Hypoxia Induces High-Mobility-Group Protein I(Y) and Transcription of the Cyclooxygenase-2 Gene in Human Vascular Endothelium

From the Sealy Center for Molecular Cardiology, Department of Medicine, The University of Texas Medical Branch, Galveston, Tex.

	Abstract

Top Abstract Introduction Materials and Methods Results Discussion References

 
Abstract—Cyclooxygenases catalyze a rate-limiting step in the synthesis of vascular endothelial prostaglandins. Expression of the inducible cyclooxygenase-2 (COX-2) gene is increased by hypoxia in human vascular endothelial cells via the nuclear factor (NF)-B p65 transcription factor, which is necessary but not sufficient to fully induce COX-2 transcription in response to hypoxia. After finding that cytoplasmic NF-B p65 and IB (an inhibitory protein that binds NF-B p65 precursors) levels are not changed by hypoxia, we hypothesized that other factors might play a role in regulating the COX-2 promoter, like the high-mobility-group (HMG) I(Y) family of proteins, which features multiple A·T hooks and is associated with NF-B–mediated transactivation. Nuclear protein obtained from human umbilical vein endothelial cells (HUVECs) was supplemented with HMG I(Y) during electrophoretic mobility shift assays using an NF-B-3' element probe. These data suggested that HMG I(Y) proteins interact with NF-B p65 to induce COX-2 promoter activity. We also found that TATA-box DNA demonstrated increased electrophoretic shifting indicative of DNA binding after incubation with either hypoxic HUVEC nuclear protein or normoxic nuclear protein supplemented with HMG I(Y). Transfection of HUVECs with an expression vector containing the COX-2 promoter ligated to HMG I(Y) cDNA demonstrated positive feedback on COX-2 promoter activity in hypoxia. We confirmed that COX-2 is transcriptionally regulated by hypoxia using a nuclear runoff assay. Hypoxia increased steady-state cellular levels of HMG I(Y) mRNA as an early event, corresponding with increases in HMG I(Y) protein. Overexpression of HMG I(Y) was associated in a dose-response relationship with increasing prevalence of the COX-2 protein in hypoxic HUVECs. Furthermore, sense (and antisense) HMG I(Y) overexpression caused stimulation (or inhibition) of COX-2 promoter activity as measured by luciferase reporter gene expression. The physiological significance of these findings was demonstrated by cyclooxygenase-dependent release of prostaglandin E₂ by HUVECs in hypoxia. We concluded that hypoxia increases expression of HMG I(Y) proteins while facilitating transactivation of the COX-2 promoter. The HMG I(Y) family of proteins may therefore function as part of a hypoxia-induced enhanceosome that helps to promote transcription of COX-2.

Key Words: endothelium • hypoxia • HMG I(Y) • cyclooxygenase-2 • prostaglandin E₂

	Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

 
The high-mobility-group protein family (HMG) I(Y) is a group of nonhistone chromosomal proteins.^{1 2} HMG I(Y) binds to A·T-rich regions of DNA with A·T hooks,³ inducing DNA conformational changes. HMG I(Y) can also regulate the affinity of some transcriptional factors to their DNA-binding elements while interacting with other transcriptional regulators.^{4 5 6 7} As a architectural protein regulator, HMG I(Y) can either facilitate the binding of some factors to their DNA-binding elements and increase promoter activity or interfere with the binding of some other factors and inhibit promoter activity.^{8 9 10 11 12 13 14 15} The affinity of HMG I(Y) binding to DNA depends on the structure of the DNA binding site.^{4 16 17} Nuclear factor (NF)-B is among those factors considered to interact with HMG I(Y).⁸

The inducible cyclooxygenase, cyclooxygenase-2 (COX-2), is considered a rate-limiting enzyme in the synthesis of the prostaglandins from arachidonic acid.¹⁸ COX-2 can be induced rapidly by many physical and chemical stimuli in human endothelial cells, including shear stress, cobalt, mitogenic agents, and cytokines.^{19 20 21} The COX-2 gene has been cloned, and there are several potential response elements upstream from the transcription start site of the COX-2 gene, including Sp1, activator protein-2, NF-B, NF-IL6, cAMP response element, E box, and TATA.^{22 23} In addition, IL-1ß regulates the translation of COX-2 gene by binding the 3' untranslated region and stabilizing mRNA.²⁴ The NF-IL6 site is responsible for induction of COX-2 by lipopolysaccharides and phorbol ester in vascular endothelial cells.²² A cAMP response element mediates the effect of v-src on COX-2 expression in fibroblasts.²⁵ In a previous study, we showed that hypoxia induced the expression of COX-2 gene via NF-B p65 factor in human endothelial cells, demonstrating that hypoxia increases the binding of NF-B p65 to the relatively 3' consensus element in the upstream promoter of the COX-2 gene.¹⁶ We found that the binding of the NF-B is a necessary, if not sufficient, step for the response of COX-2 to hypoxia. This finding left us with questions as to what other factors could be conveying hypoxia-specific signals to the nucleus, making the p65 interaction with the COX-2 promoter a sufficient stimulus for transactivation.

We hypothesized that the increase in COX-2 gene expression by NF-B in hypoxia was sensitive to interaction between NF-B or TATA elements and the HMG I(Y) family of proteins given this background information. We examined the expression of HMG I(Y) gene under hypoxic conditions and found evidence indicating that HMG I(Y) plays a role in the induction of COX-2 expression in hypoxia.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

 
Cell Culture and Hypoxia Treatment
Human umbilical vein endothelial cells (HUVECs) (Clonetics) were cultured in medium 199 with 2.2 g NaHCO₃/L (Life Technology, Inc), 5% FBS (Hyclone Laboratories), 50 µg/mL endothelial cell growth supplement (Collaborative Biomedical Products), 50 µg/mL heparin, and 1.0 µg/mL hydrocortisone. HUVECs were studied within 5 passages from primary culture. HUVECs were exposed to ambient oxygen concentration of 1% (hypoxia) and 21% (normoxia) when the cells were grown to 60% to 80% confluence. The medium was preequilibrated to the environmental gas conditions overnight before cellular exposure. The possibility of generating reoxygenation artifacts after removal of the cells from the incubator was carefully prevented by immediately placing cells on ice and replacing hypoxic medium with lysis buffer.

Construction of Plasmids
The full-length human HMG I cDNA (pBS-HMG-I) was a gift from Dr Raymond Reeves. A set of deletion constructions of the COX-2 promoter, extending from -1800, -304, -245, and -45 bp upstream from the transcription start site to +65, were produced with polymerase chain reaction. These deletions were inserted into vector pGL2-basic (Promega) containing a luciferase reporter gene and are referred to as pD4, pD3, pD2, and pD1, respectively. An HMG I(Y) expression plasmid pD3-HMG was driven by the COX-2 promoter through replacement of luciferase reporter gene (HindIII-EcoRV fragment) of pD3 with the HMG I cDNA insert of pBS-HMG-I. The pSV40-HMG(+) and pSV40-HMG(-) plasmids were generated by removing the luciferase reporter gene in pGL-2 promoter (Promega) through restriction endonuclease digestion with EcoRV and HindIII and subsequently ligating HMG I cDNA in sense (+) or antisense (-) orientation with the SV40-driven remnant of pGL-2. The correctness of the recombinant plasmids was confirmed by restriction digest analyses.

Western Blotting
Cell culture dishes were briefly washed with prechilled PBS before adding cell lysis buffer (50 mmol/L HEPES [pH 7.4], 10 mmol/L KCl, 1 mmol/L EDTA, 1 mmol/L EGTA, 1 mmol/L DDT, 0.5 mmol/L phenylmethylsulfonyl fluoride, 1 µg/mL pepstatin A, 1 µg/mL leupeptin, 10 µg/mL trypsin inhibitor, 10 µg/mL aprotinin, and 0.5% Triton). Protein concentration was measured using Bio-Rad protein assay reagent. After electrophoresis in 12% SDS-PAGE gel, the 30 µg cell lysate was blotted to a nitrocellulose membrane. Nonspecific binding was blocked by soaking the membrane in a TBST buffer (20 mmol/L Tris base [pH 7.5], 137 mmol/L NaCl, and 0.1% Tween 20) containing 5% nonfat milk for 30 minutes at 37°C. The membrane was then incubated with polyclonal IgG to NF-B p65 (1:1500 dilution), the inhibitory protein that binds NF-B p65 precursors (IB, 1:1000 dilution), or HMG I(Y) (1:1000 dilution) (Santa Cruz Biotechnology) for 30 minutes at 37°C and washed in TBST for 5 minutes (5 times each). After a similar incubation with horseradish peroxidase–conjugated anti-IgG, the membrane was washed and incubated in luminol ECL detection reagents (Amersham) and exposed to the film. Incubation with IgG monoclonal antibody to chicken -tubulin (Sigma) at 1:2500 dilution was also performed for comparative purposes.

Electrophoretic Mobility Shift Assay
Nuclear protein extraction and gel shift assays were performed as previously described for NF-B-3'.¹⁶ The duration of HUVEC exposure to hypoxia was 2 hours. The binding reaction was performed with end-labeled oligonucleotide from the COX-2 promoter and 2 to 6 µg of nuclear extract. The sense strand of the double-stranded TATA box oligo used for electrophoretic mobility shift assay (EMSA) was also taken from the COX-2 promoter region: 5'-TTGGTTTTCAGTCTTATAAAAAGG-3'.

Transfection for Reporter Gene Assay
We performed transfection analysis using luciferase reporter gene constructions in HUVECs, which were grown in 35-mm 6-well plates to 70% to 80% confluence. For each well, 12 µL of lipofectin reagent (GIBCO BRL) was added into 88 µL of Opti-MEM-1 medium (Life Technologies, Inc) and incubated at room temperature for 45 minutes. Then 2.0 µg of pD1-pD4 and 0.5 µg of pSV2PAP, containing the alkaline phosphatase reporter gene downstream from an SV40 promoter, were added into 100 µL Opti-MEM-1 medium. The lipofectin and DNA solution were mixed and incubated at room temperature for 15 minutes, and then another 800 µL of Opti-MEM-1 medium was added to the mixture. After the HUVECs were rinsed once with the Opti-MEM-1 medium, the lipofectin and DNA mixture was overlaid on the cells and incubated at 37°C for 8 hours. We then replaced the mixture with the original medium and let the cells recover overnight. HUVEC cells were exposed to hypoxia or further normoxia for 24 hours before reporter gene assay. An additional 0.5 to 4.0 µg of HMG I expression plasmid (as indicated in Figure 5) was added into the DNA mixture. A control plasmid containing only the COX-2 promoter but no reporter gene was used to make the DNA quantity constant. Luciferase activity was determined in cell lysates through single photon counting on a microplate scintillation counter. Alkaline phosphatase activity was determined as previously described to control transfection efficiency.¹⁶ The values for luciferase (counts per second) and alkaline phosphatase (optical density) were reported as a ratio, thereby correcting for discrepancies in transfection efficiency.

View larger version (24K): [in this window] [in a new window]   Figure 5. Reporter gene assays using deletion constructs demonstrate that HMG I(Y) upregulates the expression of COX-2 under hypoxic conditions. HUVECs were cotransfected with either 2 µg of pD3, a luciferase reporter that demonstrates full COX-2 promoter activity in hypoxia (A and C), or pD1, a similar luciferase reporter ligated to a basal COX-2 promoter (B), and either 0.5 to 4 µg of pD3-HMG expression plasmid (A and B) or the constitutively active pSV-HMG(+) overexpression vector in the sense orientation (C). The nonexpressing plasmid pBS-HMG was used to control DNA concentration so that the total for all conditions was equal. Also, 0.5 µg of the alkaline phosphatase plasmid pSV2PAP was used in each experiment for correcting the efficiency of transfection. Four to six independent transfections were performed for each dose. Cell lysates were harvested 24 hours after treatment with hypoxia and assayed for luciferase activity and alkaline phosphatase activity. Data were expressed as mean±SEM of LUC/PAP ratios.

Northern Blotting
Total RNA was extracted from HUVECs treated with normoxia and hypoxia with the use of RNAzol B (TEL-TEST, Inc). Total RNA (20 µg/well) was loaded in a 1.2% formaldehyde agarose gel. After electrophoresis, the nucleic acids were transferred to a nylon membrane by blotting. A SacI fragment of HMG I cDNA (1186 bp) was randomly labeled with ³² P as a probe. Randomly labeled 18S RNA was also used for comparative purposes.

Nuclear Runoff Assay for COX-2 Transcripts
Nuclei were harvested from HUVECs as follows: cells were rinsed twice with ice-cold PBS, and then 5 to 8x10⁷cells were scraped into a 50-mL tube with 4 mL of ice-cold PBS, centrifuged for 5 minutes at 500g and 4°C, loosened by vortexing gently, and then lysed with 4 mL of NP-40 lysis buffer (10 mmol/L Tris-HCl [pH 7.5], 1.5 mmol/L MgCl₂, 140 mmol/L NaCl, and 1% NP-40). We incubated lysed cells for 3 to 5 minutes on ice, checked the cell lysate with a phase-contrast microscope, and then centrifuged the nuclei at 500g at 4°C for 5 minutes. The nuclei were washed twice in 1 mL of 20 mmol/L Tris-HCl (pH 8.0), 20% glycerol, 140 mmol/L KCl, 10 mmol/L MgCl₂, 1 mmol/L MnCl₂, and 14 mmol/L ß-mercaptoethanol. After centrifugation at 650g for 4 minutes, we resuspended the nuclei in 200 µL of this buffer and stored them in liquid nitrogen. Frozen nuclei were thawed to 25°C, and nascent transcripts were labeled with 20 µL of 10 mCi/mL [-³²P]UTP(800 Ci/mmol) and 2.5 µL of 100 mmol/L ATP, CTP, and GTP for 30 minutes at 30°C. RNA was extracted from the nuclei by adding 0.9 mL RNazol B and 0.12 mL chloroform. The aqueous content was separated, and RNA was precipitated with isopropanol and washed with 70% ethanol. The RNA was resuspended in 100 µL Tris-EDTA solution and 1 µL of RNase inhibitor. Limited alkaline hydrolysis was achieved with the addition of 20 µL of 1 mol/L NaOH and incubation on ice for 10 minutes, followed by neutralization with 40 µL of 1 mol/L HEPES at pH 7.9. RNA was precipitated with sodium acetate and ethanol. RNA was hybridized to nitrocellulose to which DNA for COX-1 and COX-2, along with a negative control, had been slot-blotted. The radiolabeled RNA transcripts that hybridized to the membrane were identified by exposure to film for 4 days.

GST-HMG Fusion Protein
The full-length HMG I(Y) cDNA coding region was inserted into the glutathione-S-transferase (GST) fusion protein expression vector pGEX-4T-1 by ligation to EcoRI and SmaI restriction sites. Escherichia coli BL21, transformed with the recombinant plasmid, was grown in LB medium supplemented with 100 µg/mL ampicillin at 37°C overnight. The cells were then diluted 1:10 in fresh LB and grown for another hour. Expression of the GST-HMG fusion protein was induced by incubation with 1.0 mmol/L isopropyl-ß-D-thiogalactoside for 4 hours. The cells were then resuspended in PBS and lysed by sonication. The proteins were brought into solution with 1% Triton X-100. A crude extract was separated by centrifugation and added to a 50% slurry of glutathione Sepharose 4B equilibrated with PBS. After gentle agitation at room temperature for 30 minutes, the matrix was sedimented and washed with PBS. Then GST-HMG was eluted by 10 mmol/L reduced glutathione, and the fusion protein was analyzed by SDS-PAGE.

Radioimmunoassay of PGE₂
HUVEC cells were grown to 70% to 80% confluence and exposed to 2 hours, 6 to 10 hours, 24 hours, and 48 hours of normoxic or hypoxic conditions. Immediately after stimulation, the culture medium was collected into prechilled polypropylene tubes coated with a solution of 4.5 mmol/L EDTA combined with 10 µg/mL indomethacin to inhibit further prostaglandin synthesis. Prostaglandin E₂ (PGE₂) levels were determined using a commercially available ¹²⁵I radioimmunoassay kit (Dupont). PGE₂ levels were assayed directly from medium as follows: 100 µL sample, 100 µL rabbit anti-serum, and 100 µL tracer (iodinated analogue of PGE₂) were incubated together overnight at 4°C. Then 1 mL cold precipitating reagent (16% PEG 6000 and 0.05% sodium azide in 50 mmol/L phosphate buffer, pH 6.8) was added to precipitate the antibody-bound tracer. After centrifugation, the pellet containing the antibody-antigen complex was counted in a gamma counter. Results from a serial dilution of standard concentrate were used to construct standard (dose-response) curves from which the unknowns were read by interpolation. The value in the medium at time zero was subtracted from each value to yield an index of cellular prostanoid production above that baseline over time. Protein concentration in each cell culture dish was measured after 2 washings with PBS. NaOH (0.62 mol/L) was added to solubilize the protein. The protein concentrations from each dish were then measured with a modified Bradford method (Bio-Rad) using BSA as a standard.²⁶ The results of PGE₂ measurements were reported as picograms per milligram of protein.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

 
Search for Hypoxia-Related Regulatory Elements in COX-2 Promoter
Hypoxia is known to regulate transcription via multiple factors. For instance, hypoxia-inducible factor 1 (HIF-1), which mediates erythropoietin induction in hypoxia, also regulates the transcription of genes encoding glycolytic enzymes. The presence of a HIF-1 binding site is necessary but not sufficient to direct the hypoxia-induced transcription of these genes.²⁷ There are several potential cis-acting elements within the 350 bases upstream from the COX-2 transcription start site (Figure 1A). The known hypoxia-inducible element 1 (HIE-1), which binds HIF-1, has not been found upstream or downstream from the COX-2 coding region. We performed several transfections with a set of deletions of the COX-2 promoter to search for a hypoxia regulatory element (Figure 1B). The D4 region is the intact COX-2 promoter. It displayed full COX-2 promoter activity and a full reporter gene response to hypoxia induction when transfected in a vector. Transfection using pD3, which features 1496 bp less of the COX-2 upstream promoter than does D4, showed no significant decrease of the promoter activity. The smallest construct, pD1, containing the basal promoter of COX-2 and a TATA box, displayed a minimal response to hypoxia. Although insignificant in itself, this minimal response suggested to us that hypoxia may induce the COX-2 gene through not only the NF-B binding site but also the TATA box and other DNA-protein binding sites.

View larger version (21K): [in this window] [in a new window]   Figure 1. The search for hypoxia-related regulatory elements within the COX-2 promoter region. A, Schematic representation of the promoter regions of COX-223 and COX-139 relative to transcriptional start sites. Potential factor binding sites are shown on both the direct (up) and complementary (down) strands within -350 bp. CRE indicates cAMP response element. B, Transfection of HUVECs with plasmids driven by COX-2 promoter deletion constructs ligated to the luciferase reporter gene. Regions from -1800 bp, -304 bp, -245 bp, and -45 bp to +65 bp of the COX-2 promoter were cloned into pGL2-basic as pD4, pD3, pD2, and pD1, respectively. After transfection by 2 µg of each construct and 0.5 µg of pSV2PAP and subsequent stimulation by hypoxia for 24 hours, luciferase activity (LUC) and alkaline phosphatase activity (PAP) were assayed. The results are presented as ratios of LUC to PAP. Descriptive statistics (mean±SEM) were based on 3 to 5 independent experiments.

Mechanism of NF-B Activation by Hypoxia
Previous work from our laboratory indicated that hypoxia activates the binding of NF-B p65 to a consensus NF-B oligonucleotide. We noticed that the NF-B 5'-binding site in the COX-2 promoter, which is -440 bp upstream from the transcription start site, did not have any contribution to the COX-2 promoter activity but did constitutively bind with normoxic and hypoxic HUVEC nuclear protein extracts. By comparison, hypoxia enhances protein binding to the NF-B-3' binding site in the COX-2 promoter. We examined IB- and NF-B p65 by Western blot in HUVEC cell lysates (Figure 2). -Tubulin was used as an equivalent loading control. The protein level of both factors did not change between normoxia and hypoxia (from 30 minutes to 24 hours). This suggests that the activation of NF-B by hypoxia is regulated by posttranslational modification and/or binding factors such as IB- in HUVECs. We reasoned that other factors like HMG I(Y) may play a role in the hypoxia-specific binding of NF-B transactivation factors to certain NF-B elements. This hypoxia-specific interaction might occur through cooperative binding of HMG I(Y) to the A·T region, facilitating binding of NF-B p65 to the NF-B-3' element.

View larger version (46K): [in this window] [in a new window]   Figure 2. Western blot analysis of NF-B p65 and IB- protein in HUVECs. Each lane contains 30 µg cell lysate from HUVECs treated with a time course of hypoxia and normoxia as indicated. The same blot was hybridized to anti–NF-B p65 and then to anti–IB- antibody. The hybridization to anti–-tubulin was used as a loading control. NF-B and IB- did not show any change during hypoxia. The results represent 3 independent experiments.

Hypoxia, HMG I(Y), and Nuclear Protein Binding to NF-B-3' and TATA
Hypoxic nuclear protein caused our radiolabeled NF-B-3' oligonucleotide probe to shift electrophoretically. The combination of nuclear protein from hypoxic cells and the GST-HMG fusion protein produced a synergistically enhanced shift, as seen in Figure 3. The range of GST-HMG protein used in a series of EMSA experiments revealed a maximal shift at 1 to 3 ng with a definite decrease in NF-B-3' shift above this amount. We also recorded that a GST protein without HMG had no effect on electrophoretic mobility. When hypoxic nuclear protein was used in this experiment, the shift intensified and was observed at lower doses of GST-HMG. When a TATA oligo was used for EMSA (Figure 4), multiple bands were shifted, corresponding to the shift of HMG I(Y) and TATA-binding protein (TBP) recorded in a previous study of TATA DNA-protein interactions.¹² Hypoxia induced increased shifting, as did HMG I(Y), suggesting a functional correlation between hypoxic induction of transcription and DNA binding with HMG I(Y) proteins at the TATA box as well as at the NF-B-3' element of the COX-2 promoter. However, we cannot conclude that this DNA-protein interaction is of functional significance given the limitations of EMSA; therefore, further experiments were performed to examine the effect of HMG I(Y) proteins on COX-2 promoter function and gene expression.

View larger version (74K): [in this window] [in a new window]   Figure 3. Hypoxic nuclear protein (H) and HMG I(Y) both bind to NF-B by EMSA, with the use of 2x104 cpm NF-B-3' oligonucleotide as a probe. Nuclear protein of HUVECs treated with continued normoxia or 2 hours of hypoxia was assayed: 2.7 µg was used in lanes 1 to 9, 4.0 µg was used in lanes 11 to 13, and 2.5 µg was used in lanes 14 to 16. Lanes 1 to 4 demonstrate the synergy between hypoxic nuclear protein and GST-HMG fusion protein in shifting the NF-B-3' probe. NF-B-3' was maximally shifted in normoxic nuclear protein (N) experiments by 3.0 ng GST-HMG fusion protein (range, 0.3 to 30.0 ng in lanes 5 to 8). When hypoxic nuclear protein was added to NF-B-3' probe and GST-HMG (0.3 to 30 ng in lanes 14 to 16), more shifting occurred with less GST-HMG. A 100-fold excess of cold NF-B-3' oligo was added to block specific binding in lane 9. BSA (5 µg) was added to lane 10 instead of nuclear protein. Lanes 11 to 13 demonstrate that GST protein (without the HMG fusion) has no effect on DNA-protein binding. S indicates specific shift bands; NS, nonspecific shift band.

View larger version (82K): [in this window] [in a new window]   Figure 4. Hypoxia and HMG I(Y) both increase nuclear protein binding to the COX-2 promoter TATA box by EMSA, with the use of 2x104 cpm double-stranded TATA oligonucleotide as a probe. The top band appears to be attributable to HMG I(Y) binding to TATA DNA. Nuclear protein of HUVECs (6.0 µg) treated with continued normoxia or 2 hours of hypoxia was used. Lanes 1 to 2 demonstrate increased DNA-protein interaction in hypoxia. Lanes 3 to 6 show that HMG I(Y) protein can mimic the effect of hypoxia in a dose-dependent manner. Cold probe (100x) in lane 7 blocks shifting. Heat treatment of the DNA-protein binding product at 60°C for 5 minutes (lanes 8 to 9) reduced the amount of HMG I(Y)–attributable gel shifting. Lanes 10 and 11 are controls, in which BSA and HMG I(Y) protein were used alone. S indicates specific shift bands; NS, nonspecific shift band; N, normoxic nuclear protein; and H, hypoxic nuclear protein. Note that the nonspecific band (lanes 8 and 9) is heat labile when compared with the specific bands.

Upregulation of COX-2 Promoter Activity by HMG I(Y)
To understand how HMG I(Y) regulates the expression of the COX-2 gene under hypoxia, we constructed an HMG I expression plasmid by replacing the luciferase gene with HMG I cDNA in the COX-2 promoter construct pD3, making pD3-HMG, in which the expression of the HMG I gene was driven by the COX-2 promoter. COX-2 promoter constructs ligated to luciferase reporters or HMG I were used to cotransfect the HUVECs. Expression of HMG I(Y) remarkably increased the COX-2 promoter activity in a dose-dependent manner under hypoxia (Figure 5A). Since HMG I(Y) itself is upregulated by hypoxia, as described later, the synergistic effect of HMG I(Y) and hypoxia greatly increased the expression of reporter gene compared with baseline. This stimulation is conspicuous and as strong as that of phorbol ester and lipopolysaccharide²² or 20% FCS.²⁰ pD1 contains the basal promoter of the COX-2 gene. We found that increased HMG I(Y) could also upregulate the basal promoter activity of COX-2 in cotransfected HUVECs under hypoxia (Figure 5B). This effect was not dependent on the linkage between hypoxia-sensitive elements in the COX-2 promoter and HMG I(Y) expression, as modeled with the pD3-HMG vector. Repeating these experiments with a constitutively active expression vector, pSV40-HMG(+), demonstrated that a sufficient amount of HMG I(Y) could stimulate reporter gene expression by the COX-2 promoter as much as hypoxia could (Figure 5C).

Transcription of COX-2 in Hypoxia
Hypoxia had been previously shown to regulate the transcription of COX-2 via the NF-B p65 transactivation protein. We confirmed this finding using a nuclear runoff assay by hybridization of radiolabeled nascent transcripts to immobilized pBS-hgCOX-2-3', containing the 3' end (bases 4490 to 10488) of the human genomic COX-2 sequence.²⁸ The positive control was demonstrated by hybridization to pBS-hCOX-1, containing a 1750-bp insert of COX-1 cDNA²⁹ ; the negative control was obtained using cDNA for pBluescript II SK(+). Figure 6 illustrates the comparison between COX-1 and COX-2 transcripts obtained from nuclei from HUVECs that had been treated for 30 minutes, 2 hours, and 4 hours of hypoxia versus normoxia. Although transcription of the COX-1 gene did not change during hypoxia, transcription of the COX-2 gene was increased by hypoxia.

View larger version (26K): [in this window] [in a new window]   Figure 6. Nuclear runoff of nascent RNA transcripts reveals increased transcription of the COX-2 gene in hypoxia. Radiolabeled RNA from nuclei obtained from HUVECs that had been exposed to normoxia or hypoxia was hybridized to immobilized pBS-hgCOX-2-3', containing the 3' end (bases 4490 to 10488) of the human genomic COX-2 sequence.28 The positive control was demonstrated by hybridization to pBS-hCOX-1, containing a 1750-bp insert of COX-1 cDNA.29 COX-2 transcription was demonstrably increased during hypoxia; no effect was seen on the positive control COX-1. The negative control pBluescript II SK(+) is not shown.

Regulation of HMG I(Y) Gene Expression by Hypoxia in HUVECs
We investigated how hypoxia regulated the HMG I(Y) gene expression in HUVECs. Data are shown in Figure 7. We found that steady-state levels of HMG I(Y) mRNA were nearly doubled by hypoxia by 2 hours (Figure 7A). The peak prevalence of HMG I(Y) mRNA was reached between 2 and 6 hours with the hypoxia treatment and subsequently tailed off. The change in mRNA was relatively small compared with the >5-fold in increase HMG I(Y) protein caused by hypoxia (Figure 7B). We observed that protein prevalence also began to decrease after 6 hours of hypoxia.

View larger version (31K): [in this window] [in a new window]   Figure 7. Upregulation of HMG I(Y) gene expression by hypoxia in HUVECs and the consequences of this upregulation on COX-2 expression. A, Northern blot analysis of HMG I(Y) mRNA. Total RNA (20 µg) was loaded in each lane. The total RNA was extracted from HUVECs treated with normoxia and a time course of hypoxia as indicated. Hybridization of 18S RNA was performed as a comparative control using a probe with fewer radioactive counts than the HMG I(Y) probe for clarity. The steady-state levels of HMG I(Y) mRNA nearly doubled during hypoxia. This blot is representative of 3 separate experiments; the bar graph represents the corresponding mean±SEM phosphorescent image density corrected for 18S as an index of lane loading (*P<0.05 vs normoxia). B, Western blot analysis of HMG I(Y) protein. Whole-cell lysate (30 µg) was loaded in each lane. HUVECs were treated with normoxia and a time course of hypoxia before lysis. Hybridization with anti–-tubulin was performed as a lane-loading control. HMG I(Y) protein was increased 5-fold by hypoxia, as illustrated in a bar graph of relative band density (mean±SEM, corrected for -tubulin) from 3 separate experiments (*P<0.05 vs normoxia).

Effects of Sense and Antisense HMG I(Y) on COX-2 Gene Expression
We expressed sense and antisense HMG I(Y) RNA by transfecting HUVECs with constitutively active vectors [pSV40-HMG(+) and pSV40-HMG(-), respectively]. The effect of pSV40-HMG(+) on hypoxia-mediated increases in COX-2 immunoreactive protein are illustrated in Figure 8. All cells were exposed to equal amounts of DNA through balance transfection with empty vectors. This was important, since we found that transfected cells produced less COX-2 than did cells that had not been transfected with a given amount of DNA. We found that overexpression of HMG proteins led to progressive increases in COX-2 due to hypoxia. Since there was virtually no COX-2 produced at baseline, we could not decrease COX-2 by antisense overexpression in our model. However, using relative reporter gene activity as an index as illustrated in Figure 9, we documented that pSV40-HMG(-) inhibited and pSV40-HMG(+) stimulated luciferase reporter gene activity linked to the COX-2 promoter. These experiments indicate a specific role for HMG I(Y) proteins in the regulation of COX-2 gene expression by hypoxia.

View larger version (44K): [in this window] [in a new window]   Figure 8. Effect of increasing HMG I(Y) expression on COX-2 protein in hypoxic HUVECs. Cells were transfected with 1.0 µg total DNA [pSV40-HMG(+) or a blank pGL-2]. COX-2 protein was progressively increased by hypoxia and HMG overexpression. The bar graph demonstrates the direction of the dose-response relationship of hypoxia, HMG I(Y), and COX-2 immunoreactive protein (corrected for -tubulin) from 3 separate experiments.

View larger version (17K): [in this window] [in a new window]   Figure 9. Cotransfection of HUVECs with pD3, a luciferase promoter vector ligated to the COX-2 promoter region, and either pSV40-HMG(-) or pSV40-HMG(+) is illustrated in a bar graph of luciferase reporter gene activity, corrected for alkaline phosphatase (AP) activity, from 3 separate experiments. We found that the antisense HMG I(Y) vector led to decreased COX-2 promoter activity and that the sense HMG I(Y) vector increased COX-2 promoter activity (*P<0.05 vs control).

Endothelial PGE₂ in Hypoxia
The radioimmunoassay for PGE was performed on HUVEC media directly after 2 hours, 10 hours, 24 hours, and 48 hours of hypoxic or normoxic stimulation. Direct assay of cell culture medium for immunoactive PGE₂ levels has been reported previously.³⁰ Assays on the normoxic cells were performed as controls for basal synthesis of PGE₂. Parallel hypoxic cells at each point were also pretreated with 4 µg/mL indomethacin to block the cyclooxygenase activity. As shown in Figure 10, the PGE₂ level was not changed after 2 hours of hypoxic stimulation but was slightly increased by hypoxia after 10 hours and 24 hours. PGE₂ was significantly increased after 48 hours of hypoxia compared with normoxia. Furthermore, pretreatment of cells with indomethacin eliminated the hypoxia-induced enhancement of the PGE₂ synthesis. The time course of the PGE₂ response also corresponds to that of COX-2 protein induction by hypoxia,¹⁶ suggesting that the enhancement of the PGE₂ synthesis by hypoxia is dependent on increased COX-2 protein.

View larger version (25K): [in this window] [in a new window]   Figure 10. HUVEC production of PGE2 is increased during hypoxia by a cyclooxygenase-dependent mechanism. Cells were incubated for 2 hours, 10 hours, 24 hours, and 48 hours under normoxic or hypoxic conditions. Some hypoxic cells were also pretreated with 4 µg/mL indomethacin to block the cyclooxygenase activity. PGE2 levels were assayed by radioimmunoassay of culture medium. PGE2 levels from normoxic cells are presented as a control. Results are expressed in picograms per milligram protein as the mean±SEM (n=3). *P<0.05 compared with normoxia and hypoxia+indomethacin.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

 
The present study reveals for the first time that HMG I(Y) is upregulated by hypoxia in HUVECs. The time course of the response indicates that HMG I(Y) upregulation is an early event. The data herein suggest that HMG I(Y) proteins are constituents of an enhanceosome that regulates hypoxia-mediated transcription of COX-2 in a manner analogous to the synergistic roles of HMG I(Y) and NF-B protein factors in induction of the human interferon-ß gene.^{14 31} The specific role of the HMG I(Y) protein family is indicated by direct effects of this protein on DNA-protein interactions and COX-2 gene expression, along with the ability of antisense RNA of HMG I(Y) to decrease COX-2 promoter activity.

The cyclooxygenases catalyze the rate-limiting step in the synthesis of prostaglandins from arachidonic acid. Since the time course of the PGE₂ response to hypoxia in our experiments corresponds with that of COX-2 protein in hypoxia¹⁶ and the change in PGE₂ is inhibited by the cyclooxygenase inhibitor indomethacin, we propose that hypoxia increases the PGE₂ level in human vascular endothelial cells via inducible COX-2 synthesis. Other studies have shown that the transcription of the HMG I(Y) gene is inducible in human lymphoid cells by phorbol esters and calcium ionophore stimulation.² However, the time course of the HUVEC HMG I(Y) response to hypoxia is shorter than that in response to phorbol ester in lymphoid tissue. HMG I(Y) has been linked to carcinogenesis in pulmonary hamartomas³² and uterine leiomyomata.³³ We speculate, on the basis of our data, that HMG I(Y), COX-2, and PGE₂ are part of a proliferative, proinflammatory, and angiogenic response to hypoxia in cardiovascular diseases.

COX-2 mRNA metabolism has been linked to the "AUUUA" motif in its 3' untranslated region.²³ There does not appear to be such a motif in the HMG I(Y) 3' untranslated region.¹ Therefore, another negative-feedback mechanism may be responsible for the reduction of HMG I(Y) RNA in HUVECs after prolonged hypoxia (24 hours). The transcriptional regulation afforded by HMG I(Y) is a function of the prevalence of the HMG I(Y) protein.^{17 34 35} Our results have shown that HMG I(Y) protein facilitates nuclear protein binding with NF-B-3' and with TATA box elements. The increased HMG I(Y) expressed in hypoxic HUVECs appears to alter the conformation of the protein-DNA complex and the interaction between the basal transcription complex and promoter activity of genes including COX-2. Beyond a certain point, excess HMG I(Y) diminishes both NF-B-3' and TATA binding to nuclear protein in hypoxia. This finding correlates with previous work documenting repression of transcription by excess HMG I(Y).⁹

High-level expression of HMG I(Y) is also found during embryonic development.³⁶ Previous studies demonstrate that HMG I(Y) stimulates the binding of Sp1, NF-B, and activating transcription factor-2 to DNA and that it can interact with TBP.^{5 9 10 31} We propose that HMG I(Y) mediates the induction of COX-2 by hypoxia on 3 levels: (1) HMG I(Y) may bind to the TATA box and interact with TBP, altering basal promoter activity; (2) HMG I(Y) may facilitate Sp1, NF-B, and activating transcription factor-2 binding to DNA as a cofactor, increasing COX-2 enhanceosome function at the promoter; and (3) binding of HMG I(Y) to A·T-rich elements may change the conformation of DNA itself, focal-ly altering the helical structure of DNA and thereby acting as a facilitator (or, when in excess, an inhibitor) of enhanceosome-mediated transcription.

HMG I(Y) is a unique multifunctional regulator because it interacts with both DNA and many other factors. HMG I(Y) is known to cooperate with NF-B and compete with NF-AT factors in binding to DNA.¹¹ HMG I(Y) can directly recognize and alter the structure of localized regions of chromatin,^{17 35} thereby facilitating, inhibiting, or replacing the binding of some factors.^{9 11 12 37 38} The interaction between HMG I(Y) and NF-B p65 in the induction of COX-2 transcription by hypoxia provides new insight into the mechanisms of promoter-specific transcriptional regulation and cell-specific gene expression. Our data do not point to a one-protein/one-enhancer mechanism of transcription. In the case of COX-2, the enhanceosome model provides a template for our proposal that hypoxia-mediated transcription in vascular endothelium occurs when a specific combination of protein factors binds to specified sites within a promoter region.

	Acknowledgments

 
This study was supported by the John Sealy Memorial Endowment Fund for Biomedical Research. The authors extend their gratitude to Marschall S. Runge for his helpful advice.

	Footnotes

 
Reprint requests to John F. Schmedtje, Jr, MD, Section on Cardiology, Wake Forest University School of Medicine, Medical Center Blvd, Winston-Salem, NC 27157.

Received December 30, 1997; accepted May 6, 1998.

	References

Top Abstract Introduction Materials and Methods Results Discussion References

 

1. Johnson KR, Lehn DA, Reeves R. Alternative processing of mRNAs encoding mammalian chromosomal high-mobility-group proteins HMG-I and HMG-Y. Mol Cell Biol. 1989;9:2114–2123.[Abstract/Free Full Text]
2. Friedmann M, Holth LT, Zoghbi HY, Reeves R. Organization, inducible-expression and chromosome localization of the human HMG-I(Y) nonhistone protein gene. Nucleic Acids Res. 1993;21:4259–4267.[Abstract/Free Full Text]
3. Reeves R, Nissen MS. The AT-DNA-binding domain of mammalian high mobility group I chromosomal proteins. J Biol Chem. 1990;265:8573–8582.[Abstract/Free Full Text]
4. Maher JF, Nathans D. Multivalent DNA-binding properties of the HMG-I proteins. Proc Natl Acad Sci U S A. 1996;93:6716–6720.[Abstract/Free Full Text]
5. Falvo JV, Thanos D, Maniatis T. Reversal of intrinsic DNA bends in the IFNß gene enhancer by transcription factors and the architectural protein HMG I(Y). Cell. 1995;83:1101–1111.[Medline] [Order article via Infotrieve]
6. Leger H, Sock E, Renner K, Grummt F, Wegner M. Functional interaction between the POU domain protein Tst-1/Oct- 6 and the high-mobility-group protein HMG-I/Y. Mol Cell Biol. 1995;15:3738–3747.[Abstract]
7. John S, Reeves RB, Lin J-X, Child R, Leiden JM, Thompson CB, Leonard WJ. Regulation of cell-type-specific interleukin-2 receptor -chain gene expression: potential role of physical interactions between Elf-1, HMG-I(Y), and NF-B family proteins. Mol Cell Biol. 1995;15:1786–1796.[Abstract]
8. Wood LD, Farmer AA, Richmond A. HMGI(Y), and Sp1 in addition to NF-B regulate transcription of the MGSA/GRP gene. Nucleic Acids Res. 1995;23:4210–4219.[Abstract/Free Full Text]
9. Du W, Maniatis T. The high mobility group protein HMG I(Y) can stimulate or inhibit DNA binding of distinct transcription factor ATF-2 isoforms. Proc Natl Acad Sci U S A. 1994;91:11318–11322.[Abstract/Free Full Text]
10. French SW, Schmidt MC, Glorioso JC. Involvement of a high-mobility-group protein in the transcriptional activity of herpes simplex virus latency-active promoter 2. Mol Cell Biol. 1996;16:5393–5399.[Abstract]
11. Klein-Hessling S, Schneider G, Heinfling A, Chuvpilo S, Serfling E. HMG I(Y) interferes with the DNA binding of NF-AT factors and the induction of the interleukin 4 promoter in T cells. Proc Natl Acad Sci U S A. 1996;93:15311–15316.[Abstract/Free Full Text]
12. Ge H, Roeder RG. The high mobility group protein HMG1 can reversibly inhibit class II gene transcription by interaction with the TATA-binding protein. J Biol Chem. 1994;269:17136–17140.[Abstract/Free Full Text]
13. Abdulkadir SA, Krishna S, Thanos D, Maniatis T, Strominger JL, Ono SJ. Functional roles of the transcription factor Oct-2A and the high mobility group protein I/Y in HLA-DRA gene expression. J Exp Med. 1995;182:487–500.[Abstract/Free Full Text]
14. Thanos D, Maniatis T. The high mobility group protein HMG I(Y) is required for NF-B-dependent virus induction of the human IFN-ß gene. Cell. 1992;71:777–789.[Medline] [Order article via Infotrieve]
15. Aizawa S, Nishino H, Saito K, Kimura K, Shirakawa H, Yoshida M. Stimulation of transcription in cultured cells by high mobility group protein 1: essential role of the acidic carboxyl-terminal region. Biochemistry. 1994;33:14690–14695.[Medline] [Order article via Infotrieve]
16. Schmedtje JF Jr, Ji Y-S, Liu W-L, DuBois RN, Runge MS. Hypoxia induces cyclooxygenase-2 via the NF-B p65 transcription factor in human vascular endothelial cells. J Biol Chem. 1997;272:601–608.[Abstract/Free Full Text]
17. Reeves R, Wolffe AP. Substrate structure influences binding of the non-histone protein HMG-I(Y) to free and nucleosomal DNA. Biochemistry. 1996;35:5063–5074.[Medline] [Order article via Infotrieve]
18. Goppelt-Struebe M. Regulation of prostaglandin endoperoxide synthase (cyclooxygenase) isozyme expression. Prostaglandins Leukot Essent Fatty Acids. 1995;52:213–222.[Medline] [Order article via Infotrieve]
19. Wu KK. Cyclooxygenase 2 induction: molecular mechanism and pathophysiologic roles. J Lab Clin Med. 1996;128:242–245.[Medline] [Order article via Infotrieve]
20. Xie WL, Herschman HR. Transcriptional regulation of prostaglandin synthase 2 gene expression by platelet-derived growth factor and serum. J Biol Chem. 1996;271:31742–31748.[Abstract/Free Full Text]
21. Zembowicz A, Jones SL, Wu KK. Induction of cyclooxygenase-2 in human umbilical vein endothelial cells by lysophosphatidylcholine. J Clin Invest. 1995;96:1688–1692.
22. Inoue H, Yokoyama C, Hara S, Tone Y, Tanabe T. Transcriptional regulation of human prostaglandin-endoperoxide synthase-2 gene by lipopolysaccharide and phorbol ester in vascular endothelial cells: involvement of both nuclear factor for interleukin-6 expression site and cAMP response element. J Biol Chem. 1995;270:24965–24971.[Abstract/Free Full Text]
23. Appleby SB, Ristimäki A, Neilson K, Norko K, Hla T. Structure of the human cyclo-oxygenase-2 gene. Biochem J. 1994;302:723–727.
24. Srivastava SK, Tetsuka T, Daphna-Iken D, Morrison AR. IL-1ß stabilizes COX II mRNA in renal mesangial cells: role of 3'-untranslated region. Am J Physiol.. 1994;267:F504–F508.[Abstract/Free Full Text]
25. Xie W, Fletcher BS, Andersen RD, Herschman HR. v-src induction of the TIS10/PGS2 prostaglandin synthase gene is mediated by an ATF/CRE transcription response element. Mol Cell Biol. 1994;14:6531–6539.[Abstract/Free Full Text]
26. Bradford MM. A rapid and sensitive method for the quantitation of microgram quantities of protein using the principle of protein-dye binding. Anal Biochem. 1976;72:248–254.[Medline] [Order article via Infotrieve]
27. Semenza GL, Jiang B-H, Leung SW, Passantino R, Concordet J-P, Maire P, Giallongo A. Hypoxia response elements in the aldolase A, enolase 1, and lactate dehydrogenase A gene promoters contain essential binding sites for hypoxia-inducible factor 1. J Biol Chem. 1996;271:32529–32537.[Abstract/Free Full Text]
28. Hla T, Neilson K. Human cyclooxygenase-2 cDNA. Proc Natl Acad Sci U S A. 1992;89:7384–7388.[Abstract/Free Full Text]
29. Eberhart CE, Coffey RJ, Radhika A, Giardiello FM, Ferrenbach S, DuBois RN. Up-regulation of cyclooxygenase 2 gene expression in human colorectal adenomas and adenocarcinomas. Gastroenterology. 1994;107:1183–1188.[Medline] [Order article via Infotrieve]
30. Andley UP, Hebert JS, Morrison AR, Reddan JR, Pentland AP. Modulation of lens epithelial cell proliferation by enhanced prostaglandin synthesis after UVB exposure. Invest Ophthalmol Vis Sci. 1994;35:374–381.[Abstract/Free Full Text]
31. Thanos D, Maniatis T. Virus induction of human IFN ß gene expression requires the assembly of an enhanceosome. Cell. 1995;83:1091–1100.[Medline] [Order article via Infotrieve]
32. Xiao S, Lux ML, Reeves R, Hudson TJ, Fletcher JA. HMGI(Y) activation by chromosome 6p21 rearrangements in multilineage mesenchymal cells from pulmonary hamartoma. Am J Pathol. 1997;150:901–910.[Abstract]
33. Williams AJ, Powell WL, Collins T, Morton CC. HMGI(Y) expression in human uterine leiomyomata: involvement of another high-mobility group architectural factor in a benign neoplasm. Am J Pathol. 1997;150:911–918.[Abstract]
34. Kel OV, Romaschenko AG, Kel AE, Wingender E, Kolchanov NA. A compilation of composite regulatory elements affecting gene transcription in vertebrates. Nucleic Acids Res. 1995;23:4097–4103.[Abstract/Free Full Text]
35. Nissen MS, Reeves R. Changes in superhelicity are introduced into closed circular DNA by binding of high mobility group protein I/Y. J Biol Chem. 1995;270:4355–4360.[Abstract/Free Full Text]
36. Chiappetta G, Avantaggiato V, Visconti R, Fedele M, Battista S, Trapasso F, Merciai BM, Fidanza V, Giancotti V, Santoro M, Simeone A, Fusco A. High level expression of the HMG I(Y) gene during embryonic development. Oncogene. 1996;13:2439–2446.[Medline] [Order article via Infotrieve]
37. Zappavigna V, Falciola L, Citterich HM, Mavilio F, Bianchi ME. HMG1 interacts with HOX proteins and enhances their DNA binding and transcriptional activation. EMBO J. 1996;15:4981–4991.[Medline] [Order article via Infotrieve]
38. Farnet CM, Bushman FD. HIV-1 cDNA integration: requirement of HMG I(Y) protein for function of preintegration complexes in vitro. Cell. 1997;88:483–492.[Medline] [Order article via Infotrieve]
39. Yokoyama C, Tanabe T. Cloning of human gene encoding prostaglandin endoperoxide synthase and primary structure of the enzyme. Biochem Biophys Res Commun. 1989;165:888–894.[Medline] [Order article via Infotrieve]