Characterizing and harnessing tumor-reactive T cells in the brain

Projektbeschreibung

Personalisierte T-Zell-Therapie bei Hirntumoren

Krebszellen exprimieren auf ihrer Oberfläche Moleküle, die möglicherweise vom Immunsystem erkannt werden können. Die Auswahl der immunogensten Epitope dieser Antigene aus den Tausenden, in jedem Tumor vorhandenen potenziellen Kandidaten bleibt eine Herausforderung, was aufgrund des begrenzten Gewebezugangs und der geringen Anzahl tumorinfiltrierender T-Zellen insbesondere bei Hirntumoren gilt. Das Team des ERC-finanzierten Projekts CENTRIC-BRAIN verlagert den Schwerpunkt von der Ermittlung von Tumorantigenen auf die Nutzung vorhandener, potenziell wirksamer T-Zell-Rezeptoren von tumorinfiltrierenden T-Zellen. Das Projekt baut auf predicTCR auf, einem auf maschinellem Lernen und künstlicher Intelligenz basierenden System, das die Tumorreaktivität von T-Zell-Rezeptoren vorhersagen kann, und es ist geplant, es zur Entwicklung personalisierter adoptiver Zelltherapien mit technisch manipulierten T-Zellen einzusetzen.

Ziel

Personalized tumor immunotherapy - vaccines and T cell therapies - requires the selection of few immunogenic epitopes recognized by tumor-reactive T cell receptors (TCRs) from thousands of potential tumor antigens in each tumor. This selection is a huge challenge, particularly in brain tumors, where access to tumor tissue is limited and tumor-infiltrating T cells (TILs) are sparse. CENTRIC-BRAIN will reverse this epitope-centric approach by directly harnessing tumor-reactive orphan TCRs from TILs for personalized brain tumor immunotherapy. Using human brain tumor tissue we have developed and experimentally validated predicTCR, a novel approach using machine learning and explainable AI to derive a classifier that predicts tumor-reactivity of such orphan TCRs based on the expression of signature genes with > 90% accuracy across multiple tumour entities. CENTRIC-BRAIN will employ predicTCR to develop a personalized adoptive cell therapy using transgenic T cells with tumor-specific TCRs and improved function.
CENTRIC-BRAIN hypothesizes that the signature genes underlying predicTCR determine the phenotypic and functional properties of tumor-reactive T cells and that predicted tumor-reactive orphan TCRs can be employed for adoptive therapy with personalized TCR-transgenic T cells.
Aim 1 will refine predicTCR by characterizing the transcriptional programs and spatial distribution of tumor-reactive T cells in human brain tumor samples. Aim 2 will define the functional relevance of transcriptional programs for tumor-reactive T cells in syngeneic humanized and patient-derived xenograft brain tumor models. Aim 3 will validate the predicted tumor-reactive orphan TCRs by adoptive transfer of T cells genetically engineered to express predicted tumor-reactive TCRs.
CENTRIC-BRAIN will thus revolutionize precise and efficient implementation of personalized adoptive cell therapy for brain tumors using T cells expressing tumor-reactive patient-derived orphan TCRs.