The importance of integrated genomics to uncover clinically relevant cancer driver genes

Abstract

Tumor heterogeneity hinders the discovery of true cancer driver genes. The recently identified role for YY1-associated protein 1 (YY1AP1) as a critical oncoprotein that is activated in a particular subtype of hepatocellular carcinoma (HCC) was elucidated through integrative genomic profiling followed by functional studies. This strategy proved effective for finding a functional cancer driver with prognostic impact.

Cancer is a consequence of defective adaption involving accumulating genetic or functional alterations and selection. Driver genes, which represent ideal targets in cancer therapy, are defined as initiating alterations that drive or critically support the maintenance of an oncogenic state leading to cellular dependence, known as oncogene addiction.¹ Although cancer driver genes are increasingly becoming appreciated in clinical applications, especially for patients who have tumors at advanced stages, it is challenging to unambiguously identify clinically relevant cancer driver genes from a vast number of cancer-related abnormalities. This is because tumor evolution has led to the acquisition of hundreds of genomic and epigenomic alterations along the way. A majority of cancer-related abnormalities are passengers that do not necessarily function to promote tumor growth or are historical events that may support cancer initiation at an early stage of carcinogenesis. These events may have been important at a particular stage in tumor development, but are no longer functional during tumor progression.² Therefore these cancer-associated genes are not optimal drug targets. Currently, a major impediment to identifying driver genes is tumor heterogeneity, which presents as both inter- and intra-tumor variation.

Many factors can contribute to tumor heterogeneity. Various etiological factors, such as hepatitis B or C virus, obesity, and alcohol consumption, may contribute to both inter- or intra-tumor heterogeneity that likely arises from a proposed branched evolution model during tumor growth and maintenance.^3,4 However, these obstacles may be overcome by achieving the following: (1) accurate identification of homogeneous cancer subgroups; and (2) integrated genomic and functional strategies to decipher and characterize driver genes within each subgroup.

The idea that tumors originate from cancer stem cells (CSCs)⁵ and that elimination of this small subset of cells can potentially cure cancer is supported by studies in intestinal adenomas and a glioblastoma mouse model.^6,7 The first experimental evidence in humans showing CSCs as the origin of malignancy was recently provided in myelodysplastic syndromes (MDS) by backtracking the lineage of genetic lesions to MDS stem cells.⁸ Genome-wide studies have enabled us to stratify patients into different subgroups with common hallmarks. In 2008, Yamashita et al. stratified hepatocellular carcinoma (HCC) patients based on α-fetoprotein (AFP) and epithelial cellular adhesion molecule (EpCAM) expression into 4 subgroups, which displayed distinct gene expression patterns with features resembling certain stages of hepatic lineages.⁹ Among them, the hepatic stem cell (HpSC)-like HCC subtype is defined by EpCAM⁺ AFP⁺, a molecular signature of a tumor with features of hepatic progenitor cells and poor prognosis, whereas the mature hepatocyte (MH)-like HCC subtype is defined by EpCAM⁻ AFP⁻, a molecular signature of a tumor with features of mature hepatocytes and good prognosis. These two subtypes represent 2 ends of lineage-restricted differentiation. Theoretically, by comparing HpSC and MH groups, we should be able to identify genomic alterations that are critical to drive and maintain the stemness status. However, intra- and inter-tumor heterogeneity are still evident within each defined subtype, calling for more stringent criteria in order to minimize the effect of tumor heterogeneity.

In this regard, we recently reported a new strategy involving selection of only a very small, homogeneous subset of patients within the HpSC and MH subgroups with an extremely high or extremely low level of EpCAM and AFP detected in the tumor, which we called x-HpSC and x-MH, respectively. By employing an integrated genomic screening and detailed functional validation, we revealed that YY1-associated protein (YY1AP1) is a key driver gene of cancer stemness.¹⁰

The successful identification of a cancer stemness driver gene illustrates the power and feasibility of a strategy integrating genomic profiling with functional investigation and analysis. Selecting patients representative of 2 homogenous populations within a hepatocytic lineage, representing the end points in the spectrum of cell lineage, allowed for easier identification of targets to study at the molecular level and robust confirmation of a gene driving tumorigenesis. In this study, 2 groups of patients with high homogeneity in the composition of their tumor were identified. Comparing these 2 distinct groups using genomic analysis followed by specific functional investigation lead to the identification of a key stemness driver gene, YY1AP1. Although this gene was identified by comparing small groups of homogeneous patients, it is encouraging to see that YY1AP1 abnormality is applicable to all patients whose tumor originates from cancer stem cells, despite the heterogeneous features of the tumor (Fig. 1). Moreover, we extensively characterized YY1AP1 as an oncogenic driver and potential druggable molecular target through integrative genomics and molecular analysis.

Figure 1.

Sequential acquisition of genetic lesions in hierarchically organized hepatocytes. (A) Upper, 2 hepatic cancer stem cell-originated tumor types; lower, 2 non-hepatic cancer stem cell-originated tumor types. Well-defined groups represent tumors bearing only simple initiating mutation or non-branched mutations that display high homogeneity. (B) A strategy of linear comparison of 2 well-defined groups results in the discovery of drivers of cancer stemness.

Taken together, these data show that the strategy of selecting well-defined patient populations and applying integrative genomics is successful in revealing key cancer driver genes while making tumor heterogeneity more understandable. We suggest that this strategy has great potential to enhance our genome-wide efforts to identify clinically relevant genes and may be applicable to other tumor types with extensive intra- or inter-tumor heterogeneity.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by the Intramural Research Grants of the Center for Cancer Research, National Cancer Institute.