet toward the sample to be spotted (D).
Analytical Chemistry, Vol. 78, No. 3, February 1, 2006
829
Table 1. Characterization of Droplet Size for a Series of Acetonitrile/Water Mixtures Calculated from a Quantitative Fluorescence Assaya water, %
droplet vol, pL
droplet diam, µm
100 90 70 50
181 ( 12 179 ( 9 151 ( 2 122 ( 16
70.2 69.9 66.1 61.6
a The confidence interval for the droplet volume was calculated on the 95% level. The relatively large errors are the result from four independent measurements, which required exchanging the liquid in the reservoir and retuning. The standard error for the assay was better than 2%.
were summed in 40-spectra segments using a random search pattern confined within the area of the spots. Exclusion criteria rejected spectra with poor signal intensity (base peak
