In living organisms, the compositions of biomolecules such as proteins, nucleic acids, sugars, lipids, and metabolites change continuously and dynamically in response to a variety of environmental factors. The levels of these molecules in living tissues are precisely regulated to maintain homeostasis. Therefore, in many cases, the distribution of these molecules in tissues or cells can provide valuable information for basic biological research, diagnosis of certain diseases, and identification of therapeutic targets. Conventionally, the biodistribution of such molecules has been examined by cell/tissue ‘homogenization’ methods, which provide only biochemical information, not information on the spatial distribution of target molecules in tissues. On the contrary, spectroscopic visualizations such as fluorescence tissue imaging can reveal the spatial distribution of molecules; however, they do not allow for molecular analysis. For these reasons, imaging mass spectrometry (IMS) has been developed and extensively studied; IMS allows for both visualization of tissues to give positional information on targets and molecular analysis of targets by affording the molecular weights (1-3). In addition, the advantages of mass spectrometry (MS) are manifold: i) MS provides chemical and structural information on targets, ii) MS can easily discriminate between true analytes and background and, therefore, eliminate false positive signals, and iii) MS can be used to monitor multiple analytes simultaneously. Matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) MS is particularly effective and is generally used for IMS due to its suitability for tissue analysis and the ‘soft’ ionization of large biomolecules such as proteins and oligonucleotides (4, 5). In general, a solution of organic matrix, including 2,5-dihydroxybenzoic acid, sinapinic acid, a-cyano-4-hydroxycinnamic acid, and 2,4,6-trihydroxyacetophenone, depending on the types of analytes present, is deposited on a thin section of a tissue, which is then scanned by MALDI-TOF MS to give a raster image of the distribution of biomolecules as revealed by relative signal intensities. However, the requirement for an organic matrix limits the applicability of MALDI IMS. The signals in MALDI IMS, in many cases, are strongly affected by the choice of a proper matrix and solvents, co-crystallization of a matrix and analytes, and, particularly, homogeneity of matrix deposition, leading to poor shot-to-shot and sample-to-sample reliability. Therefore, peak intensities, i.e. raster images, would not represent the real spatial distribution of target molecules in tissues. In addition, the requirement for an organic matrix hampers imaging of small molecules such as drugs and their metabolites owing to interference of the matrix in the low-mass region. Thus, various organic matrix-free LDI IMS systems have been developed to avoid the problems described above, mostly utilizing nanostructured surfaces and inorganic nanoparticles (NPs) as an alternative to the organic matrix.

This minireview starts with a brief introduction to organic matrix-free LDI MS for small molecule analysis and focuses on the progress that has been made in organic matrix-free LDI IMS methods, which are categorized into i) nanostructured surface-assisted LDI IMS and ii) inorganic nanoparticle-assisted LDI IMS. In addition, IMS methods using other ‘soft’ ionization methods such as desorption electrospray ionization (DESI) and laser ablation electrospray ionization (LAESI) will be discussed, followed by a brief discussion on the use of secondary ion mass spectrometry (SIMS), which have the advantages of little or no preparation, ease of implementation, and simplified analysis in ambient environments.

The main purpose of this minireview is to give the reader an unbiased description of the approaches for molecular level analyses of biological sample surfaces using MS, particularly for small molecules. This minireview, therefore, will not only offer a starting point for students and researchers entering this field but also be valued by active researchers requiring small molecule analyses of tissues in various areas including disease pathology, diagnostics, drug delivery systems, metabolomics, lipidomics, and pharmacokinetics.

Organic matrix-free LDI MS mostly utilizes nanostructured surfaces and inorganic NPs and is known as surface-assisted LDI (SALDI) MS (6-8). SALDI materials transfer sufficient energy from the irradiated laser to analytes for desorption/ionization without damaging the analytes and causing fragmentation, and therefore have been used as matrices for analysis of small molecules. In addition, nanostructures of the SALDI materials can provide efficient loading capacities due to large surface area, and analytes can be concentrated on the NPs by ionic strength, hydrophobic interactions, covalent binding, or bio-specific interactions through surface modifications, resulting in high sensitivity (9). As a pioneering work, Siuzdak and co-workers reported desorption-ionization on a porous silicon (DIOS) surface which was produced from flat nanocrystalline silicon through a simple etching procedure (10). Small molecule analytes including peptides (m/z 500-2000), small organic molecules (m/z 150-650), and saccharides (m/z 200-350) can successfully be analyzed using an organic matrix-free format on a porous silicon surface. In addition to DIOS, tailored surfaces with nanostructures such as layer-by-layer films of gold nanoparticles (AuNPs) (11), Si nanowires (12), and graphite-coated films (13) have reportedly been used as nanostructured surfaces. Inspired by the first example by Tanaka et al., in which 30 nm cobalt NPs mixed with glycerol were used as a matrix (4), various inorganic NPs including Au, Ag, Pt, SiO₂, TiO₂, Fe₂O₄, and ZnO have been widely used and examined for their compatibility in LDI MS (14). Particularly, AuNPs are most commonly used in biological studies owing to advantageous optical and physiochemical properties. There are pros and cons to using surface-type and particle-type organic matrix-free formats; for example, surface-type formats are robust and can be manufactured as a target plate for MALDI-TOF MS instruments, whereas particle-type formats facilitate protocol optimization for MS analysis and can act as a solid phase for extraction/enrichment of analytes in complex samples.

The following two sections describe organic matrix-free LDI IMS with grafting, the distinct feature of SALDI MS in small molecule analysis, as discussed above: i) nanostructured surfaces on which thin tissues are deposited, and ii) inorganic NPs which are sprayed onto thin tissues (Fig. 1).

Figure 1. LDI IMS with organic matrix-free systems harnessing nanostructured surfaces or nanoparticles.

Noncarbon-based surfaces

By utilizing the DIOS technique, which allows for small molecule analysis, Liu et al. accomplished imaging of small molecules in biological tissues at the cellular level on a porous silicon substrate (15). In that study, phosphatidylcholine in mouse liver tissue was analyzed by LDI MS, and then further indirect visualization of mammalian cells was demonstrated by constructing an ion map of the detected phosphatidylcholine. Rudd et al. investigated the biological role of two classes of secondary metabolites, brominated indoles and choline esters, in reproduction of Muricidae molluscs by observing changes in distribution of metabolites in mollusc tissue using DIOS (16, 17). IMS facilitated detection of the metabolites at different stages of the reproductive cycle of mature female D. orbita, one of molluscs, and proposed the biological roles of the metabolites based on temporal changes in their distribution. The DIOS substrates were further employed for the detection of drugs, lipids and metabolites (18, 19).

As an advanced use of DIOS, nanostructure initiator MS (NIMS) was introduced by the Siuzdak group for porous silicon-based mass analysis (20). NIMS uses a nanostructured silicon surface composed of roughly 10 nm pores to trap initiator molecules, such as fluorinated siloxane, lauric acid, and polysiloxane. The NIMS surface is exposed to laser irradiation resulting in vaporization or fragmentation of the initiator molecules and subsequent desorption/ionization of the absorbed analyte on the NIMS surface. The lipids (m/z 700-800) on the mouse embryo tissue section were visualized by NIMS (20), and this method was further extended to clinical applications for analysis of xenobiotics (m/z 200-350) and endogenous metabolites (m/z 150-350) in brain tissue and fluids with high sensitivity, no fragmentation, and no background interference (21). Although NIMS is well suited for the detection of various biological molecules in tissues, detection of carbohydrates and steroids is challenging because of their poor ionization efficiency. For this reason, Patti et al. combined NIMS and spray deposition of NaCl or AgNO₃, which provided a uniform environment with a source of cations ([M+ Na]⁺ or [M+Ag]⁺), for analysis of carbohydrates (m/z 180-400) and steroids (m/z 493) in Gerbera jamesonii stem and mouse brain tissue, respectively (22). This technique was further developed for imaging with essentially single-cell resolution, and NIMS was used to monitor drug exposure and metabolite biotransformation in single cells for the study of cancer metabolism (23). In addition, Ag films were sputter-coated thinly on porous silicon substrates, resulting in a dramatic improvement in the mass accuracy of fingerprint NIMS (24). For a discussion of the basic principles of NIMS for tissue imaging and various surface modifications and alternative nanostructured surfaces for NIMS, see the review by Calavia et al. (25).

Recently, Li et al. prepared a nanostructured surface consisting of a silver nanoisland on nanodiamonds which had a localized surface plasmon resonance wavelength very close to the laser wavelength of MALDI MS instruments, therefore demonstrating the enhanced efficiency of LDI for small molecule analysis (26). Those authors went on to apply the silver nanoisland surface to hexagonal boron nitride for IMS of ischemic brain damage with both small metabolites (m/z > 500) and sulfatides resulting in enhanced signal intensity of many small molecules and dramatic improvement in the visibility of raster images (27).

Carbon-based surfaces

In addition to porous silicon-based surfaces, carbon-based surfaces have also been used for matrix-free IMS. Many types of carbon-based materials, including functionalized carbon nanotubes and graphene oxides, have been suggested as alternative matrices for matrix-free IMS due to their absorption properties and efficient energy transfer to analytes. Kim et al. developed an LDI platform using a double layer of graphene oxide (GO) and aminated multi-walled carbon nanotube (MWCNT) for mouse brain tissue imaging (28). The deposition of aminated MWCNTs to the GO-coated surface yielded enhanced surface roughness and surface area for analyte adsorption, and, thus, increased LDI efficiency. This double layer effectively converted the absorbed UV light into thermal energy, allowing for imaging of glycerophosphocholine (m/z 800-860) and phosphatidylcholine (m/z 730-755) in mouse brain tissue as well as MS analysis of various biochemical small molecules. They also prepared multilayers of alternating MWCNT and GO and investigated the effect of thickness, assembly sequence, and surface roughness on LDI efficiency for small molecule analysis and IMS (29). Huang et al. demonstrated an interesting approach in which the protein mucin1, which is overexpressed in most adenocarcinomas, was utilized as a target for tumor cell analysis with LDI IMS (30). In their approach, a mucin1-binding aptamer was conjugated to AuNPs which were subsequently immobilized on a GO-coated surface. The resulting surface provided the mucin1-binding aptamers with ultrahigh density and high flexibility for cooperative and multivalent binding of mucin1 on cell membranes. By using LDI-MS to monitor Au cluster ions, four different mucin1 expression cell lines were analyzed on this surface, and this platform could be utilized further as a labeling agent for tumor tissue imaging. As additional carbon-based surfaces, a pulse laser-engineered functional graphene paper with graphitic nanospheres and a graphene-coated glass substrate combined with a continuous-wave laser for atmospheric pressure mass spectrometric analysis were demonstrated for IMS (31, 32).

Inorganic nanoparticles

The performance of inorganic NPs as matrices for LDI MS depends strongly on their size, morphology, composition, and concentration. In this respect, AuNPs and AgNPs are the most often studied and widely used materials in LDI-MS because their size is readily tunable and various shapes and compositions can be prepared depending on the researcher’s purpose. Goto-Inoue et al. visualized the distribution and localization of glycosphingolipids (m/z < 950) – amphiphilic molecules involved in various biological processes – in mouse brain sections using AuNPs as a matrix (33). Compared with dihydroxybenzoic acid, a conventional organic matrix for glycosphingolipids, AuNPs provided approximately 20 times more sensitive detection of glycosphingolipids and successfully enabled visualization of ionic images of glycosphingolipids including 14 kinds of sulfatides and 10 kinds of gangliosides in mouse brain samples. Recently, Phan et al. compared three different sample preparation methods, including sublimation with two conventional organic matrices followed by recrystallization with trifluoroacetic acid, and surface modification with AuNPs to profile and image the lipids (m/z 500-900) in Drosophila brain tissue, which is an important model organism used in biological and neurological studies (34). They suggested that the different sample preparation methods made with a particular matrix material are suitable for detection of different biomolecules in the fly brain, and therefore, complementary analysis using a suitable matrix will allow for precise and diverse imaging of lipids in tissue samples.

AgNPs are also actively used for IMS. Hayasaka et al. harnessed AgNPs modified with alkylcarboxylate and alkylamine to visualize fatty acids (m/z < 300), such as stearic, oleic, linoleic, arachidonic, and eicosapentaenoic acid, and palmitic acid in mouse liver that are not detectable using a dihydroxybenzoic acid matrix (35). Additionally, they prepared mouse retinal tissue sections with a thickness of 10 mm and analyzed them with a scan pitch of 10 mm using AgNPs, revealing the six-zonal distribution of fatty acids in different layers of the retina. The Wood group introduced an AgNP implantation method where AgNPs were formed with a magnetron and accelerated to 50 eV to be deposited across entire tissue sections for uniform and reproducible AgNP layer formation (36, 37). Gas phase AgNP ions were generated by magnetron sputtering and grown to 0.5-15 nm diameter, and then selected particles at 6 nm were implanted in the tissue section. This method produced high spatial resolution LDI images of various lipid species (m/z 600-1700) in rat heart and kidney tissues. Similarly, Dufresne et al. conducted simultaneous imaging of cholesterol along with olefins using metallic silver coatings at nanometer thicknesses with a sputter coating system (38). The Yeung group performed profiling and imaging of metabolites (m/z 400-700) on plant leaves, flowers, and roots by utilizing IMS with spray deposition of AgNPs (39, 40).

Since the first introduction of the graphite surface-assisted detection of proteolytic digests (m/z 150-1700) of cytochrome c by Sunner et al. (41), colloidal graphite materials have also been utilized for IMS. Yeung’s group used an aerosol spray of colloidal graphite for the detection and localization of cerebrosides (m/z 800-880) in rat brain tissue (42) and small metabolites (m/z 130-450) such as organic acids, flavonoids, and oligosaccharides in fruits (43). Furthermore, they analyzed flavonoids (m/z 285-755) and cuticular waxes (m/z 280-615) in intact leaves of plants such as Arabidopsis, one of the most important model systems in plant biology (44).

Metal oxide nanoparticles

NPs of metal oxides have been utilized for SALDI materials due to their unique structures and compositions. The Setou group demonstrated IMS of lipids and peptides at cellular resolution (15 mm) using extremely small iron oxide NPs (3.7 nm in diameter) flanked by amorphous silicates with hydroxyl and amino groups on their surfaces (45). These hydrophilic functional groups facilitate the ionization of adsorbed analytes through not only efficient energy transfer but also preferential sodium/potassium adduct formation. This material was further used to determine the distribution of sulfatide (m/z < 910) in the dentate gyrus region of the hippocampus (46). They also utilized TiO₂ NPs for LDI MS identification and visualization of low molecular weight metabolites in mouse brain tissue (47). In their work, IMS identification of the metabolites using TiO₂ NPs, AuNPs, and dihydroxybenzoic acid as matrix molecules were compared. While only 4 signals were specific to dihydroxybenzoic acid, 179 metabolites were specific to TiO₂ NPs. In addition, TiO₂ NPs provided a higher number of molecular signals than AuNPs without any NP-related peaks in visualization of mouse brain tissues.

Recently, metal oxide laser ionization (MOLI) MS was reported as an organic matrix-free system for lipid analysis using powders of various metal oxides such as ZnO, MgO, Fe_xO_y, Co_xO_y, and CuO as an alternative to organic matrices (48). In this method, lipid molecules were ionized by protonation or sodiation which can be attributed to Lewis acid-base interactions between analytes and metal oxide. As such, the MOLI MS method can offer a new approach for the analysis of lipids. For example, CaO as a matrix replacement provided reproducible lipid cleavage, enabling lipid profiling for bacterial identification (49, 50). MOLI MS was also applied to IMS by Basu et al., who used cerium(IV) oxide to induce laser-catalyzed fatty acyl cleavage for bacterial identification, and to detect and image fatty acids in brain tumor tissues (51).

Electrospray ionization-based IMS

The Cooks group first introduced the DESI method for analysis of diverse analytes including small non-polar compounds, peptides, and proteins (52). In this method, solvent microdroplets were electrosprayed onto the sample surface, and the impact of electrosprayed charged particles on the surface resulted in desorption and ionization of analytes. The resulting desorbed gas-phase ions were then transferred to a distant mass spectrometer which gave mass spectra similar to normal ESI MS. Using DESI MS, small molecule RDX (m/z 334) and coniceine (m/z 125) were observed on a porcine leather surface and Conium maculatum seed section, respectively. Furthermore, the same group carried out in vivo sampling of living tissue surfaces by analyzing the antihistamine loratadine on the human finger, suggesting that DESI can be used for IMS of biological materials in ambient conditions. As an expansion of DESI MS to IMS, the Cooks group reported the direct and specific determination of the distribution of epinephrine (m/z 184.1) and norepinephrine (m/z 170.1) along with various phospholipids (m/z 750-1000) in the porcine adrenal gland (53). In addition, localization of lipids in human prostate cancer and injured rat spiral cord tissue (54, 55) and secondary metabolites in plant materials (56) were also verified. Furthermore, three-dimensional images were constructed from a suitable set of two-dimensional images obtained using DESI IMS (57). For more details regarding DESI IMS, see the recent review by Parrot et al., which includes discussion of ionization mechanisms, sample preparations, and applications (58).

Another approach to electrospray ionization-based IMS, the combination of infrared (IR) laser ablation with electrospray ionization (LAESI), was introduced by the Vertes group (59). In LAESI-MS, biological and medical analytes and organisms with sufficient water content are analyzed using a mid-IR laser at 2940 nm, corresponding to the frequency of the O-H bond’s vibration in water, resulting in strong absorption of the wavelength by the water. Because the sample absorbs mid-IR energy, a gas phase plume is created from the sample surface. These laser-ablated particulates from the sample surface then interact with electrospray droplets, which provide a source of ions, allowing ionization of the laser-ablated particulates and subsequent analysis by a detector. Using this method, excretion of the antihistamine fexofenadine (m/z 502.3) in urine samples of humans who had taken fexofenadine caplets orally was analyzed without the use of organic matrices (59). In addition, in vivo spatial metabolite profiling of French marigold seedlings was performed to observe the various metabolite peaks (m/z 150-770) on the leaf, stem, and root. As a further extension of this strategy for atmospheric pressure IMS, that group reported the distribution of various lipids (m/z 130-770) and metabolites (m/z 100-900) in mouse brain and plant leaf tissue sections (60, 61). The LAESI method was also applied for in situ profiling of metabolites in single cells, which is challenging due to the complexity and small size of the samples. Cells exhibit a large degree of metabolic diversity depending on age, nutrition, and environmental factors, and therefore, chemical imaging and analysis of individual cells in a cell population would have broad applicability in biomedical research and clinical diagnostics. The Vertes group utilized LAESI for metabolic profiling (62) and in situ cell-by-cell imaging (63) of single cells in different cell populations. The localization of various metabolites (m/z 100-650) was determined in onion and daffodil epidermal cells.

IMS with TOF-SIMS

SIMS is a desorption and ionization technique used to analyze the composition of surfaces by sputtering the surfaces with an energetic primary ion beam and analyzing secondary ions emitted from the surfaces (1, 64, 65). These secondary ions can directly produce high-resolution chemical images, so this platform is well-suited for the analysis of surface composition of biological materials. Although LDI- and ESI-based methods are widely used for visualization of molecular distribution on biological surfaces due to their efficiency and simplicity, these methods produce limited-resolution raster images. In this respect, SIMS ionization is advantageous over these methods, as it allows not only high mass resolution but also high spatial resolution of low molecular weight analytes. In terms of spatial resolution, a commonly used LDI method is capable of resolution as small as 5-100 mm because it uses laser light which has a focused spot size as small as 1 mm. However, SIMS offers enhanced resolution because it uses a primary ion beam that can be focused as sharply as 10 nm, allowing IMS of single cells and even different organelles within cells. For example, Todd et al. reported organic ion imaging of rat brain tissues with an enhanced resolution using TOF-SIMS (66). In general, the monatomic primary ions (Ar⁺, Ga⁺, In⁺, Au⁺, Xe⁺, Bi⁺) are commonly used for SIMS, which sometimes causes extensive fragmentation of analytes, hampering highly sensitive IMS of tissue surfaces. This drawback can be overcome by using lower energetic polyatomic primary ion beams such as C₆₀⁺, SF₅⁺, Bi₃⁺, Au_n⁺, and Cs_n⁺ for IMS, which does not result in extensive fragmentation (67, 68). Furthermore, TOF-SIMS has been successfully employed in 3?D IMS of biological systems. Fletcher et al. reported the visualization of cellular features and 3?dimensional mass spectral imaging using C₆₀⁺ as a primary ion source (69). Phosphocholine and adenine in HeLa cells were detected and analyzed for MSI of single cells and identifying cellular organelles such as the nucleus and endoplasmic reticulum. Thus, TOF-SIMS can provide favorable conditions for tissue imaging as an alternative to LDI- and ESI-based IMS techniques owing to the highly enhanced resolution and capability of 3D imaging. On the other hand, TOF-SIMS suffers from the drawback of fragmentation of surface molecules and the high cost of equipment compared with LDI- and ESI-based facilities resulting in limited accessibility. Therefore, researchers must use caution when determining which analytical tools to use depending on the specifics of their biological samples and target molecules. Analytical methods and target analytes discussed in this review are summarized in Table 1.

Table 1

Table 1 Analytical methods and target analytes discussed in this review

Analytical Method			Analyte	Imaging Target	Ref.
LDI-MS	DIOS	Porous silicon surface	Phosphatidylcholine (PC)	Mouse liver tissues, HEK 293 cells	15
			Metabolites	Molluscs tissues	16, 17
				Fingerprints	18, 19
	NIMS	Initiator coated surface	Lipids	Mouse embryo tissue	20
				Fingerprints	24
			Metabolites (Clozapine, Ketamine)	Mouse brain tissues	21
			Glucose, steroids	Gerbera jamesonii stem	22
			Cholesterol	Mouse brain tissues	23
	Carbon-based Surface	Graphene oxides, Carbon nanotubes	Glycerophosphocholine, phosphatidylcholine	Mouse brain tissues	29
			Mucin 1	Tumor tissues	30
			Adenine	Hippocampal tissues	32
	Inorganic NP	Au	Glycosphingolipids	Mouse brain tissues	33
			Lipids	Drosophila brain tissues	34
		Ag	Fatty acids	Mouse liver and retinal tissues	35
			Lipids	Rat heart tissues, Rat kidney tissues	36, 37, 38
			Metabolites	Plants (flower, root)	39, 40
		Fe3O4	Lipids, peptides	Rat cerebellum tissues	45
			Sulfatides	Rat hippocampal tissues	46
		TiO2	Metabolites (putrescine, uracil, ornithine)	Mouse brain tissues	47
		Colloidal graphite	Proteolytic digests	Cytochrome c	41
			Cerebroside, metabolite, oligosaccharides	Rat brain tissues, fruits	42, 43
			Flavonoids, cuticular wax	Arabidopsis intact leaf	44
ESI-MS	DESI	Electrosprayed microdroplets	RDX, coniceine	Porcine leather, Conum maculatum seed	52
			Epinephrine, norepinephrine	Porcine adrenal gland	53
			Sulfatides, phosphatidylserine, phosphatidylinositol	Mouse brains	54, 55, 57
			Hyperforin, hypericin	Plant (leaf)	56
	LAESI	Infrared laser ablation	Fexofenadine (antihistamine)	Urine (human)	59
			Lipids	Rat brain tissues	61
			Metabolites	Plant (leaf)	60
				Epidermal cells	62, 63
SIMS		Primary ions	Lipopolysaccharides	Rat brain tissues	66
			Phosphocholine and Adenine	Single cells	69

This paper was supported by Konkuk University in 2017.

The authors have no conflicting interests.