08/10 – SEMINÁRIO DE DADOS (Mestrado) – THIAGO STEINDORFF KRUGER

SEMINÁRIO DE DADOS (Mestrado)

THIAGO STEINDORFF KRUGER  

Lysosomal Gene Expression in Glioblastoma: A Marker of Macrophage Infiltration and Prognosis

Orientadora: Dra. Ursula da Silveira Matte

Avaliador: Dr. Fabrício Figueiró (PPG Bioquímica – UFRGS)

08 de outubro de 2025 (quarta-feira)

12h

Presencial: Anfiteatro do Departamento de Genética da UFRGS (Campus do Vale, prédio 43312)

Abstract

As a first step, we conducted a systematic review of 496 studies published in 2023 and 2024 that used TCGA data in pan-cancer transcriptomic analyses, identifying major methodological reporting flaws, such as the absence of description of the differential expression algorithm in 16.26% of the papers and the lack of definition of a cut-off for survival analysis in 40.71%. We also observed a firm heterogeneity across tumors: while in cholangiocarcinoma (CHOL), more than 80% of genes appeared differentially expressed, in pancreatic adenocarcinoma (PAAD) and thymoma (THYM), it was less than 10%; in survival analysis, clear cell renal carcinoma (KIRC) and lower grade glioma (LGG) had about 60% of genes associated with prognosis, versus only 1.35% in testicular germ cell tumor (TGCT). We replicated the analyses using edgeR (for DEGs, with p < 0.05 and |logFC| > 1) and GEPIA2 (for SRGs, with p < 0.05), confirming extreme discrepancies among tumors, and demonstrated, through evaluation of independent datasets, that DEG proportions vary more by dataset than by tumor biology. Therefore, we propose that pan-cancer comparisons be made using proportional methods, such as Fisher’s exact test, rather than raw counts of DEGs or SRGs—an approach we validated with housekeeping genes (fewer DEGs than expected) and cell cycle genes (more DEGs than expected). Applying this method to lysosomal genes (KEGG hsa:04142), we found that, although generally less differentially expressed, these genes were strongly associated with survival in several tumors, with GBM standing out. In GBM, 22 lysosomal genes were identified as SRGs, all of which correlated with a worse prognosis. GSEA confirmed that this set was the most enriched KEGG pathway for prognosis in GBM. To further investigate, we constructed an integrated scRNA-seq atlas of GBM, comprising 307,138 cells from 101 samples, which revealed increased expression of 14 out of 22 survival-related lysosomal genes (SRLGs) in myeloid cells, particularly macrophages. This finding was validated by deconvolution of TCGA bulk samples and the scRNA-seq datasets in the Human Protein Atlas. Using WGCNA, we identified a co-expression module (yellow) comprising 9 SRLGs, which is strongly enriched in myeloid-related functions. Transcription factor analysis highlighted PU.1 (SPI1) as a central regulator, with ChIP-seq validation showing PU.1 binding at the promoter regions of most SRLGs in the yellow module. Finally, cell-cell communication analysis with cellchat indicated that lymphoid cells may recruit macrophages into the GBM microenvironment via CCL5, which was correlated with myeloid infiltration and worse prognosis, suggesting that inhibiting this pathway represents a more promising strategy for macrophage depletion in GBM than classical targets such as MIF, CSF1, or CX3CL1.