Seed becoming soil: A new paradigm of the Ewing sarcoma tumor microenvironment

Summary

Cells in the tumor tumor microenvironment, including cancer-associated fibroblasts (CAFs), contribute to tumor growth and immune evasion. A recent study of Ewing sarcoma identified “CAF-like” tumor cells that mimic the pro-tumorigenic features of CAFs. These findings highlight the role of cell plasticity in tumor growth.

In this issue of Clinical Cancer Research, Wrenn and colleagues describe a state of Ewing sarcoma cells, in which the cancer cells exhibit characteristics normally attributed to mesenchymally derived stromal cells known as cancer-associated fibroblasts (CAFs) (1). Since Paget put forth the “seed and soil” hypothesis in 1889 (2), clinicians and cancer biologists have come to recognize that tumor cells modulate, and are modulated by, the surrounding non-tumor tissue. This “tumor microenvironment” (TME) includes innate and adaptive immune cells, vascular and lymphatic networks, and an extracellular matrix (ECM) of collagen fibrils, proteoglycans and other factors produced by tumor and stromal cells (3,4), and mesenchyme-derived mesenchymal stromal cells (MSCs) and CAFs (5,6).

CAFs are a heterogeneous group of cells that may be derived from resident fibroblasts in the tissue in which tumors arise, from MSCs recruited by the tumor, or, via epithelial-mesenchymal transition (EMT) or other transdifferentiation pathways, from epithelial cells, pericytes, adipocytes and other cell types (6). One study used differentiation of induced pluripotent stem cells to cancer stem cell (CSC)-like cells to suggest that CAF-like cells may arise directly from CSCs (7). However, to date there is little direct evidence that CAFs originate from tumor cells themselves.

From a clinical perspective, understanding of the full complexity of the TME and the role of CAFs is critical, because of the role of the TME in promoting chemoresistance, metastasis and a suppressive immune environment (8,9). Tumor/TME interactions are highly complex and investigating these interactions requires multiple complementary techniques, including genomics, histopathology, single cell and spatial transcriptomics and in vivo models. The contribution by Wrenn et al. is an excellent example of how such a multidisciplinary approach can provide unexpected insights, in this case focusing on Ewing sarcoma.

Ewing sarcoma (EwS) is a malignant bone and soft tissue tumor of children, adolescents and young adults. Although patients with localized disease have a 5-year overall survival rate of >80%, prognosis for patients with metastatic disease remains poor (10), and survivors experience significant late adverse effects of therapy. New therapeutic approaches that are more effective and less toxic are needed. Despite the success of immunotherapy in other cancer types, similar success has not yet translated to EwS (11), in part due to our poor understanding of the EwS TME. Several studies have suggested that M2 macrophages may be important immunomodulators in EwS (12,13). Most recently, Cillo et al. profiled the TME of EwS through multiplex immunofluorescence of EwS tumor samples and single-cell RNAseq (scRNAseq) of sorted CD45+ cells from peripheral blood and tumors (14). This analysis revealed distinct CD14+CD16+ macrophage subsets and CD8+ effector T cells in EwS. A zebrafish EwS model demonstrated a role for ECM proteoglycans in promoting tumor growth in vivo (15).

In contrast to the relative paucity of data on the EwS TME, understanding of EwS tumor cell biology has advanced more rapidly. Molecularly, Ewing sarcoma is characterized by one of several reciprocal chromosomal translocations that fuse a FET family protein with an ETS transcription factor (16). The most common of these, EWSR1::FLI1, has both gene activating and repressing functions. Among the gene sets normally repressed by EWSR1::FLI1 are those associated with mesenchymal development, and this, in combination with EWSR1::FLI1 knockdown studies, has led to a model in which variations in EWSR1::FLI1 function generate phenotypic plasticity of EwS cells: EWSR1::FLI-high cells display cell-cell adhesion properties, whereas EWSR1-FLI1-low cells adopt mesenchymal features and are more migratory (17,18). How might this plasticity contribute to EwS tumor growth in vivo?

Wrenn et al. addressed this question by first identifying transcripts that are normally repressed by EWSR1::FLI1 and expressed by mesenchymal-high state tumor cells. From this analysis, they identified NT5E, which encodes the cell surface protein CD73. CD73+ cells could be detected in EwS cells lines and PDXs. Using CITE-Seq, the authors showed that CD73+ cells are enriched in EMT signatures compared to CD73- cells, with increase in mesenchymal and ECM genes that are normally repressed by EWSR1::FLI1, consistent with a classic “EWSR1::FLI1-low” cell state. Unexpectedly, however, Wrenn et al. identified a “hybrid” state of the cells, in which CD73+ cells maintain expression of EWSR1::FLI1 protein, and expression of EWSR1::FLI1-activated genes. Thus, by an unknown mechanism, genes normally repressed by the fusion are selectively derepressed in CD73+ cells.

To test whether this hybrid transcriptional state can occur in tumors, the authors performed digital spatial profiling of xenograft tumors, integrated with transcriptome analysis of selected regions. The balance of EWSR1::FLI1-mediated activation and repression was heterogeneous: the interior of tumors demonstrated the usual EWSR1::FLI1-high pattern (EWSR1::FLI1 activated genes ON, repressed genes OFF), and an invasive focus showed the opposite, EWSR1::FLI1-low signature. Significantly, a number of other regions exhibited the CD73+ hybrid transcriptional state, indicating that de-coupling of the gene activating and repressive functions of EWSR1::FLI1 occurs in vivo.

The mesenchymal nature of CD73+ cells and their transcriptional profile, which is enriched for ECM genes, was similar to that identified in CAFs across a number of cancer types (6). To investigate the role of these “CAF-like” cells in human tumors, the authors identified genes whose expression was correlated with that of NT5E in several EwS tumor gene expression datasets. Comparison with the in vitro studies identified a core set of genes (including NT5E) that correlated positively with expression of both ECM organization and EWS::FLI1 repressed gene signatures. Staining of tumor resection specimens with antibodies against CD73 and TNC (a protumorigenic ECM protein) confirmed the strong association of CD73+ cells with deposition of ECM, further highlighting the role of tumor derived CD73+ cells in functions traditionally attributed to CAFs.

The findings of Wrenn and co-authors greatly expand our perspective on the complex tumor/TME interaction, showing that tumor cells may directly mimic pro-tumorigenic properties of surrounding stromal cells (Figure 1). The study raises a number of fascinating questions about these “CAF-like” cells. What are the biochemical and/or epigenetic mechanisms that allow the apparently selective de-repression of EWSR1::FLI1 target genes in CD73+ cells? How plastic are the cell states in vivo? Referring back to Paget’s initial observations, which were based on analysis of patterns of metastasis, what is the role of CD73+ cells in metastatic spread of EwS? If CD73+ cells are more invasive and perhaps have increased metastatic propensity, is their CAF-like state maintained in established metastatic tumor—is their presence needed to maintain the “soil” for metastasizing tumor cells? Or do the cells revert to a more mixed phenotype? In this regard, future orthotopic xenograft models may allow for a more detailed comparison of primary and metastatic tumors. Given the significant overlapping gene expression between CAF and CAF-like CD73+ cells, it will be important to investigate the interactions of CAF-like EwS cells with other immune and non-immune cells. Better understanding of these interactions will require high-resolution spatial studies of tumors, as well as immunocompetent animal models such as the fish model (15) and humanized mouse models (19). The insights of Wrenn and colleagues now pave the way to a new and better understanding of complex tumor/TME interactions, and ultimately to more effective treatments for EwS and other cancers.

Figure 1: Contribution of CAF-like tumor cells to the tumor microenvironment.

Ewing sarcoma (EwS) tumors recruit immune cells and stromal cells including cancer-associated fibroblasts (CAFs). EwS cells with low activity of the EWSR1::FLI1 fusion oncogene enter a migratory, mesenchymal state. Wrenn et al. (1) identify tumor-derived “CAF-like” cells that maintain expression of EWSR1::FLI1-activated genes while mimicking pro-tumorigenic functions of CAFs. (Adapted from an image created with Biorender.com)