Gel-assisted mass spectrometry imaging enables sub-micrometer spatial lipidomics

Abstract

A technique capable of label-free detection, mass spectrometry imaging (MSI) is a powerful tool for spatial investigation of native biomolecules in intact specimens. However, MSI has often been precluded from single-cell applications due to the spatial resolution limit set forth by the physical and instrumental constraints of the method. By taking advantage of the reversible interaction between the analytes and a superabsorbent hydrogel, we have developed a sample preparation and imaging workflow named Gel-Assisted Mass Spectrometry Imaging (GAMSI) to overcome the spatial resolution limits of modern mass spectrometers. With GAMSI, we show that the spatial resolution of MALDI-MSI can be enhanced ~3-6-fold to the sub-micrometer level without changing the existing mass spectrometry hardware or analysis pipeline. This approach will vastly enhance the accessibility of MSI-based spatial analysis at the cellular scale.

Fig. 1. Gel-assisted mass spectrometry imaging (GAMSI).

Schematics showing the principle and generalized workflow of GAMSI. Samples are (1) treated (optionally) with cleavable small-molecule linkers and polymerized to form a superabsorbent hydrogel composite, (2) homogenized and expanded, (3) immobilized onto a sample plate, (4) subject (optionally) to cleavage of the small-molecule linkers to release the analytes via, for example, a photocleavage event (dotted box), and (5) coated with a matrix (if used) and analyzed on a mass spectrometer. The magnified schematics (solid circles) show the reversible tethering/release of biomolecules and molecular probes to/from the hydrogel polymer chains in the sample-hydrogel composite.

Fig. 2. Lipid GAMSI.

a Fluorescence intensity per normalized unit tissue area of fresh-frozen and GAMSI-processed (PFA/GA-fixed) mouse cerebellum white matter, fluorescently labeled using a phospholipid dye [bar height, mean; black dots, individual data points; error bar, standard error of the mean (SEM); n  =  15 regions of interest (ROIs) from three brain slices from one animal]. The unit tissue area was normalized to the pre-expansion scale. b Averaged mass spectra (m/z = 650-925) of fresh-frozen and GAMSI-processed (PFA/GA-fixed) mouse cerebellum. c Spatial distributions of selected lipids in the fresh-frozen (left) and GAMSI-processed (PFA/GA-fixed, ~3-fold expanded) (right) mouse cerebellum. The instrument pixel size (raster distance) (AB SCIEX 4800) was set at 100 µm. Scale bar: 1 mm (3 mm). Here and after, unless otherwise noted, color scale bars for mass spectrometry images represent the relative intensity of the signals, m/z (mass-to-charge ratio) values are provided for singly charged deprotonated ions [M-H]^-, and scale bars are provided at the pre-expansion scale (with the corresponding post-expansion size indicated in the brackets). The instrument pixel size refers to the physical size of a pixel as determined by the instrument specifications, whereas the effective pixel size is given as the instrument pixel size divided by the sample expansion factor. PC: phosphatidylcholine; PE: phosphatidylethanolamine; PI: phosphatidylinositol. Source data are provided as a Source Data file.

Fig. 3. GAMSI is applicable to any MSI pipelines.

Spatial distributions of selected lipids in (a) fresh-frozen and (b) GAMSI-processed (PFA-fixed, ~4-fold expanded) mouse cerebellum. Both samples were imaged with the instrument pixel size (Bruker rapifleX, with 9AA as the default matrix for lipid imaging) set at 50 µm. Scale bars: 500 µm (1.9 mm). DG: diacylglycerol; PS: phosphatidylserine; PA: phosphatidic acid; PE: phosphatidylethanolamine; PG: phosphatidylglycerol; PI: phosphatidylinositol; SHexCer: sulfatide hexosylceramide; HexCer: hexosylceramide; ST: sterol; TG: triglyceride.

Fig. 4. Multiplexed lipid-protein GAMSI.

Spatial distributions of (a–b) mass reporters of photocleavable mass-tags (PC-MTs) targeting MBP and SYN-I, and (c–t) selected lipids in a ~ 4-fold expanded mouse cerebellum lobule (PFA-fixed). u–w Overlays of proteins and selected lipids. The instrument (Bruker rapifleX) pixel size was set at 50 µm. Scale bars: 200 µm (760 µm). MBP: myelin basic protein; SYN-I: Synapsin I; DG: diacylglycerol; PS: phosphatidylserine; PA: phosphatidic acid; PE: phosphatidylethanolamine; PG: phosphatidylglycerol; PI: phosphatidylinositol; SHexCer: sulfatide hexosylceramide; HexCer: hexosylceramide; ST: sterol; TG: triglyceride.

Fig. 5. Lipid GAMSI at sub-micrometer resolution.

a Lipid GAMSI of ~4-fold expanded mouse cerebellum (PFA-fixed) with the instrument (Bruker rapifleX, with DAN as the matrix) pixel size set at 10 µm (left). The effective pixel size is ~2.5 µm. Fluorescence image of the same tissue slice (immunostained against calbindin) and an overlay of the GAMSI and fluorescence images are shown to the right. The arrows in the fluorescence image indicate individual Purkinje cells. Scale bars: 25 µm (100 µm). The experiment was repeated twice. b Lipid GAMSI of ~4-fold expanded mouse cerebellum (PFA-fixed) with the instrument (Bruker timsTOF) pixel size set at 20 µm (bottom), 10 µm (top left), and 5 µm (top right). The corresponding effective pixel size is ~5 µm, ~2.5 µm and ~1.3 µm, respectively. Scale bar: 500 μm (1.9 mm). (Insets) Magnified views of the boxed regions. Scale bars: 50 µm (190 µm). The apparent tissue tears are from cryo-sectioning. Unless otherwise noted, all the Bruker timsTOF images were collected using MALDI-1 mode. c Lipid GAMSI of ~6-fold expanded mouse cerebellum (PFA-fixed) with the instrument (Bruker timsTOF) pixel size set at 5 µm. The effective pixel size is ~880 nm. Scale bars: 100 µm (570 µm), and 30 µm (171 µm) for the magnified view of the boxed region in the overlay image. The apparent tissue tears are from cryo-sectioning. The experiment was repeated once. WM: white matter; GL: granular layer; ML: molecular layer; PC: Purkinje cell; PI: phosphatidylinositol; PA: phosphatidic acid; ST: sterol; PI-Cer: ceramide phosphoinositol.