ABSTRACT: Recent fate mapping studies concluded that EMT is not required for metastasis of carcinomas. Here we challenge this conclusion by showing that it failed to account for possible crosstalk between EMT and non-EMT cells that promotes dissemination of non-EMT cells. In breast cancer models, EMT cells induce increased metastasis of weakly metastatic, non-EMT tumor cells in a paracrine manner in part by non-cell autonomous activation of the GLI transcription factor. Treatment with GANT61, a GLI1/2 inhibitor, but not with IPI-926, a Smoothened inhibitor, blocks this effect, and inhibits growth in PDX models. In human breast tumors, the EMT-TFs strongly correlate with activated Hedgehog/GLI signaling, but not with the Hh ligands. Our findings indicate that EMT contributes to metastasis in part via non-cell autonomous effects that activate the Hh pathway. While all Hh inhibitors may act against tumors with canonical Hh/GLI signaling, only GLI inhibitors would act against EMT-induced GLI activation. In recent years, immunohistochemical analyses and multiplex high-throughput single cell sequencing of human tumor cells have shown that tumors are composed of diverse cell subpopulations containing different driver mutations, gene and protein expression profiles, growth rates, and responses to chemotherapeutics. Such heterogeneity is exacerbated by cellular plasticity, where some cells may undergo oncogenic epithelial-to-mesenchymal transition (EMT), resulting in loss of cell-cell adhesion and polarity, reduced epithelial and elevated mesenchymal protein expression, increased migration and invasion, and enhanced dissemination from the primary tumor. Because metastases in patients appear epithelial, the reverse process, mesenchymal-to-epithelial transition (MET), may occur to allow tumor cell colonization in secondary metastatic sites, establishing cellular plasticity as an important aspect of tumor progression. However, the role of EMT in carcinoma metastasis is controversial. Recent lineage tracing studies argue against the requirement of EMT for metastasis, as reporter-tagged cells that underwent a previous EMT were not found at the secondary site. However, these studies did not address the potential cooperation between EMT and non-EMT cells during the metastatic process, as EMT cancer cells may enable non-EMT cells to gain access to the secondary site, leading to macrometastatic growth. Thus, metastasis can be influenced by intratumoral heterogeneity: where a small proportion of primary tumor cells that have undergone an EMT may influence neighboring, non-EMT tumor cells. Twist1, Snail1, and Six1 are EMT-inducing transcription factors (EMT-TFs) that have all been associated with breast cancer metastasis. All three EMT-TFs regulate critical developmental processes such as cell survival, migration and invasion; in part by influencing EMT. In addition, they are downregulated post-embryogenesis, but re-expressed in various cancers where they cell autonomously induce EMT, resulting in increased percentages of tumor initiating cells (TICs) and enhanced metastasis. In carcinomas, Twist1 and Snail1 transcriptionally repress E-Cadherin (E- Cad) and upregulate mesenchymal genes. Similarly, Six1 overexpression induces EMT by regulating E-Cad localization and altering other EMT markers. During development and cancer, EMT-TFs act in concert with several signaling networks including TGFβ, Wnt, and Hedghog (Hh). The Hedgehog (Hh) signaling pathway is a prominent regulator of embryonic development, where Hh ligands function as morphogens to control numerous developmental processes. Interestingly, in Drosophila eye development, hh is a direct target of sine oculis (the homolog of Six1) and Six1 regulates Hh/GLI signaling during lung development and in fibroblasts. Additionally, Twist1 and Hh/GLI signaling are intimately linked during development, and recently Twist1 and Snail1 were associated with the Hh pathway in TICs. In mammals, canonical activation of Hh/GLI signaling involves binding of one the Hh ligands such as(either, Desert (DHH), Indian (IHH), or Sonic Hedgehog (SHH) to Patched-1 (PTCH1) or Patched-2 (PTCH2) receptors, relieving the inhibitory activity of PTCH on Smoothened (SMO). When inhibition is relieved, levels of the transcriptional activator forms of one or more GLI transcription factors (GLI1, 2, or 3) increase in the nucleus, resulting in activation of Hh target genes 12. Non-canonical activation of the GLI TFs can occur in a Hh- or SMO-independent manner via secreted factors like TGF-β19. Importantly, autocrine and paracrine Hh-mediated crosstalk between tumor cells and the tumor microenvironment 20 results in increased proliferation, stem cell self-renewal and metastasis in various cancers 21. In basal cell carcinoma (BCC) and medulloblastoma, activated Hh signaling is often due to mutations in pathway components such as PTCH and SMO, while in other tumor types including breast, mutations are not observed at high frequency. Instead there is evidence for Hh ligand-dependent pathway hyperactivity. Because numerous studies link Hh signaling to progression in multiple tumor types, derivatives of cyclopamine (e.g. GDC-0449 and IPI926), a naturally occurring plant-derived steroidal alkaloid which targets SMO, are in clinical trials for select patients with BCC and medulloblastoma with promising efficacy. However, these inhibitors have not shown promise in breast cancer despite evidence for activation of this pathway. Herein, we demonstrate that EMT-TFs Twist1, Snail1 and Six1 influence neighboring carcinoma cells in a non-cell autonomous manner, by increasing EMT features and aggressive properties of cells not expressing these TFs. Six1 is a key mediator of the non-cell autonomous effects downstream of Twist1 and Snail1, and can induce metastasis of non-Six1, non-EMT cells. All three EMT-TFs function non-cell autonomously by activation of GLI-mediated transcription in non-EMT cells, but employ different mechanisms of pathway activation, some of which are Hh ligand and SMO-independent. We find that pharmacological inhibition of GLI, but not SMO, in the non-EMT cells efficiently inhibits the non-cell autonomous phenotypes imparted by all three EMT-TFs. Importantly, we demonstrate that in selected patient derived breast cancer xenograft (PDX) models endogenously expressing high levels of these EMT-TFs (but not Hh ligands), GANT61 (a GLI1/2 inhibitor) inhibits tumor growth whereas IPI926 (a SMO inhibitor) did not. Our data suggest that upstream SMO inhibitors may not prove efficacious in tumors where a proportion of cells activate GLI-TF via EMT-TFs, and instead argue that inhibitors directly targeting GLI may be more effective.
