Experiments conducted in this study provide clear evidence that hypoxia and its mimetics induce the expression of NDRG1
gene at both the RNA and protein level (Figs. , , and ). It is hypothesized that nickel induces this gene by creating hypoxia-like conditions in cells. Support for this hypothesis came with the discovery of the first molecular oxygen sensors in mammalian cells, namely proline and asparagine hydroxylase enzymes which regulate the oxygen-dependent post-translational modification of HIF-1α protein and thereby change its stability and transcriptional activity. Prolyl hydroxylase PHD hydroxylates HIF-1α at residue Pro564
in the presence of oxygen which creates a signal for pVHL to bind it, causing consecutive ubiquitination and proteasomal degradation of HIF-1α protein [16
]. Likewise, in the presence of oxygen asparaginyl hydroxylase enzyme FIH-1 (factor inhibiting HIF-1) hydroxylates C-TAD domain of HIF-1α which in turn prevents it binding to coactivator p300/CBP and limits transactivation ability [18
]. Under hypoxic conditions these modifications of HIF-1α by above mentioned enzymes do not occur, and the transcription of hypoxia-responsive genes are promoted [22
The hypoxia-responsive element (HRE) is a specific five nucleotide-HIF binding-DNA sequence (5'-RCTCG-3') which is common in all hypoxia-responsive genes [25
gene has three HIF-1 binding sites in its non-coding sequence, one in its promoter and the other two in the 3' untranslated region [27
]. It is known that a HIF-1 binding site in the 3' region of the erythropoietin gene regulates the transcription of this hypoxia-responsive gene [28
]. Conceivably, NDRG1
is likely to be regulated by HIF-1 through the binding sites in its untranslated sequences.
HIF-1 modifier enzymes, PHD and FIH-1, both have non-heme iron centres (29), and transition metals nickel and cobalt can interact with these centres, subsequently inhibiting the enzymes (30). By their effects on HIF-1 modifying enzymes, nickel and cobalt have the capacity of creating constitutive hypoxia-like conditions in cells [31
Our hypothesis relating nickel, oxygen-sensing, hypoxia, HIF-1, pVHL, and NDRG1 expression may be elaborated as follows: when we expose cells to nickel, internalized nickel inhibits PHD enzyme interacting with its iron centre. This prevents the hydroxylation of proline residue in the ODD domain of HIF-1α and subsequent pVHL binding, rescuing HIF-1α from proteasomal degradation (Fig. , upper part). Rescued and accumulated α subunits of HIF-1 form stable heterodimers with β subunits and translocate to the nucleus, and HIF-1αβ heterodimers bind to hypoxia responsive elements (HRE) of the NDRG1 gene and promote the transcription of the gene. Nickel also inhibits the FIH-1 enzyme and subsequent hydroxylation of C-TAD domain of HIF-1α, and this in turn results in the recruitment of coactivators of HIF-1 to the NDRG1 gene regulatory sequences thereby further stimulating the expression of the gene (Fig. , lower part). However, since the experiments to show the specific interaction between metals and HIF-1 modifying enzymes are yet to be executed, this aspect of the hypothesis remains unproven.
The illustration of the hypothetical mechanism by which nickel and cobalt upregulate the expression of the NDRG1 gene
Ndrg1 protein outlasts HIF-1 after hypoxia. Despite being a major regulator of hypoxia response, HIF-1 transcription factor is a very unstable protein which is rapidly degraded under normoxic conditions; the half-life of HIF-1α in post-hypoxic cells is less than 5 minutes. On the other hand our results showed that Ndrg1 protein levels remain high at least 16 hours after returning to normoxic conditions (Fig. ). Similar results have been reported by Lachat et al. (34) who showed that it took 48 hours for Ndrg1 levels to return to pre-anoxic levels after the cessation of hypoxia. Their experiment carried out in colon carcinoma cells (SW480) supports our results, indicating that the high stability of Ndrg1 protein is not cell specific.
Our study of relationship between HIF-1 and NDRG1 expression indicated that the induction of the gene was primarily dependent on this transcription factor (Fig. ). Neither RNA nor the protein product of NDRG1 gene was induced in HIF-1α-/- cells upon short-term exposures to hypoxia. In the long-term hypoxia experiment we detected some amount of Ndrg1 protein in HIF-1α-/- cells starting from the second day, but the levels of protein accumulation on the second and third days were considerably higher in HIF-1α+/+cells than those in HIF-1α-/- cells. However, these results indicate that in chronic hypoxic conditions such as cancer, other factors additional to HIF-1 could be involved in the regulation of NDRG1 gene expression. Several HIF-1 independent pathways have been described to date as being effective under hypoxic conditions (35–40).
Our work in several human tissues showed that in the majority of these organs Ndrg1 protein was differentially overexpressed in cancers compared to normal tissues (Fig. , Table ). Normal tissue samples of certain organs (lung and brain) were almost completely free of Ndrg1 expression, whereas these samples showed HIF-1 protein expression to some extent. In some cases (especially glioblastoma of brain) the expression of Ndrg1 coincided with HIF-1 protein, indicating that induction of HIF-1α by hypoxia probably resulted in Ndrg1 accumulation in these cancers. In most of the other cases though, diffuse and strong Ndrg1 expression did not coincide with HIF-1 protein expression. These differences in the detection of two proteins may be explained by (i) considerably higher stability of Ndrg1 protein compared to that of HIF-1, and (ii) reflection of HIF-1 independent hypoxia response by Ndrg1. With these features, Ndrg1 has the capacity of reflecting tumour hypoxia in a broader spectrum than does HIF-1 and could be considered as a better signature for hypoxic tumour cells than HIF-1. Therefore, despite the proposal of HIF-1 as a tumour marker [41
], we present its down-stream product Ndrg1 as a stronger candidate of cancer marker especially for certain tissues (lung, brain, and skin).
Several normal tissue samples showed some Ndrg1 expression albeit at lower levels than in cancer samples of similar tissues. In their comprehensive study, Lachat et al. (34) showed the expression of Ndrg1 protein in normal human tissues, reporting also the intensities and sub-cellular localizations of the stainings. We observed similar staining patterns in several tissues; more emphasized stainings in the glandular, acinar, ductal, and tubular cells of normal breast, prostate, colon, and kidney tissues. We also share the observation that Ndrg1 exist in all three locations of the cells-cytoplasm, nucleus, and membrane. But, with the exception of colon mucosa, when we stained the cancerous tissues of above mentioned organs, the staining was more intense. However, since the differential expression of Ndrg1 between normal and cancer tissues of lung, brain, and skin was much starker (Table ), we propose Ndrg1 be initially tried as a marker for these tissues. For the reasons that are unknown, Ndrg1 was expressed at lower levels in colon cancer than it was in normal colon. Similar results have also been reported by others [42
]. This could be due to fact that colon epithelium is a dynamic structure being continuously renewed. Studies addressing the mechanism of Ndrg1 down-regulation in colon cancers will shed more light on the function of Ndrg1 protein and its relation to cancer development.
Another common finding between our study and Lachat et al's (34) is no expression of Ndrg1 in normal brain and lung epithelium. Lachat et al. (34) showed at the transcriptional level that normal brain and lung express NDRG1, in fact these and many other tissues expressing NDRG1 mRNA did not contain detectable levels of the protein product of this gene. This could be due to degradation of the mRNA under normal conditions. Stabilizing the mRNAs of hypoxia-responsive genes is one way cells promote the expression of these genes under hypoxic conditions (38, 39). More studies are needed to resolve how Ndrg1 levels are managed in normal cells, during hypoxia, and as well as in cancer cells.
Masuda et al. [40
] report the down-regulation of NDRG1
gene by VHL tumour suppressor protein (pVHL). They state that no hypoxia-responsive element exists on the 5' flanking sequence of the gene and thus underplay the role of HIF-1 in the regulation of NDRG1
. As mentioned previously, NDRG-1
gene has three HIF-1 binding sites, one in its promoter and two in the 3' untranslated region. Therefore, the down-regulation of NDRG1
gene by pVHL is likely to be mediated through the HIF-1 pathway. The observation of NDRG1
being down-regulated by a major tumour suppressor further supports our observation that it is up-regulated in several cancer tissues and could be used as a marker.
Hypoxia-responsive pathway (HRP) allows tumour cells to overcome harsh microenvironment conditions associated with tumour growth. The protein products of induced by this pathway (e.g. EPO, VEGF, several glycolytic enzymes) allow clones of tumour cells to gain growth advantage under unfavourable conditions, and this concept is pivotal in switching to a more malignant phenotype. Although the exact functions of Ndrg1 are still unknown, as another effector of HRP, it is also likely to help tumour cells establish themselves. Therefore, the use of drugs that specifically disrupt the functions of Ndrg1 protein may provide new cancer therapies. It is thus of interest to investigate the effects of the elimination of this protein on the cancer cell survival and proliferation. However, first Ndrg1 expression in normal hypoxic tissue (such as infarct tissues) should be determined to show the cancer specificity of the protein. Second, the potential side effects of its elimination should be assessed since several normal tissues express Ndrg1 ubiquitously.
Hypoxia is also an important determinant for the success of chemotherapy and radiotherapy [43
]. Masuda et al. [40
] even argue that Ndrg1 could be involved in limiting sensitivity to anti-cancer drugs. However hypoxia is an equally likely limiting factor in anti-cancer therapy, and Ndrg1 may be simply the signature of the hypoxic state.