Single-cell lineage and transcriptome reconstruction of metastatic cancer reveals selection of aggressive hybrid EMT states

Abstract

Metastatic cancer remains largely incurable due to an incomplete understanding of how cancer cells disseminate throughout the body. However, tools for probing metastatic dissemination and associated molecular changes at high resolution are lacking. Here we present multiplexed, activatable, clonal, and subclonal GESTALT (macsGESTALT), an inducible lineage recorder with concurrent single cell readout of transcriptional and phylogenetic information. By integrating multiple copies of combined static barcodes and evolving CRISPR/Cas9 barcodes, macsGESTALT enables clonal tracing and subclonal phylogenetic reconstruction, respectively. High barcode editing and recovery rates produce deep lineage reconstructions, densely annotated with transcriptomic information. Applying macsGESTALT to a mouse model of metastatic pancreatic cancer, we reconstruct dissemination of tens-of-thousands of single cancer cells representing 95 clones and over 6,000 unique subclones across multiple distant sites, e.g. liver and lung metastases. Transcriptionally, cells exist along a continuum of epithelial-to-mesenchymal transition (EMT) in vivo with graded changes in associated signaling, metabolic, and regulatory processes. Lineage analysis reveals that from a majority of non-metastatic, highly epithelial clones, a single dominant clone that has progressed along EMT drives the majority of metastasis. Within this dominant clone a parallel process occurs, where a small number of aggressive subclones drive clonal outgrowth. By precisely mapping subclones along the EMT continuum, we find that size and dissemination gradually increase, peaking at late-hybrid EMT states but precipitously falling once subclones are highly mesenchymal. Late-hybrid EMT states are selected from a predominately epithelial ancestral pool, enabling rapid metastasis but also forcing extensive and continuous population bottlenecking. Notably, late-hybrid gene signatures are associated with decreased survival in human pancreatic cancer, while epithelial, early-hybrid, and highly mesenchymal states are not. Our findings illuminate features of metastasis and EMT with the potential for therapeutic exploitation. Ultimately, macsGESTALT provides a powerful, accessible tool for probing cancer and stem cell biology in vivo.

Interconnected High-Dimensional Landscapes of Epithelial-Mesenchymal Plasticity and Stemness

“…These results are reminiscent of observations in breast cancer that knockdown of E-cadherin, which can induce a full-blown EMT, can restrict the metastatic potential of cells [82]. Reinforcing observations were recently reported in a mouse model of pancreatic cancer, where the "late hybrid" cells, but not the "most mesenchymal" cells were found to be metastatically advantageous with specific proliferative, metabolic, and signalling processes associated with them [9]. Therefore, in prostate and pancreatic cancer, "the more the EMT, the more the stemness" dogma does not hold true; instead, the 'stemness window' seems to lie on an epithelial or hybrid E/M range of values on the 'EMP axis'.…”

“…The number of hybrid states can also vary depending on the cancer subtype. Of course, a large enough number of discrete states may be better approximated as a continuum [9]. Notwithstanding the difference in number of such states reported, a more important question is the functional relevance of these hybrid E/M states.…”

Single-Cell Approaches to Deconvolute the Development of HSCs

“…Specifically, the barcode vector contains evolving barcode regions and static barcode parts. The researchers confirmed the practicality of macsGESTALT in vitro with limited dilution and expansion and then, proved its ability to reconstruct colonies and identify the EMT continuum in PDAC-bearing mice [ 45 ]. However, this technique has not been applied to understanding the development of HSCs.…”

Cancer: More Than a Geneticist’s Pandora’s Box

“…Put together, fast and reversible adaptations at a phenotypic (i.e., (Shachaf and Felsher, 2005) non-genetic) level seem to be playing an instrumental role in enabling cancer metastasis as compared to slow and irreversible adaptations available at a genetic level. Hence, it is not surprising that while cells from various sub-clones of the primary tumour have been seen in circulating tumour cells and capable of metastasizing (Lyberopoulou et al, 2015;Simeonov et al, 2020), no unique mutational foot-prints have yet been deciphered, unlike other hallmarks of cancer, for which mutations in various oncogenes and/or tumour suppressor genes have been pinpointed (Hanahan and Weinberg, 2011;Mantovani et al, 2019). Such plasticity during metastasis can lead to phenotypic (non-genetic) heterogeneity, as witnessed in circulating tumour cells (CTCs) across cancer types (Bocci et al, 2021;Yu et al, 2013).…”