High Resolution Mass Spectrometry of Cystine-Containing Neuropeptides in Histological Sections of Human FFPE Tissue Banks

Abstract

Using our earlier developed protocol, mass spectrometry imaging of small endogenous peptides (and a selection of small metabolites) can be successfully performed directly in tissue sections of formaldehyde-fixed paraffin-embedded (FFPE) samples, such as those available in Homo sapiens biobanks. In analogy with immunohistochemistry (IHC) which employs antibodies as detection probes, this method was designated mass spectrometry histochemistry (MSHC) as it solely relies on (top-down) mass spectrometry for analyte detection. We demonstrate that MSHC enables the localization of cystine-containing neuropeptides in histological sections of human FFPE biobanked tissue and illustrate this on pituitary adenomas and non-diseased pituitary tissues archived for several years in an academic hospital pathology biobank. The instrumental setup consists of high-resolution mass spectrometers (several orbitrap systems and one dedicated hybrid TOF instrument) fitted with atmospheric pressure (AP) scanning matrix-assisted laser desorption/ionization MALDI. Currently, the best spatial resolution routinely achievable with such (MALDI) apparatus is 5 μm. The high mass spectrometric resolution obtained allows revealing the full isotope envelopes of the peptides. As such both reduced and oxidized cysteine-containing ‘proteoforms’ of e.g., the neurosecretory nonapeptides vasopressin and oxytocin can be visualized in biobanked FFPE tissue, demonstrating yet a novel application of MSHC.

1. Introduction

In a chapter of a book dedicated to the amino acid cysteine (Cys), it is probably redundant to emphasize the importance of this amino acid residue in protein/peptide biology. Yet, we would like to highlight one of the main reasons for our long-term interest in (endogenous) Cys-containing peptides. These peptides, often encoded by polypeptide precursor genes, are representatives of an extremely important class of secretory biomolecules, which fulfill crucial roles in intercellular communication. As such their functions in the organism belong to the (if not the) most essential and fundamental characteristics of ‘life’ [1, 2]. More than 40 years back, when the senior author of this chapter (P.V.) engaged in secretory peptide research, the most elegant way of demonstrating (and localizing) secretory biomolecules in tissues was by conducting immunohistochemistry-based experiments (see e.g. [3, 4]). Molecular characterization of newly discovered secretory peptides initially required laborious (if not heroic) extractions and chromatographic separations followed by tedious amino acid sequencing by automated Edman degradation (e.g. [5, 6]). With the advent of protein/peptide (tandem) mass spectrometry (MS), the identification and molecular characterization of secretory peptides became much more efficient (e.g. [7]), especially when it became evident that direct tissue matrix-assisted laser desorption/ionization (MALDI) MS enabled the immediate identification of secretory (neuro)peptides (e.g. [8, 9]). Several years later, the technology had developed thus far that scanning MALDI MS today allows in situ, i.e., on-tissue, analysis of secretory peptides, directly on histological sections of biological tissues, including samples that have been archived in tissue banks for a number of decades [10]. In this chapter, we illustrate the type of data that can be obtained from two different Cys-containing neuropeptides by modern high-resolution (HR) MS coupled to the latest atmospheric pressure scanning microprobe MALDI sources on well-documented samples from a hospital tissue bank.

We like to underline at this point that the sample histological sections we employed in this study all originated from formaldehyde-fixed paraffin-embedded tissue blocks that had been stored at ambient temperature in a hospital tissue bank for more than a few years. This remarkably contradicts several reports on MSI studies claiming that, without specific antigen retrieval steps (with their inherent risk of analyte delocalization and signal intensity reduction) and/or the use of special ‘reactive’ matrices for on-tissue chemical derivatization, top-down MS analysis of FFPE is destined to fail [11, 12]. Indeed, it has been suggested that formaldehyde fixation should preferably be avoided for MSI analyses [13].

2. Materials and methods

2.1 Samples

2.1.1 Sample origin and conservation

Homo sapiens pituitary-derived tissues were carefully selected from the large pathological collection at Leuven University Hospital FFPE biobank. The tissues were surgical specimens from patients who had undergone (anterior) pituitary tumor resection. All tissues had been fixed in formaldehyde and embedded in paraffin according to the hospital’s standard operating procedure. Special attention was made to select tissue blocks that partly contained posterior pituitary nerve terminals, known to store the nonapeptides vasopressin and oxytocin, as validated by histochemical stains, including hematoxylin/eosin (H&E) and immunohistochemical (IHC) staining with nonapeptide antibodies (see below).

2.1.2 Sample processing

2.1.2.1 Sectioning

Paraffin blocks containing the tissue samples were trimmed and microtome sectioned at 5 μm thickness. For use in the orbitrap MS systems, the tissue sections were collected on regular [2.5 (or 2.6) × 7.5 cm] glass microscope slides and stored at room temperature until deparaffinization. For use in the TOF system, sections were mounted on conductive (indium tin oxide coated) slides.

2.1.2.2 Deparaffinization

Deparaffinization was performed essentially as described [1], the glass slides were sequentially immersed in a brief series of solvents, i.e., 2 times in xylene (100%) for 2 and 1 min respectively and two times in ethanol (absolute) for 1 min each. They were left to dry at room temperature for a minimum of 30 min before MALDI matrix application.

2.1.2.3 MALDI matrix coating

Before MALDI MS imaging, tissues were coated with MALDI matrix using automated pneumatic spraying or sublimation. For spray-coating, a solution of 2,5-dihydroxybenzoic acid (DHB, 50 mg/ml) was prepared in 50% acetonitrile, 50% H₂O containing 0.1% TFA. Two different spraying devices were employed yielding equal-quality matrix depositions. The sublimation instrument (see below) was operated using DHB powder (>99.0% pure) right out of the vial (Sigma-Aldrich #85707-1G-F).

2.1.2.3.1 M5 sprayer (HTX)

M5 sprayer settings were as follows: nozzle temperature: 75°C; nozzle height: 40 mm; flow rate: 0.1 ml/min; velocity: 1200 mm/min; track spacing: 1 mm; 3 passes; pressure: 10 psi; gas flow rate: 3 l/min; drying time: 2 sec.

2.1.2.3.2 Suncollect sprayer (Sunchrom)

Suncollect sprayer settings were as follows: z-distance: 25 mm; flow rate: 0.1 ml/min; velocity: 1200 mm/min; track spacing: 2 mm; 3 passes; pressure: 2.5 bar.

2.1.2.3.3 iMLayer matrix sublimation system (Shimadzu)

DHB was applied through an automated system as described [14] with sublimation resulting in a 2 μm matrix thickness (thinness) layer. To minimize matrix crystal size and assure analyte incorporation into the matrix crystals, a quick recrystallization step was performed (1.5 min) in a closed chamber at 75°C with a paper tissue with 500 μL MeOH:H₂O 5:1000.

2.2 Mass spectrometry

2.2.1 Ionization: MALDI

Different scanning atmospheric pressure (AP) MALDI sources were employed, depending on the mass spectrometer used.

2.2.1.1 AP-SMALDI

A Q-Exactive HF hybrid quadrupole/orbitrap mass spectrometer (ThermoFisher Scientific (Bremen) GmbH, Bremen, Germany) was connected to an AP-SMALDI⁵ AF high-resolution MALDI imaging source (TransMIT GmbH, Giessen, Germany). The laser focus diameter on the sample was 5 μm and the sample was scanned with a 7 μm step size, resulting in an image pixel size of 5 μm × 7 μm.

2.2.1.2 AP/MALDI

Different generation orbitrap (ThermoFisher Scientific (Bremen, Germany and San Jose, CA USA) instruments were used in connection with an AP/MALDI (ng) UHR ion source (MassTech, Columbia, MD, USA). These comprised ion trap/orbitrap hybrid systems, including an LTQ Orbitrap Velos and an LTQ Orbitrap Velos Pro Elite; the single stage Exactive Plus, and the latest generation quadrupole/orbitrap Exploris 480. Laser energy was typical between around 5% (Nd-YAG laser), and laser focus was set below 10 μm for the Velos, Elite, and Exactive systems, to allow for a 10 × 10 μm² pixel size to be sampled in constant speed raster mode. The sample stage movement was programmed in the Target control software (MassTech) to be synchronized with the MS scan rate, to allow for, at least, 1 MS scan per pixel to be recorded.

2.2.1.3 Integrated AP MALDI

An iMScope TRIO (Shimadzu) ion-trap time-of-flight (IT-TOF) mass spectrometer Shimadzu Corporation (Kyoto, Japan) was also used, essentially as described before [14]. This instrument has a fully integrated atmospheric pressure MALDI source. Laser focus was <5 μm; laser intensity was set at 16.0 a.u., laser frequency was 1000 Hz and 200 shots were accumulated per pixel. For data acquisition, Imaging MS Solution 1.1 was used.

2.2.2 MS acquisition to imzML data format

2.2.2.1 QExactive HF

Mass spectra were acquired in positive-ion mode within a mass range of 500 to 2000 m/z and at mass resolution of 240,000 at 200 m/z. Internal lock-mass calibration on a DHB matrix cluster ion assured a mass accuracy of less than 2 ppm (root mean square error). Ion injection time was set to 500 ms, the S-lens level was maintained at 100 arbitrary units, and the ion transfer capillary temperature was 250°C. Raw MS data files (*.raw) and corresponding position files (*.udp) from the AP-SMALDI source controller software were obtained as inputs for subsequent processing. The latest imzML converter ((c) University of Giessen and TransMIT GmbH) was employed for efficient data conversion. The converter utilizes an advanced algorithm transforming initial profile data into centroid data, reducing data dimensionality without loss of essential molecular information. Critical image dimension parameters including number of pixels (in x and y) and pixel size are automatically selected from the files, enhancing speed and accuracy of data processing, by minimizing manual intervention. Upon conversion process, the raw MS data are transformed into the imzML format. This generic format assures flexible data accessibility and facilitates downstream analysis, providing a foundation for robust interpretation and visualization of spatially-resolved molecular information [15].

2.2.2.2 Exactive Plus

The Exactive Plus orbitrap FT-MS system (ThermoFisher Scientific) was operated in positive-ion mode in a mass range of 150–1500 m/z at a mass resolution of 140,000. Ion transfer capillary temperature was set at 375°C.

2.2.2.3 Exploris 480

The Exploris 480 FT-MS system (ThermoFisher Scientific) was operated in positive-ion mode in a mass range of 400–1500 m/z at mass resolution of 60,000. Ion transfer capillary temperature was set at 375°C. Laser at 20% energy, 100 Hz. Max injection time 200 ms. AGC was on.

2.2.2.4 LTQ orbitrap Velos systems

We previously reported data obtained on early generation LTQ Orbitrap Velos PRO Elite and LTQ Orbitrap Velos systems, which were able to detect the secretory neuropeptides in question [16, 17, 18, 19], but in this chapter we will concentrate on the younger ThermoFisher MS systems mentioned above.

2.2.2.5 iMScope TRIO

The iMScope TRIO IT-TOF mass spectrometer (Shimadzu) was operated in positive-ion mode in a mass range of 550–1500 m/z. Sample voltage was set at 3.5 kV; 200 laser shots were accumulated per pixel (laser frequency 1000 Hz).

To try and confirm the neuropeptide identity of the MSHC detected ion signals, tissue collision-induced dissociation (CID) MS/MS analyses were performed using m/z 1084.445 and 1007.443 ± 1.000 amu as precursor ion setting and adjusting the mass range to resp. 300–1100 m/z and 250–1100 m/z.

Spectra were processed using the Imaging Mass Solution software (v.1.1; Shimadzu).

2.3 Histochemistry

Two types of histochemical staining methods were used for light microscopic analysis of the tissue sections. These included classical H&E staining as well as immunohistochemistry (IHC).

2.3.1 Hematoxylin/eosin (H&E) staining

To facilitate histological recognition of the different cell and tissue types, an adjacent (serial) section of each paraffin block was stained with hematoxylin and eosin. For this the respective microscope slide containing a deparaffinized section was immersed in 70% ethanol (5 min), followed by a brief dip/wash in distilled water. Slides were then stained in Harris hematoxylin solution for 8 min, after which they were washed in running tap water for 5 min. Subsequently, sections were counterstained in eosin-phloxine solution for 1 min, followed by conventional dehydration and cover-slipping; dehydration through 70% and 96% ethanol (2 changes of 5 min each), clearing steps in 2 changes of xylene (5 min each), and mounting with xylene based mounting medium.

2.3.2 Immunohistochemical (IHC) staining

Two different rabbit polyclonal antisera were used. Neuropeptide immunohistochemistry was performed using rabbit anti-vasopressin and anti-oxytocin polyclonal antibodies (a kind gift by Prof. L. Arckens; Laboratory for Neuroplasticity and Neuroproteomics, University of Leuven, Belgium). Both antisera had been used before in immunohistochemical studies to demonstrate vasopressin- and oxytocin-like immunoreactive peptides in neuronal tissues of various animal species [3]. Details of the immunohistochemical technique employed and antibody incubation parameters were described earlier [3, 4].

3. Results

To illustrate the specificity of our HRMS1 analyses, we can compare the experimentally obtained spectra with the calculated (theoretic) isotopic distribution of (a) vasopressin and (b) oxytocin respectively. Figure 1 shows these both in oxidized (cystine, closed S∙S ring) and reduced state (open disulfide ring; two free thiol S∙H groups).

Figure 1.

Calculated (theoretical) isotopic distribution of vasopressin (a) and oxytocin (b). Note that, at observed m/z range, highest isotope peaks are invariably those with all C atoms as C12, followed by ions with one C13 atom, and, thirdly, ions with two C13 atoms in their composition.

In the majority of the pixels sampled no neuropeptide ion signals are detected, which is in line with the known tissue distribution of the 2 nonapeptides focused on in this paper. Depending on the position where the tissue mass spectra are acquired, different (combinations of) ion signals are detected within the determined neuropeptide mass range. Some typical examples are shown in Figure 2. These comprise pixels (Figure 2a,d and e) that yield more or less complete neuropeptide ion isotope patterns; whereas in other pixels merely the most intense (monoisotopic) peaks are evident, such as for the (lower intensity) oxytocin ions (Figure 2b and c).

Figure 2.

Typical examples of on-tissue detected single-pixel isotopic patterns of vasopressin and oxytocin [M + H]⁺ ions (Exactive Plus data), showing different spectra depending on peptide quantities/signal intensities obtained. To facilitate comparison, insert in Figure 2d shows a theoretic isotope envelope of oxytocin with its cystine disulfide bridge formed.

Using AP-SMALDI⁵ AF—QE HF MSHC images were obtained at 7 μm lateral resolution. The distribution of vasopressin and oxytocin-related peptide ions is partly overlapping, but differential (Figure 3 lower panel). Immunohistochemical staining demonstrated that vasopressin- as well as oxytocin-immunoreactive substances, are detected with a similar tissue distribution to the MSHC detected ions (Figure 3 upper panel).

Figure 3.

AP-SMALDI⁵ AF—QE HF MSHC images (7 × 7 μm² pixel size) generated in METASPACE (lower panel). Lower left image is an MSHC overlay of pixels containing protonated vasopressin (blue), protonated & sodiated oxytocin (resp. magenta & red), and protonated protoporphyrin (PP; [C₃₄H₃₄N₄O₄ + H]⁺; precursor of heme; green). Upper panel shows serial sections of same paraffin block stained with H&E and IHC. Legend: H&E, hematoxylin/eosin; IHC, immunohistochemistry; MSHC, mass spectrometry histochemistry; OT, oxytocin; VP, vasopressin. Center lines show primary structures of VP (middle) and OT (right), with only 2/9 amino acid residues different.

Using the AP/MALDI—Exploris 480 MSHC combination, images could be acquired at 5 μm lateral resolution on similar FFPE tissue from pituitary adenoma resection (Figure 4).

Figure 4.

METASPACE-generated MSHC image (left) obtained of biobanked human pituitary FFPE tissue with AP/MALDI 480 (5 × 5 μm² pixel size), with serial sections of the same paraffin block stained with anti-vasopressin antibodies (middle; VP-IHC) and hematoxylin/eosin (right; H&E). Abbreviations: ah, adenohypohysis; nh, neurohypophysis; H, heme b (m/z 616.16); OT, oxytocin (m/z 1007.44); VP, vasopressin (m/z 1084.44). Yellow pixels represent specific DHB matrix cluster ions, which appear to form predominantly outside of tissue.

Equally, ions representing both peptides are detected when the AP/MALDI source was mounted on earlier generation orbitrap systems, including the Exactive Plus (Figure 2, MSHC image not shown), and the 15 years-old LTQ Velos (previously published MSHC images [16, 17]).

Finally, also using the fully integrated iMScope TRIO can be used to generate images in the same spatial resolution range, as shown in Figure 5. The sample analyzed here was from a different pituitary adenoma patient than the one who donated the sample imaged in Figures 3 and 4. Immunohistochemistry of adjacent sections to the one imaged by MSHC yielded the expected images with no obvious distinction between the anti-vasopressin and anti-oxytocin staining (Figure 5 upper panel).

Figure 5.

Lower panel shows MSHC images (10 × 10 μm² pixel size) of resected human pituitary adenoma tissue archived in FFPE biobank, obtained after analysis with iMScope TRIO and generated using imaging mass solution software (v.1.1; Shimadzu). The numbers below represent selected m/z values of ions imaged, with far-left image being an overlay image of vasopressin ions in green and oxytocin ions in red. Numbers at the y-axis indicate ion intensity color scales. Upper panel left shows very same tissue section prior to MSHC analysis (and before DHB matrix application). Middle and right images are two serial adjacent sections immunostained with anti-vasopressin and anti-oxytocin antibodies. Legend: ah, adenohypophysis; nh, neurohypophysis; IHC, immunohistochemistry; OT, oxytocin; VP, vasopressin.

CID tandem MS data were obtained both from the precursor ion at m/z 1084.445 and of that at m/z 1007.443 (Figure 6).

Figure 6.

On-tissue CID MS/MS spectra of precursor ions at m/z 1084.445 (a) and m/z 1007.443 (b). Respective inserts show vasopressin (a) and oxytocin (b) sequence with predicted fragment ion masses (online ProteinProspector MS Product output [https://prospector.ucsf.edu/prospector/cgi-bin/msform.cgi?form=msproduct]). Most fragile peptide bond (aminoterminally from Pro residue, carboxyterminally from disulfide bridge-linked cystine ring) is marked in green, yielding, for both nonapeptides, b6 and y3 as most dominant ions in the tandem MS spectrum.

4. Discussion

As stated in our previous publication [18], many biologically active secretory peptides contain one or multiple disulfide bridges, in part due to/their contribution to the peptides’ stability and activity. We concluded that the presence of Cys can be used as selection criterion for biologically important biomolecules in animal secretions. We calculated the prevalence of cysteines (6.91% of all residues) in the animal toxin sub-database of UniProt relative to the overall occurrence of cysteines in the whole UniProtKB/Swiss-Prot (release 2015_1) database (1.37% of all residues). We have recalculated these numbers based on the current versions of the respective databases (Figure 7). For this calculation, the latest version of the UniProt database (release 2024_1) was downloaded from https://www.uniprot.org/uniprotkb/statistics. The Tox-Prot database was derived from this using the following query: “taxonomy_id:33208) AND ((cc_tissue_specificity:venom) OR (keyword:KW-0800)) AND (reviewed:true)” as described at https://www.uniprot.org/help/Toxins. We cannot but conclude that our previous observation [18] still stands, with 6.66% of all residues being Cys in Tox-prot, whereas only 1.39% of all residues are Cys in UniProt.

Figure 7.

Bar graphs illustrating importance of Cys as residue in secreted, bioactive proteins. (a) Upper panel compares relative amino acid abundance (in %, y-axis) in UniProt (red bars) with that in tox-Prot database (blue bars) as of January 2024 (a). Note especially the relatively high abundance of the cysteine residues in the Tox-Prot database (6.66%) compared to the Uniprot database (1.39%). (b) Lower panel shows an alternative representation of Cys importance by plotting relative amino acid abundance increases in Tox-Prot database compared to Uniprot database, calculated by (%Tox-Prot/%Uniprot) × 100%. Generally speaking, relative amino acid abundances are comparable between both databases, except for Cys which exhibits an almost 500% increase in relative abundance.

Since this publication, we have continued our work on yet another dimension to investigate endogenous peptides (Cys-containing as well as other secretory) in complex biological samples. We refer to the spatial dimension, and when applying mass spectrometry imaging technology directly to histological sections, one could refer to this approach as mass spectrometry histochemistry (MSHC). Whereas we originally studied fresh samples, minutes after prelevation/resection from the living organism [9, 20], in the past 7 years we have been focusing mainly on the type of material that is more easily (and ethically) accessible, i.e. formaldehyde-fixed, paraffin-embedded (FFPE) samples [10, 16, 17, 21].

We here focus on the data obtained for 2 known Cys-containing neurosecretory nonapeptides (vasopressin and oxytocin) which can easily be detected in Homo sapiens biobanked neuronal tissues [16, 17, 21].

The chemical structure of both vasopressin and oxytocin with their disulfide bridge formed is shown in Figure 8.

Figure 8.

Homo sapiens vasopressin (VP, left) and oxytocin (OI, right) in their predominant (proteo)forms, i.e., a carboxyterminally amidated nonapeptide with two cysteines at residues 1 and 6 linked by a disulfide bridge. Note: primary structure of vasopressin is CYFQNCPRG-NH₂ [elemental composition C₄₆H₆₅N₁₅O₁₂S₂], and of oxytocin is CYIQNCPLG-NH₂ [elemental composition C₄₃H₆₆N₁₂O₁₂S₂].

In this paper, we focus on MS1 data, which, on their own, obviously are debatably sufficient to validate the peptide identifications. However, we previously published that, using synthetic peptide analogs (both oxidized as well as reduced (open disulfide bridge) versions, tandem MS as well as ion mobility fully confirm, at least for vasopressin, our peptide identifications (see [21])).

Depending on the position of the tissue where mass spectra are acquired, different (combinations of) ion signals in the zoomed-in neuropeptide mass range are observed in the respective pixels, with typical examples shown in Figure 2. It shall be clear that not in all pixels, full isotope profiles are evident (Figure 2b and c). Indeed, it can be expected that in certain pixels some peptides may not be present at concentrations sufficient for the second (let alone third) isotopes to exceed the limit of detection. The fact, however, that the monoisotopic peaks detected exhibit the expected mass defect, and share the typical tissue distribution with the pixels that do contain enough neuropeptide ion signal for multiple isotopes to be observed, strengthens our interpretation that, also in these single isotope peak spectra true neuropeptide ions (albeit in very low quantities) are detected. This is also in line with the well-known very low abundance of typical signaling (secretory) peptides in tissues.

It follows as well that not all pixels may generate enough ion signals to allow for MS/MS identification/confirmation of the peptide primary structure. Yet when selecting areas with the highest MS signal for the respective precursor m/z, sufficient tandem MS information can be deduced from fragmentation spectra to allow for unambiguous annotation of the spectra to the respective neuropeptide. MALDI fragmentation spectra generated with the iMScope TRIO IT-TOF system are shown in Figure 6. These fully agree with the earlier described fragmentation spectra, when, in addition to the typical MS1 characteristics of typical endogenous peptide ions [16], we also used ion mobility to confirm the identity of human pituitary nonapeptides extracted from a single FFPE tissue section [21].

Even using one of the earliest generation ion trap-orbitrap combinations (our LTQ Orbitrap Velos, first operated in 2010, being the oldest HRMS instrument tested), vasopressin as well as oxytocin ions can be detected [16, 17]. However, signal intensities seen with the later generation orbitraps [this chapter] are markedly higher, suggesting that, in the older-generation instruments, the low abundant endogenous peptide analyses appear close to the limit of detection. In our view, this also explains why older-generation instruments may lack the sensitivity to reveal the complete isotopic envelope of a detected peptide. This then yields ambiguous data, which may not be conclusive, in particular when untargeted biomolecular discovery is envisaged, such as for novel biomarker identifications. We, therefore, recommend employing an HR mass spectrometer with the highest sensitivity possible for such analyses. We remark here that present-generation hybrid TOF instruments, such as the performant iMScope TRIO apparatus are also able to yield sufficient resolution as well as sensitivity to allow for MSHC detection/discovery of neuropeptides and/or neuropeptide proteoforms, even including on-tissue CID tandem MS confirmation of neuropeptide primary structures (Figure 6). This suggests that also its current successor, the iMScope QT, based on a QTOF mass spectrometer, may be well suited for this.

It is well-known that, in traversing the endoplasmic reticulum and Golgi apparatus, secretory proteins/peptides undergo a variety of posttranslational modifications (PTMs), with disulfide bond formation often being a PTM necessary for protein/peptide (secondary) structure and/or enhancing the peptide’s half-life by hindering protease (such as trypsin) digestion. It follows that a relatively high proportion of bioactive secretory peptides share the attribute of having one or more Cys-residues in their primary structure. We have used this feature in the past to filter bioactive peptide candidates from complex LC/MS datasets of animal secretions [18]. The conventional method of determining the presence of disulfide bonds in a peptide is by simply counting the mass shift before and after reduction [22] or after reduction and alkylation [23]. However, such an approach requires the comparison of two separate analyses, which is not always feasible or desirable. To quickly assign the presence of cysteines in a peptide of unknown nature without the need for disulfide bond reduction, we, therefore, developed an HRMS filtering approach [18], in which we employ two intrinsic characteristics of the sulfur atom to select peptides with cysteine residues. The relatively large negative mass defect (the difference between the isotopic and nominal or integer mass) in combination with the positive isotopic shift (difference between average and monoisotopic mass) [24] allows for rapid identification of sulfur-containing peptide candidates from LCMS datasets of a complex mixture of unknown Cys-containing samples. Normalization of these two shifts results in two non-additive and independent peptide properties, which had previously been introduced as normalized nominal mass defect (NMD) and normalized isotopic shift (NIS), both of which have been used as a data representation method after de novo sequencing [25].

In the present chapter, we show that the detection and localization of Cys-containing secretory peptides can be done directly on the very same histological section as prepared by the histopathologist for H&E and/or IHC stains. The extremely high mass accuracy obtained by the various HR (Fourier transform, FT) MS instruments employed in these experiments allows for direct elemental composition calculation. The results generated are sufficient for database identification, such as those provided by the online automated mass spectrometry imaging data annotation platform METASPACE (https://metaspace2020.eu/ [26]).

Comparing the on-tissue isotope distributions of the neuropeptide-related ions with the theoretical calculations, we observe a-typical patterns. Indeed, in a significant number of pixels, the 3rd (and 4th) isotopic peak of vasopressin appears higher than expected (Figure 2a). We previously substantiated, using orthogonal validation with synthetic peptide analogs, tandem MS and ion mobility, that this is due to the presence of reduced neuropeptide, i.e. vasopressin with its [Cys1-Cys6] disulfide bridge opened [21]. In this chapter we show (single pixel) on-tissue MS spectra suggesting that a similar phenomenon is evident for oxytocin (Figure 2c and d). One biological assumption can be that we are observing the (in Golgi?) maturation of the neuropeptides, where in the process of peptide synthesis first the linear version of the peptide and its neurophysin are produced after which they are fully post-translationally processed, by C-terminal amidation and cystine formation. As it can be expected that the linear versions of both nonapeptides exhibit different binding characteristics to their respective neuropeptide receptors, the fact that we can now establish which neuropeptide proteoform is present in which tissue (compartment) may prove to be biologically relevant.

We also emphasize here that, although the tissues have been fixed in formaldehyde and stored in paraffin for many years (some were over a decade old), remarkably little formaldehyde fixation modifications are evident in our datasets. Only the pixels with the highest vasopressin concentration show signals at m/z 1096.44, being the Schiff base ion [M + C + H]⁺ [particularly detected in the Exploris and Exactive series mass spectrometers; data not shown], a well-known intermediate reaction product of the formaldehyde fixation [27].

In general, we show here that MSHC results can be obtained on a wide variety of orbitrap systems (ThermoFisher Scientific), but also the fully integrated MS imaging IT-TOF instrument (Shimadzu). In combination with our previously published work [10, 21] which also employed Bruker mass spectrometers, this exemplifies that MSHC is full ‘vendor-neutral’, and that the unique sample preparation method is key to retaining the respective analytes of interest in their tissue location and thus allow for successful MSHC experiments.

Finally, we highlight that in terms of spatial resolution achievable with the different AP MALDI sources employed and using the very MSHC sample preparation described here, it was possible to generate human neuropeptide ion images from pixels as small as 5 × 5 μm² (of a 5 μm thin FFPE tissue section).

5. Conclusions

We conclusively demonstrate that MSHC allows for different neuropeptides to be imaged on 5 μm thick histological sections of year-long biobanked FFPE tissue. MSHC in addition allows for the visualization of the tissue localization of certain classes of metabolites (such as protoporphyrin and related biomolecules) as well. (These will be extensively reported elsewhere). The MSHC analyses yield sufficient resolution to show the on-tissue presence of both oxidized (cystine-containing) and reduced (separate cysteine residues) neuropeptides.

Whereas FFPE tissue is traditionally not considered, let alone recommended for MSI applications in pathology, using this combination of MSHC sample preparation and atmospheric pressure MALDI analysis, we demonstrate the value and potential of analyzing easily accessible FFPE biobanked tissue. MSHC analyses, therefore, have great promise to help understand pathological processes involving peptides, which today can only be studied with microscopic histo(patho)logy. Hence we can paraphrase the conclusion of Ref. [28] that MSHC as described here adds to the promise towards integration of MSI in routine molecular pathology for clinical diagnostics.

Acknowledgments

We would like to express our gratitude to Dr. Kerstin Strupat (ThermoFisher Scientific (Bremen) GmbH) for her continued support in bringing MALDI imaging and orbitrap mass spectrometry together. Prof. Dr. Uwe Karst is gratefully acknowledged for providing us access to his iMScope TRIO at the University of Münster (Germany). We are indebted to Dr. Venkat Panchagnula (MassTech, US) and Dr. Charles C. Liu (ASPEC Technologies, China) for their continued support and enthusiasm. We would also like to thank Anatech (South Africa) for their technical assistance and support.

PV would like to dedicate this book chapter to the memory of Dr. Martin Robert Lloyd Paine, who sadly passed away due to colon cancer in the middle of the COVID pandemic in May 2021. Marty, you will never be forgotten!