Improved matrix coating for positive- and negative-ion-mode MALDI-TOF imaging of lipids in blood vessel tissues

High-quality matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI MSI) of lipids in biological tissue relies on the fabrication of a homogeneous matrix coating featuring best possible analyte integration. This communication addresses a matrix vapor deposition/recrystallization process for the application of 1,5-diaminonaphthalene (1,5-DAN) onto slices of human aortic tissue. The matrix coating is compatible with both positive- as well as negative-ion-mode MALDI MSI facilitating a significantly enhanced detection of lipid-related signals in different cell layers of blood vessel walls. Graphical abstract Electronic supplementary material The online version of this article (10.1007/s00216-019-01826-x) contains supplementary material, which is available to authorized users.


Matrix vapor deposition and recrystallization on small substrates
Matrix deposition on small sized stainless steel or ITO targets was carried out using a commercially available sublimation apparatus ( Figure S1). The apparatus was filled with about 200 mg of 1,5-DAN, while the stainless steel or ITO target with an aortic tissue section was fixed at the cooling finger using adhesive tape. The sublimation apparatus was coupled to a rough pump, a liquid chiller and was placed in an oil bath. Sublimation was performed at an oil bath temperature of 144 °C and a rough vacuum <2x10 -3 mbar for 2.5 minutes. The sublimation apparatus was then opened, the matrix removed and the flask was filled with 2 mL of e.g. acetonitrile, chloroform or toluene. The cooling finger was inserted into the flask with the boiling solvent to expose the matrix coating to the vapor for about 10 seconds. The metal slide was fixed on a MTP ground steel target using an UV curing glue and directly transferred into the vacuum system of the mass spectrometer.

Fig. S1
Experimental setup for matrix vapor deposition/recrystallization on small sized stainless steel or ITO glass slides. The commercially available sublimation apparatus on top of an oil bath, the attached rough pump and the chiller are not shown (1); small sized stainless-steel target with an aortic tissue section fixed at the cooling finger using an adhesive tape (2); small sized stainless-steel target fixed on the stainless steel MALDI target using an UV curing glue (3)

Matrix vapor deposition and recrystallization on ITO slides
Matrix deposition on ITO glass slides was carried out using in-house built sublimation apparatus ( Figure S2). A petri dish on top of a heating plate was filled with about 200 mg of 1,5-DAN, while the ITO slide with an aortic tissue section was fixed on the cooling plate using adhesive tape. The sublimation apparatus was closed and coupled to a rough pump and a liquid chiller (15°C). Sublimation was performed within 5 minutes at a rising heating plate temperature between 123 -136 °C and a rough vacuum <2x10 -3 mbar. According to a method described by Yang et al. [1], the amount of deposited matrix was determined to be 0.111±0.004 mgxcm -2 . The sublimation apparatus was then opened, the petri dish with the matrix removed and the hot plate quickly cooled down to about 20°C above the boiling point of the solvent for recrystallization. Another petri dish filled with 5 mL of e.g. acetonitrile, chloroform or toluene was placed on the hot plate and the sublimation apparatus was closed for about 30 seconds to expose the matrix coating to the vapor of the boiling solvent. The ITO slide was fixed in a Bruker MTP Slide Adapter II and directly transferred into the vacuum system of the mass spectrometer.

Fig. S2
Experimental setup for matrix vapor deposition/recrystallization on standard ITO glass slides. The in-house built sublimation apparatus on top of a commercially available hot plate, the attached rough pump and the chiller are not shown (1); ITO glass slide with an aortic tissue section fixed at the cooling plate using adhesive tape (2); ITO glass slide inside the Bruker MTP Slide Adapter II (3)   Table S1 Validation of different sample preparation procedures. Open source mass spectrometry tool mMass [2] was used to process average spectra (whole sample). After automated peak picking (with a signal to noise threshold >6) and deisotoping a Kendrick mass defect plot [3] was utilized to determine the number of lipid related signals as well as to exclude signals, which are obviously not related to lipid compounds  Pheophytin "a", a colored natural product, which is perfectly soluble in toluene, was dissolved in toluene and spotted onto four different ITO glass slides. After 5 minutes of DAN matrix vapor deposition microscopic images of selected spots were taken (panels A, D, G and J). After 45 sec, 30 sec and 15 sec of matrix recrystallization using toluene further microscopic images of the same spots were taken (panels B, E and H) and MALDI MSI in the positive ionmode (as described in Materials and Methods part of the main text) was performed at a lateral resolution of 35 µm: (C) 45 sec recrystallization using toluene resulted in an unacceptable lateral analyte diffusion of approx. 100 µm; (F) optimum conditions were found at 30 sec recrystallization resulting in an acceptable lateral diffusion below 20 µm and high signal-tonoise ratios throughout the sample; (I and K) very short or no recrystallization showed no lateral analyte diffusion on the one side, but caused signal suppression due to lack of orthogonal analyte integration at the peripheral zones of the pheophytin spots

Fig. S7
Effect of recrystallization on the signal-to-noise ratio. For a better comparison two adjacent cryosections of human aortic tissue (8 μm thick) were thaw-mounted on the same ITO slide and coated with 1,5-DAN. While one section was covered with a piece of paper, the coating of the second section was recrystallized using toluene. The samples were subjected to MALDI MSI analysis and rastered in positive as well as negative ionmode under equal conditions (identical laser fluence; lateral resolution was set to 70 µm; 20 shot per spot; random walk after 5 consecutive shots). Average spectra were obtained from about ¼ of each aortic section comprising all layers of the tissue (black line = without recrystallization; red line = with recrystallization). In the negative ionmode (top) absolute signal intensities of lipid related signals were found to be increased by 30-65%, in the positive ion-mode (bottom) absolute signal intensities of lipid related signals were found to be increased by 5-23%

Processing MSI data and data visualization
MSI raw data were first exported to the common Analyze 7.5 format using FlexImaging (Bruker Daltonics) and then processed utilizing the open-source software mMass [2] as well as MALDIquant (Version 1.16.2) [4] and Cardinal (Version 1.8.0) [5] packages for the R environment. An in-house R script was developed to generate average spectra, preprocess and visualize the MSI data set.  Table S2. All images are plotted using a color scale from black (0%) to red (100%). Corresponding negative ion-mode images are depicted in Fig. S9B  Fig. S9B Consecutive negative ion-mode MALDI-TOF MSI of lipids in human aortic tissue. A cryosection (8 μm thick) was thaw-mounted on a stainless steel slide and coated with 1,5-DAN. After matrix recrystallization using toluene the sample was rastered in the positive and then, with an offset of approx. 50µm, in the negative mode at a lateral resolution of 110 μm (laser spot size was set to 35 μm). From top left to bottom right: All negative mode ion images of lipids listed in Table S3. All images are plotted using a color scale from black (0%) to red (100%). Corresponding positive ion-mode images are depicted in Fig. S9A

MALDI MSI: Assignment of lipids
Software mMass [2] was used to process average spectra obtained from an aortic tissue section washed with NH4HCO2 solution prior to DAN matrix sublimation and recrystallization using toluene. After automated peak picking (signal-to-noise threshold >6) and deisotoping a Kendrick mass defect plot [3,6,7] was utilized to identify lipid related signals. The signal assignment is based on a study from Doppler et al. [8], who published a quantification of metabolites in the same aortic tissue samples using AbsoluteIDQ1 p150 kit (BIOCRATES Life Sciences AG, Innsbruck, Austria) and was further validated by an LC-MS/MS analysis (see below).

Identification of Lipids in human aortic tissue using LC-MS/MS
Lipid extraction of human aortic tissue slices Two human aortic tissue slices (8 µm each) were detached from glass slides using a small spatula and transferred into an Eppendorf microcentrifuge tube. 100 µL of MeOH were added and the suspension was sonicated for 5 minutes and vortexed for additional 20 sec. The suspension was then centrifuged at 13400 rpm for 10 min. The lipid containing organic supernatant was transferred into a fresh tube and stored at -20°C until use.