A short video about our idea -- https://youtu.be/Fq87pAqMpus
Schwann cells are the cells-of-origin of neurofibroma. Neoplastic Schwann cells in benign plexiform neurofibromas can transform into a malignant state into a malignant peripheral nerve sheath tumor (MPNST). This transformation involves the step-wise accumulation of oncogenic mutations. It has been hypothesized that malignantly transformed Schwann cells assume a stem-cell-like or a de-differentiated state but the molecular characteristics of this stem-like state is not known.
For many tumors, the malignantly transformed version of a normal cell is often a reversion to an embryonic state. This reversion involves signalling pathways that are part of the normal development of the cell. This suggests that characterizing the transcriptomes of intermediate stages of Schwann cell development provides us with the tools to understand the stem-like state of MPNST tumors.
Our approach is to integrate single-cell transcriptomic datasets across different stages of neural crest development in mice [1-3], and create a trajectory of Schwann cell development from the neural tube all the way to mature myelinating and non-myelinating Schwann cells. We derive gene signatures from each of the cell types in Schwann cell development and characterize the state of de-differentiation of Schwann cells in plexiform neurofibromas [4], as well as evaluate the transcriptomes of bulk human MPNST tumors [5].
- We successfully construct a trajectory of Schwann cell development. The validity of our trajectory can be assessed from the fact that Schwann cell precursors arise from the autonomic lineage of the mouse neural crest, while myelinating and non-myelinating Schwann cells descend from Schwann cell precursors. We derive signature genes from each stage of neural crest development.
- Schwann cells in plexiform neurofibroma resemble myelinating Schwann cells. We score single-cell transcriptomes of Schwann cells from plexiform neurofibromas [4] using the above neural crest signatures. We find that these Schwann cells display high transcriptomic activity of genes characteristic of mature myelinating Schwann cells, suggesting that they are not yet at a de-differentiated stage.
- MPNST tumors display activation of migratory neural crest and neutral tube cells. We scored multiple human MPNST bulk RNA-seq samples [5] using our neural crest signatures. All tumor samples showed activiation of non-myelinating Schwann cells. A majority, however, also displayed characteristics of neural tube and migratory neural crest cells. A subset of these tumors also recapitulated a Schwann cell progenitor state (present in mouse embryos at the E13.5 stage).
The Jupyter notebook and accompanying code for our submission is available here.
The processing_functions.R library contains the functions needed for loading all requisite data into the notebook for analysis, as well as the helper functions used for computing trajectories and deriving marker genes for evaluating tumor samples. Some of the datasets needed to run this notebook are quite large and do not fit into the data/ folder available here. We will make this data available in a shareable form as soon as we can.
The notebook contains all the commands and calls needed to compute the trajectory tree computing for all developing cells as well as the signature heatmap of all MPNST tumor samples that we processed.
Data Sources :
[1] Soldatov, Ruslan, et al. "Spatiotemporal structure of cell fate decisions in murine neural crest." Science 364.6444 (2019). https://science.sciencemag.org/content/364/6444/eaas9536.abstract . This was the source of the E9.5 trunk neural crest data.
[2] Furlan, Alessandro, et al. "Multipotent peripheral glial cells generate neuroendocrine cells of the adrenal medulla." Science 357.6346 (2017).https://science.sciencemag.org/content/357/6346/eaal3753.abstract . This was the source of the E12.5 and E13.5 Schwann cell precursors.
[3] Wolbert, Jolien, et al. "Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity." Proceedings of the National Academy of Sciences 117.17 (2020): 9466-9476. https://www.pnas.org/content/117/17/9466.short . This was the source of the nmSC and mySC (non-myelinating and myelinating Schwann cells).
[4] Fletcher, Jonathan S., et al. "Cxcr3-expressing leukocytes are necessary for neurofibroma formation in mice." JCI insight 4.3 (2019). https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6413799/ . This was the source of the Scwhann cell single-cell RNA-seq profiles from a mouse plexiform neurofibroma tumor.
[5] https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000792.v1.p1 . Source of the bulk human MPNST tumor samples.
Methods References :
[6] Aibar, Sara, et al. "SCENIC: single-cell regulatory network inference and clustering." Nature methods 14.11 (2017): 1083-1086. The AUCell package was used for scoring neural crest and Schwann cell signatures in Schwann cells from [4].
[7] Stuart, Tim, et al. "Comprehensive integration of single-cell data." Cell 177.7 (2019): 1888-1902. The Seurat package.
[8] Qiu, Xiaojie, et al. "Reversed graph embedding resolves complex single-cell trajectories." Nature methods 14.10 (2017): 979. Monocle3 package.