We devoted a lot of effort to develop adequate models and imaging technologies to reliably capture the invasive behaviour of DIPG observed in patients. This resulted in an IUE in vivo model that was specifically designed to mimic both the genetic background, as well as early disease onset in the developing brain. We performed in depth characterisation of the tumour and its immediate environment, using a technology that allows for extensive molecular and spatial profiling. This confirmed a highly diffuse pattern of tumour growth, as seen in patients. Moreover, we found a clear resemblance of the tumour and its direct surroundings compared to tumours found in patients. This demonstrates the suitability of our model for understanding human disease, and particularly the role of the tissue environment in dictating tumour cell behaviour. We exploited this model and our spatially resolved transcriptomic dataset to study interactions between different tumour cells subsets and immune cell types in their immediate environment. This revealed spatially restricted interrelations between immune and tumour cell subsets, showing that a specific tumour cell population preferentially located with infiltrating immune cells. Receptor-ligand inference revealed specific communication pathways involved in these interactions, identifying two therapeutic targets that when inhibited could reduce tumour progression.
To characterise the tumour and its cellular environment in great detail, we advanced our imaging technology to include 8 fluorescent labels. These 8 markers can be used to stain for various molecules expressed on cells and, thereby, identify different cell types present. To make sense of the resulting high-dimensional datasets, we furthermore developed a novel AI-based analysis method. This tool automatically finds each cell present in the dataset and subsequently extracts all the acquired information from this cell. This includes molecule expression, identified by our 8-colour imaging, but also spatial information, such as the location of the cell in the tissue or its particular size or shape. We demonstrated the discovery power of this analysis by identifying new tumor cell sub-populations (van Ineveld, Kleinnijenhuis et al. Nature Biotechnology, 2021; van Ineveld, Collot et al. Nature Protocols, 2022)
We also established a live cell imaging and computational framework to classify T cell behaviour and identify their underlying molecular signatures (Dekkers, Alieva et al. Nature Biotechnology, 2023; Alieva et al. Nature Protocols, in press). This has led to the identification of highly potent engineered T cells and further combinatorial therapy and a selection strategy to enhance anti-tumour activity of engineered T cells for DMG and other cancer indications. By following single cells over time, this method can now also be used to record active tumour cell characteristics, such as their speed and direction of movement. Together with our 8-colour imaging method, this allows us to identify highly invasive tumour cells and associate this behaviour to specific cell types present in their direct environment. In doing so, we aim to identify environmental factors contributing to this invasive behaviour that could be targeted to counteract the tumour’s invasive spread and improve disease outcome.
