Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

An integrative genomics screen uncovers ncRNA T-UCR functions in neuroblastoma tumours


Different classes of non-coding RNAs, including microRNAs, have recently been implicated in the process of tumourigenesis. In this study, we examined the expression and putative functions of a novel class of non-coding RNAs known as transcribed ultraconserved regions (T-UCRs) in neuroblastoma. Genome-wide expression profiling revealed correlations between specific T-UCR expression levels and important clinicogenetic parameters such as MYCN amplification status. A functional genomics approach based on the integration of multi-level transcriptome data was adapted to gain insights into T-UCR functions. Assignments of T-UCRs to cellular processes such as TP53 response, differentiation and proliferation were verified using various cellular model systems. For the first time, our results define a T-UCR expression landscape in neuroblastoma and suggest widespread T-UCR involvement in diverse cellular processes that are deregulated in the process of tumourigenesis.


Tumourigenesis is driven by (epi-)genetic alterations that result in gene expression changes. Besides protein-coding genes, these alterations also affect various classes of non-coding RNAs such as microRNAs (miRNAs) and long intergenic non-coding RNAs (lincRNAs). miRNAs function as negative regulators of gene expression through imperfect binding to target mRNAs, whereas lincRNAs associate with chromatin-modifying complexes to alter gene expression (Bartel, 2009; Guttman et al., 2009; Khalil et al., 2009). Both miRNAs and lincRNAs have been implicated in a number of oncogenic and tumour-suppressor pathways (O'Donnell et al., 2005; He et al., 2007; Guttman et al., 2009), whereby miRNAs have been established as key components in tumour biology (Esquela-Kerscher and Slack, 2006).

Recently, another class of non-coding RNAs called transcribed ultraconserved regions (T-UCRs) has been associated with the process of tumourigenesis (Calin et al., 2007). T-UCRs are transcribed from 481 ultraconserved regions (UCRs) defined as being at least 200 bp in length and 100% conserved between the human, mouse and rat genomes (Bejerano et al., 2004). Genomically, UCRs are located both within genes and in regions lacking apparent protein-coding features (Bejerano et al., 2004; Calin et al., 2007). The original UCR annotation by Bejerano et al., focusing on overlap with protein-coding genomic regions, annotated 111 UCRs as exonic, 256 as non-exonic, whereas the remaining 114 UCRs, for which the evidence of overlap with a protein-coding sequence was inconclusive, were termed possibly exonic (Bejerano et al., 2004). As with miRNAs and lincRNAs, the high degree of conservation implies that UCRs are of functional importance in mammalian cell biology. Indeed, one of these UCRs is contained in an enhancer upstream of the DACH1 gene (Nobrega et al., 2003; Bejerano et al., 2004), which has a role in development, and non-exonic UCRs that lie intronically are often associated with developmental genes (Bejerano et al., 2004). UCRs are extensively transcribed and their expression patterns have been shown to discriminate between normal and cancer tissues (Calin et al., 2007). Furthermore, UCRs are frequently located at fragile sites and genomic regions involved in cancers (Calin et al., 2007). However, the functional relevance of the vast majority of UCRs remains elusive.

In this study, we aimed at examining the possible functions of T-UCRs using neuroblastoma as a tumour model. Neuroblastoma is an aggressive childhood tumour accounting for 15% of all pediatric cancer deaths. Neuroblastoma patients present with a highly variable clinical course and aggressive tumours are characterized by a combination of various genetic abnormalities, including 1p or 11q deletions, 17q gain and MYCN amplification (Maris et al., 2007). We profiled all 481 T-UCRs on a large and well-characterized panel of representative neuroblastoma tumours and analysed their expression with respect to genomic location and co-regulation with host and surrounding genes. We show that T-UCR expression is related to important clinical and genetic parameters in neuroblastoma and provide evidence that T-UCRs have prognostic value in neuroblastoma. Using a functional genomics approach, based on the integration of T-UCR and mRNA expression data, we assigned putative functions to each T-UCR and validated these for a subset of T-UCRs using cellular model systems. Our results reveal for the first time a deregulated T-UCR expression landscape in neuroblastoma and implicate T-UCRs in a wide range of cellular functions.


Re-annotation and selection of independently expressed T-UCRs

For T-UCR quantification, we designed 481 primer sets for reverse transcription quantitative PCR (RT–qPCR). All primer pairs were validated in silico and a representative selection of primers was tested experimentally for their specificity (n=384) and efficiency (n=481). Amplicon sizing indicated good concordance between theoretical and experimental amplicon lengths (r2=0.961) (Supplementary Figure 1A), and single-curve efficiency estimates, as determined by PCR miner (Zhao and Fernald, 2005), show that 84% of the tested primers have efficiencies between 90 and 110 and 97% have efficiencies above 80% (Supplementary Figure 1B).

A significant number of UCRs is located exonically within coding (host) genes. For a number of these, the measured expression value could be confounded by the expression of the host gene. For further analyses, we therefore decided to select T-UCRs whose expression profile is independent from that of the host gene. For this correlation analysis, accurate information on the genomic location of each T-UCR relative to known protein-coding genes is pivotal. As the currently available T-UCR annotation is based on the old hg17 genome assembly and as only three T-UCR categories were used (exonic, non-exonic and possibly exonic) (Bejerano et al., 2004) (, we re-annotated all T-UCR sequences using the more recent genome build hg18 and re-organized them into five different categories by matching their location to that of the human RefSeq genes (Pruitt et al., 2009). The new genomic categories (intergenic, intronic, exonic, partly exonic, exon containing) are unambiguously defined and provide a more detailed genomic annotation for each T-UCR (Figure 1, Supplementary Table 1, Supplementary File 1). We categorized 38.7% of T-UCRs as intergenic, that is, located between genes rather than within a coding segment, 42.6% as intronic, 4.2% as exonic, 5% as partly exonic and 5.6% as exon containing (Figure 2a). For a few T-UCRs (3.9%), the genomic annotation varied because of host gene splice variants. These T-UCRs were categorized as ‘multiple’ (Figure 2a). To define which of the intragenic T-UCRs are expressed independently of their host and flanking protein-coding genes, we quantified T-UCR (RT–qPCR) and mRNA (exon array) expression levels in a well-characterized cohort of neuroblastoma tumour samples (n=49 for T-UCRs and n=40 for mRNA). Next, we calculated correlations between the expression of each intragenic T-UCR and that of its host gene in the tumour cohort. For intragenic T-UCRs with a significant positive correlation to the host gene (Spearman's rank, P<0.05), we concluded that the T-UCR RT–qPCR assay measured the expression of the host gene (or that of the host gene and the T-UCR) rather than that of the T-UCR alone (Figure 2b). Not surprisingly, the largest fraction of positive correlations was found for exonic T-UCRs (95%), followed by the exon-containing (66%) and partly exonic T-UCRs (66.7%) (Figure 2a). We found approximately half of all T-UCRs (237 T-UCRs, both inter- and intragenic) to be expressed independently and we therefore selected these for further analysis. None of the intragenic T-UCRs showed a negative correlation to the host gene, whereas 17 T-UCRs (both intra- and intergenic) were negatively correlated to one of the flanking up- or downstream-coding genes (the first flanking up- and downstream gene on both strands). Expression levels of the different categories of independent T-UCRs were highly variable with intergenic T-UCRs showing significantly lower expression compared with intragenic T-UCRs (Mann–Whitney test, P<0.0001) (Figure 2c).

Figure 1

Re-annotation of UCRs. Representation of the different UCR classes according to their genomic location with respect to protein-coding genes defined by Refseq. An example of each class is shown.

Figure 2

Features of T-UCR expression in neuroblastoma. (a) Organization of T-UCRs in five categories according to their position with respect to all known human Refseq genes. Distributions for all T-UCRs (black) and for the T-UCRs expressed independently of their host/flanking gene (grey) are shown. (b) Representative positive correlation between the expression of a T-UCR (uc.75) and its host gene (ZEB2) (n=40). (c) Expression distribution of all independently expressed T-UCRs (n=237) in each of the five different categories (whiskers: Tukey). Number of T-UCRs per category is indicated between brackets. T-UCRs that belong to more than one category are termed ‘multiple’.

T-UCRs are associated with histone marks for active transcription

Little is known on how T-UCR transcription is initiated and regulated. In addition, T-UCRs are frequently located within protein-coding genes, often resulting in highly correlated expression between the T-UCR and the host gene. To evaluate whether transcriptional initiation of T-UCRs shares characteristics with that of protein-coding genes, we sought for an association between T-UCRs and histone marks for transcription initiation. The genomic location of all H3K4me3 marks (a marker for transcriptional initiation) in four different cell lines was obtained from the UCSC genome browser and was used to determine the distance between the centre of each of the 237 independent T-UCRs and that of the nearest active H3K4me3 mark in any of the four cell lines (Bernstein et al., 2005, 2006; Mikkelsen et al., 2007). As a comparison, this procedure was repeated for either a set of random genomic locations with a similar chromosomal distribution as that of the T-UCRs or for the transcription start sites (TSS, genomic location indicating start of gene transcription) of all RefSeq genes or for all miRNA genes. As expected, the TSS of the RefSeq genes are strongly associated with active H3K4me3 marks when compared with random locations (Kolmogorov–Smirnov test, P<0.0001) (Figure 3). Strikingly, a similar association is observed for T-UCRs (P<0.001) (Figure 3a). To exclude that this is due to the fact that a substantial number of T-UCRs are located within coding genes and thus in close proximity to TSS, we evaluated the H3K4me3 distance distributions of intergenic and intragenic T-UCRs separately. Both intergenic (Kolmogorov–Smirnov test, P<0.05) and intragenic (Kolmogorov–Smirnov test, P<0.0001) T-UCRs were significantly associated with active H3K4me3 marks (Figures 3b and c), but with a different distribution as compared with protein-coding genes (Kolmogorov–Smirnov test, P<0.0001), suggesting a difference in transcriptional organization between T-UCRs and protein-coding genes. In addition, miRNAs were closely associated with active H3K4me3 marks (P<0.0001) (Figure 3d). It is interesting that H3K4me3 distance distributions for miRNAs and T-UCRs appear similar (no indication for a different distribution, Kolmogorov–Smirnov, P>0.05), suggesting common features of transcription initiation for these two classes of non-coding RNAs.

Figure 3

H3K4me3 distance distributions. Distributions of the distance to the nearest histone mark for active transcription (H3K4me3) measured for the transcription start sites of all RefSeq genes (blue), a set of random genomic positions (green), (a) all independently expressed T-UCRs (red), (b) all independently expressed intragenic T-UCRs (red), (c) all independently expressed intergenic T-UCRs (red) and (d) all human miRNAs (red). For visualization purposes, only distances up to 10 000 bp are shown.

T-UCR expression is correlated to clinicogenetic parameters in neuroblastoma

We next examined whether deregulated T-UCR expression might be implicated in known clinicogenetic neuroblastoma subgroups. Differential T-UCR expression was evaluated with respect to MYCN amplification status and UCR copy-number status. MYCN amplification status distinguishes the highly aggressive MYCN-amplified (MNA) tumours from MYCN-non-amplified (MNN) tumours. We found a signature of seven T-UCRs (uc.347, uc.350, uc.279, uc.460, uc.379, uc.446 and uc.364) significantly upregulated in MNA tumours (n=18) compared with MNN tumours (n=31) (Mann–Whitney test, P<0.0001) (Figure 4a, Supplementary Figure 2). Four of the upregulated T-UCRs were intergenic, whereas three were intronic. No T-UCRs were downregulated in the MNA tumours. For a random selection of three of seven T-UCRs, the expression was evaluated on an independent large neuroblastoma tumour cohort containing 366 samples. From the three T-UCRs (uc.279, uc.364, uc.460) that were profiled on this cohort, two (uc.279, uc.460) were significantly upregulated in the MNA tumours (Mann–Whitney test, P<0.05). To evaluate whether any of these seven T-UCRs are induced by MYCN, we profiled their expression in the SHEP-MYCN-ER cellular model system, which allows MYCN activation on the addition of 4-hydroxy tamoxifen (Schulte et al., 2008). Upon MYCN activation, three of seven T-UCRs (uc.460, uc.350 and uc.379) were induced at least twofold (Figure 4b), suggesting that these are MYCN responsive. DNA copy-number changes are known to affect coding gene expression. We therefore examined UCR copy-number changes with respect to T-UCR expression for known critical regions in neuroblastoma (Vandesompele et al., 2005; Lazcoz et al., 2007). We identified seven T-UCRs whose expression correlated to the UCR copy-number status (Supplementary Table 2). These findings demonstrate that deregulated T-UCR expression is associated with genomic aberrations in neuroblastoma tumours.

Figure 4

T-UCR expression in clinicogenetic neuroblastoma subgroups. (a) Expression of a T-UCR signature (n=7) in MYCN-amplified (MNA) and MYCN-non-amplified (MNN) neuroblastoma tumours (n=49), measured by means of the pathway score of all seven T-UCRs (whiskers: Tukey). (b) Expression fold change of uc.350, uc.379 and uc.460 48 h on MYCN activation in the SHEP-MYCN-ER cell line.

Integrative genomics uncovers putative T-UCR functions

Having established differential T-UCR expression patterns in neuroblastoma tumours, we sought to assign putative functions to each T-UCR. To this purpose, we adapted a functional genomics approach, recently proposed by Guttman et al. (2009), based on the integration of multi-level transcriptome data. The independently expressed T-UCRs were correlated to functionally annotated protein-coding genes across the neuroblastoma tumours. The correlation values were subsequently used for Gene Set Enrichment Analysis (GSEA) (Subramanian et al., 2005), thus providing a means to associate each T-UCR to Gene Ontology-based functional classifications, as well as to published experimental data (see Supplementary Methods). The T-UCR annotations serve as a functional resource and are available as supplementary material (Supplementary Files 2, 3 and 4). Significant enrichments (FDR <5%) were extracted and results were clustered to visualize classes of T-UCRs with similar predicted functions (Figure 5a and Supplementary Figure 3). For a large number of T-UCRs, we observed widespread association to numerous cancer-related cellular functions and pathways such as proliferation, apoptosis and differentiation. For example, the most prominent cluster identified using this methodology contained several T-UCRs significantly related to the expression of protein-coding genes involved in the inter-related processes of cell cycle, DNA replication and DNA repair (Figure 5b).

Figure 5

Functional annotation of T-UCRs. (a) Hierarchical clustering of T-UCRs (x axis) and Gene Ontology Biological Process (y axis) showing significant (FDR <5%) positive correlations (blue), significant negative correlations (red) and no correlation (white) as determined by a Gene Set Enrichment Analysis-based approach. The predominant functional cluster is boxed. (b) Gene Ontology Biological Process categories of the boxed cluster with indication of the number of significantly correlated T-UCRs.

To assess the validity of the functional predictions, we sought independent experimental validation for a subset of inferred T-UCR functions. We decided to examine the T-UCRs that were annotated to the TP53 response pathway according to association with TP53-related gene sets from the molecular function and chemical and genetic perturbations collections. To this end, a human neuroblastoma cell line (NGP) lentivirally transfected with a short hairpin RNA against human TP53 (LV-h-p53) or murine TP53 (LV-m-p53, used as a negative control) (Van Maerken et al., 2006) was treated with nutlin-3 for 24 h and profiled for T-UCR expression. On nutlin-3 treatment, MDM2 activity is antagonized leading to TP53 stabilization and activation of function in cells with wild-type TP53 (Vassilev et al., 2004; Van Maerken et al., 2006). We found 40 T-UCRs to be responsive to nutlin-3 treatment in NGP-LV-m-p53 cells, but not in NGP-LV-h-p53 cells, indicating a TP53-dependent expression. Almost three quarters (29 of 40) were annotated to the TP53 response pathway (Fisher's exact test, P<0.05), thus confirming the functional predictions. Earlier, Calin et al. (2007) reported that T-UCR uc.73 was implicated in the apoptotic response in colon cancer cell line COLO320. Interestingly, we found uc.73 to be annotated to the TP53 response pathway supporting its reported role in apoptosis and further corroborating our workflow.

Functional T-UCR expression network in neuroblastoma

Highly co-regulated genes often share a common activating/repressing feature, that is, they belong to the same cellular process or are transcribed in response to a common internal or external stimulus. As previously reported, several of the T-UCRs are located in or around genes related to differentiation and splicing (Bejerano et al., 2004), thus by genomic location alone, groups of related T-UCRs have been identified. To further examine putative shared functions among the 237 independently expressed T-UCRs, we first visualized all pairwise T-UCR correlations (Supplementary Figure 4). As several groups of co-expressed T-UCRs could be identified, we used a network approach in which significantly correlated T-UCRs (Pearson's correlation, P<0.001) are connected. Individual T-UCRs with less than two significant connections were excluded, resulting in a highly interconnected network of related T-UCRs (Figure 6). Four major clusters, consisting of 9, 11, 9 and 6 T-UCRs respectively, were identified (Figure 6). To assign putative common functions, each cluster was annotated based on the GSEA results. Significant T-UCR GSEA terms from the biological process, molecular function, as identified above, were tested for association with the network clusters using Fisher's exact test. For cluster 1, we identified DNA damage response as a significant GSEA-based annotation. Cluster 2 was predominantly associated with cell-cycle regulation and proliferation, whereas cluster 3 appeared to be implicated in differentiation. For cluster 4, development and immune response were among the significant GSEA terms. DNA damage response is highly related to TP53 activation. To validate the annotation of cluster 1, we evaluated its expression by means of the cluster 1 pathway activity score in the nutlin-3-treated NGP-LV-h-p53 and NGP-LV-m-p53 cells. No significant differential pathway activity score was observed in the NGP-LV-h-p53 cells (fold change <1.5), whereas the pathway activity score markedly changed (fold change >1.5, Mann–Whitney test, P<0.01) in the NGP-LV-m-p53 cells, 24 h upon nutlin-3 treatment (Figure 7a). These findings suggest that cluster 1 represents a subset of co-expressed T-UCRs that are responsive to TP53 activation. As sustained nutlin-3 treatment of NGP-LV-m-p53 cells results in TP53-dependent neuronal differentiation (Van Maerken et al., 2006), we evaluated cluster 3 pathway activity in NGP-LV-h-p53 and NGP-LV-m-p53 cells, 5 days upon nutlin-3 treatment. NGP-LV-m-p53 cells, but not NGP-LV-h-p53 cells, showed marked increase in neuronal differentiation after 5 days of nutlin-3 treatment (Figure 7b). Differential activity of cluster 3 was observed in NGP-LV-m-p53 cells after nutlin-3 treatment (fold change >1.5, Mann–Whitney test, P=0.074), but not in NGP-LV-h-p53 cells, suggesting a role for cluster 3 T-UCRs in p-53-dependent neuronal differentiation (Figure 7c). We further evaluated cluster pathway scores in the neuroblastoma tumour cohort with respect to patient survival. Samples were divided into quartiles according to the pathway activity score (Fredlund et al., 2008). Interestingly, Kaplan–Meier analysis based on these quartiles revealed a significant correlation (P<0.01) between the activity of cluster 4 and overall and event-free patient survival (Figure 7d). These results implicate co-expressed T-UCR clusters in different aspects of neuroblastoma disease and outcome.

Figure 6

T-UCR expression network in neuroblastoma. Interconnected expression network of T-UCRs in neuroblastoma. Pairs of T-UCRs with a significant correlation (P<0.001; Pearson's correlation) were connected followed by exclusion of T-UCRs with less than two significant connections. Positive correlations are indicated in green, negative correlations in red. Four individual clusters are apparent from the network.

Figure 7

T-UCR expression clusters are correlated to neuroblastoma biology and outcome. (a) T-UCR cluster 1 is implicated in TP53 response in neuroblastoma. Cluster 1 expression is summarized by means of its pathway activity score. Pathway activity score fold changes on treatment of NGP-LV-h-p53 and NGP-LV-m-p53 with 16 μM nutlin-3 for 1 day, relative to vehicle treatment, are plotted. Fold changes >1.5 are indicated in red. (b) Morphology of NGP-LV-m-p53 and NGP-LV-h-p53 cells at day 5 of treatment with 16 μM nutlin-3 (5d+) or vehicle control (5d−). NGP-LV-m-p53 cells show an extensive neurite outgrowth after 5 days of nutlin-3 treatment. NGP-LV-h-p53 cells did not show any signs of neuronal differentiation. (c) T-UCR cluster 3 is implicated in differentiation of neuroblastoma cells. Pathway activity fold changes were calculated as in (a). (d) Kaplan–Meier plots for overall (OS) and event-free (EFS) survival of neuroblastoma patients based on the pathway activity score of T-UCR cluster 4, represented as quartiles. Increased activity of cluster 4 correlated to a poor overall and EFS. (e) Mean expression (log 2) of T-UCR cluster 2 in control (grey) and serum-starved (red) BJ cells (error bars reflect s.e.m.).

Finally, we asked whether the inferred T-UCR functions would hold true for tissue types other than neuroblastoma. To this end, we cultured immortalized human fibroblast BJ cells in starvation medium (0.1% fetal calf serum) to induce cell-cycle arrest and thus block proliferation. Next, we examined the expression of T-UCR cluster 2, which was annotated to cell-cycle regulation and proliferation, in control and serum starved BJ cells. Interestingly, we observed a significant differential expression of cluster 2 on serum starvation (Mann–Whitney test, P<0.05) (Figure 7e), suggesting that our proposed functional T-UCR annotations are also valid for human fibroblasts and may potentially be applied to other cell types.


Non-coding RNAs have emerged as an important component of the human transcriptome. Significant progress is being made on the functional annotation of a particular class of small non-coding miRNAs, although the functionality of others, such as T-UCRs, is still elusive. In this study, we aimed at characterizing T-UCR expression and function, based on an integrative analysis in an aggressive childhood tumour. We designed an RT–qPCR based T-UCR-profiling platform for measurement of the expression of all 481 T-UCRs. RT–qPCR is the gold standard for small RNA profiling (Chen et al., 2005; Mestdagh et al., 2008) and has a superior specificity, sensitivity and flexibility compared with array-based expression profiling platforms.

As UCRs are defined solely on the basis of sequence similarity, cautious interpretation of their measured transcription is warranted. Perfectly conserved sequences that are located exonically might very well represent important features of the host gene instead of an independently expressed T-UCR. We therefore determined which T-UCRs were expressed independently from their genomic environment. Strikingly, about half of the T-UCRs showed strong positive correlations to the expression of their host gene. Not surprisingly, this was mainly the case for T-UCRs that overlapped with an exon of a protein-coding gene, suggesting that some UCRs represent a conserved feature of the host gene rather than coding for an independent transcriptional unit.

To gain further insight into the initiation and regulation of T-UCR transcription, we evaluated the chromatin state of the T-UCR genomic neighbourhood. Actively transcribed genes, both coding and non-coding, are marked by trimethylation of lysine 4 of histone H3 (H3K4me3) at their promoter (Mikkelsen et al., 2007; Guttman et al., 2009), a feature that has been used to identify lincRNAs (Guttman et al., 2009). We hypothesized that comparing the distribution of active H3K4me3 marks between T-UCRs on one hand and miRNAs and protein-coding genes on the other could reveal insights into the general structure of T-UCR transcriptional units and their initiation. We observed an association between H3K4me3 marks and T-UCRs, independent of host-gene-associated H3K4me3 marks. The H3K4me3 distance distribution was different from that of the protein-coding genes but showed a striking similarity to that of miRNAs. Compared with protein-coding genes, miRNA promoters have similar features but the organization of the transcriptional unit is more complex as miRNAs can be transcribed from promoters located several kilobases (up to 40 kb) away (Corcoran et al., 2009). The correspondence in H3K4me3 distance distribution between miRNAs and T-UCRs suggests a similar transcriptional organization with initiation sites located several kilobases away from the T-UCR. Further experimental evaluation is necessary to validate these observations.

The high conservation of T-UCRs across species almost inevitably implies functionality. Previously, Calin et al. (2007) reported differential T-UCR expression between normal and cancerous tissues and identified one T-UCR, uc.73, to be oncogenic in colon cancer. In this study, we show that T-UCRs are widely expressed in neuroblastoma tumours and that their expression correlates to important clinicogenetic parameters such as MYCN amplification status. In addition, DNA copy-number changes that are associated with neuroblastoma disease were shown to affect T-UCR expression, which is in line with the observation that T-UCRs are frequently located at fragile sites or genomic regions involved in cancer (Calin et al., 2007). To gain further insight into the pathways and processes in which T-UCRs are involved, we implemented an integrative genomics workflow to infer putative T-UCR functions using Gene Set Enrichment Analysis. In support of the workflow, T-UCRs predicted to be involved in apoptosis and differentiation were experimentally validated using a cellular model system. Furthermore, our predictions annotated uc.73 to the TP53 response pathway. This result is in concordance with published results showing that RNA interference-mediated knockdown of uc.73 induced apoptosis in a colon cancer cell line (Calin et al., 2007), thus again confirming the validity of our workflow.

Our further analyses of T-UCR expression patterns uncovered an interconnected network consisting of four major clusters, functionally associated with cancer-related cellular processes such as proliferation, apoptosis and differentiation. Moreover, expression of cluster 4 correlated to patient outcome, thus further indicating T-UCRs as a factor in neuroblastoma biology. Interestingly, cluster 4 was associated with development and immune response, which is in agreement with our previously reported observation that neuroblastoma tumours, as compared with normal precursor neuroblasts, are characterized by an overrepresentation of genes involved in immune response (De Preter et al., 2006). In addition, we have shown that the inferred T-UCR functions are not confined to neuroblastoma cells but are also valid for immortalized human fibroblasts. This suggests that T-UCRs are of general relevance in cell biology and opens perspectives to use these annotations as a functional resource in future T-UCR studies.

In conclusion, our results show that T-UCRs are widely expressed in neuroblastoma tumours and correlate to clinicogenetic parameters. Functional T-UCR annotations, inferred through a functional genomics approach and validated using cellular models, reveal associations with several cancer-related cellular processes such as apoptosis and differentiation. Additional studies are needed to further elucidate T-UCR function and to unravel the transcriptional programmes mediating their expression. T-UCRs make up an interesting class of non-coding RNAs and could prove attractive targets for treatment or diagnosis.

Materials and methods

Patient samples

A total of 49 neuroblastoma tumours were collected at the Ghent University Hospital (Ghent, Belgium) and at the Medical School of Valencia (Valencia, Spain) before treatment (Supplementary Table 3). An additional cohort of 366 neuroblastoma tumours, originally described by Vermeulen et al. (2009), were also included. All samples were obtained at diagnosis. Informed consent was obtained from the patient's relatives. Patients were staged according to the International Neuroblastoma Staging System (Brodeur et al., 1993).

Cellular models

NGP-LV-h-p53 and NGP-LV-m-p53 cells (Van Maerken et al., 2006) were cultured in RPMI 1640 (Invitrogen, Carlsbad, CA, USA) supplemented with 15% fetal calf serum and treated with 16 μM nutlin-3 (Cayman Chemical, Ann Arbor, MI, USA) or vehicle control (ethanol) for 1 and 5 days before harvesting. SHEP-MYCN-ER cells (Schulte et al., 2008) were treated with 4-hydroxytamoxifen or vehicle control (ethanol) for 2 days before harvesting. BJ cells were serum starved for 5 days in Dulbecco's modified Eagle's medium (DMEM) supplemented with 0.1% fetal calf serum. Medium was refreshed after 3 days. Non-starved BJ cells were cultured in DMEM (10% fetal calf serum) for 5 days.

T-UCR primer design

RT–qPCR primers for 481 T-UCRs were designed using Primer3 (stand alone or implemented in Beacondesigner) (Rozen and Skaletsky, 2000) and validated through an in silico primer analysis pipeline (Lefever et al., 2009). Designs were selected according to four different criteria: absence of stable secondary structures in the primer-annealing regions, specificity, absence of SNPs in the primer-annealing regions and 3′ GC content. All assays are available from PrimerDesign Ltd (Southampton, UK). Primer efficiencies were determined using PCR miner (Zhao and Fernald, 2005). To evaluate primer specificity, amplicons were sized on a Caliper LC90 (Caliper Life Sciences, Hopkinton, MA, USA).


For detailed reaction conditions see Supplementary Methods. Expression data were normalized using the mean expression value per sample (Mestdagh et al., 2009). For the independent validation cohort of 366 tumours, RT–qPCR data were normalized using qbasePLUS v1.2 as described previously (Vermeulen et al., 2009). T-UCR RT–qPCR data are available in rdml format (Lefever et al., 2009) (Supplementary File 5). Compliance of qPCR experiments with the MIQE guidelines (Bustin et al., 2009) ( is listed in the MIQE checklist (Supplementary File 6).

Exon array

Total RNA was isolated from 40 tumour samples and was hybridized to Human Exon 1.0. ST array (Affymetrix, Santa Clara, CA, USA) at the microarray facility of the University Hospital of Essen according to the manufacturer's protocol. To obtain expression information per gene, exon data were merged by transcript clusters. Exon array data were normalized according to the RMA-sketch algorithm using Affymetrix Power Tools (Affymetrix).

Array comparative hybridization

Array CGH was performed for detection of T-UCR copy-number alterations using a custom 44K array enriched for regions with recurrent imbalances in neuroblastoma (1p, 2p, 3p, 11q, 17) and T-UCR genes (Agilent Technologies, Palo Alto, CA, USA). See Supplementary Methods for detailed description.


See Supplementary Methods for detailed description.


  1. Bartel DP . (2009). MicroRNAs: target recognition and regulatory functions. Cell 136: 215–233.

    CAS  Article  Google Scholar 

  2. Bejerano G, Pheasant M, Makunin I, Stephen S, Kent WJ, Mattick JS et al. (2004). Ultraconserved elements in the human genome. Science 304: 1321–1325.

    CAS  Article  Google Scholar 

  3. Bernstein BE, Kamal M, Lindblad-Toh K, Bekiranov S, Bailey DK, Huebert DJ et al. (2005). Genomic maps and comparative analysis of histone modifications in human and mouse. Cell 120: 169–181.

    CAS  Article  Google Scholar 

  4. Bernstein BE, Mikkelsen TS, Xie X, Kamal M, Huebert DJ, Cuff J et al. (2006). A bivalent chromatin structure marks key developmental genes in embryonic stem cells. Cell 125: 315–326.

    CAS  PubMed  Google Scholar 

  5. Brodeur GM, Pritchard J, Berthold F, Carlsen NL, Castel V, Castelberry RP et al. (1993). Revisions of the international criteria for neuroblastoma diagnosis, staging, and response to treatment. J Clin Oncol 11: 1466–1477.

    CAS  Article  Google Scholar 

  6. Bustin SA, Benes V, Garson JA, Hellemans J, Huggett J, Kubista M et al. (2009). The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clin Chem 55: 611–622.

    CAS  Article  Google Scholar 

  7. Calin GA, Liu CG, Ferracin M, Hyslop T, Spizzo R, Sevignani C et al. (2007). Ultraconserved regions encoding ncRNAs are altered in human leukemias and carcinomas. Cancer Cell 12: 215–229.

    CAS  Article  Google Scholar 

  8. Chen C, Ridzon DA, Broomer AJ, Zhou Z, Lee DH, Nguyen JT et al. (2005). Real-time quantification of microRNAs by stem-loop RT-PCR. Nucleic Acids Res 33: e179.

    Article  Google Scholar 

  9. Corcoran DL, Pandit KV, Gordon B, Bhattacharjee A, Kaminski N, Benos PV . (2009). Features of mammalian microRNA promoters emerge from polymerase II chromatin immunoprecipitation data. PLoS One 4: e5279.

    Article  Google Scholar 

  10. De Preter K, Vandesompele J, Heimann P, Yigit N, Beckman S, Schramm A et al. (2006). Human fetal neuroblast and neuroblastoma transcriptome analysis confirms neuroblast origin and highlights neuroblastoma candidate genes. Genome Biol 7: R84.

    Article  Google Scholar 

  11. Esquela-Kerscher A, Slack FJ . (2006). Oncomirs—microRNAs with a role in cancer. Nat Rev Cancer 6: 259–269.

    CAS  Article  Google Scholar 

  12. Fredlund E, Ringner M, Maris JM, Pahlman S . (2008). High Myc pathway activity and low stage of neuronal differentiation associate with poor outcome in neuroblastoma. Proc Natl Acad Sci USA 105: 14094–14099.

    CAS  Article  Google Scholar 

  13. Guttman M, Amit I, Garber M, French C, Lin MF, Feldser D et al. (2009). Chromatin signature reveals over a thousand highly conserved large non-coding RNAs in mammals. Nature 458: 223–227.

    CAS  Article  Google Scholar 

  14. He L, He X, Lim LP, de Stanchina E, Xuan Z, Liang Y et al. (2007). A microRNA component of the p53 tumour suppressor network. Nature 447: 1130–1134.

    CAS  Article  Google Scholar 

  15. Khalil AM, Guttman M, Huarte M, Garber M, Raj A, Rivea Morales D et al. (2009). Many human large intergenic noncoding RNAs associate with chromatin-modifying complexes and affect gene expression. Proc Natl Acad Sci USA 106: 11667–11672.

    CAS  Article  Google Scholar 

  16. Lazcoz P, Munoz J, Nistal M, Pestana A, Encio IJ, Castresana JS . (2007). Loss of heterozygosity and microsatellite instability on chromosome arm 10q in neuroblastoma. Cancer Genet Cytogenet 174: 1–8.

    CAS  Article  Google Scholar 

  17. 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.

    CAS  Article  Google Scholar 

  18. Maris JM, Hogarty MD, Bagatell R, Cohn SL . (2007). Neuroblastoma. Lancet 369: 2106–2120.

    CAS  Article  Google Scholar 

  19. Mestdagh P, Feys T, Bernard N, Guenther S, Chen C, Speleman F et al. (2008). High-throughput stem-loop RT-qPCR miRNA expression profiling using minute amounts of input RNA. Nucleic Acids Res 36: e143.

    Article  Google Scholar 

  20. Mestdagh P, Van Vlierberghe P, De Weer A, Muth D, Westermann F, Speleman F et al. (2009). A novel and universal method for microRNA RT-qPCR data normalization. Genome Biol 10: R64.

    Article  Google Scholar 

  21. Mikkelsen TS, Ku M, Jaffe DB, Issac B, Lieberman E, Giannoukos G et al. (2007). Genome-wide maps of chromatin state in pluripotent and lineage-committed cells. Nature 448: 553–560.

    CAS  Article  Google Scholar 

  22. Nobrega MA, Ovcharenko I, Afzal V, Rubin EM . (2003). Scanning human gene deserts for long-range enhancers. Science 302: 413.

    CAS  Article  Google Scholar 

  23. O'Donnell KA, Wentzel EA, Zeller KI, Dang CV, Mendell JT . (2005). c-Myc-regulated microRNAs modulate E2F1 expression. Nature 435: 839–843.

    CAS  Article  Google Scholar 

  24. Pruitt KD, Tatusova T, Klimke W, Maglott DR . (2009). NCBI reference sequences: current status, policy and new initiatives. Nucleic Acids Res 37: D32–D36.

    CAS  Article  Google Scholar 

  25. Rozen S, Skaletsky HJ . (2000). Primer3 on the WWW for general users and for biologist programmers. In: Krawetz S and Misener S (eds). Bioinformatics Methods and Protocols: Methods in Molecular Biology. Humana Press: Totowa, NJ. pp 365–386.

    Google Scholar 

  26. Schulte JH, Horn S, Otto T, Samans B, Heukamp LC, Eilers UC et al. (2008). MYCN regulates oncogenic microRNAs in neuroblastoma. Int J Cancer 122: 699–704.

    CAS  Article  Google Scholar 

  27. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA et al. (2005). Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA 102: 15545–15550.

    CAS  Article  Google Scholar 

  28. Van Maerken T, Speleman F, Vermeulen J, Lambertz I, De Clercq S, De Smet E et al. (2006). Small-molecule MDM2 antagonists as a new therapy concept for neuroblastoma. Cancer Res 66: 9646–9655.

    CAS  Article  Google Scholar 

  29. Vandesompele J, Baudis M, De Preter K, Van Roy N, Ambros P, Bown N et al. (2005). Unequivocal delineation of clinicogenetic subgroups and development of a new model for improved outcome prediction in neuroblastoma. J Clin Oncol 23: 2280–2299.

    CAS  Article  Google Scholar 

  30. Vassilev LT, Vu BT, Graves B, Carvajal D, Podlaski F, Filipovic Z et al. (2004). in vivo activation of the p53 pathway by small-molecule antagonists of MDM2. Science 303: 844–848.

    CAS  Article  Google Scholar 

  31. Vermeulen J, De Preter K, Naranjo A, Vercruysse L, Van Roy N, 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.

    CAS  Article  Google Scholar 

  32. Zhao S, Fernald RD . (2005). Comprehensive algorithm for quantitative real-time polymerase chain reaction. J Comput Biol 12: 1047–1064.

    CAS  Article  Google Scholar 

Download references


We are indebted to all the members of the SIOPEN and the GPOH for providing tumour samples or the clinical history of patients. This research was funded by the Gent University Research Fund (BOF 01D31406 to PM, BOF 01F07207 to FP, BOF 01Z09407 to J Vandesompele), the Belgian Kid's Fund and the Fondation pour la recherche Nuovo-Soldati (J Vermeulen), RD06/0020/0102 from RTICC/ISCIII to RN, the American Cancer Association to EF, the Swedish Cancer Society to MR, the Fund for Scientific Research (grant number: G.0198.08 and 31511809) and the Belgian Foundation Against Cancer, found of public interest (project SCIE2006-25). KDP is a postdoctoral researcher with the Fund for Scientific Research-Flanders. CK is supported by a doctoral grant from the Institute for the Promotion of Innovation by Science and Technology in Flanders (IWT - 081373). We acknowledge the support of the European Community under the FP6 (project: STREP: EET-pipeline, number: 037260) and FP7 (ONCOMIRS, grant agreement number 201102). This publication reflects only authors' views; the commission is not liable for any use that may be made of the information herein. This article presents research results of the Belgian programme of Interuniversity Poles of Attraction, initiated by the Belgian State, Prime Minister's Office, Science Policy Programming.

Author information



Corresponding author

Correspondence to J Vandesompele.

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

Mestdagh, P., Fredlund, E., Pattyn, F. et al. An integrative genomics screen uncovers ncRNA T-UCR functions in neuroblastoma tumours. Oncogene 29, 3583–3592 (2010).

Download citation


  • neuroblastoma
  • T-UCR
  • non-coding RNA

Further reading


Quick links