Deep topographic proteomics of a human brain tumour

The spatial organisation of cellular protein expression profiles within tissue determines cellular function and is key to understanding disease pathology. To define molecular phenotypes in the spatial context of tissue, there is a need for unbiased, quantitative technology capable of mapping proteomes within tissue structures. Here, we present a workflow for spatially-resolved, quantitative proteomics of tissue that generates maps of protein abundance across tissue slices derived from a human atypical teratoid-rhabdoid tumour at three spatial resolutions, the highest being 40 µm, to reveal distinct abundance patterns of thousands of proteins. We employ spatially-aware algorithms that do not require prior knowledge of the fine tissue structure to detect proteins and pathways with spatial abundance patterns and correlate proteins in the context of tissue heterogeneity and cellular features such as extracellular matrix or proximity to blood vessels. We identify PYGL, ASPH and CD45 as spatial markers for tumour boundary and reveal immune response-driven, spatially-organised protein networks of the extracellular tumour matrix. Overall, we demonstrate spatially-aware deep proteo-phenotyping of tissue heterogeneity, to re-define understanding tissue biology and pathology at the molecular level.


Figure S1 -
Figure S1 -Clinicopathological data of AT/RT case (A) Pre-operative MRI T1 with gadolinium in a female patient.(B) SMARCB1 immunohistochemistry staining on biopsy tissue, showing negative staining in tumour cells and retained nuclear expression in host-derived endothelial cells (brown reaction product); lumen of a blood vessel top right.(C) H&E stain of postmortem tissue block showing relative homogeneity and medium-sized tumour blood vessels.

Figure S2 -
Figure S2 -Tissue area titration across two LC-MS/MS systemsThe number of proteins identified from a titration of AT/RT tissue area on two LC-MS/MS systems.Source data are provided as a Source Data file.

Figure S3 -
Figure S3 -Distribution of identified and quantified proteins.

(
A) 833 µm resolution data.Violin plots showing distributions of the number of identified and quantified proteins per voxel.Solid vertical lines represent the median value.Dashed vertical lines represent upper and lower quartiles.(B) 833 µm resolution data.Histogram showing the number of voxels where each protein was identified/quantified as per panel (A).(C) 350 µm data as in panel (A).(D) 350 µm data as in panel (B).(E) 40 resolution µm data.Violin plot shows distribution of the number of quantified proteins per voxel.Solid vertical line represents the median value.Dashed vertical lines represent upper and lower quartiles.(F) 40 µm resolution data.Histogram showing the number of voxels where each protein was quantified.Source data are provided as a Source Data file.

Figure S4 -
Figure S4 -Aggregate intensity distributions at 833 µm spatial resolution Maps of the log2 transformed (A) summed and (B) mean intensities of each voxel.

Figure S5 -
Figure S5 -Immunohistochemistry validation of AT/RT proteomic maps Proteomic maps of proteins targeted for follow-up IHC staining (n = 1 per protein) and IHC images (A,C,E) Normalised protein intensity maps with their corresponding Moran's Index of spatial autocorrelation (I).Normalised protein intensities are scaled separately for each protein.Grey = not detected.Rectangles depict the approximate location displayed in IHC images.(B,D,F) AT/RT tissue stained and visualised by IHC.All scale bars = 1 mm.Full images of the slides are contained in supplementary data file 5.

Figure S6 -
Figure S6 -Proteomic maps of immune cell markers at 833µm spatial resolution H&E Stained Image of AT/RT Tumour (A).Proteomic maps of immune cell-marker proteins at 833 µm resolution (B).Normalised protein intensities are scaled separately for each protein.Grey = not detected.

Figure S7 -
Figure S7 -Pathway enrichment in immune-cell clusters at 833 µm spatial resolution GSEA of clusters 6 & 13 from Figure 4A.Gene set membership is indicated by colouring the cell with that protein's log2 fold-change.

Figure S8 -
Figure S8 -Dimensionality reduction and gene-set enrichment at 40 µm spatial resolution (A) Measurement map of distance from each cell to the nearest blood vessel, reproduced from Figure 5B for reference.(B) Map of cluster assignment based on hierarchical clustering and the dynamic tree cut algorithm (spatially unaware), reproduced from figure 5D for reference.(C) UMAP embedding of data coloured by cluster assignment in (B).(D) Enriched MSigDB Hallmark gene sets within marker proteins (two-sided Wilcox test, Benjamini-Hochberg multiple testing correction threshold of 1 %) of clusters shown in the cluster map.Significantly enriched hallmarks (one-sided hypergeometric test, Benjamini-Hochberg multiple testing correction threshold of 5 %) for each cluster are indicated by the presence of circles.The size and colour of the circles represent the number of proteins contributing to that term and the adjusted p value of the enrichment, respectively.Cluster 3 did not show significant pathway enrichment.Source data are provided as a Source Data file.

Figure S9 -
Figure S9 -Aggregate intensity distribution of core matrisome proteins at 833 µm spatial resolution Maps of the log2 transformed (A) summed and (B) mean intensities of core matrisome proteins as defined by MatrisomeDB for each voxel.Grey voxels indicate no core matrisome proteins were detected.

Figure S10 -
Figure S10 -Distribution of Moran's I in matrisome and non-matrisome proteins 833 µm spatial resolution Moran's I distribution of proteins annotated in Matrisome DB as core matrisome (blue), matrisome associated (yellow), and all other proteins (grey).Source data are provided as a Source Data file.