Three-Dimensional (3D) Imaging with Infrared Matrix-Assisted Laser Desorption Electrospray Ionization (IR-MALDESI) Mass Spectrometry

Abstract

Mass spectrometry imaging as a field has pushed its frontiers to three dimensions. Most three-dimensional mass spectrometry imaging (3D MSI) approaches require serial sectioning that results in a loss of biological information between analyzed slices and difficulty in reconstruction of 3D images. In this contribution, infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) was demonstrated to be applicable for 3D MSI that does not require sectioning because IR laser ablates material on a micrometer scale. A commercially available over-the-counter pharmaceutical (OTC) was used as a model to demonstrate the feasibility of IR-MALDESI for 3D MSI. Depth resolution (i.e., z-resolution) as a function of laser energy levels and density of ablated material was investigated. The best achievable depth resolution from a pill was 2.3 µm at 0.3 mJ/pulse. 2D and 3D MSI were performed on the tablet to show the distribution of pill-specific molecules. A 3D MSI analysis on a region of interest of 15 × 15 voxels across 50 layers was performed. Our results demonstrate that IR-MALDESI is feasible with 3D MSI on a pill and future work will be focused on analyses of biological tissues.

INTRODUCTION

Mass spectrometry imaging (MSI) reveals a glimpse into complex biological processes by simultaneously localizing different chemical species. MSI has been successfully applied in proteomics[1], lipidomics[2], metabolomics[3], pharmaceutical forensics[4], and other fields[5]. For example, monitoring how a drug or disease marker is spatially distributed helps to understand the mechanism of a disease[6]. Since biology is a three-dimensional phenomenon, three-dimensional mass spectrometry imaging (3D MSI) has the potential to uncover unique information that cannot be collected using traditional two-dimensional (2D) MSI approaches.

3D MSI has developed rapidly with emerging ionization methods, such as matrix-assisted laser desorption ionization (MALDI)[7], secondary ion mass spectrometry (SIMS)[8], desorption electrospray ionization (DESI)[9], and laser desorption ionization (LDI)[10]. There are two primary modes of 3D MSI: 1) serial-section-based and 2) ablation-based with the former being more popular[11]. In a typical 3D MSI study, the sample is sliced serially and individual 2D MSI experiments are performed on each section. Then consecutive 2D images are registered to reconstruct a 3D ion heat map for each analyte. This technique is an “add-on” to all regular mass spectrometry imaging methods because it puts no demand on instrumentation whereas the real difficulty resides on sample preparation, registration of images, and data-processing. Although many results have been reported using this protocol, a significant amount of sample is not analyzed because imaging occurs in a series of discrete layers. Also, registration of 2D images to a common coordinate system can be challenging due to sample deformation during sectioning, orientation of the sample layers, and natural variation in tissue shape with depth. Since 3D images have to be reconstructed, computational algorithms have been developed to stack images based on the original morphological information obtained with magnetic resonance imaging (MRI) and histological staining[12, 13]. These data processing steps make 3D MSI a time-consuming method.

The problems mentioned above for serial-section-based 3D imaging can be avoided with ablation-based 3D MSI, where no sectioning is needed. In the latter mode, an energetic ion or light beam ablates materials from the sample, continuously exposing new surface for imaging in the x-y plane. As ablation goes deeper, chemical changes across the z-direction are revealed in recorded mass spectra. Ablation-based MSI mode has been successfully applied in SIMS and laser ablation electrospray ionization (LAESI)[14, 15]. 3D SIMS MSI utilizes two beams: the first beam ejects atoms, molecules, and secondary ions from the surface while the second beam sputters the already analyzed surface to create a new plane for imaging the next layer[16]. These two ion beams are used iteratively to create 3D chemical maps of species. Based on this protocol, Castellanos and coworkers revealed that triacylglycerides were abundant in oocyte region of sugar-fed female Aedes aegypti mosquitoes[17]. Despite high spatial resolution in both the z and x-y directions[18], SIMS is not as soft as other ionization sources such as MALDI and ESI. Moreover, SIMS requires vacuum so it cannot be used to investigate volatile molecules. Alternatively, Nemes et al. used a laser to ablate material in a 3D manner with an ambient ionization technique named LAESI[15]. LAESI 3D MSI was achieved by creating depth profiles at each grid point in a region of interest (ROI). Six-layer 3D images for metabolites in Iynise leaf tissue were constructed with resolution of approximately 300 µm and 30 µm in the x-y and z directions, respectively.

For the first time, we used infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) for 3D MSI. IR-MALDESI was introduced by Muddiman in 2006[19] and is an atmospheric ionization source that combines features of laser desorption and electrospray ionization (ESI). Neutral materials are desorbed by an infrared laser and are then partitioned into charged droplets generated by an orthogonal ESI. IR-MALDESI allows for direct detection with a lateral resolution of 50 µm without oversampling[20]. IR-MALDESI has been used for 2D imaging of neurotransmitters[21], peptides[22], and other species[23], whereas 3D IR-MALDESI MSI has never been demonstrated before.

In this work, 3D IR-MALDESI MSI was performed by repeatedly collecting 2D images over the same ROI. In this proof-of-principle experiment, over-the-counter (OTC) pharmaceuticals produced in the form of multiple unit pellet system (MUPS) were used as models to test the feasibility of IR-MALDESI for 3D MSI. Spherical components inside each pill (Figure 1c) have different chemical compositions compared to that of surrounding powder outside spherical components and can be spatially resolved in 3D MSI. Depth resolution was explored on a full pill followed by 2D and 3D MSI performed on a half pill.

Figure 1.

Pills used as a model for demonstration; a) optical image of a full pill, where z-resolution at different energy levels was determined; b) optical image of a pill trimmed in half for 2D and 3D MSI; small circles are due to MUPS formulation; c) schematic of a pellet and its components where each color indicates different components.

EXPERIMENTAL

Materials

Prilosec OTC® (Cincinnati, OH, USA), an omeprazole delayed-release tablet, was purchased from a local pharmacy. All the tablets used in this study were produced in a lot number 8124171971. In the first part of the experiment, which determines depth resolution on a full pill (Figure 1a), no sample preparation was needed prior to imaging. In the second part of the experiment, the tablet was trimmed flat (Figure 1b) with a Leica CM1950 cryomicrotome (Buffalo Grove, IL, USA) for 2D and 3D MSI. HPLC-grade methanol and water were purchased from Burdick and Jackson (Muskegon, MI, USA). MS-grade formic acid was purchased from Sigma-Aldrich (St. Louis, MO, USA). Burn paper for laser focusing was purchased from ZAP-IT (ZFC-23; ZAP-IT, Concord, NH, USA).

IR-MALDESI System

An IR-MALDESI source coupled to a Q Exactive Plus mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) was used for all the imaging experiments presented. The set-up has been previously described in detail by others[24]. In short, neutrals ablated by laser partition into charged droplets from an orthogonal electrospray and are sampled into the mass spectrometer, with an injection time of 25 ms and automatic gain control (AGC) disabled. Mass spectra were collected across m/z 150–600 in positive mode with lock mass at m/z 371.1012 and mass resolving power 140,000 at m/z 200. Electrospray flow-rate was set to be 2 µL/min. A focused laser (IR-Opolette 2731, Opotek, Carlsbad, CA, USA) with a wavelength of 2940 nm was fired one laser pulse at each point within the region of interest (ROI). Repetition rate was set to be 20 Hz. More information about the IR-MALDESI interface can be found in Supplemental Table 1.

Depth Resolution Determination

Laser energies at 0.3 and 1.2 mJ/pulse were used to investigate dependence of ablation depths on laser energy. Laser energy was adjusted using a Q-switch which triggers laser pulses (Supplemental Table 2). Energy at target was measured with a laser power meter (Nova 2, Ophir, Jerusalem, Israel). Depth profiles after ablation of more than one layer were accurately measured with a confocal laser scanning microscope (VK-X1100, Keyence, Itasca, IL, USA). Depth profiles were used to calculate depth resolution, or in other words, depth of ablation region after firing one laser pulse per voxel.

2D MSI

The trimmed flat tablet was adhered to the imaging stage using a double-sided tape. ROI on the half pill was 20 × 45 voxels, with a spot-to-spot spacing of 150 µm. Laser energy at target was 1.2 mJ/pulse. We chose the highest energy level for 2D MSI to mimic parameters that we usually use for 2D IR-MALDESI MSI.

3D MSI

3D MSI was performed on a trimmed pill at a ROI of 15 × 15 voxels with a step size of 52 µm, using one of the lowest laser energies (0.3 mJ/pulse) to generate highest possible spatial resolution in z-direction. Prior to 3D IR-MALDESI, we estimated depth resolution by measuring depth profiles from ablation of 5 and 10 layers without stage height adjustment. To preserve laser focus during ablation of 50 layers in 3D IR-MALDESI MSI, the stage was moved up manually after analysis of every 3 layers.

Data processing

XCalibur raw data was converted to imzML format using an open source application raw to imzML converter[25]. Then imzML files were analyzed with MSiReader, a Matlab-based software developed in the Muddiman group[26, 27].

RESULTS AND DISCUSSION

Depth Resolution Determination on a Pill

Spatial resolution dictates the quality of generated ion heatmaps. Depth resolution was determined by measuring the depth of ablated spots. Lower laser energy and denser analyzed surface result in higher depth resolution (less material ablated per laser shot). Depth resolutions at the highest (1.2 mJ) and lowest (0.3 mJ) energy levels were explored. Optical and laser images after ablation of 1, 5, 10, 15 and 20 layers were measured with confocal laser scanning microscope (Figures 2a and 2c) and show that the ablation depth increased with the number of ablated layers. From side views of laser images in Figures 2a and 2c, the crater was deeper in the center compared to the edges due to the Gaussian nature of the beam. The ablation depth was measured by averaging depth profiles from 20 lines (at 2 µm increments) drawn across each ROI. Depth profiles show a linear relationship with layer numbers for both energy levels (Figures 2b and 2d). The average thickness of the outer pink enteric coating (65 µm) was measured using a Leica optical microscope. 3 and 30 layers had to be ablated to completely desorb the outer layer with the laser energy of 1.2 and 0.3 mJ/pulse, respectively. This is consistent with optical results shown in Figures 2a and 2c, where white powder shows up after ablation of 5 layers at 1.2 mJ/pulse energy level but never appears within 20 layers ablated at 0.3 mJ/pulse energy level.

Figure 2.

Ablation depth, optical and laser images were obtained using a confocal laser scanning microscope after ablation of 1, 5, 10, 15, and 20 layers. Optical and laser images show profiles after ablation at a) 1.2 mJ/pulse and c) 0.3 mJ/pulse. Color scale bars show ablation depth in µm. Linear relationship between ablation depth and layer number is shown for two laser energy levels: b) 1.2 mJ/pulse and d) 0.3 mJ/pulse. Depth resolution (depth of ablated region per layer) was estimated as weighted ablated depth: 20.1 µm for 1.2 mJ/pulse and 2.3 µm for 0.3 mJ/pulse. Spot size was 100 µm at 1.2 mJ/pulse and 78 µm at 0.3 mJ/pulse.

2D MSI on a Pill Microtomed in Half

2D MSI was performed on the half pill to resolve the chemical compositions of the pill. Three pill-specific molecules (Figure 3) were found inside the pill and used for demonstration of 3D IR-MALDESI MSI. Monomer of starch (C₆H₁₀O₅, m/z 163.0601) was distributed in the core of pellet as well as the surrounding powder outside pellets (Figure 4a); triethyl citrate (C₁₂H₂₀O₇, m/z 277.1282), a marker of enteric coating for delaying drug release, was distributed in the outer layer of pellet and surrounding powder among the pellets (Figure 4b); omeprazole (C₁₇H₁₉N₃O₃S, m/z 346.1220), the active ingredient, was distributed in the inner layer of pellet (Figure 4c). These three markers were successfully resolved in a half pill (Figure 4i) and their colocalization (Figure 4d) agrees with the pellet’s model scheme shown in Figure 1c.

Figure 3.

Representative mass spectra for a) active ingredient omeprazole (C₁₇H₁₉N₃O₃S), b) triethyl citrate (C₁₂H₂₀O₇), and c) monomer of starch (C₆H₁₀O₅). MMA denotes mass measurement accuracy.

Figure 4.

Three representative molecules were selected to show their distributions in the half pill: a) starch (m/z 163.0601), b) triethyl citrate (m/z 277.1282), and c) active ingredient omeprazole (m/z 346.1220). Figure 4d shows a colocalization of three molecules. Figures 4e and 4f show the distribution of A+1 ¹³C₁ (m/z 347.1253) and A+2 ³⁴S₁ (m/z 348.1178) isotopologues of omeprazole, respectively; Figure 4g shows a colocalization of monoisotopic peak A (red), A+1 ¹³C₁ (green) and A+2 ³⁴S₁ (blue) isotopologues of omeprazole with same gain; i) optical image of ROI. The laser spot size was 120 µm.

For the A+2 peaks of omeprazole, ³⁴S₁ (m/z 348.1176) and ¹³C₂ (m/z 348.1291) isotopologues are baseline resolved (Figure 3a). We can count the number of sulfurs and this approach allows for the determination of unique elemental compositions[28, 29]. It is important to note that although sulfur counting was not required here, given that the components in the OTC are known, that it is a very useful for unknowns. The number of sulfurs in omeprazole was calculated using Equation 1 and was estimated to be 1.00[28–30], using a theoretical sulfur abundance of 0.042 as reported by IUPAC[31].

$$Sulfur\hspace{0.17em}number=\frac{A+2_{Exp}}{A_{Exp}}/0.042$$ Equation 1

Also, the distributions of A+1 ¹³C₁ (m/z 347.1253, Figure 4e) and A+2 ³⁴S₁ (m/z 348.1178, Figure 4f) isotopologues of omeprazole colocalize with the monoisotopic peak (Figure 4g) with the same gain confirming the sensitivity of the imaging method.

3D MSI on the Pill Microtomed in Half

3D MSI was performed on a trimmed pill using the lowest and stable energy level of 0.3 mJ/pulse (Supplemental Table 2). Prior to 3D experiment, 5 and 10 layers (Supplemental Figures 1, 2) were ablated from the hall pill without stage height adjustments to determine the depth resolution which was calculated as weighed average of 16.3 μm, with ablation depth of 83.1 μm for 5 layers and 161.0 μm for 10 layers. To preserve laser focus during ablation of 50 layers in 3D fashion, the height of translational stage was adjusted after every three ablated layers. We didn’t adjust the stage height after each ablated layer because of the stage precision in the z-direction (Supplemental Table 1). One tick mark on the adjustment knob is 25 µm while depth resolution is 16.3 µm per layer. After every three layers imaged (16.3 × 3 = 48.9 µm), we moved the stage up by two tick marks (~50 µm). This is the smallest increment we could achieve for this experiment. Ablation depth after 50 layers was measured as 840.4 μm, with an average depth resolution of 16.3 μm/layer with stage height adjustment, compared to 7.9 μm/layer without stage height adjustment.

The heat maps for three representative molecules (Figure 5) and their colocalization (Supplemental Figure 3b) show the process that one pellet was trimmed away and then another pellet was imaged. The distribution of ³⁴S isotopologue (Supplemental Figure 3a) confirms the presence of omeprazole across 50 layers. Constructed 3D heat maps (Figure 6) show distribution of three representative markers in three-dimensional space.

Figure 5.

Ion abundance and distribution for A peaks of a) starch), b) triethyl citrate and c) omeprazole on the half pill across 50 layers, with laser spot size 80 µm and depth resolution 16.3 µm.

Figure 6.

Three-dimensional intensity maps for a) starch, b) triethyl citrate, c) omeprazole; d) colocalization of three markers in the pill, with laser spot size 80 µm and depth resolution 16.3 µm.

In this work, we demonstrated utility of IR-MALDESI for 3D MSI, which offers an alternative method to traditional 3D MSI. No excessive sample preparation steps nor assistive morphological information (e.g., MRI) is needed for imaging registration. The main source of variance comes from inconsistent laser ablation performance from layer to layer, resulting in sharp ablation profiles as shown in Figures 2a and 2c. This is due to the Gaussian laser beam, where higher energy is distributed in the center. Inconsistent ablation profiles could be avoided with beam homogenizer, which creates uniform beam profiles. However, the objective of this work was to test that the current IR-MALDESI system is applicable for 3D MSI, rather than to develop a robust method for 3D IR-MALDESI.

All in all, the presented work provides workflow for ablation-based 3D IR-MALDESI by presenting 3D MSI analysis on a 15×15 ROI across 50-layers within 70 minutes. IR-MALDESI has proven to be feasible for 3D MSI on hard OTC tablets. Future work will be focused on development of 3D IR-MALDESI MSI for analyses of biological samples.

CONCLUSIONS

The OTC tablet that was formulated using MUPS was analyzed using IR-MALDESI source coupled with a Q Exactive Plus mass spectrometer to evaluate the feasibility of the former for ablation-based 3D MSI. We also investigated the depth of ablated spots at different laser energy levels to determine a range for depth resolution. The highest and lowest depth resolutions were measured as 2.3 µm at 0.3 mJ/pulse and 20.1 µm at 1.2 mJ/pulse, respectively. Minimal to no sample preparation makes IR-MALDESI an alternative MSI tool for analyses of biological samples. Therefore, future work will be focused on developing reproducible methods for 3D IR-MALDESI MSI of biological tissues.