The discovery of AVIL as a bona fide oncogene in glioblastoma

ABSTRACT

Our work in rhabdomyosarcoma led us to the discovery of a novel oncogene, Advillin (AVIL) in glioblastoma. Multiple lines of evidence support that AVIL is an Achilles heel of glioblastoma, with its specific targeting potentially an effective treatment approach for the disease. A new signaling axis was also established.

Glioblastoma (GBM) is one of the most deadly forms of human cancer. GBMs are aggressive, highly invasive, and resistant to chemotherapy and radiotherapy. The mean survival duration of patients with GBM, the most malignant form of glioma, is approximately 1 year with no effective therapy to date. The highly invasive and proliferative nature of GBM renders the tumor relapse and incurable. The current standard option, radiation plus temozolomide, only displayed a 2.5 month better survival rate.¹ Clearly, a better understanding about the disease and effective therapeutic targets are urgently needed.

Advillin (AVIL) is a member of the villin/gelsolin family of actin-regulatory proteins. It binds actin and may play a role in the development of neuronal cells that form ganglia.² Recently, we showed for the first time, that it is a bona fide oncogene, and plays a critical role in the tumorigenesis in GBMs.³ This study is an extension of our previous work. In the previous study, we discovered AVIL forming a gene fusion with a housekeeping gene, Methionyl-TRNA Synthetase (MARS), in a pediatric tumor, rhabdomyosarcoma.^4–6 Suspecting that other cancers may dysregulate AVIL using different mechanisms, we found that AVIL gene locus is amplified in 15%–18% of glioblastoma cases in The Cancer Genome Atlas (TCGA) studies.⁷ However, at protein level, we observed AVIL overexpressed in all of the GBMs we tested. More impressively, we showed that AVIL is expressed at even higher levels in GBM stem/initiating cells (GSC/GIC). In contrast, AVIL is hardly expressed or expressed at a much-reduced level in astrocytes, neural stem cells, or normal brain tissues. In addition, AVIL expression correlates with patient prognosis, with high AVIL expression being associated with worse outcomes.

To investigate the effect of AVIL in GBM tumorigenesis, we conducted both loss- and gain-of-function studies. In the loss-of-function studies, we found that silencing AVIL killed GBM including GSC/GIC cell cultures, and dramatically inhibited in vivo xenografts in mice, but had no effect on control astrocytes or neural stem cells. Live cell imaging showed also reduced migration activity associated with AVIL silencing. Conversely, in the gain-of-function studies, overexpressing AVIL promoted cell proliferation and migration in GBMs and astrocytes in vitro, enabled fibroblasts to escape contact inhibition, and most importantly, transformed immortalized astrocytes. These evidences support that AVIL is a bona fide oncogene.

Mechanistically, AVIL binds to F-actin and regulates F-actin dynamics. We found that this interaction is crucial for the oncogenic activity of AVIL, as the point mutants at the headpiece domain, which disrupts F-actin binding, are defective in rescuing AVIL silencing. When we performed whole transcriptome analyses merging data from both AVIL silencing and AVIL overexpression systems, a few candidates jumped into our eyes including Lin-28 Homolog B (LIN28B). It has been reported before that LIN28B can promote proliferation, and invasion of cancer cells,⁸ and it negatively regulates the group of tumor-suppressive microRNAs, Lethal-7 (let-7). Rescue experiments and downstream analysis of let-7 expression further connected AVIL with LIN28B and let-7. However, one enigma appeared. How does an F-actin regulating factor regulate transcription of LIN28B? We suspect that an intermediate factor is involved. When we carefully examined the list of genes that showed the opposite trend of expression with AVIL silencing and overexpression, we noticed that a large number of them are known targets of Forkhead Box M1 (FOXM1).⁹ Indeed, an unbiased Binding Analysis for Regulation of Transcription (BART) analysis put FOXM1 the top transcription factor upstream of AVIL targets. FOXM1 itself is not on the list of AVIL targets as no effect at its RNA level was observed with AVIL perturbation. It turned out that AVIL regulates FOXM1 protein stability. Further experimental data supports the signaling axis of AVIL-FOXM1-LIN28B-let-7 (Figure 1). How exactly does AVIL and F-actin regulate FOXM1 protein stability requires further investigation.

Figure 1.

Potential mechanisms of Advillin (AVIL)-driven gliomagenesis. Signaling axis of AVIL/Forkhead Box M1 (FOXM1)/Lin-28 Homolog B (LIN28B)/Lethal-7 (let-7) is depicted. Three downstream targets of let-7 are included: High Mobility Group AT-Hook 2 (HMGA2), Insulin-Like Growth Factor Binding Protein 1 (IGFBP1), and Insulin-Like Growth Factor Binding Protein 3 (IGFBP3). AVIL regulates F-actin and the stability of FOXM1. We assumed that regulation mainly occurs in cytoplasm. Whether the translocation of FOXM1 between cytoplasm and nucleus is regulated by AVIL not clear

Even though large sets of genomic and transcriptomic data are available to facilitate the identification of driver mutations in adult cancers including GBMs, true signals are often buried in a large number of passenger events. In contrast to adult cancers, pediatric tumors tend to have fewer point mutations and structural changes. Because gene fusions often result in the aberrant expression of a proto-oncogene, our strategy was to start from a gene fusion discovered in pediatric cancer, then extending to adult GBMs, which then led to the discovery of a bona fide oncogene to which GBMs are addicted.³ Numerous significant discoveries in cancer have stemmed from studying pediatric tumors. The “two-hit” theory and the discovery of the Retinoblastoma (RB) gene both came from studying retinoblastoma in children.

AVIL is located 45kb away from Cyclin-Dependent Kinase 4 (CDK4), and around 10MB from Mouse Double Minute 2, Human Homolog of MDM2. Given the importance of CDK4 and MDM2 in tumorigenesis, it is possible that AVIL amplification is only a byproduct of a larger fragment amplification. However, AVIL expression is upregulated in all the GBMs including the ones without copy number gain through additional mechanisms. Furthermore, both the loss- and gain-of-function studies provided direct evidence that AVIL overexpression is involved in GBM tumorigenesis rather than being solely a passenger event of a consequence of CDK4 and MDM2 locus amplification.

To evaluate the transformative ability and to probe potential collaborative effect between AVIL and other pathways, we performed the classic focus assay on NIH3T3 cells. We observed significantly larger and higher numbers of foci in cells transfected with AVIL than with an empty vector control. Impressively, AVIL overexpression alone triggers similar if not more colonies than the combination of AVIL with three other oncogenic pathway perturbations,¹⁰ Tumor Protein P53 (TP53) silencing, RB silencing, and Epidermal Growth Factor Receptor (EGFR) mutant. Furthermore, RNA-Seq analysis revealed that AVIL overexpressed astrocytes activated all three signaling pathways known to GBM tumorigenesis. These findings support AVIL being a potential nexus critical for GBM, and that targeting AVIL may be an effective approach for GBM therapy.

Funding Statement

This work was supported by the National Cancer Institute [CA240601].