Development of tissue-imaging mass spectrometry approach using an atmospheric pressure MALDI source

(1)

1

MSc Chemistry

Analytical Science Track

Method development and optimization

of a tissue imaging mass spectrometry (MSI) approach

using an atmospheric pressure matrix assisted laser

desorption ionization source (AP-MALDI)

Master Thesis by Helene Fiedler

Student number: UvA 11997060

VU 2634385

Supervisor & 1

st

_Examiner

_{prof. dr. Pim Leonards}

2

nd

_Examiner

_{prof. dr. Marja Lamoree}

2

nd

_Supervisor

_{dr. Erika R. Amstalden}

Vrije Universiteit Amsterdam

Department of Environment & Health

Universiteit van Amsterdam

May 2020

(2)

2

Abstract

In environmental research analytical techniques play a key role in understanding biological structures, metabolic mechanisms, and the biodistribution of chemicals. Small endogenous and exogeneous molecules are studied in situ in cells, tissues or organisms to provide functional

information on the physiological state of an organism and their reaction to e.g. environmental stress. Lipids are ubiquitous throughout organisms and highly abundant in many cell types. They fulfil essential biological functions such as energy storage, structural composition and cell signalling. Understanding their involvement in e.g. neurodegenerative disorders, cancer, cardiovascular and many other diseases, might reveal more detailed insight for better understanding and improved treatment methods. Initial techniques, such as LC-MS (liquid chromatography mass spectrometry) or GC-MS (gas chromatography mass spectrometry) were used to evaluate the metabolites in

homogenized samples, but are incapable of providing information on the location of single

compounds within tissues. Mass spectrometry imaging (MSI) is a powerful tool to correlate spectral information and molecular profiles with anatomical features, organism’s functioning, phenotype, and chemical exposure by visualizing the spatial distribution of compounds within biological samples. For targeted imaging of certain molecular entities various ionization techniques can be used that are suited to different applications. In this study an atmospheric pressure matrix assisted laser

desorption ionization (AP-MALDI) source was used and coupled with a high resolution QqToF-MS (quadrupole- time of flight- mass spectrometry) system. Important instrumental parameters of the ion source and the mass spectrometer were optimised using an experimental design approach for the detection of lipids, focusing on the mass range between 600-900 Da. Also the influence of AP-MALDI parameters, including the laser alignment, source settings, laser frequency, and laser power were evaluated. It was found that, next to the alignment of the ion source and proper sample preparation, the source parameters capillary voltage and dry gas flow rate have the greatest impact on the signal intensity of the lipid standards. Best detection, in positive as well as negative ionisation mode, was achieved at 3000-3500 V applied on the capillary in combination with a low dry gas flow rate around 2 l/min. The laser fluence was adjusted to 45 %- 55 % laser power (depending on the matrix used) and a laser frequency between 1000 - 2000 Hz to avoid high background and baseline noise.

The developed method was then applied on tissue, analysing slices of zebrafish with three different matrices, α-Cyano-4-hydroxycinnamic acid (α-CHCA), 2,5-Dihydroxybenzoic acid (DHB) and 1,5-Diaminonaphthalene (DAN), in positive and negative ionisation mode. After appropriate data conversion, the measurements revealed molecular specific images, showing sharply separated biological features (e.g. the eye, heart, gills) and well resolved spatial characteristics within the tissue at pixel sizes from 100 µm down to 20 µm.

(3)

3

Abbreviations

QqTOF-MS quadrupole- time of flight- mass spectrometry

MSI mass spectrometry imaging

IS ion suppression

TIC total ion current

EIC extracted ion current

BP/BPC base peak, base peak current

cnts counts (absolute signal intensity in numbers) [c] concentration (molar concentration, M) isCID in source collision induced dissociation cap.V capillary voltage

DoE design of experiments

(6)

6

1. Introduction

After Dr. Richard Caprioli et al pioneered MALDI-MSI (matrix-assisted laser desorption/ionization MSI) for the visualisation of large biomolecules in cells and tissue in 1997 (Cobice et al., 2015), an advanced method was introduced appliying the fundamentals of MALDI in an ambient environment only three years later (Takats et al., 2018). Atmospheric pressure matrix-assisted laser

desorption/ionization (AP-MALDI) techniques rapidly developed and became a competitive alternative to traditional vacuum MALDI instruments in research oriented MSI studies.

(AP) MALDI MSI is a powerful tool for targeted or untargeted in situ analysis of molecular species in biological samples. It allows to correlate metabolic information to the histological and morphological feature in cells or tissue (Baker et al., 2017). Besides non laser-based methods like desorption electrospray ionization (DESI) and secondary ion mass spectrometry (SIMS) MSI and non-MS based techniques (NMR (nuclear magnetic resonance), CARS (coherent anti-Stokes Raman scattering) spectroscopy, and MRI and microscopy (Buchberger et al., 2018; Svatoš, 2010), matrix-assisted laser desorption/ionization (MALDI) MS is the most widespread scanning (microprobe) method for the visualization of chemical distributions and metabolic changes (Svatoš, 2010). Microprobe MSI uses a focused ionization beam, which for MALDI is either UV or IR laser radiation, to desorb, ablate and ionize molecules from a well-defined area (pixel or voxel). Therefore, samples are covered with a solvent-matrix mixture before analysis to support the extraction and ionisation of the target molecules, as will be covered in more detail in section 3.2.2.3. The spectral data of thousands of spots with defined spatial coordinates are assembled into a tissue characteristic map for certain molecules of interest. The signal of individual ions can be plotted in two-dimensional (2D) or three-dimensional (3D) graphics and visualized as typically pseudo-colour images (Svatoš, 2010).

Furthermore, the structural information of molecules can be annotated by using exact masses together with tandem MS (MS/MS).

The aim of this project is the development of a mass spectrometry imaging approach for the application on biological tissue using an AP MALDI source, focusing on the detection of lipids. Therefore, the ion source is coupled with a high resolution QqTOF MS instrument, which provides high selectivity, low picogram sensitivity, mass accuracy down to 1ppm (compact, 2020), high sampling speed and comes with the ability of performing tandem MS for further method

development and targeted compound identification. To the best of our knowledge, this arrangement of devices has not been reported yet. Recent studies mostly utilised the AP-MALDI source in

combination with ion mobility (Ryumin and Cramer, 2018) and different trap mass spectrometers, such as an orbitrap (Cao et al., 2018; Jackson et al., 2018) or hybrid instruments (Sussman et al., 2018). Due to the lack of extensive background knowledge existing methods of the similar MALDI source will be adjusted and re-evaluated for the measurements under ambient conditions. Additionally, special tools were developed and must be used to ensure optimal usage of the AP-MALDI source, such as the alignment of the device using the alignment needle. More detailed information on the AP-MALDI source and mass spectrometer is given in chapter 2. in order to understand the fundamental mechanisms, plan the optimisation procedure, and correctly evaluate the results.

To achieve the aim of the thesis, optimisation of the analysis of lipids with the AP-MALDI QqTOF-MS is carried out in several steps:

1. Experimental work on lipid standards to facilitate the optimisation of parameters, using the common dried droplet method. Special focus is laid on:

• The reproducibility of the results and generation of stable signals from analyte-matrix mixtures. This includes the evaluation of the sample preparation method and the source alignment as important contribution factors.

(7)

7

• Evaluation of different matrices for optimized measurements in both positive and negative ionisation mode.

• The adjustment of different instrumental parameters of the ionisation source and mass spectrometer. Especially laser settings and source parameters vastly influence the accuracy and sensitivity of the experimental results.

In analytical chemistry, using statistical tools is recommended for refined tuning of technical parameter and increases the efficiency of the optimisation procedure.

Experimental design (DoE, Design of experiment) approaches substantially facilitate the simultaneous evaluation of several parameter and their interaction effects and reduce the workload by condensing the number of measurements without losing informative value.

2. Application of the method with optimised settings for the detection of lipids on zebrafish tissue with increasing spatial resolution. Especially the zebrafish (Danio rerio) is currently a popular model vertebrate for metabolomic studies and an important subject in the research of the environment and health department at the VU Amsterdam in various projects. The AP-MALDI is expected to provide information on the spatial distribution of the endogeneous compounds and toxicants in zebrafish embryos and correlated metabolite changes, to better understand the molecular toxicity pathways.

3. (Imaging) software, data processing, evaluation of results and extraction of scientifically relevant knowledge.

This project is solely focusing on the analytical aspects of method development, optimisation, and application, including the comparison to MALDI MSI, and highlighting the advantages and limitations of measurements at atmospheric pressure. A detailed evaluation of the biological background and findings is not carried out. Nevertheless, the new method promises a wide range of application to all kinds of surfaces and insight into spatial dimensions of biological processes on a molecular level. Suggestions for further development and alternative methods, based on the findings of this project, are given later in section 6.

(8)

8

2. Theory of mass spectrometry imaging

MSI is a multi-step method that, after minimal but crucial sample preparation, proceeds with the desorption and ionization of the analyte molecules from a dried sample spot, tissue section or cell that was applied on a stainless steel or glass sample plate. This takes place in the external AP-MALDI ion source, that shares many fundamental aspects with a conventional vacuum MALDI, including the usage of different matrices as a key part of the method, as well as proportions of matrix-to-analyte, sample preparation procedures and the energy density of the laser beam at the target surface (Laiko et al., 2000). The small and compact ionisation source allows simple and fast interchange with other atmospheric pressure devices, like ESI and APCI sources (Sussman et al., 2018), and can be linked to all types of mass spectrometers that provide an interface for ambient sampling (Moskovets et al., 2016; Trap et al., 2017). Atmospheric pressure conditions facilitate the introduction and handling of samples and enable the analysis and utilisation of volatile compounds and matrices (Sussman et al., 2018; Chen et al., 2018). Optimization development significantly improved the ion transmission efficiencies and sensitivity to a level comparable with vacuum MALDI as reported in several studies (Chen et al., 2018). The particularly soft ionization technique produces low internal energy molecular ions inducing minimal fragmentation, which makes AP-MALDI particularly important for the analysis of biological samples (Laiko, Baldwin and Burlingame, 2000).

Imaging via AP-MALDI and MS readout/detection in combination with compound and tissue specialized sample preparation methods can be applied to detect almost any kind of molecules in a huge mass range, in all kind of biochemical studies (Ferguson et al., 2014). It has the capability to image thousands of molecules in a single experiment, such as metabolites, lipids, peptides, proteins, and glycans, and offers a label-free alternative to other methods (Buchberger et al., 2018; Cobice et

al., 2015). AP-MALDI MSI also shows a high tolerance to salts, buffers and various sample

contaminations (Moskovets et al., 2016). Its sensitivity ranges down to femtomole levels and high throughput analysis in combination with TOF-MS instruments. The generated spectra are

uncomplicated and comparably easy to interpret due to the almost exclusive generation of single charged ions and little to no fragmentation (Calvano et al., 2018). To strengthen the biological conclusions MSI can also be combined with other complementary imaging modalities, such as microscopy and staining methods or Raman spectroscopy (Buchberger et al., 2018).

2.1 Principle and instrumentation

2.1.1 AP-MALDI (ng) UHR source

The monolithic source housing of the compact AP-MALDI contains all important component and provides the space for the ionization process to take place: laser, optics and lenses, x-y stage and target plate holder, camera, and control electronics. It is attached to the atmospheric pressure interface flange of the mass spectrometer (Trap et al., 2017). The target plate, supporting the prepared sample, can be simply inserted onto the target plate holder and the door of the AP-MALDI closed to activate the ion source to enable measurements. A more precise description of the sample preparation can be found in later section 3.2.2.

(9)

9 A solid-state Nd:YAG laser irradiates the

solidified/crystalized components from the surface at a 28 ̊angle, with a wavelength of 355 nm, output frequency up to 10 kHz and a laser spot size down to 10 µm (Trap et al., 2017; Schneider, Lock and Covey, 2005). The distance between sample plate and the inlet of the MS capillary extender is approximately 2 mm and can be optimized or adjusted through different spacer plates mounted onto the sample plate holder for appropriate laser focusing on the sample. A simple layout is

illustrated in figure 1.

To facilitate an appropriate operation of the AP-MALDI source an integrated CCD camera and imaging optics enable the user to monitor the target plate motion and the sample ablation

processes on the computer screen. A source of visible light inside the case housing illuminates the target plate surface generating a good view onto the sample. The camera image is displayed on a small window in the Masstech Target Software, via which important settings and parameter of the AP-MALDI source, such as the plate motion and laser firing, can be adjusted and controlled. During MSI experiment, the XY coordinates of the laser spot on the sample and the acquisition time stamp are recorded, which are then used along with the MS data file to generate ion images by imaging software (Chen et al., 2018). The whole internal arrangement of all major AP-MALDI (ng) UHR can be seen in figure 2 below.

Figure 2: Simplified AP-MALDI (ng) UHR from MassTech (without HV connection to the sample plate) (Trap et al., 2017)

28 ̊

Figure 1: Instrumental and laser spot arrangement in the AP-MALDI source (Tan et al., 2004)

Capillary extender, MS inlet Laser

(centred position) Sample

spot

(10)

10

2.1.2 Ion formation process

MALDI is a complex physical and chemical phenomenon, comprising the formation of protonated, deprotonated, cationized and radical species. These processes happen within a time scale of

nanoseconds (Schulz et al., 2006) and are relatively independent of the solvent composition, solution pH and analyte acid–base properties (Karas et al., 2000). The general concept comprises a huge variety of experimental conditions that can be adjusted to fit the detection of many different classes of analyte molecules. Incident angles of the laser beam, laser wavelengths, pulse energies, and pulse widths are properties strongly influencing the outcomes of a measurement. For example, high temperatures in the plume greatly define the dynamics in MALDI processes and might cause

metastable fragmentation of molecular ions induced by ion-neutral collisions in the hot environment (Moskovets et al., 2016). This can be influenced by the chosen matrix and the laser fluence. The variety of available matrices and co-matrices in combination with countless analyte molecules generates even more possibilities of matrix-analyte combinations. Distinct fundamental mechanisms might be active in different matrices at a certain excitation wavelength (Knochenmuss, 2016). Additionally, different methods of sample preparation and mass analysers with different observation time scales, acceleration fields and sample temperatures can be used to record a mass either in the positive or negative ion mode depending on the analyte. All these parameters affect the final mass spectrum, which is only a snapshot of the ion population remaining some microseconds after the laser pulse and reaching the detector and does not provide time resolution on successive MALDI mechanism (Schulz et al., 2006). Nowadays more processes are understood but scientists are still searching for correlations between performance and possible contributory factors. No property itself provides a satisfactory and no single mechanism can explain all ions observed explanation and the performance in MALDI within a unified model (Knochenmuss, 2016). It remains a high demand on better understanding the ionization mechanisms that are involved in the MALDI to improve ion yields, control fragmentation, and to control the charge state during the formation process. Furthermore, enhanced knowledge could provide rational guidelines for the matrix selection to develop analyses for new classes of compounds and specific analytical problems (Zenobi and Knochenmuss, 2002). In quantitative approaches enhanced knowledge could explain possible reasons for discrimination effects in MALDI experiments (Zenobi and Knochenmuss, 2002). The fundamental process of laser light absorption and mechanism of ion formation under atmospheric pressure are similar to those explored in vacuum instruments (Trap et al., 2017). Additionally, specific atmospheric pressure processes, including thermalization of vibrationally excited ions and ion-ion or ion-molecule reactions, might happen over an extended time period (Laiko et al., 2000). The prevalent scientific idea defines the ionization as a combination of primary and secondary processes. The former arises due to laser irradiation, while the latter emerges during the pulse extraction by collision with other molecules within the particle plume. These two processes are independent mechanisms that can occur in parallel and both influenced by the choice of the matrix (Schulz et al., 2006).

In primary ionization processes the first generation of free ion is formed from neutral molecules in the sample in both positive as well as negative ion mode. This early phase takes place close to the samples surface and leads to initial charge separation of mostly matrix-derived ions species (Zenobi and Knochenmuss, 2002), since they are usually much more abundant than the analyte and highly energized by the laser, while heavier analytes are rather inert with respect to primary ion formation (Knochenmuss, 2016).

In AP-MALDI the most important contribution to analyte final internal energy is due to secondary processes. After desorption and first ionization reactions the molecules are transferred from a solid state into the gas phase ion cloud that is referred to as the laser plume (Laiko et al., 2000). The material in the expanding plume is still several % of the solid density and packed with primary protonated and deprotonated ions, neutral molecules, matrix ions and radicals, electrons, hydrogen

(11)

11

ions and analyte molecules (Schulz et al., 2006; Moskovets et al., 2016). The moderately hot cloud expends in the direction towards the incoming laser beam with an approximate cosine intensity distribution along this axis (Zenobi and Knochenmuss, 2002),

as shown in more detail later in

section 2.2.2.4.

The following cascade of secondary reactions between matrix ions and molecules can comprise single or multiple steps leading to the formation of analyte ions that are evetually observed at the detector of the mass spectrometer (Knochenmuss, 2016). The rate of these charge and energy transfer reactions, including dissociative electron-capture reactions releasing hydrogen atoms, energy pooling mechanism and neutralisation, proton transfer or other charge compensation reactions (Zenobi and Knochenmuss, 2002), depends on the plume density and temperatures, and thus does the ion yield (Knochenmuss, 2016).

Secondary reactions can take place in both polarities, between matrix and analyte, analytes and analyte and ambient gas molecules and are assumed to be reversible reactions (Knochenmuss, 2016). The most important ionization mechanisms are those leading to protonated and cationized analyte including proton-transfer reactions, cationization reactions and electron-transfer reactions (Zenobi and Knochenmuss, 2002). Shortly after the analyte ions have been formed they collide with gas molecules at atmospheric pressure that cools them down and decreases their internal energy (Schulz

et al., 2006). The rate of analyte-matrix collisions becomes negligible compared to collisions rates of

the analyte with ambient gas molecules as soon as the dense plume has dissipated (Schulz et al., 2006). The rapid thermalization of the ions prevents high-energy collision induce dissociation (CID) and fragmentation of excited ions or decay into metastable ions. Thus, preserving intact precursor ions of compounds such as thermally labile proteins, peptides, metabolites or other biochemicals and pharmaceuticals (Moskovets et al., 2016; Laiko et al., 2000). Therefore, AP-MALDI gives more

reproducible results than conventional vacuum-MALDI-TOF in terms of fragmentation extent (Schulz

et al., 2006) and allows one to study the sole contribution of the primary ionization processes and

the role of the matrix in energy-transfer, since the corresponding fragment can only be formed in the very early stages (Schulz et al., 2006).

On the other hand, these collisions also induce effective ion neutralization in the plume (Karas et al., 2000; Zenobi and Knochenmuss, 2002). These processes are most effective for highly charged cluster ions and high molecular weight analytes with extended colission cross section (Knochenmuss, 2016; Karas et al., 2000), explaining the increasing difficulties of generating mass spectra from e.g. large proteins (Karas et al., 2000; Moskovets et al., 2016). Ion suppression (IS) is another mechanism that leads to a decrease in signal of certain analyte ions depending on their surrounding. It is

straightforward evidence for ionization via exciton pooling, since the product will be either matrix ion or analyte ion if present (Zenobi and Knochenmuss, 2002). If enough analyte is present and the laser flux does not exceed a certain threshold all matrix ions can be suppressed (Knochenmuss, 2016). Suppresion effects not only occur between matrix and analyte, but also between different analytes from similar or different compound species (Knochenmuss, 2016). Generally, analytes with low ionization potential (IP) values, high ionization efficiency or a faster reaction rate with the matrix suppress those with higher IP values (Knochenmuss, 2016; Moskovets et al., 2016). This effect occurs either in positive or in negative mode, but not in both (Zenobi and Knochenmuss, 2002). Therefore, it remains difficult to certainly assign the observed ions in the mass spectrum to primary preforming ionization mechanisms or if they’re resulting from secondary gas-phase chemistry in the plume (Zenobi and Knochenmuss, 2002).

Different models have been developed that tried to explain MALDI on a quantum chemical level, such as the coupled physical and chemical dynamics (CPCD) model of the lucky survivor model (Knochenmuss, 2016; Robinson et al., 2018) within one approach. But considerable uncertainties still exist and the underlying mechanisms are under constant debate (Calvano et al., 2018). Next to electromagnetic energy from the laser thermal energy might be able to make up for deficiencies between a photon excitation and the ionization potential of the matrix compound as suggested by Allwood, Dyer and Dreyfus (Zenobi and Knochenmuss, 2002), but ionization processes are found to

(12)

12

not happen at thermal equilibrium and below the very high temperatures that would also adversely affect fragmentation (Knochenmuss, 2016).

The near-complete lack of knowledge regarding the local environment in a MALDI sample is a major limitation for rapid development and inhibits the implementation of consistent mechanisms (Zenobi and Knochenmuss, 2002). The ionisation, as well as the desorption, is a collective phenomenon of photoionization and photochemical reactions and ion loss processes, from which the final ions and quantitative ratios that one observes in the mass spectrum could completely differ from the primary formed ions (Zenobi and Knochenmuss, 2002).

2.1.3 Mass spectrometry

Mass spectrometry is an analytical technique for the detection and identification of molecules in their ionized states. The methods high sensitivity, low detection limits to far below femtomole levels, the speed of analysis and its diversity of its application have resulted MS to be an outstanding technique in aspects of e.g. biochemical research, including proteomics, metabolomics and high throughput screening in drug discovery (Hoffmann and Stroobant, 2007). In principle the molecules are separated according to their mass-to-charge (m/z) ratio (Shariatgorji et al., 2014) after they have been ionized in the ion source and transferred to the mass analyser. Nowadays a variety of ionization methods exist to deal with all kinds of samples, which may be solid, liquid or gaseous. In this study the AP-MALDI instrumentation is detachable from the mass analyser and detection system and can be exchanged with other ion sources, such as ESI or APCI.

After the ionization procedure, the charged particles are separated in the mass analyser according to their m/z and detected as a proportion to their abundance (Hoffmann and Stroobant, 2007). From this point all processes take place under high vacuum conditions (low pressure). Charge state as well as the mass of an ion influence its behaviour in the mass analyser. The mass determined by the mass spectrometer depends largely on its resolution and accuracy of the analyser. The resulting mass spectrum, in which the signal intensity (y-axis, relative abundance) is plotted as a function of the mass-to-charge ration m/z (unitless) on the x-axis, visualizes the elemental and isotopic composition of a sample and helps to determine the masses and structure of chemical compounds. Different mass analysers utilize various specific analytical principles in which they separate the charged particles due to their characteristics either in space or in time. The basic differences between the common types are defined by how they use static or dynamic electric and magnetic fields to achieve separation. They can be divided into groups based on different criteria. Scanning analysers, such as quadrupole instrument, only allow the ions of a given m/z to go through at a given time, successively scanning through the determined mass range. While, TOF or ion trapping analysers allow for the simultaneous transmission of all ions throughout the entire time scale. Each mass analyser has its advantages and limitations. The m/z range that can be analysed is theoretically limitless, but restrictions arise from the inability to desorb/ionize very large analytes, efficient transmission through the different stages and from detectors which are inefficient for very large ion detection (Kabarowski, 2014). Therefore, the sensitivity of MS for molecules in the especially the low weight range (< 1000 Da) is higher (Ferguson et al., 2014). While improvements in mass accuracy positively affects the resolution, the optimization of one parameter for specific analytical purposes is often a compromise that might adversely affecting other properties of the instrument and a matter of expenses for the machinery.

(13)

13

2.1.3.1 QqTOF-MS (quadrupole-quadrupole-orthogonal acceleration time-of-flight mass

spectrometer)

(Bruker Daltonik GmbH, 2013)

In this study the AP-MALDI source was coupled to a QqTOF instrument for the development and optimization of imaging techniques on zebrafish tissue. The aim was to perform MSI with high spatial resolution, high mass

resolution, high mass accuracy, high selectivity and good sensitivity for compound

identification to fit the purpose of chemical exposure and toxicity studies. The QqTOF mass spectrometer utilizes instrumental capabilities from both a tandem quadrupole and a time-of-flight arrangement by connecting them in a consecutive arrangement in one instrument. With this setup ions can be efficiently guided through the different stages and transferred into the analyser while neutral gas molecules are removed by the pumping system. This enables measurements in an extended mass range and higher selectivity and sensitivity compared to the simpler oaTOF (orthogonal acceleration time-of-flight) (Laiko et al., 2000).

Several technical components are incorporated in each stage of the compact QqTOF through which the ions and gas molecules pass, as shown in the model of the Bruker instrument in figure 3. This includes an ion transfer stage, a Q-q-stage containing a collision cell, a TOF assembly with

implemented reflectron and the detector at the end, supported by five vacuum stages that build up the required low-pressure conditions. In the earliest stage of the MS system the ions are guided from the capillary exit to the first mass analyser through two ion funnels and one hexapole, on which effective potentials (radio frequency (RF) voltages, direct and alternating currents (DC, AC)) are applied. Neutral gas and solvent molecules escape from the ion path through the gaps between the funnel plates electrodes and are removed by the pumping system. An axial DC gradient promotes the collection of ions exiting the capillary and pushes them towards an orifice at each funnel’s exit leading to the hexapole ion guide.

The following analytical quadrupole consists of three quadrupole segments representing the first mass analyser in the compact. While the middle segment provides for high mass resolution the outer two segments optimize the ion transfer efficiency. In the RF only mode, the Q-q-stage functions as an additional ion guide for transition of a narrow or broad mass range. Operating the quadrupoles as a mass filter only, one can isolate a certain mass or defined mass range of analyte ions prior to mass analysis in the TOF MS. By varying the ratio of AC/DC from low to high one can scan through the whole mass range (m/z) to detect present molecules by selecting threshold masses at which ions will either pass between the rods or hit at some point to become neutralized.

Operating the instrument in MS/MS mode (tandem MS, 2D-MS, n_{MS) enables indirect structural}

analysis by collision induced dissociation (CID) of the ions and yields additional molecular information for the identification of low abundant, structural complex molecules from the soft AP-MALDI

ionisation (Ferguson et al., 2014). Therefore, the second quadrupole, that is enclosed by a collision cell (reaction chamber), is filled with a collision gas.

The narrow mass range containing the desired compounds molecular ion gets isolated in the 1st_MS

stage by the earlier mass resolving quadrupole. The isolated ion then enters the 2nd_quadrupole

Figure 3: Layout in the QqTOF-MS from Bruker Daltonik (Bruker Daltonik GmbH, 2013) Q-q-stage collision cell TOF reflectron detector ion funnel capillary hexapole analytical quadrupole orthogonal accelerator

(14)

14

inside an enclosed chamber, which is now acting as a collision cell/reaction chamber. It is filled with a neutral inert gas, such as N2 or argon, with which the analyte ions collide. When the internal energy exceeds the dissociation energy of the bonds in the analyte compound the ion dissociates into fragments. Complex ions can dissociate via different reaction pathways with respective dissociation energy. The compound specific fragmentation pattern facilitate the identification of e.g.

biomolecules (Sussman et al., 2018). The quadrupole in the collision cell acts as a 2D ion trap that ends with a gate lens and a transfer lens. For an adjustable time slot, the gate lens temporarily blocks the transmission to the TOF stage for efficient accumulation of ions. Afterwards, the voltage changes to supports the extraction, transmission and injection of the accumulated fragments, reaction products and daughter ions to be analysed in the 2nd_{MS mass analyser (TOF). The three main TOF}

components are the orthogonal acceleration stage, the dual stage reflectron and the detector. Ions are accelerated orthogonally into the flight tube and towards the reflectron by an electrostatic field acting on their charge. Afterwards the ions enter a field free drift region in which they are separated over time. Ions defined by a fixed Ekin and different m/z ratio reach different velocities with which

they travel uniformly trough the drift region of known length. The underlying process can be easier imagined with the fundamental TOF equation:

t=constant*√m/z

with the constant being defined as 2eU/L2

Variations in the flight time that can arise for molecules of the same charge and mass can be compensated by implementation of a reflectron, which consists of a series of plates with increasing voltages (Kabarowski, 2014). Ions are decelerated, stopped, and deflected, reaccelerated, and redirected back into the flight tube and towards the detector. This design enables simultaneous time and spatial ion focusing by compensating for ions of different energies but same m/z and increase mass accuracy and resolution in TOF MS (Kabarowski, 2014).

At the final stage of the analytical instrumentation a microchannel plate detector converts the signal from the mass analysers into an electrical signal that can be read out by the computer. The results can be visualized with the appropriate software numerically or in form of spectra.

2.2 AP-MALDI MSI

2.2.1 Advantages/disadvantages & differences to vacuum MALDI

In general, the AP-MALDI has many features in common with the conventional vacuum MALDI, such as their high tolerance to salts and other contaminants (Moskovets et al., 2016). No affinity tagging, staining or expression markers are necessary for either methods, which would limit the amount of chemical info that can be obtained from a sample. Images of both known and unknown targets are obtained without the need of developing and preparing specialized labels and extensive knowledge about the individual compound that will be analysed (Shariatgorji, Svenningsson and Andrén, 2014). The overlay onto histological images of the same tissue section enables the identification of

correlations between changes in the abundance and localization of specific chemical species with a high spatial resolution. (Shariatgorji et al., 2014). Even the ablation after thin layer chromatography (TLC) or other surfaces is feasible and straightforward with (AP)-MALDI (Ferguson et al., 2014). Similarities between the MALDI methods, such as sample preparation procedures, nature of the matrix, the proportions of matrix and analyte, and the laser energy that is used, accelerate, and simplify the optimization process for the latter method. One can simply fall back onto already established methodologies (Laiko et al., 2000).

(15)

15

Certainly, each ionization method has its own advantages and disadvantages that can be used to gain complementary information (Buchberger et al., 2018). Atmospheric conditions facilitate the handling of the sample and the source, since the stand-alone machine can be attached to multiple MS

instruments offering e.g. higher mass accuracy and resolution for confident identification of small molecule metabolites or high throughput (Sussman et al., 2018). It gives access to analyse additional molecules and labile compounds that are not compatible with vacuum MALDI sources by enabling the use of otherwise sublimating matrices (Jackson et al., 2018; Sussman et al., 2018). The sample no longer needs to be placed under vacuum prior to analysis, which combined with the 10 kHz laser and the advantages of different mass analysers significantly increases the speed of image acquisition compared to MALDI instruments with lower repetition rates (Sussman et al., 2018). This gain in time gets compensated by the precise alignment of the ion source, that is essential for AP-MALDI.

Furthermore, more tuning and careful optimization of parameters is required to generate

comparable performances to the vacuum MALDI (Schneider et al., 2005). As the transfer of ions into the vacuum system is relatively inefficient the total sample consumption is higher under atmospheric conditions (Laiko et al., 2000). The even softer ionisation method gives rise to extensive levels of analyte-salt and analyte-matrix adducts, potentially spreading the appearance of the target ion over several low intensity signals (Moskovets et al., 2016).

Most studies report a lower sensitivity and inferior signal intensity in AP experiments of

approximately half the vacuum MALDI’s intensities (Schneider et al., 2005; Sussman et al., 2018), and more detectable m/z for each matrix (Sussman et al., 2018). Nevertheless, Schneider and colleaques proved that the AP-MALDI offers sufficient sensitivity for many biological applications by measuring on an QqLIT (quadrupole-linear ion trap) instrument to generate sub-femtomole MS and MS/MS data for protein digests. The intensity of signals of ions below 1200 Da approached that of vacuum

systems and similar S/N ratios were achieved over an extended mass range with both of the two MALDI sources (Schneider et al., 2005; Sussman et al., 2018). The distribution of the shared m/z between MALDI and AP-MALDI were similar but the AP-MALDI provides images with a spatial resolution surpassing MALDI performances (Sussman et al., 2018).

2.2.2 Qualitative analysis of lipids with AP-MALDI MS(I)

When operated appropriately the AP-MALDI comes with the ability to analyse a huge number of endogenous and exogeneous molecules over a wide mass range directly from a complex biological sample, since no specific staining or tagging approach is required and all ionized molecules can theoretically be detected. (Ferguson et al., 2014). The visualisation of compounds at µm spatial distribution in biological matter goes along with simultaneous identification of ions of interest by using high mass resolution MS instruments (Shariatgorji et al., 2014). So far, AP-MALDI has already been used to detect drugs, pesticides, oligosaccharides, peptides and proteins (Sussman et al., 2018) and multiple lipidomic and imaging studies of all major lipid classes in several tissue types (Jackson et

al., 2018; Sussman et al., 2018; Desbenoit et al., 2017). Lipids are ubiquitous throughout organisms

and highly abundant in many cell types, comprising internal and external membrane (lipid bilayers) and subcellular organelles, such as mitochondria and surrounding nuclear membrane (Ferguson et al., 2014) Furthermore, these amphipathic compounds fulfil essential biological functions and are involved in neurological processes and diseases, including neurodegenerative disorders or tumours. A list of the diverse classes and best methods for their detection can be found in the paper of Leopold et al (Leopold et al., 2018) and by Ferguson and team (Carly N. Ferguson et al., 2014). Localisation and concentrations changes of endogenous compounds can lead to further inside and understanding of those processes and indicate malfunction (Shariatgorji et al., 2014). Therefore, their spatial information is as important as their chemical properties, exceeding the analytical

capabilities of traditional methods such as chromatography (Seeley and Caprioli, 2013). An advantage of lipids is their inherent ability to ionise readily in either positive as well as negative ionisation mode, due to typical functional groups such as phosphate anions or nitrogen centred cations (Carly N. Ferguson et al., 2014; Sussman et al., 2018). They can be detected in samples without specific

(16)

16

treatment. Yet, the analysis of mixtures comes along with suppression problems, favouring the ionisation of certain lipid classes over others.

While PC lipid species might dominate the mass spectra in positive ionisation mode, the negative mode normally reveals signals of phosphorylated and sulphated lipid species (Leopold et al., 2018). For the development of novel tools and analytical techniques one must be able to overcome two main challenges: first the preservation of the structural and locational information of individual lipids and second, providing high sensitivity and selectivity for molecular specification from the diversity of lipid structures (Ferguson et al., 2014).

2.2.2.1 Sample preparation

Clearly highlighted throughout all (AP) MALDI studies, one of the most crucial steps for a successful analysis is the sample preparation. The basic procedure is minimal and only contains a few steps, yet highly affects the reproducibility, sensitivity, acquisition speeds, spatial resolution and reliable interpretation of data (Buchberger et al., 2018). Already existing matrices, application techniques and modification methods can be copied from vacuum MALDI and further optimised for the respective instrumentation. The results are highly depending on the purity of the matrix as well as environmental contaminants and conditions, such as the surrounding temperature during

preparation. One must also maintain the integrity of the molecular distribution and localization of the analytes for MSI in case of washing the tissue or using specific sample modification methods for enhanced detection of a certain compound. Generally, washing of the tissue slice for the analysis of lipids is not necessary, but can be useful for the modification of the target ion to simplify the spectra via enhanced formation of e.g. positive salt adducts with phospholipids or removal of interfering compounds that promote the chemical noise (Ferguson et al., 2014). Other strategies have been reported to change the physicochemical properties of the target molecules to improve the ion yield by reducing discrimination effects in multicomponent analysis. A few examples, such as chemical derivatisation (Zenobi and Knochenmuss, 2002; Shariatgorji et al., 2014), on tissue labelling (Buchberger et al., 2018) and on tissue digestion (predominantly used for proteins) (Shariatgorji et

al., 2014; Buchberger et al., 2018) have been introducing and used for different analyte classes.

2.2.2.2 Matrix application

To gain an evenly coated tissue surface the process of matrix application must take place under controlled conditions. Only then one can ensure sufficient extraction and co-crystallization of the analytes from the sample and minimal diffusion, since already the surrounding temperature affects the quality of the homogeneity of the solid material (Buchberger et al., 2018). For non-imaging measurements an extracted analyte solution is premixed with an excess of matrix solution and spotted onto a target plate as 0,5-1 µL droplets (Moskovets et al., 2016). After short drying the droplets can be analysed, making this technique quick and easy for fast screening of compounds. The application with a painter’s airbrush is prevalently used for standard MALDI imaging measurements and easy and cheap but cannot guarantee high quality and reproducibility of the imaging results. Inconsistent spray velocity, duration of the cycles and variations in the thickness and wetness of the matrix layer can induce extreme aberrations between scientists and cause analyte diffusion,

especially in small molecule studies (Gemperline et al., 2014). The automatization of this process using a spraying robot in a closed chamber makes up for exactly these disadvantages and enables high resolution imaging for reproducible results across individuals and laboratories (Buchberger et

al., 2018). Electrospray devices additionally allow the control of the crystal size (Buchberger et al.,

2018). The solvent-free application via sublimation of the matrix onto the tissue presents another low-cost option for high resolution imaging. Both latter techniques were reported to yield a greater number of metabolite ions and less analyte diffusion, providing complementary matrix peak profile

(17)

17

results, and possibly the detection of additional compounds at best spatial resolution (Van Nuffel et

al., 2018). Yet, the dry method of sublimation comes with a lower sensitivity, and detection of fewer

analyte ions (Buchberger et al., 2018; Gemperline et al., 2014).

2.2.2.3 Matrices for the analysis of lipid with AP-MALDI

The key for a successful analysis of the enormous diversity of endogenous and exogeneous

compounds is the selection of an appropriate matrix, taking into consideration compound properties such as proton affinity, absorption spectra and functional groups (Jackson et al., 2018). Even amongst classes of lipids finding the optimal matrix candidate can distinguish between satisfying signal or no detection at all. The mostly organic, crystalline matrices, of lower molecular mass around 150 to 200 Da (in this project), fulfil multiple roles during the analysis.

First the analyte molecules must be extracted from the tissue section and co-crystallize with the matrix by evaporation of the solvent. This mainly depends on the availability of the molecules, the application method, choice of matrix and the solvent content in the mixture. If the target molecules did not incorporate well in the crystal structures, it is unlikely that they will be detected (Buchberger

et al., 2018). Furthermore, the ‘wetness’ of the matrix solution (solvent content) during the

application influences the extraction, diffusion, integration and crystal size on the biological material. The latter vastly limiting the spatial resolution in imaging by exceeding the size of the laser spot (Ferguson et al., 2014). The second task of the matrix is the efficient absorption of the laser energy via electronical excitation. It acts as a mediator for controlled energy deposition onto the analytes molecules to ionise (Ferguson et al., 2014). The appropriate amount of energy must be balanced between minimal internal energy deposition, to softly overcome the limit of ionisation, and

extensive, ‘hard’ ionisation causing fragmentation, which disturb the results or could possibly be of help in e.g. structural studies (Schulz et al., 2006). And as third in the list, the matrix supports and improves the efficiency of the ionisation procedure via e.g. collective energy pooling and proton transfer reactions. Again, one must seek out the appropriate matrix concentration and softness’ (or respectively ‘hardness’) of the matrix for the sensible balance between minimal internal energy deposition for soft ionisations of the ions. The protonation of analyte molecules is

thermodynamically favourable if the proton affinity (PA) of the compound is higher than the PA of the chosen matrix, but excess exothermal energy will be deposited into the reaction partners as internal excitation that can lead to increased analyte fragmentation (Zenobi and Knochenmuss, 2002). One should of course assure appropriate solubility properties of the matrix with the analyte and chemical inertness to prevent manipulative reactions of the compounds. The mass range of the target compounds must be considered to avoid signal suppression by the intense interferences from the saturated matrix and chemical noise, mostly originating from impurities in the lower mass range (Moskovets et al., 2016). In every mixture analysis the competition of charge and favoured ionisation of e.g. more basic analytes (in positive ionisation mode) may produce unequal peak intensities and discrimination of signals also from equimolar compositions (Kabarowski, 2014). The differences in the detectability even among only phospholipids (PL) make mixture analysis complicated in both positive and negative ionisation mode (Leopold et al., 2018). It has been found that higher-molecular-weight analytes can be analysed better at lower matrix-analyte ratios (Knochenmuss, 2016). Due to AP-MALDIs soft ionisation, matrix cluster can form depending on the chemical nature of the analyte (A+M+H, A+2M+H,…)(Laiko et al., 2000). In general the distribution of the analytes ion current over multiple peaks should be avoided since it reduces the sensitivity of the method (Laiko et al., 2000). The migration of the lipids and capacity of the matrix depends of the fluidity (wetness, solvent content) of the mixture at the respective preparation temperature and can vary between different kinds of lipids as well (Van Nuffel et al., 2018). Cholesterol, for example, has been found to migrate in the presence of other fluidised lipids and tends to colocalise with the monoacylglycerols (MAG) and diacylglycerols (DAG). Whereas triacylglycerols (TAG) and phospholipids, that can be found in the mass range between approx. 700−900 Da, tend to migrate with a delay compared to MAG and DAG and therefore depend on the interaction time and thickness of the matrix cover (Van Nuffel et al., 2018). Some lipid classes, such as TAG, are simply not detectable at all if an unsuitable matrix is used

(18)

18

(Leopold et al., 2018). Several matrixes have gained popularity for their universal application possibilities. Additionally, the exploration and synthesis of new matrices, targeting on more specific analytical questions, brought up interesting methods such as nanoparticle-assisted laser

desorption/ionization (NALDI) for improved spatial resolution (Ferguson et al., 2014), or liquid matrices that circumvent issues regarding solubility and co-crystallization of the mixtures and largely smoothened the hotspot phenomenon (Zenobi and Knochenmuss, 2002). Three of the most

commonly used matrices were used in this project. α-cyano-4-hydroxycinnamic acid (CHCA) and 2,5- dihydroxybenzoic acid (DHB) function predominantly in positive ionisation mode, while

1,5-Diaminonaphthalene (DAN) was chosen as a versatile matrix that can be utilised also in negative ionisation mode.

α-Cyano-4-hydroxycinnamic acid (α-CHCA)

Amongst the most popular and universal matrices since the early days of (AP) MALDI is α-cyano-4-hydroxycinnamic acid (α-CHCA)(figure 4). It has been initially utilised for the analysis of peptides and proteins and later also for metabolites in positive ionisation mode (Buchberger et

al., 2018). Especially for smaller molecules up to only a few thousand

Da, α-CHCA is a good matrix to try, due to its high ionisation efficiency

for a broad range of compounds and formation of a homogeneous layer of matrix crystals

(Kabarowski, 2014; Moskovets et al., 2016). Its low pKa of 1,17 (Porta et al., 2011) and proton affinity (PA) between 812-842 kJ/mol (Schulz et al., 2006) makes this matrix a really acidic candidate and strong ionisation partner. Furthermore, α-CHCA forms small crystals for homogeneous distribution of compounds in dried droplets and even cover of the surface for imaging experiments (Gemperline et al., 2014).

2,5-Dihydroxybenzoic acid (DHB)

Another important matrix, 2,5-dihydroxybenzoic acid (DHB)(figure 5), has been used both in positive and negative ionisation mode (Kabarowski, 2014), with increasing focus on the detection of small molecules in metabolomic studies. Currently, it is known as the most popular and universal matrix for lipid analysis, since it was found to enables the detection of virtually all lipid species ranging from apolar to very polar lipids (Dong et al., 2013; Leopold et al., 2018; Buchberger et al., 2018). With a pKa value around 2,94 (Leopold et al., 2018) and a PA between 850-860 (kJ/mol) (Schulz et al., 2006), DHB is considered to be a ‘cool’ matrix

compared to α-CHCA. Imparting only little internal energy onto the analyte’s molecule enables softer ionisation that leads to less fragmentation and therefore facilitates analysis of the molecular ions. On the other hand it promotes the formation of more pronounced adduct ions with e.g. sodium, other cations or the matrix itself, which might reduce the sensitivity of detection due to split in signal (Dong et al., 2013) in AP-MALDI experiments (Schulz et al., 2006). Yet, DHB has been reported to be the most appropriate partner for lipid analysis with a low yield of photo- chemically generated matrix ions, hence little interferences and clean spectra (Calvano et al., 2018). MALDI studies of cholesterols and CE (cholesterol esters) lipids were conducted using DHB, as well as it has been utilised for the analysis of TAGs and a variety of phospholipids (PL) and sphingomyelin (SM) (Leopold et al., 2018). However, this matrix also comes with slight disadvantages: It is recommended to not be used in negative ionisation mode due to its significant background. It was also found to be not as sensitive in detection compared to other matrices (Leopold et al., 2018). Lastly, the application of the matrix is more critical and should be handled in a sublimation chamber or automated sprayer to form small crystals for high quality MSI (Gemperline et al., 2014).

Figure 5: Molecular structure of DHB Figure 4: Molecular structure of α-CHCA

(19)

19 1,5-Diaminonaphthalene (DAN)

Due to the huge structural variability and diversity of lipid classes analytes in the same tissue might require different polarities for acquisition.

1,5-Diaminonaphthalene (DAN) (figure 6) is a versatile matrix that can be used in either positive as well as negative ionisation mode (Dong et al., 2013) (Perry et

al., 2020). Its high basicity (pKa = 12,1 of 1,8-bis(dimethylamino)naphthalene,

value for 1,5-DAN not found (Ribeiro da Silva et al., 2010)) and proton affinity (PA) of 942 kJ/mol (Kevin Demeure, Valérie Gabelica, 2010) make this matrix a potential proton sponger, thus induces the formation of predominantly deprotonated molecular ions of the analyte via gas-phase reactions (Zenobi and Knochenmuss, 2002). The analysis of metabolites in negative ionisation

mode has been reported successful in various papers (Kabarowski, 2014; Buchberger et al., 2018), highlighting the sensitivity for especially phospholipids (PL) (including PI, PS, PA, CL, as well as PE, which are readily deprotonated) and yield of representative MS profiles of the natural PL mixtures compared to 9-AA in MSI on tissue (Dong et al., 2013; Leopold et al., 2018; Perry et al., 2020). Its strong UV absorption allows for the use of rather low laser power and much lower matrix

concentrations than 9-AA and DHB for the detection of phospholipids (Dong et al., 2013). Further comparisons between these matrices on the intensities of various lipid subclasses and adduct formation patterns can be found in the recent study of Perry et al (Perry et al., 2020).

2.2.2.4 Experimental parameters of AP-MALDI

Besides the matrix type and the matrix-to-analyte ratio a number of other parameters play a role in optimal detection of the target ions. They can be grouped by assigning each to a certain part of the instrumentation or individual functioning within the AP-MALDI MS.

First of all, the geometry of the ionisation source aligning with the inlet orifice of the capillary extender is not per se listed as an individual parameter in many literatures on the atmospheric source, since different MS instrument provide various connective flange systems. Nevertheless, when mounted onto the Bruker QqTOF-MS, slight irregularities of the sources position of some millimetres can arise. For a good performance with the AP MALDI especially the alignment of the plume relative to the mass spectrometer inlet is critical for ions to be efficiently drawn into the vacuum and not scattered by the

inhomogeneous electric field near the capillary inlet (Schneider et al., 2005; Moskovets et al., 2016) Therefore, one has to consider that the expansion will not be

perpendicular to the sample plate but along the axis of the incoming laser beam, as shown in the figure 7. The alignment has a large impact on the ion transfer into the MS and therefore, vastly influence the sensitivity and the detection of ions (Laiko et al., 2000).

Secondly, adjustment in the laser energy arriving at the targets surface can dramatically affects the ions signal intensity and limit of detection (LOD) as well (Moskovets et al., 2016). Considering the matrix’s fundamental physical-chemical properties and ionisation threshold for the target compound group one can adjust the energy per area via the laser power and the laser frequency (repetition rate), since the laser wavelength is a fixed parameter. Studies generally reported higher ion yields with elevated laser irradiance (Zenobi and Knochenmuss, 2002)

.

Moskovets et al. experienced a

Figure 6: Molecular structure of DAN

Figure 7: Model of the plume expansion in the AP-MALDI source (only indicating the dimensions and angles of the ion cloud and laser beam) (Tan, Laiko and Doroshenko, 2004).

̊

Capillary extender, MS inlet Laser (centred position) Sample spot Grounde d

(20)

20

saturation of the signals of peptides at laser energies approximately twice the ionisation threshold and no continues improvement in analytes signal (Moskovets et al., 2016). High laser power can induce fragmentation due to excess internal energy, called in-source decay (ISD) and post-source decay (PSD). The loss of the backbone and intensification of chemical noise and possible

fragmentation reactions are indications for these reactions (Hinou, 2019).

The third settings group comprises the three source parameter. A major challenge in AP-MALDI is the efficient transmission of ions from the atmospheric pressure zone into the vacuum of the mass spectrometer, which limits the sensitivity of the ion source (Laiko et al., 2000). Before the ions enter the MS through the narrow-bore, heated capillary with a limited gas intake, the plume of particles disperses once created at atmospheric pressure in the approximately 2 mm gap between the sample plate and the MS inlet (Tan et al., 2004; Sussman et al., 2018). The most important of those source parameters is the voltage that is applied to the metal capillary extender and shielded by a non-conductive heating sleeve. A static electric field is formed that promotes the charge separation and ion transport between the target plate and the capillary inlet (Moskovets et al., 2016). Additionally, a steady gas flow of nitrogen, which is provided by the mass spectrometer inlet, creates a suction which supports the electric field by aerodynamically focusing the expanding matter and dragging molecules towards the mass analyser (Trap et al., 2017; Laiko et al., 2000; Tan et al., 2004;

Moskovets et al., 2016). Along with the dry gas temperature they define the conditions in the source housing and are mainly responsible for the transfer of the ions from the atmospheric pressure interface into the vacuum of the MS. Yet, especially the capillary voltage also influences the kind of ions that can be detected. At increasing voltages the large forces of the inhomogeneous field around the capillary extender inlet overcomes the focusing of the aerodynamic drag, explaining the loss of ions, such as matrix clusters, by hitting the MS orifice (Moskovets et al., 2016). In general, the efficiency of focusing via the dry gas flow rate near the inlet is more pronounced for heavier

molecules and high rates might even adversely affect the signal of lipids and lighter ions (Moskovets

et al., 2016). The dry gas temperature on the heated capillary extender was not mentioned as an

essential parameter that would directly influence the formation of lipid ions significantly in any studies.

Last in the row of instrumental parameter that must be considered are the MS tune settings, divided into the transfer, quadrupole and collision cell voltages and energies that are applied to the different components of the QqTOF-MS, as explained in the earlier section 2.1.3. While lower voltages and frequencies generally suffice for the detection of high energetical light ions, heavier analyte ions demand more support and guidance throughout their travel. Furthermore, one can narrow the mass range to focus on only certain compounds and thereby improve the selectivity and sensitivity by optimising the MS tune. Since the early ion stage and internal energy of the compounds can have an enhanced influence on their detection. Phenomena, such as in-source decay before ions enter the mass spectrometer, were reported to vastly change the outcomes for measurements on e.g. proteins (Sellami et al., 2012) and other compounds (Hinou, 2019). Therefore, the isCID was selected

individually from the MS tune parameters and evaluated in separated experiments.

Regarding imaging, some more specific parameters vastly define the outcome of the measurement, amongst which the laser spot size and pixel dimensions are extremely important. Improving the spatial resolution requires a higher focus of the laser onto a smaller spot and, therefore, decreases the area of tissue that is getting ionised. Additionally, the rate of bimolecular reactions such as pooling and neutralization in the MALDI plume depend on its density, making the ion yield also a function of spot diameter (Knochenmuss, 2016). One has to consider the trade of between high spatial resolution or sensitivity (Buchberger et al., 2018). To tackle these issues studies have reported the optimisation of instruments geometry and optics to decrease laser spot size with an equal energy distribution at the samples surface. Additionally, adequate sample preparation methods must be used to diminish resolution losses due to the crystal size of the matrix (Buchberger et al., 2018).

(21)

21

2.2.3 Application of MSI on a tissue sample

After sample collection and dissection of the organism, the biological tissue should be immediately flash-frozen in order to stop chemical activity (e.g. enzymatic reactions), thus, reducing degradation and delocalization (e.g. diffusion across the tissue) of the molecules (Buchberger et al., 2018). The sample is sectioned into thin slices between commonly 10-20 µm in a cryotome. In AP-MALDI imaging, also thicker sections could be utilized, since the sample thickness insulating properties have no effect on the imaging performance (no voltage is applied on the sample plate). Furthermore, thicker slides are less susceptible to the mechanical force of the microtome blade and maintain the spatial integrity of the fragile tissue slice better (Kabarowski, 2014). Embedding techniques, enclosing the biological material in e.g. agar, gelatine, Tissue-Tek® and OCT (optimal cutting temperature compound), polyvinyl alcohol and polyethylene glycol polymers, are not commonly used in MALDI. Their incompatibility and cross-linking effects with some materials and analyte classes can suppress ion formation in MALDI (Kabarowski, 2014; Shariatgorji et al., 2014; Villacrez et al., 2018). But a recent study has reported the development of new methods for lipid analysis via formalin fixation only (without paraffin embedding) (Buchberger et al., 2018).

After cutting the tissue, it is thaw-mounted onto an appropriate sample holder, e.g. a microscope glass slides or metal plate, without any gluing material (Buchberger et al., 2018). In MSI with an AP-MALDI of the new generation normal glass or metal sample plates without conductive surface e.g. ITO (indium tin oxide) coating can be utilized, since no potential is applied on the target plate anymore. For measurements including more extensive sample preparation techniques like washing, O’Rourke et al coated the sample holder slide in nitrocellulose to glue the tissue on and prevent it from flaking or being washed off (Buchberger et al., 2018). A detailed description on the tissue preparation process can be found in section 3.2.2.

2.2.4 (Imaging) software, data processing & evaluation

Dedicated software and statistical programs are handy tools that facilitate not only the data processing of experimental outcome evaluation, but also the development and optimisation of analytical methods. Simple open source programs, such as Chemoface that was used in this project, will save a lot of work due to automatised generation and evaluation of e.g. experimental design approaches (design of experiment, DoE).

The MS data sets that result from a single imaging measurement of a big area or high spatial resolution comprise thousands of molecule-specific images with a high degree of dimensionality (Shariatgorji et al., 2014). Increasing complexity and exponential growth in file size makes them challenging to process and demands extensive computational capacity. In the standard conversion procedure, an imzml format file is generated from the raw instrumental data. It is compatible with most imaging software and reduces the data sets for more efficient storage. The AP-MALDI does not come with an instrument specific conversion and imaging software package, such as other

instruments including the fleximaging software of the MALDI-TOF/TOF from Bruker or the SCiLS Lab program that provide an imzml export function. Thus, several programs must be utilised to convert the raw MS files of spectra over the whole measurement time (chromatogram) into a XY-coordinate congruent molecular map. One approach includes the intermediate creation of an mzml file that is subsequently combined with the spatial data with imzMLConverter. The imzml format file can then be visualised with a variety of imaging programs that offer different tools for detailed evaluation. DatacubeExplorer (developed by FOM-AMOLF, Amsterdam) is a fast, straight-forward option that enables dynamic scrolling through masses in the average spectrum of the dataset and selected analysis of regions-of-interest in the tissue section. Additionally, programs like MSiReader (open source software developed by North Carolina State University), Biomaps (by Markus Stöckli) and SCiLS can be used to analyse multiple imaging data sets, perform comparative and colocalization

Development of tissue-imaging mass spectrometry approach using an atmospheric pressure MALDI source

MSc Chemistry

Analytical Science Track

Method development and optimization

of a tissue imaging mass spectrometry (MSI) approach

using an atmospheric pressure matrix assisted laser

desorption ionization source (AP-MALDI)

Master Thesis by Helene Fiedler

Supervisor & 1

Examiner

prof. dr. Pim Leonards

2

Examiner

prof. dr. Marja Lamoree

2

Supervisor

dr. Erika R. Amstalden

Vrije Universiteit Amsterdam

Department of Environment & Health

Universiteit van Amsterdam

May 2020

Abstract

Table of Contents

Abbreviations

1. Introduction

2. Theory of mass spectrometry imaging

2.1 Principle and instrumentation

2.1.1 AP-MALDI (ng) UHR source

2.1.2 Ion formation process

as shown in more detail later in

section 2.2.2.4.

2.1.3 Mass spectrometry

2.1.3.1 QqTOF-MS (quadrupole-quadrupole-orthogonal acceleration time-of-flight mass

spectrometer)

2.2 AP-MALDI MSI

2.2.1 Advantages/disadvantages & differences to vacuum MALDI

2.2.2 Qualitative analysis of lipids with AP-MALDI MS(I)

2.2.2.1 Sample preparation

2.2.2.2 Matrix application

2.2.2.3 Matrices for the analysis of lipid with AP-MALDI

2.2.2.4 Experimental parameters of AP-MALDI

.

2.2.3 Application of MSI on a tissue sample

2.2.4 (Imaging) software, data processing & evaluation

_Examiner

_{prof. dr. Pim Leonards}

_Examiner

_{prof. dr. Marja Lamoree}

_Supervisor

_{dr. Erika R. Amstalden}