Background
Neuroblastoma (NB) is a pediatric tumor derived from the sympathoadrenal lineage of neural crest progenitor cells and represents the most common malignancy in early childhood [
1]. DNA and RNA aberrant profiles have been shown to identify mechanisms behind the clinical outcome of NB as the expression of several genes involved in proliferation, differentiation and metastasis that negatively impact on therapy success. Despite recent improvements in survival in randomized trials, nearly 50% of children with high-risk disease is refractory to therapy or suffer a relapse [
2‐
4]. High-risk tumors are characterized by un-differentiated phenotype, age at diagnosis ≥18 months and harbor a very low rate of recurrent somatic mutations in both nuclear and mitochondrial DNA [
5‐
9].
Hypoxia is an important factor in the pathology of many human diseases, including cancer, diabetes, aging, and stroke/ischemia. Low oxygen levels represent an important microenvironment condition to affect the activation status of signaling pathways as drug resistance mechanism. Indeed the increased expression of Hypoxia-Inducible-Factor HIF-1α mRNA (HIF1A) in tumors is relevant to establish resistance to therapeutic approaches as radiotherapy [
10,
11]. We have recently reported that high HIF1A expression may stratify high-risk NB patients with poorer prognosis and low HIF1A expression enhances neuronal differentiation signaling pathways activation and response to differentiating agents [
12]. The identification of factors able to influence the expression levels of HIF1A could allow greater therapeutic success. Recent reports suggest that HIF1-α protein might be degraded in VHL-independent manner following intracellular accumulation of methylglioxal (MGO), a highly reactive α-oxoaldehyde formed as a by-product of glycolisis [
13,
14]. Polymorphisms in glucoxylase I enzyme (GLOI) results in down-regulation of GLOI enzyme that play important role in MGO detoxyfication and favor damage from oxydative stress and the degradation pathway of HIF1A [
15]. Indeed, conditions with increased availability of glucose, such as diabetes or down-regulation of GLOI highlight the importance of mechanisms to disrupt cell response to hypoxia.
Tumor cells respond to repeated oxygen levels fluctuations in tumor microenvironment through epigenetic control. Epigenetic regulatory mechanisms are coordinated at several levels: i) DNA, by (hydroxy) methylation of CpG islands (CGI), ii) RNA, through involvement of regulatory noncoding RNA, and iii) proteins, by activation of epigenetic regulators and posttranslational modificators of histones. Their concerted action in hypoxia drives tumor plasticity through the acquisition of local or global chromatin modifications, which allow the accessibility of hypoxia-responsive elements (HRE) loci or of new active DNA regions at hypoxia inducible factors [
16].
Epigenetic regulation of gene expression by DNA methylation plays a central role in determining tissue specific gene expression and chromosome instability. In cancer, the DNA methylation landscape is very complex: promoter CGIs hypermethylation is associated to inactivation of tumor suppressors as well as the presence of DNA hypomethylation blocks and contiguously hypermethylated CGIs at telomeric regions [
17,
18]. Several studies show HIF1A expression can control DNA hypomethylation status of HRE. Interestingly, more than half of histone demethylase belonged to Jumonji C family genes were up-regulated by hypoxia and four of them (JMJD1A, JMJD2B, JMJD2C, PLU-1) were reported to be direct HIF1A targets and may result in increased HIF-1α binding to the HRE [
19,
20].
Tumor hypoxia acts as a novel regulator of DNA methylation independently of HIF1A activity. High levels of hypoxia metabolites as succinate and fumarate altered the global DNA methylation patterns via significant DNA hypermethylation [
21]. Activity of ten-eleven translocation (TET) enzymes that catalyze DNA demethylation through 5-methylcytosine oxidation depends directly on oxygen shortage. Indeed, TETs activity is reduced by tumor hypoxia in human and mouse cells [
22]. Although HIF1A plays a role in defining DNA methylation status of its targets, its role in the global hypermethylation induced by hypoxia remains to be explored [
23].
To shed light on the molecular mechanisms by which hypoxia reshapes gene expressions of tumors, we have performed an integrated analysis of gene expression and DNA methylation in NB cells upon HIF1A inhibition in normoxia and hypoxia conditions. We found that HIF1A transcription response in hypoxia is driven by epigenetic control of low oxygen levels and can upgrade high-risk tumor features. Interestingly, HIF1A targets expressed in both normoxic and hypoxic areas may provide novel targets to eradicate solid tumors.
Discussion
Increased expression of HIF1A in tumors is relevant to establish resistance to therapy [
10,
11]. Interestingly, we have previously reported that high HIF1A expression may stratify high-risk NB patients with poorer prognosis [
12].
Currently, targeting of hypoxia signaling has limitations in clinics with regard to changeable oxygen concentrations in solid tumor areas and HIF1A direct compounds do not show clinical efficiency. Indeed, the identification of HIF1A target genes and deep insights into the mechanisms of HIF1A driven gene expression may provide novel risk factors to meliorate survival/therapeutic successes in patients with high-risk tumors that lack of precisely genomic causes.
In the present study, we have investigated HIF-1 driven transcription activity in both hypoxic and normoxic conditions in NB cells depleted of HIF1A expression. The analysis of pathways regulated by HIF1A exclusively in normoxic NB cells shows a role of HIF1A in metabolic process necessary for tumor cells viability. Particularly, the global down-regulation of gene expression in absence of HIF1A suggests that NB cells slow down their metabolic activity, thus becoming less proliferating. HIF1A involvement in basic cellular activity, like glycolytic pathways, has been described [
29].
Contrary, in hypoxic cells the absence of HIF1A affects the activation of neuronal differentiation pathways in line with literature data showing that low oxygen in environments causes de-differentiation of NB cells towards an immature and neural-crest-like phenotype [
30]. We have previously highlighted HIF1A involvement in NB neuronal differentiation pathways activation and response to differentiating agents [
12].
Interesting to note, mostly of genes regulated by HIF1A in both normoxic and hypoxic areas belong to MAPK pathways. This pathway is frequently altered in high-risk NB at relapse and at diagnosis and multiple drugs aimed to target MAPK signaling are used in current clinical trials for the treatment of metastatic tumors [
5,
8,
31]. Indeed, HIF1A target genes in both normoxic and hypoxic areas may provide potential targets for a precision therapy. HIF1A is not the unique player to define the whole picture of hypoxia-regulated gene expression. In effect, we report that NB cells adapt to hypoxia by HIF1A-dependend and HIF1A-indipendent driven response. These findings help us to understand how oxygen is sensed at NB cellular levels.
We assume that HIF1A driven transcriptional response in hypoxia is a consequence of the epigenetic control of low oxygen levels at DNA methylation status. We have observed that hypoxia exposure induces a global DNA hypermethylation in NB cells and HIF1A itself might control DNA methylation status. A global DNA hypermethylation has been previously linked to poor NB prognosis as site-specific DNA hypermethylation of tumor suppressor genes to optimize the environment for cancer initiation and progression [
32,
33]. The hypoxia epigenetic controls at the levels of RNA and proteins still remain to be explored.
Despite the stereotype, DNA methylation does not appear to play a major role in gene regulation from 5’CGI promoters of most genes in hypoxia. Indeed, few genes show a correlation between expression and methylation status of close regulatory regions and some correlations were validated in NB samples. Hypoxic gene signatures generated from this correlation analysis are able to stratify NB patients in two risk categories. Although numerous prognostic gene signatures have been developed to classify NB patients, none has been introduced into clinical risk stratification systems [
2,
34,
35]. To overcome these limitations, the establishment of gene signatures that take into account the effects of oxygen levels in tumor bulk more than clinical and genetic markers may be an innovative strategy for NB stratification at diagnosis. Of course, these findings need independent validations.
Conversely, low oxygen levels and HIF1A affect the methylation status of probes located in intragenic and intergenic regions [
36‐
38]. Most probes are located in NB active regulatory regions and the different methylation status correlates to different expression of distant candidate targets associated with NB survival. These genes have been previously associated to therapy resistance and cancer progression and may represent potential markers for NB.
CDC20 is a component of the mammalian cell-cycle mechanism and activates the anaphase-promoting complex (APC); its inhibition may enhance radio sensitivity in nasopharyngeal carcinoma cells [
39].
SNRPE (small nuclear ribonucleoprotein polypeptide E) has oncogenic effects in prostate cancer [
40]. TDP1 (Tyrosyl-DNA Phosphodiesterase 1) is DNA repair enzyme potential therapeutic target for the treatment of colorectal cancer [
41].
FOXM1 (Forkhead Box M1) transcription factor regulates the expression of cell cycle genes and plays an important role in NB tumorigenicity through maintenance of cells undifferentiated state [
42]. Interestingly, FOXM1 overexpression in hypoxia has been already documented in cancer [
43].
DMAP1 (DNA Methyltrasferase 1 Associated Protein 1) contributes to epatocarcinoma malignancy [
41].
YBEY (C21orf57) is a highly conserved metalloprotein not-well characterized in cancer.
High-throughput sequencing-based studies have shown low mutations frequency in coding-portion of NB genome and high recurrence of structural rearrangement. Previous genome-wide association studies revealed that many loci associated with NB susceptibility lie in non-coding regions of the genome [
35,
44‐
46]. Based on these evidences, it is reasonable to expect that recurrent non-coding somatic mutations could have a regulatory effect in NB tumorigenesis. In light of all this, our results further underline the role of non-coding regulatory elements in driving NB tumorigenesis through epigenetic regulation in hypoxia. How epigenetic landscape in hypoxia contributes to transformations and how these alterations complement other acquired somatic mutations need to be elucidated.
One limitation of this study is the use of established cell lines that reflects limited aspects of in vivo tumor microenvironments. It lacks geometrical complexity, cellular components including immune cells and organ-specific stromal cells, and extracellular matrix components. Here, our aim was to establish a HIF1A-based method useful in the investigation of undiscovered mechanisms of neuroblastoma tumorigenesis under hypoxic microenvironments. However, our results still need to be confirmed by functional validation and mechanistic studies which could further improve in vitro cell line models predictive validity.