Advances in High-Resolution MALDI Mass Spectrometry for Neurobiology

Abstract

Research in the field of neurobiology and neurochemistry has seen a rapid expansion in the last several years due to advances in technologies and instrumentation, facilitating the detection of biomolecules critical to the complex signaling of neurons. Part of this growth has been due to the development and implementation of high resolution Fourier transform (FT) mass spectrometry (MS), as is offered by FT ion cyclotron resonance (FTICR) and Orbitrap mass analyzers, which improves accuracy of measurements and helps resolve the complex biological mixtures often analyzed in the nervous system. The coupling of matrix-assisted laser desorption/ionization (MALDI) with high-resolution MS has drastically expanded the information that can be obtained with these complex samples. This review discusses notable technical developments in MALDI-FTICR and MALDI-Orbitrap platforms and their applications toward molecules in the nervous system, including sequence elucidation and profiling with de novo sequencing, analysis of post-translational modifications, in situ analysis, key advances in sample preparation and handling, quantitation, and imaging. Notable novel applications are also discussed to highlight key developments critical to advancing our understanding of neurobiology and providing insight into the exciting future of this field.

Introduction

Neuroscience is a complex field with many different facets that come together to study and understand the nervous system. Characterizing the composition and biological interaction in the nervous system has long been pursued to understand its function and structure, but no current tool can match the speed and structural elucidation capabilities of mass spectrometry (MS). With millions of possibilities for molecule composition, it is crucial to be able to know with high confidence the exact identity of a molecule to even begin to study how it interacts to alter functions in the nervous system. MS is a powerful analytical technique to help understand this multidisciplinary field, enabling the characterization of various biomolecules, from lipids and metabolites to large proteins. The use of MS allows for the elucidation of many structural properties to identify and understand complex proteins, such as amino acid sequence, state of folding, and post-translational modifications (PTMs).

Since the introduction of matrix-assisted laser desorption/ionization (MALDI) MS in 1985 (Karas, Bachmann & Hillenkamp, 1985), this technique has rapidly grown to be the second most widely used ionization method behind electrospray ionization (ESI). As a soft ionization method, MALDI enables the MS detection of intact biomolecules by maintaining covalent bonds during ionization (Karas, Bachmann & Hillenkamp, 1985; Karas & Kruger, 2003). A matrix compound facilitates ionization by absorbing the energy of a laser, resulting in the desorption and ionization of analytes from a surface. In the field of biological sciences, MALDI-MS is regarded as a highly versatile tool able to analyze a range of samples including biomolecules (Caprioli, Farmer & Gile, 1997; Chen et al., 2009; DeKeyser et al., 2007; Ham, Jacob & Cole, 2005), biopolymers (Bahr, Karas & Hillenkamp, 1994; Caprioli, Farmer & Gile, 1997; Fagerquist et al., 2010; Li et al., 1999), and single cells (Garden et al., 1998; Li et al., 1999; Li, Garden & Sweedler, 2000; Li et al., 2000) either extracted or directly profiled from the biological matrix.

As MALDI is a solid-state ionization technique, it preserves the localization of molecules and enables the high spatial resolution MS imaging (MSI), where an intact tissue section can simply be coated with a thin layer of matrix prior to MS analysis. In an MSI experiment, the MALDI laser beam systematically ablates regions of a sample at a pre-defined pixel size, retaining the spatial coordinates of spectra acquired at each pixel. The pixel size, or spatial resolution, is limited by the instrument’s laser diameter and the size of the matrix crystals, as well as by other instrumental considerations related to the stage mechanics. Generally, a spatial resolution of 20 or 30 μm is achievable on commercial instruments (Ye, Greer & Li, 2012; Zavalin et al., 2015). However, newer commercial instruments are approaching higher spatial resolutions, such as the timsTOF flex, which can provide 10 μm spatial resolution (Spraggins et al., 2019). Custom instruments can provide even higher spatial resolution with smaller pixel size (Kompauer, Heiles & Spengler, 2017a; Niehaus et al., 2019). Images can then be created for specific mass-to-charge (m/z) values extracted from each acquired spectrum, enabling the visualization of analyte abundance at each location in the tissue through the generation of a heat-map image. The ability for in situ imaging by MALDI-MS enables the localization of thousands of biomolecules to specific tissue regions, useful for determining the function of key molecules within a tissue. Specifically, in the field of neurobiology, this means the localization of thousands of potential biomarkers to specific regions in the brain to determine neurological relevance. MSI has been beneficial in the field of neurobiology, allowing for the visualization of drug penetration (Ntshangase et al., 2019; Quiason & Shahidi-Latham, 2015; Tang et al., 2019; Vallianatou et al., 2018) and disease biomarker distribution (Hulme et al., 2020; Nielsen et al., 2016; Paine et al., 2019; Ye et al., 2014), among other applications, in brain tissues.

A notable limitation to MALDI compared to ESI is its propensity to generate primarily singly charged ions. The lack of multiply-charged ions makes tandem MS (MS/MS) fragmentation relatively inefficient compared to ESI-MS/MS fragmentation, requiring many compounds to be identified through accurate mass matching – matching the observed intact mass to a database of known compounds (Buchberger et al., 2018). This reliance on a database is particularly challenging for neuropeptides with a large number of PTMs, as this increases the complexity of the database, leading to the risk of an inflated false discovery rate. MALDI has most commonly been coupled with time-of-flight (TOF) instruments for the analysis of biological compounds. However, the limited mass resolution and accuracy offered by MALDI-TOF-MS analysis makes it difficult to confidently assign an identity to molecules, as an exact mass is difficult to obtain. This limitation has been overcome with the coupling of high performance mass analyzers, most notably the Fourier transform ion cyclotron resonance (FTICR) and Orbitrap, due to their increased mass accuracy, resolving power, and capability for performing improved MS/MS experiments compared to TOF mass analyzers, although MS/MS spectra are still difficult to acquire due to reasons that are discussed later in this review. These high resolution accurate mass (HRAM) MS platforms have been beneficial in the analysis of biomolecules, ensuring higher confidence assignments through accurate mass matching and de novo sequencing. This increased identification confidence enables isoforms and modified molecules to be distinguished beyond the capabilities of a TOF analyzer (Dilillo et al., 2017a; Ye et al., 2013b). To directly compare the spatial and accurate mass capabilities of MALDI in conjunction with different mass analyzers, Rompp et al. performed experiments on different MALDI instrument platforms, highlighting the complementarity of different mass analyzers and displaying improved mass measurement accuracy during the imaging analysis of mouse brain tissue with the FTICR and Orbitrap platforms compared to TOF (Rompp et al., 2015a).

Developments in HRAM MS have increased our ability to identify and localize biomolecules that have altered function and effects on neurological systems. With MALDI-FTMS, isoforms can be distinguished, and high spatial resolution enables the localization of proteins and peptides differentially expressed and/or modified to specific regions in the brain to help understand roles in neurological disease and function. This review will discuss the utility of MALDI high resolution mass spectrometry (HRMS) in advancing our understanding of neurobiology and the technological developments that make it possible. The intent is to highlight recent advances of the MALDI-FTMS platform in the analysis of neurologically relevant biomolecules in all areas of the workflow, including sample preparation, molecular characterization, quantitation, ionization, and data analysis, as depicted in Figure 1. We also comment on notable applications that have been employed that expand our current understanding of complexities in the field of neuroscience. While this is not a comprehensive review of all recent developments, emphasis is placed on demonstrating the direction the field is progressing toward, specifically in the field of neurobiology.

Figure 1.

Depiction of the typical workflow for high-resolution MALDI-MS analysis, including sample preparation for both extraction and imaging experiments, ionization, MS acquisition, and analysis with both structural information and image construction.

Sample Preparation

Appropriate sample preparation steps are important to any MS experiment, but this is particularly true with MALDI-MS experiments as the quality and reproducibility of mass spectra depend on matrix crystallization. Furthermore, separation steps are more difficult to couple with MALDI, exacerbating the need for proper sample preparation to decrease ion suppression and spectral complexity. These considerations are especially crucial for MSI, as even subtle differences throughout a tissue can lead to differences in analyte intensity, total ion abundance, and matrix crystallization, all of which can lead to inaccurate results regarding the presence or localization of analytes of interest. Furthermore, as the nature of imaging limits the range of sample preparations steps that can be performed, much effort has been directed toward furthering techniques to improve sensitivity and reproducibility. Some interesting developments in this area specifically related to high-resolution MALDI-MS and MALDI-MSI are summarized in this section.

Improving Ionization

While the capabilities of a HRAM instrument allow for confident mass measurement and sequencing, it can only detect what is present; therefore, optimizing extraction procedures, when used, is very important prior to analysis. As such, enhanced dissection and extraction techniques for rat neuropeptides were investigated to decrease post-mortem protein degradation and increase neuropeptidome coverage (Dowell, Heyden & Li, 2006). When studying the neuropeptidome of the Jonah crab Cancer borealis, the Li lab created an optimized microextraction method via acidified methanol. They attributed an increase in detected peptides to more efficient peptide extraction/ionization, ruling out protein cleavage from the acetic acid addition (Kutz, Schmidt & Li, 2004). In some cases, erroneous results may arise from MS analysis due to sample preparation, as was considered in the previous study. While investigating the orcokinin neuropeptide family in the lobster Homarus americanus, a modification of endogenous neuropeptides occurring during the extraction procedure was seen (Stemmler et al., 2013). This highlights the importance of considering the potential chemical and biochemical reactions introduced during the extraction procedure. Additionally, a nanosecond photochemical reaction-based click chemistry was recently developed for removal of matrix interference upon ionization, which enhanced neuropeptide identification by serving as a simplified cleanup step (Li et al., 2019c). Demonstrating the importance of proper sample preparation, Ly et al. tested different approaches to create an optimized MSI protocol, from dissection to analysis, to determine neuropeptide distributions in the retrocerebral complex of the American cockroach, Periplaneta Americana (Ly et al., 2019).

A key challenge that is often exacerbated with imaging is the ability to ionize analytes for detection. This is particularly challenging for neurotransmitters, metabolites, and other small molecules, which are diverse classes of molecules often containing species that do not ionize easily, or yield ions by matrix peaks. Several derivatization strategies have been developed to better detect a broader range of these molecules. A subset of these derivation methods have been systematically evaluated for application to neurotransmitters in brain tissue and optimized methods were developed for maximizing sensitivity and minimizing background interference for these low-abundance molecules (Esteve et al., 2016). It was also found that by combining derivatization of primary amines with pyrylium salt and derivatization-free methods, signal complexity was reduced and complementary ionization of neurotransmitters was obtained (Cao et al., 2019). Another research group combined multiple chemical derivatizations, including coniferyl aldehyde for primary amines, Girard’s reagent T for carbonyl groups, and 2-picolylamine for carboxylic acids, on consecutive tissue sections to increase coverage of metabolites (Duenas, Larson & Lee, 2019). These results indicate that utilizing a multifaceted approach improves coverage of these diverse molecules, as it was found that a vast majority of metabolites possess at least one of these functional groups and most of the others contain an alcohol group, which is already easily ionized without derivatization. Additionally, fluoromethylpyridinium-based reactive matrices have been implemented with notable success for improving detection limits of metabolites containing phenolic hydroxyl or primary or secondary amine groups (Shariatgorji et al., 2019). The matrices were able to facilitate charge-tagging of these types of neurotransmitters, which led to the identification of numerous neurotransmitters in Parkinsonian brain tissue.

Targeted methods have also been explored using immunoaffinity-based methods to enrich for specific molecule classes. For example, enrichment for FMRFamide-related peptides in crustaceans prior to analysis by MALDI-FTMS resulted in increased ion signal intensity of peptides from this family relative to other neuropeptides (Ma et al., 2009b). On-tissue digestion of proteins has also been pursued with the MALDI-FTMS platforms. Rat brain tissue slices were imaged on MALDI target plates pre-coated with trypsin and α-cyano-4-hydroxycinnamic acid (CHCA) for direct enzymatic digestion to determine the spatial distribution of proteins, tryptic peptides, and lipids with increased throughput of analysis (Zubair et al., 2016). Schober and colleagues also performed direct on-tissue digestion of proteins on mouse brain tissue for AP-MALDI-FTMS imaging by depositing trypsin with a pneumatic sprayer (Schober et al., 2012). To increase reproducibility of the results from on-tissue digestion of mouse brain, systematic optimization of digestion incubation times was performed in order to achieve complete, uniform on-tissue digestion, which led to a decrease in morphology-induced measurement bias (Heijs et al., 2015b).

Decreasing Spectral Complexity

Many samples analyzed by MALDI cannot be subjected to a separation technique like liquid chromatography. Consequently, many sample preparation techniques have been investigated to decrease spectral complexity and increase analyte ionization efficiency. One direction has been the reduction of observed ion adduct peaks in spectra, as these adducts add spectral complexity and decrease peptide signal intensity due to the distribution of analyte signal amongst multiple peaks. In-trap MALDI-FTICR-MS cleanup procedures have been developed to decrease matrix adducts from on-tissue analysis and to increase sensitivity and simplify the spectral profile for better detection of high molecular mass neuropeptides (Ma et al., 2009a). While traditionally the formation of adduct ions are unfavorable, in certain cases, they can provide valuable information. Stemmler et al. observed using MALDI-FTMS that red pigment-concentrating hormone (RPCH) from the sinus gland of crustaceans only appears as sodium adducts. Therefore, they pursued approaches to identify and increase detection of cationized species by increasing the abundance of metal ions to unmask sodiated peaks and increase signal response (Stemmler et al., 2006). This enhanced cationization was achieved by salt-doping – applying an acidified, aqueous solution of a neutral salt – to a previously crystallized dried droplet tissue sample and was successful in improving RPCH detection. Additionally, there have been steps made to incorporate separation with MALDI using MS imaging, such as was done with capillary electrophoresis (CE) to detect neuropeptides with MALDI (DeLaney & Li, 2020). Instead of determining spatial localization within a tissue, MS imaging was used to retain the separation resolution in an offline CE experiment by localizing peak migration profiles of a continuous trace, negating the need for fraction collection while improving separation resolution.

Direct tissue MS is challenging because chromatography is precluded and tissue cleanup is difficult, as localization on tissue needs to be maintained. As a result, an analyte signal is often masked by interfering agents. One way to improve the signal of these analytes is by washing tissues with specific solvents prior to analysis. Common washing solvents include ethanol, chloroform, xylene, and isopropanol, among other organic solvents, though ethanol has been particularly useful for neuropeptide analysis (Buchberger et al., 2020; Ly et al., 2019) and ammonium formate for removing salt adducts (Angel et al., 2012; Erich et al., 2019; Harris et al., 2020). However, there have been other interesting methods, such as using supercritical fluid CO₂ to remove lipids (Matsushita et al., 2017) and simply using xylene followed by ethanol for removal of paraffin from formalin-fixed, paraffin-embedded tissue (Paine et al., 2018). Multiple washing steps can also be implemented to remove more interfering artifacts, but these may also wash away some of the analyte(s) of interest (Hermann et al., 2020).

Matrix Application

The last step in any MALDI-MS experiment prior to MS analysis is the application of matrix. The most common matrices used are 2,5-dihydroxy benzoic acid (DHB) and CHCA for positive mode, 9-aminoacridine (9AA) for negative mode, and 1,5-diaminonaphthalene (DAN) for both positive and negative mode. The effectiveness of these matrices have been evaluated recently for several molecule classes (Perry et al., 2020; Romero-Perez, Takei & Yao, 2014). However, a variety of new matrices have been evaluated for use with MALDI-FTMS to improve ionization efficiency, observed signal intensity, and reproducibility to fully take advantage of the high spatial and mass measurement resolution. This includes a photoreactive matrix for double-bond localization of phospholipids in mouse brain (Waldchen, Spengler & Heiles, 2019), N-butyl-4-hydroxy-1,8-naphthalimide for the analysis of small molecules and peptides in situ in rat brain slices (Cheng et al., 2017). After considering multiple candidate matrix compounds (Figure 2), a dual-polarity matrix for metabolites, nucleotides, fatty acids, and multiple classes of lipids in brain tissue has been identified for positive and negative mode detection, broadening the scope of molecules that can be detected from a single tissue section (Li et al., 2019a). Non-organic matrices, such as silver nanoparticles, have also been investigated and found to be able to profile 10 classes of lipids in the brain through MALDI-FTICR-MS analysis (Guan et al., 2018). Nanoparticles have also been employed for AP-MALDI-MSI analysis, a high spatial resolution ionization method, to localize and identify lipids and metabolites influenced by drugs in hippocampal tissue (Kim et al., 2017). Ice was also evaluated as a matrix for high resolution imaging of mouse brain through the modified infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) MSI platform (Ekelof & Muddiman, 2018; Robichaud, Barry & Muddiman, 2014). Matrix-free imaging approaches have also been developed. For example, a range of phospholipids were identified and localized in mouse brain and kidney sections while maintaining a simplified sample preparation workflow (Goodwin et al., 2011b; Lee et al., 2012).

Figure 2.

(a) UV absorption of candidate matrix compounds help to understand the variable ability for (b) on tissue ionization in positive (left) and negative (right) ion modes, demonstrating the importance of careful matrix consideration. Adapted and reprinted with permission from (Li et al., 2019a).

Another matrix-free method involves using silver nanopost array structures as substrates, which has been effective for a variety of samples, including peptides and metabolites (Korte, Morris & Vertes, 2019; Stopka et al., 2016). While matrix selection is important, appropriate deposition of the matrix is an equally critical consideration. Proper application of matrix must be employed to ensure small, homogenous crystallization. The two methods that enable the smallest, most uniform crystal size are sublimation and electrospray deposition. The sublimation technique was improved upon recently by the addition of a recrystallization step, which combined the reproducibility and spatial resolution of sublimation with improvements in detection by recrystallization for the detection of metabolites (Morikawa-Ichinose et al., 2019). Electrospray deposition has also been used for studying analytes in mouse brain, resulting in 1 μm homogenous crystals, greatly increasing the capacity for high spatial resolution imaging (Li et al., 2016; Wang et al., 2019). Additionally, the use of robotic sprayers has become a common way to achieve suitable matrix crystal sizes, particularly with dry application with organic solvent systems in which the sprayer is heated such that the solvent evaporates prior to interacting with the sample’s surface (Gemperline, Rawson & Li, 2014). Solvent-free dry matrix application methods that rely on sublimation of powdered matrix were also implemented for the MALDI-FTICR-MSI analysis of dopamine antagonists and other targets in rat brain tissue (Caughlin et al., 2017; Goodwin et al., 2011a). Pre-coating the slide with matrix and derivatization reagents such as trypsin prior to placing the tissue on the slide greatly simplifies the imaging workflow and achieves good results for proteins up to 300 kDa in rat brain (Basu et al., 2019; Zubair et al., 2016). While there are many factors associated with preparing a sample for MALDI-MS or MALDI-MSI analysis, the optimization and consideration that goes into each individual step is important for more sensitive, accurate, and reproducible results.

Acquisition and Analysis

Ionization Sources

MALDI is the most commonly used ionization source for MSI experiments due to its ease of use, high sensitivity, and applicability to a wide variety of biological molecules. There have been many improvements recently to MALDI sources in a variety of capacities to further expand its capabilities. Interestingly, simply adjusting the ionization parameters without any instrument modifications can have highly desirable effects on performance, as demonstrated by a recent study that systematically optimized various parameters based on the number of peaks, resolution, and mean error (Ferey et al., 2019). The effects were evaluated and the final optimized detection parameters resulted in improvements in all three areas for both positive and negative polarities for imaging metabolites in brain tissue (Ferey et al., 2019). MALDI has been further improved through a novel ionization interface with ESI for coupling to Orbitrap instruments (Belov et al., 2017). With this interface, MALDI was performed at sub-atmospheric pressure and ions were introduced into an electrodynamic dual funnel interface, resulting in improved sensitivity while being easily integrated with ESI (Belov et al., 2017). Another sub-atmospheric pressure MALDI source developed by MassTech has also been recently evaluated for neuropeptides and was found to, with the appropriately optimized parameters, improve sensitivity for imaging with reduced ion suppression (Li et al., 2019b). Furthermore, adjusting the gas flow and pressure in a commercial MALDI source to decrease off-axis gas disruption of ion focusing improved transmission of large protein ions, resulting in substantial improvements to signal-to-noise (S/N) ratios (Prentice et al., 2018). These modified setups show the potential for pushing the sensitivity of MALDI to further extend its current experimental limits.

The pressure can be even further elevated to atmospheric pressure (AP) for analyses closer to ambient conditions. High-resolution AP-MALDI sources have been developed for ultra-high spatial resolution with simple interfacing to high mass resolution analyzers, further expanding the scope of imaging analysis (Chen et al., 2018; Guenther et al., 2011; Kompauer, Heiles & Spengler, 2017a). Such sources have been applied to imaging a variety of molecules, most notably neuropeptides (Chen et al., 2018) and metabolites (Khalil et al., 2017) in the brain of model organisms. One interesting example includes using a high-resolution AP scanning microprobe MALDI setup to obtain high resolution and mass accuracy with on-tissue MS/MS to obtain metabolomics information in region-specific areas of Drosophila melanogaster (Khalil et al., 2017). Another notable study involved using an AP-MALDI source to image a diverse class of molecules, including neuropeptides, lipids, metabolites, and small proteins, from crustacean and rat brain to produce high mass and spatial resolution MS and MS/MS images to localize low abundance and fast degrading neuropeptides (Chen et al., 2018). Another interesting development with AP-MALDI that has a potential within the field of neuroscience is the utilization of a source with a long-distance laser triangulation that enables 3D chemical topography of tissue sections (Kompauer, Heiles & Spengler, 2017b). The source has been demonstrated its suitability for applications with plants and parasites, but it is likely to find a useful role in the field of neuroscience as well.

In addition to adjusting the pressure in MALDI-MSI analysis, implementing laser post-ionization, denoted as MALDI-2, can lead to an enhanced sensitivity in imaging experiments. The method works by performing a second laser irradiation of the initial MALDI plume, ionizing more molecules from the original plume and thus improving the sensitivity (Spivey et al., 2019). The method has been shown to yield up to a 100-fold increase in sensitivity for various molecules (Barre et al., 2019; Ellis et al., 2017). This improved sensitivity has enabled the practicality of improved spatial resolution by mitigating the tradeoff between laser ablation size and ion abundance (Spivey et al., 2019). This novel ionization source has been demonstrated successfully in imaging of mouse cerebellum tissue sections, showing its potential in neuroscience research (Niehaus et al., 2019).

The soft ionization ability of MALDI to desorb analyte molecules from a surface with relatively low fragmentation and to produce predominantly singly charged ions leads to one of the largest current challenges to the method: decreased gas-phase fragmentation efficiency to create abundant fragment ions for MS/MS characterization. As MS/MS spectra are critical to sequence assignment, two interesting techniques, laserspray ionization (Trimpin et al., 2010) and matrix-assisted ionization vacuum (MAIV) (Inutan & Trimpin, 2013), alter the traditional MALDI setup to generate multiply charged ions, expanding the detectable mass range, increasing fragmentation efficiency, and improving the capacity for de novo sequencing for protein identification. Laserspray ionization, which involves the laser ablation of a solid mixture off a surface in electric field-free atmospheric pressure conditions, was shown to achieve in situ peptide and protein identification, localization, and characterization from complex rat brain tissue (Chen, Lietz & Li, 2014). MAIV, on the other hand, produces multiply charged ions desorbed off a surface without the use of a laser or high voltage (Inutan & Trimpin, 2013). Studies using MAIV coupled to an Orbitrap demonstrates enhanced characterization abilities (Chen, Lietz & Li, 2018; Chen et al., 2016).

To further decrease the limitations of MALDI analysis, liquid extraction surface analysis (LESA) MS was developed for in situ tissue analysis, taking elements from both ESI and MALDI (Quanico et al., 2013; Ryan et al., 2018). Using liquid micro-extraction on tissue, spatial information can be retained while gaining the ability for online LC separation post tissue extraction and the use of ionization by electrospray. Liquid micro-extraction has been particularly useful in obtaining a more complete chemical profile compared to MALDI, particularly with single-cell analysis (Aerts et al., 2014; Comi et al., 2017). The method has been used with success to monitor dopamine (Gill et al., 2017) and sterols (Yutuc et al., 2020) in mouse brain tissue. Coupled with FTICR, incorporation of liquid micro-extraction enabled the acquisition of an MS/MS spectrum with fragment peaks that yielded confident identification (Kim et al., 2018). When compared directly to MALDI, the technique has been shown to be powerful in elucidating the role of precipitated morphine in inflammatory response. While MALDI is a powerful technique that will likely remain useful for a long time, innovative modifications can greatly increase the scope of research possibilities.

Acquisition Parameters

The high resolution offered by FT instrumentation has drastically improved the power of MSI, as evidenced by direct comparisons between high-resolution MALDI-MSI and MALDI-TOF/TOF MSI for crustacean and rat brain sections (Ye et al., 2013b). A comparison of the same region of interest in rat CNS tissue shows a more resolved peak with HRMS than with TOF-MS, enabling confident assignment of guanosine monophosphate. Furthermore, several low-abundance peaks of similar masses were resolved in the Orbitrap instrument, while these peaks were merged in the TOF-MS spectrum. In situ analysis of complex mixtures can be difficult but the use of a FTICR mass analyzer can provide significant S/N improvement through in-cell accumulation (ICA), where ions are trapped in the ion cyclotron resonance (ICR) cell and stored from multiple MALDI ionization events prior to analysis (Goodlett et al., 2000; O’Connor et al., 2004). To further improve the detection and level of information obtained from imaging experiments, modifications in instrumental acquisition parameters have been established, several of which have been particularly useful for neuroscience research. One such notable example is executing a pseudo gas-phase fractionation in order to perform both high resolution MSI and data-dependent acquisition (DDA) MS/MS in a single experiment to profile neuropeptides (OuYang, Chen & Li, 2015). Obtaining quality MS/MS spectra of neuropeptides in situ is difficult due to their low abundance in the presence of a complex sample matrix. By dividing the entire m/z range into segments based on abundance and optimizing various parameters, quality MS/MS spectra and localization information were obtained for low-abundance neuropeptides in crustacean brain tissue (OuYang, Chen & Li, 2015). Sequence information was obtained on-tissue for neuropeptides on a MALDI-Orbitrap platform through the careful optimization of all sample preparation, acquisition, and analysis parameters (Verhaert et al., 2010). Sensitivity was also improved by modifying the source gas flow and pressure of MALDI-FTICR in order to maximize ion transmission, which resulted in improved detection of large proteins in rat brain tissue (Prentice et al., 2018). With this method, the S/N ratio improved by an order of magnitude and proteins up to 22 kDa were detected, essentially doubling the mass range of the instrument platform (Prentice et al., 2018). These studies show how optimizing instrument parameters can greatly improve detection quality of molecules that have historically been difficult to analyze.

Software

With the improvements in both spatial resolution and mass resolution afforded by MALDI-MSI, there emerges a challenge in handling the large data files produced by these types of imaging runs. There have been numerous software advances specifically designed to overcome this challenge by streamlining the way data is handled and preprocessed in order to reduce computation time without losing important data. One approach to handling this challenge has been through the implementation of distributed computing with a continuous mode mosaic datacube for processing FTICR-MSI data, which resulted in an 8-fold improvement in processing time (Smith et al., 2015). Other notable developments have been made to address the challenges of large data specific to three dimensional MSI through hierarchical stochastic neighbor embedding (Abdelmoula et al., 2018) and multimodal or multi-center imaging with SpectralAnalysis software (Race et al., 2016). Additionally, a hierarchical clustering scheme denoted SpecPlot was used to analyze the large amounts of data acquired to identify and map neuropeptides in crustacean nervous tissue (Schmidt et al., 2008). The method successfully used a bioinformatics approach for high-throughput comparative peptidomics.

In addition to challenges associated with the large size of the data, there are also concerns regarding the reproducibility of imaging data, particularly when originating from different laboratories and instrument platforms. Several groups have worked to reduce this variability by evaluating experimental parameters on different instruments and discussing key considerations that should be made when performing multicenter studies, e.g. preparation of tissue samples and handling of data (Rompp et al., 2015a). Proper reporting guidelines have been established for what information is necessary to include when describing new MSI experiments in order to further improve reproducibility between research groups (McDonnell et al., 2015). Ensuring consensus on the information required to be revealed for publication facilitates repeatability of experiments in other laboratories. The transparency of MS imaging data is further improved by making use of public repositories, such as has been established by Rompp et al. (Rompp et al., 2015b). Maintaining transparency through adhering to uniform reporting guidelines, processing data in a common data format, e.g. imzml, and using public data repositories ensures the reliability of MS imaging data and provides confidence in the biological conclusions that can be made. There have also been developments made to reduce experimental repeatability within experiments. For example, Tu & Muddiman performed a systematic evaluation of normalization strategies in order to reduce the variation associated with long imaging runs (Tu & Muddiman, 2019). They found that using a local median normalization approach to be a robust method that reduced the relative standard deviation by up to 22% (Tu & Muddiman, 2019). As with any other experimental method, reproducibility should be prioritized with MS imaging experiments.

There have been numerous other notable advancements in software and data analysis capabilities for MSI, facilitating handling of high spatial and high mass resolution data offered by FT instrumentation. One such development has been the newest version of MSiReader, a MATLAB-based software that has been continuously evolving to meet the growing demands of the field (Robichaud et al., 2013). The newest version enables absolute quantification to be performed with reduced analysis times, as well as updates in features to include multimodal imaging capabilities, polarity filtering, and quality assurance features (Bokhart et al., 2018). Another MATLAB software has also been recently developed using digital image recognition methods to rank images by spatial correlation, with an image similarity scoring algorithm, resulting in rapid analysis of imaging datasets (Ekelof et al., 2018). Machine learning and multivariate regression were employed through a predictive image fusion method (Van de Plas et al., 2015) to improve spatial resolution (Prentice et al., 2018). For clinical applications, integrative clustering has recently been applied to multilevel molecular MSI data in order to stratify patient cohorts, successfully detecting patient subgroups (Balluff et al., 2019).To better visualize the heterogeneity of neurological responses through the brain, a dynamic 3D reconstruction of an entire mouse lung was created by Jones and colleagues, utilizing the high spatial resolution and high mass accuracy MALDI-FTICR-MSI platform in conjunction with histological staining for high confidence visualization of various lipid species (Jones et al., 2017a). This work was further expanded to profile lipid changes associated with traumatic brain injury via MALDI-LTQ-Orbitrap-XL-MSI (Mallah et al., 2018). A high-resolution 3D representation, see Figure 3, was created through the superposition of consecutive tissue images to visualize response post traumatic brain injury and reveal and localize lipid signatures of injured and uninjured brain tissue (Mallah et al., 2018). The numerous developments made in handling these traditionally-computationally expensive imaging experiments have substantially improved throughput and the depth of information able to be obtained, which demonstrates promising potential for making these techniques more ubiquitous in clinics.

Figure 3.

3D representation of imaged mouse brain tissue superimposed to demonstrate heterogeneity at different points enabling the visualization of tissue lesions. Reprinted with permission from (Mallah et al., 2018).

Multimodal Imaging

While high-resolution MSI is a powerful technique with continuously growing applications and capabilities, it does not always provide enough information for interpretation of biological processes in a given system. It is often advantageous to couple MSI with other localization-based techniques, such as histology to provide context to interpret high spatial resolution images, and sampling techniques such as laser-capture microdissection for deeper profiling. Laser capture microdissection, followed by LC-MS/MS can be performed on the regions of interest identified by MSI. Detailed chemical information can be obtained using the localized proteomics method for specific regions of the tissue (Dilillo et al., 2017b). The primary challenge with combining these different modalities together is co-registration to ensure that the data are precisely complementing each other; significant research effort has been devoted to address this challenge. The three modalities, MSI, histology, and laser microdissection, have recently been combined through the segmentation of the MSI data and passing the data to laser microdissection through the 3 co-registration steps (Dewez et al., 2019). The MSI data is then able to guide the laser microdissection for microproteomics of extracts, resulting in comprehensive characterization of cancerous tissue (Dewez et al., 2019). Histology and MSI were also combined to profile peptides and lipids in the human primary visual cortex and visualize differential localization of the biomolecules within the visual cortexes (Gonzalez de San Roman et al., 2018). To further improve histology-guided MSI, a new tool has recently been developed to use histology from adjacent tissues to focus analysis on specific areas of interest, instead of relying on analysis of large areas of tissue (Heijs et al., 2015a). The histology tissue section can then be annotated and used to define the area to be imaged with MSI, minimizing the time for data acquisition and analysis (Heijs et al., 2015a). Computational automated registration of MSI to autofluorescence microscopy has also been demonstrated to be powerful for streamlining in-depth molecular characterization (Patterson et al., 2018a; Patterson et al., 2018b). Those advances in throughput can help translating MSI techniques to the clinic for more routine analysis.

Furthermore, multimodal imaging can take advantage of the high spatial resolution offered by histology to enhance MSI spatial resolution beyond the limitations of the laser, sensitivity, and other experimental factors. This has been made possible by a patch-based super resolution method, where the histology information guides reconstruction of the MS image by relying on image redundancy of neighboring pixels (Scupakova et al., 2019). This method has been validated on mouse cerebellum with high qualitative and quantitative accuracy (Scupakova et al., 2019). Increased spatial resolution was further made possible through an image fusion prediction method to correlate MSI and microscopy data to predict ion distributions for a higher spatial resolution than can be measured (Van de Plas et al., 2015). Another notable advance in multimodal imaging comes from the development of a workflow enabling high resolution MSI and infrared imaging of the same tissue (Neumann et al., 2018). The precise spatial alignment enabled advanced image fusion and pan sharpening of MS images (Neumann et al., 2018). Other modalities have also been incorporated with MS imaging and histology. For example, using 3D MALDI-MSI combined with stimulated Raman scattering microscopy, magnetic resonance imaging (MRI), and histology, glioblastoma tumors in mouse brain were studied to characterize the pharmacokinetics and pharmacodynamics of a potential drug (Randall et al., 2018). Further work was performed to develop an automated scheme to co-register 3D MSI and MRI data (Abdelmoula et al., 2019). These improvements demonstrate the potential of multimodal imaging for enhancing the capabilities and applications of MALDI-MSI experiments.

Molecular Profiling

Calibration

The most prominent advantage of FT mass analyzers is the high resolving power and mass accuracy afforded, which enables both well-resolved peaks in complex samples and high confidence for assignment of individual analytes. However, optimal mass accuracy is dependent on careful and appropriate calibration of the mass analyzer, and while MALDI-FTICR experiments typically maintain mass errors below 5 parts-per-million (ppm), there can be large mass errors of over 100 ppm when critical factors are not taken into account. Of the several factors influencing mass accuracy of MALDI-FTICR measurements, one of the most critical is space-charge effects, occurring when a large number of ions repel one another and cause an impairment of measurement accuracy. The effect of space-charge was examined by systematically varying ion density independently of other factors (Easterling, Mize & Amster, 1999). Importantly, it was found that by accounting for total ion intensity of both external calibrant and analyte, the mass error can be decreased from 100 ppm to 0.1 ppm (Easterling, Mize & Amster, 1999). Another method for correcting for space-charge effects used ¹⁵N-labeled peptides (Jing & Amster, 2012). A mass error of 1 ppm was achieved by using the mass difference between ¹⁴N and ¹⁵N peak pairs after data collection (Jing & Amster, 2012).

Long imaging experiments on heterogeneous tissue require special consideration when it comes to mass calibration. Recently, calibration strategies were systematically evaluated for FTICR-MSI experiments, including external, internal, lock mass, and abundance corrected calibration. The calibration methods were evaluated for their performance and ease of implementation, and it was found that abundance correction was the most suitable, enabling sub-ppm mass error by correcting for fluctuations in ion abundance over entire tissue sections (Smith et al., 2012). Calibrant location on an MALDI target plate can also be a factor affecting mass accuracy due to differences in ion population, as is often a concern with MSI. Smith et al. demonstrated drastic improvements when simply using an internal calibrant versus an external calibrant for a new MALDI imaging source, including better mass accuracy and less variability, as shown in Figure 4 (Smith et al., 2011).

Figure 4.

Comparison of mass accuracy for internal versus external calibration for MALDI-MSI of lipids in a rat brain, demonstrating the improvement in both accuracy and reproducibility afforded by internal calibration. Reprinted with permission from (Smith et al., 2011).

By optimizing mass calibration techniques, novel developments were able to be made in the field of neuroscience. MALDI-FTICR-MS was used in conjunction with in-cell accumulation techniques for the in situ analysis of neuropeptides from the Jonah crab Cancer borealis with peptide sequencing capability (Kutz, Schmidt & Li, 2004). Kutz et al. demonstrated an increase in S/N ratio and improved mass measurement accuracy by incorporating calibrants on a spot separate from the analyte and by broadening the spectral mass range (Kutz, Schmidt & Li, 2004). This provides enhanced sensitivity for the detection of trace-level neuropeptides and increased confidence in peptide identification in complex neuronal samples. Stemmler and colleagues studied the Maine lobster, Homarus americanus, and demonstrated that MALDI-FTMS metastable fragments from Asp-Xxx cleavages can be used to facilitate orcokinin peptide detection during direct tissue analysis (Stemmler et al., 2005). Internally-calibrated spectra provided improved mass measurement accuracy and precision, in addition to predicted fragment ion fingerprints acquired in situ (Stemmler et al., 2005). The usefulness of proper mass calibration was demonstrated by correlating MALDI-MS images of proteins up to 12 kDa with LC-MS/MS data, drastically improving confidence and biological information of the proteins (Spraggins et al., 2015). In general, when care is taken to select an appropriate calibrant and to pay attention to specific experimental details, MALDI-FTICR can achieve mass accuracy comparable to or superior to its complementary ionization technique, ESI.

De novo sequencing

Interpreting mass spectra through peptide de novo sequencing, the act of assigning spectral fragment ions to specific amino acids, has enabled unknown and novel compounds to be identified. In typical proteomics workflows, a sequence database is used for matching MS/MS spectra in order to assign peptide identifications. When a database is unavailable, such as when there is no genome information available for an organism, de novo sequencing is used in order to identify peptides with minimal prior knowledge about the sample. Identifying neuropeptides with de novo sequencing is particularly advantageous for discovering novel peptides that have not previously been reported in the literature. By combining de novo sequencing with MSI, information can simultaneously be obtained for peptide sequence and specific tissue localization. In this way, more information is gained about novel peptides in a single analysis. With the advent of high mass accuracy and high resolution FT mass analyzers, the ability to make confident assignments has increased greatly. Increased mass resolution also facilitates the analysis of complex spectra and differentiation of fragment ion peaks with similar m/z values. Demonstrating the ability for de novo sequencing by MALDI-FTICR, Kutz et al. were able to identify neuropeptides with high confidence directly from neuroendocrine tissue (Kutz, Schmidt & Li, 2004). Furthering these results, the Li lab used multiple MS platforms, including MALDI-FTMS on tissue and tissue extract analysis, to enable de novo sequencing of 34 novel peptides with potential neurological function in the American lobster, Homarus americanus (Ma et al., 2008). The increased de novo sequencing capabilities for neuropeptide identification were also pursued and achieved through a multiplex MSI method coupled with DDA analysis using the HRAM MALDI-Orbitrap platform, demonstrated in Figure 5 (OuYang, Chen & Li, 2015). Applied to other biomolecules, it can improve the throughput for high confidence de novo sequencing.

Figure 5.

MS/MS spectrum and tissue image of a novel neuropeptide identified by de novo sequencing. Reprinted with permission from (OuYang, Chen & Li, 2015).

While HRAM MALDI techniques are useful for sequence elucidation, ion fragmentation, a requirement for thorough de novo sequencing, tends to be less efficient due to the production of predominantly singly-charged precursor ions. Hence, it is very beneficial and common to see de novo sequencing performed by MALDI in conjunction with orthogonal ESI analysis, enhancing identification of novel compounds in the brain. De novo sequencing capabilities of MALDI-LTQ-Orbitrap, MALDI-FTICR, MALDI-TOF/TOF and nanoLC-ESI-QTOF mass spectrometers were evaluated for crustacean neuropeptide analysis (Chen et al., 2010). The direct comparison demonstrated the advantages of higher mass accuracy and mass spectral resolution provided by the Orbitrap mass analyzer to yield the most confident sequence assignments of the MALDI platforms, with ESI providing complementary results. The use of ESI enables LC separation to be used prior to MS analysis, simplifying complex mixtures, and leads to the formation of multiply charged gas phase ions. This increases fragmentation efficiency and provides more informative MS/MS spectra and is therefore commonly used for de novo sequencing. However, the spatial resolution afforded by MALDI cannot be maintained, hence a multi-platform analysis is commonly performed for the sequencing and localization of unknown biomolecules.

The analysis and de novo sequencing of intact proteins on tissue by MALDI-FTMS has been limited by the finite mass range traditionally afforded by FT mass analyzers, thus ESI has primarily been used for intact protein sequencing. However, successful on-tissue characterization has been achieved with MALDI-FTMS. For example, the Caprioli lab demonstrated the use of MALDI-FTICR-MS to detect intact proteins (up to 12kDa) in rat brain and kidney tissue with high resolution and accurate mass, able to distinguish ions with different modifications and charge states and overlapping isotopic distributions with other neighboring ions, which is not possible with the TOF platform (Spraggins et al., 2015). The high-confidence identifications made by MALDI in situ analysis of the rat brain tissue enable their correlations with LC-ESI-MS/MS de novo sequencing data obtained from rat brain extract (Spraggins et al., 2015). The high mass resolving power and mass accuracy of FTMS mass analyzers enabled the differentiation, identification, and localization of intact proteins in a multifaceted approach with ESI de novo sequencing.

Post-Translational Modification Analysis

High resolution MS is used to profile a large class of molecules, including those that undergo post-translational modification (PTM). Identifying the presence, type, and location of a modification is of interest, as changes can lead to different neurological functions or binding in the brain and is often characterized through MS fragmentation. Commonly studied PTMs include phosphorylation, acetylation, glycosylation, and oxidation. Glycosylation is widely studied as it is a very diverse PTM with numerous glycan classes possible and a magnitude of combinations of glycan composition, structure, length, and branch position. As each of variations and structural heterogeneity affects glycan function which in turn affects activities of proteins and peptides through structural or regulatory changes, identifying a glycosylation modification and its site can elucidate its role in disease. In a multifaceted MS approach, 6 expected N-glycosylated sites were confirmed on the murine neural cell adhesion molecule from the immunoglobulin superfamily, identified in tryptic peptide fragments analyzed via MALDI-FTICR-MS (Albach et al., 2004). Using PNGase F to deglycosylate the glycopeptides, as unmodified peptide generally has a better ionization efficiency than its glycosylated counterpart, the authors were able to detect and identify each glycosylation site and glycoform present (Albach et al., 2004).

Phosphorylation, one of the most common PTMs, is associated with a variety of functional changes for many biological purposes, including regulating enzyme activity. While modified peptides are of low abundance, MS analysis in negative ion mode generates phosphorylated peptide ions with a higher S/N ratio, although the overall signal tends to be lower than when the positive ion mode is used (Hou et al., 2010). These modifications can occur at many different amino acid sites and can be present on multiple sites simultaneously, making it more difficult to characterize from fragment ions. Enabled by the high resolving capability of MALDI-FTICR-MS, Becker & Przybylski investigated phosphorylation modification of tau protein, a protein found in neurons, linked to Alzheimer’s disease (AD), and were able to identify and distinguish multiply-phosphorylated peptides and their sites of modification (Becker & Przybylski, 2007). Understanding where these PTMs occur is important as changes in phosphorylation are associated with the aggregation of tau in fibrillary tangles of brain tissue, as seen in AD patients. Phosphorylation was also studied by Bauer and colleagues, along with oxidation in the goldfish brain for its effects on neuronal protein reggie-2, a protein associated with the plasma membrane (Bauer et al., 2001). MALDI-FTICR-MS analysis was carried out to detect and characterize several phosphorylation and oxidation sites. Modifications were also investigated using a TOF, but due to its lower mass accuracy, the only assignable ions acquired belonged to unmodified partial sequences (Bauer et al., 2001). This comparison highlights the importance and need for a high resolution FTMS method for confident sequence characterization and identification of complex and modified samples. The ability of MALDI-FTMS to distinguish these PTM isoforms enables characterization of changes in the brain associated with neurological disorders/diseases.

While PTMs are studied widely due to their association with disease, there is some difficulty since these modifications are often labile, such as tyrosine sulfation. While some peptide hormones and neuropeptides are known to be sulfated, there are many cases where its neurological roles are unclear (Seibert & Sakmar, 2008). The biological roles for the sulfation of tyrosine are uncertain in many instances, but it has been linked to protein-protein interaction and function (Seibert & Sakmar, 2008). Characterization through MS is difficult as MALDI and ESI analysis in the negative ionization mode can result in the partial or complete neutral loss of the sulfo moiety, while positive ionization mode analysis leads to fully desulfated ions, impeding the detection and localization of this labile PTM. The high acidity of the sulfoester group however leads to a more stable deprotonated ion when analyzed in negative ion mode, though only a single sulfoester group can be stabilized at a time through MALDI, where singly charged ions are routinely observed (Seibert & Sakmar, 2008). The high mass accuracy of FTMS can be leveraged to localize tyrosine sulfation through the differentiation in mass shift between phosphorylation and sulfation of 0.0095 Da (Taylor et al., 2008). While ESI has been reported to provide less sulfation fragmentation (Onnerfjord, Heathfield & Heinegard, 2004), tyrosine sulfation can be characterized in MALDI with alternative fragmentation methods available, such as in-source decay, which enable more complete sulfation fragmentation. A top-down MALDI in-source decay FTICR method was developed to successfully localize the sulfation of proteins and antibodies (Tyshchuk et al., 2019). Another group demonstrated the use of ultraviolet, infrared, and high-low energy photodissociation for FTMS characterization of sulfo-peptides (Halim et al., 2018).

To further combat the difficulties of analyzing labile PTMs through MALDI-FTMS, studies have been performed to develop a softer ionization technique: MAIV, a modified MALDI technique for ionization forgoing the use of laser ablation or high voltage (Chen et al., 2016). This has been applied on a modified MALDI-LTQ-Orbitrap XL platform and its softer ionization enabled labile PTM analysis of intact peptides from rat brain extract and in situ. This technique was compared to MALDI when evaluating the use of MAI, MAIV at atmospheric pressure (Chakrabarty et al., 2013), and electron transfer dissociation on an Orbitrap Elite to further probe labile PTMs in rat brain tissue extract and peptide/protein standards containing known neuropeptides (Chen, Lietz & Li, 2018). Preservation of labile PTMs such as glycosylation was observed to a larger degree with MAI compared to MALDI (Chen, Lietz & Li, 2018). Although comparisons show that ESI preserves better the GalNAc attachment specifically, MAI had the benefit of higher throughput for sampling, with less sample consumption and carry over, demonstrating its ability as an alternative to MALDI or ESI for labile PTM analysis.

Quantitation

MALDI has been widely accepted as a powerful ionization source for profiling a wide variety of molecules and gaining detailed information. However, for many years MALDI was used predominantly for gaining qualitative information, as the lack of uniformity in sample and in matrix crystallization made the technique inherently irreproducible. It was not until recently that MALDI was embraced as a quantitative measurement technique, when new developments were made in chemical labeling tags and computational algorithms to improve the accuracy and robustness of ion abundance measurements. The high specificity offered by its coupling with high resolving power mass analyzers has led to numerous interesting research findings in the field of neuroscience.

Label-free methods for quantifying molecules with MALDI have been applied with notable success, particularly when applied to imaging tissues (Dilillo et al., 2017a; Heijs et al., 2015b). One such example utilized a novel nonlinear calibration model for the quantification of drug targets using computational methods that enabled improved data fitting and accounted for nonlinearities within the method (Abu Sammour et al., 2019). However, to accommodate for the crystallization heterogeneity in MALDI, stable isotopes are often employed using chemically reactive tags that enable more accurate relative abundance measurements. One such technique, using reductive methylation with isotopes of formaldehyde, has been used successfully for the relative quantification of neuropeptides in crustacean tissue (DeLaney, Buchberger & Li, 2018). Recently, changes in neuropeptide abundance from pH stress has been investigated with the method on a MALDI-Orbitrap platform (Liu et al., 2019). This study included a comparison of results with LC coupled to ESI to MALDI-MS, as shown in Figure 6. While the researchers found detection bias toward different families, the quantitation remained mostly consistent between the two methods, confirming the suitability of stable isotope labeling for MALDI quantitative analysis (Liu et al., 2019). The technique has also been applied to crustacean neuropeptide studies of temperature stress (Chen et al., 2014) and feeding-related changes (Zhang et al., 2018), the latter coupling CE with MALDI-FTICR for improved neuropeptidomic coverage as well as quantitative accuracy (Zhang et al., 2018). Hydrophilic interaction liquid chromatography (HILIC) has also been coupled with MALDI for performing relative quantitation of N-glycans with stable-isotope labeling (Chen et al., 2017). The platform used MS imaging as an alternative to fraction collection and subsequent spot analysis to improve separation resolution while maintaining high sensitivity and quantitative accuracy (Chen et al., 2017). Stable isotope labeling has also been used to study metabolites in mouse samples, incorporating isotopically labeled standards. One interesting implementation compared MALDI and silicon nanopost array-laser desorption/ionization and found good quantitative agreement between the two ionization sources (Korte, Morris & Vertes, 2019). Another recent development was quantifying metabolites in mouse tumors using MALDI-MS imaging, where stable isotopes were included for each analyte and k-means clustering was used for region-specific pixel-by-pixel quantitation (Swales et al., 2018). IR-MALDESI has also been shown to be effective for performing quantitative analysis of small molecules such as neurotransmitters in tissue with the inclusion of isotopically-labeled internal standards (Bagley et al., 2018; Nazari et al., 2018). The results showed good agreement with LC-ESI-MS for relative quantification (Nazari et al., 2018) and demonstrated the ability to reproducibly image small molecules without derivatization (Bagley et al., 2018). These developments highlight the broadened range of applications now possible for MALDI-based quantitative analysis.

Figure 6.

Quantitative results for stable isotope-labeled neuropeptides in various crustacean tissues analyzed with both MALDI and ESI, indicating good agreement between the two complementary ionization sources. Reprinted with permission from (Liu et al., 2019).

There are several software packages developed specifically for MALDI that facilitate quantitative experiments and improve the throughput and reproducibility of data analysis. One such example is MALDIquant, an R package that includes a complete pipeline for all steps in MALDI-MS analysis, including importing raw data, preprocessing steps, peak alignment, and calibration, in a modular fashion that allows customization for different types of experiments (Gibb & Strimmer, 2012). Another package, DeepQuanTR, has been specifically developed for LC-MALDI-MS in order to perform label-free quantification of proteins from various samples (Fugmann, Neri & Roesli, 2010). The software tool includes steps such as normalization to internal standards, extraction of peptide features, retention time adjustment, and relative quantification. The developers report a protein quantitation value that allows for inter-sample comparison and annotation of peptides to the corresponding proteins for improved quantitative accuracy (Fugmann, Neri & Roesli, 2010). Specifically designed for MALDI-MS imaging experiments, the msIQuant software suite enables rapid analysis and visualization of imaging data from large datasets, making it particularly useful for handling the large data files generated from high-resolution MS experiments (Kallback et al., 2016). The software’s quantitation feature enables the generation of calibration curves from internal standards in order to obtain absolute quantification (Kallback et al., 2016). These software tools highlight some of the advancements made particularly for handing the large volume of data generated from quantitative MALDI-MS experiments, further enhancing the potential for what can be measured.

Applications in Neuroscience

Neuropeptides and Peptides

Peptides related to the nervous and endocrine system, notably neuropeptides — peptides known to elicit a neurological response, have long been studied through MS, although HRAM mass analyzers have increased the ability to deconvolute complex samples and spectra. To directly evaluate the increased capabilities of MALDI-FTICR analysis compared to MALDI-TOF, Taban and colleagues sliced rat brain tissue down the center and imaged adjacent sections with each platform (Taban et al., 2007). Demonstrating the enhanced mass measurement capabilities of the FTICR, they identified many different endogenous peptides, neuropeptides, and lipids unable to be distinguished through parallel analysis by the TOF mass analyzer (Taban et al., 2007). This enables the elucidation of sequences of unknown compounds that may be potential biomarkers in the brain. Another study demonstrates the ability of MALDI-FTICR-MS to profile neurological molecules that cannot be characterized by TOF analysis. Using high-resolution mass spectrometry, Christie et al. were able to fully-sequence a peptide whose peak had been detected using MALDI-TOF 3 years prior, but its sequence was unable to be elucidated at that time (Christie et al., 2006). A novel SIFamine neuropeptide family member was identified from the stomatogastric nervous system of the American lobster, Homarus americanus with a Gly to Val substitution. Further investigation led to the discovery of its presence in the crayfish Procambarus clarkii and clawed lobster Homarus gammarus as well (Dickinson et al., 2008). This neuropeptide, partially responsible for modulating pyloric rhythm, and many other neurologically important biomolecules could not have been identified without the high mass accuracy and mass resolution capabilities of FTMS mass analyzers.

To further take advantage of the abilities of HRAM mass analyzers, a systematic and comprehensive neuropeptidomic investigation of the rat habenular nuclei using LC-ESI Orbitrap, LC-ESI-FTICR, and MALDI-FTICR-MS was performed by the Sweedler Lab. They identified and localized 262 neuropeptides from 27 prohormones in the medial habenula region and 177 neuropeptides produced from 20 prohormones in the lateral habenula region to better understand interaction in the brain and the role of the habenular nuclei (Yang et al., 2018). These large-scale studies can be very powerful in characterizing regions of the brain to find potential biomarkers for disease, although targeted approaches can be valuable as well. For example, deposits of Aβ peptides were detected and localized in the autopsied brain tissue images of Alzheimer’s disease patients due to their association with the disease (Ikegawa et al., 2019). By scrutinizing the various lengths of Aβ peptides, the authors were able to localize the deposition of specific Aβ peptides to leptomeningeal vessels, while other Aβ peptide variants were deposited as senile plaque in cerebral parenchyma (Ikegawa et al., 2019). These studies highlight the importance in studying the presence of specific peptides as well as their locations to discover potential biomarkers and better understand disease progression.

MALDI-FTMS has also been used to study and characterize model animal systems for neuropeptide analysis. Most notably, the crustacean neuropeptidome has been profiled with MALDI-FTMS in various studies (DeLaney, Buchberger & Li, 2018; Hui et al., 2012; Ma et al., 2010; Ma, Kutz-Naber & Li, 2007; Ma et al., 2009c; OuYang, Chen & Li, 2015; Stemmler et al., 2007b; Wang et al., 2009; Wang et al., 2008; Ye et al., 2013a). The first step in determining the physiological roles of neuropeptides is characterizing these diverse molecules to obtain a clear understanding of the molecules present in the nervous system. Neuropeptide transmitters and hormones in the nervous system and neuroendocrine organs of the American lobster Homarus americanus have thus been characterized via MALDI-FTMS (Ma et al., 2008). A larger neuropeptide profiling study was performed using MALDI-FTMS on commissural ganglia tissues from 32 crustacean species from seven decapod infraorders (Stemmler et al., 2007a). The ultra-high mass resolution, accurate-mass measurements, and in-cell accumulation capabilities enabled identification of many neuropeptides and detection of a few common peptides present in all species (Stemmler et al., 2007a), contributing to a list of species-conserved neuropeptides identified by MALDI-FTMS (Dickinson et al., 2009a; Dickinson et al., 2009b; Stemmler et al., 2010). While MALDI-FTMS has propelled the characterization of the crustacean neuropeptidome, peptide identification capabilities were further increased by coupling membrane-assisted capillary isoelectric focusing to MALDI-FTMS, enhancing the observed signal and leading to the identification of a number of observed orcokinin family neuropeptides in Cancer borealis and Callinectes sapidus brain extracts (Zhang et al., 2011).

Neurotransmitters, Metabolites, and Other Small Molecules

Neurotransmitters and metabolites are essential signaling molecules in the central nervous system. Detecting responses and changes in abundance or expression of these molecules, while also considering their localization to specific regions in the brain, can facilitate functional characterization. This can be crucial to understand the effects of drugs or chemicals on the neurological changes that lead to responses. To investigate this, metabolic profiles of 24 endogenous compounds found in mice with a neurodegenerative disease, multiple sclerosis, were investigated for spatial and semi-quantitative information after exposure to a promising drug, teriflunomide (Rzagalinski et al., 2019). Performing MSI with high spatial resolution of brain coronal sections enabled the capacity for the drug to cross the blood brain barrier to be assessed, as well as the locations of metabolites affected by drug treatment. The high resolving power of the FTICR allowed for tentative identifications of metabolic pathways interacting with the drug (Rzagalinski et al., 2019). To elucidate the link between region-specific changes and neural disease, neurotransmitter responses to exposure to the neurotoxic chemical tetrabromobisphenol A, a common flame retardant, was studied by the Muddiman group using an IR-MALDESI Orbitrap platform to image specific brain tissue regions of exposed rats (Bagley et al., 2018). While a mouse brain was profiled after exposure to ubiquinol via MALDI-TOF, MALDI-FTICR was employed to take advantage of the higher mass resolution for identification to detect biomarkers for cardiac and neurodegenerative disease (Tatsuta et al., 2017). Knowing the distribution of neurotransmitters and metabolites throughout the brain and their responses to drugs offers an improved understanding of neurological signaling, helping to assign functions to certain regions of the brain and the nervous system.

While the extensive characterization of the crustacean neuropeptidome has been described in the previous section, it is also important to note that studies have been performed to understand the roles of neurotransmitters and metabolites in the crustacean model as well. Identification and localization of crustacean neurotransmitters and metabolites found in the brain using MALDI-LTQ-Orbitrap-MS were performed to understand metabolic signaling in the nervous system. Cao et al. were able to identify neurotransmitters only localized to specific regions on the brain while others were distributed throughout the brain, although in varying abundances (Cao et al., 2019). The high-resolution mass spectra from MSI analysis allowed for confident identification and localization while LC-MS/MS was also performed to further confirm structural assignment for identified neurotransmitters. Similar work was performed by the same lab previously to localize metabolites and neurotransmitters in the crustacean and rodent central nervous systems (Ye et al., 2013b). The distributions of these small signaling molecules in complex biological tissue provides important insight into neuroscience-related research.

Proteins

The use of the high spatial resolution and mass accuracy capabilities of MALDI-FTMS for the detection of proteins and phospholipids was demonstrated through imaging mouse brain (Rompp et al., 2015a). Proteins have also been largely studied for their role in neurobiology through MALDI-FTMS, some of which include facilitating neuronal communication and regulating transport through the blood brain barrier. Using MALDI-Orbitrap-MS/MS and LC-ESI-MS/MS (LTQ Orbitrap), Bell and colleagues investigated the effect of glaucomatous serum and an antibody on proteins produced by neuroretinal cells to understand the degenerative disease, glaucoma (Bell et al., 2015). They were able to elucidate the proteomic changes of cells treated with primary open-angle glaucoma serum, noting differences in regulation of proteins involved in cell regulation. Incubation with a targeting antibody lead to the observation of abundance(?) changes of several proteins beneficial to diseased cell apoptosis. As proteins are crucial for cell communication, single cell MALDI-FTMS analysis of dorsal root ganglia neuronal cell bodies have been performed to identify proteins, along with peptides and lipids, to better characterize cell heterogeneity (Do et al., 2018). Single-cell analysis is particularly challenging due to the severely limited sample size, requiring techniques with both high sensitivity and high specificity. However, studying the discrete chemical composition of individual cells can provide improved insight into disease mechanisms and therapeutics. This important advancement in single-cell analysis with MALDI-FTMS provides insight into the cooperative function of morphologically similar cells with distinct protein expression. These differences in expression can best be observed through thorough studies which are enabled by the increased capability of MALDI-FTMS to identify and localize proteins.

In a notable study demonstrating the capabilities of a HRMS platform, Dilillo et al. were able to differentiate between proteins with overlapping m/z distributions that were expressed differently in the tumor and healthy regions of coronal mouse brain sections (Dilillo et al., 2017a). As shown in Figure 7, these tumor region specific proteins and proteoforms could not have been identified using a TOF mass analyzer. While the presence of protein isoforms can greatly complicate MS spectra, the use of FTMS enables the ability to accurately identify these modified proteins as potential biomarkers. The analysis of these proteins was correlated with other biomolecules through a multimodal MSI approach involving laser ablation inductively coupled plasma mass spectrometry imaging (LA-ICP-MSI), a hard ionization technique (Gonzalez de San Roman et al., 2018). LA-ICP-MSI and MALDI-MSI was used to identify and elucidate the distributions of 13 proteins, 71 peptides, and 123 lipid species in human brain primary visual cortex tissue (Gonzalez de San Roman et al., 2018). Applying similar comprehensive approaches to other proteomic imaging studies of the brain and related tissue can lead to further discoveries in the field.

Figure 7.

(a) Average spectra aquired via MALDI-TOF-MSI (top) and MALDI-FTICR-MSI (bottom) of mouse brain tissue containing a tumor. (b) Scanned image of the tissue highlight the tumor region. (c) The average TOF (left) and FTICR (middle) mass analyzer aquired spectra, and overlay of the average FTICR aquired spectra from tumor (pink) and healthy (blue) regions which shadded regions highlighting tumor specific proteoforms only able to be identified via FTICR analysis. (d) The MALDI-FTICR platform is able to distinguish two proteoforms of Histone H2B and another protein. (e) The FTICR platforms enables the differentiation of two m/z overlapping proteins with interspered and isobaric isotopomers with different localization in the tumor and healthy tissue. (f) Average mass spectra of tumor and healthy regions, showing potential tumor biomarker shown to be undetected by TOF analysis. Reprinted with permission from (Dilillo et al., 2017a).

Lipids

Numerous lipid classes are present throughout the brain in varying abundance with changes in response to stressors. Lipids commonly occupy similar m/z values with high chemical diversity, making them difficult to identify. By utilizing the HRAM FTMS platforms to distinguish individual lipid species, their neurological responses and their localization to regions in the brain can be studied. Such investigations may lead to a better understanding of disease mechanisms and to make progresses towards the prevention and cure of diseases. Gaucher disease infected mouse brain tissue was analyzed via MALDI-FTICR-MSI and LC-ESI-MS/MS to determine glycosphingolipid abundance and localization (Jones et al., 2017b). With the aid of immunohistochemistry, the observed brain inflammation regions with altered autophagy and degradation of proteins were correlated with the increased accumulation of specific identified glycosphingolipids (Jones et al., 2017b). In a notable study, glycerophosphoglycerols and sphingolipids were identified as classifiers to rapidly and accurately distinguish medulloblastoma and pineoblastoma, two pediatric tumor types commonly mistaken for each other due to histopathological similarities (Clark et al., 2018). In another imaging study, changes in ganglioside and ceramide expression and distribution throughout mouse brain was monitored after exposure to blast induced traumatic brain injury from different distances and time points post-blast via MALDI LTQ-Orbitrap-XL-MSI (Woods et al., 2013). The localization of biomolecule response after trauma to specific regions of the brain is key to a deeper understanding of neurobiology for development of a treatment. Enabled by sensitive MALDI-FTICR-MSI techniques, single cell lipid analyses of rodent cerebellar cells were pursued to characterize cell-to-cell heterogeneity (Neumann et al., 2019). Certain lipid features were localized in only a few of many studied cells, potentially indicating different functions by specific brain cells, again highlighting the importance of localization and identification for lipid analysis.

Conclusion

The coupling of MALDI and high-resolution mass analyzers has experienced rapid growth in recent years, particularly within the field of MSI. As a result, there have been numerous developments in all aspects of the experimental workflow from sample preparation to data analysis. These developments have served to enhance the molecular information obtainable from a single tissue section, both in terms of breadth of identifications and characterization of individual molecules. Advances in sample preparation have largely involved considerations of derivatization methods, on-tissue sample cleanup steps, and matrix composition and application techniques for improved ionization and reproducibility. MALDI-MS has been used to gain structural information about the chemical makeup of molecules through developments in de novo sequencing and PTM analysis, as well as functional information through both label and label-free quantitative approaches. Improvements in ionization and instrumentation have additionally been established to enhance detection capabilities, particularly of low abundance signaling molecules in the nervous system. Finally, novel and improved data analysis approaches have been employed to expand the information we can gain from these experiments. These notable advancements have led to a variety of recent interesting applications in the field of neuroscience, including those related to neuropeptides, proteins, lipids, and small molecules such as neurotransmitters and metabolites, though many directions remain to be pursued. As challenges in these respective fields are being addressed, it will become increasingly important to keep discoveries up to date in the field of neurobiology.

While the high mass resolution offered by Fourier transform mass analyzers, such as the FTICR and the Orbitrap, has greatly expanded the capabilities of a single analysis, it also increases the size of data generated. This result, combined with the large data requirements for high spatial resolution imaging, makes data analysis a current bottleneck in these experiments. Experiment time is also extended by the longer acquisition times required for HRMS compared to TOF instruments. Furthermore, the limitations of MALDI and similar laser-based ionization methods, including matrix interference and difficulties in incorporating a separation component, prevent high-resolution MALDI-MSI from generating the high-specificity MS/MS information offered by ESI methods. However, many researchers are making substantial progress towards overcoming these challenges, and so it likely remains only a matter of time before MALDI-FTICR and MALDI-Orbitrap workflows are ubiquitous for molecular analysis. We can expect to see additional exciting developments in the areas of both acquiring large volumes of data and handling this data from a computational standpoint. The field has been moving steadily forward in improving both spatial and mass resolution and throughput, which will inevitably increase both the size and amount of data generated. Adding additional layers of information, such as through high-quality MS/MS or multi-omics approaches, will enhance the depth of these analyses. As these improvements are made towards acquiring data of increasing chemical complexity, sophisticated computational advances will continue to be made to handle the large data sets. As the molecular complexity of various facets of neuroscience rely on MS for the elucidation of the structure and function of important biological molecules such as neuropeptides, neurotransmitters, lipids, and proteins, the field has been propelled forward by advancements in high-resolution MALDI-MS. We anticipate that more information will soon be unveiled about the deeper complexities of the nervous system with this powerful technique.

Abbreviations

(9AA)

9-aminoacridine

(AD)

Alzheimer’s disease

(AP)

atmospheric pressure

(CE)

capillary electrophoresis

(CHCA)

α-cyano-4-hydroxycinnamic acid

(CO₂)

carbon dioxide

(DDA)

data dependent acquisition

(DHB)

2,5-dihydroxy benzoic acid

(ESI)

electrospray ionization

(FT)

Fourier transform

(FTICR)

Fourier transform ion cyclotron resonance

(HILIC)

hydrophilic interaction liquid chromatography

(HRAM)

high resolution accurate mass

(HRMS)

high resolution mass spectrometry

(ICA)

in-cell accumulation

(ICR)

ion cyclotron resonance

(IR-MALDESI)

infrared matrix-assisted laser desorption electrospray ionization

(LA-ICP-MSI)

laser ablation inductively coupled plasma mass spectrometry imaging

(LESA)

liquid extraction surface analysis

(m/z)

mass-to-charge ratio

(MAI)

matrix-assisted ionization

(MAIV)

matrix-assisted ionization vacuum

(MALDI)

matrix-assisted laser desorption/ionization

(MALDI-TOF/TOF)

matrix-assisted laser desorption/ionization tandem time-of-flight

(MRI)

magnetic resonance imaging

(MS)

mass spectrometry

(MS/MS)

tandem mass spectrometry

(MSI)

mass spectrometry imaging

(PTM)

post-translational modification

(RPCH)

red pigment-concentrating hormone

(S/N)

signal-to-noise

(TOF)

time-of-flight