Sequential activation of hypoxia-inducible factor 1 and specificity protein 1 is required for hypoxia-induced transcriptional stimulation of Abcc8

Abstract

Cerebral ischemia causes increased transcription of sulfonylurea receptor 1 (SUR1), which forms SUR1-regulated NC(Ca-ATP) channels linked to cerebral edema. We tested the hypothesis that hypoxia is an initial signal that stimulates transcription of Abcc8, the gene encoding SUR1, via activation of hypoxia-inducible factor 1 (HIF1). In the brain microvascular endothelial cells, hypoxia increased SUR1 abundance and expression of functional SUR1-regulated NC(Ca-ATP) channels. Luciferase reporter activity driven by the Abcc8 promoter was increased by hypoxia and by coexpression of HIF1α. Surprisingly, a series of luciferase reporter assays studying the Abcc8 promoter revealed that binding sites for specificity protein 1 (Sp1), but not for HIF, were required for stimulation of Abcc8 transcription by HIF1α. Luciferase reporter assays studying Sp1 promoters of three species, and chromatin immunoprecipitation analysis in rats after cerebral ischemia, indicated that HIF binds to HIF-binding sites on the Sp1 promoter to stimulate transcription of the Sp1 gene. We conclude that sequential activation of two transcription factors, HIF and Sp1, is required to stimulate transcription of Abcc8 following cerebral ischemia. Sequential gene activation in cerebral ischemia provides a plausible molecular explanation for the prolonged treatment window observed for inhibition of the end-target gene product, SUR1, by glibenclamide.

Introduction

The sulfonylurea receptor 1 (SUR1)-regulated NC_Ca-ATP channel has a critical role in the formation of edema induced by cerebral ischemia. Blockade of SUR1-regulated channels by the specific, highly selective sulfonylurea inhibitor, glibenclamide, confers significant protection in rat models of stroke, including reductions in lesion volume, edema, brain swelling, and mortality from malignant cerebral edema (Simard et al, 2006, 2009, 2010).

The treatment window for glibenclamide in rat models of stroke exceeds 6 hours (Simard et al, 2009, 2010). The prolonged treatment window for glibenclamide is thought to result from the fact that SUR1-regulated NC_Ca-ATP channels are not constitutively expressed, but must be transcriptionally upregulated for their pathophysiological effects to be manifested (Simard et al, 2006, 2009, 2010). After an ischemic insult, 3 hours are required before mRNA for Abcc8, the gene that encodes SUR1, is increased 2.5-fold, and 8 hours are required before SUR1 protein is increased 2.5-fold.

Transcriptional mechanisms responsible for SUR1 upregulation after cerebral ischemia are poorly understood. The transcription factor, specificity protein 1 (Sp1), has a critical role in basal expression of the gene (Ashfield and Ashcroft, 1998; Hernandez-Sanchez et al, 1999). In rat models of stroke, Sp1 increases within 2 hours of onset of ischemia, and the nuclear abundance of Sp1 increases 3.5-fold by 3 hours, with no change in Sp3 (Simard et al, 2006; Yeh et al, 2011). The increase in abundance of Sp1 is accompanied by an increase in binding of Sp1 to the Abcc8 promoter (Simard et al, 2006). However, the specific role of Sp1, and the mechanism for its increase, have not been determined.

Hypoxia-inducible factor 1 (HIF1) and HIF2 are transcription factors that regulate hypoxia-inducible genes (Wenger et al, 2005). Hypoxia-inducible factors form heterodimeric complexes consisting of an α and a β subunit (Wang et al, 1995). Three HIFα subunits (HIF1α, HIF2α/endothelial PAS domain protein 1, and HIF3α), and three HIFβ subunits (HIF1β/ARNT1, ARNT2, and ARNT3) are known. The most widely expressed α subunit in mammalian tissues is HIF1α, with the other HIFα subunits having more specialized or tissue-specific functions (Semenza, 2000). Hypoxia-inducible factor α proteins possess an oxygen-dependent degradation domain and two transcription activation domains. Under normoxic conditions, two conserved proline residues in the oxygen-dependent degradation domain are hydroxylated by prolyl hydroxylase. The hydroxylated HIFα proteins are recognized, polyubiquitinated, and undergo degradation through the 26S proteasome (Lee et al, 2004). Under conditions of hypoxia, proteosomal degradation of HIFα ceases, which allows it to dimerize with omnipresent, naturally stable HIFβ subunits and translocate to the nucleus (Huang et al, 1996). Once formed, the heterodimer binds to the DNA sequence, 5′-RCGTG-3′ (where R is A or G), which forms the hypoxia response elements (HREs) in the 5′-flanking regions (promoters) of target genes.

Hypoxia-inducible factor has been identified as a critical element in the transcriptional program activated by cerebral ischemia (Witt et al, 2005; Li et al, 2007). One or more of several stimuli associated with ischemia, including hypoxia, oxyhemoglobin desaturation, reactive oxygen species, nitric oxide, and inflammatory stimuli may be responsible for HIF1α stabilization (Hellwig-Burgel et al, 1999; Wenger et al, 2005; Carver et al, 2007). In cerebral ischemia, HIF serves an important role in adaptive responses involving glucose metabolism, angiogenesis, cell survival, and other protective functions (Acker and Acker, 2004). However, HIF is also associated with potentially deleterious effects, including upregulation of pro-apoptotic genes and delayed neuronal cell death (Halterman et al, 1999; Wenger et al, 2005; Fan et al, 2009). Notably, pharmacological as well as gene suppression of HIF are protective in cerebral ischemia/hypoxia (Helton et al, 2005; Chen et al, 2007, 2008, 2010). Given the potentially harmful role of HIF in the context of cerebral ischemia/hypoxia, coupled with the harmful role of SUR1 in edema formation induced by cerebral ischemia, we hypothesized that HIF might be involved in transcriptional upregulation of Abcc8. Here, we report that HIF promotes transcription of Abcc8, but that it does so via a complex mechanism of sequential gene activation that requires Sp1.

Materials and methods

Rat Model of Ischemic Stroke

All experiments were performed in accordance with the relevant guidelines and regulations as stipulated in the United States National Institutes of Health Guide for the Care and Use of Laboratory Animals. The experimental protocol was approved by the Institutional Animal Care and Use Committee of the University of Maryland. We made every effort to minimize the number of rats used and their suffering. Fasted male Wistar rats (250 to 300 g; Harlan, Indianapolis, IN, USA) were anesthetized (60 mg/kg ketamine and 7.5 mg/kg xylazine intraperitoneally) and allowed to ventilate air spontaneously. Cerebral ischemia was induced by middle cerebral artery occlusion (MCAO), which we performed with laser Doppler flowmetry monitoring, as described (Simard et al, 2009). Four groups of rats were studied (five rats per group): (1) 15 minutes ischemia/no reperfusion; (2) 60 minutes ischemia/no reperfusion; (3) 105 minutes ischemia/1 hour reperfusion; and (4) 105 minutes ischemia/3 hours reperfusion. In all cases, the presence of ischemia was confirmed by >80% reduction in laser Doppler flowmetry signals. The first group (15 minutes MCAO) was used to examine early HIFα stabilization; the second group (60 minutes MCAO) was used for chromatin immunoprecipitation; the third and fourth groups (105 minutes MCAO/1 to 3 hours reperfusion) were used to examine SUR1 protein upregulation.

Cell Culture

Human brain microvascular endothelial cells (hBMECs) and endothelial cell medium were purchased (ScienCell Research Laboratories, Carlsbad, CA, USA). The hBMEC were maintained in the endothelial cell medium containing 5% fetal bovine serum, endothelial cell growth supplements, and penicillin/streptomycin solution. HepG2 and mouse brain microvascular endothelial (bEnd.3) cells were cultured in minimal essential medium and Dulbecco's modified eagle medium, respectively, containing 10% fetal bovine serum, 100 unit/mL of penicillin, and 100 μg/mL of streptomycin. All cells were cultured in 5% CO₂ at 37°C unless otherwise indicated.

For hypoxia experiments, the cells were transferred to a chamber (C-chamber, BioSperics, Lacona, NY, USA) equipped with an O₂ and a CO₂ controller (ProOx P110 O₂ controller and ProCO₂ P120 CO₂ controller; BioSperics). The chamber was placed in an incubator at 37°C. The CO₂ controller regulated the delivery of CO₂ and the O₂ controller regulated the delivery of N₂ to maintain an atmosphere in the chamber of 1.0%±0.1% O₂, 5.0%±0.1% CO₂, and 94% N₂.

For Sp1 inhibition experiments, bEnd.3 and HepG2 cells were treated with WP631 methanesulfate (Sigma, St Louis, MO, USA) (50 or 500 nmol/L for 24 hours). HepG2 cells exposed to tumor necrosis factorα (20 ng/mL for 24 hours) to activate nuclear factor-κB were used as a negative control.

Immunohistochemistry

Cryosections of the rat brains were processed and immunolabeled as described (Simard et al, 2009, 2010) using primary antibodies directed against: SUR1 (SC-5789; Santa Cruz Biotechnology, Santa Cruz, CA, USA), HIF1α (SC-12542), HIF2α (IW-PA1129; IHC World LLC, Woodstock, MD, USA), and rat IgG (SC-2011). Fluorescent-labeled, species appropriate secondary antibodies (Invitrogen, Carlsbad, CA, USA) were used for visualization. Omission of primary antibodies and competition with antigenic peptides, when available, were used as negative controls. Sections were coverslipped with polar mounting medium containing antifade reagent and the nuclear dye, 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen), and were examined using epifluorescence microscopy.

Patch Clamp Electrophysiology

Nystatin whole-cell recordings and inside–out single-channel recordings of SUR1-regulated NC_Ca-ATP channels were performed as described (Chen et al, 2003). All recordings were obtained using Cs⁺ as the charge carrier, to block all potassium channels including K_ATP channels.

Promoter Cloning and Luciferase Reporter Plasmid Construction

The ∼600 bp-long 5′-flanking regions of the human and rat Abcc8 genes (−521 to +62 and −570 to +80, respectively; numbers relative to the transcription start site) and ∼2 kb-long 5′-flanking regions of mouse, rat, and human Sp1 genes were amplified by polymerase chain reaction (PCR) and cloned into luciferase reporter plasmids (pGL3-basic). We used an overlapping PCR method (Aiyar et al, 1996) to mutate the hypoxia responsive element (5′-RCGTG-3′) and the Sp1-binding sites (5′-GGGCGG-3′) in the 5′-flanking regions of human Abcc8 and rat Sp1. Briefly, two rounds of PCR were performed; the first PCR round amplified two overlapping mutated DNA fragments surrounding the mutation sites, and the second PCR round generated full-length mutant fragments by joining the two DNA fragments from the first PCR round. All the mutated DNA fragments were cloned into pGL3-basic plasmids. To construct reporter plasmids containing various 5′-flanking regions of the human Abcc8 and rat Sp1 genes, PCR was performed using appropriate primers and the amplified DNA fragments were cloned into pGL3-basic plasmids. All the plasmids constructed using PCR were verified by sequencing before transfection.

Transfection and Analysis of Luciferase Expression

Photinus luciferase reporter plasmids containing various promoter regions of the Abcc8 and Sp1 genes were transfected into HepG2 cells using Lipofectamine 2000 (Invitrogen). To normalize transfection efficiency, each Photinus luciferase reporter construct was transfected along with control plasmid in which the cytomegalovirus promoter drives expression of Renilla luciferase. After transfection, the cells were maintained in normoxic conditions for 24 hours and then switched to hypoxic conditions or were maintained in normoxic conditions for another 24 hours. For HIF1α and Sp1 cotransfection, empty vector or expression plasmid containing HIF1α and Sp1 cDNA under the control of cytomegalovirus promoter was transfected along with the luciferase plasmids (Huang et al, 1996). The activities of the Photinus and Renilla luciferases in extracts of the transfected cells were measured using the Dual Luciferase Reporter Assay System (Promega, Madison, WI, USA). The Photinus luciferase activity was divided by the Renilla activity from the same sample to normalize the transfection efficiency. The divided values were then expressed relative to those from cells transfected with chicken β-actin promoter-driven luciferase in each condition. To obtain the fold induction of luciferase activity by hypoxia or HIF1α coexpression, the values with hypoxia or HIF1α were divided by those from corresponding control samples of normoxia or empty vector transfection (Woo et al, 2002).

RNA Isolation and Reverse Transcription Polymerase Chain Reaction

To determine the abundance of Abcc8 mRNA, total RNA was extracted from bEnd.3 cells using Trizol Reagent (Invitrogen), and the concentration of total RNA was determined by measuring the optical density at 260 and 280 nm. To avoid contamination by genomic DNA, the RNA was further purified with Amplification Grade DNase I (Invitrogen). cDNA was synthesized from 1 μg of total RNA of each sample using SuperScript III Reverse Transcriptase (Invitrogen). The abundance of Abcc8 mRNA in the samples was determined by real-time PCR (ABI PRISM 7300; Applied Biosystems, Carlsbad, CA, USA). The abundance of S18 mRNA was measured to normalize the samples. The primers used were 5′-GCCAGCTCTTTGAGCATTGG-3′ (forward) and 5′-AGGCCCTGAGACGGTTCTG-3′ (reverse) for Abcc8; 5′-CGGAAAATAGCCTTCGCCATCAC-3′ (forward) and 5′-ATCACTCGCTCCACCTCATCCT-3′ (reverse) for S18.

Chromatin Immunoprecipitation Analysis

The brain tissues from the ischemic MCA territory and from the contralateral MCA territory were dissected, minced, and fixed with 1% formaldehyde for 10 minutes at room temperature. Aliquots of chromatin were prepared and immunoprecipitated using rabbit anti-HIF1α antibody (Abcam, Cambridge, MA, USA) using a commercial kit (Active Motif, Carlsbad, CA, USA). Negative control samples were prepared by chromatin immunoprecipitation using control rabbit IgG. Immunoprecipitated chromatins were analyzed by PCR using primers for rat Sp1 promoter targeting individual HREs. The sequences of primers used were 5′-CCTGGTGGTTGACTGGTTTC-3′ (forward)/5′-GCTCTCTCTCTGAGGCGTGT-3′ (reverse) for HREa/b, 5′-TTCCAAGTTCCAGCCTTCAC-3′ (forward)/5′-CCGGAATAGGACAGGATCAA-3′ (reverse) for HREc, and 5′-GGCTTCACGAAGTTCTCCAG-3′ (forward)/5′-ACGAAAACCCAAGTCCACTG-3′ (reverse) for HREd. PCR settings included one cycle at 95°C for 5 minutes, followed by 35 cycles of 94°C for 1 minutes, 58°C for 1 minutes, and 72°C for 30 seconds, and terminated with one additional cycle at 72°C for 5 minutes. PCR products were analyzed by agarose gel electrophoresis.

Data Analysis

Data are expressed as mean±s.e. Comparisons between multiple groups (Figure 3) were made using one-way analysis of variance with a Fisher's post-hoc test; comparisons between pairs (Figures 4, 5, 6, before normalization) were made using Student's t-test. Values with P<0.05 were considered to be statistically significant.

Results

Sulfonylurea Receptor 1 in the Brain Capillaries Early After Ischemia

We examined SUR1 upregulation in the MCA territory after 105 minutes MCAO plus 1 to 3 hours reperfusion. Regions with early cerebral edema were identified by immunolabeling for extravasated endogenous IgG (Remmers et al, 1999; Simard et al, 2009) (Figure 1A). Sulfonylurea receptor 1 was found to be upregulated first in microvascular endothelial cells, where SUR1-regulated channels have been implicated in the formation of cerebral edema (Simard et al, 2006, 2009, 2010). Upregulation of SUR1 was evident as early as 1 hour after the start of reperfusion (Figure 1B, left panel). Double labeling for von Willebrand factor confirmed that the elongated structures involved were microvessels (Figure 1C). After 3 hours of reperfusion, expression of SUR1 became more widespread, with upregulation in neurons (Figure 1B, right panel; Figure 1D), as previously reported (Simard et al, 2006).

Figure 1.

Cerebral ischemia leads to early stabilization of hypoxia-inducible factor 1α (HIF1α) and upregulation of sulfonylurea receptor 1 (SUR1) in cerebral microvessels. (A, B) Immunolabeling of the middle cerebral artery (MCA) territory of the rat brain for IgG (A) or SUR1 (B) 1 and 3 hours after reperfusion, following 105 minutes ischemia; the pattern of labeling showing elongated structures is consistent with early involvement of microvessels (B, left); regions sampled in panel B are indicated with arrows in panel A. (C, D) Magnified views of double labeling experiments showing expression of SUR1 (red) in microvessels at 1 hour, colabeled with von Willebrand factor (green) (C), or in neurons at 3 hours, colabeled with NeuN (green) (D); superimposed images are also shown, with nuclei labeled using 4′,6-diamidino-2-phenylindole (DAPI) (right-hand panels). (E) Immunolabeling of the MCA territory of the rat brain for HIF1α 15 minutes after onset of ischemia; the pattern of labeling showing elongated structures is consistent with early involvement of microvessels (right panel). The data shown are representative of five experiments each.

Cerebral ischemia also was associated with HIF1α stabilization. At 15 minutes after onset of MCAO, HIF1α stabilization was observed only in microvessels (Figure 1E) but not in neurons, where it is reported to be upregulated at later times (Althaus et al, 2006; Li et al, 2007). Hypoxia-inducible factor 2α immunolabeling was not observed (not shown).

Hypoxia Is an Early Signal for Upregulation of Sulfonylurea Receptor 1 in the Brain Microvascular Endothelial Cells

Human brain microvascular endothelial cells were cultured under hypoxic condition (1% O₂ for 24 hours), resulting in nuclear accumulation of HIF1α (Figure 2A). Under the same hypoxic conditions, upregulation of SUR1 was evident in plasmalemmal membranes (Figure 2B). These findings with endothelial cells in vitro recapitulated the observations made on endothelial cells in vivo (Figures 1B and 1E).

Figure 2.

Hypoxia causes upregulation of functional sulfonylurea receptor 1 (SUR1)-regulated NC_Ca-ATP channels in the human brain microvascular endothelial cells (hBMEC). (A, B) Immunolabeling for hypoxia-inducible factor 1α (HIF1α) (A) and for SUR1 (B) under conditions of normoxia (20% O₂) or hypoxia (1% O₂ for 24 hours); HIF1α and SUR1 are shown in red, with nuclei labeled using 4′,6-diamidino-2-phenylindole (DAPI); nuclear localization of HIF1α is indicated by pink nuclei; the data shown are representative of five experiments each. (C, D) Whole-cell currents during ramp pulses (4/min; holding potential (HP), −50 mV) (C) or during continuous recording (D), before and after ATP-depletion by Na azide plus 2-deoxyglucose (2-DG) in cells exposed to normoxic (C, left) or hypoxic (1% O₂ for 24 hours) conditions (C, right; D); ‘difference currents' are also shown (red). (E) Single-channel recordings of inside–out patches with Cs⁺ as the principal cation, with channel openings inhibited by ATP on the cytoplasmic side; slope conductance, 34 pS; seven patches. (F) Whole-cell currents in a cell exposed to hypoxia (1% O₂ for 24 hours) during repeated ramp pulses (4/min; HP, −50 mV) (left) or during individual ramp pulses (right), before and after ATP-depletion by Na azide plus 2-DG, before and after addition of glibenclamide (300 nmol/L); the current after glibenclamide is shown in red. The whole-cell recordings in panels C, D and F are representative of five to seven recordings each.

Hypoxia-induced upregulation of SUR1 in hBMEC was associated with the appearance of functional ion channels with properties characteristic of SUR1-regulated NC_Ca-ATP channels (Chen et al, 2003; Simard et al, 2006). As previously, all patch clamp recordings were performed using Cs⁺ as the charge carrier, to block currents from potassium channels, including K_ATP channels. Because the SUR1-regulated NC_Ca-ATP channel is sensitive to the intracellular concentration of ATP, we used ATP-depletion to activate it. In control hBMEC not exposed to hypoxia, ATP-depletion did not activate any ionic current (Figure 2C, left panel), consistent with the absence of constitutive expression of the channel. After exposure to hypoxia (1% O₂ for 24 hours), ATP-depletion induced an inward current carried by Cs⁺ that reversed near 0 mV, characteristic of a nonselective cation channel (Figure 2C, right panel; Figure 2D). Single-channel recordings showed a single-channel conductance of 34 pS (Figure 2E). The current activated by ATP-depletion was blocked by the SUR1-selective blocker, glibenclamide (Figure 2F). The biophysical and pharmacological features of the channel induced by hypoxia in endothelial cells were identical to those of SUR1-regulated NC_Ca-ATP channels previously reported in neurons and astrocytes (Chen et al, 2003; Simard et al, 2006).

Together, these findings indicated that hypoxia, possibly acting via HIF, leads to an increase in abundance of SUR1 protein and de novo expression of SUR1-regulated NC_Ca-ATP channel in the brain microvascular endothelial cells.

Hypoxia and Hypoxia-Inducible Factor Stimulate Transcription of Abcc8

We speculated that upregulation of SUR1 was mediated through hypoxia-induced transcription of Abcc8 by HIF1. To investigate the role of hypoxia and HIF1 in transcriptional regulation of Abcc8, HepG2 cells were transfected with a luciferase reporter plasmid containing the 5′-flanking region of human Abcc8. Luciferase reporter plasmids containing the human EpoE (erythropoietin enhancer) (Huang et al, 1996) or the chicken β-actin promoter were used as positive and negative controls, respectively (Figure 3).

Figure 3.

Transcriptional stimulation of the Abcc8 promoter by hypoxia and by coexpression of hypoxia-inducible factor 1α (HIF1α). Relative luciferase activity of chicken β-actin promoter (Actin), erythropoietin enhancer (EpoE), and −521/+62 (numbers relative to the transcription start site) region of human Abcc8 promoter (Abcc8), each tested with cotransfection of empty vector (CTR), coexpression of HIF1α, after exposure to hypoxia alone, and after exposure to hypoxia plus cotransfection with HIF1α; n=5 for each promoter; ^*P<0.05; ^**P<0.01; ^***P<0.001.

As expected, luciferase activity driven by EpoE was increased by hypoxia and by coexpression of HIF1α (Huang et al, 1996). The combination of hypoxia plus HIF1α coexpression further stimulated the EpoE-driven luciferase activity by 5.5-fold. This effect was specific, as luciferase activity driven by the chicken β-actin promoter was not affected by hypoxia or by HIF1α coexpression.

Luciferase activity driven by the human Abcc8 promoter was also increased by hypoxia and by coexpression of HIF1α, and hypoxia combined with HIF1α coexpression yielded a cooperative effect, increasing luciferase activity by 3.2-fold. These data confirmed that hypoxia and HIF1 stimulate transcription of Abcc8.

Specificity Protein 1-Binding Sites, but not Hypoxia Response Elements, Mediate the Stimulatory Effect of Hypoxia-Inducible Factor 1 on Abcc8 Transcription

We speculated that direct binding of HIF1 to the 5′-flanking region of Abcc8 would be necessary for transcriptional stimulation of the gene, as shown for transcriptional regulation of erythropoietin by HIF (Semenza et al, 1991). This conjecture was supported by sequence analysis of the 5′-flanking regions of mouse, rat and human Abcc8, which revealed that each promoter contained several HREs, the canonical binding sites for HIF. In the human Abcc8 gene, five HRE are located within 600 bp of the transcription start site (Figure 4A). In the corresponding regions of the mouse and rat genomes, four HRE are present.

Figure 4.

Identification of the cis-regulatory elements responsible for the stimulatory effect of hypoxia-inducible factor (HIF) on Abcc8 transcription. (A–C) Schematic diagrams of luciferase reporter plasmids containing various regions of the human Abcc8 promoter (left), and relative luciferase activity (mean±s.e.) observed for each construct following cotransfection of empty vector (Vector) or coexpression of HIF1α (horizontal bars, right), showing: (A) the effect of mutations of the hypoxia response element (HRE); n=14; (B) the effect of serial deletion of the distal segment of the promoter; n=6; (C) the effect of mutations of the specificity protein 1 (Sp1)-binding sites; n=8; ^*P<0.05; ^**P<0.01. In the schematics, the numbers and symbols indicate cloned regions and mutations made in the reporter plasmids, with binding sites for HIF (HRE, filled ellipses), Sp1 (filled boxes), and for mutated sites (empty symbols) depicted; HRE and Sp1-binding sites were mutated from the consensus sequence, 5′-RCGTG-3′ (R is A or G) to 5′-RCaTa-3′, and from 5′-GGGCGG-3′ to GGGaaa-3′, respectively. (D) Fold induction of luciferase activity driven by the human Abcc8 promoter (−521/+62 region) in response to HIF1α coexpression and tumor necrosis factor α (TNFα) treatment (20 ng/mL), in the presence or absence of the Sp1 inhibitor, WP631 methanesulfonate (500 nmol/L); n=7. (E) Fold increase in Abcc8 mRNA in response to hypoxia (1% O₂ for 24 hours) in the presence or absence of WP631 methanesulfonate (50 nmol/L); n=3.

To test the hypothesis that HIF1 binds to the HRE in the Abcc8 promoter to stimulate transcription of the gene, the five HRE located in the human Abcc8 promoter, named HRE1 to HRE5, were inactivated singly and in combination by site-directed mutagenesis, with each mutation verified by sequencing (Figure 4A). Surprisingly, mutation (m) of each individually (mHRE1, mHRE2/3, mHRE4, and mHRE5) or of the five in combination (mHRE1–5) had no effect on luciferase activity stimulated by HIF1α coexpression; in all cases, coexpression of HIF1α still yielded a doubling of luciferase activity (Figure 4A).

To confirm the foregoing findings, and to identify the cis-regulatory element(s) involved in transcriptional stimulation of Abcc8 by HIF1α, we performed serial deletions of distal segments of the promoter (Figure 4B). Luciferase reporter plasmids containing various portions of the 5′-flanking region of the Abcc8 promoter were studied. Consistent with our previous findings, HIF1α coexpression increased luciferase activity in cells transfected with plasmids lacking all of the five HRE (−100/+62 and −50/+62); in all cases, coexpression of HIF1α still resulted in a doubling of luciferase activity (Figure 4B). Moreover, the observation that the −50/+62 promoter yielded a doubling of luciferase activity, similar to that observed with the −521/+62 promoter, indicated that it was unlikely that noncanonical HIF1-binding sites in the −521 to −50 region accounted for HIF1α-mediated activation of the Abcc8 promoter.

Further deletion of the promoter resulted in complete loss of luciferase activity. In cells transfected with the +1/+62 plasmid, HIF1α coexpression resulted in no increase in luciferase activity (not shown). These results indicated that the promoter region between −50 to +1 contains the cis-regulatory element(s) required for basal transcription as well as for mediating the stimulatory effect of HIF1α on Abcc8 transcription.

Sequence analysis of the human Abcc8 promoter identified five binding sites (5′-GGGCGG-3′) for the transcription factor, Sp1, in the −100 to +62 region (Figures 4B and 4C). Specificity protein 1 has a major role in Abcc8 transcription (Ashfield and Ashcroft, 1998; Hernandez-Sanchez et al, 1999; Xu et al, 2000; Simard et al, 2006).

To test the hypothesis that Sp1 is involved in the stimulatory effect of HIF1α coexpression on Abcc8 transcription, the five Sp1-binding sites in the −100/+62 plasmid were mutated (Figure 4C). Mutation of these sites resulted in a dramatic decrease in luciferase activity, confirming that Sp1 has a critical role in basal expression of Abcc8 (Ashfield and Ashcroft, 1998; Hernandez-Sanchez et al, 1999). Moreover, mutation of the Sp1 sites abolished the effect of HIF1α coexpression on luciferase activity (Figure 4C).

To further evaluate the role of Sp1-binding sites, we studied the effect of the Sp1-family-specific inhibitor, the bisintercalating anthracycline, WP631 methanesulfonate (Mansilla and Portugal, 2008), on the stimulatory effect of HIF1α on Abcc8 transcription. Coexpression of HIF1α-stimulated luciferase activity driven by the Abcc8 promoter 2.2-fold (Figure 4D). This stimulatory effect was abolished by WP631 methanesulfonate (Figure 4D). At the same concentration, WP631 methanesulfonate did not inhibit basal activity driven by the Abcc8 promoter itself (not shown), and did not affect the stimulatory effect of tumor necrosis factorα/nuclear factor-κB on Abcc8 transcription (Simard et al, 2007) (Figure 4D), consistent with a specific effect related to transcriptional stimulation by Sp1.

We also used WP631 methanesulfonate to investigate the role of Sp1 in Abcc8 transcription driven by hypoxia. We measured native Abcc8 transcript in the brain microvascular endothelial cells under normoxic versus hypoxic conditions. Exposure to hypoxia (1% O₂ for 24 hours) resulted in a fourfold increase in Abcc8 mRNA in vehicle-treated cells (Figure 4E). By contrast, in cells incubated with WP631 methanesulfonate, the Abcc8 response to hypoxia was blocked (Figure 4E).

Together, these results indicated that binding of Sp1 to its binding sites in the Abcc8 promoter is required to mediate the stimulatory effect of hypoxia and HIF1α on transcription of the Abcc8 gene. Moreover, the data with WP631 methanesulfonate in bEnd.3 cells suggest that other transcription factors known to regulate Abcc8, such as FoxA2, β2, and STAT3 (Kim et al, 2002; Gorogawa et al, 2004; Lantz et al, 2004), are unlikely to be involved in the effect of hypoxia.

Hypoxia-Inducible Factor Stimulates Transcription of Specificity Protein 1

We and others have shown that cerebral ischemia and hypoxia induce nuclear accumulation of Sp1 (Xu et al, 2000; Simard et al, 2006). Hypoxia-inducible factor has been reported to stimulate one of its target genes via transcriptional activation of the Sp1 gene (Tai et al, 2009, 2010), although specific interactions between HIF1 and the Sp1 promoter were not examined. Sequence analysis of the 5′-flanking regions of Sp1 from the mouse, rat, and human genomes revealed multiple HRE located in these regions (Figure 5A). We hypothesized that HIF stimulates transcription of Abcc8 indirectly via transcriptional activation of Sp1.

Figure 5.

Identification of cis-regulatory elements responsible for the stimulatory effect of hypoxia-inducible factor (HIF) on Sp1 (specificity protein 1) transcription. (A, B) Schematic diagrams of luciferase reporter plasmids containing ∼2 kb-long 5′-flanking regions of human, mouse, and rat Sp1 (A), and various regions of the rat Sp1 promoter (B) (left), and relative luciferase activity (mean±s.e.) observed for each construct following cotransfection of empty vector (Vector) or coexpression of HIF1α (horizontal bars, right), showing: (A) the stimulatory effect of HIF1α coexpression on Sp1 promoters in the three species; n=4; (B) the effect of serial deletion of the distal segment of the rat promoter; n=9; ^*P<0.05; ^**P<0.01. In the schematics, the numbers and symbols indicate cloned regions in the reporter plasmids, with binding sites for HIF (hypoxia response element (HRE), filled ellipses) depicted.

To test the hypothesis that HIF stimulates transcription of Sp1, we examined the effect of HIF1α coexpression on Sp1 transcription. Luciferase activities in cells transfected with reporter plasmids containing ∼2 kb-long 5′-flanking regions of mouse, rat, and human Sp1 were all increased two to threefold by coexpression of HIF1α (Figure 5A).

To identify the cis-regulatory element(s) involved in the stimulatory effect of HIF on Sp1 transcription, the distal segments of the rat Sp1 promoter were serially deleted from the luciferase reporter construct (Figure 5B). Six HRE, named HREa to HREf, are located within 2 kb upstream of the transcription start site of rat Sp1. Deletion of HREe and HREf did not affect luciferase activity stimulated by HIF1α coexpression (−1200/+46) (Figure 5B). By contrast, deletion of HREc and HREd dramatically reduced induction by HIF1α coexpression (−700/+46 and −500/+46) (Figure 5B). Luciferase activity from cells transfected with the −500/+46 plasmid was not significantly increased by HIF1α coexpression (Figure 5B). These results indicated that two HRE, HREc and HREd, are critical cis-regulatory elements responsible for the stimulatory effect of HIF on Sp1 transcription.

DNA-Binding Activity of Hypoxia-Inducible Factor to the Specificity Protein 1 Promoter Following Cerebral Ischemia

To examine the in vivo role of HIF in transcriptional regulation of Sp1, we determined the DNA-binding activity of HIF to the Sp1 promoter using tissues from a rat MCAO model. The sequence and location of the four HRE (HREa–d) examined are shown (Figure 6A). Chromatin immunoprecipitation using anti-HIF1α antibody was performed to measure DNA-binding activity of HIF to these four HRE; only three specific pairs of primers were used to identify involvement of the four HRE, since two of them, HREa and HREb, (‘HREa/b') overlap. Hypoxia-inducible factor was found to bind specifically to HREc and HREd, but not to HREa or HREb in the Sp1 promoter (Figure 6B). Moreover, the DNA-binding activity of HIF to HREc and HREd increased in ischemic regions, but not in contralateral regions (Figure 6B). Consistent with our previous findings (Figure 5), these results indicated that HIF binds to HREc and HREd in the rat Sp1 promoter to stimulate transcription of the gene following cerebral ischemia.

Figure 6.

In vivo confirmation of cis-regulatory elements responsible for the stimulatory effect of hypoxia-inducible factor (HIF) on Sp1 (specificity protein 1) transcription in cerebral ischemia. (A) Sequences and locations (relative to the transcription start site) of four hypoxia response element (HRE) in the rat Sp1 promoter. (B) Chromatin immunoprecipitation analysis using HIF1α antibody targeting the four HRE located in the rat Sp1 promoter; genomic DNA was isolated from the brain after ischemia (I) and from the contralateral side (C), and immunoprecipitated using rabbit anti-HIF1α antibody or rabbit control IgG; DNA before (Input) and after immunoprecipitation with the antibody indicated was amplified by PCR using primers targeting the four HRE, and the resulting DNA was analyzed on an agarose gel. (C) Relative luciferase activity driven by a 650-bp-long rat Abcc8 promoter (−570 to +80 relative to transcription start site), tested with cotransfection of empty vector (Vector) or coexpression of Sp1; n=9; ^*P<0.05.

Specificity Protein 1 Stimulates Transcription of Rat Abcc8

Specificity protein 1 has been shown to stimulate transcription of mouse and human Abcc8 (Ashfield and Ashcroft, 1998; Hernandez-Sanchez et al, 1999), but this has not been reported for rat Abcc8. We examined the effect of Sp1 coexpression on luciferase activity driven by a 650-bp-long 5′-flanking region of rat Abcc8 (Figure 6C). Luciferase activity driven by the rat Abcc8 promoter was increased 8.8-fold by Sp1 coexpression, confirming the role of Sp1 in this species as well.

Discussion

The major finding of the present study is that Abcc8 is a novel downstream target gene of HIF, with hypoxia and HIF stimulating Abcc8 gene expression. Previously, transcriptional regulation of Abcc8 had been studied almost exclusively in the context of diabetes, since SUR1 is the regulatory subunit of the K_ATP channel involved in insulin secretion in pancreatic β cells (Ashfield and Ashcroft, 1998; Hernandez-Sanchez et al, 1999). Several transcription factors, including Sp1, FoxA2/HNF3β, β2/NeuroD, and STAT3, were implicated in transcription of Abcc8 in pancreatic β cells (Ashfield and Ashcroft, 1998; Hernandez-Sanchez et al, 1999; Kim et al, 2002; Gorogawa et al, 2004; Lantz et al, 2004). Among these, Sp1 was found to have a critical role in basal expression of the gene (Ashfield and Ashcroft, 1998; Hernandez-Sanchez et al, 1999). Data presented here, as well as previously (Simard et al, 2006), reaffirm the critical role of Sp1 in expression of Abcc8 in rat, mouse, and human, and extend previous findings in pancreatic β cells to the brain microvascular endothelial cells.

The present report is the first to implicate HIF in the regulation of Abcc8. Nuclear accumulation of HIF1α was evident in microvessels as early as 15 minutes after onset of ischemia. After cerebral ischemia, the earliest upregulation of SUR1 was detected in microvascular endothelial cells, with subsequent expression of SUR1 expanding to involve other cells including neurons. Experiments with cultured cells demonstrated that hypoxia and HIF1 stimulate transcription of Abcc8 and cause upregulation of SUR1-regulated NC_Ca-ATP channels in the brain microvascular endothelial cells. Together, these data are consistent with previous observations that upregulation of SUR1 in endothelial cells has a key early role in cerebral edema following cerebral ischemia (Simard et al, 2006, 2009).

A second major finding of the present study is that the stimulatory effect of HIF on Abcc8 is mediated not directly but indirectly via upregulation of Sp1 transcription. Cooperative interactions between HIF and Sp1 have been reported to occur by one of two mechanisms: (1) hypoxia-induced activation of retinoic acid receptor-related orphan receptor a4 and glyceraldehyde-3-phosphate dehydrogenase genes depends on one or more HRE plus an adjacent Sp1-binding GC-box (Miki et al, 2004; Higashimura et al, 2011) and (2) hypoxia-induced activation of cyclooxygenase-2 and phenylethanolamine N-methyltransferase genes occurs via sequential activation of Sp1 transcription by HIF (Xu et al, 2000; Tai et al, 2009, 2010). Our experiments with multiple mutations and deletions of the Abcc8 and Sp1 promoters indicated that HIF regulation of Abcc8 occurs by the second of these mechanisms, a two-step, sequential activation process with Sp1 having a critical intermediary role. Our data expand on previous reports regarding this mechanism by identifying the two cis-regulatory elements on the Sp1 promoter that are involved.

Our data do not exclude the possibility that HIF interacts directly with the Abcc8 promoter to have a role in transcription. It remains possible that (1) an interaction occurs between HIF and Sp1 that amplifies the effect of Sp1 at Sp1-binding sites and (2) unidentified noncanonical HIF-binding sites could be present in the proximal promoter of Abcc8 that overlap with the Sp1-binding sites that we mutated. Also, our data do not completely exclude the possibility that HIF interacts with other transcription factors to influence Abcc8 transcription, although our finding that WP631 methanesulfonate blocks hypoxia-induced upregulation of Abcc8 mRNA in bEnd.3 cells (Figure 4E) suggests otherwise. Overall, our data indicate that Sp1 and the Sp1-binding sites in the proximal promoter region of Abcc8 are necessary and sufficient for the stimulatory effect of HIF on Abcc8 transcription.

Sequential transcriptional activation is a process that likely amplifies and prolongs the original signal from a short-lived transcription factor such as HIF, and also delays transcription of the end-target gene, here Abcc8. This mechanism provides a novel molecular explanation for the prolonged treatment window observed for inhibition of the end-target gene product, SUR1, by glibenclamide in cerebral ischemia (Simard et al, 2009, 2010).

Sulfonylurea receptor 1 forms the regulatory subunit of both the K_ATP and the NC_Ca-ATP channels. Transcriptional regulation of SUR1 is independent of transcriptional regulation of the pore forming subunits of these two channels, and appears to depend on context. Promoter analysis revealed no putative binding sites for HIF in the promoter regions of either Kir6.1 or Kir6.2, although one report indicated that Kir6.1, but not Kir6.2, is weakly upregulated by hypoxia (Melamed-Frank et al, 2001). In central nervous system ischemia and trauma, upregulation of SUR1 mRNA is not accompanied by upregulation of protein or mRNA for Kcnj8 and Kcnj11, which encode the pore forming subunits of K_ATP channels, Kir6.1 and Kir6.2, respectively (Chen et al, 2003; Simard et al, 2006; Patel et al, 2010). This pattern of transcription involving SUR1 but not Kir6.x was shown previously to be associated with de novo upregulation of SUR1-regulated NC_Ca-ATP channels (Simard et al, 2006), as we reaffirmed here in the brain microvascular endothelial cells exposed to hypoxia.

Specificity protein 1-mediated transcription can be regulated either by increasing Sp1 or by decreasing Sp3 (Discher et al, 1998; Suske, 1999). Cerebral ischemia leads to an increase in nuclear localization of Sp1 and increases the Sp1/Sp3 ratio by increasing Sp1 without decreasing Sp3 (Xu et al, 2000; Simard et al, 2006). Our data indicated a 3.5-fold increase in nuclear Sp1 measured 3 hours after onset of ischemia. Cerebral ischemia increases Sp1 by several redundant mechanisms. It was recently reported that an early rise in H₂O₂ during cerebral ischemia results in a 30% increase in Sp1 within 1 hour of onset of ischemia, with no change in mRNA levels or Sp1 stability, but via activation of an internal ribosome entry site that regulates rapid and efficient translation of existing mRNA (Yeh et al, 2011). In addition, HIF stimulates transcription of Sp1 (Tai et al, 2009, 2010), a finding that we confirmed here and that we expanded on by identifying the specific cis-regulatory elements involved. Finally, the early rise in Sp1 autoactivates the Sp1 promoter in a feed-forward manner (Nicolas et al, 2001), further increasing late transcription of Sp1. The existence of multiple redundant mechanisms suggests an important role for Sp1 in the response to cerebral ischemia.

In cerebral ischemia, Sp1 has a critical role in transcriptional regulation not only of SUR1 but also of caspase-3 (Liu et al, 2002). Caspase-3 is an end-executioner of apoptosis and has been linked to cell death in cerebral ischemia (Broughton et al, 2009). De novo expression of SUR1 is required to form functional SUR1-regulated NC_Ca-ATP channels, which mediate the formation of cerebral edema (Simard et al, 2006, 2009, 2010), and which act as end-executioners in oncotic (necrotic) cell death (Simard et al, 2006). Transcriptional regulation of these two molecular entities is undoubtedly complex, but to have both of these mechanisms of cell death under partial control of a single transcription factor argues strongly for the importance of Sp1 upregulation following ischemic injury to the brain.

In summary, we show that HIF has a major role in transcriptional activation of the Abcc8 gene whose activation is required for formation of NC_Ca-ATP channels implicated in pathological responses following cerebral ischemia. The finding that sequential gene activation involving Sp1 is required for the effect of HIF provides a novel molecular mechanism that may explain the prolonged treatment window observed with glibenclamide in the context of cerebral ischemia.

The authors declare no conflict of interest.