Classical (CL) and basal (BL) subtypes of PDAC are recapitulated in pancreatic cancer cell lines, organoids, and PDX cells and their phenotype correlates with the expression of transcription factors (TFs), such as GATA6, HNF1A, p63 or MYC. Super-enhancers regulating key transcription-factors for cell identity, such as GATA6 or MYC, will further drive acquisition of CL or BL PDAC programs. Interconversion between subtypes is likely to happen as a result of the combined effect of acquired mutations in the tumour cells and additional signals from the microenvironment. This phenotypic plasticity has clinical implications because some subtypes respond differently to standard therapies. Since epigenetic changes at SEs are reversible, modulation of epigenetic writers, readers or erasers may convert resistant tumours into responder tumours which - in turn - may improve patient survival.
To understand better what is the contribution of the tumour microenvironment to the diversity and plasticity of PDAC subtypes I have used subtype-specific reporter cell models (monitoring of subtype dynamics) and culture systems where the direct and indirect communication with stromal cells can be addressed (effect of TME on subtype plasticity).
In our results we could observe that some PDAC cells show enrichment of subtype specific signature (classical and basal), however the expression of key marker genes is variable - some cells express high level of CL or BL genes, however in others those genes are low expressed and are considered to have intermediate state. To follow subtype plasticity in an easy and fast way, we introduced a cDNA coding for a fluorescent protein into genes that are key markers of subtype identity. The fluorescent protein expression in this system is found under the endogenous promoter of target gene, therefore every time marker gene is expressed, or its expression is changes it is reflected at the level of fluorescent gene/protein. When GFP was inserted into the target gene of a classical cell line and positive cells selected by sorting, we observed heterogeneous level of fluorescent protein expression and approximately 50% of cells were marked as target protein/fluorescent protein positive. Since the GFP signal in sorted cells was not very bright, it is possible that the gating strategy we applied was not sufficiently stringent, thus some negative GFP cells were collected and after expended in culture. When we made single cell clones from edited cells and checked GFP expression we indeed found some clones with high GFP and some with low or negative signal. In experiments where target gene and protein expression could be modulated, the GFP signal followed these changes similarly. Overall, we generated a cell model where subtype specific marker expression can be easily monitored, allowing us to study the influence of stromal cells on subtypes plasticity. When PDAC cell from different subtypes were treated with condition media from PDAC-derived cancer-associated fibroblast we observed context dependant effect on cell proliferation. iCAF condition media increased proliferation of only CL cells (Patu8988S), but this effect was absent when myCAF condition media was used. BL cells (PL45) were inert to the effect of both CAF subtypes. However, there was no significant change in the expression of CL/BL marker genes, regardless weather PDAC cells were threated with condition media from both iCAFs and myCAFs or plated in transwells. These experiments were performed in culture systems where cells are attached to the plastic (2D), which might not recapitulate the real scenario found within tumour. Instead, 3D model might be more permissive for signals received from surrounding cells. Moreover, it is possible that the cell models we used, like Patu8988S or PL45 are found on the far extremes of their classical or basal program, so less prone to switch. Indeed, our gene expression analysis showed a panel of cell lines with intermediate gene expression feature, which might be easier to push to another subtype programme. Further experiments, using other cell models and 3D co-culture systems, are planned to respond to these questions. We disseminated our findings at several internal meetings or participated in outreach events, such as Noche de los Investigadores organized within CNIO or by Madri+D (2020,2021 and 2022). Moreover, we discussed our findings at the PAN3DP conference 2022 held in London.
