• neuroblastoma;
  • differential gene expression;
  • local variant;
  • metastatic variant;
  • oligonucleotide microarray


  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Metastasis is the primary cause of mortality in Neuroblastoma (NB) patients, but the metastatic process in NB is poorly understood. Metastsis is a multistep process that requires the coordinated action of many genes. The identification of genes that promote or suppress tumor metastasis can advance our understanding of this process. In the present study, we utilized a human NB xenograft model comprising local and metastatic NB variants, which was recently developed in our laboratory. We set out to identify molecular correlates of NB metastasis and to determine the clinical relevance of these molecules. We first performed genome-wide expression profiles of metastatic and nonmetastatic NB variants that have an identical genetic background. We found that some of the proteins highly expressed in the metastatic NB variants are localized in the cytoplasm and endoplasmic reticulum. Other proteins are linked to metabolic processes and signaling pathways, thereby supporting the invasive and metastatic state of the cells. Subsequently, we intersected the differentially expressed genes in the human xenografted variants with genes differentially expressed in Stage 1 and Stage 4 primary tumors of NB patients. By using the same gene-expression platform, molecular correlates associated with metastatic progression in primary NB tumors were identified. The resulting smaller gene set was clinically relevant as it discriminated between high- and low-risk NB patients.

Neuroblastoma (NB) is the most commonly occurring extracranial tumor in childhood. These tumors originate in embryonic neural crest precursor cells of the sympathetic nervous system. NB accounts for ∼8% of all malignancies in patients younger than 15 years, and occurs most frequently in the adrenal gland.1 The tumors can regress spontaneously, particularly in infants. However, children older than 1 year of age, with a widespread metastatic disease or with a large, aggressive, localized tumor, have a poor long-term survival rate of ∼30%.2, 3

Metastasis is the primary cause of mortality in cancer patients, but despite of that it is one of the most poorly understood processes in cancer biology. This is a multistep process that requires the coordinated action of many genes.4–6 A systematic identification of genes that promote or suppress tumor invasion and metastasis can advance the understanding of this process.

We recently described the generation and characterization of novel local and metastatic human NB variants.7 Lung-metastasizing MHH-NB-11 cells were isolated from mice bearing orthotopically inoculated adrenal tumors. “Local” variants were isolated from the adrenal tumors. “Lung metastatic” variants were generated by repeated cycles of in vivo passages. These cells displayed an aggressive and metastatic phenotype in vivo, portend a poor prognosis and exhibited unique properties in vitro.7

Oberthuer et al. recently described a robust, gene-expression-based classifier, which reliably predicts NB tumor behavior and can aid physicians in choosing the most appropriate form of first-line treatment.8 For this study, they constructed a NB-specific oligonucleotide-array utilizing several platforms of gene-expression data. This array comprises 10,163 (11K) probes for the 8,155 Unigene Cluster considered to be important in the development and progression of NB.

Using this array, we performed genome-wide expression profiles of our genetically identical local and metastatic human NB variants in nude mice. Differentially expressed genes were detected in these variants and a “NB metastatic gene-expression signature” could be identified. We show that this signature identifies high-risk NB patients and suggest that these genes could be used as therapy targets or prognostic markers in NB.

Material and Methods

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Cell cultures

The MHH-NB-119 NB parental cell line was kindly provided by Dr. T. Pietsch, Department of Neuropathology, University of Bonn Medical Center, Bonn, Germany. MHH.Ad and SY5Y.Ad NB local variants and MHH.Lu3 and SY5Y.Lu3 NB metastatic variants were maintained as previously described.7

RNA preparation

For gene expression analysis of NB cells, total RNA was isolated using the Lysing Matrix D tubes containing TRIzol reagent. RNA integrity was assessed using the 2100 Bioanalyzer considering only samples with an RNA Integrity Number ≥ 7.5.


The study comprised 134 patients (Stage 1 or Stage 4) out of the 251 patients of the German NB Trials NB90-NB2004, diagnosed between 1989 and 2004. Patient's parameters (e.g., age at diagnosis, stage, response to treatment etc.) as well as RNA preparation and gene-expression analysis of tumor samples were described previously.8

Gene expression analysis

Gene-expression profiles of NB cell variants were generated as dye-flipped dual-color replicates using customized NB-related oligonucleotide microarray comprising 10,163 oligonucleotide probes (11K) as previously described.8 For each sample, 1 μg of linearly amplified Cy3- and Cy5-labeled cRNA, respectively, was hybridized with a common reference of 1 μg of reverse-color Cy-labeled cRNA of a total RNA pool of 100 NB tumor samples8 using Agilent's Low-RNA-Input-Linear Amplification Kit and In Situ-Hybridization Kit-Plus. After washing and scanning, raw microarray data of the local and metastatic variants were normalized using linear and LOWESS method preformed by Agilent's feature extraction software. Cell variants data from dye-flipped chip pairs were considered as 2 separated technical repeats. Multidimensional scaling (MDS),10 hierarchical clustering analyses and analyses of variance (ANOVA) were performed using the Rosetta Resolver® Gene-Expression software (Rosetta Biosoftware, Seattle, WA). To identify candidate genes that were differentially expressed in NB local and metastatic variants, we applied the 1-way ANOVA statistic model. Genes with a ≥ 1-fold (in log10 scale) increase or decrease in expression and 2-tailed p values smaller or equivalent to 0.01 were selected.

Later analyses were performed in order to refine the metastasis-associated gene expression signature to a smaller set of genes and to associate it with the clinical outcome of NB patients. First, the same criteria, as described for the in vivo generated variants, were applied on the gene expression data of primary tumors from Stage 1 and Stage 4 NB patients. Then, we intersected between the deferentially expressed gene lists of the in vivo generated variants and of the primary tumors.

Gene ontology

Gene and protein names, symbols and accession numbers were identified using Bioinformatics databases (National Center for Biotechnology Information (NCBI) RefSeq:; UniProt: Gene Ontology (GO) annotations were analyzed with the Panther Protein Classification System ( to identify functional annotations that were significantly enriched in a gene set compared to the entire human genome. Prediction of subcellular localization was assigned by using DAVID Functional Annotation Tool, ( and pTARGET (


The following antibodies were used for Western blotting assays: mouse anti-human Paracingulin (CGNL1, Zymed Laboratories, San Francisco, CA); goat anti-human Hexokinase 2 (HK2), goat anti-human Damage-specific DNA Binding Protein 2 (DDB2) and rabbit anti-human ERK2 polyclonal antibody (all from Santa Cruz Biotechnology, Santa Cruz, CA); mouse anti-human Thymidine Kinase 1 (TK1; Novus Biologicals, Littleton, CO). Horseradish peroxidase-conjugated goat anti-mouse, goat anti-rabbit and goat anti-donkey antibodies were used according to the manufacturer's instructions (Jackson ImmunoResearch Laboratories, West Grove, PA).

Western blotting

Cell lysates were incubated for 20 min on ice, and cleared by centrifugation for 20 min at 16,000g, 4°C. After the addition of Laemmli sample buffer the lysates were boiled for 10 min, resolved on SDS-PAGE and transferred onto nitrocellulose membrane. The membrane was blocked at room temperature with 3% BSA or 5% nonfat dry milk diluted in TBS-Tween for 1 hr. The target proteins were detected by using a relevant primary antibody and suitable HRP-conjugated secondary antibodies, as described above. Ponceau staining prior to blocking of the membrane or ERK2 protein were used for loading control. The bands were visualized by chemoluminescence-ECL reactions (Amersham, Buckinghamshire, United Kingdom) and autoradiography by exposure to Fuji film. The amount of the relevant protein in the lanes was estimated by densitometry and was calculated in reference to the loading control in the lane as measured by densitometry using Scion Image software. The protein expression values of the metastatic variant were normalized to the protein expression values of the local variant.

Cellular fractionation

Cell pellets were washed twice with phosphate-buffered saline, then twice with washing buffer (20 mM Hepes, 5 mM KCl, 150 mM NaCl) and cleared by centrifugation at 900g for 5 min. Cell pellets were lysed and homogenized in lysis buffer A [5 mM HEPES, 1 mM sodium pyrophosphate pH 8.0, 1 mM NaF, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride (PMSF), 2 μg/ml leupeptin and 2 μg/ml aprotinin] with 30 strokes of an homogenizer on ice. Following centrifugation at 500g for 10 min and collection of cleared supernatants (consisting the cytoplasmatic and membrane fraction), pelleted nuclei were resuspended with lysis buffer C (350 mM NaCl, 20 mM HEPES, pH 8.0, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM sodium orthovanadate, 1 mM PMSF, 4 mM β-glycerol phosphate, 2 μg/ml leupeptin and 2 μg/ml aprotinin). Lysates were incubated for 30 min on ice, vortexed and centrifuged for 30 min. The cleared supernatant contained the nuclear fraction. Detection of DDB2 (nuclear protein) in the highly enriched nuclear fraction was performed by Western blot analysis.


  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

This study aimed at identifying genes associated with NB progression and metastasis, and to determine their potential clinical relevance. To this end, we used a recently established orthotopic model of human NB local and metastatic variants originating from the same NB tumor.7 The working hypothesis is that there are differentially expressed genes between local and metastatic NB variants having the same genetic background. In this study, we utilized the MHH system, the metastatic variants of which displayed unique properties and demonstrated a metastatic phenotype in vivo.7

Gene expression signature associated with the ability of NB cells to metastasize to the lung

Utilizing a NB-related customized oligonucleotide microarray,8 we generated gene-expression profiles of the parental NB cell line MHH-NB-11, the local tumor variant MHH.Ad, the 2nd cycle lung metastatic variant MHH.Lu2 and the 3rd cycle lung metastatic variant MHH.Lu3. Four technical repeats were preformed for the local and the metastatic variants. One out of four technical repeats of the local variant was excluded due to poor quality of the resulting gene-expression profile. As a first step in evaluating our hypothesis, we examined the entire set of data using unsupervised MDS analysis10 (Fig. 1a). This tool visualized the genetic similarities or dissimilarities between the variants depicted as distance in 3 dimensions. These relations measured over all genes on the array between all variants. This analysis revealed only marginal separation between the local tumor variant and the 2nd cycle of the lung metastatic variant, and a distinct separation between the local tumor variant and the 3rd cycle lung metastatic variant. Two technical repeats of the parental cell line were added in order to emphasize the gradual dissimilarities in global expression during the generation of the metastatic variants.

thumbnail image

Figure 1. (a) Differential gene expression between NB local and metastatic variants. MDS10 of all gene expression profiles. The spatial locations of the various cell variants in the 3D plot are presented as dots. Each dot represents the gene expression profile of a single technical repeat. (b) Two-dimensional hierarchical cluster diagram of 116 probes representing 112 genes, differentially expressed between local and metastatic variants. Comparison of gene expression profiles of the metastatic variants (MHH.Lu.Lu and MHH.Lu.Lu.Lu) and the local variant (MHH.Ad) identifies 116 probes. The gene-expression data representing 112 differentially expressed genes were applied to all samples. Rows represent samples, columns represent genes. Varying expression values are indicated as log10 ratios ranging from −0.8 (blue) to +0.8 (red). For the gene list, please refer to Supporting Information file 1.

Download figure to PowerPoint

Successive in vivo passages of tumor cells increase the metastatic capacity of such cells.14–18,48 The results of our MDS analysis clearly indicate that progressive alterations in global gene expression of the NB cells occur as a result of each in vivo passage. In view of the fact that the differences in gene expression patterns between the metastatic variants generated from the 2nd and the 3rd in vivo passage cycles were rather minimal, we considered, in later analyses, the lung metastatic variants of the 2nd and the 3rd cycles to be a single group.

Subsequently, we compared the transcriptional profiles of the local tumor variant (MHH.Ad) and the lung metastatic variants (MHH.Lu2 and MHH.Lu3) in order to identify differentially expressed genes, By calculating an ANOVA, 116 probes (Fig. 1b) representing 112 genes (listed in Supporting Information file 1) were described as differentially expressed by our filtering and statistical comparison criteria.

Biological properties of the differentially expressed genes

To classify the 112 genes into biological categories, we separated them into 2 sets: 64 genes that showed higher expression levels in the metastatic variants than in the local variant and 39 genes with higher expression levels in the local variant than in the metastatic variants. Nine genes that exhibited only very subtle differences in expression levels between local and metastatic variants were excluded from this analysis.

We analyzed the GO annotations of each set of genes with the Panther Protein Classification System compared to the entire NCBI reference list of the human genome.11 As shown in Tables 1 and 2, the Panther system found more categories that were significantly enriched in the gene set of the metastatic variants as compared to the local variant (p-value < 0.05, as determined by binomial statistics19). The 64 genes whose expression was higher in the metastatic variants than in the local variant were classified by the Panther system into genes coding for 19 biological processes; for 13 molecular functions and for 21 signaling pathways. Furthermore, 3 of the identified categories (carbohydrate and other polysaccharide metabolism pathways and the p53 pathway) remained significantly represented in this set of genes even after the application of the Bonferroni correction for multiple testing. (Table 1, see p-value with Bonferroni correction).

Table 1. Panther classification of biological processes, molecular functions and pathways significantly enriched in a set of 64 genes highly expressed in the metastatic variants as compared to the local variant
inline image
Table 2. Panther classification of biological processes, molecular functions and pathways significantly enriched in a set of 39 genes highly expressed in the local variant as compared to the metastatic variants
inline image

The 42 genes whose expression in the local variant was higher than in the metastatic variants were classified by the Panther system into those coding for 10 biological processes; for 3 molecular functions and for 3 signaling pathways. However, none of the categories were significantly represented in this set of genes after Bonferroni correction for multiple testing.

Several genes that were more highly expressed by the metastatic variants are included in categories which were previously reported to be associated with tumor progression. Among these categories are genes coding for biological processes such as carbohydrate metabolism, other polysaccharide metabolism, lipid, fatty acid and steroid metabolism, cell cycle control and cell structure and motility. In addition, Panther detected genes coding for several pathways such as p53 pathway, inflammation mediated by chemokine and cytokine signaling pathway, PI3 kinase pathway, interleukin signaling pathway and angiogenesis. Interestingly, out of these categories Panther significantly identified the carbohydrate metabolism, other polysaccharide metabolism as well as P53 pathway as those possibly involved in NB progression and metastasis.

Using the DAVID Functional Annotation Tool, the subcellular localization of 82 genes (out of 112) predicted that the majority of the genes code for intracellular proteins. Interestingly, 27 and 29% of these proteins are localized at the cytoplasm and nucleus, respectively. As demonstrated in Figures 2a and 2b, respectively, genes that were localized in the cytoplasm represent 33% of the genes that were more highly expressed in the metastatic variants and only 14% of the genes that were more highly expressed in the local variant. In a similar manner, more genes are localized at the endoplasmic reticulum of the metastatic variants (11%) compared to the local variant (4%).

thumbnail image

Figure 2. Subcellular localization prediction of genes highly expressed in local or in metastatic NB variants. Categorization based on known or predicted subcellular distributions using the pTARGET database of (a) genes highly expressed in the metastatic variants than in the local variant and (b) genes highly expressed in the local variant than in the metastatic variants, identified by DAVID Functional Annotation Tool.

Download figure to PowerPoint

Metastasis associated genes in primary NB tumors

It is reasonable to assume that a biologically meaningful and clinically relevant gene expression profile that mediates metastasis would be uniquely expressed by a subgroup of patients suffering from disease progression and tumor dissemination. To test this assumption, further analysis was performed using gene-expression profiles generated from patients who were enrolled in the German NB Trials NB90-NB2004 who presented divergent clinical courses of the disease.8 Comparing transcriptional profiles of localized Stage 1 (n = 67) and metastasized Stage 4 (n = 67) tumors that were generated using the same oligonucleotide microarray, we detected 666 differentially expressed probes by ANOVA. Following the intersection of these 666 probes with the 112 genes retrieved previously using the metastatic and local variants, 15 probes, representing 14 genes, were revealed to be shared between the NB patient material and the NB variants-generated in nude mice. Two-dimensional hierarchical clustering based on these 15 probes identified 2 main clusters of all 107 patients (Stages 1 and 4). As demonstrated in Figure 3a, 1 cluster represents patients currently considered to be high-risk and the other cluster represents patients currently considered to be low-risk. The high-risk patient's cluster consists mostly of Stage 4 patients who showed progression of disease and/or MYCN amplification.8 In contrast, most of the Stage 1 patients, none with MYCN amplification, were grouped in the low-risk cluster.

thumbnail image

Figure 3. (a) Genes associated with NB metastasis in primary tumors from NB patients. A 2-dimensional hierarchical cluster analysis of 107 NB patients was performed (Stage 1 and Stage 4) using the expression data of 15 shared probes in patient tumors and in the in vivo-generated variants. Rows represent Stage 1 and Stage 4 NB primary tumors; columns represent the 15 probes. Two main clusters were identified. One cluster contains patients currently considered to be at high-risk and the other cluster contains patients currently considered to be at low-risk. Varying expression values are indicated as log10 ratios ranging from −0.5 (blue) to +0.5 (red). The column on the right end of the figure indicates the course of the disease in which red represents patients with progression, relapse or death; white represents event-free survival. (b) Expression of differentially expressed genes at the protein level. (A) MHH and (B) SY5Y local and metastatic variants were utilized for the validation of HK2, DDB2, TK1 and CGNL1 expression at the protein level. HK2, DDB2, TK1 and CGNL1 expression were detected by Western blot analysis. Ponceau staining and ERK2 expression were used as loading control. Depicted are the results of 1 representative blot out of at least 4 independent experiments. The bars (on the right panel of each blot) represent the normalized protein expression calculated in reference to the loading control in the lane as measured by densitometry.

Download figure to PowerPoint

The relative expression of the 14 differentially expressed genes is summarized in Table 3. Six out of the 14 genes demonstrated a similar pattern of the relative expression in the patient-derived tumors and in the variants. However, only 4 (out of 6) genes (HK2, DDB2, TK1 and CGNL1) were identified by the NCBI database. Three genes show higher expression levels and the expression level of 1 gene was lower, both in the metastatic variants and in the tumors from high-risk patients than in the local variant or in the tumors from the low-risk patients.

Table 3. Fourteen shared genes in primary tumors from NB patients and the variants V: Expression in the metastatic variants relative to the local variant, P: Expression in stage 4 primary tumors relative to stage 1 primary tumors from NB patients, [UPWARDS ARROW]: High expression, [DOWNWARDS ARROW]: Low expression
inline image

Validation of differentially expressed genes at the protein level

The local and the metastatic variants of the MHH tumor used for the gene-expression profiles were utilized for the validation of HK2, DDB2, TK1 and CGNL1 expression at the protein level. Although higher expression of DDB2, TK1-dimer and HK2 was demonstrated in the MHH.Lu3 metastatic variant as compared to the local MHH.Ad variant (Fig. 3bA), the expression of CGNL1 in these cells was lower as compared to the MHH.Ad local variant.

SY5Y local and metastatic variants that previously were reported to have low or high metastatic potential in immunodeficient mouse models7 were also utilized in these validation assays. As shown in Figure 3bB, the SY5Y variants exhibit a similar expression pattern of only 2 out of 4 genes (HK2 and CGNL1).


  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Metastasis is the leading cause of death of cancer patients including those with NB. Since the cellular and the molecular events leading to NB metastasis are largely unexplored, we need to broaden our understanding of this process. The goal of the present study was to discover molecular correlates of NB progression.

NB disseminates most frequently to bone marrow (87%) and to bone (66%).1, 37 NB pulmonary metastases are rare at diagnosis and represent a terminal stage.37–40 Accordingly, lung metastasis indicates the presence of biologically more aggressive cells and portends a poor prognosis.

In a previous study, we described a novel in vivo xenograft model for human NB metastasis. This model comprises variants that form local, nonmetastasizing tumors in the adrenal of nude mice being the orthotopic site of NB and variants that metastasize to the lungs. Both types of variants originated from the same human NB tumor, thus ensuring a common genetic background.7 In the present translational study, we used the MHH xenograft model and a specific NB oligonucleotide microarray8 to compare the transcriptomes of the local and metastatic MHH variants. This enabled the identification of molecular correlates associated with NB progression in patients.

One hundred twelve genes were differentially expressed in the local and metastatic NB variants. These differences were at a low fold-change threshold, but statistically significant. We hypothesize that this set of differentially expressed genes is associated with the ability to metastasize to the lung, and we further assume that these genes were responsible for the difference in the in vivo metastatic potential between the local and the metastatic variants of the MHH tumor and for the unique phenotypic properties of these variants in vitro.7 Apparently, metastatic outgrowth does not require extensive alterations in gene expression patterns. These results are in line with the findings of others that metastatic outgrowth is not accompanied by major changes in the gene expression of the tumor.41, 42

GO analysis is a powerful tool, especially with respect to the biological function of a large set of genes.43 To underscore the functional differences between the local and metastatic NB variants as well as to link NB metastasis with essential biological processes and pathways, we applied GO analysis to our 2 sets of genes (genes with higher expression levels in the metastatic variants than in the local variant and genes with higher expression levels in the local variant than in the metastatic ones). This analysis underscored the fact that significantly more biological categories were classified by the genes highly expressed in the metastatic variants as compared to the local one. Interestingly, most of these biological processes are directly or indirectly associated with metabolic processes, for example, carbohydrate metabolism, other polysaccharide metabolism, lipid, fatty acid and steroid metabolism and electron transport. A major number of categories in the metastatic variants were associated with intercellular signaling and pathways. A relatively high number of genes are localized in the cytoplasm and endoplasmic reticulum of the metastatic variants.

Altogether, these data support previous reports that indicate the importance of metabolic and signaling pathways for the induction and maintenance of an activation state required for invasion and metastasis.44

Further analysis has been performed in order to translate the xenograft model system to the clinical setting. We asked if any of the differentially expressed genes is also differentially expressed in Stage 1 and Stage 4 human NB primary tumors. In contrast to the xenograft NB variants that all share the same genetic background, the NB tumors originate from patients of different genetic backgrounds. Fourteen out of 112 differentially expressed genes (12.5%) found in the xenografted cell variants, are also differentially expressed in Stage 1 and Stage 4 tumors of NB patients., The clinical relevance of this gene expression pattern is demonstrated through its ability to discriminate between high- and low-risk patients in the human NB primary tumors that were examined. The 14 genes associated with metastasis can be classified according to the novel classification that discriminates between 3 groups of differentially expressed genes: metastatsis initiation genes, metastasis progression genes and metastasis virulence genes.45 Accordingly, we suggest that genes exhibiting low expression patterns in the metastatic variants and high expression patterns in Stage 4 tumors are classified as “metastasis initiation genes.” These genes provide an advantage in primary tumors and, in doing so, pave the way for the tumor cells to escape into the circulation. Therefore, the expression of these genes is critical in the initial steps of metastasis, and less important in later steps of the process. The second group of genes exhibits a similar pattern of expression in our metastatic cell variants, as well as in Stage 4 NB tumors. These genes are suggested to be “metastasis progression genes.” They fulfill certain rate-limiting functions in primary tumor growth; they are distinguished from oncogenes and can be found within gene-expression signatures that correlate with certain primary tumors with the risk of organ-specific dissemination. We further propose that the 3rd group of genes exhibiting a high expression pattern in the metastatic variants and a low expression pattern in Stage 4 tumors are “metastasis-virulence genes.” They are defined as genes that provide a selective advantage in secondary sites, but not in the primary tumor. Thus, their expression may be low in primary tumors, but are highly important and thus strongly upregulated once the cells reach the metastatic site.

Out of the 14 differentially expressed genes we focused on 4 genes, HK2, TK1, DDB2 and CGNL1, which exhibited a similar pattern of expression in the Stage 4 patient-derived tumors and in the xenografted metastatic variants. The microarray expression pattern of these genes in the variants could be validated at the protein level.

The availability of the local and metastatic NB variants described in a previous publication7 and the profiling of these variants in the present study provide clues as to the molecular biology of NB progression. It would be of interest to find out if the set of 4 “NB metastasis-associated genes” mentioned above (HK2, TK1, DDB2 and CGNL1) are a manifestation of the initial oncogenic transformation event; arose during the development of the primary tumor or during progression to metastasis. It is also unclear whether these “NB metastasis-associated genes” are necessary or sufficient for metastasis formation. Obviously, these genes may interact with various microenvironmental factors and thereby be involved in the regulation of a panel of downstream genes that may also interact with the microenvironment. The 4 “NB metastasis-associated genes” as well as the genes they regulate could, thus control cell behavior by delivering selective pressures and inductive signals to the tumor cells thereby promoting progression and metastasis. Functional studies are required in order to determine whether these 4 genes contribute to the aggressiveness and the metastatic potential of primary NB or just correlate with its progression. It is interesting to note that one of the 4 NB “metastais-associated genes”—hexokinase 2—is functionally involved in the pathogenesis of various forms of cancer exhibiting a “Warburg effect.” As such, it serves as a target for cancer therapy.46 However, to date nothing is known about the association of this enzyme and NB tumorigenesis or metastasis. Much less is known about the functional significance of the other “NB metastasis-associated genes” in cancer.

The present study is characterized by 4 unique features. We constructed an orthotopic metastatic model of NB that mimics the entire metastatic process of such tumors. Second, our local and metastatic variants have the identical genetic background. The differential phenotype of the local and metastatic variants can thus be ascribed solely to differences in gene expression between these variants. Third, we compared gene expression profiles of metastasis and of the respective local tumor rather than of metastasis and the respective parental cell line.14, 47 Such a comparison has high prospects to detect subtle differences in gene expression that may be involved in metastasis formation. Fourth, we translated our results to the clinical setting by intersecting the differentially expressed genes of variants derived from the xenograft model with the patient primary tumors. This association enabled to determine the extent to which NBs in the clinic express the differential gene signature detected in the model and to accentuate clinically relevant metastatic genes.

On the whole, the data summarized in this study emphasize the importance of decoding and understanding the genomic, phenotypic and functional diversity in nonmetastatic versus metastatic NB. The ability to synthesize the knowledge from genome-wide analyses of human tissue, laboratory animal models and classical in vitro systems will enable the development of more powerful and effective NB treatments in the future.


  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information
  • 1
    Brodeur GM, Castleberry RP. Neuroblastoma. In: PizzoPA, PoplackDG, eds. Principles and practice of pediatric oncology, Chapter 29, 3rd edn. Philadelphia, PA: Lippincott-Raven Publishers, 1997. 76183.
  • 2
    Vasudevan SA, Nuchtern JG. Gene profiling of high risk neuroblastoma. World J Surg 2005; 29: 31724.
  • 3
    Brodeur GM. Neuroblastoma: biological insights into a clinical enigma. Nat Rev Cancer 2003; 3: 20316.
  • 4
    Ara T, DeClerck YA. Mechanisms of invasion and metastasis in human neuroblastoma. Cancer Metastasis Rev 2006; 25: 64557.
  • 5
    Fidler IJ. Critical determinants of metastasis. Semin Cancer Biol 2002; 12: 8996.
  • 6
    Gupta GP, Massague J. Cancer metastasis: building a framework. Cell 2006; 127: 67995.
  • 7
    Nevo I, Sagi-Assif O, Edry Botzer L, Amar D, Maman S, Kariv N, Leider-Trejo LE, Savelyeva L, Schwab M, Yron I, Witz IP. Generation and characterization of novel local and metastatic human neuroblastoma variants. Neoplasia 2008; 10: 81627.
  • 8
    Oberthuer A, Berthold F, Warnat P, Hero B, Kahlert Y, Spitz R, Ernestus K, Konig R, Haas S, Eils R, Schwab M, Brors B, et al. Customized oligonucleotide microarray gene expression-based classification of neuroblastoma patients outperforms current clinical risk stratification. J Clin Oncol 2006; 24: 50708.
  • 9
    Pietsch T, Gottert E, Meese E, Blin N, Feickert HJ, Riehm H, Kovacs G. Characterization of a continuous cell line (MHH-NB-11) derived from advanced neuroblastoma. Anticancer Res 1988; 8: 132933.
  • 10
    Khan J, Wei JS, Ringner M, Saal LH, Ladanyi M, Westermann F, Berthold F, Schwab M, Antonescu CR, Peterson C, Meltzer PS. Classification and diagnostic prediction of cancers using gene expression profiling and artificial neural networks. Nat Med 2001; 7: 6739.
  • 11
    Mi H, Lazareva-Ulitsky B, Loo R, Kejariwal A, Vandergriff J, Rabkin S, Guo N, Muruganujan A, Doremieux O, Campbell MJ, Kitano H, Thomas PD. The PANTHER database of protein families, subfamilies, functions and pathways. Nucleic Acids Res 2005; 33: D284D288.
  • 12
    Dennis G,Jr, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC, Lempicki RA. DAVID: Database for Annotation, Visualization, and Integrated Discovery. Genome Biol 2003; 4: P3.
  • 13
    Guda C, Subramaniam S. pTARGET [corrected] a new method for predicting protein subcellular localization in eukaryotes. Bioinformatics 2005; 21: 39639.
  • 14
    Minn AJ, Gupta GP, Siegel PM, Bos PD, Shu W, Giri DD, Viale A, Olshen AB, Gerald WL, Massague J. Genes that mediate breast cancer metastasis to lung. Nature 2005; 436: 51824.
  • 15
    Morikawa K, Walker SM, Jessup JM, Fidler IJ. In vivo selection of highly metastatic cells from surgical specimens of different primary human colon carcinomas implanted into nude mice. Cancer Res 1988; 48: 19438.
  • 16
    Bruns CJ, Harbison MT, Kuniyasu H, Eue I, Fidler IJ. In vivo selection and characterization of metastatic variants from human pancreatic adenocarcinoma by using orthotopic implantation in nude mice. Neoplasia 1999; 1: 5062.
  • 17
    Onn A, Isobe T, Itasaka S, Wu W, O'Reilly MS, Ki Hong W, Fidler IJ, Herbst RS. Development of an orthotopic model to study the biology and therapy of primary human lung cancer in nude mice. Clin Cancer Res 2003; 9: 55329.
  • 18
    Myers JN, Holsinger FC, Jasser SA, Bekele BN, Fidler IJ. An orthotopic nude mouse model of oral tongue squamous cell carcinoma. Clin Cancer Res 2002; 8: 2938.
  • 19
    Thomas PD, Kejariwal A, Guo N, Mi H, Campbell MJ, Muruganujan A, Lazareva-Ulitsky B. Applications for protein sequence-function evolution data: mRNA/protein expression analysis and coding SNP scoring tools. Nucleic Acids Res 2006; 34: W645W650.
  • 20
    Mathupala SP, Ko YH, Pedersen PL. Hexokinase II: cancer's double-edged sword acting as both facilitator and gatekeeper of malignancy when bound to mitochondria. Oncogene 2006; 25: 477786.
  • 21
    Pedersen PL. Warburg, me and Hexokinase 2: multiple discoveries of key molecular events underlying one of cancers' most common phenotypes, the “Warburg Effect”, i.e., elevated glycolysis in the presence of oxygen. J Bioenerg Biomembr 2007; 39: 21122.
  • 22
    Kattan Z, Marchal S, Brunner E, Ramacci C, Leroux A, Merlin JL, Domenjoud L, Dauca M, Becuwe P. Damaged DNA binding protein 2 plays a role in breast cancer cell growth. PLoS One 2008; 3: e2002.
  • 23
    Wittschieben BO, Wood RD. DDB complexities. DNA Repair (Amst) 2003; 2: 10659.
  • 24
    Hannigan BM, Barnett YA, Armstrong DB, McKelvey-Martin VJ, McKenna PG. Thymidine kinases: the enzymes and their clinical usefulness. Cancer Biother 1993; 8: 18997.
  • 25
    Munch-Petersen B, Cloos L, Jensen HK, Tyrsted G. Human thymidine kinase 1. Regulation in normal and malignant cells. Adv Enzyme Regul 1995; 35: 6989.
  • 26
    Citi S, Amorosi A, Franconi F, Giotti A, Zampi G. Cingulin, a specific protein component of tight junctions, is expressed in normal and neoplastic human epithelial tissues. Am J Pathol 1991; 138: 7819.
  • 27
    Ohnishi H, Nakahara T, Furuse K, Sasaki H, Tsukita S, Furuse M. JACOP, a novel plaque protein localizing at the apical junctional complex with sequence similarity to cingulin. J Biol Chem 2004; 279: 4601422.
  • 28
    Li L, Hung AC, Porter AG. Secretogranin II: a key AP-1-regulated protein that mediates neuronal differentiation and protection from nitric oxide-induced apoptosis of neuroblastoma cells. Cell Death Differ 2008; 15: 87988.
  • 29
    Fontan-Gabas L, Oliemuller E, Martinez-Irujo JJ, de Miguel C, Rouzaut A. All-trans-retinoic acid inhibits collapsin response mediator protein-2 transcriptional activity during SH-SY5Y neuroblastoma cell differentiation. FEBS J 2007; 274: 498511.
  • 30
    Kodama Y, Murakumo Y, Ichihara M, Kawai K, Shimono Y, Takahashi M. Induction of CRMP-2 by GDNF and analysis of the CRMP-2 promoter region. Biochem Biophys Res Commun 2004; 320: 10815.
  • 31
    Glass TL, Raabe TD, Garcia DM, Koke JR. Phosphorylated neurofilaments and SNAP-25 in cultured SH-SY5Y neuroblastoma cells. Brain Res 2002; 934: 438.
  • 32
    Oyler GA, Polli JW, Higgins GA, Wilson MC, Billingsley ML. Distribution and expression of SNAP-25 immunoreactivity in rat brain, rat PC-12 cells and human SMS-KCNR neuroblastoma cells. Brain Res Dev Brain Res 1992; 65: 13346.
  • 33
    Schmitt-Graff A, Muller H, Rancso C, Ahnert-Hilger G, John M, Riecken EO, Stein H, Wiedenmann B. [Molecules of regulated secretion are differentiation markers of neuroendocrine tumors]. Verh Dtsch Ges Pathol 1997; 81: 15761.
  • 34
    Wen W, Taylor SS, Meinkoth JL. The expression and intracellular distribution of the heat-stable protein kinase inhibitor is cell cycle regulated. J Biol Chem 1995; 270: 20416.
  • 35
    Oehlrich N, Devitt G, Linnebacher M, Schwitalle Y, Grosskinski S, Stevanovic S, Zoller M. Generation of RAGE-1 and MAGE-9 peptide-specific cytotoxic T-lymphocyte lines for transfer in patients with renal cell carcinoma. Int J Cancer 2005; 117: 25664.
  • 36
    Chang R, Xu X, Li MD. Molecular cloning, mapping and characterization of a novel mouse RING finger gene, Mrf1. Gene 2002; 291: 2419.
  • 37
    Cheung NV, Cohn SL. Neuroblastomaed. Heidelberg, Germany: Springer, 2005.
  • 38
    Cowie F, Corbett R, Pinkerton CR. Lung involvement in neuroblastoma: incidence and characteristics. Med Pediatr Oncol 1997; 28: 42932.
  • 39
    Towbin R, Gruppo RA. Pulmonary metastases in neuroblastoma. AJR Am J Roentgenol 1982; 138: 758.
  • 40
    Kammen BF, Matthay KK, Pacharn P, Gerbing R, Brasch RC, Gooding CA. Pulmonary metastases at diagnosis of neuroblastoma in pediatric patients: CT findings and prognosis. AJR Am J Roentgenol 2001; 176: 7559.
  • 41
    Perou CM, Sorlie T, Eisen MB, van de Rijn M, Jeffrey SS, Rees CA, Pollack JR, Ross DT, Johnsen H, Akslen LA, Fluge O, Pergamenschikov A, et al. Molecular portraits of human breast tumours. Nature 2000; 406: 74752.
  • 42
    Weigelt B, Glas AM, Wessels LF, Witteveen AT, Peterse JL, van't Veer LJ. Gene expression profiles of primary breast tumors maintained in distant metastases. Proc Natl Acad Sci USA 2003; 100: 159015.
  • 43
    Thomas PD, Mi H, Lewis S. Ontology annotation: mapping genomic regions to biological function. Curr Opin Chem Biol 2007; 11: 411.
  • 44
    Kroemer G, Pouyssegur J. Tumor cell metabolism: cancer's Achilles' heel. Cancer Cell 2008; 13: 47282.
  • 45
    Nguyen DX, Massague J. Genetic determinants of cancer metastasis. Nat Rev Genet 2007; 8: 34152.
  • 46
    Mathupala SP, Ko YH, Pedersen PL. Hexokinase-2 bound to mitochondria: cancer's stygian link to the “Warburg Effect” and a pivotal target for effective therapy. Semin Cancer Biol 2009; 19: 1724.
  • 47
    Xu L, Shen SS, Hoshida Y, Subramanian A, Ross K, Brunet JP, Wagner SN, Ramaswamy S, Mesirov JP, Hynes RO. Gene expression changes in an animal melanoma model correlate with aggressiveness of human melanoma metastases. Mol Cancer Res 2008; 6: 7609.
  • 48
    Halachmi E, Witz IP. Differential tumorigenicity of 3T3 cells transformed in vitro with polyoma virus and in vivo selection for high tumorigenicity. Cancer Res 1989; 49: 23839.

Supporting Information

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Additional Supporting Information may be found in the online version of this article.

IJC_24889_sm_suppinfotable.xls46KSupporting Information Table

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.