Oncogenic activation of FOXR1 by 11q23 intrachromosomal deletion-fusions in neuroblastoma

Abstract

Neuroblastoma tumors frequently show loss of heterozygosity of chromosome 11q with a shortest region of overlap in the 11q23 region. These deletions are thought to cause inactivation of tumor suppressor genes leading to haploinsufficiency. Alternatively, micro-deletions could lead to gene fusion products that are tumor driving. To identify such events we analyzed a series of neuroblastomas by comparative genomic hybridization and single-nucleotide polymorphism arrays and integrated these data with Affymetrix mRNA profiling data with the bioinformatic tool R2 (http://r2.amc.nl). We identified three neuroblastoma samples with small interstitial deletions at 11q23, upstream of the forkhead-box R1 transcription factor (FOXR1). Genes at the proximal side of the deletion were fused to FOXR1, resulting in fusion transcripts of MLL–FOXR1 and PAFAH1B2–FOXR1. FOXR1 expression has only been detected in early embryogenesis. Affymetrix microarray analysis showed high FOXR1 mRNA expression exclusively in the neuroblastomas with micro-deletions and rare cases of other tumor types, including osteosarcoma cell line HOS. RNAi silencing of FOXR1 strongly inhibited proliferation of HOS cells and triggered apoptosis. Expression profiling of these cells and reporter assays suggested that FOXR1 is a negative regulator of fork-head box factor-mediated transcription. The neural crest stem cell line JoMa1 proliferates in culture conditional to activity of a MYC-ER transgene. Over-expression of the wild-type FOXR1 could functionally replace MYC and drive proliferation of JoMa1. We conclude that FOXR1 is recurrently activated in neuroblastoma by intrachromosomal deletion/fusion events, resulting in overexpression of fusion transcripts. Forkhead-box transcription factors have not been previously implicated in neuroblastoma pathogenesis. Furthermore, this is the first identification of intrachromosomal fusion genes in neuroblastoma.

Introduction

Forkhead-box transcription factors are a large evolutionarily conserved family of transcriptional regulators that share a highly conserved winged helix DNA-binding domain. Outside of this domain, forkheads have diverged into sub-families and incorporated a variety of other domains conferring on them a plethora of functions (Myatt and Lam, 2007). This divergence is especially relevant within the context of cancer where forkhead-box genes have been found operating as both oncogenes and tumor suppressors via a variety of mechanisms. Among the tumor suppressors, the FOXO family is the most intensely studied. Chromosomal translocations involving FOXO1/3/4 have been implicated in the genesis of pediatric malignancies. In alveolar rhabdomyosarcoma FOXO1 is commonly fused to the PAX3 and PAX7 (Galili et al., 1993; Davis et al., 1994). In mixed lineage leukemia FOXO3 and FOXO4 are fused to the MLL gene and these fusion products are sufficient for transformation of murine myeloid progenitors (So and Cleary, 2002, 2003). Recently, triple knockout of FOXO1/3/4 in mice has demonstrated that FOXOs are bona fide tumor suppressors (Paik et al., 2007). Additionally, FOXOs are commonly post-translationally inactivated in a wide range of cancers by the activation of the PI3K/Akt pathway, which generally results from loss of the tumor suppressor phosphatase and tensin homolog or constitutive activation of receptor tyrosine kinase signaling (Brunet et al., 1999; Yuan and Cantley, 2008). Forkheads are not always tumor suppressive as FOXM1 and FOXP1 demonstrate. FOXM1 is crucial for cell-cycle progression and mitotic spindle integrity and has been found in various cancers to act oncogenically (Schüller et al., 2007; Wierstra and Alves, 2007). FOXP1 has been identified in diffuse large B-cell lymphoma to be over-expressed as the result of intrachromosomal translocations and high expression has been linked to poor prognosis (Wlodarska et al., 2005; Hoeller et al., 2010). This raises the possibility that other forkheads may be of comparable importance in oncogenesis. Forkhead-box R1 (FOXR1, previously named FOXN5) located on 11q23.3, is another recently described member of the forkhead family (Katoh and Katoh, 2004c). On the basis of sequence alignment, FOXR1 is most similar to the FOXP sub-family (Myatt and Lam, 2007). FOXR1 expression is restricted to the early stages of embryogenesis as is its ortholog FOXR2 (FOXN6), which is located on the X chromosome (Katoh and Katoh, 2004a, 2004b; Schuff et al., 2006).

Chromosomal deletions within the 11q23 region are a common abnormality found in a wide range of cancers (Hampton et al., 1994; Lee et al., 2000; Pulido et al., 2000). In addition to deletions, translocations have also been identified. The most frequent translocations occur within leukemia's, which exhibit a multitude of MLL gene fusion products generated by a variety of chromosomal translocations (Liu et al., 2009). Neuroblastomas are pediatric tumors that arise from neuroblasts of the sympathetic nervous system. Despite intense multi-modal therapy neuroblastoma remains the second most common cause of cancer deaths among children (van Noesel and Versteeg, 2004). Previously, deletion of the 11q23 region has been described in neuroblastoma with the smallest region of overlap mapping to 11q23.3. LOH of 11q23 is associated with poor prognosis in neuroblastoma and characteristics such as advanced stage disease and unfavorable histopathology (Guo et al., 1999). Despite the clear link to prognosis in neuroblastoma tumors only two candidate tumor suppressors at 11q23 have been proposed to date. These are cell adhesion molecule 1 (TSLC1) and the microRNA mir34c (Cole et al., 2008; Nowacki et al., 2008).

Here we describe that a subset of neuroblastomas have intrachromosomal 11q23 deletions that create PAFAH1B2FOXR1 or MLLFOXR1 fusion genes. FOXR1 is silent in all normal tissues studied, but the fusion gene is highly expressed in these neuroblastoma samples. Re-expression of FOXR1 is sufficient to drive the proliferation of non-malignant mouse neural crest stem cells, whereas silencing of FOXR1 expression in the osteosarcoma cell line, HOS, results in growth inhibition and increased cell death. Analysis of Affymetrix profiling data from this knockdown experiment demonstrates significant de-repression of genes containing the canonical forkhead-box DNA-binding element (DBE) within their proximal promoter regions, such as the cell-cycle inhibitor and validated FOXO target gene p27Kip1. Suppression of DBE-driven expression by FOXR1 was confirmed in luciferase reporter assays with synthetic DBE-driven reporter constructs. These data strongly suggest that FOXR1 expression has an oncogenic role in neuroblastoma via the inhibition of the transcription of forkhead-box family target genes.

Results

Intrachromosomal deletions create FOXR1 fusion genes in neuroblastoma

In an ongoing effort to characterize genomic aberrations and corresponding gene expression patterns in neuroblastoma tumors we combine mRNA expression profiling by Affymetrix microarrays and DNA copy number analysis by comparative genomic hybridization (CGH) and single-nucleotide polymorphism array profiling for a series of 225 tumor samples. These data were analyzed using the R2 web application (http://r2.amc.nl). aCGH and single-nucleotide polymorphism profiling identified three neuroblastomas with micro-deletions in the 11q23 region. Large deletions of the q arm of chromosome 11 are frequently found in neuroblastoma and are reasoned to lead to inactivation or haploinsufficiency of tumor suppressor genes. Alternatively, intrachromosomal aberrations could lead to the formation and over-expression of fusion genes. The CGH and single-nucleotide polymorphism array profiles of tumor sample NRC0041 revealed that NRC0041 only contains an intrachromosomal deletion within one copy of 11q23 of 0.5 Mb and was otherwise free of genomic copy number alterations (Figure 1a). This micro-deletion has a distal breakpoint in front of the FOXR1 locus and a proximal breakpoint within the MLL gene, which creates a potential MLLFOXR1 fusion gene. Overlaying the Z-scores of Affymetrix expression data on the genomic map shows that the FOXR1 gene is highly expressed in this sample (Figure 1a). In the second tumor with an 11q23 deletion, NRC0129, the micro-deletion also created a potential MLL–FOXR1 fusion from a 0.5 Mb deletion with breakpoints highly similar to NRC0041 as shown by single-nucleotide polymorphism array (Supplementary Figure S1). Surprisingly, the third tumor with 11q23 deletion (NRC0051) also exhibited a FOXR1 fusion resulting from a 1.8 Mb deletion. However, the proximal breakpoint was located within the PAFAH1B2 gene (Supplementary Figure S1).

Figure 1
figure1

The MLL–FOXR1 fusion results in re-expression of FOXR1. (a) The top panel of this figure shows the aCGH overview of all chromosomes of patient NRC0041. The aberration on chromosome 6p is a known copy number variation in the control donor DNA. The second panel is a focus of the 11q23 region showing a combination of single-nucleotide polymorphism array b allele frequencies (top), aCGH log-fold copy number changes (middle) and Affymetrix mRNA Z-score data (bottom) for NRC0041. The MLL and FOXR1 genes are indicated by arrowheads. The vertical dashed lines indicate the boundaries of the 11q23 interstitial deletion. The red circle indicates the FOXR1 probe that detects the mRNA expression that results from the MLL–FOXR1 fusion. Z-scores are based on our core panel of 88 NB tumors. (b) mRNA expression of FOXR1 across a large panel of neuroblastomas (red), pediatric cancer-derived cell lines (pink) as well as other tumor types (blue) and normal tissues (green) visualized with the R2 MegaSampler program.

We hypothesized that the FOXR1 expression resulted from the formation of a fusion gene with MLL or PAFAH1B2. As a consequence the expression of FOXR1 would no longer be regulated by the FOXR1 promoter but by the promoters of MLL or PAFAH1B2. Support for this idea was found by using the MegaSampler program of R2 (http://r2.amc.nl). R2 is a web-based collection of bioinformatic applications designed for the analysis of gene expression and genomic data derived from various microarray platforms. R2 also contains an extensive database of publicly available gene expression data sets. Using the MegaSampler application of R2 it is possible to view the expression of a gene across a user-defined panel of data sets. This showed that MLL and PAFAH1B2 are relatively highly expressed in neuroblastoma cell lines and tumors relative to other cancers and normal tissues (Supplementary Figure S2). This indicated that the promoters of these genes may indeed be driving the observed FOXR1 expression. We used MegaSampler to analyze the expression of FOXR1 in 287 neuroblastomas and in a broad panel of 504 normal tissues, 1803 other cancers and a collection of 86 pediatric cancer cell lines. FOXR1 is highly expressed in only four neuroblastoma tumors (three tumors with the confirmed 11q23 micro-deletions and one additional tumor sample), one endometrial tumor and the osteosarcoma cell line HOS. For the additional neuroblastoma with high FOXR1 expression, NBAF29, no genomic DNA was available to analyze for micro-deletions. Expression of FOXR1 is essentially silent in all other samples, including those of normal tissue (Figure 1b). This is consistent with previous findings that FOXR1 expression is only detectable in the early stages of embryogenesis (Katoh and Katoh, 2004a, 2004b; Schuff et al., 2006). To further explore the extent of FOXR1 expression in neuroblastoma, we performed quantitative (qPCR) on an additional 362 neuroblastoma tumor samples and identified three more tumors exhibiting very high FOXR1 expression (Table 1). A total of 7 neuroblastomas with strong FOXR1 over-expression were, therefore, identified in 649 neuroblastomas, which in all three analyzable cases were caused by micro-deletion and potential fusion events.

Table 1 Summary of neuroblastoma tumor samples that exhibit FOXR1 expression as measured by either microarray or qPCR

To confirm the existence of FOXR1 fusion genes, cDNA was generated and PCR performed to sequence the products using forward primers within the first exons of MLL and PAFAH1B2 and reverse primers within the FOXR1 coding sequence. In all, three tumors with intrachromosomal 11q23 micro-deletions and high expression of FOXR1 we could sequence verify cDNA fusion products (Figures 2a, b and c). The MLL-FOXR1 cDNA of NRC0129 contained the complete exon 1 of MLL (422 bp) in addition to 87 bp of the FOXR1 promoter and 226 bp of the FOXR1 5′ untranslated region. The cDNA of NRC0041 also contained exon 1 of MLL (422 bp), 15 bp of intron 1 of MLL, 30 bp of the FOXR1 promoter and 226 bp of the FOXR1 5′ untranslated region. The cDNA of NRC0051 contained exons 1 and 2 of PAFAH1B2 (177 bp), 87 bp of the FOXR1 promoter and 226 bp of the FOXR1 5′ untranslated region. For all fusion genes the FOXR1 coding sequence was in-frame with the coding sequences of MLL or PAFAH1B2. Fusion cDNAs contained the full-length FOXR1 coding sequence. In addition to the fusion genes containing full-length FOXR1 we could identify splice variants missing exon 5 or exon 4 and 5 of FOXR1. Both FOXR1 splice variants have been reported in the Genbank database as expressed sequence tags. The supplemental contains the actual cDNA sequences for these fusion products.

Figure 2
figure2

Verification of the MLL–FOXR1 and PAFAH1B2–FOXR1 fusion transcripts. (a) Schematic diagrams of the sequence results of the FOXR1 fusion transcripts identified in three tumors. Arrowheads indicate locations of primers used for real-time (RT)–PCR and sequencing. (b) RT–PCR products generated from MLL–FOXR1 fusions. (c) RT–PCR product generated from PAFAH1B2–FOXR1 fusion.

FOXR1 can substitute for MYC-driven proliferation in mouse neuroblasts

It is known that most oncogenic MLL fusion genes involve large and functionally defined regions of MLL. The MLLFOXR1 fusion gene does not include such a region. The three different fusion genes all contained various parts of MLL or PAFAH1B2 while all included full-length FOXR1. This suggests that the tumor driving event is the over-expression of the FOXR1 gene product. Therefore, we sub-cloned the full-length FOXR1 open reading frame (ORF) from the NRC0129 cDNA into the pMSCVpuro expression vector for further characterization. To evaluate the oncogenic potential of FOXR1 over-expression we used mouse JoMa1 neuroblasts stably expressing a 4-hydroxytamoxifen (4-OHT) inducible ER-c-myc fusion gene (Maurer et al., 2007). Normally, cultured mouse neuroblasts undergo apoptosis or differentiate if cultured in vitro but JoMa1 cells maintain a MYC-dependent proliferative phenotype when treated with 4-OHT. Withdrawal of 4-OHT results in MYC exclusion from the nucleus and cell cycle arrest. In order to explore the role of FOXR1 in neuroblast proliferation we over-expressed either FOXR1 or green fluorescent protein in JoMa1 cells. After withdrawal of 4-OHT FOXR1 was able to fully compensate for MYC-driven proliferation in these neuroblasts (Figures 3a and b). High FOXR1 expression, therefore, maintains the proliferative capacity of primary neuroblasts. This finding strongly supports the genomic evidence and implicates aberrant FOXR1 expression in neuroblastoma oncogenesis.

Figure 3
figure3

Constitutive over-expression of FOXR1 maintains the proliferation of JoMa1 cells following withdrawal of 4-OHT. (a) Light microscopy images of JoMa1 mother line, JoMa1-FOXR1 and JoMa1-green fluorescent protein (GFP) lines with and without 4-OHT. (b) MTT assay quantification of cell viability with and without 4-OHT addition. Cells were seeded at 1000 cells per well in 96-well plates and 4-OHT was withdrawn. After 72 h, cell viability was measured by MTT assay.

Silencing of FOXR1 in osteosarcoma HOS cells results in cell cycle arrest and cell death

Because the JoMa1 results demonstrate that FOXR1 can drive proliferation we next asked if knockdown of endogenously expressed FOXR1 could inhibit it. From the previous use of MegaSampler we had identified the osteosarcoma cell line HOS as expressing FOXR1 (Figure 1c). We transduced HOS cells with five lentiviral shRNAs targeting FOXR1 from the Sigma lentiviral TRC library and a scrambled control shRNA (SHC002) at equal multiplicity of infection for all shRNAs. The knockdown of FOXR1 was confirmed by qPCR, which identified the best shRNAs as B4 and B8 (Figure 4a). As expected, FOXR1 knockdown with both shRNAs resulted in a very large reduction in growth as assessed in a 3T3 assay (Figures 4b and c). At 144 h post-transduction, HOS nuclei were isolated and analyzed by fluorescence-activated cell sorter. Silencing of FOXR1 resulted in a dramatic increase of the sub-G1 fraction, suggesting that the reduced HOS growth was at least in part because of an increase in apoptosis (Figure 4d).

Figure 4
figure4

Knockdown of FOXR1 expression by lentiviral shRNA reduces HOS cell proliferation and induces apoptosis. (a) Time-course qPCR analysis of FOXR1 knockdown in HOS cells. HOS cells were transduced with control shRNA SHC002, FOXR1 shRNA B4 or B8. We also included a non-transfected control sample. Total RNA was harvested at the indicated time points (time point 0 being untransduced cells). qPCR was performed to measure FOXR1 expression and the inverse dCT is plotted in the figure. (b) Images of control and FOXR1 shRNA-transduced HOS cells at 72 h post-transduction. (c) 3T3 assay of SHC002 and FOXR1 shRNA transduced HOS cells. Cells were transduced at an multiplicity of infection (M.O.I.) of 3 with the indicated shRNAs. After 24 h infection, 1 μg/ml puromycin was added. After 24 h,the medium was refreshed again without puromycin. Cells were then counted and re-seeded at 3-day interval over the course of 13 days. All assays were performed in triplicate with the individual results plotted in matching colors for each shRNA. (d) Fluorescence-activated cell sorter (FACS) analysis of SHC002 and FOXR1 shRNA transduced HOS cells. Cells were transduced or not with control and FOXR1 shRNAs at an M.O.I. of 3. Transduced cells were selected on puromycin for 24 h and then the medium was changed. At 72 h nuclei were harvested and stained with propidium iodide. The assay was carried out in triplicate for each shRNA. A total of 30 000 nuclei were counted per sample. All assays were performed in triplicate.

FOXR1 suppresses forkhead-box family target genes

To further delve into FOXR1 function, we performed Affymetrix profiling of the FOXR1 knockdown in HOS. We isolated a time series of RNAs at 0, 16, 24, 48 and 72 h after lentiviral-mediated silencing of FOXR1 or transduction with the SHC002 control virus. A data set was generated and analyzed using the built-in R2 time series module as well as identifying the strongly regulated genes by analysis of variance analysis. This analysis provided further confirmation of our FOXR1 knockdown by both shRNAs (Figure 5a). Additionally, we identified 180 genes upregulated and 148 genes downregulated following FOXR1 silencing (see Materials and methods for analytic method). Among the strongly activated genes were the cell-cycle inhibitors p21Cip1, p27Kip1 and the mTORC2 component Rictor (Figure 5a). Interestingly, these three genes are well-characterized FOXO family target genes normally upregulated by FOXO activity (Stahl et al., 2002; Gomis et al., 2006; Chen et al., 2010).

Figure 5
figure5

Forkhead-box transcription factors are activated following shRNA knockdown of FOXR1 in HOS cells. (a) R2 time series module plots of genes regulated by FOXR1 silencing. HOS cells were transduced at an multiplicity of infection (M.O.I.) of 3 with control SHC002 and FOXR1 shRNAs. Post-transduction total RNA was harvested at 16, 24, 48 and 72 h time points. Time point 0 was without transduction. (b) Analysis of transcription factor binding site (TFBS) content of the promoters of regulated time series genes. From the genes defined as differentially regulated in the time series experiment 167 promoters were evaluated for the 180 genes (only 1 promoter per gene, some genes having no reliable TSS information) upregulated by FOXR1 silencing. These promoters were compared to 463 randomly chosen promoters from a set of HOS cell line expressed genes (background, BG), excluding genes from among the 167 (foreground, FG) set (see Materials and methods for microarray and TFBS analysis). (c) Luciferase reporter assay after FOXR1 knockdown in HOS cells. Cells were first transduced at an M.O.I. of 3 with the control and FOXR1 shRNA B4. The following day, they were selected on puromycin for 24 h then counted and re-seeded at equal density in 24-well plates. The cells were then co-transfected in triplicate with the following combinations of vectors: (1) CMV-Renilla/pGL3-1 × DBEmut and (2) CMV-Renilla/pGL3-1 × DBEwt. After 24 h, the cells were harvested and luciferase and renilla signals quantified. For analysis, the luciferase/renilla ratio was calculated for each well and the resultant values averaged within each triplicate. For each shRNA transduction, the pGL3-1 × DBEwt reporter activity was then divided by the pGL3-1 × DBEmut reporter activity to calculate the fold change of the 1 × -DBEwt relative to the 1 × -DBEmut reporter.

This finding raised the possibility that forkhead-box transcription factors and particularly the FOXO family of transcription factors were inactivated by FOXR1 and re-activated upon FOXR1 silencing. To assess the extent of this finding, we carried out a statistical analysis to identify, which transcription factor binding motifs were enriched in the genes activated and repressed by FOXR1 knockdown. This analysis was carried out using the proximal promoters (2500 bp upstream and 500 bp downstream of the TSS) of the regulated genes and a randomly chosen set of promoters as background. We found the set of upregulated genes to be highly enriched for forkhead motifs, prevalent among these being FOXO1/FOXO and FOXJ2 motifs (Figure 5b). The genes downregulated following FOXR1 silencing showed no enrichment for forkhead motifs. This is probably because both FOXO and FOXJ2 are known transcriptional activators and generally not associated with direct transcriptional repression of their target genes (Furuyama et al., 2000; Gómez-Ferrería and Rey-Campos, 2003). These findings suggest that FOXR1 may function as a repressor of forkhead-box-mediated transcription by either directly or indirectly blocking the activity of other forkheads that are transcriptional activators.

To test the idea that forkhead-box transcription factors can be modulated by FOXR1, we generated a pair of synthetic luciferase reporter vectors each containing one copy of either a functional FOXO-binding element (1 × DBEwt–‘IndexTermTTGTTTAC’) or a mutant element (1 × DBEmut—‘IndexTermTGTTCTAT’; Furuyama et al., 2000). We then transduced HOS cells with FOXR1 shRNA B4 or SHC002 control virus and subsequently co-transfected the cells with the luciferase reporters and a renilla expression plasmid as an internal control. After 24 h transfection, lysates were harvested and luciferase/renilla signals quantified. After normalization of luciferase to renilla values, the values of the 1 × DBEwt reporter were normalized to the 1 × DBEmut reporter values. Relative to the SHC002 control virus, we observed a twofold increase in forkhead-box-mediated transcriptional activity after FOXR1 silencing (Figure 5c). These data strongly suggest that FOXR1 either directly or indirectly inhibits the transcriptional activation of forkhead-box target genes, some of which are well proven FOXO target genes. This finding implies that FOXR1 may impair the transcriptional activity of FOXO family members.

Clinical characteristics of FOXR1 over-expressing tumors

Neuroblastoma tumors with large deletions involving 11q23 are known to have a poor prognosis (Guo et al., 1999). This is thought to be functionally related to reduced expression of tumor suppressor genes that are located on the 11q deleted region and which result in increased aggressiveness of the tumor and therapy resistance. The group of 11q-deleted tumors consists of a subset of tumors with terminal 11q deletions and a subset with interstitial deletions. We now hypothesize that in a subset of the tumors with interstitial deletions of 11q, the tumor driving event is not deletion of a tumor suppressor but over-expression of a fusion gene. As a consequence the clinical behavior of these tumors could be different. Therefore, we correlated over-expression of FOXR1 to clinical characteristics. We have a series of 287 neuroblastoma tumors for which we have mRNA profiling data and a series of 352 for which we have performed FOXR1 qPCR. We selected samples with an increased expression of FOXR1 (Affymetrix: MAS 5.0 corrected present calls above 100/qPCR qbasePLUS determined CNRQ above 5). Surprisingly 6 out of 7 are low-stage tumors, and 5 are younger than 18 months. For six tumors we have follow-up information. Five patients were in complete remission for over 50 months and only one patient died from his disease (Table 1). These findings are in contrast with reports on 11q-deleted tumors, which are known to be high stage tumors and have a very poor prognosis.

Discussion

Using an integrated array CGH and mRNA expression analysis approach of neuroblastoma tumors we have discovered the oncogenic activation of a forkhead-box transcription factor, FOXR1. The over-expression of FOXR1 is achieved by 11q23.3 intrachromosomal deletion and subsequent fusion of either MLL or PAFAH1B2 first exons to the proximal promoter region of FOXR1. This shows that combining aCGH and mRNA expression data is a powerful method to identify DNA aberrations that lead to over-expression of tumor driving genes. Notably, this is the first ever fusion gene event identified in neuroblastoma. In support of the hypothesis that this is an oncogenic event in neuroblastoma we provide evidence in primary mouse neuroblasts that over-expression of FOXR1 is sufficient to maintain neuroblast proliferation after sustaining MYC activation is withdrawn. In order to further elucidate the function of FOXR1 in a tumorigenic background we identified the HOS osteosarcoma cell line as having high FOXR1 expression by R2-facilitated microarray analysis. Using lentiviral shRNAs that target FOXR1 we performed stable knockdowns in the HOS cell line. Two different shRNAs caused acute downregulation of FOXR1 as shown by both qPCR and microarray analysis. This FOXR1 knockdown resulted in an increase in the sub-G1 population as shown by fluorescence-activated cell sorter analysis as well as a dramatic decrease in proliferation as measured by 3T3 growth assays.

Subsequent Affymetrix profiling of a time series of FOXR1 knockdown combined with transcription factor binding site enrichment analysis revealed that many putative and known forkhead-box target genes were upregulated following FOXR1 knockdown. Among them were the well-known FOXO target genes p27Kip1, p21Cip1 and Rictor (Stahl et al., 2002; Gomis et al., 2006; Chen et al., 2010). This strongly indicated that FOXO transcription factors may have been activated following FOXR1 knockdown. This suggests the potential for FOXR1 to either directly or indirectly blocks FOXO target gene expression. To test this possibility we cloned a FOXO reporter system containing one copy of the canonical forkhead-box/FOXO target sequence DBE or a mutated element. The reporter assays revealed that FOXR1 knockdown was indeed capable of activating forkhead-box/FOXO transcriptional activity in HOS cells. Given the identity of the target genes involved FOXO appears to be the most likely sub-family of forkhead-box transcription factors that were activated although other members such as FOXJ2 cannot be definitively excluded. Given the similarity of FOXR1 to the FOXP sub-family by sequence alignment it seems reasonable that FOXR1 may also be a transcriptional repressor such as FOXP (Myatt and Lam, 2007). This suggests that FOXR1 may have the ability to directly repress genes usually activated by other forkhead-box transcription factors. Within the context of these aberrant over-expressions in neuroblastoma and the osteosarcoma cell line HOS this repression appears to be an oncogenic event.

11q23 deletion has long been considered a prognostic and tumor-driving event in neuroblastoma (Guo et al., 1999; Buckley et al., 2010). Usually interstitial deletions lead to the loss of tumor suppressors, which potentiate oncogenesis. Here we describe the re-expression of a forkhead-box oncogene capable of inhibiting the activity of tumor suppressive forkhead-box transcription factors such as FOXO in a subset of 11q23-deleted tumors. Notably, these tumors with 11q23 micro-deletions do not contain any other 11q aberrations. This observation classifies a small subgroup of neuroblastoma with intrachromosomal 11q deletions that lead to activation of an oncogene, which is distinct from the majority of neuroblastoma where the 11q deletions most likely lead to inactivation of a tumor suppressor or several tumor suppressors. This is of importance when determining the smallest region of overlap of 11q deletions in search of important tumor suppressors. In addition, analysis of clinical characteristics shows that these patients have a remarkable good prognosis compared with other neuroblastoma patients with 11q deletions.

This is the first report on the involvement of forkhead-box transcription factors in neuroblastoma tumors. which highlights the central importance of forkhead-box-mediated transcriptional activity in the determination of cell fate and regulation of tumorigenesis. It is also the first to describe a bona fide fusion gene in neuroblastoma. These mechanisms may be especially relevant in pediatric cancers given the various mutations identified to date involving this class of transcription factors, especially within the FOXO family, which is replete with gene fusion events. Therefore, further elucidation of the mechanisms underlying FOXR1-mediated tumorigenesis is warranted and may uncover additional nodes of importance in the forkhead-box transcriptional networks.

Materials and methods

Patient material and cell lines

All samples were derived from primary tumors of untreated patients. Material was obtained during surgery and immediately frozen in liquid nitrogen. The mRNA expression and DNA copy number analysis were performed using several different platforms on samples derived through various collaborations as given in the Supplementary Materials and Methods Table 1. The Affymetrix expression data from adult tumors and normal tissues were derived from the Expression Project for Oncology database from the International Genomics Consortium (http://www.intgen.org/expo/). The HOS cell line was a kind gift from Prof M Serra. The cell line was cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 20 mM L-glutamine, 10 U/ml penicillin and 10 μg/ml streptomycin. Cells were maintained at 37 °C under 5% CO2.

PCR analysis fusion genes

cDNA from the tumors and cell line samples was isolated using Trizol (Invitrogen, Carlsbad, CA, USA). Real-time–PCR was performed using 1 μg total RNA, 125pM oligodT12 primers, 0.5 mM deoxyribonucleotide triphosphates, 2 mM MgCl2, RT-buffer (Invitrogen) and 100U Superscript II (Invitrogen) in a total volume of 25 μl. cDNA of 1 μl volume was used for PCR analysis of the breakpoints. The primers used for PCR amplification and sequencing of the MLL-FOXR1 breakpoints were 5′-IndexTermCTCGTCTTCGTCTTCGTCATC-3′ (fwd) and 5′-IndexTermCTTTCCAGGGGGATACACAAT-3′ (rev). The PAFAH1B2-FOXR1 breakpoint was amplified with 5′-IndexTermCGCACGGACCCTCTACTTC-3′ (fwd) and 5′-IndexTermTAAGGGGAGGTGAGATGTGG-3′ (rev). For exons 1–5 of FOXR1 from NRC0051 5′-IndexTermAGCTCCAACACCTCGACTTCT-3′ (fwd) and 5′-IndexTermGGCACTTTCTCAAAGCTGTCTC-3′ (rev) were used. To capture the full-length ORF of the MLL fusion from NRC0129 cDNA 5′-IndexTermCTTCACGGGGCGAACATG-3′ (fwd) and 5′-IndexTermGCATTAGCTGGCTGTTGCTTC-3′ (rev) were used.

Expression and luciferase reporter constructs

For over-expression studies, the full-length FOXR1 ORF was sub-cloned into a Clonase (Invitrogen) compatible pENTR/D-TOPO vector (Invitrogen) modified to contain the MCS 5′-IndexTermAAGCTTGGTACCGAGCTCGGATCCACTAGTCCAGTGTGGTGGAATTCTGCAGATATCCAGCACAGTGGCGGCCGCTCGAGTCTAGA-3′ (creating pENTR-MCS). After FOXR1 ORF amplification with the following primers: 5′-IndexTermTATATAAAGCTTGCTCCAACACCTCGACTTCT-3′ (fwd) and 5′-IndexTermTATATAGAATTCGCATTAGCTGGCTGTTGCTTC-3′ (rev) from NRC0129 cDNA, the product was digested with HindIII (5′ end) and EcoRI (3′ end) and ligated into similarly digested pENTR-MCS. Using the LR Clonase II enzyme (Invitrogen), the FOXR1 ORF was then introduced into the vector pMSCVpuro (Clonetech, Mountain view, CA, USA) by recombination. Previously, a gateway cassette was introduced into the pMSCVpuro vector using the Gateway Conversion Kit (Invitrogen), to make it compatible with the Invitrogen Gateway system and Clonase-mediated recombination.

JoMa1 cell experiments

JoMa1 cells were cultured in a modified neural crest culture medium (NCC medium; Maurer et al., 2007). Dulbecco's modified Eagle’s medium (4,5 mg/ml glucose, L-glutamine, Pyruvate) and Ham’s F12 medium were mixed 1:1 and supplemented with 1% N2-Supplement (Invitrogen), 2% B27-Supplement (Invitrogen), 10 ng/ml epidermal growth factor (PromoKine, Heidelberg, Germany), 1 ng/ml FGF (Millipore, Billerica, MA, USA), 100 U/ml Penicillin–Streptomycin (Invitrogen) and 10% chick embryo extract (Pajtler et al., 2010). To ensure nuclear localization of c-MycERT and concomitant proliferation in JoMa1 cells, 200 nM 4-OHT (Sigma, Deisenhofen, Germany) was added to NCC medium. JoMa1 cells were transfected by electroporation and subsequent puromycin selection. Cells were seeded at a density of 1000 cells per well for MTT cultured in standard NCC medium. MTT assays were performed as described previously (Schulte et al., 2009). Cell viability was assessed 72 h after withdrawal of 4-OHT.

Lentiviral shRNA silencing

The lentiviral shRNA expression vectors were obtained from Sigma (MISSION shRNA Lentiviral library). Lentiviral particles were produced in HEK293T cells by co-transfection of lentiviral vector containing the short hairpin RNA (shRNA) with lentiviral packaging plasmids pMD2G, pRRE and pRSV/REV using FuGene HD. Supernatant of the HEK293T cells was harvested at 48 and 72 h after transfection, which was purified by filtration and ultracentrifugation. The concentration was determined by a p24 enzyme-linked immunosorbent assay. HOS cells were counted and 12 500 cells were plated in 6-well plates in 2 ml of culture medium. The culture medium was changed after 16 h and the cells were infected with either the shFOXR1-lentivirus (Sigma TRC FOXR1B4 and FOXR1B8) or the control lentivirus (non-target shRNA control SHC002). An multiplicity of infection of 3 was used for the experiments. Medium was refreshed 24 hours post-infection.

Fluorescence-activated cell sorter analysis and 3T3 assays

For fluorescence-activated cell sorter analysis HOS cells were grown for 24 h in 6-well plates and transduced with shRNA as described above. After 24 h transduction, the medium was refreshed. At 144 h after infection nuclei were isolated and stained with propidium iodide by incubation in a hypotonic solution of phosphate buffered saline, dH2O, propidium iodide and RNAse. A total of 10 000 nuclei per sample were counted. For 3T3 assays all cell counting experiments were performed in triplo using a Beckman Coulter counter. HOS cells were seeded at a density of 50 000 cells per 6 cm dish. After 24 h, cells were transduced with the shRNAs as above. The next day 1 μg/ml puromycin was added with medium refresh. At time points 3, 6, 10 and 13 days cells were counted and re-seeded.

qPCR analysis

1 μg of Trizol isolated RNA was used for cDNA synthesis as described above. A volume of 1 μl of this cDNA was used for QPCR. A fluorescence-based kinetic real-time PCR was performed using the real-time iCycler PCR platform (Biorad, Hercules, CA, USA) in combination with the intercalating fluorescent dye SYBR Green I. The IQ SYBR Green I Supermix (BioRad) was used in accordance with the manufacturer's instructions. The primers used for the qPCR to measure FOXR1 knockdown were as follows: 5′-IndexTermCCACATCTCACCTCCCCTTA-3′ (fwd) and 5′-IndexTermCTTTCCAGGGGGATACACAA-3′ (rev). The control primers to β-actin were as follows 5′-IndexTermACATCTGCTGGAAGGTGGAC-3′ (fwd) and 5′-IndexTermGCAAAGACCTGTACGCCAAC-3′ (rev).

FOXR1 gene expression in 362 neuroblastoma tumor samples was performed according to a procedure described elsewhere (Vermeulen et al., 2009). In brief, a qPCR assay was designed for FOXR1 and five reference sequences (HMBS, HPRT1, SDHA, UBC and Alu) and validated using our in silico analysis pipeline (Lefever et al., 2009). Real-time qPCR was performed in a 384-well plate instrument (LC480, Roche, Germany) and data analyzed using qbasePLUS 1.4 (http://www.qbaseplus.com; Hellemans et al., 2007).

Luciferase reporter assays

Single-copy luciferase reporter constructs were generated by cloning the forkhead-box DBEwt ‘IndexTermTTGTTTAC’ or a mutated version (DBEmut) ‘IndexTermTGTTCTAT’ into the pGL3-SV40 vector (Promega, Madison, WI, USA) double digested with SmaI (5′ end) and BglII (3′ end) to create pGL3-1xDBEwt and pGL3-1 × DBEmut. The DBEwt oligos were as follows: 5′-IndexTermCTGGATTTGTTTACGTCCCGGGTGCA-3′ (fwd) and 5′-IndexTermGATCTGCACCCGGGACGTAAACAAATCCAG-3′ (rev). The DBEmut oligos were as follows: 5′-IndexTermCTGGATTGTTCTATGTCCCGGGTGCA-3′ (fwd) and 5′-IndexTermGATCTGCACCCGGGACATAGAACAATCCAG-3′ (rev). The reporter assays were performed using the Dual-Luciferase Reporter Assay System (Promega cat. #E1980). Lysates were harvested in 100 μl 1 × passive lysis buffer from 24-well plates. A volume of 50 μl of each lysate was loaded into 96-well opaque plates for injection with 50 μl of each substrate; signal detection was performed on a Synergy HT Multi-Mode Microplate Reader (Biotek, Winooski, VT, USA).

References

  1. Brunet A, Bonni A, Zigmond MJ, Lin MZ, Juo P, Hu LS et al. (1999). Akt promotes cell survival by phosphorylating and inhibiting a Forkhead transcription factor. Cell 96: 857–868.

  2. Buckley PG, Alcock L, Bryan K, Bray I, Schulte JH, Schramm A et al. (2010). Chromosomal and microRNA expression patterns reveal biologically distinct subgroups of 11q- neuroblastoma. Clin Cancer Res 16: 2971–2978.

  3. Chen CC, Jeon SM, Bhaskar PT, Nogueira V, Sundararajan D, Tonic I et al. (2010). FoxOs inhibit mTORC1 and activate Akt by inducing the expression of Sestrin3 and Rictor. Dev Cell 18: 592–604.

  4. Cole KA, Attiyeh EF, Mosse YP, Laquaglia MJ, Diskin SJ, Brodeur GM et al. (2008). A functional screen identifies miR-34a as a candidate neuroblastoma tumor suppressor gene. Mol Cancer Res 6: 735–742.

  5. Davis RJ, D'Cruz CM, Lovell MA, Biegel JA, Barr FG . (1994). Fusion of PAX7 to FKHR by the variant t(1;13)(p36;q14) translocation in alveolar rhabdomyosarcoma. Cancer Res 54: 2869–2872.

  6. Furuyama T, Nakazawa T, Nakano I, Mori N . (2000). Identification of the differential distribution patterns of mRNAs and consensus binding sequences for mouse DAF-16 homologues. Biochem J 349: 629–634.

  7. Galili N, Davis RJ, Fredericks WJ, Mukhopadhyay S, Rauscher III FJ, Emanuel BS et al. (1993). Fusion of a fork head domain gene to PAX3 in the solid tumour alveolar rhabdomyosarcoma. Nat Genet 5: 230–235.

  8. Gómez-Ferrería MA, Rey-Campos J . (2003). Functional domains of FOXJ2. J Mol Biol 329: 631–644.

  9. Gomis RR, Alarcon C, He W, Wang Q, Seoane J, Lash A et al. (2006). A FoxO-Smad synexpression group in human keratinocytes. Proc Natl Acad Sci USA 103: 12747–12752.

  10. Guo C, White PS, Weiss MJ, Hogarty MD, Thompson PM, Stram DO et al. (1999). Allelic deletion at 11q23 is common in MYCN single copy neuroblastomas. Oncogene 18: 4948–4957.

  11. Hampton GM, Mannermaa A, Winqvist R, Alavaikko M, Blanco G, Taskinen PJ et al. (1994). Loss of heterozygosity in sporadic human breast carcinoma: a common region between 11q22 and 11q23.3. Cancer Res 54: 4586–4589.

  12. Hellemans J, Mortier G, De PA, Speleman F, Vandesompele J . (2007). qBase relative quantification framework and software for management and automated analysis of real-time quantitative PCR data. Genome Biol 8: R19.

  13. Hoeller S, Schneider A, Haralambieva E, Dirnhofer S, Tzankov A . (2010). FOXP1 protein overexpression is associated with inferior outcome in nodal diffuse large B-cell lymphomas with non-germinal centre phenotype, independent of gains and structural aberrations at 3p14.1. Histopathology 57: 73–80.

  14. Katoh M, Katoh M . (2004a). Germ-line mutation of Foxn5 gene in mouse lineage. Int J Mol Med 14: 463–467.

  15. Katoh M, Katoh M . (2004b). Identification and characterization of human FOXN6, mouse Foxn6, and rat Foxn6 genes in silico. Int J Oncol 25: 219–223.

  16. Katoh M, Katoh M . (2004c). Identification and characterization of human FOXN5 and rat Foxn5 genes in silico. Int J Oncol 24: 1339–1344.

  17. Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM et al. (2002). The human genome browser at UCSC. Genome Res 12: 996–1006.

  18. Lee AS, Seo YC, Chang A, Tohari S, Eu KW, Seow-Choen F et al. (2000). Detailed deletion mapping at chromosome 11q23 in colorectal carcinoma. Br J Cancer 83: 750–755.

  19. Lefever S, Vandesompele J, Speleman F, Pattyn F . (2009). RTPrimerDB: the portal for real-time PCR primers and probes. Nucleic Acids Res 37: D942–D945.

  20. Liu H, Cheng EH, Hsieh JJ . (2009). MLL fusions: pathways to leukemia. Cancer Biol Ther 8: 1204–1211.

  21. Mahony S, Benos PV . (2007). STAMP: a web tool for exploring DNA-binding motif similarities. Nucleic Acids Res 35: W253–W258.

  22. Martinez MJ, Smith AD, Li B, Zhang MQ, Harrod KS . (2007). Computational prediction of novel components of lung transcriptional networks. Bioinformatics 23: 21–29.

  23. Matys V, Kel-Margoulis OV, Fricke E, Liebich I, Land S, Barre-Dirrie A et al. (2006). TRANSFAC and its module TRANSCompel: transcriptional gene regulation in eukaryotes. Nucleic Acids Res 34: D108–D110.

  24. Maurer J, Fuchs S, Jager R, Kurz B, Sommer L, Schorle H . (2007). Establishment and controlled differentiation of neural crest stem cell lines using conditional transgenesis. Differentiation 75: 580–591.

  25. Myatt SS, Lam EW . (2007). The emerging roles of forkhead box (Fox) proteins in cancer. Nat Rev Cancer 7: 847–859.

  26. Nowacki S, Skowron M, Oberthuer A, Fagin A, Voth H, Brors B et al. (2008). Expression of the tumour suppressor gene CADM1 is associated with favourable outcome and inhibits cell survival in neuroblastoma. Oncogene 27: 3329–3338.

  27. Paik JH, Kollipara R, Chu G, Ji H, Xiao Y, Ding Z et al. (2007). FoxOs are lineage-restricted redundant tumor suppressors and regulate endothelial cell homeostasis. Cell 128: 309–323.

  28. Pulido HA, Fakruddin MJ, Chatterjee A, Esplin ED, Beleno N, Martinez G et al. (2000). Identification of a 6-cM minimal deletion at 11q23.1-23.2 and exclusion of PPP2R1B gene as a deletion target in cervical cancer. Cancer Res 60: 6677–6682.

  29. Schones DE, Smith AD, Zhang MQ . (2007). Statistical significance of cis-regulatory modules. BMC Bioinformatics 8: 19.

  30. Schuff M, Rossner A, Donow C, Knochel W . (2006). Temporal and spatial expression patterns of FoxN genes in Xenopus laevis embryos. Int J Dev Biol 50: 429–434.

  31. Schüller U, Zhao Q, Godinho SA, Heine VM, Medema RH, Pellman D et al. (2007). Forkhead transcription factor FoxM1 regulates mitotic entry and prevents spindle defects in cerebellar granule neuron precursors. Mol Cell Biol 27: 8259–8270.

  32. Schulte JH, Lim S, Schramm A, Friedrichs N, Koster J, Versteeg R et al. (2009). Lysine-specific demethylase 1 is strongly expressed in poorly differentiated neuroblastoma: implications for therapy. Cancer Res 69: 2065–2071.

  33. So CW, Cleary ML . (2002). MLL-AFX requires the transcriptional effector domains of AFX to transform myeloid progenitors and transdominantly interfere with forkhead protein function. Mol Cell Biol 22: 6542–6552.

  34. So CW, Cleary ML . (2003). Common mechanism for oncogenic activation of MLL by forkhead family proteins. Blood 101: 633–639.

  35. Stahl M, Dijkers PF, Kops GJ, Lens SM, Coffer PJ, Burgering BM et al. (2002). The forkhead transcription factor FoxO regulates transcription of p27Kip1 and Bim in response to IL-2. J Immunol 168: 5024–5031.

  36. van Noesel MM, Versteeg R . (2004). Pediatric neuroblastomas: genetic and epigenetic ′danse macabre′. Gene 325: 1–15.

  37. Vermeulen J, De PK, Naranjo A, Vercruysse L, Van RN, Hellemans J et al. (2009). Predicting outcomes for children with neuroblastoma using a multigene-expression signature: a retrospective SIOPEN/COG/GPOH study. Lancet Oncol 10: 663–671.

  38. Wakaguri H, Yamashita R, Suzuki Y, Sugano S, Nakai K . (2008). DBTSS: database of transcription start sites, progress report 2008. Nucleic Acids Res 36: D97–101.

  39. Wierstra I, Alves J . (2007). FOXM1, a typical proliferation-associated transcription factor. Biol Chem 388: 1257–1274.

  40. Wlodarska I, Veyt E, De PP, Vandenberghe P, Nooijen P, Theate I et al. (2005). FOXP1, a gene highly expressed in a subset of diffuse large B-cell lymphoma, is recurrently targeted by genomic aberrations. Leukemia 19: 1299–1305.

  41. Yuan TL, Cantley LC . (2008). PI3K pathway alterations in cancer: variations on a theme. Oncogene 27: 5497–5510.

Download references

Acknowledgements

This work was supported by the Stichting Koningin Wilhelmina Fonds (KWF), Stichting Kindergeneeskundig Kankeronderzoek (SKK) and the Stichting Kinderen Kankervrij (KiKa).

Author information

Correspondence to J J Molenaar.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Additional information

Supplementary Information accompanies the paper on the Oncogene website

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Santo, E., Ebus, M., Koster, J. et al. Oncogenic activation of FOXR1 by 11q23 intrachromosomal deletion-fusions in neuroblastoma. Oncogene 31, 1571–1581 (2012). https://doi.org/10.1038/onc.2011.344

Download citation

Keywords

  • FOXR1
  • FOXO
  • forkhead-box
  • neuroblastoma
  • MLL
  • 11q23

Further reading