High-Speed MALDI-TOF Imaging Mass Spectrometry: Rapid Ion Image Acquisition and Considerations for Next Generation Instrumentation
A prototype matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometer has been used for high-speed ion image acquisition. The instrument incorporates a Nd:YLF solid state laser capable of pulse repetition rates up to 5 kHz and continuous laser raster sampling for high-throughput data collection. Lipid ion images of a sagittal rat brain tissue section were collected in 10 min with an effective acquisition rate of roughly 30 pixels/s. These results represent more than a 10-fold increase in throughput compared with current commercially available instrumentation. Experiments aimed at improving conditions for continuous laser raster sampling for imaging are reported, highlighting proper laser repetition rates and stage velocities to avoid signal degradation from significant oversampling. As new high spatial resolution and large sample area applications present themselves, the development of high-speed microprobe MALDI imaging mass spectrometry is essential to meet the needs of those seeking new technologies for rapid molecular imaging.
Key wordsMass spectrometry TOF MALDI Imaging Microprobe Lipids Continuous laser raster
The emergence of matrix-assisted laser desorption ionization imaging mass spectrometry [MALDI imaging mass spectrometry (IMS)] as a prominent analytical tool  has helped drive advancements in instrumentation, lasers, data acquisition methods, and data processing packages. MALDI imaging mass spectrometry began at a time when 20 Hz lasers were standard and acquiring ion images took many hours and even days in some cases to collect data and deal with rough and unproven software. The technology has rapidly developed into a powerful analytical tool able to spatially resolve biologically relevant molecules such as pharmaceuticals, metabolites, lipids, peptides and proteins [2, 3, 4, 5, 6], providing a new molecular dimension to classic histology. Technological advancement has been central to the development of MALDI IMS, and developing next generation mass spectrometry instrumentation is essential to expanding its capabilities even further.
MALDI imaging mass spectrometry provides spatially resolved multichannel (m/z) information for intact molecular ions. For biological applications, the experiment is performed by adhering thinly sliced tissue sections onto a target and applying an energy absorbing matrix in a manner that minimizes delocalization of the analyte. Ions are generated as the sample moves through the path of a stationary laser beam and images are constructed by plotting the x and y coordinates of the ablations spots (pixels) with respect to ion signal intensity as detected by the mass spectrometer. Each pixel contains an entire mass spectrum allowing hundreds of images to be constructed from a single acquisition. This is commonly referred to as microprobe imaging mass spectrometry.
As microprobe MALDI IMS has developed, throughput has been a major issue particularly as the technology advances to higher spatial resolution and larger image areas. For example, a 1 mm2 tissue section imaged with a spatial resolution of 100 μm produces an image with 100 total pixels, a manageable number of spectra. However, much work is being done to image tissue sections at higher spatial resolutions [7, 8, 9] with 5 μm being the current applicable limit for microprobe MALDI IMS. If the same 1 mm2 tissue section is imaged at 5 μm spatial resolution, 40,000 pixels would be generated taking many hours to acquire. The same consideration applies to sampling larger areas for applications such as determining molecular distributions in whole-body tissue cross sections [10, 11, 12, 13] and biomarker discovery using tissue microarrays [14, 15, 16]. Developing high-throughput imaging mass spectrometric techniques is essential as MALDI IMS advances.
Laser technology has been central to the evolution of high-speed imaging. MALDI is performed by co-crystallizing or coating the sample with a matrix that efficiently absorbs the energy of the incident laser beam . Matrix molecules often have acidic functionality so that in the plume, following irradiation, protons are efficiently transferred to the analyte resulting in intact protonated molecular ions. Traditionally, N2 lasers have been used for MALDI imaging, but because of their low repetition rates (50 Hz) these lasers are not ideal for this purpose. Recently, there has been a shift to using frequency tripled solid-state lasers using gain mediums such as Nd:YAG, Nd:YLF, and Nd:YVO4. Although the Gaussian beam profile for solid-state lasers is not ideal, they have proven effective for MALDI applications and the ability to operate at pulse repetition rates >1 kHz with much longer life spans (typically >109 shots) than N2 lasers make them preferred for high-throughput experiments [7, 18]. A recent report described a modified hybrid quadrupole time-of-flight MS with a Nd:YVO4 laser capable of pulse repetition rates up to 20 kHz . Although higher laser repetition rates were possible, it was reported that operating between 5 and 10 kHz provided the best results, allowing for a 150 μm spatial resolution lipid ion image of a sagittal rat brain tissue section to be acquired in roughly 40 min. This is several times faster than the performance of most commercial MALDI IMS instrumentation.
Vertically, spatial resolution is determined simply by the distance between laser raster rows. Although continuous laser raster sampling has been shown to significantly decrease image acquisition time, much has yet to be done to understand the fundamentals of the process. In particular, determining the relationship among continuous laser raster sampling, high repetition rate lasers and data quality is essential in developing next generation high-speed MALDI IMS technology. Additionally, the use of high repetition rate lasers and high-throughput image acquisition also brings the need for advanced methods for handling huge data files and subsequently mining of these data.
With this contribution we report ultra high-speed acquisition times for microprobe MALDI imaging mass spectrometry. A prototype high-speed MALDI-TOF instrument utilizing continuous laser raster sampling was used to acquire 100 μm ion images of a sagittal rat brain tissue section in 10 min. A detailed discussion of the advantages and limitations of continuous laser raster sampling and experimental results highlighting the optimal experimental conditions for ion imaging experiments are presented. With a systematic approach to understanding the fundamentals of continuous laser raster sampling and the development of new instrumentation, the next technological step towards high-throughput MALDI imaging mass spectrometry can be accomplished.
2.1 Sample Preparation
MALDI matrices 2,5-dihydroxybenzoic acid (DHB) and α-cyano-4-hydroxycinnamic acid (CHCA) were purchased from Acros-Organics (Morris Plains, NJ, USA). Angiotensin II (AngII) and all peptides used for external calibration of the instrument were purchased from Sigma-Aldrich (St. Louis, MO, USA). Fresh frozen adult rat brain was purchased from Peel-Freez Biologicals (Rogers, AR, USA). Solutions were prepared using HPLC grade methanol (Fisher Scientific, Fair Lawn, NJ, USA) and Millipore filtered deionized water (Millipore, Bedford, MA, USA). Tissue samples were cryosectioned (12 μm thickness), thaw-mounted on gold-coated plates, and dried in a vacuum dessiccator (~2 h) prior to application of DHB by sublimation (120°C, 50 mTorr, 5.5 min) . For comparison, a serial tissue section was thaw-mounted onto a glass slide and stained with hematoxylin and eosin (H and E). Continuous laser raster sampling was tested using standard homogenous coatings of AngII (100 nmol/mL) and CHCA (10 mg/ml, 60% methanol/0.1% TFA) spray coated (120°C/10 PSI N2 sheath gas, 400 μL/min flow rate, four passes) onto gold plates using a TM-Sprayer (LEAP Technologies, Carrboro, NC, USA).
2.2 Imaging Mass Spectrometry
All experiments were performed on a prototype MALDI time-of-flight mass spectrometer designed and built by Marvin Vestal and his team at Virgin Instruments (Sudbury, MA, USA). The instrument is a positive mode reflector TOF MS with an effective path length of 3.2 m operated at 8 kV. Briefly, ions are generated using a 349 nm, diode-pumped, frequency-tripled Nd:YLF laser (Spectra-Physics, Santa Clara, CA, USA) capable of operating at laser repetition rates up to 5 kHz. Laser energy is controlled by adjusting the current applied to the diode while keeping laser attenuation constant. All reported laser energies (μJ/pulse) are taken prior to attenuation. Generated Ions are extracted by pulsed acceleration using grid-less ion optics. The ion beam is redirected through two sets of deflector plates on the way to a two-stage ion mirror and detector (0.5 ns electron multiplier). The two deflectors are necessary to move the ion beam from the path of the laser beam that is incident perpendicular to the sample stage and direct the ions to the mirror at the proper angle to maximize resolution. A specified number of laser shots were “hardware” averaged on the acquisition card (model U1082A/AP 240; Acqiris USA, Monroe, NY, USA) prior to writing data to the hard disk. External calibration was performed using a mixture of standard peptides including bradykinin 1–7 (m/z 757.85), angiotensin II (m/z 1046.54) and [Glu1]-fibrinopeptide B (m/z 1570.68) in addition to CHCA clusters (m/z 379.09 and m/z 568.14). Theoretical and instrumental design considerations have been described previously [22, 23, 24].
The instrument is designed to use continuous raster sample interrogation. Ion images of rat brain tissue sections were acquired using both typewriter (acquisition in one lateral direction only) and serpentine (acquisition in both lateral directions) patterns. Tissue images were obtained using a laser repetition rate of 3 kHz (actual: 3056 Hz) at 100 μm spatial resolution as determined vertically by the motor step size between continuous laser raster rows and laterally by setting the stage velocity and hardware average to the appropriate values (see Eq. 1). Best results were observed using a 5 mm/s sample stage velocity and 60 laser shots/spectrum hardware average. Data analysis was done using MiniView analysis software (Virgin Instruments, Sudbury, MA, USA), and all images were color scaled individually from 100%–0% for the relative ion signal intensity.
Lipid identifications were made using mass accuracy data from a MALDI FT-ICR mass spectrometer (9.4 Tesla Apex-Qe; Bruker Daltonics, Billerica, MA, USA) equipped with an Apollo II dual ion source and a 355 nm solid-state laser. External calibration was done using a peptide mixture of bradykinin 1–7, angiotensin II and [Glu1]-fibrinopeptide with the MALDI matrix CHCA. Lipid classifications were determined by comparing mass accuracy data (<1 ppm) with the LIPID MAPS database (Nature Lipidomics Gateway, www.lipidmaps.org).
2.3 MALDI Raster Sampling Experiments
To better understand continuous laser raster sampling as a method for MALDI surface interrogation, uniform spray coated layers of AngII/CHCA were analyzed at various sample stage velocities and laser repetition rates. In both cases, data was acquired in a similar fashion as if performing an imaging mass spectrometry experiment by rastering over discrete rectangular areas varying a specific instrumental parameter from one area to another. To understand the relationship between signal intensity and sample stage velocity, all imaged areas were collected using a laser repetition rate of 3 kHz, 5 μJ laser power, 50 laser shots/spectrum, and step size of 200 μm between raster lines. A different sample stage velocity, ranging from 1 to 10 mm/s, was used to sample each area. Similarly, to test the effect laser repetition rate has on signal intensity, areas were sampled using 2 mm/s sample stage velocity, 5 μJ laser power, 50 laser shots/spectrum, and step size of 100 μm between raster lines while varying laser repetition rate for each rectangular area. It should be noted that actual laser repetition rates are slightly different from input values. Optical images of raster lines were taken with an Olympus BX50 20× microscope (Center Valley, PA, USA) and Image-Pro Plus ver. 7 software.
3 Results and Discussion
3.1 High-Speed MALDI TOF Imaging Mass Spectrometry
3.2 Continuous Laser Raster Sampling
As imaging mass spectrometry technology advances, high repetition rate lasers will become essential for high-resolution imaging experiments. Figure 4c shows the relationship between laser shots/unit area and laser beam diameter at different laser repetition rates at a fixed sample stage velocity (5 mm/s). It is apparent from this plot that as the technology moves towards using smaller laser spot sizes (<10 μm diameter) to acquire higher spatial resolution ion images laser spot overlap will not be significant. High pulse repetition rate lasers (>10 kHz) can then be used to acquire images quickly with very little detriment to signal intensity from substantial oversampling. There are, however, new problems that arise when using such high laser repetition rates. For example, a sample moving through the path of a 5 μm diameter laser operating at a pulse repetition rate of 20 kHz at 5 mm/s will have a laser spot overlap of 20 laser shots/unit area. Assuming the pulsed extraction voltages can be operated at 20 kHz, acquiring a 5 μm spatial resolution ion image using these experimental conditions (20 kHz laser repetition rate, 5 mm/s stage velocity, 20 laser shots/spectrum hardware average) would require an acquisition rate of 1000 spectra/s, which is not currently attainable. To circumvent this issue while maintaining 20 kHz laser repetition rates and the same spatial resolution the hardware average would need to be increased and the sample stage velocity decreased (Eq. 1). Unfortunately, this would then potentially cause ion signal intensity to diminish from increasing laser shot overlap. Laser repetition rates >10 kHz may not be necessary and in many cases may actually non-optimal due to either oversampling or the inability to save data fast enough.
Angiotensin II (m/z 1046.5) Signal Intensity as a Function of Stage Velocity Using Raster Sampling. Ion Image Areas were Averaged and Compared to Calculated Values for Spatial Resolution and Shots/Unit Area. Loss of Signal at Slower Sampling Speed is Shown to Correlate to Oversampling as Confirmed by Optical Images
Angiotensin II (m/z 1046.5) Signal Intensity as a Function of Repetition Rate Using Raster Sampling. Ion Image Areas were Averaged and Compared to Calculated Values for Spatial Resolution and Shots/Unit Area. Loss of Signal at Higher Laser Repetition Rates is Shown to Correlate to Oversampling as Confirmed by Optical Images
The experimental results described above are consistent with the hypothesis that for continuous laser raster sampling severe oversampling can occur under certain experimental conditions, leading to low ion signal intensities. As highlighted by the theoretical calculations (Figure 4a and b), with decreasing sample stage velocity and/or increasing laser repetition rate laser spot overlap increases rapidly, leading to poor spectra quality from oversampling. Although the experimental results are expected to vary slightly for different MALDI matrices, matrix deposition methods, and sample thickness, our results suggest that to prevent ion signal degradation experimental conditions should be such that laser shot overlap is <50 laser shots/unit area. These experiments set the framework for using continuous laser raster sampling to generate high quality ion images for MALDI imaging mass spectrometry.
As imaging mass spectrometry adapts to handle higher spatial resolution and larger sample area applications time and throughput become significant issues. We have used a prototype high-speed MALDI TOF mass spectrometer with continuous laser raster sampling capable of acquiring ion images more than 10× faster than commercially available MALDI imaging mass spectrometers. Sagittal rat brain lipid ion images acquired with 100 μm spatial resolution were collected in 10 min using a laser repetition rate of 3 kHz in typewriter motion raster sampling. These images were effectively acquired at a rate of 30 pixels/s. A systematic study of continuous laser raster sampling showed that poor ion signal intensity can result from severe oversampling due to exceedingly high laser repetition rates and/or slow sample stage velocity. These results suggest that for optimal performance, laser shot overlap should not exceed 50 laser shots/unit area. Theoretical calculations suggest that for moderate laser diameters (20–80 μm), laser repetition rates <5 kHz may be preferred and laser repetition rates >10 kHz can be unfavorable, even for small laser beam diameters, because of oversampling and the inability to currently write data to disc at sufficiently high rates. These results help in the development of the next generation of MALDI imaging mass spectrometry technology and its ability to handle the new technical and biological challenges imposed by speed, spatial resolution, and sample size.
The authors gratefully acknowledge financial support from NIH grants 5R01GM58008–12 and 1P41 031461-01. They thank Marvin Vestal, Matt Gabeler-Lee, Kevin Hayden, George Mills, and P. Jane Gale at Virgin Instruments for their help and advice.
- 12.Trim, P.J., Henson, C.M., Avery, J.L., McEwen, A., Snel, M.F., Claude, E., Marshall, P.S., West, A., Princivalle, A.P., Clench, M.R.: Matrix-Assisted Laser Desorption/Ionization-Ion Mobility Separation-Mass Spectrometry Imaging of Vinblastine in Whole Body Tissue Sections. Anal. Chem. 80, 8628–8634 (2008)CrossRefGoogle Scholar
- 15.Morita, Y., Ikegami, K., Goto-Inoue, N., Hayasaka, T., Zaima, N., Tanaka, H., Uehara, T., Setoguchi, T., Sakaguchi, T., Igarashi, H., Sugimura, H., Setou, M., Konno, H.: Imaging Mass Spectrometry of Gastric Carcinoma in Formalin-Fixed Paraffin-Embedded Tissue Microarray. Cancer Sci. 101, 267–273 (2010)CrossRefGoogle Scholar
- 19.Trim, P.J., Djidja, M., Atkinson, S.J., Oakes, K., Cole, L.M., Anderson, D.M.G., Hart, P.J., Francese, S., Clench, M.R.: Introduction of a 20 kHz Nd:YVO4 Laser into a Hybrid Quadrupole Time-of-Flight Mass Spectrometer for MALDI-MS Imaging. Anal. Bioanal. Chem. 397, 3409–3419 (2010)CrossRefGoogle Scholar
- 20.Simmons, D. A. Improved MALDI-MS Imaging Performance using Continuous Laser Rastering. Applied Biosystems Technical Note. 2008.Google Scholar
- 24.Vestal, M. L.; Hayden, K. M. High-Performance MALDI-TOF Mass Spectrometry. Proceedings of the ASMS Conference on Mass Spectrometry and Allied Topics, Indianapolis, USA, June (2007)Google Scholar