Regulation of stem cell self-renewal and differentiation by Wnt and Notch are conserved throughout the adenoma-carcinoma sequence in the colon

Normal gut homeostasis is ensured by intestinal stem cells (ISCs), which rely on integration of various morphogenic pathways including bone morphogenetic protein, Hedgehog, Notch and Wnt signaling cascades [1]. Colorectal cancer (CRC) is a multistep process in which genetic hits induce transition from normal mucosa via adenomatous lesions to an invasive carcinoma [2]. Adenoma initiation is suggested to occur via deregulation of these morphogenic pathways. First and foremost, mutations leading to activation of the Wnt cascade are seen in all CRC patients [1, 3]. However, these activating mutations never result in a complete activation, but rather in a finely-tuned Wnt activity level [4]. Furthermore, cell-to-cell variation in pathways activity is frequently observed. In agreement with these concepts are our recent observations that have pinpointed Wnt pathway regulation as the basis of cancer stemness in CRC [5]. Functionally marked by high Wnt pathway activity, colon cancer stem cells (CSCs) have self-renewal capacity and the potential to differentiate into all cell lineages present in cancerous tissue [6]. These similarities between ISCs and CSCs may therefore point to the intriguing possibility that the response to morphogenic signals is not lost during tumorigenesis. Despite the wealth of data regarding the role of morphogenic pathways in controlling cell fate and proliferation in intestinal tissue, to our surprise few studies have attempted to comprehensively assess their role throughout the stages of neoplastic progression. For example, whereas Notch inhibition leads to goblet cells accumulation in adenomas [7], little is known on the role of Wnt signaling in this stage.

Using a number of human and mouse intestinal-derived organoid cultures from normal, adenoma and cancerous tissues [8, 9], we provide a comprehensive overview on the role of both Wnt and Notch morphogenic pathways in the different stages of CRC development. We find that Notch inhibition decreases stemness and enhances goblet-like differentiation in all stages of disease. Conversely, Wnt activity is associated with stemness in normal, adenoma and carcinoma tissue. Our results point to the conclusion that colon carcinogenesis, at early and late stages consistently retain numerous characteristics of the normal intestine.

Normal intestinal epithelial cultures rapidly self-organized into typical single layered organoid structures composed of epithelial cells (EpCAM⁺) surrounding a central lumen (Figure 1). As previously shown, organoids contained the various differentiated gut cell types, including neuroendocrine cells (Chromogranin A, (CHGA)) (Additional file 1: Figure S1A), goblet cells (MUC2) and enterocytes (Villin) (Figure 1A) [8, 9]. Adenoma cultures contained scattered goblet-like cells judged by MUC2 staining (Figure 1B), but appeared to lack CHGA⁺ cells (Additional file 1: Figure S1A) consistent with previous observations [10] and stainings on the corresponding primary tumor material (Additional file 1: Figure S1B). Intriguingly, when analyzing CRC spheroid cultures, derived from a single cell clone and reportedly enriched in CSCs [11], we detected the presence of various lineages (Figure 1C) including the CHGA⁺ neuroendocrine cell lineage. This demonstrates that differentiation towards the distinct epithelial lineages occurs even in late stage cancers and, whenever present, this process might be similarly orchestrated at all stages of the disease.

To address this, we first assessed the effect of pharmacological inhibition of the Notch pathway in the distinct cultures. As previously reported, dibenzazepine (DBZ) mediated Notch inhibition resulted in goblet cell differentiation in cultured mouse intestinal epithelium and adenoma (Additional file 2: Figure S2). Importantly, also treatment of human epithelial organoid cultures derived either from normal, adenoma or adenocarcinoma tissues with DBZ, caused a clear increase in goblet cells (Figure 2A-B) and a significant induction of MUC2 mRNA level (Figure 2C). This goblet cell-like differentiation appeared to be observed only in cultures already expressing low amounts of MUC2 (Additional file 3: Figure S3). More importantly, the clonogenic potential of all cultures was clearly decreased upon DBZ treatment, which is consistent with the loss of (cancer) stem cells (Figure 2D and Additional file 2: Figure S2), implying that Notch pathway-dependent self-renewal and cell fate decision are maintained at all stages of tumor development.

Secondly, we carefully assessed the Wnt pathway activity in all derived cultures using the TCF/LEF reporter construct (TOP-GFP) [5]. As mentioned above, deregulation of the Wnt pathway is frequently observed in CRC development and we have shown that high Wnt activity can be used as a functional CSCs marker in adenocarcinomas [5]. In addition, recent reports have highlighted the heterogeneous expression of the Wnt target and stem cell marker LGR5 in both adenomas and carcinomas [12, 13]. As expected, a hierarchy in Wnt pathway activity in healthy epithelial and adenocarcinoma organoid cultures could be detected by distinct GFP patterning (Figure 3A). Intriguingly, adenoma cultures also display heterogeneous Wnt activity, with TOP-GFP^high adenoma cells displaying high levels of stem cell marker genes. More importantly, TOP-GFP^high adenoma cells represented the clonogenic fraction. This indicates that the functional hierarchy of the Wnt pathway, which is known for normal epithelium and adenocarcinoma CSCs, also identifies stemness in adenomas (Figure 3C-D).

Previous studies have highlighted the role of Wnt and/or Notch in distinct stages of colon cancer. For instance, Notch inhibition has been shown to induce goblet differentiation in mouse adenomas and block stemness in colon CSCs [7, 14]. Similarly, we have previously shown that Wnt activity determines stemness in adenocarcinoma-derived colon CSCs. This present study, however, for the first time provides a comprehensive overview of these pathways at all stages and extends these surprising findings by showing that Wnt pathway heterogeneity persists at all stages despite evident mutations in the Wnt pathway and that Notch signals maintain a balance between self-renewal and cell lineage decisions. Currently, the molecular mechanisms underlying such heterogeneity remain unclear. Nevertheless, the organization appears to be purely intrinsic to epithelial cells as the mesenchymal niche is not present in vitro. Differential Wnt activity in the normal epithelial organoid cultures was suggested to exist due to the presence of Wnt/Notch ligand producing paneth cells [15]. Likewise, tuning of Wnt signaling in adenoma or adenocarcinoma might be achieved by a similar mechanism. In this respect, it is interesting to note that paneth-like cells have been reported in mouse adenomas and reside in close proximity to the LGR5⁺ adenoma stem cells [13]. Furthermore, microenvironment derived signals such as HGF and Jagged-1 [5, 16] could also be responsible for this Wnt hierarchy.

Our data on a unique panel of cultures, representing normal, pre-malignant neoplastic lesions as well as adenocarcinoma tissue from both mouse and human origin, shows that at all stages of disease, the tissue is organized in a highly similar hierarchical fashion and present with cell types encompassing multiple differentiation lineages. Here we find that the Wnt and Notch pathways are not only crucial for the orchestration of this hierarchy but their function in regulating self-renewal and cell fate is also conserved during cancer development. Furthermore, our data provide evidence that the robust mechanism orchestrating hierarchy in normal intestinal epithelium is strongly wired in the epithelial cells and its regulation is sustained throughout carcinogenesis (Figure 4). The Wnt and Notch pathways, even though subjected to mutation or aberrant regulation, maintain their central position in the regulation of this hierarchy. These parallels between normal and neoplastic tissue notably makes targeted therapies towards these pathways challenging, as it will predictably impact normal hierarchy. Understanding similarities, but above all the differences in molecular mechanisms will help in highlighting crucial differences that can be exploited for therapeutic design.