Molecular Profiling of Glioblastoma Patient-Derived Single Cells Using Combined MALDI-MSI and MALDI-IHC

Abstract

In recent years, mass spectrometry-based imaging techniques have improved at unprecedented speeds, particularly in spatial resolution, and matrix-assisted laser desorption/ionization (MALDI) mass spectrometry imaging (MSI) experiments can now routinely image molecular profiles of single cells in an untargeted fashion. With the introduction of MALDI-immunohistochemistry (IHC), multiplexed visualization of targeted proteins in their native tissue location has become accessible and joins the suite of multimodal imaging techniques that help unravel molecular complexities. However, MALDI-IHC has not been validated for use with cell cultures at single-cell level. Here, we introduce a workflow for combining MALDI-MSI and MALDI-IHC on single, isolated cells. Patient-derived cells from glioblastoma tumor samples were imaged, first with high-resolution MSI to obtain a lipid profile, followed by MALDI-IHC highlighting cell-specific protein markers. The multimodal imaging revealed cell type specific lipid profiles when comparing glioblastoma cells and neuronal cells. Furthermore, the initial MSI measurement and its sample preparation showed no significant differences in the subsequent MALDI-IHC ion intensities. Finally, an automated recognition model was created based on the MALDI-MSI data and was able to accurately classify cells into their respective cell type in agreement with the MALDI-IHC markers, with triglycerides, phosphatidylcholines, and sphingomyelins being the most important classifiers. These results show how MALDI-IHC can provide additional valuable molecular information on single-cell measurements, even after an initial MSI measurement without reduced efficacy. Investigation of heterogeneous single-cell samples has the potential of giving a unique insight into the dynamics of how cell-to-cell interaction drives intratumor heterogeneity, thus highlighting the perspective of this work.

Figure 1

Optimized workflow for single cell MALDI-IHC and molecular profiling on a single slide. First, lipid profiles are obtained from the PDCL GBM single cells using high-resolution MALDI-MSI. After matrix removal, the cells are stained with cell-specific MALDI-IHC probes, for cell characterization. Fluorescent markers on the probes allow for optical confirmation of cellular locations. Finally, MALDI-IHC and MALDI-MSI images can be overlaid to coregister single cells between measurements and extract cell type specific lipid spectra. Created in BioRender.com.

Figure 2

Patient-derived glioblastoma tumor cells stained with a 14-plex MALDI-IHC antibody panel. (A) Three different cell markers are highlighted: GLUT1, for GBM cells, in red, NF-L for neurofilaments in neurons, in yellow and SYN-I, as a synaptic marker, in blue. Pixel size = 5 × 5 μm². (B) Single pixel spectra from three different cells stained by the markers shown in (A). Color-coded accordingly. (C–F) Shows the effect of prior measurements on the efficacy of subsequent MALDI-IHC measurements. (C–E) MALDI-IHC images of the mass corresponding with GLUT1 in cells that had undergone no prior treatment (C), cells that had undergone MALDI-MSI sample prep (D) and cells that had been measured with MALDI-MSI prior to the MALDI-IHC procedure (E). Pixel size = 5 × 5 μm². (F) Mean peak areas, based on 10 single cells each, for each marker and condition. Each ROI, indicative of one cell, was manually determined. One-way ANOVA revealed no significant difference between the 3 groups (p = 0.73). Error bars indicate SE. Intensity of the markers NeuN, IBA-1, nicastrin and MAP2 were too low to detect single cells. All data presented in Figure 2 is obtained from n = 1 cell covered ITO slide as described in Supporting Figure S1. Replicate MALDI-IHC measurements can be seen in Supporting Figure S2. Data to support Figure 2F can be seen in Supporting Tables S2 and S3.

Figure 3

Multimodal imaging of single cells. (A–D) MALDI-IHC stained single cells (A + C), visualized with the marker for neurofilaments (NF-L) were correlated to images of the same cells obtained with fluorescence imaging (B) and MALDIMSI lipid imaging of m/z 754.5 (PC 34:4) (D). Two different field-of-views are visualized (A + B, C + D). Fluorescence images were detected in the range of 478–533 nm. MALDI-MSI images were obtained in positive-ion mode with a pixel size of 10 × 10 μm². MALDI-IHC images were obtained at 5 × 5 μm² pixel size. White and green arrows indicate correlating cells between images. Lipid IDs are based on LC-MS/MS from GBM tissue, as described in.

Figure 4

Molecular profiling of single cells. (A, B) MALDI-MSI images of lipid distributions in single cells. Pixel size = 10 × 10 μm. (C) MALDI-IHC image of single cells visualizing markers for GLUT1 (green) and NF-L (purple). Pixel size = 5 × 5 μm². Single cells were correlated to lipid distributions in cells obtained with MALDI-MSI (white and green arrows). The white arrow indicates a cell correlating with the NF-L marker, while the green arrow indicates a cell correlating with GLUT1. The correlated cells show a different lipid expression profile, highlighted by m/z 780.5 (PC 36:5) not being detected in the GLUT1 expressing cell (green circle, (A, B)). (D) Mass spectra from each of the correlated cells with select masses highlighted as different between the spectra. Lipid IDs are based on LC-MS/MS from GBM tissue, as described in.

Figure 5

Automatic cellular recognition model of neuronal and glial cell types. (A) MALDI-MSI lipid distribution of single cells, with recognition model labels overlaid on each cell. Each cell-related ROI for the model was selected based on the lipid distribution. Blue pixels indicate a GLUT1-associated (GBM cells) spectrum. Cyan pixels indicate a GFAP-associated (astrocyte) spectrum. Orange pixels indicate an NF-L-associated spectrum. White pixels indicate poly-l-lysine-associated spectrum, indicative of background signal. (B–D) Cell type specific spectra from each recognized marker, respectively. For each cell type, an example of the recognition model overlay for a single cell is shown (left small image), as well as an overlay of the MALDI-IHC marker (cool-to-warm scale), on the MALDI-MSI (viridis-scale) image (right small image). Pixel size = 10 × 10 μm².