Vol. 53 No. 4/2006, 789–799 Regular paper on-line at: www.actabp.pl

Previously we have shown that hypoxia strongly induces the expression of 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase-3 and -4 (PFKFB-3 and PFKFB-4) genes in several cancer cell lines via a HIF-dependent mechanism. In this paper we studied the expression and hypoxic regulation of PFKFB-4 and PFKFB-3 mRNA as well as its correlation with HIF-1alpha, HIF-2alpha, VEGF and Glut1 mRNA expression in the pancreatic cancer cell line Panc1 and two gastric cancer cell lines MKN45 and NUGC3. This study clearly demonstrated that PFKFB-3 and PFKFB-4 mRNA are expresses in MKN45, NUGC3 and Panc1 cancers cells and that both genes are responsive to hypoxia in vitro. However, their basal level of expression and hypoxia responsiveness vary in the different cells studied. Particularly, PFKFB-3 mRNA is highly expressed in MKN45 and NUGC3 cancer cells, with the highest response to hypoxia in the NUGC3 cell line. The PFKFB-4 mRNA has a variable low basal level of expression in both gastric and pancreatic cancer cell lines. However, the highest hypoxia response of PFKFB-4 mRNA is found in the pancreatic cancer cell line Panc1. The basal level of PFKFB-4 protein expression is the highest in NUGC3 gastric cancer cell line and lowest in Panc1 cells, with the highest response to hypoxia in the pancreatic cancer cell line. Further studies showed that PFKFB-3 and PFKFB-4 gene expression was highly responsive to the hypoxia mimic dimethyloxalylglycine, a specific inhibitor of HIF-alpha hydroxylase enzymes, suggesting that the hypoxia responsiveness of PFKFB-3 and PFKFB-4 genes in these cell lines is regulated by the HIF transcription complex. The expression of VEGF and Glut1, which are known HIF-dependent genes, is also strongly induced under hypoxic conditions in gastric and pancreatic cancer cell lines. The levels of HIF-1alpha protein are increased in both gastric and pancreatic cancer cell lines under hypoxic conditions. However, the basal level of HIF-1alpha as well as HIF-2alpha mRNA expression and their hypoxia responsiveness are different in the MKN45 and NUGC3 cancer cells. Thus, the expression of HIF-1alpha mRNA is decreased in both gastric cancer cell lines treated by hypoxia or dimethyloxalylglycine, but HIF-2alpha mRNA expression is not changed significantly in NUGC3 and slightly increased in MKN45 cells. Expression of PFKFB-4 and PFKFB-3 was also studied in gastric cancers and corresponding nonmalignant tissue counterparts from the same patients on both the mRNA and protein levels. The expression of PFKFB-3 and PFKFB-4 mRNA as well as PFKFB-1 and PFKFB-2 mRNA was observed in normal human gastric tissue and was increased in malignant gastric tumors. The basal level of PFKFB-4 protein expression in gastric cancers was much higher as compared to the PFKFB-3 isoenzyme. In conclusion, this study provides evidence that PFKFB-4 and PFKFB-3 genes are also expressed in gastric and pancreatic cancer cells, they strongly respond to hypoxia via a HIF-1alpha dependent mechanism and, together with the expression of PFKFB-1 and PFKFB-2 genes, possibly have a significant role in the Warburg effect which is found in malignant cells.

The transcription factor HIF is central in coordinating many of the transcriptional adaptations to hypoxia and a necessary mediator of the hypoxic effect in mammalian cells (Wenger, 2002;Bitlon & Booker, 2003;Greijer et al., 2005).HIF is a heterodimeric transcription factor composed of two subunits: a constitutively expressed β-subunit and an α-subunit.Their expression and activity are controlled by intracellular oxygen concentration (Wenger, 2002).In mammals HIF-α subunit exists as multiple isoforms with different biological properties.Three principal isoforms (HIF-1α, HIF-2α and HIF-3α) are encoded by three distinct genetic loci, further diversity being generated by alternative promoter usage and alternative splicing (Semenza et al., 2001;Makino et al., 2001;2002).Expression of these genes varies in different organs and cell types but HIF-1α is the widely expressed and major functional isoform (Bitlon & Booker, 2003;Hu et al., 2003;Huang & Bunn, 2003;Sowter et al., 2003;Elvidge et al., 2006).The HIF-1α and HIF-2α isoforms are closely related, but HIF-3α is significantly different and is expressed as a number of alternative spliced variants (Makino et al., 2001).One of them, inhibitory PAS domain protein (IPAS), is a hypoxia-inducible splicing variant of the hypoxia-inducible factor-3α locus.It is a negative regulator of hypoxia-inducible gene expression.IPAS functions as a dominant negative regulator of hypoxia-inducible transcription factors by forming complexes with those proteins that fail to bind to hypoxia response elements of target genes.The alternatively spliced transcript of HIF-3α was only observed in the heart and lungs under hypoxic conditions (6% oxygen for 6 h), thus defining a novel mechanism of hypoxiadependent regulation of gene expression (Makino et al., 2002).Importantly, this mechanism may establish a negative feedback loop regulation of adaptive responses to hypoxia/ischemia in these tissues.There are also several alternative spliced variants of HIF-1α (HIF-1α 516 , HIF-1α 557 , HIF-1α 735 ) that terminate translation respectively at codons 516, 557 and 735, resulting in the absence of both N-activation domain and C-activation domain or C-activation domain only (Wenger, 2002).There are data that overexpression of an exogenous testis-specific splice isoform of human HIF-1α (hHIF-1αTe) resulted in the inhibition of the endogenous HIF-1 transcriptional activity, demonstrating that the testis-specific hHIF-1αTe isoform is a dominant-negative regulator of normal HIF-1 activity (Depping et al., 2004).
The expression of the transcriptional complex HIF is tightly coupled to oxygen concentration.Whereas the HIF-1β subunit is constitutively expressed, HIF-1α is highly unstable in normoxic conditions, being rapidly degraded by the ubiquitinproteasome system (Epstein et al., 2001;Ivan et al., 2001;Metzen & Ratcliffe, 2004).Recently, the oxygen sensors monitoring this hypoxic response were identified as prolyl-and asparaginyl-hydroxylase enzymes, which during normoxia (normal physiological levels of oxygen) mediate the rapid degradation of HIF-α and prevent transcriptional recruitment of the cofactor CBP/p300, respectively (Epstein et al., 2001;Ivan et al., 2001;Metzen & Ratcliffe, 2004).HIF-1 activation depends on the hydroxylation of specific prolyl and asparaginyl residues in the α subunit of HIF complex that controls the survival and transcriptional activity of this protein (Ivan et al., 2001;Min et al., 2002;Schofield & Ratcliffe, 2004).These recently described HIF hydroxylases are a family of non-haem iron and oxoglutarate-dependent dioxygenases that define a novel mechanism of protein modification that transduces the oxygen-sensing signal and controls hypoxic gene activation (Semenza, 2001;Lando et al., 2002;Min et al., 2002;Masson & Ratcliffe, 2003;Mole et al., 2003;Appelhoff et al., 2004).There is data (Goyal et al., 2004) that upregulation of NADPH oxidase 1 under hypoxia activates hypoxia-inducible factor 1 via an increase in reactive oxygen species.The HIF pathway is also responsible for patterns of gene expression as well as cell growth creatic cancer cells, they strongly respond to hypoxia via a HIF-1α dependent mechanism and, together with the expression of PFKFB-1 and PFKFB-2 genes, possibly have a significant role in the Warburg effect which is found in malignant cells.
Thus, 6-phosphofructo-2-kinase/fructose-2,6bisphosphatase is a key regulatory enzyme of glycolysis both in normal and hypoxic conditions because hypoxia induces the expression of PFKFB-1, PFKFB-2, PFKFB-3 and PFKFB-4 genes in various cell lines (Chesney et al., 1999;Marsin et al., 2002;Minchenko et al., 2002;2003;2004;2005c).However, regulation of the expression of these PFKFB isoenzymes following hypoxic treatment is different and occurrs in a cell-specific manner.Moreover, the rapid activation of glycolysis by fructose-2,6-bisphosphate as well as by hypoxia in activated monocytes is regulated by phosphorylation -dephosphorylation of PFKFB isoenzymes at the N-or C-terminus by AMP-activated and several other protein kinases in different signaling pathways (Marsin et al., 2002;Hue et al., 2003;Rider et al., 2004).High expression of PFKFB-4, PFKFB-3 and its inducible isoform was observed in various human cancers (Chesney et al., 1999;Atsumi et al., 2002;Minchenko et al., 2004;2005a;2005b).Moreover, there is data supporting a potential role for the phosphorylation of PFKFB-3 protein in enhanced glycolysis, the progression of cancer and angiogenesis (Bando et al., 2005).Highly phosphorylated PFKFB-3 was found in human tumor cells, vascular endothelial cells, and smooth muscle cells, as determined by immunostaining with an anti-phospho-PFK-2 (PFKFB-3) antibody (Bando et al., 2005).However, little is known about the expression and hypoxia-responsiveness of PFKFB-4 and PFKFB-3 in pancreatic and gastric cancer cells which significantly differ from many other malignant cell lines.
This study provides evidence that the PFKFB-4 and PFKFB-3 genes are expressed in human pancreatic and gastric cancer cell lines, strongly respond to hypoxia via an HIF-dependent mechanism and are overexpressed in gastric malignant tumors.Hypoxic induction of HIF-1α protein in these cell lines correlates with reduction HIF-1α mRNA expression.

Materials.
Dimethyloxalylglycine was obtained from Frontier Scientific, Inc. (Logan, UT, USA).[α 32 P]UTP (800 Ci/mmol) and Hyperfilm ECL were obtained from Amersham Biosciences.Cell lines and culture conditions.Human gastric cancer cell lines MKN45 and NUGC3 as well as pancreatic cancer cell line Pank1 were obtained from the American Type Culture Collection (Rockville, MD, USA) and grown according to the supplier's protocols.The cells were incubated at 37 o C before harvesting under normoxic (21% oxygen and 5% carbon dioxide) or hypoxic (1% oxygen, 5% carbon dioxide and 94% nitrogen for 6 h) conditions or exposed for 6 h to 1 mM dimethyloxalylglycine.
RNA isolation.Total RNA was extracted using Trizol reagent according to the manufacturer's protocols (Invitrogen, Carlsbad, CA, USA).RNA pellet was washed with 75% ethanol, dissolved in nuclease-free water and used for ribonuclease protection assays.
In vitro transcription to prepare antisense probes for ribonuclease protection assay.The synthesis of radiolabeled probes for ribonuclease protection assay was according to the BD Biosciences protocol using T7 RNA polymerase (BD Biosciences Pharmingen, San Diego, CA, USA) and [α 32 P]UTP.For ribonuclease protection assays water solutions of total RNA were dried under vacuum and dissolved in 20 µl of 80% formamide hybridization buffer containing labeled probes.Samples were preincubated for 5 min at 85 o C to denature RNA and then incubated for 16 h at 45 o C as described previously (Minchenko & Caro, 2000).Single-strand RNA was removed by digestion with ribonuclease T1 at 30 o C for 60 min and extracted with phenol/chloroform.Protected probe fragments were precipitated with 2.5 volumes of ethanol with the addition of 10 µg of transfer RNA/sample.The samples were denatured and protected fragments separated on a 6% polyacrylamide sequencing gel in Tris/borate/EDTA buffer for 2 h at 50 mA as described previously (Minchenko et al., 1994).The expression of different mRNAs was determined using Fujix BAS 2000 Bio-Image Analyzer (Fuji Photo Film Co., Japan).The intensity of each mRNA band was normalized for 18 S ribosomal RNA levels and results are expressed as the ratio of mRNA to 18S rRNA (percent of control).
The protein complexes were visualized by enhanced chemiluminescence reagents (Amersham Biosciences) as described previously (Minchenko et al., 2004).β-Actin was used to ensure equal loading of the samples.
Statistical analysis.The results are expressed as mean ± standard error of the mean (S.E.M.) of four independent experiments.Comparison of two means was performed by the use of unpaired Student's t-test.P value of < 0.05 was regarded as significant.

RESULTS
In this work, we studies the expression, hypoxic regulation and mechanisms of hypoxic regulation of PFKFB-3, PFKFB-4, HIF-1α, HIF-2α, VEGF and Glut1 genes in human pancreatic and gastric cancer cell lines as well as in human gastric cancers.

Effect of hypoxia and dimethyloxalylglycine on PFKFB-4 and PFKFB-3 gene expression in gastric and pancreatic cancer cell lines
To examine the effects of hypoxia and dimethyloxalylglycine on the expression of the PFKFB-4 gene in gastric and pancreatic cancer cell lines, mRNA levels were measured by ribonuclease protection assays.As shown in Fig. 1, the transcript PFKFB-3 and PFKFB-4 gene expression in pancreatic and gastric cancer cells level of PFKFB-4 isozyme is very low but detectable in the human gastric cancer cell line MKN45 growing under normoxic conditions.These cells also express PFKFB-3.However, the basal transcript level of PFKFB-3 is much higher as compared to PFKFB-4.Exposure of MKN45 gastric cancer cells to hypoxia (1% oxygen for 6 h) greatly stimulated the expression of both PFKFB-4 and PFKFB-3 mRNA.The ribonuclease protection assays analysis at PFKFB-4 and PFKFB-3 mRNA expression in the MKN45 gastric cancer cells was also highly responsive to dimethyloxalylglycine.
In the human gastric cancer cell line NUGC3 the basal transcript level of PFKFB-4 and PFKFB-3 and their hypoxia responsiveness were different.Thus, the transcript level of PFKFB-4 mRNA in these cancer cells growing under normoxic conditions was much lower as compared to the PFKFB-3 isoform.The expression of both PFKFB-4 and PFKFB-3 mRNA in NUGC3 gastric cancer cells was increased by hypoxia; however, hypoxia responsiveness of PFKFB-3 was much higher as compared to PFKFB-4.Expression of both PFKFB-4 and PFKFB-3 mRNA in NUGC3 gastric cancer cells was highly responsive to the hypoxia mimic dimethyloxalylglycine.
We also studied the expression and hypoxia responsiveness of PFKFB-4 and PFKFB-3 genes on the mRNA level in the human pancreatic cancer cell line Panc1 (Fig. 2).The transcript level of the PFKFB-4 isoenzyme in this cancer cell line growing under normoxic conditions was low but slightly higher as compared to PFKFB-3 mRNA.Exposure of these cancer cells to hypoxia stimulated the expression of PFKFB-4 and, to a lesser extent, PFKFB-3 mRNA.Dimethyloxalylglycine also strongly induce the PFKFB-4 mRNA expression in Panc1 cells.The expression of PFKFB-3 gene in the pancreatic cancer cells was increased by dimethyloxalylglycine but the hypoxia responsibility of this isoform of PFKFB was much lower as compared to PFKFB-4 mRNA.
The results of four independent ribonuclease protection experiments were quantified using a Fujix BAS 2000 Bio-Image Analyzer and expressed as the ratio between PFKFB-4 or PFKFB-3 mRNA and 18S rRNA.As shown in Fig. 2, the increase of the PFK-FB-4 and PFKFB-3 transcript levels in the MKN45 gastric cancer cells by hypoxia was similar: 406% (P < 0.01) and 398% (P < 0.01), respectively.However, in the gastric cancer cell line NUGC3 hypoxia increase of the PFKFB-3 transcript level more strongly that than of as compared to PFKFB-4: PFKFB-3 by 1198% (P < 0.001) and PFKFB-4 by 306% (P < 0.01).Quantification of ribonuclease protection assays of PFKFB-4 and PFKFB-3 mRNA levels in the pancreatic cancer cells shows that hypoxia induced PFKFB-4 transcript by 12-fold (P < 0.001) and PFKFB-3 mRNA level -twice (P < 0.05).
Further investigation showed, that hypoxia or dimethyloxalylglycine induced the expression of PFKFB4 protein measured by Western blot analysis with a rabbit polyclonal anti-PFKFB4 antibody in both gastric and pancreatic cancer cells (Fig. 3).However, both the basal and hypoxia-induced expression of PFKFB-4 protein differed in both gastric and pancreatic cancer cell lines.A strongest induction of PFKFB-4 protein level both by hypoxia and dimethyloxalylglycine was observed in the pancreatic cancer cells.

Effect of hypoxia and dimethyloxalylglycine on HIF-1α and HIF-2α gene expression in gastric and pancreatic cancer cell lines
As shown in Fig. 4, hypoxia induces the protein level of HIF-1α in the both gastric cancer cell lines MKN45 and NUGC3 as well as in the pancreatic cancer cell line Panc1.However, the basal HIF-1α protein level differs in the MKN45 and NUGC3 gastric cancer cell lines: it is much higher in the MKN45 cells.In the NUGC3 and Panc1 cell lines the basal and hypoxia inducible levels of HIF-1α protein were similar.However, the transcript levels of HIF-  1α were decreased in both gastric cancer cell lines under hypoxic conditions (Fig. 5).The MKN45 and NUGC3 gastric cancer cells treated with dimethyloxalylglycine have also decreased levels of HIF-1α mRNA.Analysis of HIF-2α mRNA expression and its hypoxia responsiveness in the MKN45 and NUGC3 cancer cells showed that the basal level of HIF-2α mRNA was much lower as compared to HIF-1α mRNA (Fig. 5), but showed a positive response to hypoxia or dimethyloxalylglycine treatment in MKN45, while in the NUGC3 cancer cells was not responsive to hypoxia and slightly decreased HIF-2α mRNA expression after dimethyloxalylglycine treatment.

Effect of hypoxia and dimethyloxalylglycine on VEGF and Glut1 gene expressions in gastric and pancreatic cancer cells
We also studied the hypoxic regulation of Glut1 and VEGF mRNA expression to compare the hypoxia responsiveness of the PFKFB-4 and PFK-FB-3 genes with known HIF-1 dependent genes.Results of this study clearly demonstrated that hypoxia and dimethyloxalylglycine strongly induced the transcript levels of Glut1 and VEGF genes both in the gastric and pancreatic cancer cell lines, like it was obtained for PFKFB-4 and PFKFB-3 genes (Fig. 6).However, some differences were found in the induction of Glut1 and VEGF genes by hypoxia or dimethyloxalylglycine treatment in the MKN45 and NUGC3 gastric cancer cell lines: the expression of Glut1 mRNA was increased much more strongly in the MKN45 cells, while the expression of VEGF mRNA -in the NUGC3 gastric cancer cells.

Expression of PFKFB-1, PFKFB-2, PFKFB-3, PFKFB-4, Glut1 and VEGF mRNA in human gastric malignant tumors
The PFKFB-4 and PFKFB-3 gene expression was studied in gastric cancers and corresponding nonmalignant tissue counterparts from the same patients on both the mRNA and protein levels.Moreover, we also analyzed the expression of two other PFKFB genes -PFKFB-1 and PFKFB-2.As shown in    Fig. 7, the transcript level of PFKFB genes in gastric non-malignant tissue was maximal for PFKFB-2, slightly less for PFKFB-3 and minimal for PFKFB-1 and PFKFB-4.Analysis of the expression of PFKFB-1, PFKFB-2, PFKFB-3 and PFKFB-4 mRNA in malignant gastric tumors showed that expression of all these mRNAs was increased in the tumor but the strongest induction was observed for PFKFB-3 mRNA.The expression of known HIF-1-dependent genes Glut1 and VEGF was also increased in gastric malignant tumors.
Using Western blotting we measured the protein levels of PFKFB-4 and PFKFB-3 as well as the levels of the inducible isoform of the PFKFB-3 isozyme.As shown in Fig. 8, the level of PFKFB-4 protein expression in gastric cancers was much higher as compared to the PFKFB-3 isoenzyme.The expression of PFKFB-4, PFKFB-3 and iPFKFB-3 proteins was increased in malignant gastric tumors as compared to corresponding nonmalignant tissue.Thus, the main protein isoform of PFKFB expressed in both gastric malignant tumors and nonmalignant tissue is PFKFB-4.

DISCUSSION
This study clearly demonstrates that 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase-3 and -4 mRNA is expressed in the pancreatic cancer cell line Panc1 and in two gastric cancer cell lines MKN45 and NUGC3 and that both genes are responsive to hypoxia in vitro.The ribonuclease protection analysis also showed that PFKFB-4 and PFKFB-3 mRNA expression in both gastric and pancreatic cancer cell lines is highly responsive to dimethyloxalylglycine in normoxic conditions.However, the basal transcript level of PFKFB-4 and PFKFB-3 and their hypoxia responsiveness vary in the different cell lines studied.Moreover, no strong correlation is present between PFKFB-3 or PFKFB-4 mRNA and protein expression in the gastric and pancreatic cancer cells, both in normoxic and hypoxic conditions.This data concurs with our previous results (Minchenko et al., 2002;2004;2005c).
Thus, we have previously shown that hypoxia strongly induces the expression of PFKFB-3 and PFKFB-4 genes in many cancer cell lines via HIF-dependent mechanism in a cell-specific manner.Moreover, the analysis of PFKFB-4 transcript and protein levels in different mammary gland malignant cell lines clearly demonstrated that no strong correlation is present between PFKFB4 mRNA and protein expression, both in normoxic and hypoxic conditions (Minchenko et al., 2005c).Thus, the basal and hypoxia-inducible expression of PFKFB-4 mRNA is higher in the T47D mammary gland adenocarcinoma cell line as compared the MCF7 mammary gland adenocarcinoma cell line.However, PFKFB-4 protein expression is higher in the MCF7 cell line as compared to T47D cells, both in normoxic and hypoxic conditions (Minchenko et al., 2005c).Moreover, SKBR3 and MDA-MB-468 mammary gland adenocarcinoma cell lines have similar PFKFB-4 mRNA expression both in normoxic and hypoxic conditions.Unexpectedly, no significant levels of PFKFB4 protein were observed in the MDA-MB-468 cancer cell line in the experimental conditions used for SKBR3 or other cell lines.In contrast, the basal and hypoxia-inducible expression of PFKFB-4 mRNA is correlated with PFKFB-4 protein level in the SKBR3 and BT549 mammary gland adenocarcinoma cell lines (Minchenko et al., 2005c).The results of this study demonstrate a similar basal expression of PFKFB-4 mRNA and a significantly different basal expression of PFKFB-4 protein in gastric and pancreatic cancer cell lines.Thus, the basal level of PFKFB-4 protein is the lowest in the Panc1 cell line and highest in NUGC3 gastric cancer cells.The level of hypoxic regulation of PFKFB-4 mRNA and protein varies in different gastric cancer cell lines as well as in the Panc1 cell line.
It is interesting to note that there is an inverse correlation between the basal levels of PFKFB-4 protein expression and the hypoxic responsiveness of PFKFB-4 mRNA expression, although its biological significance remains to be determined.Indeed, the lowest basal level of PFKFB-4 protein, which is observed in the pancreatic cancer cell line Panc1, is correlated with the highest hypoxic induction of PFKFB-4 mRNA and protein expression; however, the hypoxia-induced level of PFKFB-4 protein in these cells is in fact lower when compared to the basal level of PFKFB-4 protein expression in NUGC3 cells.Moreover, the highest basal level of PFKFB-4 protein, which we observed in the gastric cancer cell line NUGC3, is correlated with the lowest hypoxic induction of PFKFB-4 mRNA and protein expression; on the other hand, the hypoxia-induced level of PFKFB-4 protein in these cells is very high as compared to the basal or hypoxia-induced levels of PFKFB-4 protein in both Panc1 and MKN45 cells.Thus, the results of Western blot analysis with specific anti-human PFKFB-4 antibodies revealed that the constitutive levels of PFKFB-4 protein in the MKN45 and NUGC3 gastric cancer cell lines as well as in the Panc1 pancreatic cancer cell line are different and are increased by hypoxia.These observations suggest that the increase in PFKFB-4 protein expression was not reflected at the mRNA level.It is possible that this discrepancy between PFKFB-4 mRNA and protein levels, which were found in the different gastric and pancreatic cancer cell lines, is related to the mechanism which controls stability of PFKFB-4 protein.However, the precise molecular mechanism for these inverse correlations is complex and possibly includes post-translational modification and stability of PFKFB-4 enzyme in a cell type-specific manner and warrants further investigation.
The induction of PFKFB-3 mRNA expression in the NUGC3 gastric cancer cell line by hypoxia and dimethyloxalylglycine is much stronger as compared to PFKFB-4 mRNA expression.It is important to note that PFKFB-3 gene in the Panc1 pancreatic cancer cells has the lowest hypoxia responsiveness as compared to both gastric cancer cell lines.We have previously shown that the hypoxic induction of PFKFB-3 mRNA expression in mammary gland cancer cells is much stronger in MCF7 and T47D breast cancer cells (estrogen receptor-positive cell lines) as compared to SKBR-3 and MDA-MB-468 cells (estrogen receptor-negative cell lines) (Minchenko et al., 2005c).The different sensitivity of the PFKFB-3 isoenzyme to hypoxic induction was shown recently for many other cell lines (HeLa, Hep3B, RPE, and fibroblasts), while hypoxic induction of Glut1 gene was similar in these cell lines (Minchenko et al., 2002).
The results of this investigation provide clear evidence that the hypoxic induction of PFKFB-4 gene expression was replicated by dimethyloxalylglycine in the different gastric and pancreatic cancer cell lines, suggesting that the hypoxia responsiveness of this gene is regulated by HIF proteins.Dimethyloxalylglycine (an oxoglutarate analog) is a specific inhibitor of prolyl hydroxylases, protects the HIF-1α protein from proteasomal degradation and significantly increases its level (Epstein et al., 2001;Schofield & Ratcliffe, 2004).Inhibition of these enzymes can induce the levels and transcriptional activity of HIF-1α under normoxic conditions and mimics hypoxic conditions (Metzen & Ratcliffe, 2002).Thus, using dimethyloxalylglycine we can identify the HIF-dependent effects of hypoxia.
Recently was shown that hypoxic induction of PFKFB-4 gene transcription is mediated by the hypoxia responsive element located in the 5'-pro-moter region of the human PFKFB-4 gene (293-300 bp upstream from the GATA site) and that this hypoxia responsive element has homology with the same described in other hypoxia responsive genes (Gleade & Ratcliffe, 1998;Minchenko & Caro, 2000;Wenger, 2002;Fukasawa et al., 2004;Minchenko et al., 2005c).Recently the hypoxia responsive element was identified in the promoter region of PFKFB-3 gene which mediates hypoxia responsiveness on the transcriptional level (Fukasawa et al., 2004;Obach et al., 2004).
The expression of VEGF and Glut1, which are known HIF-dependent genes, is also strongly induced under hypoxic conditions in gastric and pancreatic cancer cell lines.The significant induction of HIF-1α protein in both gastric and pancreatic cancer cell lines under hypoxic conditions supports the HIF-1α-dependent character of the induction of expression of the genes studied.However, the basal level of HIF-1α as well as HIF-2α mRNA expression and their hypoxia responsiveness is different in the MKN45 and NUGC3 cancer cells.Thus, the expression of HIF-1α mRNA is decreased in both gastric cancer cell lines treated by hypoxia or dimethyloxalylglycine, but no significant changes of HIF-2α mRNA expression were found in the NUGC3 gastric cancer cell line under hypoxia.However, the expression of HIF-2α mRNA in the MKN45 gastric cancer cell line was slightly induced by dimethyloxalylglycine and hypoxia.A similar pattern of HIF-1α and HIF-2α mRNA expression in hypoxic conditions was shown in the A 549 lung adenocarcinoma cell line (Uchida et al., 2004).Importantly, there is an inverse correlation between the hypoxic responsiveness of HIF-1α mRNA and protein expression.These observations suggest that the increase in HIF-1α protein expression was not reflected at the mRNA level.Moreover, the expression of HIF-1α mRNA is significantly decreased both under hypoxic condition and by dimethyloxalylglycine action.It is possible that this discrepancy between HIF-1α mRNA and protein levels, which was found in different gastric and pancreatic cancer cell lines, is related to the divergence in mechanisms which control the stability of HIF-1α mRNA and protein.Thus, the hypoxic induction of HIF-1α protein expression is a result of its stabilization which is mediated by specific oxygen-and iron-dependent prolyl hydroxylase enzymes that utilize oxoglutarate as a co-substrate (Epstein et al., 2001;Lando et al., 2002;Mole et al., 2003;Schofield et al., 2004).There is data that hypoxia and dimethyloxalylglycine decrease HIF-1α mRNA levels in a cell type-specific manner (Minchenko & Caro, 2000;Marti et al., 2002;Uchida et al., 2004;Minchenko et al., 2006).The dimethyloxalylglycine responsiveness of HIF-1α mRNA expression suggests that HIF-1α protein can mediate HIF-1α mRNA expression dur-ing hypoxia.Uchida et al. (2004) have shown that hypoxia induce HIF-1α protein and reduce HIF-1α mRNA expression in A 549 lung carcinoma cell line, but HIF-2α mRNA is slightly increased.This data is closely correlated with the results of the present investigation: hypoxia reduces HIF-1α and slightly increases HIF-2α mRNA expression in the gastric cancer cell line MKN45.This effect is cell-specific because it is not observed in the NUGC3 gastric cancer cell line.
Cancer cells show elevated glycolytic rates, produce high levels of lactate and pyruvate (the Warburg effect), and this correlates with an increased expression of glycolytic enzymes and glucose transporters via a HIF-dependent mechanism (Lu et al., 2002;Hopfl et al., 2004).Since PFKFB isoenzymes catalyze the synthesis and degradation of fructose-2,6-bisphosphate, they control glycolysis and play a significant role in the Warburg effect, which is typical for tumor cells (Chesney et al., 1999;Minchenko et al., 2002;2004;Rider et al., 2004).The major finding reported here is that several isoenzymes of 6phosphofructo-2-kinase/fructose-2,6-bisphosphatase with different kinetics and regulatory properties are overexpressed in solid gastric tumors as compared to nonmalignant tissue counterparts from the same patients, suggesting that all four isoenzymes may contribute to the Warburg effect.Our study provides evidence that the PFKFB-4 gene is overexpressed in most cancer cells, is strongly responsive to hypoxia, and may have a significant role in the Warburg effect, much like PFKFB-3.Recently, we have shown that lung cancers also overexpress all of the PFKFB genes (Minchenko et al., 2005b).Of interest is that the level of different PFKFB gene expression differs in lung and gastric cancers and nonmalignant tissues.These observations suggest that different PFKFB genes are expressed in nonmalignant tissues and cancers in a cell-specific manner.Thus, in lung cancers expression of PFKFB-4 and PFKFB-3 mRNA is increased much more strongly than in gastric cancers as compared to nonmalignant tissue counterparts (Minchenko et al., 2005b).
Moreover, this study provides evidence that the high level of PFKFB-4 protein expression was detected in gastric malignant tumors.Similar results were shown in the colon, breast and lung cancers (Minchenko et al., 2005a;2005b).However, the level of PFKFB-4 protein in the lung and breast nonmalignant tissues is very low in comparison to the gastric normal tissues, but its expression much higher in the lung and breast malignant tumors, as opposed to the gastric and colon cancers (Minchenko et al., 2005a;2005b).It is interesting to note that there is an inverse correlation between the levels of PFKFB-4 protein expression in normal (nonmalignant) tissues and its induction in cancers: induction of PFKFB-4 protein expression is much stronger in malignant tumors from organs which have low levels of PFKFB-4 protein in nonmalignant tissues.
There is data that the PFKFB-3 isoenzyme is highly expressed in transformed cells and malignant tissues, suggesting that it may contribute to the high glycolytic rate observed in tumors (Chesney et al., 1999;Atsumi et al., 2002;Minchenko et al., 2005a;2005b).Interestingly, the level of PFKFB-4 protein in gastric, lung, breast and colon cancers is much higher as compared to the PFKFB-3 isoenzyme.Thus, the PFKFB-4 isoenzyme should be considered a tumorspecific enzyme in the lung, breast, gastric and colon cancers and may also contribute to the high glycolytic rate observed in tumors, much like PFKFB-3.There is also data that a molecule of PFKFB-3 in which EDTA is covalently linked to ADP is a good starting molecule for the development of new cancer-therapeutic molecules (Kim et al., 2006).It is important to note that most isoenzymes of the PFKFB gene family may find clinical utility as novel targets for the development of new anticancer agents.
In conclusion, this study provides evidence that PFKFB-4 and PFKFB-3 genes are expressed in gastric and pancreatic cancer cells and strongly respond to hypoxia via a HIF-1α dependent mechanism but no strong correlation is present between PFKFB-3 or PFKFB-4 mRNA and protein expression in these cancer cells, both in normoxic and hypoxic conditions.Moreover, there is an inverse correlation between the basal levels of PFKFB-4 protein expression and the hypoxic responsiveness of PFKFB-4 mRNA expression, although the biological significance of this is currently unknown and warrants further investigations.Hypoxic induction of HIF-1α protein in these cell lines correlates with a reduction of HIF-1α mRNA expression.

Figure 1 .
Figure 1.Effect of hypoxia and hypoxia mimic dimethyloxalylglycine on the expression of PFKFB-3 and PFKFB-4 mRNA in human gastric cancer cell lines MKN45 and NUGC3 and pancreatic cancer cell line Panc1, measured by ribonuclease protection assay.The cells were exposed to hypoxia (H) or treated with dimethyloxalylglycine (I) for 6 h.N, control (normoxic) cells.Intensities of different mRNA bands were normalized to 18S rRNA.The data are representative of four experiments.

Figure 2 .
Figure 2. The effect of hypoxia on PFKFB-4 and PFKFB-3 mRNA expression, measured by ribonuclease protection assay, in human gastric cancer cell lines MKN45 and NUGC3 and pancreatic cancer cell line Panc1.The cells were exposed to hypoxia (H) for 6 h.N, control (normoxic) cells.Intensities of different mRNA bands were normalized to 18S rRNA.The bar heights are mean values ± standard errors of the mean.

Figure 3 .
Figure 3.Western analysis of the expression of PFKFB-4 protein in human gastric cancer cell lines MKN45 and NUGC3 and in human pancreatic cancer cell line Panc1: effect of hypoxia (H) and dimethyloxalylglycine (I).The cells were exposed to hypoxia or treated with dimethyloxalylglycine for 6 h.N, control (normoxic) cells.Intensities of PFKFB-4 protein bands were normalized to β-actin.The data are representative of four experiments.

Figure 4 .
Figure 4. Effect of hypoxia on the expression of HIF-1α protein in human gastric cancer cell lines MKN45 and NUGC3 and in pancreatic cancer cell line Panc1, measured by Western analysis.The cells were exposed to hypoxia (H) for 6 h.N, control (normoxic) cells.Intensities of HIF-1α protein bands were normalized to β-actin.The data are representative of three experiments.

Figure 5 .
Figure 5.Effect of hypoxia and hypoxia mimic dimethyloxalylglycine on the expression of HIF-1α and HIF-2α mRNA, measured by ribonuclease protection assay, in human gastric cancer cell line MKN45 and NUGC3.The cells were exposed to hypoxia (H) or treated with dimethyloxalylglycine (I) for 6 h.N, control (normoxic) cells.Intensities of different mRNA bands were normalized to 18S rRNA.The data are representative of three experiments.

Figure 6 .
Figure 6.Effect of hypoxia and hypoxia mimic dimethyloxalylglycine on the expression of VEGF and Glut1 mRNA in human gastric cancer cell lines MKN45 and NUGC3 and pancreatic cancer cell line Panc1, measured by ribonuclease protection assay.The cells were exposed to hypoxia (H) or treated with dimethyloxalylglycine (I) for 6 h.N, control (normoxic) cells.Intensities of different mRNA bands were normalized to 18S rRNA.The data are representative of three experiments.

Figure 7 .
Figure 7. Expression of PFKFB-1, PFKFB-2, PFKFB-3, PFKFB-4, Glut1 and VEGF mRNA in human gastric malignant tumors (T) and corresponding nonmalignant tissue (N) from the same patients, as measured by ribonuclease protection assay.Intensities of different mRNA bands were normalized to 18S rRNA.The data are representative of three experiments.