Mass Spectrometry Imaging of Therapeutics from Animal Models to Three-Dimensional Cell Cultures

Abstract

Mass spectrometry imaging (MSI) is a powerful label-free technique for the investigation of the spatial distribution of molecules at complex surfaces and has been widely used in the pharmaceutical sciences to understand the distribution of different drugs and their metabolites in various biological samples, ranging from cell-based models to tissues. Here, we review the current applications of MSI for drug studies in animal models, followed by a discussion of the novel advances of MSI in three-dimensional (3D) cell cultures for accurate, efficient and high-throughput analyses to evaluate therapeutics.

In the process of searching for new medicines to improve the quality of human life, detailed understanding of a drug’s absorption, distribution, metabolism and excretion (ADME) has fundamental significance in determining the efficacy or toxicological properties of the pharmaceutical substance. To understand a drug’s distribution, imaging approaches are needed. Autoradiography¹, positron emission tomography (PET)², magnetic resonance imaging (MRI)³ and fluorescence imaging⁴ are commonly used approaches to examine the spatial distribution of a drug in laboratory animals. These traditional imaging methods require either radioisotopes, magnetic or imaging probes.

In autoradiography methods, a radiolabel is attached to the pharmaceutical of interest prior to dosing. In the experiment, the radiolabel is imaged, not the compound itself. Thus, one intrinsic limitation of this technology is the inability to distinguish the localization of a parent drug and its potential metabolites. There are two types of radiography commonly used to image drugs and their metabolites. Macroautoradiography (MARG) is a histological technique that entails a detailed spatial imaging of radioactivity in tissues at the cellular level (typical spatial resolution 10 µm)⁵. By contrast, whole body autoradiography (WBA), with typical spatial resolutions of 100 µm, has the ability to show the distribution of radiolabeled substances in all organs of an intact animal carcass⁴. Due to its more limited spatial resolution, WBA is often combined with liquid chromatography mass spectrometry (LC-MS)⁶, which is capable of identification and quantification of the compound and metabolites from tissue extracts, to gain specific distribution information.

Both PET and MRI require the use of labels. Radioactive isotope labels are used in the PET technique. For MRI analysis, drugs typically need to contain ¹⁹F or be enriched in ¹³C to be visualized in vivo as the sensitivity depends on the magnetic properties of the monitored nucleus⁷. Though PET and MRI can be employed to perform drug-imaging studies, they suffer from millimeter spatial resolution as well as low specificity and limited molecular information. Other labeled imaging techniques, such as fluorescent microscopy, require a drug molecule to be modified with a chemical tag, which can affect its pharmacological properties, drug uptake, and localization within the body.

MSI is an interesting alternative used in a growing number of ADME studies and pharmaceutical compound investigations. Many excellent MSI reviews describing drug studies have been published in recent years^8–14. MSI possesses nanometer scale spatial resolution. It provides the chemical specificity of mass spectrometry to allow direct mapping of a parent drug or prodrug without any labeling. MSI also enables simultaneous detection of the active and toxic metabolites of a pharmaceutical compound to gain a better understanding the efficacy and toxicity properties of the drug.

Briefly, in MSI experiments, sectioned biological specimens are placed in a mass spectrometer and an ordered array of mass spectra are acquired from a raster of positions across the sample surface. Ion density maps can then be reconstructed for any selected peak of a particular mass from the dataset in which each pixel consists of a mass spectrum or multiple MS/MS transitions. Various mass spectrometry ionization techniques have been developed for MSI in biological samples. Due to space limitations, a few ionization techniques will be discussed, but only matrix-assisted laser desorption/ionization (MALDI) will be described in detail below. The novel application of MALDI-MSI in 3D cell cultures as a platform for drug evaluation will then be introduced.

MALDI and its Applications

MALDI is one of the most widely employed ionization techniques used to visualize the spatial arrangement of drugs and metabolites in tissue^15–19. In a typical MALDI-MSI experiment, a UV-absorbing matrix is carefully applied on the surface of a tissue section to assist desorption and ionization of analytes in the sample. Following irradiation by a laser beam in an array of discrete points, mass spectra are generated for each x, y coordinate and images of interest are generated.

Troendel et al. first used MALDI-MSI for direct profiling of the drug paclitaxel in tissues in 1999²⁰. In 2003, the first MS images of drug distribution in a rat brain were reported by the Caprioli group²¹. Various pharmaceutical compounds such as clozapine²², erlotinib²³, olanzapine^24–25, oxaliplatin²⁶, and terfenadine²⁷ have subsequently been investigated and their metabolites have been imaged in different tissue sections and cells. Figure 1a shows whole-body MSI images generated with a hybrid MALDI-LTQ-Orbitrap-MS to detect the distribution of reserpine and its metabolites at 2 h following a 20 mg/kg oral dose²⁸. The results show that reserpine was found to mostly localize in the stomach and testis of the rat. In addition, the accumulation of the known reserpine metabolites methyl reserpate and reserpic acid in the intestines, muscle, thoracic cavity and liver indicates extensive metabolism of reserpine. The simultaneous imaging of both the drug and the resulting metabolites shows a significant advantage of MALDI-MSI over conventional WBA.

Figure 1.

Mass spectrometry imaging technologies from whole body to subcellular level. (a) Extracted ion images at 5 ppm mass tolerance: [I]optical image of sagittal section of 20 mg/kg reserpine rat 2h postdose; [II]reserpine (m/z 609.281); [III] methyl reserpate (m/z 415.223); [IV] reserpic acid (m/z 401.207). (b) MALDI FTICR MSI images of: [I]different metabolites from mouse liver treated with erlotinib under 60 µm resolution;[II]heme,erlotinib and metabolites in mouse brain tissue (10 µm thickness) with U87 xenograft tumor.Images were acquired at 50 µm resolution for a total of 14000 spectra, and sister sections were H&E stained. (c) Application of MALDI MSI in MCTS: [I] structure of MCTS; [II] localization of irinotecan within a 2 µm section of an irinotecan-treated (20.6 µM, 24h) HCT 116 MCTS obtained by MALDI MSI. (d) Exemplary ¹²C¹⁴N⁻, ³⁴S⁻, ³¹P⁻, and ¹⁹⁴Pt⁻ secondary ion signal intensity distribution maps acquired from semi-thin sections of SW480 cells treated with 25 µM of cisplatin. Regions of interest (ROIs) were manually defined within the cytoplasm, the nucleus, the nucleolus and the chromatin. According to the ¹⁹⁴Pt⁻ and ³⁴S⁻ signals, the drug is accumulated in small cytoplasmic, sulfur-rich aggregates (encircled in red), as well as in the nucleoli. The scale bars are 5 µM.

There have been great advances in the use of MALDI-MSI in recent years to expand its applications, and here we review some methodological developments that can be helpful for successful drug evaluation, including improvements in sample preparation, instrumentation, data acquisition and analysis.

Cryosectioning and Washing

After harvesting biological samples for MALDI-MSI, they are snap frozen and stored at –80 °C until sliced into thin sections ranging from 3 µm to 50 µm with a cryostat¹¹. The thickness of the sample can be a significant factor, which influences the spectrum quality, such as the signal to noise ratio, especially for molecules present at low concentrations. Tissue slices thicker than 20 µm can result in a degraded spectrum, which also affect the quality of imaging results²⁹. Thus thickness of tissue slices needs to be optimized for best detection efficiency and spectral quality. The sections are then thaw-mounted onto a MALDI stainless steel plate, electrically conductive or nonconductive glass slide depending on the sample preparation requirements of the specific MALDI mass spectrometer.

For the evaluation of therapeutics, washing the slides is not advised to avoid delocalization or substantial/total loss of the drug compounds in organic solvents. However, a pH-controlled tissue washing protocol recently developed for MALDI-MSI has been shown to help improve the limits of detection for low-molecular weight compounds, like the drug cimetidine, imipramine and a protease inhibitor compound C, in sections³⁰. This protocol involves washing the slides with a buffer solution. This buffer solution is pH adjusted so that the target molecules have low solubility in order to remove suppressant species from the tissue and improve the signal-to-noise ratio without significant delocalization of the analytes of interest during the cleanup procedure. As a result, the signal-to-noise ratio was increased by a factor of 10. This protocol can be beneficial for researchers in pharmaceutical fields using MALDI-MSI.

Matrix Selection and Application

Before performing MALDI-MSI, the choice and application of matrix is significant to obtain high quality imaging results and should be optimized for each type of experiment. Small molecule (< 700 Da) analysis is sometimes challenging due to the background noise of organic matrices, which can suppress or overlap with the m/z of the analytes. For this reason, different matrices should be tested when investigating a specific small compound in order to select the one that provides the greatest selectivity and sensitivity by promoting ionization of the analytes of interest without increasing the background ions. The most commonly used matrices in drug and metabolite imaging studies are 2,5-dihydroxybenzoic acid (DHB), and α-cyano-4-hydroxycinnamic acid (CHCA), while many new organic matrices have been developed to date including N-(1-naphthyl) ethylenediamine dinitrate³¹,9-AA³², 1,5-diaminonaphthalene (1,5-DAN)³³, hydroxyflavones³⁴, and N-(1-naphthyl) ethylenediamine dihydrochloride (NEDC)³⁵.

Compared to the classical crystalline matrices, ionic liquid matrices (ILM) have shown some advantages in producing more homogenous sample spots and thus higher reproducibility^36–37. ILM are equimolar mixtures of conventional matrices like DHB, or CHCA combined with organic bases like pyridine (Py) or N,N-dimethylethylenediamine (DMED). Nanoparticles have also become attractive new matrix subjects. Silver³⁸, TiO₂^39–40, and gold⁴¹ nanostructured materials have been successfully applied in surface assisted laser desorption/ionization mass spectrometry (SALDI MS). The results show that they simplify sample preparation, improve reproducibility of peak intensities for analytes and eliminate background signals from matrix degradation. All of these properties make them useful matrices for the identification of small molecules. Graphene (G) has emerged as another novel matrix alternative for analysis of low mass compounds⁴². With recent advances in nanotechnology, G based nanomaterials^43–45 have shown great improvements as a substrate for matrix-free LDI MS to detect and image diverse molecules with high sensitivity, with minimal background interference, especially in the low mass range.

Matrix application strategies range from spotting to sublimation methodologies. Direct spotting of matrix on tissue surfaces allows for excellent extraction of molecules, but results in a lower lateral spatial resolution of around 200 µm and poor spot-to spot and shot-to-shot reproducibility. Matrix spraying either using an artistic airbrush or a robotic sprayer such as ImagePrep (Bruker Daltonics Inc.), SunCollect sprayer (SunChrom, Friedrichsdorf), and TM-Sprayer (HTX Technologies Inc.) offers a resolution between 10–20 µm, though wetting of the tissue will delocalize the drug of interest and also cause tissue shrinkage if spraying conditions are not carefully controlled. Sublimation is an alternative solvent-free technique to deposit high purity matrix directly on the sample while avoiding the possibility of analyte diffusion and spreading in the specimen. Sublimation and rehydration generates homogenous small crystals of matrix thus giving high image resolution, but has lower sensitivity for the analysis of lipids and small molecules. The solvent-free matrix dry-coating approach⁴⁶ can be another alternative way to overcome tissue shrinkage with improved matrix crystal distribution and enhanced signal uniformity. However, in this dry-coating method, extraction of analytes is less effective and molecules undergo limited ionization, especially in the case of pharmaceutical compounds. To obtain both homogenous matrix coverage and improve ionization efficiency, the two-step matrix application was developed. This approach employs the airbrush spaying or sublimation in the first step to produce small matrix crystals, followed by the deposition of droplets of matrix solution using automatic dispensers, or spraying to achieve re-crystallization^47–49. As a result, improved crystal morphology and size drastically enhances signal intensity, reproducibility and imaging quality. Therefore, it is also suggested that an appropriate matrix application method needs to be optimized for different biological tissue samples to achieve uniform matrix distribution, reduced crystal size and controlled solvent effects.

The choices of matrix and application method are crucial steps to improve sensitivity and spatial resolution for specific analytes of interest. These advances in matrix types and application approaches greatly assist with the identification and mapping of drugs and metabolites in tissue samples. Careful optimization and planning of sample preparation before performing any large-scale analysis are necessary to generate high-quality MALDI-MSI data.

Spatial Resolution

Spatial (lateral) resolution, or pixel resolution, generally refers to the center-to-center distance between spectra. While sample preparation factors like matrix crystal size or delocalization of analytes during matrix application contribute, the physical limitation of spatial resolution is also determined by the diameter of the laser beam⁸. Several instrumental approaches have been developed to improve the spatial resolution. One method involves the use of a 20-µm pinhole filter to reduce the diameter of the laser beam⁵⁰. The spatial resolution was increased from 20 µm to 5 µm on two MALDI-TOF instruments with this approach. In other examples, a transmission geometry vacuum ion source was developed to provide sub-micron laser spot sizes by irradiating the backside of the sample⁵¹; while a Gaussian beam profile has been designed using optical single mode fibers to give a small laser beam diameter⁵². Oversampling is another method to improve spatial resolution. This approach involves using a raster increment smaller than the width of the laser beam coupled with complete matrix ablation so that only a fraction of the beam is generating the analyte signal. In this method, resolution is not limited to the size of the laser beam.

Spatial resolution is an important parameter to consider in the design of a therapeutic imaging experiment. For a whole body MSI study in rats, a spatial resolution of several hundred microns is sufficient to provide an overview of distribution patterns of drugs and metabolites in major organs or tissues with the advantage of speed (sample area/time) and sensitivity. A resolution between 20–200 µm can be used for MSI of specific tissues or organs, while a higher resolution (< 20 µm) is necessary for imaging applications at the cellular level. Therefore it is necessary to determine the proper spatial resolution needed to answer the question at hand before analyzing a sample. All of the developments in improving the spatial resolution will have great benefits for the MALDI-MSI analyses of therapeutics at the cellular and sub-cellular levels.

Detection and Identification of Drugs and Metabolites

A great challenge for the identification of drugs and metabolites in biological samples is the interference caused by numerous endogenous species as well as the matrix ion signals. Detection of molecules with similar m/z values or overlapping isotopic clusters can lead to ambiguous molecular identification. There are several solutions to eliminate this interference. If the matrix peaks overlap with the m/z of the analytes of interest, the analyte signals can be unmasked by either deuterating the matrix to generate a mass shift⁵³ or substituting the matrix with an alternative substance. In other cases, the use of standards for drugs and major metabolites can aid in the development of a targeted imaging approach to increase analytical sensitivity for specific compounds. The use of imaging MS/MS to generate ion density maps for product ions, which are matched to those of the precursor ion, also increases confidence in the identification process. The rapid development of high resolving power as well as sub-ppm mass accuracy MS instruments such as the Orbitrap⁵² and FTICR^54–56 can also help differentiate peaks of similar mass. In Figure 1b, the high mass accuracy of FTICR acquisition (<1 ppm mass tolerance) gives a confident assignment and spatial distribution of the drug erlotinib, as well as its four metabolites, within the liver of a drug-treated mouse without requiring MS/MS. Recently, the application of ion mobility analyses enable a post-ionization separation step that assists in resolving peaks of isobaric species based on their distinct molecular cross sections. By incorporating this methodology, the drug vinblastine was successfully distinguished from endogenous lipid molecules in whole body rat sections⁵⁷.

Quantification

Quantification requires that the magnitude of the measured signal intensity is proportional to the concentration or amount of the analyte in the sample matrix. The relative quantitative information acquired from MALDI-MSI experiments should be taken with special consideration. Ion suppression, heterogeneity of target surface, matrix crystal size and morphology, ionization efficiency, stability of analytes, and fluctuations in detector response can all affect the results. One way to address such issues is the normalization of the drug signals to the total ion current or endogenous species⁵⁸. However, uneven distribution of the endogenous species can lead to artifacts if the intensities of the pharmaceutical compounds are normalized against them. An alternative approach is to normalize to a homogenously applied standard, which has similar properties and ionization efficiencies as the compound(s) being analyzed⁵⁹. Compound signals that increase relative to that of the internal standard when different samples are compared can be considered to represent increases in the amounts of the analytes, while decreases reflect reduced levels. This approach has been shown to be an effective way to compensate for inherent variability in sample preparation and ionization process.

For absolute quantification, the measured ion intensities cannot be directly compared to an absolute quantity without a reference standard due to the reasons described above. To date, the majority of approaches for drug quantification using MALDI-MSI involve generating a standard curve for the compound by spotting a range of concentrations onto the surface of a control tissue section⁹. By comparing the ion intensities from the dosed sample, the drug concentration is then estimated. This is a quick and simple method, however, there are still limitations in terms of the uniformity of spotting standards for reproducibility and accurate reflection of the extraction efficiency of the drug in treated samples. To overcome this challenge, a novel method using a mimetic tissue model, including a set of tissue homogenates spiked with varying drug concentrations, has been employed⁶⁰. The frozen model was sectioned and analyzed together with the dosed tissues. Differences in analyte extraction and ion suppression were taken into consideration in this methodology. Two drugs, lapatinib and nevirapine, were successfully quantified with this approach and the results were comparable to the LC-MS quantification.

Data Processing, Analysis and Available Software

Data processing, performed to find m/z values that correspond to analytes of interest (e.g. pharmaceutics and their metabolites), serves as a bridge between the raw dataset and data visualization or analysis. The initial step in data analysis requires pre-processing: baseline correction, noise reduction, normalization, spectra smoothing and recalibration, which allows cleaning of the spectra from baseline and background noise, facilitating the selection of peaks of interest⁶¹. The next step involves data representation using multivariate statistical analysis of the processed spectra and spatial segmentation based on clustering^61–62.

Commercially available software for MALDI imaging data acquisition and visualization from vendors includes FlexImaging (Bruker Daltonics), HDI (Waters), ImageQuest (Thermo Scientific), TissueView (AB Sciex), Intensity Mapping (Shimadzu), and WinCadence (Physical Electronics). In addition to vendor-provided software, Biomap (Novartis) and Datacube Explorer V0.7 (AMOLF) can be used for visualization and basic statistical analysis. Other imaging generic tools, written in e.g. ImageJ, or Matlab like OmniSpect, or MSiReader, have also been developed. Recently, an open-source R-based software package is available for processing and visualization of spectra, followed by statistical segmentation such as Principal Component Analysis (PCA), Spatially Aware, and Spatial Shrunken Centroids segmentation methods, and classification with Partial Lease Squares Discriminant Analysis, and Orthogonal Projections to Latent Structures Discriminant Analysis of the resulting images⁶³. As well as the commercial imaging software packages, in-house developments for data analysis are excellent alternatives in the MALDI imaging field. Our lab recently developed an analysis pipeline within the MATLAB programming platform employing a combination of PCA, clustering methods, and linear discriminant analysis to visualize differences in protein expression levels across a 3D tumor spheroid section⁶⁴. Due to the flexibility of the popular computing environment like MATLAB or R, such in-house developed analysis methods can be easily utilized and adapted by other researchers.

Overall, developments in sample preparation protocols, instrumentation and data analysis have vastly improved the speed, spatial resolution and sensitivity of MALDI-MSI, which is rapidly maturing as a routine bioanalytical technique with great potential for pharmaceutical discovery and development investigations.

Other Major Ionization Techniques

To avoid matrix interference and simplify sample preparation procedures, matrix-free techniques have also been developed and applied in therapeutic MSI studies. Secondary ion mass spectrometry (SIMS) was one of the first matrix-free desorption/ionization methodologies for the analysis of metabolites⁶⁵. SIMS uses a beam of high-energy primary ions to bombard and eject secondary ions from the sample surface, which are then guided into the mass spectrometer for analysis. SIMS MSI can be divided into two categories, static and dynamic SIMS. Static SIMS has a low primary-ion count resulting in about 1% of the uppermost surface monolayer being sampled, while in the dynamic mode, high primary ion current densities are used to sample all of the upper monolayers of the surface⁶⁶. SIMS typically desorbs and ionizes small molecules and has high spatial resolution (down to 50 nm) to reach subcellular levels for imaging⁶⁷. SIMS MSI instruments can also perform depth-profiling experiments with depth resolution greater than 50 nm by an erosive process of sputtering with the primary-ion beam to determine the molecular localizations beneath the sample surface, and consequently enabling 3D imaging. Although SIMS possesses high spatial resolution, it suffers from limited sensitivity at >500 Da and extensive surface fragmentation generated by high-energy primary ions⁶⁸ making it unpopular for tissue samples. However, used in combination with other techniques, such as MALDI-MSI¹, scanning electron microscopy⁶⁹, atomic force⁷⁰, or fluorescence microscopy⁷¹, SIMS has great potential to expand its application in therapeutic studies. Recently, an advanced type of dynamic SIMS, nano-scale SIMS combined with confocal laser-scanning microscopy, was designed for trace element and isotope analysis⁷². It was successfully applied for subcellular mapping of metal-based compounds like Platinum (Pt)-complexes in cell culture samples exposed to cisplatin (Figure 1d)⁷².

Siuzdak et al. first demonstrated nanostructure-initiator mass spectrometry (NIMS) in 2007⁷³. NIMS is another matrix-free ionization technique with simple sample preparation using a liquid initiator to facilitate desorption⁷³. During ionization, the nanostructured surface absorbs the laser energy, resulting in rapid surface heating, thus causing vaporization of the trapped initiator and triggering desorption of the absorbed analytes without fragmentation⁷⁴. NIMS was shown to be sensitive for the localization of clozapine and its metabolite N-desmethylclozapine from rat brain sections⁷⁵. More recently, it was used to detect intracellular uptake of 3’-deoxy-3’-fluorothymidine (FLT) and rapamycin, drug metabolism (FLT conversion to FLT-monophosphate), and downstream pharmacodynamic responses at the level of single lymphoma cells⁷⁶. This relatively new matrix-free approach provides soft ionization, molecular specificity, high sensitivity, and high cell-level spatial resolution, which are important criteria in pharmaceutical studies and in optimization of existing therapies.

Several ambient ionization techniques have also been introduced and utilized in MSI studies of drugs and metabolites. Desorption electrospray ionization (DESI) is one such technique. It is performed by the interaction of charged droplets with a surface, producing gaseous ions that are directed into the mass spectrometer by use of an ion transfer line⁷⁷. This method has been used on several occasions for the imaging of drugs like clozapine^78–79 and propranolol⁸⁰ in rat tissue sections and whole mouse bodies. However, the poor spatial resolution of ~250 µm for DESI limits its application. Recent improvements in sprayer flow rate, sprayer surface angle and sprayer diameter⁸¹ have enhanced the resolution to less than 50 µm. Currently, development of nano-DESI has achieved a spatial resolution of 20 µm while analyzing small molecules across rat brain tissue⁸².

Liquid Extraction Surface Analysis (LESA) is also an ambient solvent-based ionization method. In LESA, a droplet of solvent is placed on the sample at a specific position to extract analytes, followed by electrospray ionization. Currently, LESA lacks high spatial resolution. The spatial resolution is limited by the size of the pipette nozzle to deposit and aspirate the extraction solvent and typically 1 mm diameter is sampled⁸³. However, it does not require additional sample preparation and has high sensitivity when used in selected reaction monitoring mode. It has been used as a rapid complementary technique to MALDI-MSI when detailed spatial distribution is not necessary. LESA has been applied to map the distribution of drugs like terfenadine⁸³, moxifloxacin⁸⁴, chloroquine and its metabolites⁸⁵ in sections from whole body, organs or tissues of rat, and the data was validated against WBA, DESI, or MALDI-MSI approaches. In other studies, LESA MSI also enabled semi-quantification of drugs and their metabolites, and the results correlated with quantitative WBA data^86–87.

Similar to DESI, laser ablation ESI (LAESI) is another ambient ionization technique that is less destructive to the sample surfaces under investigation⁸⁸. LAESI employs a mid-IR laser beam at 2940 nm. The IR laser beam excites the hydroxyl vibrations of water molecules and ablates the surface in concert with electrospray ionization⁸⁸. LAESI was first introduced by Nemes in 2007⁸⁸, and used in biomedical imaging in 2010⁸⁹. So far it has been used to map lipids and metabolites in animal organs^89–90, and this approach has great potential to be applied in pharmaceutical imaging studies.

Laser ablation inductively coupled mass spectrometry (LA-ICP-MS) has been developed for the evaluation of metal containing compounds. Using LA-ICP-MS, it is possible to visualize the distribution of metals deriving from metallo-drugs, such as the Br-containing drug TMC207⁹¹, gold compounds⁹², and platinum drugs like oxaliplatin⁹³, cisplatin^94–96 and carboplatin⁹⁷ mouse tissue. This method achieves spatial resolution at sub-cellular levels⁹⁶ and also enables quantification of metal complexes⁹⁷. One disadvantage of this technique is the limited detection capability; the approach is specific for compounds containing metals. Combining it with other methodologies can expand its application in drug discovery and development⁹⁸. Advantages and limitations of the various ionization approaches are summarized in Table 1.

Table 1.

Overview of various ionization techniques for imaging mass spectrometry analysis.

Technique		Mass Range (Da)	Lateral Resolution (µm)	Advantages	Limitations	Commercialized by
MALDI		0-70,000	30–500	Soft ionization Wide mass range Atmospheric pressure and vacuum sources available High tolerance of sample (size, shape, salts and detergents) Versatile method	Molecular coverage, and imaging quality depend on matrix Quantitative analysis is challenging Interference with matrix-related peaks below m/z=300	Bruker Daltonics Shimadzu Thermo Fischer Scientific Waters Applied Biosystems AB Sciex
SIMS	Static Mode	0-2000	0.5–50	High spatial resolution Wealth of molecular information Low consumption of material Various primary-ion guns available	Source-induced fragmentation Small sampling area Low signal levels Very high surface sensitivity Samples must be compatible with an ultra high vacuum	Physical Electronics IonTOF
	Dynamic Mode	0-300	0.04–0.5	Very high spatial resolution (cellular and subcellular level) High signal intensity for ppb-level elemental analysis Quantitative imaging possible Depth profiling possible	Limited molecular information Risk of tissue damage Small sampling area Very high surface sensitivity Samples must be compatible with ultra high vacuum Few ions can be measured simultaneously	Cameca Physical Electronics Millbrook Instruments
NIMS		0-1500	15–20	Soft ionization No sample preparation Low chemical noise and high sensitivity in the low mass range below 1000 m/z Flexibility of different initiators	Risk of tissue slice damage during imaging process Surface modifications/additives are needed to improve ionization efficiencies	Not yet commercialized
DESI		0-5000	100–1000	Soft ionization No sample preparation Ambient ionization technique	Low spatial resolution Molecular coverage depends on spray composition Risk of destroying samples with high-pressure gas	Prosolia
LESA		0-2000	~1000	No sample preparation High sensitivity of detection due to efficiency of surface extraction method	Low spatial resolution limits the investigation of smaller regions within biological structure of interest Surface needs to be sufficiently hydrophobic to form a micro liquid junction between tip and sample	Advion
LAESI		0-66000	200–400	Minimal sample preparation Depth profiling possible	Low spatial resolution Water rich target is needed for ablation process Variations of water content and tensile strength can affect ion yield and imaging results	Protea Biosciences
LA-ICP		6–250	5–200	Simultaneous determination of all trace, minor, major elements and isotopes in tissue Depth profiling possible Quantitative imaging possible Excellent for metal analysis Low danger of contamination	Invasive technique Isobaric or multiply charged elements at same nominal mass can interfere with atomic ions of analytes Reference materials with similar matrix composition required for quantification	Waters Thermo Fischer Scientific Applied Spectra Agilent Technologies

A Novel Application: 3D Cell Culture Systems

Cell cultures have been proven to be indispensible during the drug development process for pre-animal and pre-clinical evaluation of promising therapeutics. For decades, most studies testing drug candidates were performed using the cells of interest cultured on compatible polystyrene in a two-dimensional (2D) monolayer format. 2D cell monolayers provide a well-controlled and homogeneous cellular environment, simplify the analysis techniques and sustain the growth and proliferation for most cell types⁹⁹. However, the results from this in vitro system have been questioned for not accurately recapitulating in vivo scenarios due to differences in cell morphology, proliferation, gene expression patterns and cell-cell or cell-matrix interactions^99–103. In addition, from a drug delivery perspective, 2D cell cultures also lack the appropriate physiological barriers arranged in a proper geometry to mimic the microenvironment in the body⁹⁹.

Alternatively, attention has shifted to 3D cell cultures, using scaffolds, hydrogel and microfabricated devices to provide enhanced models to more closely replicate the complex in vivo conditions for testing of drug delivery, metabolism and toxicity⁹⁹. Among the different 3D cell culture systems, multicellular tumor spheroids (MCTS), which are spherical self-assembled cell clusters, are one of the most commonly used models that closely mimic human solid tumors¹⁰⁴. Figure 1c (I) shows a typical structure of a MCTS with a radius greater than 200 µm. Because of oxygen and nutrient diffusion limits, all cells and metabolic waste accumulations within the tumor mimics. The large MCTS harbor concentrically layered zones of highly proliferative cells on the outer rim, quiescent cells on the inside and necrotic cells at the core region. This heterogeneous system reflects the in vivo microenvironment of proliferating tumor cells next to capillaries, while more distant cells stay quiescent or die through apoptosis.

Similar to in vivo tissues, MCTS show greater resistance than single cells or monolayer cultures to a given dose of many pharmaceuticals^105–109. This phenomena was first demonstrated by Durand and Sutherland in 1972¹⁰⁹ and later termed “multicellular resistance (MCR)”¹¹⁰ and replicates a significant obstacle to successful cancer therapy. Factors causing MCR include: intercellular molecular changes like activation of Stat3 and NF-κB signaling pathways, microenvironment alterations like changes in extracellular matrix (ECM) and varying pH, the presence of hypoxic cells and differences in cell cycle distribution within MCTS, and importantly, drug penetration within spheroids, which is the rate-limiting step for drug delivery^{104,111–112}. In addition to structural and physical properties, poor penetration of drug molecules into spheroids can be caused by binding with the outer layers of cells or physical barriers due to the ECM and cellular membranes^{108,113–114}. Additionally, pH gradients within the 3D structure sometimes protect cells in the acidic core from weakly acidic compounds such as mitoxantrones, while enhancing the effect of weakly basic drugs such as chlorambucil¹¹⁵. Due to the inherent pathophysiological gradients, the heterogeneous structure and transport limitations, MCTS are an accurate model to evaluate therapeutics, especially drug penetration limitations and metabolism properties. Nevertheless, methodologies for imaging drug distribution in spheroids were limited to approaches that required labels like fluorescent microscopy^116–117 or autoradiography¹¹⁸. More convenient and powerful methods for direct localization of drugs and their metabolites are still in great need.

In early 2011, a MALDI-MSI experiment on a 3D reconstructed skin tissue model was established to study the absorption of imipramine into skin¹¹⁹. Months later, the Hummon laboratory was the first to adapt this complementary label-free MALDI-MSI in 3D MCTS¹²⁰. This methodology was originally performed in colon adenocarcinoma HCT 116 spheroids for mapping the distribution of proteins¹²⁰, and then further applied to study the penetration of the anticancer drug irinotecan [Figure 1c (II)]¹²¹. A typical MALDI-MSI workflow for therapeutic evaluation in MCTS is shown in Figure 2. Spheroids are grown to ~ 1mm in diameter on an agarose-coated 96-well plate. Around day 13, MCTS are large enough to exhibit chemical gradients and cellular heterogeneity, and thus are ready to be dosed with pharmaceutical compounds. After a specific length of treatment time, spheroids are gently transferred with a seratological pipet and embedded in gelatin. The frozen gelatin block containing the MCTS is then sectioned at thicknesses between 10 – 14 µm on a cryostat and thaw-mounted to ITO coated glass slides. After application of an appropriate matrix, MALDI-MSI is then performed for the spatial localization of drugs and metabolites. For more details and a video of this protocol, please consult Ahlf-Wheatcraft et al¹²².

Figure 2.

General scheme of MALDI MSI technology in 3D cell cultures for evaluation of therapeutics. Cells are seeded in agarose-coated 96-well plate for MCTS formation. When MCTS grow to ~ 1mm in diameter, which is large enough for establishing cellular heterogeneity and chemical gradients to mimic in vivo tumors, they are ready to be treated with pharmaceutical compounds. After a specific time length, MCTS are gently collected and embedded in gelatin for sectioning, followed by MALDI-MSI analysis. This workflow enables simultaneous localization of drugs and metabolites in the MCTS sample without needing any pre-labeling.

With this methodology, the distribution of irinotecan (m/z 587) has been imaged in HCT 116 MCTS after 20.6-µM treatment for specified lengths of time. Irinotecan accumulation in the periphery and the center of spheroids increases with time as shown by ion density maps in Figure 3. In the first 6 h of incubation, irinotecan was present only in the periphery, followed by an enhanced drug penetration into the central part of the spheroids within 12 h. After 24 h of treatment, irinotecan was distributed across the entire structure of the MCTS, with areas of drug accumulation evident at 48 h of treatment. The identities of the irinotecan metabolites were examined by nanoflow liquid chromatography-tandem mass spectrometry (nLC-MS/MS). By analyzing the small molecules extracted from the drug-treated spheroids, ten metabolites in total were identified. MALDI-MSI provides sensitive image analysis of not only the parent drug, but also the metabolites simultaneously. Three metabolites, including the bioactive metabolite SN-38, were also mapped in treated spheroids after 72 h of drug incubation.

Figure 3.

Time-dependent penetration of irinotecan (m/z 587) in HCT 116 MCTS analyzed by MALDI-IMS. MCTS were treated with 20.6 µM irinotecan for 2, 6, 12, 24, 48, 72 hours (from left to right). For every treatment duration, color gradient intensity maps were generated from 7 consecutive 12 µm slices from a single MCTS in 120 µm vertical intervals.

This approach is valuable in assessing delivery of irinotecan and other various pharmaceutical entities in MCTS. Leucovorin or folinic acid is converted in vivo into the biologically active metabolite folate and enhances the effect of 5-fluorouracil (5-FU) in colorectal cancer. 5-FU is metabolized to the active metabolite 5-fluoro-2′-deoxyuridine 5′-monophosphate (FdUMP), binding to and inhibiting the enzyme thymidylate synthase (TS)¹²³. Folinic acid is rapidly metabolized to 5, 10-methylenetetrahydrofolate (5,10-CH₂-THF), 5, 10-methenyltetrahydrofolate (5, 10-CH=THF), and then to 5-methyltetrahydrofolate (5-MTHF)^123–124. Leucovorin has no intrinsic antitumor effect, but its metabolite 5,10-CH₂-THF forms an inhibitory ternary complex with FdUMP and TS, which enhances the inhibition of thymidylate synthesis and impairment of both DNA synthesis and repair processes¹²⁴. Figure 4 shows the penetration of folinic acid into MCTS after incubation with the compound at the concentration of 35 µM. Folinic acid (m/z 474) reaches and accumulates around the core region after 24h treatment. The metabolite 5, 10-CH=THF (m/z 474) is also detected and found to be concentrated at the center of MCTS. It has been shown that folinic acid can be avidly converted into 5, 10-CH=THF at acidic pH and this metabolite has great stability in low pH conditions^125–126. Thus the pH gradient of the MCTS could be a reason for this accumulation. The structures of both the folinic acid and 5, 10-CH=THF have been confirmed by MALDI MS/MS.

Figure 4.

Distributions of folinic acid and one metabolites within a 12 µm section of an leucovorin-treated (35 µM, 24h) HCT 116 MCTS obtained by MALDI-imaging. (a) Optical image of the MCTS section. Spatial localizations of (b) DHB (m/z 155, blue); (c) 5, 10-CH=THF metabolite (m/z 456, green); (d) folinic acid (m/z 474, red). (e) Merged image shows high relative abundance of folinic acid. (f) Mass-to-charge plot.

MALDI-MSI is also compatible with other techniques like confocal Raman microscopy (CRM)¹²⁷. This correlated imaging approach was developed to evaluate the structural and chemical diversities of various molecules within the MCTS. CRM is a non-destructive imaging approach and enables subcellular visualization of classes of biomolecules, while MSI allows the identification and localization of small molecules from discrete regions of a spheroid section. Thus, different chemical data can be obtained and correlated to further investigate cellular responses to chemotherapeutical treatment as well as studying drug penetration and metabolism.

Recently, nanoparticle-assisted LA-ICP-MS analysis of drug delivery in 3D cell cultures was performed¹²⁸. Figure 5 shows the images acquired from the TFK-1 MCTS incubated for 48 h with 8.2 µg/mL [(a) + (b)] and 40.7 µg/mL [(c) + (d)] 5,10,15,20-tetrakis(3-hydroxy-phenyl)porphyrin (mTHPP) loaded poly(lactic-co-glycolic acid) (PLGA) nanoparticles. mTHPP is a second-generation photosensitizer and was tagged by complexing with palladium before analysis. The results show that a distinct enrichment of the drug occurs predominantly within the outer layers of the spheroids, which could be attributed to the very high affinity of the low polarity compounds to proteins. This work proves the LA-ICP-MS as another promising approach to study penetration of metal containing drugs into MCTS.

Figure 5.

Images of TFK-1 tumor MCTS sections incubated with mTHPP-PdPLGA nanoparticles for 48h [(a) + (b)] 8.2 µg/mL and [(c) + (d)] 40.7 µg/mL. The scale represents a length of 200 µm in all images. The ¹⁰⁵Pt signal is used for data evaluation. Intensity is stated in counts per second (cps). Images were obtained using a laser spot diameter of 10 µm.

Together, these combinations of powerful label-free MSI techniques with 3D MCTS have great possibility to improve the predictive power for in vivo therapeutic efficacy. Not only can drug penetration be evaluated, but the drug metabolism and distribution of various analytes can also be studied simultaneously. These methodologies provide significant benefits for MCTS to become an essential screening tool for the preclinical assessment of a variety of therapeutics.

Conclusions

Recent developments of MSI in the arena of drug and metabolite localization have revealed great potential for pharmaceutical discovery processes. MALDI-MSI remains a popular approach due to its high sensitivity and simplicity of operation. Advances from sample preparation to instrumentation have vastly improved the speed, selectivity, spatial resolution and identification capabilities for MSI by MALDI. Other well-developed ionization methods like SIMS, NIMS, DESI, LESA, LAESI, and LA-ICP ensure a variety of options for mapping drugs and metabolites in various tissue or organ samples with different spatial resolutions.

Testing of pharmaceutical compounds in 3D cell cultures is increasingly regarded as an essential step during drug development. Application of MALDI-MSI and LA-ICP techniques in MCTS has shown great advantages for directly and rapidly imaging the distribution of both parent drug and multiple metabolites in a single experiment. With further optimization, it is expected that other ionization methods will have greater roles to play for drug evaluation in 3D cell cultures. In addition, co-localization of pharmaceuticals with proteins, metabolites or lipids in MCTS using MSI could be another possibility to acquire more useful information of therapeutic response directly at the action site of a drug. Moreover, future work should be directed towards combinatorial therapy studies in this model system.

By adapting MSI technologies, MCTS will become a more powerful and informative platform to screen and select drugs in a format that closely mimics the microenvironment conditions in patients. With continued developments in MSI, this system also has great potential to be a common and routine approach in the drug discovery and development workflow.