A ZEB1-HDAC pathway enters the epithelial to mesenchymal transition world in pancreatic cancer.

Epithelial to mesenchymal transition (EMT) correlates with high-grade malignancy including the competence to form metastases. In addition, EMT has recently been linked to cellular self-renewal programmes of cancer stem cells and apoptosis/anoikis resistance, which are all features of therapeutic resistance. The EMT programme is driven by several transcription factors (TFs), such as the transcriptional regulators SNAIL, SLUG, ZEB1 and ZEB2 and the basic helixeloopehelix factors E47 and TWIST. These proteins target and repress the CDH1 gene, which encodes for E-cadherin, an important caretaker of the epithelial state. Expression studies in human pancreatic cancer showed expression of SNAIL in 78% and of SLUG in 50% of cases. Although no or low levels of TWIST are expressed in pancreatic cancers, up-regulation of this gene under hypoxic condition may argue for a contribution towards tumour progression. 2 Expression of the ZEB2 gene was recently found to be silenced by promoter methylation in the majority of pancreatic cancers. This finding argues against ZEB2 as a major repressor of E-cadherin in pancreatic cancer. In addition to such expression data, the functional relevance of SNAIL and SLUG for EMT and repression of the CDH1 gene has been described in various pancreatic cancer models in vitro and in vivo. Aghdassi et al (see page 439) present new compelling evidence that the zincfinger TF ZEB1 is a repressor of E-cadherin expression. Based on the observation that 40% of pancreatic cancers have reduced Ecadherin levels and that low E-cadherin expression correlates with a poor prognosis after pancreatic cancer resection, regulation of E-cadherin was investigated. Mutations of CDH1 and hypermethylation of CpG islands in the CDH1 promoter could be excluded as general mechanisms for reduced E-cadherin expression. When they compared the levels of E-cadherin and ZEB1, Aghdassi et al detected an inverse correlation in pancreatic cancer cell lines and tumour specimens. Consequently, binding of ZEB1 to the CDH1 promoter was found specifically in cell lines lacking E-cadherin expression and inhibition of ZEB1 expression restored E-cadherin expression, grabbing ZEB1 into the row of EMT regulators and CDH1 repressors in pancreatic cancer. In accordance with these data, the Brabletz group has recently demonstrated that ZEB1 represses the expression of miR-200 family members in pancreatic cancer cells, which contributes to the activation of the tumour promoting NOTCH pathway. ZEB1 expression was especially found in undifferentiated (G3 and G4) pancreatic cancers, restricted to invasive areas with signs of EMT, which confirms the inverse correlation of E-cadherin with ZEB1 described by Aghdassi and colleagues. EMT-TFs can act in hierarchical cascades to cooperatively induce EMT. For example, TWIST can activate SNAIL and SLUG expression, whereas SNAIL and SLUG can drive ZEB1 expression. Consistent with this, Aghdassi et al observed that high SNAIL and ZEB1 expression was often directly correlated in pancreatic cancer cell lines. Furthermore, the IKK-NFkB signalling pathway, a well characterised inducer of EMT, was recently shown to induce SNAIL and ZEB1 in pancreatic cancer cells. While these observations provide some evidence for the existence of a SNAIL-ZEB1 axis, further studies are needed to precisely identify an EMT-TF hierarchy in pancreatic cancers. Considering the diverse features cells acquire during EMT, such as invasiveness, migration, stemness or anoikis resistance, it will be important to decipher where EMT-TFs possess unique competence and where and how the factors cooperate. As a consequence of the fast progress in sequencing technologies, we know that pancreatic cancer is characterised by an extreme genetic heterogeneity. Therefore, it is important to find and distinguish molecular mechanisms, which are common in pancreatic cancerdthe node concept. For the regulation of E-cadherin during EMT in pancreatic cancer cells, Aghdassi et al now demonstrate that histone deacetylases (HDACs), especially HDAC1 and HDAC2, are critically involved. HDACs deacetylate the e-amino group of lysines located at the N-terminal tail of histones. This can lead to a repressive chromatin structure (heterochromatin) and altered gene transcription. By modulating these epigenetic acetylation marks, HDACs can promote proliferation and confer therapeutic resistance. HDAC1 and HDAC2 are both overexpressed in pancreatic cancer, and especially high HDAC2 expression was observed in poorly differentiated cancers. 18 Aghdassi et al demonstrate that ZEB1 can directly bind HDAC1 as well as HDAC2 and that ZEB1 can recruit a HDAC1/2 containing repressor complex to the CDH1 promoter in pancreatic cancer cells. Accordingly, the repression of the CDH1 gene in mesenchymal murine pancreatic cancer cell lines derived from the genetic Kras model, as well as the repression of E-cadherin in TGFb-induced EMT of pancreatic cancer cells, rely onHDAC activity. Furthermore, repression of E-cadherin by HDAC1/2 was recently demonstrated with in vivo selection models of pancreatic cancer EMT. Recruitment of catalytically active HDAC1 or HDAC2 to the CDH1 promoter was demonstrated with chromatin immunoprecipitation assays in tissue sections of human pancreatic cancer, corroborating the importance of HDACdependent repression of E-cadherin in vivo. The observation that Aghdassi et al detected exclusively HDAC1 or HDAC2 recruited to the CDH1 promoter in human pancreatic cancer tissue section is intriguing. Both enzymes are highly homologous, cooperatively act in corepressor complexes, and share redundancy for many biological processes. Although the reasons for the specific recruitment of either HDAC remain unclear, an explanation might be the different ability of the II Medizinische Klinik, Klinikum rechts der Isar, Technische Universität München, München, Germany; Friedrich-Schiller-University Jena, Center for Molecular Biomedicine, Institute of Biochemistry and Biophysics, Jena, Germany