Using cell-cell interactions to unlock new cancer treatments: Forcing neural crest tumors back onto the developmental path

Project Information

KILL-OR-DIFFERENTIAT

Grant agreement ID: 856529

DOI

10.3030/856529

EC signature date 27 May 2020

Start date 1 October 2020

End date 31 January 2027

Funded under

• EXCELLENT SCIENCE - European Research Council (ERC)

Total cost

€ 9 377 151,00

EU contribution

€ 9 377 151,00

Coordinated by

MEDIZINISCHE UNIVERSITAET WIEN
Austria

Periodic Reporting for period 3 - KILL-OR-DIFFERENTIAT (Using cell-cell interactions to unlock new cancer treatments: Forcing neural crest tumors back onto the developmental path)

Reporting period: 2023-10-01 to 2025-03-31

Summary of the context and overall objectives of the project

The interactions between tumor and its microenvironment are often critical to uncovering the mechanisms of tumor survival. The tumor can also repress immune response by inducing complex interactions among dozens of immune and stromal cell types that typically make up tumor microenvironment, however those remain largely uncharacterized as we currently lack systematic approaches to uncover relevant cell-cell interactions. The alternative to killing tumor cells, either directly or through immune system, is to force them to differentiate. Such strategy is particularly promising for tumors arising due to failure of progenitor populations to follow proper differentiation cascade. Here as well, the progress has been limited by lack of understanding of the specific intercellular signals that that are disrupted in tumorigenesis. We propose a systematic approach for characterizing cell-cell interactions in complex microenvironments through joint analysis of spatially-resolved and disassociated single-cell transcriptomics. We will apply it to identify inter-cellular signals and pathways that can push tumors of neural crest origin, including as pheochromocytoma (PCC), paraganglioma (PGL) and neuroblastoma (NB), towards terminal differentiation. Building on our expertise with neural crest development, we will use single-cell profiling to map individual tumor cells onto developmental trajectory of neural crest differentiation. We propose to achieve the following objectives:

AIM 1. To profile homeostatic trajectory of neural crest differentiation and identify inter-cellular interactions guiding normal neural crest development using single-cell and spatially-resolved transcriptomics analysis.
AIM 2. Characterize tumor microenvironment composition, cell-cell interactions and intratumoral transcriptional heterogeneity in different pediatric and adult subtypes of NB, PCC and PGL tumors. Map tumor cells onto the homeostatic trajectory to identify corresponding differentiation stage and signaling state of the individual tumor cells. Identify signaling and downstream transcriptional pathways disrupted in tumors.
AIM 3. Predict signalling or transcriptional perturbations that would push tumor cells down homeostatic neural crest differentiation path. Perform pre-clinical validation of such differentiation therapy perturbations in culture and in genetically engineered mouse models and mouse xenografts.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

From the beginning of the project, we addressed Aim 1 and Aim2, and to some extent Aim3. Firstly, finalized and deeply analyzed the single cell map of the entire neural crest lineage, which became a basis for anew hi fi atlas of cell types relevant to the development and origin of a plethora of diseases, such as neuroblastoma (in the first place), but also melanoma, neurofibromatosis and many more. This atlas was already put to work for the analysis of neuroblastoma cell type heterogeneity during the first reporting period. In the parallel, we generated spatial maps of the neural crest on the tissue slices. Next, we created a principally new method to analyse clonality and cell type heterogeneity in tumors, called NUMBAT. We also created the method for analyzing transitions in cell states called scFates. This is a big achievement, as it helps to build the map of transitions between malignant plastic cell populations with different properties. Furthermore, we generated novel transgenic animal models of neuroblastoma as outlined in aim 2 and observed tumors with 100 % penetrance in these mice. In the models, we deleted KIF1Bβ and NF1 in the embryonic mouse sympatho-adrenal lineage and observed pheochromocytoma, neuroblastoma, and composite tumors arising in aged mice. Deep single-cell RNA sequencing combined with immunohistochemistry and RNA scope revealed neuroblast-chromaffin cell state transitions at embryonic and postnatal stages driving tumor plasticity. We also identified a neuroblastoma-specific transcriptional network involving YAP/TAZ, TEAD, FOSL, and RUNX family proteins that regulates mesenchymal identity and plasticity. Inhibition of YAP/TAZ reduced MES cell proliferation, while their overexpression induced mesenchymal reprogramming. Immune profiling revealed an immunosuppressive tumor microenvironment in neuroblastoma, characterized by TIGIT+ T cells interacting with NECTIN2/3-expressing stromal cells. To understand the clonal evolution in tumors and to compare it to clonality in the neural crest lineage, we adopted the lentiviral barcoding and single-cell transcriptomics to create a clonal atlas of neural crest and neuroblastoma using animal models. This enables revealing spatial and temporal dynamics in cell fate decisions. For this we further developed a new machine learning tool `Clone2vec` to capture clonal fate diversity, linking lineage identity to transcriptional states and uncovering fate biases in neural crest and tumor populations.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

The major progress beyond the state of the art includes the two main points:
1. We developed a principally new method for calling the tumor-related mutations in single cell data for building clonal trees of clonal development within tumors (this helps to understand the repertoire of malignant cell types in the tumor and their relations with each other including unwanted plasticity). This method is just published in highly prestigious Nature Biotechnology journal and is called NUMBAT:
https://github.com/kharchenkolab/numbat
It takes advantage of the recent progress in population genetics, which has enabled us to predict maternal and paternal alleles of individuals from haplotype data with high accuracy. Numbat uses this population-based phasing to significantly increase the sensitivity and specificity in detecting copy-number alternations and loss-of-heterozygosity events.
2. We developed a new computational tool for delineating fate decisions and genes mediating transitions between different cell states or fates called scFates.
https://github.com/LouisFaure/scFates
scFates goes beyond static snapshots of cellular states, enabling dynamic modeling of transitions and trajectories. This provides a richer understanding of how cells change over time or respond to external factors. By connecting trajectory data with gene expression changes, scFates helps link cellular transitions to underlying molecular mechanisms, making it easier to interpret biological processes.
3. Another major breakthrough is the discovery of human-specific aspects of sympatho-adrenal and neural crest development published in Nature Genetics paper (Kameneva et al.), which is focused on characterizing human embryological aspects of chromaffin cell formation within healthy adrenal gland. This knowledge if of high relevance to neuroblastoma research, as it helps to understand why mouse models of neuroblastoma poorly recapitulate the original human disease.
4. We developed the new machine learning tool for clonal analysis in healthy tissues and tumors called word2vec.
https://github.com/dav/word2vec
Using Word2Vec, a natural language processing tool, for clonal analysis offers innovative advantages by leveraging its ability to embed high-dimensional data into meaningful, lower-dimensional vector representations. This tool is at the core of the future expectations for this project as we are planning to apply it to understand clonality and possible transitions in modelled neuroblastomas.

key publication on the project