Introduction

In the human body, arguably the brain is of most fundamental importance because it controls thinking, feelings, and memories, and also our major actions and reactions. The brain is also an incredibly complex organ integrating many parts. Although it comprises only ca. 2% of the body weight of an adult human, it accounts for nearly 20% of the consumed energy [1]. Knowledge of brain operating systems is essential in order to seek a cure for brain disorders such as drug addiction, depression, and a series of neurodegenerative diseases.

Neurotransmitters and neuromodulators play prominent roles in brain functioning: the former by promoting the intersynaptic signal transmission and the latter by modulating postsynaptic events. Of these two, neurotransmitters are the most common class of chemical messengers in the nervous system [2]. These relatively low molecular weight molecules (<200 Da) are subdivided into two main groups according to their chemical structure (i.e., biogenic amines and amino acids). They have been studied in brain tissue mainly by indirect analyses [3], employing techniques that are also time- and labor-consuming, and commonly with poor specificity [4].

A major breakthrough in tissue analysis has been launched by mass spectrometry imaging (MSI) [5, 6]. MSI combines the speed, sensitivity, and selectivity of MS with spatial distribution analysis at the molecular level to provide a new dimension for histology. MSI has enabled the 2D visualization of the arrangement of many types of biomolecules in different tissues [7]. Recently, MSI has also been applied to the detection and spatial localization of neurotransmitters [8, 9]. Matrix-assisted laser desorption ionization (MALDI)-MSI, described initially by Caprioli and co-workers [10], has been the ionization technique used in the majority of studies on this subject. For example, acetylcholine has been imaged by MALDI MS/MS-based MSI [11], MALDI-MSI using a deuterated matrix [12, 13], and MALDI high-resolution and high-accuracy MS (HRMS) [14]. The biogenic amine dopamine [13, 15, 16] and the amino acids γ-aminobutyric (GABA) [13, 15–19], glutamic [15, 16], and aspartic acids and serine [16] have also been imaged by MSI using MALDI as the ionization technique in all but one of these studies [17].

MALDI operates under vacuum and requires a matrix as a proton donor that should be co-crystallized on the tissue surface. The quality of the crystals and the homogeneity of the matrix deposition are crucial for the image quality. Additionally, for analysis of low molecular weight compounds, such as neurotransmitters, the matrix represents a limiting factor because it produces many other isobaric interferences [20].

As an alternative, desorption electrospray ionization (DESI) has offered an ambient desorption/ionization technique suitable for MSI. As an ambient technique, DESI operates under atmospheric pressure [21]. DESI is also matrix-free and can utilize different types of spray solvents to improve analyte selectivity and nondestructively analyze tissue [22]. The absence of matrix not only simplifies the workflow but also enables the DESI method to be more straightforward in neurotransmitter detection because it is free from matrix interferences. Herein, we demonstrate that DESI-MSI coupled to a high resolution and high accuracy mass spectrometer is able to provide proper spatial distribution of GABA, glutamate, aspartate, serine, acetylcholine, dopamine and choline in rat brain slices. A preliminary report on this study has been presented [23].

Experimental

Materials and Methods

Unless otherwise stated, all chemicals and reagents were from Sigma-Aldrich, (Saint Louis, MO, USA) and used without further purification. Sodium pentobarbital was from Cristália Produtos Químicos e Farmacêuticos Ltda (Campinas, SP, Brazil).

Mixture of Standards

Stock solutions of GABA (100 mM), choline chloride (120 mM), acetylcholine acetate (90 mM), aspartic acid (20 mM), glutamic acid (14 mM), serine (100 mM), and dopamine chloride (50 mM) were prepared using distilled and deionized water (dd water) and were kept at −20 °C until use. On the day of the experiment, working solutions were made by diluting the stock solutions in dd water at a final concentration of 5 mM of each standard.

Animal Dissection

Male adult Wistar rats (300–400 g) were deeply anesthetized with sodium pentobarbital (75 mg/kg, i.p.) and were killed by decapitation. All procedures were approved by the institutional animal care committee at UNICAMP (CEUA Protocol numbers 2534–1 and 4123–1), and experiments were performed in accordance with the guidelines for animal care. After decapitation, the brain was dissected within 1 min and was immediately deeply frozen (for cryosectioning) in liquid nitrogen. Alternatively, the brain was kept in an ice bath and homogenized. The imaging experiments were conducted with brains obtained from two different rats. One brain was used for sagittal sections and the other for coronal sections. The cerebral and cerebellar extracts were prepared from a third animal.

Tissue Section Preparation

The frozen brains were cut using a Leica CM 1900 cryostat-microtome (Leica Biosystems, Nussloch, Germany). Sagittal and coronal brain sections were cut at a thickness of 14 μm. Tissue sections were transferred by thaw mounting onto conventional microscope glass slides without any surface treatment and were stored at −80 °C. Sections were desiccated at room temperature 15 min before use.

Brain Extracts

Cerebral and cerebellar homogenates were prepared according to the literature [24, 25] with modifications. After decapitation, the cerebrum (one hemisphere ~600 mg) and cerebellum (~400 mg) were collected and placed in 2 and 1.5 mL, respectively, of dd water. Tissues were manually homogenized on ice in a 5 mL Glass/PTFE Potter Elvehjem Tissue Grinder (Kimble Chase, Vineland, NJ, USA) until no chunks remained (20–30 strokes of the pestle) and were centrifuged (Smart R17; Hanil Science Industrial Co. Ltd., (Gimpo, Gyeonggi Province, South Korea)) at 14,000 rpm and 10 °C for 10 min. Supernatants were collected and stored at −20 °C until use.

Spot Analysis

Two microliters of cerebral or cerebellar homogenates or working solution of standards were spotted in an Omni Slide (Prosolia Inc., Indianapolis, IN, USA) and analyzed under DESI conditions. Each homogenate or solution was analyzed in triplicate. Initially, spots were analyzed in the positive ion mode and were then subsequentely analyzed in the negative ion mode, which makes the images of the spots in the negative ion mode look splashed.

DESI-MS and MSI Analyses

The analyses were performed in a Q-Exactive (Thermo Scientific, Bremen, Germany) mass spectrometer with a resolution of 140,000 at m/z 400 coupled with an Omni Spray Ion Source 2-D (Prosolia Inc.) for data acquisition. Firefly v.1.3.0.0 data conversion was used to generate the images, which were treated in BioMAp3804. The step size was 0.15 μm, the flow rate was 3.0 μL.min−1 and the surface scan rate was 600 μm.s−1. Analyses were performed in either the negative and or positive ion modes. The S-Lens RF level was set to 20 in order to increase the transmission of low m/z ions, the capillary temperature was 280 °C and the spray voltage was 3.4 kV. The images were acquired with 75,000 resolution. The signals attributed to the neurotransmitters were accurately assigned with errors less than 2 ppm (see Supporting Information for chemical structures and details). MSI was performed with a spatial resolution [26] of 200 μm (See Supporting Information for details - Figure S1).

Synthesis of Acetylcholine

The acetylation of choline was performed according to the literature [27] with slight modifications. Experimental synthesis and spectroscopic data (1H and 13C NMR) are available in the Supporting Information.

Results and Discussion

Neurotransmitters in rat brain slices have been imaged by MSI since 2010 [17] exclusively using MALDI as the ionization technique. One exception was the pioneering work on neurotransmitter imaging that used laser ablation electrospray ionization MS to image GABA and choline [17]. In addition, DESI-MSI was applied to image norepinephrine and epinephrine in porcine and slices of rabbit adrenal glands [28].

The relatively low concentration of neurotransmitters and the presence of other easily ionizable and abundant biomolecules such as lipids make the process of neurotransmitter imaging in brain slices similar to ‘looking for a needle in a haystack’ [9, 29, 30]. Another challenge is the presence of many isobaric molecules from the tissues or from the matrix in MALDI-MSI, which interfere with low-resolution MS analyses. In this scenario, DESI-MSI could contribute by specifically circumventing the problem of isobaric low molecular weight interferences from the matrix.

We therefore first examined ion suppression under DESI-MSI conditions for neurotransmitter analysis. For that we used an equimolar mixture of standards. Table 1 presents the ions via their m/z ratios observed in the spectra in Figure 1. We found the following preference for ionization in the positive ion mode: acetylcholine of m/z 146.118 > choline of m/z 104.107 > [dopamine + H]+ of m/z 154.086 > [GABA + H]+ of m/z 104.071 (Figure 1a). Fortunately, acetylcholine and choline are cationic species, making them readily ionized by DESI, resulting in the two more intense ion peaks. Dopamine is a strong base, is easily protonated, and produces a strong ion peak. In contrast to the isolated molecule, which produces an intense DESI-MS ion peak, the peak intensity of protonated GABA decreases to only 1% of the original value when present in the equimolecular mixture tested (not shown). This decrease indicates intense ion suppression [31, 32] suffered by GABA during DESI-MSI analysis of neurotransmitters in tissues. Peak attribution for these neurotransmitters performed by DESI-MS operating with a resolution of 140 K and a mass accuracy of less than 2 ppm (Supplementary Table S1), in rat cerebral and cerebellar extracts. In the cerebellar extract, GABA, choline, and acetylcholine could be detected in the positive ion mode, with choline producing the most intense ion peak (Figure 1b). In the cerebral extract (Figure 1c), only acetylcholine and choline were detected, with the strong prevalence of the ion peak of choline over that of the acetylcholine neurotransmitter. The analysis of the mixture of standards in the negative ion mode revealed that serine suffers strong ion suppression from both the glutamic and aspartic acids (Figure 1d). The preference for DESI(−) ionization was: [aspartate – H]− of m/z 132.029 > [glutamate – H]− of m/z 146.046 > [serine – H]− of m/z 104.035. All of these amino acids could also be detected in the cerebellar (Figure 1e) and cerebral (Figure 1f) extracts.

Table 1 Mass‐To‐Charge Ratios (m/z) Observed in Figure 1
Full size table
Figure 1
figure 1

DESI(+)-MS of neurotransmitter standards and extracts. (a) DESI(+)-MS of an equimolar mixture (5 mM) of GABA (m/z 104.071, [M + H]+), choline (m/z 104.107, [M]+), acetylcholine (m/z 146.118, [M]+), and dopamine (m/z 154.086, [M + H]+). (b) DESI(+)-MS of the cerebellar extract. (c) DESI(+)-MS of the cerebral extract. (d) DESI(−)-MS of an equimolar mixture (5 mM) of serine (104.035, [M – H]−, aspartic acid (132.029, [M – H]−), and glutamic acid (146.046, [M – H]−). (e) DESI(−)-MS of the cerebellar extract. (f) DESI(−)-MS of the cerebral extract. Color bars were added to facilitate visualization of the ion peaks of interest

Full size image

In addition to the use of HRMS Orbitrap for the unambiguous DESI detection of these neurotransmitters in cerebellar and cerebral extracts, we performed simultaneous DESI-MSI of the mixture of the standards and the extracts. Figure 2 shows the ion images generated in either the positive (Figure 2a–d) or negative ion modes (Figure 2e–f). The concomitant detection of all of these molecules in the mixtures of standards and in the extracts was observed. The direct comparison between the cerebellar (lines 2) and cerebral extracts (lines 3) shows that acetylcholine (Figure 2c) seems to be more abundant in cerebral extracts compared with cerebellar extracts. In contrast, glutamate (Figure 2g) is more abundant in cerebellar extracts (line 2). The abundance of choline is also notably larger in the extracts than in the mixture of the standards, which may occur because of better ionization of choline in the spots of the extracts due to low ionic suppression exerted by the other analytes, as by acetylcholine in particular. The increase in the concentration of choline during the preparation of the extracts may also exacerbate this effect [33].

Figure 2
figure 2

DESI-MSI of the spots of the standard solutions (first line in each Panel), cerebellar extract (second line in each Panel), and cerebral extract (third line in each Panel). (a–d) DESI(+)-MSI for (a) GABA, (b) choline, (c) acetylcholine, and (d) dopamine. (e – g) DESI(−)-MSI for (e) serine, (f) aspartic acid, and (g) glutamic acid. Scale bar: 2 mm

Full size image

The concern about misassignments of these low molecular weight neurotransmitter with a myriad of isobars (and isomers) in MSI has been previously raised [19], but the use of a HRMS Orbitrap analyzer and the concomitant image of the extracts and standards monitoring the same m/z value helps to minimize misassignments. Indeed, Figure 2 confirms that neurotransmitters can be imaged by DESI-MSI with proper visualization of their spatial distributions.

An advantage of avoiding derivatization agents is the possibility of imaging a broad range of molecules under the same conditions or in the same tissue slices while avoiding the risk of ion suppression or interfering side reactions [34]. Figure 3a shows the image of a sagittal rat brain slice generated from deprotonated sulfatide 24:1 of m/z 888.600, which is found to be strictly connected to the white matter and brainstem [35]. Figure 3f shows the image of a coronal rat brain slice generated from the deprotonated dimer of docosahexaenoic acid (DHA) [36] of m/z 654.567, which shows to be more abundant in the striatum. These images show the versatility of DESI-MSI as an alternative for the more common Nissl stain protocol (Figure 3k and l) to define brain structures. Figure 3b and g display images generated by the ion of m/z 146.118 corresponding to the cationic acetylcholine in the sagittal and coronal sections, respectively. These images show greater relative abundance of this neurotransmitter in the striatum, thalamus, and midbrain. Despite the relatively high concentration of cholinergic neurons in the brain, acetylcholine has been mapped only indirectly via its receptors or by enzymes related to their synthesis or degradation [15]. MALDI-MS/MS imaging was used to map acetylcholine in sagittal rat brain slices [11], and highest concentrations were found in the hippocampus, thalamus, and striatum. A different spatial distribution of acetylcholine was obtained by MALDI-MSI using D4-α-cyano-4-hydroxycinnamic acid (D4-CHCA) as the matrix [12], which indicated that acetylcholine is most abundant in the cortex, corpus callosum, ventral hippocampal commissure, thalamus, and cerebellum. MALDI-HRMS for rat brain slices also show major localization of acetylcholine in the cortex and brainstem [14], whereas high spatial resolution MALDI-MSI found acetylcholine mainly in the striatum, hippocampus, thalamus, pons, and other small structures. This MALDI-MSI distribution was corroborated by the intraperitoneal administration of tacrine, a central cholinesterase inhibitor that evoked a 7-fold increase in the concentration of acetylcholine-enriched regions [15]. There is, therefore, good agreement between our DESI-MSI spatial distribution of acetylcholine and several previous MALDI-MSI data from sagittal rat brain sections.

Figure 3
figure 3

(+)-DESI-MSI of neurotransmitters and metabolites in sagittal (b–e) and coronal (g–j) and (−)-DESI-MSI of lipids in sagittal (a) and coronal (f) rat brain sections. Relative abundance and spatial distribution of the ions of m/z 888.600 and 654.567 (a) and (f), respectively; STR = striatum, HPF = hippocampal formation, TH = thalamus, HY = hypothalamus, MB = midbrain, CB = cerebellum, P = pons, MY = medulla, CC = corpus callosum, CTX = cortex; acetylcholine (b) and (g); GABA (c) and (h); choline (d) and (i), and dopamine (e) and (j). Sagittal (k) and coronal (l) adjacent rat brain slices stained using the Nissl protocol. Scale bar: 5 mm

Full size image

The DESI-MS coronal view at the striatum level (Figure 3g) shows the presence of the acetylcholine neurotransmitter in this brain region extending toward the olfactory tubercle. As far as we could check, no profiles of acetylcholine in coronal sections have been reported. High spatial resolution MALDI-MSI images of acetylcholine in a sagittal rat brain section have, however, been reported, which corroborate our findings and is in accordance with the extensive axonal arborization found in the cell bodies of the striatal cholinergic interneurons [15].

The images in Figure 3c and h are for the spatial distribution of the neurotransmitter GABA, detected as its protonated molecule [GABA + H]+ of m/z 104.071. The sagittal image of Figure 3c reveals more pronounced abundances of GABA in the thalamus, midbrain, and basal forebrain. In accordance with what was observed in the sagittal section, this neurotransmitter is in the coronal section at the striatum level found most notably in the cortex and basal forebrain (Figure 3h). GABA is an inhibitory neurotransmitter that was the first to be imaged by MS [17]. More recently, the use of a derivatization strategy to obtain MALDI improved images of GABA distribution that closely resemble the images shown here has been reported [15, 16]. In these works, GABA MSI of sagittal rat brain sections revealed the prevalence of this neurotransmitter in the hypothalamus, midbrain, and basal forebrain. GABA was also found in the medial septum/diagonal band region (MSDB) of the brain [15]. GABA predominates in the hypothalamus and is also important in the midbrain and basal forebrain and its substructures [15, 37]. These data are in accordance with the DESI-MSI obtained herein.

Choline is not a neurotransmitter but is a precursor for the synthesis of phosphatidylcholines (PC), which comprises one-half of the total membrane lipid content. The increase in cell proliferation and cell membrane synthesis during tumorigenesis is known to affect the choline metabolism. Three-dimensional profiles of choline and choline-containing compounds are commonly obtained by proton magnetic resonance spectroscopy (1H-MRS) and positron emission tomography (PET) to follow tumor progression [38]. Few studies have, however, described mapping of choline in rat brains. The first image for choline distribution in a rat brain used HRMS to resolve the isobaric ions from protonated GABA and choline so as to produce selective MSI but showed very diffuse images from a coronal rat brain section with a slight prevalence of choline in the basal region [17]. Sagittal MSI of choline was further collected in two additional MSI studies [14, 39], but a discussion about the anatomical distribution of this metabolite in the brain sections was not provided. The MSI for the sagittal section of choline of m/z 104.107 in Figure 3d shows that its concentration is greater in the cortex, striatum, thalamus, midbrain, and cerebellum. In agreement with these findings, the coronal section in Figure 3i shows a high concentration of choline in the cortex and striatum.

Dopamine was also imaged (Figure 3e and j). This neurotransmitter was not observed in the sagittal rat brain sections (Figure 3e), but the coronal section showed dopamine mostly concentrated in the striatum, in agreement with previous reports (Figure 3j) [13, 15, 16, 40, 41].

Ionic suppression plays an important but not exclusive role in the intensity of the images generated. Choline ions produce the most intense images in the sagittal sections compared with the other images presented in Figure 3. In the mixture of standards, the cationic choline produces, however, an ion that is approximately one-third as abundant as that from the cationic acetylcholine (Figure 1a). The intense image of choline (Figure 3d) is related to the abundance of this cationic species in the rat brain and also to post-mortem changes, including hydrolyses of acetylcholine to produce choline, during the dissection procedure [11, 33].

The amino acidic neurotransmitters glutamate, aspartate, and serine were imaged via DESI in the negative ion mode (Figure 4). Again, the images of the ions of m/z 654.567 and 888.600 were used as references. For glutamic acid of m/z 146.046, the sagittal view indicates the prevalence of this neurotransmitter in the cortex, thalamus, and cerebellum, but it can also be observed in the striatum (Figure 4b). This image is in good agreement with the coronal view (Figure 4f), which presents a greater concentration of this neurotransmitter in the cortex and in the striatum. Few MSI studies for rat brain slices monitoring this neurotransmitter have been reported. Using 2,4-diphenyl-pyranylium tetrafluoroborate (DPP-TFB) as a derivatization reagent, MALDI-MSI showed a more prominent presence of glutamate in the cortex and striatum in rat brain coronal sections [15]. A similar profile was observed in sagittal and coronal rat brain sections using not only DPP-TFB but also p-N,N,N-trimethylammonioanilyl N-hydroxysuccinimidyl carbamate iodide (TAHS) and 4-hydroxy-3-methoxycinnamaldehyde (CA) as derivatization agents and MALDI-MSI [16]. The results from both of these MSI studies are in good agreement with the images reported herein.

Figure 4
figure 4

(−)-DESI-MSI of neurotransmitters and lipids in sagittal (a–d) and coronal (e–h) rat brain sections. Relative abundance and spatial distribution of m/z 654.567 and 888.600 (a) and (e), respectively; STR = striatum, TH = thalamus, MB = midbrain, CC = corpus callosum, PAL = pallidum; glutamic acid (b) and (f); aspartic acid (c) and (g); serine (d) and (h). Scale bar: 5 mm

Full size image

In the literature, MSI of aspartic acid and serine are also scarce. Figure 4c shows the sagittal view of aspartate via the ion of m/z 132.029, which is more abundant in the thalamus, midbrain, and cerebellum. The coronal section in Figure 4g shows prevalence of aspartate in the region of the pallidum from the striatum. MALDI images of aspartic acid obtained after derivatization with TAHS and CA have been reported, and the results are in good agreement with our DESI-MSI data [16]. Serine (m/z 104.035) was found to be more abundant in the cortex and corpus callosum of the sagittal sections (Figure 4d), and this distribution matches what is observed in the coronal section (Figure 4h). Serine derivatized with TAHS was found in the striatum in the coronal section and in the hypothalamus, thalamus, and cortex in the sagittal section [16].

Conclusion

We have shown that the more direct and simpler ambient DESI-MSI technique is also able to reveal the spatial distribution of neurotransmitters in rat brain slices. Clear and well-resolved images were obtained, whereas the use of a high-resolution mass spectrometer was shown to be essential in order to address isobars and to collect selective images. DESI-MSI can, therefore, be incorporated into neuroscience investigations for the spatial screening of neurotransmitters such as, for instance, in cases in which the abundance and/or distributions of these important biomolecules are expected to change.

When the present study was in the final stage of preparation, Bergman et al. [42] reported the absolute quantitation of the neurotransmitters acetylcholine, GABA, and glutamate in rat brain coronal slices by another ambient, matrix-free, and desorption-based ionization technique named nano-DESI-MSI. Our study corroborates and extends these findings, describing a detailed spatial distribution of the above-mentioned neurotransmitters as well as of aspartate, serine, and dopamine in coronal and sagittal rat brain slices. As we have anticipated [23], these results enlarge the application of atmospheric pressure ionization techniques to the field of neuroscience.