High throughput, spatially-resolved proteomic analysis of a human brain tumour

Simon Davis - Target Discovery Institute


Tissue-based proteomics approaches often lack spatial resolution and the data generated gives little to no knowledge about the heterogeneity of the tissue structure. The isolation of cells of a single phenotype or small regions of tissue can ease the signal averaging issues seen when taking ‘bulk’ measurements of diverse cell populations. Here we apply a sensitive, spatially resolved workflow for the proteomic analysis of a tumour to identify proteins that display spatial expression patterns within the tissue in an unsupervised manner.


We used laser capture microdissection (LCM) to isolate tissue voxels from a human brain tumour obtained post-mortem in a raster pattern. Captured voxels were proteolytically digested with trypsin and peptides analysed by LC-MS/MS on an Orbitrap Fusion Lumos using 60-minute gradients (‘low throughput’) or on a TimsTOF Pro coupled to an Evosep One using 11.5-minute gradients (‘high throughput’). Data were searched in MaxQuant using match between runs. The generated spatial proteomic data were tested for spatial expression trends across the sampled area using spatially aware statistical methods.

Preliminary Data

We identified approximately 4000 and 2000 proteins in total from the low throughput (60-minute gradient) and high throughput (11.5-minute gradient) approaches respectively, with batch run times of 8 days and 1 day. The regions sampled included the tumour solid core, brain/tumour interface and blood vessels. Using a permutation test for Moran’s I, a measure of spatial autocorrelation, 526 and 273 proteins were identified as displaying significant evidence of spatial autocorrelation in their spatial expression profiles. These proteins include examples of blood proteins, tumour and neuronal markers and these proteins show an enrichment of Gene Ontology terms related to tumour and vascular biology.

Novel Aspect

Identification of spatially expressed proteins within heterogeneous tissue using an unsupervised spatial analysis approach.