This article is mentioned in:

Introduction

Tumor marker genes earn their functional importance from the fact that they play an essential role during disease development. Thus, their detection is fundamental for outcome monitoring and targeted treatment. Different types of tumor marker genes have been proposed including prognostic (1), diagnostic (2), predictive (3), pharmacodynamic (4) and recurrence (5) markers. In different tumor types, reduced oxygen supply to the tumor microenvironment, namely hypoxia, plays an important role during disease development to more severe stages (6). Hypoxia-inducible factor-1α (HIF-1α) is known to play an essential role in the regulation of genes induced or upregulated by the hypoxic tumor microenvironment (7,8). The cytoplasmic protein N-myc downstream-regulated gene 1 (NDRG1), also known as CAP43, non-race specific disease specific protein 1 (NDR1), differentiation-related gene 1 protein (Drg-1), nickel-specific induction protein (Cap43), protein regulated by OXYgen-1 (PROXY1) and NMSL protein, belong to the α/β-hydrolase superfamily (9). NDRG1 is involved in a diversity of cellular characteristics such as specific stress responses, hormone responses, cell growth and differentiation. NDRG1 gene mutations cause the neuropathy Charcot-Marie-Tooth disease type 4D (10). Its expression may also possess the potential to be used as a prognostic factor for some types of cancer (11). Cell differentiation signals upregulate NDRG1 and result in metastasis suppression (9). However, observations regarding the role of NDRG1 in human cancer development are controversial. In hepatocellular carcinoma (HCC) upregulation of NDRG1 has been shown to correlate with tumor aggressiveness and patient survival (11); in colon cancer cell lines NDRG1 downregulation was observed and its overexpression led to induction of cell differentiation (12). Moreover, a low NDRG1 protein expression level was found to correlate with unfavorable patient prognosis in glioma (13), colorectal cancer (14), esophageal squamous cell carcinoma (15), pancreatic ductal adenocarcinoma (16), prostate (17,18) and breast cancer (19).

Under hypoxia, accumulation of HIF-1α and its translocation to the nucleus along with the co-activator CBP/p300 results in its binding to hypoxia-responsive elements (HREs) of target gene promoters and subsequent regulation of their transcription (22,23). Studies on the NDRG1 promoter identified HREs in the upstream promoter region and suggest regulation of NDRG1 by HIF-1α (24). However, these studies have revealed controversial results for the manner of NDRG1 regulation by HIF-1α under hypoxia. Various of these studies reported a marked upregulation of the NDRG1 expression level under hypoxic conditions, whereas others have shown downregulated NDRG1 under the same conditions (8,13–15,18,19,25). Consequently, the regulation of NDRG1 under hypoxia is highly complex and its cellular function is still a matter of debate. However, it is established that besides HIF-1α several other transcription factors involved in tumorigenesis are implicated in its regulation (8,20,21). Thus, it is important to clearly define the role of NDRG1 during carcinogenesis as well as the regulatory events steering this process.

It has been shown that different signal transduction pathways are modulated and controlled by different factors such as cytokines and growth factors or environmental conditions such as hypoxia for example (26). Transforming growth factor-β1 (TGF-β1) is one such signaling protein. Pathways involving Smad family members are also important regulators (27). Such proteins are crucial for the pathophysiological regulation and involved in the development of various diseases including fibrosis (28) and cirrhosis, leading to HCCs (29). Particularly, one member of the Smad family, Smad7, abrogates the TGF-β1 signaling pathway by initiating a negative feedback loop (30). Such a regulation may be involved in the process of cancer manifestation as seen for HCC (31). Smad7 is regulated by different pathways (32). One of these involves the transcription factor YB-1, a member of the cold shock family of proteins (30).

Specific transcription factor 1 (SP1) is another protein involved in the regulation of many cancer-related genes (33). In addition, it is involved in the active regulation of vascular endothelial growth factor (VEGF) which is another HIF-1α regulated hypoxia gene (34). Since the HIF-1α promoter contains SP1-specific transcription elements (35), SP1 may also be an important factor in NDRG1 regulation.

The transcription factor CCAAT enhancer binding protein α (CEBPα), which belongs to the BLZIP family (36), plays an important role in governing metabolic processes relevant for disease development including cancer (37). Although HIF-1α is well recognized as a main regulator during hypoxia (38), the role of other transcription factors during hypoxic conditions in the tumor microenvironment have not yet been clearly established. Therefore, we aimed to examine NDRG1 protein and mRNA expression under different oxygenation conditions in human glioblastoma multiforme (GBM) cell lines to gain further insight into its biological role. In addition, the expression levels of potential NDRG1 gene-regulating transcription factors HIF-1α, SP1, CEBPα Smad7 and YB-1 were investigated. At least to a certain extent, we planned to address, whether these factors may have the potential to serve as tumor markers.

Materials and methods

Cell lines, cell culture and hypoxia treatment

The human malignant glioma cell lines U87-MG and U373 were originally purchased from the American Type Culture Collection (ATCC; Rockville, MD, USA). GaMG is an established cell line from a patient with GBM (Gade Institute, University Bergen, Bergen, Norway) (39). Cell lines were grown on glass Petri dishes in Dulbecco's modified Eagle's medium (DMEM), supplemented with 10% fetal bovine serum (FBS), penicillin (100 IU/ml), streptomycin (100 µg/ml) and 2 mM L-glutamine in a humidified incubator with 5% CO2 at 37°C. For hypoxia treatment, the cells were exposed to 0.1% O2 for 1, 6 and 24 h in a Ruskinn hypoxic workstation (Ruskinn Technology Ltd., Cincinnati, OH, USA) as previously described (8). Cells incubated under aerobic conditions served as the negative control; cells treated with 100 µM of the chelating agent desferrioxamine (DFO) under aerobic conditions served as the positive control.

Nuclear extract preparation

Nuclear extracts were prepared according to previously described protocols with minor modifications (8,40). A total of 5×106 cells were used for each experiment. The cells were scratched-off the Petri dishes using 10 ml phosphate-buffered saline (PBS) and a pellet was obtained by centrifugation. The supernatant was discarded. Subsequently, the cells were transferred into a pre-chilled micro-centrifuge tube, re-suspended in 1 ml PBS and centrifuged at 14,000 × g for 45 sec at 4°C. The cells were then gently re-suspended in 400 µl ice cold hypotonic buffer [10 µM/l HEPES pH 7.9, 10 µM/l KCl, 0.1 µM/l ethylenediaminetetraacetic acid (EDTA), 0.1 µM/l ethylene glycol tetraacetic acid (EGTA), 1 µM/l phenylmethylsulfonyl fluoride (PMSF), 10 µl complete protease inhibitor cocktail (Roche, Mannheim, Germany) and 1 µmol/l dithiothreitol (DTT)] by pipetting up and down several times, and then incubated on ice for 15 min. The cells were lysed by adding 25 µl of 10% NP40 with successive vortexing for 10 sec at the highest setting. The homogenate was centrifuged for 10 min at 3,000 × g at 4°C. The supernatant, containing the cytoplasmic fraction, was transferred and saved. The pellet, containing the nuclear extract, was re-suspended in 50 µl complete cell extraction buffer [20 µM/l HEPES pH 7.9, 0.4 M/l NaCl, 1 µM/l EDTA, 1 µM/l EGTA, 1 µM/l PMSF and 0.1 µl protease inhibitor cocktail (Roche)] by shaking it for 20 min at 4°C in a tube shaker followed by centrifugation at 14,000 × g and 4°C for 5 min. The supernatant was transferred to a clean micro-centrifuge tube and stored at −80°C. The protein concentration was quantitated using the Bradford method protein Quantitation Assay kit 5000001 (Bio-Rad, Munich, Germany).

Whole-cell lysate preparation and western blotting

Whole-cell lysates were prepared with 0.1 ml RIPA buffer (Amresco, Vienna, Austria), 0.5% sodium deoxycholate, 0.1% SDS, protease inhibitors pepstatin A 1.4 µM, aprotinin 0.15 µM and leupeptin 2.3 µM and 100 µM PMSF (all from Sigma-Aldrich, Munich, Germany). Phosphatase inhibitor mix (Sigma-Aldrich) was added. The lysates were transferred to micro-centrifuge tubes, followed by 30 min incubation on ice. Subsequently, cell lysates were cleared by centrifugation at 15,000 × g for 12 min at 4°C.

Western blotting was performed using 20 µg of protein lysates as previously described (40,41). The protein lysates were size-fractionated on 8% polyacrylamide gels (NuPAGE; Life Technologies Carlsbad, CA, USA) by electrophoresis, transferred onto nitrocellulose membranes (Protran BA85; Schleicher & Schuell, Dassel, Germany), and incubated with antibodies directed against HIF-1α (BD Biosciences, Heidelberg, Germany), CEBPα (Abcam Plc., Cambridge, UK), Egr1, SP1 (both from Santacruz Biotech, Heidelberg, Germany), NDRG1, YB-1, MADH7 (all from Abcam Plc.), β-actin and β-tubulin antibodies (both from Sigma-Aldrich). Bound antibodies were detected by developing the membrane using the ECL Plus western blotting detection system for 5 min with subsequent development of the Hyperfilm (both from Amersham Biosciences, Cambridge, UK).

Total RNA isolation and semi-quantitative reverse transcription-polymerase chain reaction (RT-PCR) analysis

Total RNA was isolated from GBM cell lines as previously described (40,41). The RNA (1–5 µg) was reverse transcribed to cDNA using the RevertAid Reverse Transcription kit 1691 (Thermo Fisher Scientific GmbH, Schwerte, Germany) in a 20 µl reaction mixture for 1 h at 42°C according to the manufacturer's instructions. RT-PCR analysis then was performed utilizing PCR systems and reagents from Promega™ (Mannheim, Germany) according to the manufacturer's instructions. The primers were designed in flanking exons with Primer3 software (available online http://frodo.wi.mit.edu/cgibin/primer3/primer3_www.cgi) (Table I). The PCR products were separated on 1% agarose gels (Sigma-Aldrich) and visualized by staining with 0.07 µg/ml ethidium bromide (Bio-Rad).

Table I. Examined genes and primers used.

Table I.

Examined genes and primers used.

	Gene	Gene Bank accession no.	Forward primer (FP)	Reverse primer (RP)	PCR product size (bp)
1	SP1	NM138473.2	5′-ATGGCAAGACCTCTCACCTG-3′	5′-TCTCTTGGACCCATGCTACC-3′	526
2	Smad7	NM_005904.2	5′-CTCGGTGCTCAAGAAACTGA-3′	5′-AATCCATCGGGTATCTGGAG-3′	362
3	CEBPα	NM_004364.2	5′-AACCTTGTGCCTTGGAAATG-3′	5′-CCCTATGTTTCCACCCCTTT-3′	246
4	YB-1	NM_004559.3	5′-GGAGATGAGACCCAAGGTCA-3′	5′-GGTGCTTGCAGTTTGTTGAC-3′	551
5	Egr-1	NM_001964.2	5′-AGCTGGAGGAGATGATGCTG-3′	5′-ACAAGGTGTTGCCACTGTT-3′	348
6	HIF-1α	NM_001530.2	5′-TTACAGCAGCCAGACGATCA-3′	5′-CCCTGCAGTAGGTTTCTGCT-3′	233
7	NDRG1	NM_006096.3	5′-CTCTGTTCACGTCACGCTGT-3′	5′-CTCCACCATCTCAGGGTTGT-3′	593
8	β-actin	NM_001101.3	5′-CGTGCGTGACATTAAGGAGA-3	5′-CACCTTCACCGTTCCAGTTT'-3	668

[i] The analyzed genes, accession numbers, primer sequences and lengths of PCR products are listed.

Densitometric evaluation and statistical analysis

Western blotting band densities and signal strength of RT-PCR were analyzed with 1D Kodak Image Analysis Software (8). DNA or protein amounts were calculated as previously described (21). Statistical data analysis was performed using SPSS 15.0 (SPSS, Inc., Chicago, IL, USA). All experiments were conducted as triplets. Differences between groups were analyzed with either the Student's-test (unpaired, two-tailed) or the Mann-Whitney U test. Statistical significance was considered at p<0.05.

Results

Time- and oxygen concentration-dependent protein expression of NDRG1 and its potential transcription factors

Long-term but not short-term hypoxia leads to the induction of NDRG1 in human glioma cell lines. Previous studies have shown that such a change is associated with tumor progression and an unfavorable prognosis of glioma patients (8). Since hypoxia plays an important role in the regulation of NDRG1 expression, we examined the NDRG1 mRNA and protein expression levels and its regulatory pathways under extreme hypoxia and re-oxygenation in the human glioma cell lines U87-MG, U373 and GaMG. We assumed that the hypoxia-induced transcription factors HIF-1α, CEBPα, Egr-1, SP1, YB-1 and Smad7 may be involved in NDRG1 regulation. Therefore, we examined the protein levels of NDRG1 and of the above-mentioned transcription factors under normoxia (21% O2) and under extreme hypoxic conditions (0.1% O2), which were applied for 1, 6 and 24 h. An additional analysis of protein levels was conducted 24 and 48 h, after re-oxygenation. NDRG1 protein expression was not detectable under normoxic or hypoxic conditions for 1 h in all cell lines analyzed (Fig. 1). Only U87-MG cells exhibited low NDRG1 protein expression under these conditions. However, this cell line exhibited an increase in NDRG1 protein expression after re-oxygenation (Fig. 1B). U373 cells displayed moderate NDRG1 protein expression after 6 h of hypoxia (Fig. 1C), whereas NDRG1 expression was increased in the GaMG cells under extended hypoxia of 24 h (Fig. 1A). In all cell lines, high NDRG1 expression remained stable after re-oxygenation (Fig. 1). Among the analyzed transcription factors, Smad7 displayed a notably alternating response to hypoxia and re-oxygenation. The Smad7 protein expression was undetectable under normoxia, very low after 1 h of 0.1% O2, disappeared again after 6 h hypoxia and then increased to the highest concentration after 24 h hypoxia in GaMG cells (Fig. 1A). Re-oxygenation after 24 h of hypoxia led to a decrease in the Smad7 protein level. In the U87-MG cells, no Smad7 protein expression was detectable, except after 24 h of hypoxia, which led to moderate expression (Fig. 1A). Notably, U373 cells displayed a different expression pattern. Smad7 expression increased after 1 h of hypoxia with a moderate decrease after 6 h and a complete disappearance after 24 h (Fig. 1C). Under re-oxygenation or normoxia no Smad7 expression was detectable in this cell line (Fig. 1C). Hypoxic conditions caused an early increase of SP1 protein already after 1 h in all cell lines, although the magnitude varied from strong (U373 and U87-MG) to moderate (GaMG) (Fig. 1). Moreover, the SP1 protein expression continued to increase and reached a peak after 6 h, to be reduced again after 24 h of hypoxia. After re-oxygenation SP1 expression disappeared in the U373 cells, was low in GaMG and moderate in U87-MG cells, resembling the normoxic expression levels of these cell lines (Fig. 1C).

Figure 1. In vitro protein expression analysis in glioblastoma multiforme (GBM) cells under different oxygenation conditions. The GBM cell lines, GaMG, U87-MG and U373, were exposed to extreme hypoxic oxygenation conditions (0.1% O2) for up to 24 h and then re-oxygenated for 24 h and 48 h. Normoxia and treatment with desferoxamine (DFO; 100 µM) for 24 h served as negative and positive controls, respectively. The protein expression of NDRG1 and of the transcription factors Smad7, SP1, Egr-1, YB-1, HIF-1α and CEBPα was analyzed by western blotting. A clear O2 concentration- and time-dependent expression regulation and also cell line-specific differences were observed. (A) In GaMG, NDRG1 protein expression increase was only significant after 24 h and at the highest expression level and relatively stable upon re-oxygenation for a period of up to 48 h after 24 h of hypoxia. Smad7 protein expression (3rd row from above) pattern displayed a significant protein expression level increase after only 1 h of exposure to extreme hypoxia which decreased to the basal expression level after 6 h of hypoxia and again increased to a maximum level after 24 h of hypoxia, while, re-oxygenation resulted in a significant decrease in the expression level. SP-1 expression pattern (4th row from above) showed a significant increase in the expression level after only 1 h of extreme hypoxia that reached its maximum after 6 h of hypoxia and decreased to expression levels below that of the previous ones, but still higher than the basal expression level and of 24 h of hypoxia, while re-oxygenation up to 48 h after 24 h of hypoxia resulted in a decrease to a lower level slightly higher than those at basal expression level. Egr-1 expression level was stable in response to the different oxygenation conditions. YB-1 expression (6th row from above) was significantly increased after 1 h of hypoxia and was relatively stable up to 24 h of extreme hypoxia; while re-oxygenation for 24 and 48 h, respectively after 24 h of hypoxia resulted in the decrease to an expression level similar to the basal expression level. HIF-1α protein expression level (7th row from above) increased after 1 h of extreme hypoxic exposure and was stable up to 24 h of extreme hypoxia and its level decreased to a similar pattern as the basal expression level under normoxic conditions. CEBPα showed an expression pattern similar to that of YB-1 expression with a lower protein level expressed. (B) In U87-MG, NDRG1 expression increased significantly after 1 h of extreme hypoxia and continued to increase in its expression level up to 24 h of hypoxia and 24 and 48 h of re-oxygenation after 24 h of hypoxia and returned to basal expression level under nomoxic oxygenation conditions. Smad7 expression increased only after 24 h of hypoxia reaching its maximum. SP-1 expression pattern was similar to that in GaMG, but with a higher protein expression level. Egr-1 expression level was stable in response to the different oxygenation conditions examined. YB-1 and HIF-1α expression was displayed in a comparable pattern as in GaMG, with a higher protein level of HIF-1α, while CEBPα expression increased significantly only after 1 h of hypoxia and was at a stable level for up to 24 h of hypoxia but it decreased significantly upon re-oxygenation after 24 h of hypoxia. (C) In U373, NDRG1 expression was significantly increased after 1 h of extreme hypoxia and maintained an increase up to 24 h of hypoxia and 24 and 48 h of re-oxygenation. Smad7 expression level increased after only 1 h of hypoxia and was stable at 6 h of hypoxia but started to decrease to basal expression level after 24 h of hypoxia. SP-1 was expressed at a significant level after only 1 h of hypoxia and increased in its expression level gradually until 24 h, followed by a decline to basal level after 24 h of re-oxygenation and a relative increase after 48 h of re-oxygenation. Egr-1 again showed no change in its displayed pattern in response to the alternating oxygenation conditions. YB-1 and HIF-1α expression patterns were the same as the corresponding protein as was the case in U87-MG with a higher protein expression rate for HIF-1α, and CEBPα expression level increased significantly after 1 h of extreme hypoxic exposition and decreased after 6 h and increased again after 24 h of extreme hypoxia to its maximum level, and decreased again upon re-oxygenation to its basic expression level. One representative experiment out of three independent repetitions is shown. The housekeeping genes β-actin and β-tubulin served as loading controls.

All three examined cells lines expressed Egr-1 protein with no alterations caused by hypoxia or re-oxygenation (Fig. 1).YB-1 was not expressed under normoxia or after re-oxygenation, but hypoxic conditions induced YB-1 protein expression already after 1 h, which remained stable for at least 24 h in all three cell lines (Fig. 1). Hypoxic conditions led to an early increase in HIF-1α protein expression in all three cell lines. Re-oxygenation then caused a considerable decrease of protein expression, but not in its complete disappearance (Fig. 1).

CEBPα displayed a similar pattern under hypoxic conditions with higher expression levels in U87-MG and U373 cells compared to GaMG. Re-oxygenation led to the disappearance of detectable CEBPα protein in GaMG cells, while in U373 and U87-MG residual protein still was detectable with highest concentrations in the U87-MG cells (Fig. 1B). Desferoxamine (DFO) was used as a control and significantly induced the protein expression of all proteins analyzed (Fig. 1).

Time- and oxygen concentration-dependent mRNA expression of NDRG1 and its potential transcription factors

The mRNA expression levels of CEBPα, HIF-1α, EGR-1, Smad7, YB-1 and SP1 in the human glioma cell lines U87-MG, U373 and GaMG cultivated under different oxygenation conditions are illustrated in Fig. 2.

Figure 2. In vitro mRNA expression analysis in glioblastoma multiforme (GBM) cells under different oxygenation conditions. The GBM cell lines GaMG, U87-MG and U373 were exposed to extreme hypoxic oxygenation conditions (0.1% O2) for up to 24 h and then were re-oxygenated for 24 and 48 h. Normoxia and treatment with 100 µM DFO for 24 h served as negative and positive controls, respectively. mRNA-lysates were then analyzed by semi-quantitative RT-PCR for the expression of NDRG1 and of the transcription factors Smad7, SP1, Egr-1, YB-1, HIF-1α and CEBP-α. (A) In GaMG, NDRG1 mRNA expression (1st row, left column) increased after just 1 h of extreme hypoxic exposure and remained at a stable level up to 24 h, and 24 h of re-oxygenation after 24 h of hypoxia, but declined to an expression level nearly to that expressed under normoxia. Smad7 mRNA expression level (2nd row) increased after 1 h of hypoxia and was stable up to 24 h of hypoxia and declined upon re-oxygenation after 24 h to the basal expression level. SP-1 expression (3rd row), increased to its maximum level after 24 h of hypoxia and was stable upon re-oxygenation for 24 and 48 h, respectively, after 24 h of hypoxia. Egr-1 expression (4th row) was at a stable level under the different conditions examined. YB-1 expression level (5th row, left column) reached its maximum level after just 1 h of hypoxic exposition and declined to an expression level similar to that noted under normoxia. HIF-1α mRNA (6th row) was stable under the different aeration conditions examined. CEBPα expression (7th row) increased significantly after 6 h of hypoxia reaching its maximum after 24 h and declined to the basal expression level upon re-oxygenation. (B) In U87-MG, NDRG1 (1st row, middle column) reached its maximum mRNA expression level after only 1 h of hypoxia and was at a stable level under all examined aeration levels and declined to a basal expression level under normoxia. Smad7 mRNA expression (2nd row) behaved in a similar way as in GaMG cells with a higher expression rate. SP-1 mRNA showed a significant expression pattern (3rd row), starting with a significant increase in to a high expression level after 1 h reaching its maximum level after 6 h with a relative decline after 24 h of hypoxia and 24 h of re-oxygenation and after hypoxia and again a slight increase at 48 h of re-oxygenation after 24 h of hypoxia. Egr-1 expression (4th row) was stable with a uniform pattern during the different aeration conditions examined. YB-1 (5th row) reached its maximum expression level after just 1 h of hypoxic exposition, and was stable after 6 h and declined to a basal expression level after 24 h of hypoxia and re-oxygenation after hypoxia. HIF-1α (6th row) was expressed at a stable level under all conditions examined. CEBPα mRNA expression increased significantly after only 1 h of hypoxia, then declined to one half of the expression after 6 h of hypoxia and reached its maximum after 24 h and declined to an expression level similar to a basal expression level upon re-oxygenation after hypoxia. (C) In U373, NDRG1 mRNA (1st row), reached its maximum level after only 6 h of hypoxia and was stable under the different aeration conditions examined. Smad7 mRNA (2nd row) showed a notable expression pattern, showing a significant increase in the expression level after 1 h of hypoxia, that declined after 6 h and increased again to its maximum expression level after 24 h. Re-oxygenation after hypoxia resulted in minimization to a basal expression level. SP-1 expression (3rd row) was also interesting as it reached its maximum level after only 1 h of hypoxia and continuing with a lower expression rate for 6 and 24 h of hypoxia and 24 h re-oxygenation after 24 h of hypoxia and reaching again the maximal expression level after 48 h of re-oxygenation after 24 h of hypoxia. Egr-1 mRNA (4th row) displayed a constant expression pattern under all oxygenation conditions examined. YB-1mRNA (5th row) reached it maximum after 1 h and was stable up to 6 h and declined after 24 h as well after re-oxygenation. HIF-1α (6th row) showed a constant expression pattern under the different oxygenation conditions examined. CEBPα mRNA (7th row), was significantly expressed at its maximum expression level after only 1 h of hypoxia which kept constant up to 24 h of hypoxia and 24 h of re-oxygenation after hypoxia and declined towards the basal expression level upon re-oxygenation for 48 h after 24 h of hypoxia. One representative experiment out of three independent repetitions is shown. The housekeeping gene β-actin served as a loading control.

NDRG1 mRNA was detectable under all conditions in all cell lines analyzed. However, hypoxia procured a relative increase of mRNA levels with time. Re-oxygenation led to a slight reduction after 48 h. NDRG1 mRNA expression was highest in U373 cells, followed by U87-MG and lowest in GaMG, which also was most sensitive to re-oxygenation (Fig. 2).

Smad7 mRNA expression was clearly dependent upon the oxygenation conditions. It was not detectable under normoxia or after re-oxygenation, but highly induced already after 1 h of hypoxia. Notably, Smad7 mRNA expression disappeared after 6 h of hypoxia in the U373 cells and re-appeared after 24 h to a higher level then after 1 h (Fig. 2C). U87-MG cells displayed the strongest mRNA expression and GaMG the lowest level (Fig. 2A and B).

SP1 mRNA expression also showed significant features. In U87-MG cells, SP1 mRNA was moderately expressed and without changes under all tested conditions (Fig. 2B). However, in GaMG cells there was an increase over time under hypoxia, which remained stable at a high level after re-oxygenation (Fig. 2A). In contrast, the U373 cells had increased SP1 mRNA expression noticeably after 1 h of hypoxia, which then decreased back to basal levels over 24 h. The same basal expression was visible after 24 h of hypoxia followed by 24 h of re-oxygenation. Surprisingly, 48 h re-oxygenation caused the strongest SP1 mRNA expression in this experiment (Fig. 2C).

The same Egr-1 mRNA expression strength was detectable under all conditions in all three cell lines examined (Fig. 2).

In contrast, YB-1 mRNA expression was induced by 1 and 6 h of hypoxia, but disappeared after 24 h of hypoxia. It was not detectable under normoxia or after re-oxygenation.

HIF-1α mRNA was expressed at a constant level under all oxygenation conditions in the three cell lines, which is a common expression feature of HIF-1α (Fig. 2).

CEBPα displayed a cell line-specific expression pattern. In all three cell lines there was expression under normoxia (Fig. 2). Hypoxia induced CEBPα expression. In the GaMG cells there was very low expression detectable after 1 h hypoxia, which considerably increased after 6 h and reached the highest levels after 24 h. CEBPα expression then disappeared under re-oxygenation (Fig. 2A). In contrast, the U87-MG cells exhibited increased CEBPα mRNA expression strongly already after 1 h of hypoxia, then after 6 h the cells showed a decrease in expression by 50% to increase the expression again after 24 h. The expression remained high even after re-oxygenation (Fig. 2B). In the U373 cells, there was a significant increase in CEBPα mRNA expression after 1 h of hypoxia, which remained constant over 24 h and only disappeared after 48 h of re-oxygenation (Fig. 2C).

DFO was able to significantly induce the mRNA expression of all tested factors in the three cell lines examined (Fig. 2).

Discussion

The present study showed that NDRG1 expression was associated with long-term hypoxia, but not with short-term hypoxia in human GBM cells. NDRG1 expression may be under the control of HIF-1α in addition to the other transcription factors examined in a so-called expression-regulated time point. The significant increase in NDRG1 expression in hypoxia-exposed GBM cells confirms our previous findings of NDRG1 as a hypoxia-associated molecule with overexpression in tumor cells (21). A common survival strategy of tumor cells at the beginning of hypoxia includes Pasteur-effect remission, in this case it appears that HIF-1α binds to HER sequences of hypoxia-responsive genes (42) and thereby stimulates glycolytic transporter proteins and enzyme expression (43).

The NDRG1 gene contains two HREs within its non-coding sequence (24), indicating that its regulation is HIF-1α-dependent (44). These data support our previous findings that NDRG1 expression is induced by HIF-1α in human tumor cells (8). In addition, the HIF-1α expression precedes NDRG1 expression, indicating that it may be a prerequisite for NDRG1 upregulation since it is its major transcriptional regulator (8,21,45,46). The same holds true for other HIF-1α-regulated genes such as CA9 (47). Therefore, we suggest NDRG1 as a marker for hypoxic regions within a tumor mass (8). It may also be of prognostic relevance, since the survival rate of patients without NDRG1 expression was lower than that of patients bearing NDRG1-positive cells (13). In addition, high expression level of NDRG1 does not correlate with the overall patient survival (48).

In two of the examined GBM cell lines, NDRG1 protein expression was undetectable under normoxic conditions. Although increased NDRG1 protein expression has been reported for different types of tumor cells such as human HCC and colorectal tumors (11,49), other studies showed a decrease in its expression level such as in metastasizing colon cancer cell lines (50). Low-grade astrocytoma (LGA) displayed a lower level of NDRG1 mRNA and protein expression than GBM samples from patients (8). However, another study showed that the NDRG1 mRNA and protein expression was reduced in GBM compared to normal brain tissue (13). These observations suggest that the regulation of NDRG1 may be cancer cell-specific and also dependent on the oxygenation status of the tumor tissue.

In the present study, we showed that NDRG1 mRNA expression was downregulated upon re-oxygenation, which is in agreement with a previous study concerning decreased NDRG1 mRNA expression in the breast cancer cell line MCF-7 after re-oxygenation (51). Therefore, NDRG1 is considered to be an O2-responsive gene, linked to tumor adaptation towards re-oxygenation (51,52). However, our western blot results did not support this notion, and suggest posttranscriptional regulation of NDRG1 expression. The stability of NDRG1 after re-oxygenation following long-time exposure to hypoxic oxygenation conditions may be related to both, post transcriptional and post translational modifications. Based on our results and other research data we can postulate that a post transcriptional modification may be achieved by NDRG1 interaction with the HSP90 protein (53). The mTOR signaling pathway may also be involved, since the downstream regulator CCR4-NOT may upregulate NDRG1 (54,55). Such post transcriptional regulation may act along with post translational modification, such as e.g. sumoylation of HIF-1α. HIF-1α is destabilized and as a consequence the transcription rate of NDRG1 is reduced. In contrast, SUMO protease-1 contributes to HIF-1α stability and thereby increases the NDRG1 transcription rate (55–57). Such modifications may contribute to the modulatory process that in the end renders NDRG1 mRNA for protein synthesis.

Egr-1 displayed a constant expression pattern at the mRNA and protein level, irrespective of the tested oxygenation conditions. This suggests a constitutive functional role for this transcription factor under various physiological conditions in cancer cells (58).

In contrast, the expression of the transcription factors HIF-1α, SP1, CEBPα, YB-1 and Smad7 widely correlated with those of NDRG1. Particularly, the nuclear mRNA levels of HIF-1α, SP1 and C/EBP-α were closely correlated with those of NDRG1. It has been shown that hypoxic conditions raise the expression of these transcription factors in nuclear extracts (59), suggesting that the hypoxia-stimulated upregulation of NDRG1 may be mainly mediated by a combined effect of these factors and to a lesser extend by Smad7 and YB-1. A cooperative mechanism, as suggested in the present study, has been shown to promote the regulation of genes necessary for cell adaptation to varying microenvironments such as during starvation and hypoxia (60). Indeed, NDRG1 promoter studies revealed that a SP1 site partially is responsible for the VHL-induced suppression of NDRG1 (61,62).

The analysis of the murine NDRG1 promoter revealed an overlap between Egr-1 and SP1 binding motifs (62). Although, it has previously been shown that NDRG1 expression during hypoxia is initiated by HIF-1α (63), SP1 triggers the hypoxic promoter activation independently (64) or by interaction with HIF-1α (65). In the present study, SP1 protein expression was increased during hypoxia in the human glioma cell lines. The highest level was detected after 6 h of extreme hypoxia. Most of the transcription factors regulating metabolic events in cancer, including HIF-1α, have a binding site for SP1 in their promoter region (66). HIF-1α expression has been shown to be SP1-dependent in human prostate cancer cells (67), suggesting a crucial regulatory role in cancer metabolism as well as during hypoxia.

We showed an upregulation of CEBPα in human GBM cells in response to hypoxia. However, these results are not consistent with a previous study using breast cancer cells, which showed that the exposure of these cells to 1% O2 downregulates the CEBPα production (68) and transcription (69). These differences may be explained by the severity of hypoxia, as in our experiments the GBM cells were exposed to extreme hypoxic conditions (0.1% O2) for up to 24 h. It has been reported that the HIF-1α-bound HRE within the CEBPα promoter region is essential for CEBPα transcriptional downregulation under hypoxic conditions (62). In contrast, we observed a lack of CEBPα expression in GBM cells under normoxic conditions. CEBPα downregulation under normoxia/moderate hypoxia and its upregulation during extreme hypoxia may play a role during progression.

Different signal transduction pathways are controlled and modulated by different growth factors and environmental events such as hypoxic oxygenation conditions (26). The TGF-β pathway involves the Smad protein family (27). These proteins play an important role in triggering diseases including for example fibrosis (28) and cirrhosis, leading to HCC (29). NDRG1 overexpression could inhibit TGF-β-induced effects via reducing both Smad2 and Smad3 expression (63). However, Smad7, another member of this protein family, abrogates the TGF-β signaling pathway by a regulative negative feedback loop (30), thereby, paving the physiological conditions towards cancer disease manifestation, as HCC is an example when this negative feedback loop is hindered, while the transition to fibrosis, cirhosis and as a consequence HCC is hindered when this negative feedback loop of Smad7 remains active (31). Smad7 is regulated by different signaling pathways (32). IFNγ contributes to this regulatory scheme by inducing a member of the cold shock protein family, the transcription factor YB-1 (30,70–73) (Fig. 3).

Figure 3. Proposed regulatory signaling network for hypoxia-induced gene activation in human cancer diseases. Although HIF-1α is the main regulator of hypoxia-induced genes, other signaling pathways play an important assisting role. NDRG1 is shown as an example. (A) Under hypoxia, SP1 upregulates HIF-1α and subsequently this leads to an increase in hypoxia-induced gene transcription (NDRG1). (B) Egr-1 expression remained stable despite varying oxygenation conditions. Therefore, Egr-1 may be a cofactor for the regulation of hypoxia-induced transcription. (C) TGF-β upregulates Smad7 which functions as an antagonist via a negative feedback loop. (D) IFN-γ stimulates YB-1 overexpression which then may increase Smad7 expression, and thereby provoke the overexpression of NDRG1. (E) CEBPα expression is increased in an oxygen concentration- and time-dependent manner, and thereby functions as a positive regulator for NDRG1 expression. (F) NDRG1 expression rate is increased due to the activity of the outlined network of cytokines and transcription factors.

Our results suggest that HIF-1α regulates NDRG1 expression in GBM cells in conjunction with SP1, CEBPα, YB-1 and Smad7 under different physiological conditions, whereas Egr-1 may act as a cofactor (Fig. 4). A very similar regulatory mechanism may also exist for other hypoxia-induced genes (7), such as CA9, Epo and VEGF in human brain cancer (40,64). These factors are also controlled by HIF-1α. However, other co-factors may be involved. The knowledge concerning such regulatory processes during cancer development is important for the development and application of cancer treatment modalities.

Figure 4. Summary of pathways regulating NDRG1 gene activity under different oxygenation conditions in human brain cancer cells in vitro. SP1, HIF-1α, CEBPα, Smad7 and YB-1 play an active role in the regulation of NDRG1 while the effect of Egr-1 is neutral. The intensity and activating potential is variable and depends on the aeration conditions and time.

In conclusion, we showed for the first time that NDRG1 protein expression was significantly increased in GBM cells under hypoxia and that there was a correlation with the protein expression of HIF-1α, SP1, CEBPα, YB-1 and Smad7. We suggest that SP1 regulates NDRG1 in GBM cells under different oxygenation conditions in conjunction with HIF1α, CEBPα and, to a certain extent, YB-1 and Smad7. NDRG1 and its potential transcription factors therefore may represent target proteins for complementary treatment approaches. Combined, they also may serve as potential hypoxia tumor markers in gene expression analyses.

Acknowledgements

The authors are indebted to the Dokuz Eylul University, Graduate Institute of Health Sciences (SBE) for support. The present study was partially funded by means of the University of Würzburg, Medical Faculty, Department of Radiation Oncology. The authors are also grateful to the College of Medicine and Medical Sciences, Arabian Gulf University, Manama, Kingdom of Bahrain. We would like to thank Stefanie Gerngras and Siglinde Kühnel for their technical assistance.

References