Chemicals. Lipid standards 1‑palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine (sodium salt) (PS(16:0/18:1)) and 1‑palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-phosphocholin (sodium salt) (PC(16:0/18:1)) were from Avanti Polar Lipids (Alabaster, AL), deuterated methanol (MeOD; 99 % D) was from Acros Organics (Geel, Belgium). All other chemicals were purchased from Merck/Sigma-Aldrich (Steinheim, Germany).
Tissues. Mouse brain and kidney were dissected from 10–15-week-old female C57BL6/J mice, according to approved protocols. Whole organs were embedded in 2‑hydroxyethylcellulose (Mavg ~1,500,000 g/mol) and then snap-frozen in liquid N2. Pig brain was from a local butchery and bulk homogenate was prepared as described. 11 Briefly, brain tissue was snap-frozen in liquid N2 and 7 g of homogenized tissue was mixed with 3 g of aqueous 2-hydroxyethylcellulose polymer solution. Tissue sections of 10 µm thickness (mouse) or 20 µm thickness (pig brain homogenate) were produced with a cryotome (Jung Frigocut 2800E, Leica Biosystems, Jena, Germany) and thaw-mounted on histological glass slides (SuperFrost, Fisher Scientific, Schwerte, Germany).
Lipid standards were prepared to 1 µmol/L in 70 % acetonitrile and evenly coated onto histological glass slides using an ultrasonic sprayer (SimCoat, Sono-Tek, Milton, NY). 400 µL of standard solution were sprayed at a flow rate of 0.03 mL/min in meandering patterns at a distance of 1.8 mm between the lines. Tissue sections and slides with standards were stored at −78 °C until further use.
Deuterated matrix was synthesized by dissolving 0.2 g of DHAP in 30 mL of a mixture of deuterated methanol (MeOD) and acetone-d6 (50:50, v:v), followed by removing the solvents in a rotary evaporator (VV2011, Heidolph Instruments, Schwabach, Germany). To ensure a close to complete degree of deuterium exchange, the process was repeated three times so that in the later mass spectrometric analysis no protonated signal of the matrix was detected. The crystalline deuterated matrix was stored at -20 °C in a sealed glass flask until further use. During synthesis and storage, the deuterated matrix was kept under dry argon.
After removal from the freezer, tissue sections and standards were thawed under a stream of nitrogen before coating with the MALDI matrix in a home-build matrix sublimation chamber. 10 1.2 mL of matrix solution (DHAP: 7 mg/mL in 70 % acetonitrile (ACN); deuterated DHAP: 7 mg/mL in MeOD/acetone-d6 (50:50, v:v) were filled into the matrix reservoir of the sublimation chamber and the reservoir was heated to 135 °C. The sample slide was mounted onto a cooling socket at about 3.5 °C. The sublimation chamber was evacuated to approximately 10‑4 mbar. To stop the sublimation process, the chamber was flushed with nitrogen. 10 MALDI samples were recrystallized for 2.5 min at 70 °C in an atmosphere containing 0.5 % ethanol in H2O.
Mass spectrometer. A Q Exactive Plus Orbitrap (Thermo Fisher Scientific, Bremen, Germany), coupled with a dual-ion funnel/dual MALDI/ESI Injector (Spectroglyph, Kennewick, WA), was used as the mass analyzer. The ion source has previously been modified to enable laser-based MALDI-2. 7,14 A modified version was also used in our initial LPPI work with (s)VOCs. 25
A frequency-tripled q-switched Nd:YLF laser (Explorer One, Spectra-Physics, Mountain View, CA; emission wavelength: 349 nm; pulse width: 7-10 ns; pulse repetition rate frep, adjustable from 1 up to a maximum of 5 kHz) was used as the MALDI laser and for material ablation in the MALDI-SPICI experiments. In the presented spectra, the ablation laser was operated with a repetition rate of 300 Hz.
The default lens that is used in the Spectroglyph ion source to focus the laser beam on the target was replaced by two CaF2 planoconvex lenses (Thorlabs, Dachau, Germany) with focal lengths of 300 mm and 1000 mm. By this replacement, the combined focal length of the optical system fits the requirements of the source more precisely and enables the production of an effective spot size of ~9 µm (defined as the area of visible material ejection).
To reduce background ion signal levels, throughout our experiments the ESI inlet of the MALDI/ESI Injector was blocked with a polytetrafluoroethylene (PTFE) plug. A fine-needle valve (SS-1RS6MM, 0.37 Cv, Swagelok, Düsseldorf, Germany) was used to adjust the N2 buffer gas pressure in a range of 4-12 mbar in the region of the primary ion funnel. Optionally, dopant vapor from the headspace of a sealed reservoir flask was introduced into the ionization chamber via a PTFE tubing (outer diameter, 3/8”) and Swagelok® fittings. A second needle valve (SS-ORS3MM, 0.09 Cv, Swagelok, Düsseldorf, Germany) was used to manually control the gas flow of dopant into the funnel. For one experiment, another PTFE tubing connected a D2O chamber with the ion source in a similar manner. The D2O chamber contained a glass vial filled with D2O. The chamber was evacuated, closed by another needle valve, and the vial was broken inside the closed chamber. The pure D2O vapor was introduced at controlled amounts through the valve. The determination of the partial pressure of the introduced gases was based on an approximation by monitoring the Pirani pressure gauge of the Spectroglyph system.
During the experiments, three different mass resolving powers of the Orbitrap of Resm = 70,000, 140,000, and 280,000 (each defined for an m/z value of 200) were used. The “injection time” was set to a fixed value of 250, 500, and 900 ms for the three resolution values, respectively, resulting in data acquisition rates of 3.7, 1.9, and 0.97 pixels per second. The “AGC target” was disabled.
For higher-collision induced dissociation (HCD) tandem MS measurements, the analytical quadrupole was set to an isolation window of 0.8 Da and the collision energy (CE) was varied between 12 and 30 eV (laboratory frame). For data-dependent acquisition (DDA) of a coronal brain section, the “dynamic exclusion” was set to 9 s (for brain homogenate to 45 s) and the ACG threshold to 2 x 104. An “exclusion list” was generated prior to each experimental run with background signals of a blank spectrum acquired at equilibrated measurement conditions.
VUV module. The principle layout of the VUV SPI module has been described in detail previously. 25 A schematic of the design with measures is plotted in Figure S-1, a condensed sketch in Figure 1. Briefly, three RF-Kr discharge lamps (model PKR-106-6-14, Heraeus Noblelight, Hanau, Germany) were integrated into a custom-designed annular mount made of PEEK. The small size of the cylindrically shaped lamps with outer dimensions of 6 × 14 mm (diameter × length) allowed for positioning the lamps with optimized geometries in order to minimize perturbation of the electrical field in the central part of the ion funnel device. The distance of the emission side of the lamps to the MALDI sample was about 1 mm and the distance to the central axis of the ion funnel was ~5 mm. Adhesive copper band served to contact the lamps. The position of the electric contacts on the three lamp bodies – especially the distance between the two electrodes – needed to be essentially similar for all lamps in order to obtain homogeneous response characteristics upon pulsed electric excitation. A symmetric ring electrode connected the front-end (emission side) of the lamps serving as ground for excitation and ion extraction. A custom-made class E amplifier 25 operated the lamps via the copper electrodes at the back-end with an alternating current (AC) of 13.560 MHz and a peak-to-peak voltage of 220 V. A metal-oxide–semiconductor field-effect transistor (MOSFET; model IXZ631DF18N50, IXYS, Milpitas, CA) was triggered by an RF generator (TG2000 DDS function generator 20 MHz, Aim TTi, Huntingdon, UK) and induced a resonance circuit, which operated the lamps. Importantly, all electronic components were mounted outside the vacuum to minimize the effect of stray fields on the ion funnel. A direct current (DC) power supply (BA‑315, Bertan Associates Inc., Syosset, NY) was used to supply the symmetric ground electrode at the lamp’s front-end with a variable DC voltage of 300-500 V.
As an important modification to our previous setup, 25 the RF generator was gated by a rectangular wave function generated by the internal pulse generator of an oscilloscope (InfiniiVision DSOX3032A, 350 MHz, 4GSa/s; Keysight Technologies, Santa Rosa, CA). The repetition rate was variable between 10 to 5000 Hz. This enabled the generation of pulsed VUV light with electric pulse widths between 70 µs and the continuous wave (cw) mode without modification of the in-source hardware; the minimum pulse width was found to ignite the lamps. The function generator also served for synchronization of ablation and VUV pulses with variable repetition rates up to 5 kHz; the latter corresponds to the maximum pulse repetition rate of the laser; we note, the VUV lamps themselves could be operated with a frequency up to the continuous wave mode. A custom-made delay generator served for adjusting the delay between the actively q‑switched Nd:YLF laser and the VUV pulses.
Another key feature for achieving constant signal intensities is the symmetric design of the central ring electrode (see Figure S-1, SI for details), which connects the VUV lamp’s tips. By applying a constant DC voltage to the part of the lamps facing the plume, the electrodynamic field of the ion funnel device is affected. With a controlled symmetric “offset” being added, reproducible ion trajectories can effectively be realized. Generally, the applied DC voltage Uxtr as well as the total gas pressure Ptotal should be kept as constant as possible to ensure controlled and reproducible ion trajectories and thus to avoid variations in the ion counts or ion profiles. As shown in Figure 1f, a slight variation of Uxtr and Ptotal has only little effect on the observed ion counts and thus the exact optimization of these values plays only a minor role in the successful MALDI-SPICI experiment.
Data analysis. Xcalibur (vs. 4.0, Thermo Fisher Scientific) was used for the initial data examination and optionally for off-line recalibration of mass spectra. MSI data sets were converted to imzML-files by using ProteoWizard software 36 and uploaded to LipostarMSI (vs. 1.1; Molecular Horizon, Bettona, Perugia, Italy) for further graphical analyses. 28 Throughout this work all full scan ion images are presented using the viridis false-color scheme. The mass traces in the full scans of the DDA images were visualized in the rainbow false-color scheme and extrapolated to a five-pixel wide block by manually copying the information of signal intensities from the full scan pixels to the subsequent four pixels in the MSI picture file. MS/MS data were automatically annotated in LipostarMSI based on the DB manager – a complementary tool to LipoStar used to generate tandem MS data lists based on proprietary fragmentation rules. 37 Data sets of brain homogenate were uploaded to MZmine 2 software 38 for signal annotation of aligned lists with the tool “lipid search” and the integrated theoretical database 39. Lipid identification was furthermore conducted via the online classification system of LIPID MAPS® 28 and by comparison with literature data. 16,29-32 All assignments are based upon using a mass tolerance of 1.5 ppm for R = 280,000 and 2 ppm for the lower mass resolution settings and were in several cases further confirmed by DDA.
Statistics and reproducibility. We recorded multiple measurements of different sample sections in both positive and negative ion mode within a period of about 12 months and could achieve a high degree of reproducibility among the different measurements.
MS measurements from brain homogenate in both ion modes were conducted with at least three independently prepared samples, which were yielding comparable results; e.g. a relative standard deviation (RSD) of ~20 % in the intensities of annotated signals between measurements within 6 months and an according RSD <3.5 % within one set of measurements. For MSI experiments from brain tissue, three tissue sections from three different animals were analyzed in the positive ion mode yielding comparable results. Kidney was taken from one animal, only. Thus, in this case, only technical replicates were produced, again however yielding very comparable results. Lipid annotation was performed from averaged mass spectra of at least 10,000 pixels of the respective tissue sample and processed further as described above.
Data availability. The data that support the findings of this study, in particular all MSI data as presented in the main text, are available to download in the vendor-neutral imzML format from ____insert link____ or from the corresponding author upon request.