High Skp2 Expression Characterizes High-Risk Neuroblastomas Independent of MYCN Status

Purpose: Amplified MYCN oncogene defines a subgroup of neuroblastomas with poor outcome. However, a substantial number of MYCN single-copy neuroblastomas exhibits an aggressive phenotype similar to that of MYCN-amplified neuroblastomas even in the absence of high MYCN mRNA and/or protein levels.

Experimental Design: To identify shared molecular mechanisms that mediate the aggressive phenotype in MYCN-amplified and single-copy high-risk neuroblastomas, we defined genetic programs evoked by ectopically expressed MYCN in vitro and analyzed them in high-risk versus low-risk neuroblastoma tumors (n = 49) using cDNA microarrays. Candidate gene expression was validated in a separate cohort of 117 patients using quantitative PCR, and protein expression was analyzed in neuroblastoma tumors by immunoblotting and immunohistochemistry.

Results: We identified a genetic signature characterized by a subset of MYCN/MYC and E2F targets, including Skp2, encoding the F-box protein of the SCF^Skp2 E3-ligase, to be highly expressed in high-risk neuroblastomas independent of amplified MYCN. We validated the findings for Skp2 and analyzed its expression in relation to MYCN and E2F-1 expression in a separate cohort (n = 117) using quantitative PCR. High Skp2 expression proved to be a highly significant marker of dire prognosis independent of both MYCN status and disease stage, on the basis of multivariate analysis of event-free survival (hazard ratio, 3.54; 95% confidence interval, 1.56-8.00; P = 0.002). Skp2 protein expression was inversely correlated with expression of p27, the primary target of the SCF^Skp2 E3-ligase, in neuroblastoma tumors.

Conclusion: Skp2 may have a key role in the progression of neuroblastomas and should make an attractive target for therapeutic approaches.

The embryonal tumor neuroblastoma is the most common solid extracranial tumor in young children and is derived from primitive cells of the sympathetic nervous system. The biological behavior of neuroblastomas is extremely variable. In many patients, neuroblastoma is metastatic at the time of diagnosis and usually undergoes rapid progression with fatal outcome. Neuroblastomas tend to regress spontaneously in a portion of infants younger than age 1 year at diagnosis or to differentiate into benign ganglioneuroma in some older patients (1, 2). Deregulated MYCN is thought to play an important role in tumor progression; amplified MYCN oncogene is strongly associated with advanced-stage disease and poor outcome (3). The oncogenic properties of MYCN overexpression have been shown in various experimental systems in vitro and in vivo (4, 5). However, a considerable number of MYCN single-copy neuroblastomas exhibits an aggressive phenotype similar to that of MYCN-amplified tumors even in the absence of high MYCN mRNA and/or protein levels (6, 7), suggesting other unfavorable molecular abnormalities that could account for the inferior survival.

Recently, DNA microarray technology has been applied to study gene expression profiles in primary neuroblastomas. Patterns of differentially expressed genes among different neuroblastoma subtypes as well as gene expression classifiers have emerged, allowing a better prediction of patient's outcome than established risk markers (8–12). A consistent finding from these studies was that a subset of cell cycle genes are expressed at higher levels in high-risk neuroblastomas independent of the genomic MYCN status. In general, it is believed that elevated expression of cell cycle genes reflects higher proliferation rates of tumor cells from more aggressive subtypes. Elevated expression of cell cycle genes, which are largely regulated by E2F transcription factors, also implies an impaired Rb pathway that fails to control E2F activity. It has been reported that high E2F-1, an E2F member but also a prototypical E2F target gene, is universally found in high-risk neuroblastomas independent of genomic MYCN status (13). Although it is known that MYCN increases E2F activity and cell cycle progression, the precise molecular mechanisms by which MYCN favors G₁-S phase transition are poorly understood. Due to the absence of high MYCN levels in MYCN single-copy high-risk neuroblastomas, it is also not clear whether activation of MYCN-regulated genes plays a significant role in uncontrolled cell cycle progression in this neuroblastoma subtype.

Gene expression profiling of artificial in vitro systems allowing the ectopic expression of oncoproteins have been successfully used to characterize and predict the activity of oncogenic pathways in in vivo tumor models (14). In this study, we sought to define genetic programs evoked by ectopically expressed MYCN in vitro. We hypothesized that by analyzing the activity of these MYCN-related genetic programs in relation to patients' outcome, it should be possible to identify components downstream of MYCN that are involved in neuroblastoma tumor progression independent of genomic MYCN status. Using cDNA microarray analysis of neuroblastoma cells after conditional expression of MYCN in vitro and of 49 primary neuroblastoma tumors, we identified a subset of direct MYCN/MYC and E2F targets that provided information on patients' outcome additionally to genomic MYCN status or MYCN expression. As an alternative mechanism that enhances MYCN target gene activation in MYCN single-copy neuroblastoma tumors, we considered the deregulated activity of a cofactor involved in transcriptional activation by MYC proteins. We found Skp2, encoding the F-box protein of the SCF^Skp2 ubiqitin ligase that is required for the transcriptional activation of MYC target genes (15), highly coexpressed with MYCN/MYC and E2F target genes in high-risk neuroblastoma tumors. We analyzed the expression of Skp2 in relation to MYCN and E2F-1 in a separate cohort of 117 neuroblastoma patients using quantitative PCR (QPCR). Because Skp2 also favors G₁-S (16, 17) as well as G₂-M transition (18) via the degradation of cell cycle regulatory proteins, such as p27 and p21, we investigated the relation of Skp2 and p27 protein levels in neuroblastoma specimens using immunoblotting and immunohistochemistry.

Patients. All patients were enrolled in the German Neuroblastoma Trial and diagnosed between 1991 and 2002. The cohort available for cDNA microarray analysis consisted of 49 patients and for QPCR analysis of 117 patients (Table 1). Tumor samples were collected before any cytoreductive treatment. The only criterion for patient selection was availability of sufficient amounts of tumor material. The composition of the cohorts in terms of tumor stage, MYCN status, and age at diagnosis was in agreement with the composition of an unselected cohort of 1,741 patients diagnosed between 1990 and 2003 in Germany (ref. 19; data not shown). Standard prognostic markers were assessed in the reference laboratories of the German neuroblastoma trial in Köln, Marburg, Heidelberg, Stuttgart, and Zürich. Informed consent was obtained from patients' parents. Patients were treated according to the guidelines established by the German Neuroblastoma Trial, with risk stratification criteria as described elsewhere (20). All samples had a minimal tumor cell content of 65%, as determined by histologic examination.

Microarray experiments. Microarray experiments were done as described previously (10). The reference RNA used in the study for tumor specimens consisted of equal portions of total RNA obtained from seven human cancer cell lines (10). The microarrays used in the studies consisted of 42,578 cDNA clones, representing 25,933 UniGene clusters. Hybridization and washing of the microarrays were done as described (10). The raw data of both tumor samples and cell line experiments were normalized with the variance normalization method (21) as implemented in the R VSN package.¹⁰

The SH-EP-MYCN cell line expressing a MYCN transgene under the control of a tetracycline-repressible element (22) was used to generate gene expression profiles after targeted MYCN expression. Total RNA was collected from SH-EP-MYCN cells after targeted MYCN expression at different experimental conditions to better distinguish between direct MYCN and E2F targets (refs. 22, 23; see also Supplementary Data and Methods). The microarray data are available online (accession nos. E-CVDE-2 and E-CVDE-3).¹¹

Quantitative real-time reverse transcription-PCR. First-strand cDNA was generated from total RNA using the Superscript II First-Strand Synthesis System (Invitrogen) according to the manufacturer's directions. QPCR was done on an ABI PRISM 7700 Sequence Detection System (Applied Biosystems) with SYBR Green chemistry using the standard curve method (user bulletin no. 2, ABI PRISM 7700 SDS). For normalization, the expression level of the target genes was divided by the geometric mean of expression levels of the housekeeping genes hypoxanthine phosphoribosyltransferase 1 (HPRT1) and succinate dehydrogenase complex, subunit A (SDHA), as described previously (24, 25). The geometric mean of HPRT1 and SDHA transcript levels was shown to represent a low variability internal control for investigating differential gene expression in primary neuroblastoma by QPCR (24). Primer sequences were as follows: 5′-TCCGGAGAGGACACCCTG-3′ (MYCN-forward), 5′-GCCTTGGTGTTGGAGGAGG-3′ (MYCN-reverse), 5′-TACAGAAAGAATCTCCAGAAATCAGATC-3′ (Skp2-forward), 5′-GGAAAAATTCCTGAAAGCAGTCA-3′ (Skp2-reverse), 5′-AGCGGCGCATCTATGACATC-3′ (E2F-1-forward), 5′-TCAACCCCTCAAGCCGTC-3′ (E2F-1-reverse), 5′-TGGGAACAAGAGGGCATCTG-3′ (SDHA-forward), 5′-CCACCACTGCATCAAATTCATG-3′ (SDHA-reverse), 5′-TGACACTGGCAAAACAATGCA-3′ (HPRT1-forward), and 5′-GGTCCTTTTCACCAGCAAGCT-3′ (HPRT1-reverse).

Flow cytometric analysis. Native or snap-frozen human neuroblastoma tumor tissues as well as single cells from cell culture experiments were monitored for DNA indices and cell cycle distribution by flow cytometry (fluorescence-activated cell sorting) as described (26). Each histogram represents 30,000 to 100,000 cells for measuring DNA index and cell cycle. Histogram analysis was done with the Multicycle program (Phoenix Flow Systems). Human lymphocyte nuclei from healthy donors were used as internal standard for determination of the diploid cell population. The mean coefficient of variation of diploid lymphocytes was 0.8 to 1.0.

Protein analysis. Protein expression was assessed by immunoblotting using 50 μg of total cell lysates from tumor sections as previously described (27). Blots were probed with antibodies directed against p27 (Dako), Skp2 (Zymed), E2F-1 (BD Bioscience), and β-actin (Sigma) as loading control. Immunohistochemistry was done using the same antibodies on a paraffin multitissue array containing 141 neuroblastomas and evaluated by a pathologist (K.E.). For p27 immunostaining, a biotin-avidin method was applied, as described earlier (28). The sections were boiled in EDTA buffer (pH 9.0) for 20 min and then incubated with the primary antibody p27 at a dilution of 1:25 for 30 min at room temperature. For Skp2 immunostaining, an enhanced biotin-avidin method was used, as described earlier (29). For antigen retrieval, the slides were boiled in an autoclave in EDTA buffer (pH 8.0) for 10 min. The slides were incubated with the primary Skp2 antibody at a dilution of 1:250 overnight at 4°C. The evaluation was done semiquantitatively. The cases with <5% positive tumor cells were scored as negative and tumors with >5% of positive tumor cells were evaluated as positive.

Statistics. To test the association of MYCN in vitro clusters with clinical and genetic data (overall survival, event-free survival, genomic MYCN status, and E2F-1 gene expression), Global Test from the globaltest package of the R software package was used (Supplementary Methods; ref. 30). Because of multiple testing, Global Test P values were adjusted according to Benjamini and Yekutieli (31). Differential gene expression was evaluated for MYCN amplified versus all MYCN single-copy tumors, and also pairwise for all combinations of the following subgroups of MYCN single-copy tumors, stage 1, 2, 3/stage 4s/stage 4, using the Wilcoxon sum rank test. P values were adjusted using the Bonferroni-Holm procedure. Pearson's correlation was used to evaluate correlation among MYCN, E2F-1, and Skp2 expression. To identify a model describing the relationship between survival and Skp2 expression, the functional form of the relationship was tested by maximally selected log-rank statistics as previously described (25, 32). The resulting model was applied in further survival analyses. Multivariate Cox regression was used to investigate the prognostic power of candidate gene expression adjusting for established prognostic variables. To correct for overestimation of the hazard ratio estimate due to selection of the functional form of candidate gene expression variables, shrinkage of the variable estimate was applied (33). The Cox proportional hazards regression were done with the use of the design and survival package of the R software environment version 2.3.1.¹⁰ All reported P values are two sided.

High expression of MYCN/MYC- and E2F-regulated genes identifies high-risk neuroblastomas independent of genomic MYCN status. To identify genetic programs associated with targeted MYCN expression in vitro, we made use of SH-EP-MYCN cells that express a MYCN transgene under the control of a tetracycline-repressible element in the human neuroblastoma cell line SH-EP (22). We collected total RNA from time course experiments after targeted MYCN expression at different experimental conditions and generated gene expression profiles using genome-wide cDNA microarrays. To better distinguish MYCN from E2F target genes, we inhibited E2F activity in a subset of experiments using low serum and low-dose doxorubicin, which left MYCN target gene activation largely unaffected (Supplementary Data and Methods). We combined these time course experiments and used self-organizing maps to capture the predominant patterns of gene expression. We applied a variance filter to select gene clusters that exhibited the most prominent transcriptional changes associated with targeted MYCN expression. By this procedure, 33 clusters were identified (Supplementary Fig. S1). As adjacent clusters within the cluster map showed similar gene expression profiles, the 33 clusters were further grouped into five cluster groups (A, B, C, D, and E).

We evaluated the expression of each of the 33 selected clusters in a set of 49 neuroblastoma tumors. For this, we used the Global Test as proposed by Goeman et al. (30) and tested the association of each of the 33 in vitro clusters with overall survival and event-free survival directly, without the intermediary of single gene testing. We also tested the association of each of the 33 in vitro clusters with genomic MYCN status and E2F-1 expression in neuroblastoma tumors. Most in vitro clusters were significantly associated with poor outcome [overall survival (29 of 33, P < 0.05) and event-free survival (30 of 33, P < 0.05)] after adjusting the P values for multiple testing (Supplementary Table S1). Direct targets of MYCN/MYC transcription factors were enriched in group A clusters, as revealed by comparisons with the MYC target gene database¹²

and the literature (34–36). Figure 1 shows a two-way hierarchical cluster analysis (A) and Global Test gene plots (B) using genes from a representative group A cluster, cluster 1, in which MYCN was arranged by the self-organizing map analysis. High MYCN expression showed a lower association with poor outcome (only >2 SD above the reference line 0) than other cluster 1 genes, including Skp2, FLJ20485, MTHFD2, RG9MTD1, ILF2, SLC25A4, GMPS, and DKC1 (>4 SD above the reference line; Fig. 1B, left). Intriguingly, when the model was adjusted for genomic MYCN status, most of these genes remained significantly associated with poor outcome, whereas MYCN expression lost its association with poor outcome (Fig. 1B, right). DKC1, SLC25A4, and MTHFD2 are directly regulated by MYC transcription factors, including c-MYC and/or MYCN (34–36). Additional genes directly regulated by MYC transcription factors were found in other group A clusters, such as SFRS1, LARS, and ZNF146 (34–36), the high expression of which was also significantly associated with poor outcome independent of genomic MYCN status (data not shown). Group B clusters represent a genetic program predominantly comprising bona fide targets of E2Fs, such as MAD2L1, CDC2, CCNA2, CCNB1/2, PTTG1, and CKS2 (13, 37, 38), from cluster 176 (Fig. 1C). In line with this, group B clusters were highly associated with E2F-1 in neuroblastoma tumors (adjusted P < 0.001, Supplementary Table S1). High expression of group B cluster genes was also significantly associated with poor outcome independent of genomic MYCN status (Fig. 1D). Together, this revealed that a subset of MYCN and E2F targets are up-regulated in high-risk neuroblastoma tumors independent of genomic MYCN status and MYCN expression in MYCN single-copy tumors.

Expression of MYCN/MYC and E2F target genes does not correlate with proliferation activity in neuroblastoma tumors. To answer the question whether high expression of MYCN/MYC and E2F target genes simply reflects high proliferation activity, fluorescence-activated cell sorting cell cycle analyses of 4′,6-diamidino-2-phenylindole–stained nuclei from the respective neuroblastoma specimens were done (31 of 49 specimens were available for this analysis). Intriguingly, tumor cell proliferation indices (%S phase + %G₂-M phase) were not higher in neuroblastomas with elevated expression of MYCN/MYC (Fig. 1A) and E2F targets (Fig. 1C). Further, favorable neuroblastomas with high proliferation indices (NB17, NB7, NB26) had low expression levels of MYCN/MYC and E2F targets. This is in line with fluorescence-activated cell sorting cell cycle analyses in a large series of neuroblastoma specimens (n = 143) showing that high tumor cell proliferation indices were not associated with poor outcome in neuroblastoma tumors.¹³

Accordingly, high levels of a subset of MYCN/MYC– and E2F-regulated genes in high-risk neuroblastoma tumors were not merely due to increased proliferation of the corresponding tumor cells.

Skp2 expression correlates positively with E2F-1 and negatively with p27 in neuroblastoma tumors. Due to its central role in controlling both MYC and E2F functions, we further evaluated Skp2 in relation to MYCN and E2F-1 expression in a panel of 117 neuroblastomas by using specific quantitative reverse transcription-PCR. Relative expression values for Skp2, MYCN, and E2F-1 expression were normalized to the geometric mean of HPRT1 and SDHA expression values (24). Skp2 mRNA levels correlated tightly with E2F-1 mRNA levels as shown by QPCR [Pearson's correlation coefficient (PC) = 0.70; 95% confidence interval (95% CI), 0.60-0.79] and microarray analysis (PC = 0.84; 95% CI, 0.74-0.91). Even in the subset of MYCN single-copy tumors, Skp2 mRNA levels correlated tightly with E2F-1 mRNA levels [PC = 0.69; 95% CI, 0.57-0.78 (QPCR); PC = 0.8; 95% CI, 0.64-0.89 (microarray)]. This is in line with the proposed regulation of Skp2 by Rb-E2F (39). Expression of Skp2 was significantly higher in MYCN-amplified tumors than in MYCN single-copy tumors (P < 0.001; Fig. 2). However, Skp2 expression was poorly associated with MYCN expression in MYCN single-copy tumors [PC = 0.24; 95% CI, 0.03-0.44 (QPCR), PC = 0.21; 95% CI, −0.12-0.50 (microarray)]. In the subset of MYCN single-copy tumors, we found that Skp2 mRNA levels were significantly higher in stage 4 (P = 0.009) and 4s tumors (P = 0.03) than in localized (stage 1, 2, 3) tumors. Remarkably, whereas Skp2 mRNA levels were not different in stage IVs versus stage IV tumors (P = 0.92), MYCN mRNA levels were significantly higher in stage IVs than in stage IV MYCN single-copy tumors (P = 0.008). The primary target of the SCF^Skp2 ubiquitin ligase during cell cycle progression is the p27 protein, the expression of which is reduced in high-risk neuroblastomas (40). We investigated p27 and Skp2 protein expression using a paraffin multitissue array consisting of 141 neuroblastoma tumor specimens. Probably due to antigen masking, the detection limit for both proteins was high. We did not detect either of both proteins in 113 tumor samples. In 28 samples, we detected a mutually exclusive expression of Skp2 [n = 16; amplified = 7, stage 4 single-copy (sc) = 3, stage 3 sc = 2, stage 1/2 sc = 4] and p27 (n = 12; amplified = 1, stage 4 sc = 2, stage 3 sc = 2, stage 1/2/4s sc = 7) protein. In line with our QPCR analysis, strongest Skp2 protein staining was observed in MYCN-amplified tumors (Fig. 3A). We also did more sensitive immunoblotting for Skp2 and p27 in neuroblastoma tumors. Detection of Skp2 protein was associated with reduced p27 protein levels in neuroblastoma tumors (Fig. 3B).

Skp2 expression correlates with poor outcome in neuroblastoma patients independent of genomic MYCN status. The search for a model describing the relationship between survival and Skp2 gene expression using maximally selected log-rank statistics identified a cut point model to be most suitable: A Skp2 QPCR cutoff value of 25.7 separated two patient subgroups with significantly different outcome [maximally selected log-rank statistics, P < 0.001 (event-free survival), P = 0.002 (overall survival)]. Applying this Skp2 expression cutoff to Kaplan-Meier survival curve estimation revealed increased recurrence rate and decreased survival probability for patients with tumor expressing high levels of Skp2 (Fig. 4). As Skp2 is induced by targeted MYCN expression in vitro and its expression is significantly correlated with amplified MYCN, we asked whether Skp2 expression adds prognostic information independent of MYCN status. High Skp2 was associated with increased recurrence rate (P = 0.001) and decreased survival probability (P = 0.02) also in MYCN single-copy neuroblastomas (Fig. 4). To determine whether Skp2 expression provides additional predictive power over other established prognostic markers, multivariate survival analysis was done. The risk factors included in the model were all associated with decreased event-free survival in univariate analysis: amplified MYCN (P = 0.005), stage 4 (P < 0.001), and age at diagnosis >1.5 years (P < 0.001). Cox proportional hazards models were built based on 117 patients. Together with age at diagnosis >1.5 years, high Skp2 expression was selected as a statistically significant predictor for decreased event-free survival (Table 2). To further illustrate the prognostic independence of Skp2 expression from amplified MYCN, multivariate survival analysis was done in the subgroup of 101 tumors with single-copy MYCN. Skp2 expression, together with age at diagnosis >1.5, emerged as independent factor even in this cohort (Table 3).

In this study, we found that transcript profiling of a neuroblastoma in vitro system that allows conditional expression of MYCN delineates MYCN downstream genetic programs that are associated with patients' outcome independent of genomic MYCN status. The identified MYCN downstream genetic programs consisted of direct targets of MYC transcription factors (group A clusters, e.g., cluster 1), and of direct E2F targets (group B clusters, e.g., cluster 176). For MYCN single-copy tumors with poor outcome, high expression of MYCN/MYC target genes was unexpected, because MYCN mRNA and protein levels are usually lower in these tumors than in favorable MYCN single-copy tumors, as shown by results from independent studies (6, 7) and those presented here. In addition, expression of other MYC transcription factors is not associated with poor outcome in MYCN single-copy tumors.¹³

As an alternative mechanism for MYCN target gene activation in MYCN single-copy neuroblastoma tumors, we considered the deregulated activity of a cofactor involved in transcriptional activation by MYC proteins, namely Skp2, the F-box protein of the SCF^Skp2 E3 ligase.

High Skp2 mRNA levels strongly correlated with amplified MYCN and poor outcome as measured by cDNA microarray (n = 49) and QPCR (n = 117) in two primary neuroblastoma tumor cohorts. Notably, our multivariate survival analysis for event-free survival identified “Skp2 expression” as a predictor variable that was independent not only of the variable “amplified MYCN” but also of “stage” (stage 4 versus stage 1, 2, 2, and 4s). The prognostic independence of Skp2 expression from amplified MYCN was further shown by multivariate survival analysis within the subgroup of MYCN single-copy neuroblastomas (n = 101). This indicates that high Skp2 expression is a shared feature of MYCN-amplified and single-copy high-risk neuroblastomas.

We found that high Skp2 correlated with low p27 protein levels, the primary target of the SCF^Skp2 E3-ligase, in primary neuroblastoma tumors. This is in line with results from other cancer types that showed a correlation of high Skp2 with low p27 protein levels and advanced grade of malignancy (41). The association of low p27 protein levels with poor outcome in primary neuroblastoma tumors was described earlier (40); however, a link to deregulated Skp2 was not yet established. Our microarray and QPCR data further revealed the transcriptional changes associated with high Skp2 expression in neuroblastoma tumors: Expression of Skp2 is tightly correlated with E2F-1 and E2F activity. This is in line with previous in vitro results that established Skp2 as a direct E2F-1 target (42). Additionally, Skp2 also strongly influences E2F functions. First, Skp2 favors the degradation of p27 and p21 during G₁-S phase transition leading to cyclin-dependent kinase activation that mediates the release of E2F family proteins via Rb phosphorylation (17, 41). Second, Skp2 is required for G₂-M progression via mediating the degradation of p27, newly synthesized during S-phase progression (18). In line with these proposed Skp2 functions, we found that high Skp2 is associated with high expression of E2F targets encoding G₂-M activities [MAD2L1, CDC2, CCNB1/2, PTTG1 (cluster 176)] in high-risk neuroblastomas that often fail to express p27. Importantly, high expression of these E2F targets did not merely reflect elevated proliferation of the respective tumor cells, because expression of these E2F target genes did not correlate with tumor cell proliferation as determined by fluorescence-activated cell sorting cell cycle analysis.

It has been described that Skp2 also acts as cofactor for c-MYC–dependent transcriptional activation independent of its proteolytic function (15). Thus far, it is unclear whether high Skp2 also influences MYCN activity in neuroblastoma cells. We observed that MYCN/MYC targets genes (group A clusters, e.g., cluster 1), including DKC1, SLC25A4, MTHFD2, strongly correlated with Skp2 in neuroblastoma tumors. It is tempting to speculate that high Skp2 activity does not only favor transcriptional activation in MYCN-amplified neuroblastomas, but also in high-risk MYCN single-copy neuroblastomas even in the absence of high levels of MYCN or other MYC proteins.

An unexpected finding in this study was that mRNA levels of Skp2 and E2F-1 were not significantly different in MYCN single-copy stage 4 versus stage 4s neuroblastomas, two neuroblastoma subtypes with contrasting clinical outcome. It was observed that p27 protein levels are significantly lower in stage 4 than in stage 4s MYCN single-copy tumors (40). This argues in favor of a higher activity of the SCF^Skp2 E3-ligase in stage 4 than in stage 4s. Therefore, we suggest that Skp2 activity is differentially controlled via transcriptional independent mechanisms in these tumors. Skp2 activity is strongly inhibited via a direct interaction with the Rb protein (43). Intriguingly, this Rb function is independent of its E2F controlling function. Future studies should address the regulation of Skp2 in different neuroblastoma subtypes at the protein level.

In summary, our data suggest that Skp2 is an important player in the development and progression of neuroblastomas, and may consequently be an attractive target for therapeutic approaches including proteasome inhibitors and/or differentiation-inducing agents that suppress SCF^Skp2 E3 ligase activity.

We thank Steffen Bannert, Marc Matuszewska, Jochen Kreth, Jana Faut, Heike Düren, and Yvonne Kahlert for technical assistance; Frank Berthold, Barabara Hero, André Oberthür, and the German neuroblastoma Study Group for providing neuroblastoma tumor samples and clinical data; and Rüdiger Spitz for providing 1p fluorescence in situ hybridization data.