Cross-species array comparative genomic hybridization identifies novel oncogenic events in zebrafish and human embryonal rhabdomyosarcoma

Abstract

Human cancer genomes are highly complex, making it challenging to identify specific drivers of cancer growth, progression, and tumor maintenance. To bypass this obstacle, we have applied array comparative genomic hybridization (array CGH) to zebrafish embryonal rhabdomyosaroma (ERMS) and utilized cross-species comparison to rapidly identify genomic copy number aberrations and novel candidate oncogenes in human disease. Zebrafish ERMS contain small, focal regions of low-copy amplification. These same regions were commonly amplified in human disease. For example, 16 of 19 chromosomal gains identified in zebrafish ERMS also exhibited focal, low-copy gains in human disease. Genes found in amplified genomic regions were assessed for functional roles in promoting continued tumor growth in human and zebrafish ERMS--identifying critical genes associated with tumor maintenance. Knockdown studies identified important roles for Cyclin D2 (CCND2), Homeobox Protein C6 (HOXC6) and PlexinA1 (PLXNA1) in human ERMS cell proliferation. PLXNA1 knockdown also enhanced differentiation, reduced migration, and altered anchorage-independent growth. By contrast, chemical inhibition of vascular endothelial growth factor (VEGF) signaling reduced angiogenesis and tumor size in ERMS-bearing zebrafish. Importantly, VEGFA expression correlated with poor clinical outcome in patients with ERMS, implicating inhibitors of the VEGF pathway as a promising therapy for improving patient survival. Our results demonstrate the utility of array CGH and cross-species comparisons to identify candidate oncogenes essential for the pathogenesis of human cancer.

Figure 1. Array CGH reveals cancer-specific chromosomal abnormalities in zebrafish ERMS.

(A) Summary of common gene-containing CNA gains (green) and losses (red) in 20 animals examined. Only recurrent CNAs found in ≥3 samples are shown. The height of each bar correlates with the frequency of each aberration. Detailed view of regional gains for vegfa on chromosome 4 (B), ccnd2a on chromosome 25 (C), hoxc6a on chromosome 23 (D), and plxna1 on chromosome 6 (E). Y-axis denotes log2 ratio of the probes and X-axis denotes genomic coordinates.

Figure 2. CCND2, HOXC6 and PLXNA1 exert important roles in human ERMS cell proliferation.

(A) Western analysis following siRNA knockdown in the RD cell line. The percentage of knockdown (% KD) is indicated below each lane (representative example shown for three independent replicates). (B) Viability following siRNA gene knockdown in RD cells was assessed by cell-titer glo assay. (C) Graph summarizing the results of the cell-titer glo assay. OD-fold change over 3 days in RD cell line from knockdown by smart-pool and two individual gene-specific siRNAs is indicated. (D) Summary of EDU proliferation analysis from RD and SMS-CTR ERMS cell lines,. (Asterisks denote significant differences between siRNA knockdown compared to control siRNA (p<0.05). Each error bar in B, C, and D denotes standard deviation of 3 independent experiments.

Figure 3. Knockdown of PLXNA1 induced differentiation and impaired anchorage-independent growth of human ERMS cells.

RD cells stained with myosin heavy chain (MF20) and DAPI following culture under differentiation conditions for 72 hrs. (A) Control siRNA. (B) PLXNA1 smart-pool siRNA. (C) Control scrambled shRNA. (D) PLXNA1 shRNA-1. DAPI, blue; MF20-positive cell, green. (E) Quantification of MF-20 immunofluorescence in siRNA and shRNA-knockdown RD cells. Asterisk indicates significant differences between gene knock- down and control cells (p<0.05). Error bars denote standard deviation. (F) Western analysis of PLXNA1 shRNA stable knockdown; sc, scrambled control shRNA; 1, PLXNA1 shRNA-1; 2, PLXNA1 shRNA-2. A soft agar colony formation assay to assess PLXNA1 knockdown effects on anchorage-independent growth (G–I). (G) Control scrambled shRNA. (H) PLXNA1 shRNA. (I) Quantification of colony formation assay results. Error bar indicates standard deviation from triplicate experiments.

Figure 4. Knockdown of PLXNA1 impairs migration of human ERMS cells.

Representative images of ERMS cells transfected with gene-specific siRNAs at 0 hr (A, control siRNA; C, CCND2 siRNA; E, HOXC6 siRNA; G, PLXNA1 siRNA) and 22 hrs (B, control siRNA; D, CCND2 siRNA; F, HOXC6 siRNA; H, PLXNA1 siRNA) following gap creation. Scale bar indicates 100 µm. (I) Quantification of data from wound healing assay. Each error bar indicates standard deviation across 5–6 independent replicates. (J) A Transwell migration assay was performed in RD cells that stably express either a control shRNA or two independent PLXNA1 shRNAs. Migration was assessed after 24 hours. Each error bar indicates standard deviation across six fields at 200× magnification. Asterisks denote p<0. 05.

Figure 5. Chemical inhibition of VEGF signaling by cediranib reduces ERMS growth in vivo.

Syngeneic CG1 fish were transplanted with ERMS cells that co-expressed rag2-KRASG12D and rag2-dsRED. Fish with engrafted tumors were treated with DMSO vehicle (A–F) or 100 nM of cediranib for 7 days (G–L). Pre-treatment (A–C and G–I) and post-treatment images (D–F and J–L) of representative fish. Bright field (A,D,G,J), dsRED fluorescence (B,E,H,K) and merged image planes (C,F,I,L). Scale bar is 3 mm. (M) Quantification of relative volume change for individual animals. (N–O) fli1-GFP transgenic zebrafish were transplanted with dsRED-labeled ERMS and treated with DMSO (N) and cediranib (O). Scale bar equals 50 µm. (P) Microvessel density quantification. Asterisk indicates statistically significant difference between DMSO and cediranib-treated groups based on student t-test. Each error bar indicates standard deviation from 3 fields of microvessels for each animal. EDU incorporation analysis in DMSO (Q) or cediranib (R) treated fish. Scale bar is 50 µm. (S) Quantification of EDU analysis across each cohort of animals. Each error bar indicates standard deviation of percent EDU+ cells found within 3 fields for each animal.

Figure 6. Genes contained within low copy CNAs are expressed in primary human rhabdomyosaroma but not normal fetal muscle.

Immunohistochemistry of human primary RMS and fetal muscle tissue samples. Hematoxylin and Eosin stained sections (A–C). Expression of HOXC6, CCND2, PLXNA1 and VEGFA in embryonal rhabdomyosaroma (D, G, J, M), alveolar rhabdomyosaroma (E, H, K, N) and fetal muscle (F, I, L, O). Magnified views of staining are shown in insets. In fetal muscle, PLXNA1 and VEGFA are only expressed in the vasculature (examples indicated by arrowheads in L and O and corresponding insets). The cumulative frequency of positive staining within tumor subtype is shown in the top right corner of each panel (pediatric and adult samples combined). Scale bar (panel A) = 50 µm.

Figure 7. Elevated VEGFA expression correlates with poor clinical outcome in rhabdomyosarcoma.

Kaplan-Meier analysis was completed using microarray gene expression data for which patient outcome is available. Comparison was made in all RMS patients, ERMS patients only, and translocation-positive ARMS patients only. (A) HOXC6. (B) VEGFA.