Research

Cancers that arise in different tissues use different oncogenic mechanisms for tumor initiation and progression. However, despite the extensive genetic data that highlight the role of tissue-specificity in cancer, how factors that define tissues interact with cancer mutations to promote cancer development remains elusive.

Molecular basis of tissue-specificity in carcinogenesis

Our team’s early work discovered gene regulatory mechanisms that control metastasis gene activation and metabolism in clear cell renal cell carcinoma (ccRCC) (; ). Interestingly, these results highlighted tissue-specific mechanism of gene regulation in metastasis. Indeed, ccRCC is characterized by a unique set of genetic alterations, in particular the inactivation of the von Hippel-Lindau tumor suppressor gene (VHL), which can be detected in up to 90% of ccRCCs, but not in other common cancer types.

Capitalizing on the distinctive genetic make-up of ccRCC, we have recently used a comprehensive set of experimental approaches ranging from genetic screening, in vivo cancer models, functional genomics and human genetics, to demonstrate that the oncogenic activity of VHL mutations and the downstream activation of the hypoxia-inducible factor 2a (HIF2A), a central driver and clinically validated therapeutic target in ccRCC, is dependent on the renal developmental factor Paired box 8 (PAX8). Specifically, PAX8 and HIF2A interact at a critical gene regulatory region upstream of Cyclin D1 (CCND1), an oncogene that is required for cell cycle progression in several cancer types. We showed that the activity of this CCND1 enhancer varies between individuals due to a common genetic variant (rs7948643) that can alter PAX8 binding at this genomic site. Finally, our data show that PAX8, through the downstream mediator HNF1B, regulates the expression of MYC, another central oncogene.

These results () demonstrate that the short ~30bp stretch of non-coding DNA that harbors the PAX8 and HIF2A binding sites within a distal oncogenic CCND1 enhancer is responsible for a very significant fraction of human ccRCC burden. Most importantly, our results explain why the two most common genetic alterations that cause ccRCC, the germline variant rs7948643 and VHL mutations, have a strong association with ccRCC, but not with other cancers. Similar interaction between cancer type-specific genetic alterations and homeostatic and developmental transcriptional programs is likely to underlie tissue specificity in other cancers as well, a question we are following up on in our current work. Our results also suggest that several layers of epigenetic conditioning are required for the activation of oncogenic enhancers, but the mechanisms remain incompletely understood. Using a diverse set of state-of-the-art functional genomics tools, we are currently exploring such mechanisms.

Lineage factors as targets for cancer therapy

Our discovery of the central role of PAX8 in renal carcinogenesis suggests that PAX8 and other lineage factor pathways could be exploitable targets for novel cancer therapies. On the other hand, in some cancers lineage switching is known to induce therapy resistance, but how lineage fidelity is maintained and how it can be lost remain poorly understood. We are currently using CRISPR/Cas9-based genetic screening and experimental cancer models to explore the potential of lineage factors as therapeutic targets.

The spread of cancer to distant organs, or metastasis, cause most cancer-related deaths. However, the mechanisms that facilitate cancer metastasis remain poorly understood. We are using experimental models and functional genomics to understand how cancers become metastatic and how this process could be inhibited.

Enhancer co-option as a mechanism of metastatic cancer progression

The transcriptional traits that promote late tumor phenotypes, such as metastasis, may arise already in primary tumours (; ). Using our experimental metastasis models () together with validation in clinical samples we have discovered that the phenotypic output of a tumor-initiating pathway, i.e. the VHL-HIF2A pathway in ccRCC, can evolve in support of metastasis through co-option of specific transcriptional enhancer elements (). Thus, advanced cancer phenotypes, of which metastatic capability represents an endpoint, emerge in the primary tumor through a process that is dependent on, but not directly caused by, early oncogenic mutations. This new model explains in part how similar early driver mutations can eventually lead to different advanced cancer phenotypes and clinical behaviors, especially since specific mutations underlying metastasis have not been identified, as discussed in our recent review article (). Similar modulation of the functional output of cancer driver mutations could also be critical for the earlier steps of carcinogenesis. Using our unique experimental models and state-of-the-art genomics combined with functional analysis, our future work aims at further understanding the gene regulatory mechanisms and the role of the tumor microenvironment in the maintenance of metastatic tumors.