5.3 Desorption/Ionization Techniques used for MSI
5.3.1 Matrix-Assisted Laser Desorption/Ionization (MALDI)
MALDI is arguably the most popular and commercially developed technique deployed for MSI of lipids. In typical MALDI, analyte molecules are embedded into
a large excess of a crystalline chemical matrix, most often small organic molecules, and irradiated with nanosecond‐long laser pulses generally at 337, 349, or 355 nm.
The optical properties of the matrix and laser wavelength are matched such that the laser pulse is absorbed by the matrix, leading to an explosive disintegration of the crystal lattice. Analyte molecules, largely transparent to the employed laser light, are co‐desorbed in the rapid process and transferred to the gasphase with a low thermal burden and minimal fragmentation [34]. Ionization mechanisms in MALDI are still under lively discussion and may involve different pathways, including pho- toactivated gas‐phase charge‐transfer processes and charge separation of pre‐formed ions [35–37]. For sample preparation in conventional MALDI analysis of extracts and standards, the analyte and matrix are mixed in a solution and deposited onto a surface, resulting in analyte–matrix co‐crystallization. Upon drying, more polar compounds are built into defects within the matrix crystal lattice [38]. Because of a mismatch in solubility, less polar analyte molecules such as lipids are more likely to cover the surface of matrix microcrystals [39]. Matrix application for MALDI‐MSI is more complicated as a matrix must be deposited in a manner that minimizes ana- lyte delocalization. The microcrystalline layer must be thin and homogeneous, and the crystal size should be smaller than the intended pixel size [40]. Most commonly for MSI applications, the matrix is applied by pneumatic spraying. Here, a fine mist of matrix solution is deposited onto the sample in cycles, leading to iterative micro- extraction within each droplet and fast evaporation of the solvent [41]. The proper- ties of the resulting matrix layer such as analyte extraction, sample thickness, crystal size, and homogeneity are governed by spray parameters such as choice of solvent system, flow rate, distances between the sprayer and the sample, number of spray cycles, and spray temperature [42]. Matrix application by sublimation is also well suited for MALDI‐MSI of lipids and brings the advantages of generating smaller (submicron) crystal sizes and further minimizes delocalization because of the solvent‐free nature of the approach, making it often the method of choice for high spatial resolution analyses [43]. For this, a reservoir of the matrix and the sample are placed inside a vacuum. Heating the matrix powder leads to sublimation and re‐deposition on the surface of the cooled sample, forming a microcrystalline and porous layer. Migration into the matrix layer for lipids is facilitated by their inherent fluidity and driven by capillary forces, surface wetting, and diffusion [39]. Analyte incorporation can be improved by “recrystallization” methods that include heat treatment and interaction with solvent vapor [44].
Table 5.1 lists the most popular matrices for MALDI‐MSI of lipids and their pre- ferred field of application and polarity. The matrix choice has to be adapted to the individual sample, lipids of interest, and instrumentation [45, 46]. While the use of more volatile matrices such as 2,6‐dihydroxyacetophenone (DHAP) is restricted in high‐vacuum instruments [47], matrices such as α‐cyano‐4‐hydroxycinnamic acid (CHCA) tend to produce sizable chemical backgrounds because of the formation of clusters at elevated and atmospheric pressure conditions.
From a hardware perspective, the pixel size in MALDI‐MSI is limited by the size of the laser ablation crater that depends on the focused spot size and the required laser fluence. While commercially available instruments can routinely reach a pixel
5 Mass Spectrometry Imaging of Lipids 122
size of 10–15 μm, more specialized applications have reported pixels as small as 600 nm [48]. MALDI can detect many different lipid species, with phospholipids (PLs) and sphingolipids generally giving the highest signal intensities. The types of ions detected with MALDI across both positive‐ and negative‐ion modes are similar to other MSI approaches and described in Section 5.7.1. Demonstrating the suitabil- ity of MALDI for lipid detection, Bowman et al. tentatively assign 702 individual lipid ion species in the positive‐ion mode from a total of 2643 peaks based on accu- rate mass alone, using the ultra‐high resolving power of a 21T Fourier transform ion cyclotron resonance mass spectrometer (FTICR‐MS) and a mass accuracy of 150 ppb [49].
The sensitivity for certain lipid classes can be improved using post‐ionization approaches. In a technique termed MALDI‐2, a second laser pulse interacts with the evolving MALDI‐plume, producing many additional matrix ions via resonant mul- tiphoton ionization [50]. Under the right ion source pressure conditions (>2 mbar), subsequent gas‐phase reactions initiated by the ionized matrix distribute the charge among a wide variety of lipid classes and boost signal intensities up to several orders of magnitude [51]. Bowman et al. report a ~sixfold increase in the number of tenta- tively assigned lipids in positive‐ion mode analysis of rat liver tissue at pixel sizes between 6 and 20 μm when using MALDI‐2 [52]. Applying this approach to imaging lipids in human multiple sclerosis tissue at pixel sizes of 6 μm using oversampling 147 lipids identified at the sum‐composition level by accurate mass measurements were successfully imaged, including cholesterol ester species that localized to lipid droplets with diameters of ~10–30 μm. Other methods such as plasma‐based post‐
ionization can also yield significant increases in sensitivity and lipid coverage but have not yet been explored to the same extent as MALDI‐2 [53]. The increasing sensitivity provided by ever‐improving instrumentation and post‐ionization approaches such as MALDI‐2 and technical advancements in the miniaturization of laser spot size allow for MALDI‐MSI with a pixel size in the submicron range Table 5.1 Popular matrices for MALDI-MSI of lipids.
Substance Abbreviation Preferred polarity Vacuum stability
2,5‐Dihydroxybenzoic acid DHB Positive High
α‐Cyano‐4‐hydroxycinnamic acid CHCA/HCCA Positive High 1,5‐Diaminonaphtalene DAN Positive/Negative Medium 2,5‐Dihydroxyacetophenone 2,5‐DHA/DHAP Positive/Negative Medium 2,6‐Dihydroxyacetophenone 2,6‐DHA Positive/Negative Low 2,4,6‐Trihydroxyacetophenone THAP Positive/Negative Low
Norharmane NRM/NORH Negative High
9‐aminoacridine 9‐AA Negative Medium
N‐(1‐naphthyl) ethylenediamine
dihydrochloride NEDC Negative Medium
[18, 48, 54]. This opens the door to the analysis of intact lipids on a subcellular level as demonstrated in Figure 5.1a,b.
With a number of platforms for MALDI‐MSI commercially available, the tech- nique has been used in a large number of applications across numerous tissue types [57, 58]. Next to the targeted spatial analysis of a specific lipid species or lipid class, an untargeted approach can be used to identify potential lipid biomarkers. An example for this procedure was presented by Paine et al., utilizing the improvements in acquisition speed to record three‐dimensional data of medulloblastoma in a mouse model from a total of 49 sagittal sections (Figure 5.1c). This enabled the identifica- tion of specific lipid species such as PE(16:0_24:1) or PIP2(18:0_20:4) with decreas- ing or increasing signal intensities between metastasizing and non‐ metastasizing tumors [55]. As a non‐mammalian example, Khalil et al. used atmospheric pressure
100 àm
(b) (a)
(d)
0.6 μm 0.8 μm
1.0 μm
1.5 μm
2.0 μm
10 μm 100 μm
tMALDI-2MALDI-2
(c) LPA(18:2) 0 6.69E3SM(34:1) 0 4.70E4
LPE(16:1) 0 9.44E3SM(36:1)
PG(34:1) 0 1.11E4 0 3.62E4
LPC(18:1) 0 1.08E5PA(38:3)
PG(36:4) 0 1.22E4 0 2.48E4
PC(34:3) 0 1.00E5PE(32:1)
PC(36:2) 0 8.00E4 0 1.14E5
PI(P-34:2)0 3.67E3PI(P-34:3)
PC(36:4) 0 2.51E5 0 5.06E3
Figure 5.1 (a) Overlay portraying the distribution of blue, m/z = 577.519, [DAG(34:1)]+; orange, m/z = 784.525, PE(36:1), [M + K]+; and green, m/z = 828.692, HexCer(42:1-OH), [M + H]+ in mouse cerebellum tissue. The top five panels were recorded with the
transmission mode -MALDI-2 at the indicated pixel size. The bottom panel was recorded in top illumination geometry at a pixel size of 10.0 μm. (b) Ion intensity distribution of [PC(36:2)+H]+ in Vero-B4 cell cultures measured by t-MALDI-2-MSI using DHAP. (c) 3-D reconstructions of 49 aligned sagittal sections from a ND2:SmoA1 transgenic mouse brain containing a non-metastasizing cerebellum tumor using three representative channels: m/z 790.5 in blue, m/z 888.6 in green, and m/z 885.5 in red. Visualization of two sagittal (left) and two coronal (right) virtual sections. (d) Positive-ion AP-SMALDI-MS images obtained from a whole-body section of Anopheles stephensi. Each image represents the sum of intensities of the corresponding [M+H]+, [M+Na]+, and [M+K]+ adducts. Source: (a) Adapted with permission of Niehaus et al. [48], Springer Nature, (b) Adapted with permission of Bien et al. [18], American Chemical Society, (c) Paine et al. [55]/Springer Nature/CC BY 4.0, (d) Adapted with permission of Khalil et al. [56], American Chemical Society.
5 Mass Spectrometry Imaging of Lipids 124
MALDI‐MSI to investigate phospholipid distribution in the whole‐body section of Anopheles stephensi mosquitoes (Figure 5.1d) [56]. In the positive‐ion mode, a wide range of lipid species from a number of phospho‐ and sphingolipid classes displayed specific distributions throughout the insect’s body.