Higher Mass Accuracy MALDI-TOF/TOF Lipid Imaging of Human Brain Tissue in Alzheimer’s Disease

Abstract

Matrix-assisted laser desorption/ionization imaging mass spectrometry (MALDI-IMS) is a well-established technique for elucidating the location and relative abundance of a range of biomolecules. More recently, research into this technique has shifted from simple discovery and demonstration of utility to application in biomedical research. Here, we describe a protocol utilizing MALDI-IMS for the spatial mapping of lipids in brain tissue from normal human brains and brains from patients with Alzheimer’s disease, in the context of Alzheimer’s disease. Improved accuracy calibration of the instrument from the tissue surface is emphasized, as this allows for significantly improved mass determination in time of flight (TOF)-based instruments enabling more confident preliminary lipid identification. This improved initial result allows MALDI-IMS data to be complemented with additional instrumentation, such as liquid chromatography mass spectrometry workflows or specialized non-TOF systems such as Fourier transform cyclotron resonance instruments. This method is not limited to human tissue and can be applied to virtually any lipid-rich formalin-fixed tissue.

INTRODUCTION

By 2020, for the first time in history, there will be more people over 60 years of age than there will be under the age of 5 (World Health Organization, 2014). Increasing longevity does not necessarily mean people are living healthier lives; nearly a quarter (23%) of the global burden of death and illness has been attributed to people over 60 years (Prince et al., 2015). Age is the leading risk factor for chronic disorders such as cardiovascular disease, cancer, and neurodegenerative conditions such as dementia, now the second leading cause of death in Australia (Australian Bureau of Statistics, 2017). The most common form of dementia is Alzheimer’s disease (AD).

The dementia epidemic is driven by an aging population and the lack of disease-modifying therapies. Overall, there has been little progress in new therapeutics for neurodegenerative diseases with many pharmaceutical companies scaling down their efforts in this area (Hutson, Clark, & Cross, 2017). The major bottleneck, the so-called “valley of death,” confines most exciting findings to preclinical models due to their subsequent failure in human clinical trials. One contribution to this failure is the relative disparity in complexity between the human brain and those of lower mammals compared to all other organs.

The brain is lipid-rich compared to other organs, with white matter that consists of 70% lipid in the form of myelin sheaths. In AD, the leading risk factors are genetic variants involved in lipid and cholesterol metabolism, but until recently, there were few tools outside of liquid chromatography to assay lipids in tissue and none to look at lipid-based cytoarchitecture. Matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-IMS), also known as MALDI imaging, is a well-established methodology that has been applied to the investigation of proteins (Hankin, Barkley, & Murphy, 2007; O’Rourke, Raymond, Djordjevic, & Padula, 2015), peptides (O’Rourke, Djordjevic, & Padula, 2015), lipids (Yalcin & de la Monte, 2015), and metabolites (Shariatgorji, Svenningsson, & Andren, 2014).

MALDI, as the name suggests, is a laser-based form of molecular analysis that has some unique requirements for sample preparation. Before analytes in a tissue sample can be measured, they must first be coated with an organic acid matrix. The laser is then able to excite the matrix, which in turn excites the analytes creating a plume of ions. The ion plume is then measured by time of flight, and the resulting ion masses are reported. The pin-point accuracy of the laser is what enables MALDI to build up a two-dimensional image of a tissue, one shot at a time.

MALDI-IMS has been applied to various tissues and diseases such as cancer (Hiraide et al., 2016) and heart disease (Angel et al., 2016) and, more recently, in the brain to understand the lipid changes in alcohol-related brain damage (de la Monte et al., 2018), AD (Gόnzalez de San Román, Manuel, Giralt, Ferrer, & Rodríguez-Puertas, 2017; Mendis, Grey, Faull, & Curtis, 2016), or the lipid cytoarchitecture of regions such as the subventricular zone (Hunter, Demarais, Faull, Grey, & Curtis, 2018). We have previously described a robust methodology for maximizing peptide ionization from formalin-fixed frozen human brain tissue (O’Rourke et al., 2018). Here we modify the reproducible and inexpensive approach for peptides to assay lipids. To identify lipids, it is necessary to achieve the highest mass accuracy capable from the initial analysis, and this method achieves that accuracy without utilizing additional high-resolution instrumentation. This simple-to-follow and comprehensive means for accurate mass determination and spatial mapping of lipids directly on tissue can be combined, if needed, with complementary methodologies for more definitive characterization.

This article describes the preparation of formalin-fixed paraffin-embedded, formalin-fixed frozen, and fresh frozen tissue samples for reproducible analysis of lipids via MALDI-IMS. Sample preparation does not require robotic or automated sample preparation instruments. In preparation, tissue samples are sectioned and then mounted onto indium tin oxide (ITO) slides that have been coated with liquid nitrocellulose prepared by dissolving a nitrocellulose membrane in acetone. The sample is then sublimated with a lipid-specific matrix in an apparatus made from readily available laboratory equipment. The lipid ions are calibrated against red phosphorus, a critical step for accurate identification.

BASIC PROTOCOL 1

PREPARATION OF FORMALIN-FIXED FROZEN BRAIN TISSUE SECTIONS

The sectioning of formalin-fixed frozen or fresh frozen tissue must be done in a way to minimize potential contamination from polymers and salts, and as such embedding in optimal cutting temperature (OCT) compound should be strongly avoided. Paraffin-embedded tissue cannot be used for the analysis of lipids, as the perfusion of xylene, which is necessary for paraffin embedding, will solubilize and remove lipids.

Materials

• Brain tissue of interest

• Liquid nitrogen

• OCT embedding compound (e.g., Agar Scientific)

• Nitrocellulose-coated ITO slides (see Support Protocol 1)

• 50 mM ammonium formate, pH 6.4

• Cryostat

• Vacuum desiccator

1. Take tissue block of choice, and snap freeze for 30 sec in liquid nitrogen.

2. Section frozen block at 12 to 15 μm with a cryostat, using minimal OCT compound as an adhesive to hold the tissue in place.

3. Mount samples onto nitrocellulose-coated ITO slides by placing frozen sections on chilled (≤5°C) slide, and allow to rise to room temperature. Then leave at room temperature until dry.

   Mounted tissue can be stored at room temperature in a vacuum desiccator for several weeks prior to analysis. The desiccator is used to prevent oxidation of the lipid specimen.

4. Wash sections by pipetting ~100 to 200 μl of 50 mM ammonium formate (pH 6.4) on and off section (4 times for 5 sec per wash) as previously described (Yalcin & de la Monte, 2015).

   Sections may dislodge if slides are dipped into solution in a glass Coplin jar.

5. Dry sample thoroughly for 30 min.

SUPPORT PROTOCOL 1

PREPARATION OF NITROCELLULOSE-COATED ITO SLIDES

Nitrocellulose is used as an electrostatic substrate that helps to adhere tissue onto the surface of the slide. It is inert when in contact with the sample, does not ionize in the mass spectrometer, and is resistant to a wide range of solvents and buffers. It can also sustain the physical integrity of the tissue when subjected to ultra-high vacuums.

Materials

• Nitrocellulose membrane (e.g., Sigma-Aldrich)

• Analytical grade acetone

• ITO-coated microscope slides (0.9-mm thick)

• Standard glass microscope slides

• Dust-free box

1. To prepare liquid nitrocellulose, weigh out 40 mg nitrocellulose membrane, and add 1 ml acetone (40 mg/ml final).

2. Pipette 40 μl liquid nitrocellulose in a straight line onto the surface of the ITO slide near its short edge. Using a plain glass microscope slide, drag the nitrocellulose along the length of slide (in a similar fashion to preparing histological blood smears; see Fantham, 1925).

3. Leave to dry for 30 sec, and then store in a dust-free box until needed.

BASIC PROTOCOL 2

SUBLIMATION OF TISSUE SECTIONS

The sublimation of dry matrix onto tissue ensures that molecules are not delocalized prior to analysis while ensuring a homogenous coating and thickness.

CAUTION: The sublimation apparatus poses a risk of explosion should the glass chamber fail, so it is advised to perform this aspect of the protocol with appropriate personal protective equipment, with the sublimation apparatus placed in a fume cupboard or some other contained and shielded space. Perspex shielding (5 mm) is considered adequate.

Materials

• Sample-mounted slide (see Basic Protocol 1)

• Recrystallized 2,5-dihydroxybenzoic acid (DHB) matrix (see Support Protocol 2)

• Sand bath

• Hot plate

• 5 decimal point analytical balance, capable of measuring masses in 0.1-mg increments

• Glass petri dish

• Vacuum sublimator (e.g., Chemglass)

• Double-sided copper tape

• Metal ring

• Retort stand

• Vacuum source

• Vapor trap

• Rubber tubing

• Ice

1. Preheat sand bath to 140°C.

2. Weigh sample-mounted slide on a 5-point analytical balance and record mass.

3. Place 300 mg recrystallized DHB matrix crystals onto the bottom of a glass petri dish, and spread to create a thin, even layer of crystals (Fig. 1A).

4. Place petri dish into the bottom of the sublimator chamber (Fig. 1B).

5. Mount sample slide onto the cooling finger of the sublimator, and secure in place with double-sided copper tape so that the tissue is exposed to the DHB once the sublimator is reassembled and mounted on the sand bath (Fig. 1C).

6. Assemble the sublimator, and suspend it above the sand bath (pre-heated to 140°C) using a metal ring connected to a retort stand (Fig. 2).

7. Connect vacuum source to the sublimator, and evacuate chamber for 5 min.

8. Fill cooling finger of the sublimator with ice, and add 50 ml water. Allow to settle for a further 5 min.

9. Carefully lower chamber onto the sand bath, and leave crystals within the chamber to sublimate onto the slide for 17 min.

10. Once completed, raise sublimator from the sand bath. Turn off vacuum source, and quickly disassemble chamber.

    In order to prevent physical damage to the slide surface or condensation forming on it, the vacuum should be turned off first. The vacuum release valve is opened slowly before the chamber is quickly disassembled. The slide should be immediately removed to prevent moisture in the air condensing on the surface.

11. Re-weigh slide to ensure that there is coverage (mass of matrix bound to the slide surface equivalent to 0.2 mg/cm² of slide area; 4 mg total for a standard 7.5-cm × 2.5-cm ITO slide).

    Samples (slides with mounted tissue sections and matrix, ready for MALDI) can be stored for up to 2 days prior to analysis as long as they are kept under vacuum in the dark.

12. If coverage is insufficient, re-sublimate sample by repeating steps 4 to 9.

Figure 1.

(A) Glass petri dish with even covering of DHB matrix crystals. (B) Petri dish placed in bottom half of sublimator. (C) Glass sample slide fixed to the top of cooling finger of sublimator with copper tape.

Figure 2.

Assembled sublimation apparatus, suspended above the sand bath and connected to vacuum source.

SUPPORT PROTOCOL 2

RECRYSTALLIZATION OF MALDI MATRIX

To obtain the best possible purity of matrix and ensure that crystals sublime evenly, the matrix should be purified and recrystallized in batches prior to use. This process can also be used to increase the purity of inexpensive lower-grade DHB matrix.

Materials

• DHB matrix (e.g., Sigma-Aldrich)

• Milli-Q ultrapure water

• Activated carbon

• 2-liter conical flask

• Magnetic hot plate with stir bars

• 2-liter Buchner flask

• Rubber tubing

• Vacuum source

• 12-cm ceramic Buchner funnel

• Circle filter paper (e.g., Whatman Grade 541)

• Aluminum foil

• Vacuum desiccator

• Mortar and pestle

1. Dissolve 100 g DHB matrix in 500 ml boiling water in a 2-liter conical flask by heating the solution on a magnetic hot plate with stir bar until liquid boils and all crystals are dissolved.

2. Add 2% to 3% (w/w) activated carbon, and boil continuously for 5 min.

3. Connect Buchner flask to a vacuum source. Place funnel in the flask, and place a circle of filter paper inside the funnel.

4. Hot filter the solution to remove the activated carbon by pouring the solution onto the filter paper in the funnel while the vacuum is switched on.

   To prevent matrix crystals from crashing out in the filtration apparatus, pre-warm the apparatus by pouring 250 ml boiling water through the funnel.

5. Take the matrix solution (now in the Buchner flask), and re-boil on the magnetic hot plate with stir bar to dissolve any formed crystals. Transfer solution to a clean conical flask.

6. Once dissolved, wrap glass flask containing the solution with aluminum foil, and turn off the hot plate. Set the magnetic spinner to moderate speed.

   MALDI matrices are light sensitive and should be kept away from light sources as much as possible when not in use. Light exposure will decrease ionization efficiency of analytes resulting in low-intensity spectra.

7. Leave overnight to allow crystals to precipitate out.

8. Once precipitated, filter matrix crystals by pouring, and leave to dry in a vacuum desiccator wrapped in aluminum foil, ensuring it remains in the dark.

9. Crush dry matrix to a fine powder in a mortar and pestle. Aliquot into 250-mg masses, and store at −20°C until needed.

   The aliquots should last for up to 12 months in these conditions. If longer storage is required, aliquots can be stored at −80°C indefinitely.

BASIC PROTOCOL 3

MASS SPECTROMETRY ANALYSIS AND CALIBRATION

Accurate calibration of the instrument is critically important both for initial putative identifications of lipids and for further MS/MS analysis for definitive identification. This method focuses on using a particular instrument, the ultrafleXtreme mass spectrometer (Bruker Daltonik, Bremen, Germany), but the mass list used for calibration is applicable to any MALDI instrumentation platform.

Materials

• ITO slide with tissue sample (see Basic Protocol 2)

• Red phosphorus solution (see Support Protocol 3)

• 100% ethanol

• ultrafleXtreme MALDI TOF/TOF mass spectrometer

1. Mount ITO slide with tissue sample into the appropriate target plate slide holder for the mass spectrometer.

2. Pipette 2 μl red phosphorus solution onto one corner of the mounted tissue section not intended for imaging (Fig. 3).

   Effective calibration of time of flight is critical for accurately determining mass of the analytes and therefore assigning lipid identification. Red phosphorus needs to be on the tissue surface to ensure that the instrument is calibrated from the origin of the subsequent ion plume (and thus the beginning of the flight path). Variations in the thickness of tissue sections can affect the distance that ions have to travel, so that distance should be recalibrated using red phosphorus each time a sample is run. Reconfirming the masses of the red phosphorus once the flight distance has changed will ensure that the lipids that are subsequently analyzed will be correctly assigned.

3. Calibrate instrument using the red phosphorus [M+H] calibration list (see Table 1), and save the calibration profile.

   Not all peaks will be detected; however, it is common to get ~30 matched calibration peaks.

4. Once calibrated, image tissue at the desired pixel size. For a 1-cm × 1-cm tissue sample, it is typical to use 50 μm. If using the Bruker ultrafleXtreme, use the following settings:

   • Reflector positive mode
   • Laser power: 65%
   • Laser attenuation: 30%
   • Detector gain: 27×
   • Mass range: 400 to × 2000 M/z
   • Sample rate/digitizer: 1.25 GS/sec
   • Real-time smoothing: Off
   • Smartbeam parameter set: 2_small
   • Frequency: 1000 Hz
   • Laser shots: 500

5. Once imaged, rinse slides with ethanol to remove residual matrix, and if desired histologically stain for co-registration with cytoarchitectural or pathological features.

Figure 3.

A human post-mortem brain tissue section coated with DHB matrix with an area of red phosphorus at the opposite end of the imaged area (gray mesh square).

Table 1.

Red Phosphorus [M+H]⁺ Calibration List

Peak label	M/Z [M+H]	Tolerance (ppm)	Peak label	M/Z [M+H]	Tolerance (ppm)
RP1	31.9738	50	RP39	1208.9782	50
RP4	124.4952	50	RP40	1239.952	50
RP7	217.8166	50	RP41	1270.9258	50
RP10	310.738	50	RP42	1301.8996	50
RP11	341.7118	50	RP43	1332.8734	50
RP17	527.5546	50	RP44	1363.8472	50
RP18	558.5284	50	RP45	1394.821	50
RP19	589.5022	50	RP46	1425.7948	50
RP20	620.476	50	RP47	1456.7686	50
RP21	651.4498	50	RP48	1487.7424	50
RP22	682.4236	50	RP49	1518.7162	50
RP23	713.3974	50	RP50	1549.69	50
RP24	744.3712	50	RP51	1580.6638	50
RP25	775.345	50	RP52	1611.6376	50
RP26	806.3188	50	RP53	1642.6114	50
RP27	837.2926	50	RP54	1673.5852	50
RP28	868.2664	50	RP55	1704.559	50
RP29	899.2402	50	RP56	1735.5328	50
RP30	930.214	50	RP57	1766.5066	50
RP31	961.1878	50	RP58	1797.4804	50
RP32	992.1616	50	RP59	1828.4542	50
RP33	1023.1354	50	RP60	1859.428	50
RP34	1054.1092	50	RP61	1890.4018	50
RP35	1085.083	50	RP62	1921.3756	50
RP36	1116.0568	50	RP63	1952.3494	50
RP37	1147.0306	50	RP64	1983.3232	50
RP38	1178.0044	50	RP65	2014.297	50

SUPPORT PROTOCOL 3

RED PHOSPHORUS SOLUTION

Elemental red phosphorus is an ideal low-mass calibrant, as it is able to form an elemental cluster series without the need for any preparation. It provides >20 calibration points across the mass range of 150 to 1200 m/z. It is also able to ionize in MALDI without the addition of any matrix, meaning that it can also be used as a positive control for instrument performance.

Materials

• Elemental red phosphorus powder (e.g., Sigma-Aldrich)

• Methanol

• Mortar and pestle

• Glass scintillation vial ~22-ml)

1. Crush 100 mg red phosphorus into a fine powder in the mortar and pestle.

2. Transfer fine powder to a glass scintillation vial.

3. Add 500 μl methanol.

   The solution will settle very quickly, so the vial should be shaken to form a suspension immediately prior to use. There is no chemical interaction between the phosphorus and methanol; therefore the solution can be stored at room temperature indefinitely. If the methanol has evaporated completely, another 500 μl can be added. There is no limit to the amount of times this can be repeated.

BASIC PROTOCOL 4

ASSIGNMENT OF LIPID IDENTIFICATIONS

Tentative or preliminary identifications are generated by taking the reported lipid masses and searching them through the online database “Lipid Maps.” Once preliminary identifications are determined, further validation can be performed with complementary techniques such as liquid chromatography MS (LCMS) or by MS/MS. This protocol does not cover those techniques.

Materials

• Internet-connected computer

1. Perform lipid searching by first navigating to the online database “Lipid Maps” available at http://www.lipidmaps.org/.

2. Select the Lipid Maps Structure Database by clicking on “Browse LMSD,” and select the “Text/ontology-based search.”

3. Enter the mass of the detected molecule into the “Mass” text box.

4. Select a tolerance of 0.1 Da.

   This tolerance is independent of instrument settings used and will simply search for the entered mass +/− 0.1 Da.

5. Click “submit” to view the results.

   If more than one class of lipids is returned, then additional biological information such as the particular section of brain or other information about tissue type (or known pathology) can be used to determine the correct class. If this is still not feasible, then the identification is considered tentative pending additional experimentation.

COMMENTARY

Background Information

The protocols presented here are in regular use in the Mass Spectrometry Core Facility at the University of Sydney. The method is an adaption of the peptide-focused method first described in 2011 by Yang and Caprioli that used sublimation to apply the MALDI matrix (Casadonte & Caprioli, 2011). We have continually expanded on their initial work, and aspects of a modified technique have been previously published in two reports. The first report refined the reproducibility of the protocol, while the second characterized the key steps behind successful sample preparation (O’Rourke, Djordjevic, et al., 2015; O’Rourke, Raymond, et al., 2015).

Our gradual evolution of peptide-based MALDI-IMS has now been distilled into the analysis of lipids sharing the same levels of reproducibility and robustness previously reported. This article provides a comprehensive description of this validated and definitive protocol. It describes a flexible method that can be easily modified by other research teams to suit their needs with no individual step requiring specialized equipment or instrumentation, thereby keeping costs to a minimum. Lipid-rich tissues such as the brain are particularly amenable to MALDI-IMS, but it is also amenable in organs where there is a pathological build-up of lipid such as atherosclerosis (Martin-Lorenzo, Alvarez-Llamas, McDonnell, & Vivanco, 2016), aneurysms (Tanaka et al., 2015), colon cancer (Hiraide et al., 2016), and non-alcoholic fatty liver disease (Alamri et al., 2018).

Critical Parameters and Troubleshooting

Matrix sublimation

The even distribution of matrix in the petri dish (Basic Protocol 2, steps 3 and 4) is very important as uneven distribution will cause uneven deposition on the sample slide. This will then have a negative effect on the intensity of the resulting ions. Less matrix will decrease ionization whereas more will lead to increased intensity in ionization. The end result of an unevenly coated slide will be data that suggest that ions are in greater abundance in specific areas of the tissue compared to other areas. This could mislead a user to think that this is biologic in nature rather than an incorrectly prepared sample.

Identification of lipids

The preliminary identifications of lipids with “Lipid Maps” should not be assumed to be definitive. To fully identify the imaged lipid species, complementary LCMS or MS/MS should be used.

Anticipated Results

In human post-mortem brain tissue samples, differential lipid distribution can be most easily demonstrated between cortical gray matter and the underlying myelin-rich white matter. In neurodegenerative disorders such as AD there is considerable loss of susceptible neurons in gray matter areas such as the inferior temporal gyrus (ITG), making interpretation of case-control comparisons of post-mortem tissue difficult. However, mildly affected areas of the brain at post-mortem, such as the primary visual cortex, may act as a proxy for the ITG many years earlier in the disease process (Sutherland, Janitz, & Kril, 2011). Apart from case-control analyses, MALDI-IMS lipid data can be correlated with neurons, plaques, or neurofibrillary tangle densities in a particular region or within case comparisons carried out between regions differentially affected by the pathology to model the role of specific lipids in disease progression.

Using these protocols, calibrated spectra of lipids with clearly defined red phosphorus peaks and low ppm error across the entire mass range can be consistently generated (see Fig. 4). Ions of possible interest that might arise from lipids of a known mass or novel ions based on their differential intensity can be selected from spectra and displayed as heat maps (Fig. 5). Lipids can be subsequently identified through interrogation of databases such as http://www.lipidmaps.org. The effectiveness of calibration in this protocol allows identities to be assigned confidently. For example, the putative sterol (m/z 428.182) can be reported with an error of ~20 ppm. Albeit, positive identification of specific lipids remains within the limits imposed by the current levels of annotation in available databases (Fig. 5).

Figure 4.

Representative global lipid spectrum showing high-intensity peaks across the mass range.

Figure 5.

Heat map images generated within SCiLS Lab software (Bruker) show differential expression of two lipid species in the white and gray matter of the inferior temporal gyrus (ITG) from a neurologically normal individual (control) and a patient with Alzheimer’s disease (AD). (A) Grayscale images indicate where on the tissue section MALDI-IMS was undertaken. The abundance of m/z 428.182 (B), a putative sterol, is relatively low in the gray matter of the AD ITG, whereas m/z 689.660 (C), a putative phosphatidylcholine, has similar abundance in the white and gray matter of both individuals. Color thresholding has been carried out for visualization purposes with warmer colors representing higher abundance of ions.

Time Considerations

Preparation of nitrocellulose-coated slides takes 20 to 30 min. Cryosectioning and mounting of tissue requires ~1 hr depending on user skill. Sublimation of tissue, including setup, requires 60 to 90 min. For analysis in instrument, typically, a space of 2-cm × 2-cm will take ~10 hr at a resolution of 50 μm. Data analysis requires ~1 hr for 40 identified lipids.