Fusion RNA profiling provides hints on cell of origin of mysterious tumor

ABSTRACT

Biological samples can be grouped into separate clusters based on their gene expression profiles. This approach has yielded meaningful biological insights and facilitated biomarker discoveries. Recently, we developed another approach to study connections between biological samples based on their fusion RNA expression. We have used this approach to provide insights into the cell of origin for a mysterious tumor, alveolar rhabdomyosarcoma.

Gene expression profiling has been successfully used to cluster biological samples into meaningful groups, such as cancer versus normal, or different subtypes of cancer. Perou et al. first described 6 intrinsic subtypes of breast cancer using unsupervised analysis of gene expression profiles,¹ although 4 intrinsic subtypes, luminal A, luminal B, HER2+/ER−, and basal-like, are now widely accepted. The intrinsic subtypes correlate with known pathologic characteristics, and each subgroup has distinct clinical outcomes.²

Recently, we demonstrated that fusion RNA profiling is a novel way to group biological samples and can reveal connections between samples.³ Contradictory to the prevailing view that fusion RNAs are cancer-unique phenomena, we and others have showed that they are widespread in non-cancer cells and tissues.^4-6 We coined the phrase “fusion transcriptome," and demonstrated that “fusion RNA profiling” can cluster biological samples into separate groups of different tissues. To put the concept into use, we used alveolar rhabdomyosarcoma as a model.

Alveolar rhabdomyosarcoma (ARMS) is one of the most common neoplasms in children and adolescents. ARMS occurs mostly in limbs and has morphologic features of fetal muscle, but can also occasionally arise at sites that normally lack skeletal muscle such as the genitourinary and biliary tracts. Despite extensive research, the cell of origin for this tumor is still under debate.^7-9

The majority of ARMS have a (2;13) chromosomal translocation resulting in the formation of a gene fusion between Paired Box 3 (PAX3) and Forkhead Box O1 (FOXO1). Like many fusions in other tumor types, PAX3-FOXO1 is believed to be a driving force of ARMS tumorigenesis and therefore unique to ARMS cells. However, our recent published data challenged the dogma by demonstrating the presence of PAX3-FOXO1 RNA and protein products during normal myogenesis.¹⁰ The fact that endogenous PAX3-FOXO1 is found in some normal cells raises the possibility that these may be the cells from which ARMS originate, since they are permissive for, if not responsive to, this fusion gene product.

In our recently published work we presented further evidence to support this hypothesis.³ We performed paired-end RNA-sequencing at 4 time points throughout skeletal muscle differentiation from mesenchymal stem cells (MSCs), in addition to an ARMS cell line, RH30. We found hundreds of fusion RNAs. Interestingly, only one time point of muscle differentiation shared common fusion RNAs with RH30 cells (Fig. 1). This was also the only time point when PAX3-FOXO1 was detected. We validated this finding using 2 other MSC systems. Even though the exact time point(s) of PAX3-FOXO1 expression varied, the fusion RNA profiles of these time point(s) were most similar to that of RH30. Impressively, using qRT-PCR we found that all 18 fusion RNAs expressed in RH30 can be detected at the time point of myogensis when PAX3-FOXO1 is expressed. Among them, 7 had exactly the same expression pattern as PAX3-FOXO1.

Figure 1.

Fusion RNA profiling indicates a connection between time point 3 of myogenesis and RH30 cells. Mesenchymal stem cells (MSCs) were induced to differentiate along the skeletal muscle lineage. Four time points (T1-T4) throughout myogenesis were analyzed. RNA-sequencing was performed for these samples and for an alveolar rhabdomyosarcoma cell line, RH30. Circos plot depicts chimeric RNAs discovered across the genome; lines connect parental genes and half lines are due to the close proximity between the 2 parental genes. Outer ring: chromosomes. Innermost ring: lines denote the chimeric RNAs connecting 2 parental genes. Middle two rings: heatmap and histogram depicting gene expressions. Arrows indicate the 5 chimeras common to T3 and RH30 samples.

We further ruled out the possibility that the additional 17 fusions are downstream targets of PAX3-FOXO1 and the similarity between the myogensis time point and RH30 is merely due to PAX3-FOXO1. Histologically, rhabdomyosarcoma belongs to the group of small-blue-round-cell tumors (SBRCTs), which also includes Ewing's sarcoma and synovial sarcoma. We showed that all 7 fusions are specific to rhabdomyosarcoma. In addition, we validated expression of the fusions in a panel of clinical rhabdomyosarcoma tumor cases. All of these findings are consistent with the hypothesis that the PAX3-FOXO1 expressing cells during myogenesis are the cells of origin for ARMS.

The debate about the exact cell of origin of ARMS partially stems from the traditional “candidate cell” approach: in both cell culture and animal model systems, a certain type of cell was selected and tested to determine whether it could be transformed upon the introduction of oncogenic factors such as PAX3-FOXO1. However, the cells that are accessible and can be transformed are not necessarily the cells from which a particular tumor arises. We now propose a different strategy that involves monitoring the cancer-signature fusion RNAs associated with the tumor throughout the normal muscle cell differentiation process. We propose that the pattern of fusion RNA expression can be used as a fingerprint to identify candidates for cell of origin. This is the first time, to our knowledge, that fusion RNA profiling has been used to provide hints for the etiology of a disease.

The fusion transcriptome clustering with a binary input (present or absence of a chimera) behaved as well as traditional transcriptome clustering in human body maps³ and uncovered a connection between normal myogenesis and ARMS³ that could not be revealed by whole transcriptome analyses. Three factors may contribute to these results: (1) Gene expression profiles are highly influenced by changes in genes that are not associated with specific cell types; (2) In contrast, chimeric RNAs in general are often tightly associated with specific cancer types, and/or tissue differentiation lineages (manuscript in preparation); (3) Unlike the expression of most cellular genes, where the level of expression is used as a quantitative trait, the chimeric RNAs may be viewed as potential qualitative traits.

Several immediate questions need to be answered, and further evidence must be provided to validate the hypothesis that PAX3-FOXO1–expressing cells are the cells of origin for ARMS. What are these cells? PAX3-FOXO1–expressing cells must be separated from the other cells and characterized. Can they be transformed into ARMS? Additional molecular changes, such as dominant-negative TP53, RAS, and MYC mutations can be introduced to test whether the cells have similar characteristics to ARMS in both the cell culture systems and in vivo animal models. In addition, the connection between the fusion transcript and gene fusion at the chromosomal level is not clear. More studies are needed to understand the regulation mechanism of the fusion transcripts, including their transient nature, and their specificity.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This work is supported by St. Baldrick's V Scholar grant.