Cell fate in colon cancer stem cells: To GLI or not to GLI?

ABSTRACT

Colon cancer is a heterogeneous tumor driven by subpopulations of cancer stem cells (CSCs). We recently used patient-derived organoids (PDOs) to demonstrate that CSC survival is regulated by autocrine non-canonical Hedgehog signaling, which acts as a positive regulator of Wnt signaling to block CSC differentiation.

Recent data supports a hierarchical model of colon cancer in which subpopulations of cancer stem cells (CSCs) drive tumor growth. In addition, therapy resistant CSCs may be the source of relapse following treatment.¹ Understanding the signaling pathways that regulate the survival of CSCs may therefore lead to better treatment outcomes.

The small number of CSCs present in primary and xenograft tumor tissue has been a limiting factor in the elucidation of molecular mechanisms regulating CSCs in solid tumors. In our recent study,² we used CSC-enriched colon cancer patient-derived organoids (PDOs) to extract CSC subpopulations for molecular and functional characterization. These PDOs demonstrated both functional and molecular inter- and intratumor heterogeneity, with only a fraction of cells within each tumor being tumorigenic. We also observed a preserved functional heterogeneity between the PDOs and their tumor of origin, with the most high-grade tumors being most enriched for CSCs. These data support the validity of PDOs as models for the delineation of signaling pathways important in the regulation of CSCs and tumor biology in general.

CSCs were isolated from PDOs using the stem cell marker aldehyde dehydrogenases (ALDH) and functionally validated using in vitro spheroid and serial xenotransplantation assays before being subjected to RNA-sequencing. These studies demonstrated ALDH^Positive cells to be enriched for CSCs and developmental gene sets, including Wnt and Hedgehog signaling. Many of the signaling pathways that regulate normal tissue development are also employed by the tumors that affect them. In the intestine, Wnt signaling is highest in the stem cell compartment at the crypt base, where it drives stem cell self-renewal, and decreases as cells move up through the intestinal crypt and into the differentiated area at the top of the crypt.³ Hedgehog signaling, on the other hand, is mainly confined to the differentiated cells at the top of the crypt where it inhibits Wnt signaling and thus restricts its expression to the base of the crypt.⁴ Crosstalk between Wnt and Hedgehog signaling has also been shown to be important in the development and progression of colorectal cancer where, similar to the normal intestine, Hedgehog signaling has been shown to be a negative regulator of Wnt.⁵ It was therefore interesting to see both Wnt and Hedgehog signaling gene sets enriched in ALDH^Positive CSCs and in T-cell factor/lymphoid enhancer factor-enhanced green fluorescent protein positive (TCF/LEF-eGFP^Positive) Wnt active reporter cells. However, differential gene expression analysis demonstrated that while the Hedgehog signaling genes Sonic Hedgehog (SHH), Smoothened (SMO) and Patched Homolog 1 (PTCH1) and downstream targets such as Polycomb Group RING Finger Protein 4 (BMI1) and Bone Morphogenetic Protein 4 (BMP4) were expressed in CSCs and TCF/LEF-eGFP^Positive Wnt reporter cells, the canonical Hedgehog targets, Glioma-Associated Oncogene Family Zinc Fingers 1 (GLI1), 2 (GLI2) and 3 (GLI3) were not. In addition, immunostaining of CSC enhanced PDOs demonstrated the cells to ubiquitously express SHH and PTCH1 and to be enriched for nuclear GLI1 negative cells. Conversely, in vivo, where tumor cells are more differentiated than in the serum-free undifferentiated conditions of the cancer organoid culture system, near universal expression of nuclear GLI1 was observed. However, immunostaining of frozen xenograft sections demonstrated that rare GLI1 negative cells also exist in vivo. Interestingly, these cells were frequently located at the base of crypt-like structures within the tumor. Similar to PDOs, SHH was ubiquitously expressed in epithelial cells in vivo but was also detected in stromal cells, demonstrating that both paracrine (epithelium to mesenchyme) and autocrine Hedgehog signaling are active in vivo. Overall, these data suggested that GLI-dependent Hedgehog signaling (observed in vivo) is required for the differentiation of tumor cells but that PTCH1-dependent GLI-independent Hedgehog signaling, observed in the CSC-enriched PDOs, maintains CSCs in an undifferentiated state.

In canonical Hedgehog signaling, Hedgehog (SHH, IHH or DHH) binds to its receptor PTCH1, which is also a direct target of the Hedgehog pathway,⁶ to relieve it from its repression of SMO, which then activates a downstream signaling pathway that results in the activation of GLI family of zinc finger transcription factors GLI1, GLI2, and GLI3. These GLI proteins then translocate to the nucleus where they regulate the transcription of several target genes (Fig. 1). Non-canonical Hedgehog signaling is less well defined but generally refers to Hedgehog dependent signals that do not act via the canonical Hedgehog-to-GLI route and includes any GLI-independent cellular and tissue responses.⁷

Figure 1.

Inverse relationship between canonical and non-canonical Hedgehog signaling in colon cancer stem cell (CSC) differentiation. Hedgehog signaling in colon cancer is SHH dependent and both autocrine and paracrine signaling occurs. In canonical Hedgehog signaling, SHH binds to its receptor PTCH1 to relieve it from its repression of SMO, which then activates a downstream signaling pathway that results in the activation of GLI. GLI proteins then translocate to the nucleus where they regulate the transcription of target genes that lead to the inhibition of Wnt signaling and tumor cell differentiation. Binding of SHH to PTCH1 in colon CSCs results in SMO-independent GLI-independent non-canonical Hedgehog, which acts as a positive regulator of Wnt signaling to block CSC differentiation.

Two classes of non-canonical Hedgehog signaling have been described: Type I non-canonical Hedgehog signaling, which works through PTCH1 and is independent of SMO and Type II, which functions through SMO.⁷ In addition, SMO-independent GLI activation has also been referred to as non-canonical Hedgehog signaling.⁸

To further investigate the role of non-canonical Hedgehog signaling in colon CSCs, PDOs were treated with small molecule inhibitors (RU-SKI 43, vismodegib, cyclopamine) and RNA interference against Hedgehog signaling components Hedgehog acyltransferase (HHAT), SMO and PTCH1. Inhibition of SMO-dependent canonical Hedgehog signaling had no effect on cell survival. In contrast, targeting HHAT, which is required for the palmitoylation and activity of SHH, effectively reduced cell survival and spheroid formation in vitro and tumorigenesis in limiting dilution xenotransplantation. Furthermore, siRNA targeting of PTCH1 demonstrated the pathway to be PTCH1-dependent. Significantly, addition of recombinant SHH attenuated the effect on CSC survival. These data demonstrated non-canonical Hedgehog signaling in colon CSCs to be SHH-dependent, PTCH1-dependent and SMO-independent (Type I non-canonical Hedgehog signaling).

Targeting HHAT in colon CSCs resulted in decreased expression of stem cell associated and Wnt signaling genes and increased expression of differentiation genes. To further elucidate the relationship between Hedgehog and Wnt signaling, TCF/LEF-eGFP Wnt reporter cells were treated with SMO-targeting vismodegib, which led to an increase in Wnt signaling activity, and with HHAT inhibitor (SHH targeting) RU-SKI 43, which caused a decrease in Wnt activity. These data support a model wherein canonical SMO-dependent Hedgehog signaling leads to a downregulation of Wnt signaling and tumor cell differentiation, whereas non-canonical PTCH1-dependent Hedgehog signaling acts as a positive regulator of Wnt to maintain CSCs in an undifferentiated state (Fig. 1).

This was further supported through the analysis of clinical samples where PTCH1 expression was found to positively correlate with Wnt signaling and stem cell genes and negatively correlate with differentiation genes. GLI1 expression, on the other hand, negatively correlated with Wnt and stem cell genes and positively correlated with differentiation genes. Significantly, PTCH1 was most highly expressed in late stage tumors, suggesting that PTCH1 may be a potential new marker of prognosis in colorectal cancer.

Recent clinical trials in metastatic colorectal cancer involving the SMO antagonist vismodegib have yielded negative results.⁹ However, our research suggests that targeting non-canonical PTCH1-dependent Hedgehog signaling may offer more effective treatment strategies. For example, inducing CSC differentiation by blocking SHH signaling through the targeting of HHAT in combination with standard of care tumor debulking therapies.

Overall, these data demonstrate that colon CSC survival is regulated by SHH-dependent PTCH1-dependent non-canonical Hedgehog signaling and support HHAT, which has also demonstrated therapeutic potential in Hedgehog-dependent pancreatic cancer and breast cancers,^8,10 as a possible therapeutic target for the future development of anti-tumor treatments in colorectal cancer.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

The author apologizes to colleagues whose work could not be cited due to space limitations and thanks all the authors of the original paper for their contributions. The research leading to these results received support from the Innovative Medicines Initiative Joint Undertaking under grant agreement No. 115234 (OncoTrack), resources of which are composed of financial contribution from the European Union's Seventh Framework Programme (FP7/2007-2013) and EFPIA companies in kind contribution.