Hypoxia-induced phenotypic modulation of human neuroblastoma cells

Neuroblastoma is a childhood tumour derived from cells of the sympathetic nervous system, which are arrested at low differentiation stages. Low differentiation stage and high tumour stage correlate to poor outcome. In earlier studies we have found that hypoxia induces dedifferentiation of neuroblastoma cells, which implies that hypoxia evokes a more aggressive phenotype. Here we employ microarray analysis to investigate the hypoxic effects, in neuroblastoma cells, on the expression of a larger set of genes. Genes involved in survival and treatment resistance were upregulated, which support the concept of an aggressive hypoxic phenotype. The microarray results further strengthened the concept of hypoxia-induced dedifferentiation of neuroblastoma cells. Hypoxia-treated neuroblastoma cells were reoxygenated to determine the persistence of the hypoxic phenotype. Based on neuronal- and neural crest marker gene expression analysis, we conclude that the hypoxic phenotype persisted for at least 24 h upon reoxygenation. Hence, there was no selection for dedifferentiated cells; instead the hypoxic phenotype appears adaptable and dynamically reversible.

Major transcription factors that act in response to hypoxia are the hypoxia-inducible factors HIF-1a and HIF-2a. Intriguingly, HIF-2a, but not HIF-1a, was detected adjacent to blood vessels in neuroblastoma specimen, indicating a role for HIF-2a at more physiological oxygen levels. Analysis of HIF-protein levels in neuroblastoma cells exposed to hypoxia (1% O2) or 5% O2 (mimicking a more physiological oxygen level) revealed that HIF-1a is primarily and only transiently induced at 1% O2. HIF-2a, on the other hand, is induced at both 1 and 5% O2 and its protein levels increase over time. In a microarray analysis we extracted a set of genes with regulation patterns similar to that of the HIF-protein patterns seen in neuroblastoma cells grown at either 1 or 5% O2. We propose that the differences in HIF target gene utilization are dependent on time and oxygen conditions, rather than on target gene specificity. According to that concept, we propose that HIF-1a primarily drives gene transcription at acute hypoxia, while HIF-2a is active at prolonged hypoxia and at 5% O2, conclusions supported by HIF-1a and HIF-2a siRNA analyses and expression of the HIF-driven genes VEGF and DEC1/BHLB2.

Evaluation of a clinical neuroblastoma material showed correlation between HIF-2a immunostaining and poor patient outcome. Our results further implicate HIF-2a as a possible individual prognostic marker for neuroblastoma patients. Tumour growth of neuroblastoma cells knocked down for HIF-2a in nude mice was slower than neuroblastoma cells transfected with a scramble siRNA sequence. These observations support a role for HIF-2a in neuroblastoma progression. Immunostaining of neuroblastoma specimen revealed co-localization of VEGF and HIF-2a, suggesting a putative mechanism for neuroblastoma tumour growth due to HIF-2a-induced VEGF expression. Our results implicate an oncogenic role for HIF-2a in neuroblastoma aggressiveness, which might be exploited in the treatment of aggressive neuroblastomas.