Hypoxia-inducible factor (HIF)-1 is a transcription factor that regulates the expression of various neuroprotective genes. The goal of this study was to clarify the relationship between HIF-1 expression and subarachnoid hemorrhage (SAH) and to characterize the effects of deferoxamine (DFO)-induced increases in HIF-1 protein levels on the brainstem and the basilar artery (BA) after experimental SAH.
Rat single- and double-hemorrhage models (injected on Days 0 and 2) of SAH were used. We assessed the time courses for HIF-1 protein levels in the brainstems and the BA diameters within 10 minutes and 6 hours on Days 1 and 2 in the single-SAH model, and also on Day 7 in the double-SAH model. After induction of double hemorrhage in rats, DFO was injected intraperitoneally. We then evaluated HIF-1 protein expression and brainstem activity, BA diameter, and brainstem blood flow.
After the rats experienced SAH, HIF-1 protein expression was significantly greater at 10 minutes in the single-injection model and at 7 days in the double-injection model than at similar time points in the control group, and these increases correlated with degrees of cerebral vasospasm. DFO injection resulted in significant increases in HIF-1 protein expression and activity in the brainstems of rats with SAH, compared with the rats with SAH that were given placebos, and the rats without SAH in the double-hemorrhage model. Cerebral vasospasm and reduction of brainstem blood flow were significantly attenuated in the rats that were administered DFO.
These results show that a DFO-induced increase in HIF-1 protein level and activity exerts significant attenuation of BA vasospasm and reduction of brainstem blood flow in the rat model of SAH. DFO may be a promising agent for treating clinical SAH.
Cerebral vasospasm after ruptured aneurysmal subarachnoid hemorrhage (SAH) is a great indicator of poor prognosis. Some reports indicate that 13.5% of patients with SAH die or are disabled as a result of vasospasm (17). The limited understanding of the pathophysiological mechanisms of vasospasm in the context of SAH has resulted in a lack of targeted therapies for this complication.
Hypoxia-inducible factor (HIF)-1, which is a heterodimer of α and β subunits, is a transcription factor that mediates various processes including oxygen homeostasis (29,30,32,34), development, ischemia, and tumor angiogenesis (29) and regulates more than 40 genes including erythropoietin, vascular endothelial growth factor (VEGF), and glucose transporter-1 (33). Furthermore, HIF-1α is induced in the contexts of cerebral ischemia (4,15) and cerebral hemorrhage (14), where it exerts a neuroprotective effect. However, there have been no reports that demonstrate the role of HIF-1 on the brainstem during the period of cerebral vasospasm caused by SAH.
Deferoxamine (DFO) is an iron chelator that is used clinically for treatment of primary or secondary hemochromatosis, and experimental studies suggest that DFO may have neuroprotective properties in association with brain ischemia (26) and edema after cerebral hemorrhage (10,22) and cerebral vasospasm caused by SAH (2,7,8,18,38). In addition, DFO has a pharmacological ability to stabilize HIF-1α protein (19,39). Overexpression of HIF-1α by an iron chelator (3,27) or by hypoxic preconditioning (3,28) induces tolerance against cerebral ischemia, and thrombin preconditioning attenuates erythrocyte- and iron-induced brain edema via HIF-1α protein accumulation (9).
In this study, we present the relationship between HIF-1 expression in the brainstem and the diameter of the major cerebral arteries of rats after SAH and, and we evaluate the effects of DFO-induced overexpression in HIF-1α protein level and activity on the brainstem and basilar artery (BA) after SAH.
MATERIALS AND METHODS
Experimental Model of Subarachnoid Hemorrhage
All animal experiments were performed according to the guidelines of the Institutional Animal Care and Use Committee of Okayama University. Male Sprague-Dawley rats that weighed 350 to 400 g were allowed free access to food and water. Rats were anesthetized by an intraperitoneal injection of sodium pentobarbital (1 mg/kg) and were allowed to breathe spontaneously. A midline skin incision was made from the middle of the calvarium to the lower cervical spine with rats placed in the prone position, and the atlanto-occipital membrane was exposed. A 26-gauge needle was inserted into the cisterna magna, and 0.3 mL of autologous arterial blood was injected under sterile conditions. After induction of SAH, rats were placed in a head-down position for 30 minutes to accumulate thick clots around the BA. Control animals received the same volume of saline solution instead of autologous blood.
To evaluate the relationship between HIF-1 expression and both acute transient and delayed chronic cerebral vasospasm, rat single- and double-hemorrhage models of SAH were used for this study (35). Rats were randomly assigned to one of four groups (Group 1, single hemorrhage; Group 2, single saline injection; Group 3, double hemorrhage; and Group 4, double saline injection). After anesthetization on Day 0, all groups received 0.3 mL of either autologous arterial blood or saline. Groups 3 and 4 underwent reanesthetization 2 days later (on Day 2) and were administered a second injection of either blood or saline. Rats in Groups 1 and 2 were decapitated either at 10 minutes (Group 1, n = 29 rats; Group 2, n = 14 rats), 6 hours (Group 1, n = 25 rats; Group 2, n = 11 rats), 24 hours/Day 1 (Group 1, n = 25 rats; Group 2, n = 10 rats), or 48 hours/Day 2 (Group 1, n = 28 rats; Group 2, n = 11 rats) after injection, and rats in Groups 3 and 4 were sacrificed 5 days after the second injection on Day 7 (Group 3, n = 21 rats; Group 4, n = 10 rats). Immediately after the rats were sacrificed, the brainstems were harvested and stored in liquid nitrogen for use in the protein and messenger ribonuclieic acid (mRNA) assays.
Vasospasm was assessed by measurement of the BA lumen area in rats from Groups 1 (10 min, 6 h, Days 1 and 2) and 3 (Day 7) (n = 5 rats per group). Rat tissues were fixed by perfusion of 140 mL of phosphate-buffered saline and subsequent infusion of 140 mL of 4% paraformaldehyde in phosphate-buffered saline at physiological blood pressure. The brains were removed and stored in 4% paraformaldehyde overnight and then soaked in 30% sucrose for 48 hours at 4°C. Frozen sections (10 μm) of the BA and brainstem were cryostatically generated, and BA areas were measured under a light microscope equipped with a micrometer. BA cross sections were evaluated to record measurements at two points, namely, the midpoint between the union of the vertebral arteries and the anterior inferior cerebellar arteries, and the midpoint between the anterior inferior cerebellar arteries and the tip of the BA. The mean of the two points was used as the BA area. The ratio of the BA areas to normal BA areas (before injection) was used to assess the degree of cerebral vasospasm.
A rat double-hemorrhage model was used to assess the effects of DFO (Sigma-Aldrich, St. Louis, MO) on the brainstem and BA during delayed vasospasm after SAH. The first injection of autologous arterial blood was performed on Day 0, and the second injection was performed on Day 2. On Day 4, rats were administered 300 mg/kg of DFO intraperitoneally (SAH-DFO group, n = 23 rats). Another set of rats was treated with distilled water (SAH-placebo group, n = 22 rats) and used as controls. Both groups were compared with Group 4 (double saline injection, n = 18 rats) to evaluate HIF-1α mRNA protein expression and activity, VEGF mRNA expression, and brainstem blood flow. BA areas were measured for rats from both the SAH-DFO and SAH-placebo groups (n = 5 rats per group) on Day 7 using the same methods described above, and the effects of DFO on BA were also assessed.
Reverse-Transcriptase Polymerase Chain Reaction
Total mRNA was extracted from the brainstems using TRIzol reagent (Invitrogen Life Technologies, Carlsbad, CA). The mRNA (1 μg) was used for reverse transcription with the SuperScript First-Strand Synthesis System (Invitrogen Life Technologies) in accordance with the manufacturer's recommendations to yield 20 μL of first-strand complementary deoxyribonucleic acid (DNA) solution. HIF-1α and VEGF mRNA levels in the brainstem were assessed via polymerase chain reaction (PCR) using 25 μL of the reverse-transcriptase reaction mixture (Qiagen, Valencia, CA) containing Taq DNA polymerase, 2× Qiagen PCR buffer, 3 mmol/L MgCl2, and 400 μmol/L of each deoxyribonucleotide triphosphate in a final volume of 50 μL. Amplification was performed in a DNA cycler (GeneAmp PCR System 9700; Applied Biosystems, Foster City, CA). Rat HIF-1 oligonucleotide primer sequences were 5′-AAG TCT AGG GAT GCA GCA C-3′ and 5′-CAA GAT CAC CAG CAT CTA G-3′. To amplify HIF-1α complementary DNA, samples were kept at 95°C for 5 minutes and then subjected to thermocycling (28 cycles of 30 s at 95°C, 30 s at 53°C, and 1.5 min at 72°C, with a final extension of 5 min at 72°C). Rat VEGF primer sequences were 5′-TGC ACC CAC GAC AGA AGG GGA-3′ and 5′-TCA CCG CCT TGG CTT GTC ACA T-3′. Samples were kept at 94°C for 4 minutes and then subjected to thermocycling (26 cycles of 30 s at 94°C, 30 s at 60°C, and 2 min at 72°C, with a final extension of 5 min at 72°C). Rat β-actin primer sequences (internal controls) were 5′-TTG TAA CCA ACT GGG ACG ATA TGG-3′ and 5′-GAT CTT GAT CTT CAT GGT GCT AGG-3′. Samples were kept at 94°C for 2 minutes and subjected to thermocycling (22 cycles of 2 min at 94°C, 40 s at 94°C, and 40 s at 60°C, with a final extension of 1 min at 72°C). PCR production was analyzed by electrophoresis on 1.5% agarose gel. Gels were visualized with 0.5% ethidium bromide staining and ultraviolet transillumination. Photographs were obtained with black and white film (Polaroid, Bedfordshire, England). Relative band densities were analyzed with NIH Image 1.61 software (National Institutes of Health, Bethesda, MD).
Western Blot Analysis
For HIF-1α immunoblots, frozen brainstem tissue was homogenized in lysis buffer (1% Nonidet P-40 [Sigma-aldrich], 0.5% sodium deoxycholate, and 0.1% sodium dodecyl sulfate) containing protease inhibitor cocktail (Complete Mini; Roche Diagnostics, Mannheim, Germany). The protein concentration of the lysates was determined by the DC Protein Assay (Bio-Rad Laboratories, Hercules, CA). A double volume of 4× NuPage LDS sample buffer (Invitrogen Life Technologies) was added and boiled at 95°C for 4 minutes. The brainstem samples (100 μg) at each time point were subjected to electrophoresis in a 7.5% sodium dodecyl sulfate-polyacrylamide gel and transferred electrophoretically to a Hybond-P pure nitrocellulose membrane (Amersham Pharmacia Biotech, Buckinghamshire, England). The membranes were probed at room temperature for 2 hours with a 1-to-500 dilution of the primary antibody (mouse monoclonal anti-human HIF-1α [Novus Biologicals, Littleton, CO]), followed by exposure to a 1-to-10,000 dilution of the secondary antibody (peroxidase-conjugated rabbit anti-mouse antibody; Rockland, Gilbertsville, PA) for 45 minutes. The β-actin was analyzed as an internal control by incubation of the membrane with a 1-to-600 dilution of the primary antibody (goat polyclonal anti-human β-actin; Santa Cruz Biotechnology, Santa Cruz, CA) and subsequent incubation with a 1-to-10,000 dilution of the secondary antibody (peroxidase-conjugated donkey anti-goat antibody; Jackson Immunoresearch Laboratories, West Grove, PA). Signal development was performed using an ECL Plus detection kit (Amersham Biosciences). The relative band densities were analyzed using NIH Image software.
Electrophoretic Mobility Shift Assay
Nuclear extracts were prepared from the rat brainstem after rapid dissection and homogenization using a Nuclear Extraction Kit (Panomics, Redwood City, CA). Nuclear protein concentration was determined using the DC Protein Assay, and 5 μg of nuclear protein was used in each lane for electrophoretic mobility shift assays.
DNA-binding activity of HIF-1 was determined using the electrophoretic mobility shift assay Gel Shift Kits (Panomics) and a labeled transcription-factor probe (a biotin-labeled probe) containing the sense-strand sequence of HIF-1 consensus oligonucleotides (5′-AGCTTGCCCTACGTGCTGTCTCAGA-3′). Binding buffer, polydeoxy(inosinate-cytidylate), and consensus oligonucleotides (10 ng) were then combined and incubated. A total of 5 μg of nuclear extract was loaded onto a 6% polyacrylamide gel, and electrophoresis in Tris-borate ethylenediamine tetraacetic acid (50 mmol/L Tris buffer, 45 mmol/L boric acid, and 0.5 mmol/L ethylenediamine tetraacetic acid) was performed at 120 V for approximately 50 minutes at 4°C. After electrophoresis, the gel was transferred to a Pall Biodyne B membrane (Pall Gelman Laboratory, Ann Arbor, MI) in an electroblotting device at 300 mA for 30 minutes. The membrane was dried and blocked by incubation at room temperature with blocking buffer for 15 minutes and then probed with a 1-to-1000 dilution of streptavidin-horseradish peroxidase conjugate. The membrane was washed three times at room temperature with wash buffer and was incubated with detection buffer for 5 minutes, then with detection buffer plus a working substrate solution for 5 minutes. The membrane was exposed to Hyperfilm ECL (Amersham Biosciences) film. For the competition assay, double-stranded DNA fragments of unlabeled competitor probes (cold transcriptional factor probe) were used under the same conditions.
Brainstem Blood Flow Measurement
Rats were anesthetized and a midline skin incision of the neck was made with the rat in the supine position. Rat body temperature was maintained at 37°C using a heating pad and rectal temperature monitor. The thyroid gland was removed, and the trachea was cannulated to maintain the airway. Arterial blood partial pressure values for oxygen and carbon dioxide, and pH values were monitored and maintained in the respective physiological ranges with a monitoring device (i-STAT; i-STAT Corporation, East Windsor, NJ) and mechanical ventilation. The longus colli and longus capitis muscles beneath the esophagus were retracted bilaterally, and the exposed clivus was drilled out gently. The BA and the ventral surface of the pons were exposed over the dura mater. Brainstem blood flow was measured using laser Doppler flowmetry at 1.5 mm to the right of the BA (SAH-DFO group, n = 5 rats; SAH-placebo group; n = 5 rats, double saline injection group, n = 5 rats) (24).
All data are presented as means ± standard deviations. Comparisons of mRNA and protein levels, vessel areas, and cerebral blood flow (CBF) rates were performed using analysis of variance for multiple comparisons within and between groups, and subsequent pairwise comparisons were made using Fisher's exact test if significant variance was identified. Paired and unpaired t tests were used for comparisons between two measurements. Significant differences were considered present at P values less than 0.05.
HIF-1α and Protein and VEGF mRNA Levels in the Brainstem after SAH
Reverse transcriptase PCR revealed that HIF-1α mRNA levels were similar when SAH groups and control groups were compared at each time point in both the single- and double-hemorrhage models of SAH in rats (Fig. 1). In the rat single-hemorrhage model of SAH, Western blot analysis revealed that the HIF-1α protein level at 10 minutes was significantly greater in rats with SAH than in control rats (P < 0.01; Fig. 2). Furthermore, the VEGF mRNA level at 10 minutes in rats with SAH was significantly elevated compared with control rats (P < 0.01; Fig. 3). In the rat double-hemorrhage model of SAH, HIF-1α protein (Fig. 2) and VEGF mRNA (Fig. 3) expression in rats with SAH increased with statistical significance compared with control rats on Day 7 (P < 0.05, respectively).
Chronological Change in BA Area
In the single-injection model, acute vasospasm was maximal at 10 minutes (0.699 ± 0.109 minutes; P < 0.01, analysis of variance) and persisted for 6 hours. Mild vasospasm was present on Day 2. In the double-injection model, vasospasm was delayed, and BA area on Day 7 (0.821 ± 0.0280) was significantly reduced compared with the baseline value for BA area (P < 0.05; analysis of variance; Fig. 4).
HIF-1α mRNA and Protein and VEGF mRNA Levels in the Brainstem after Administration of DFO in a Rat Double-hemorrhage Model of SAH
There was no significant difference in HIF-1α mRNA levels when we compared the double-saline injection, SAH-DFO, and SAH-placebo groups (Fig. 5). However, HIF-1α protein and VEGF mRNA levels on Day 7 were significantly higher in the SAH-DFO group than in the double-saline injection and SAH-placebo groups (P < 0.05; Figs. 6 and 7).
Electrophoretic Mobility Shift Assay
Electrophoretic mobility shift assay studies demonstrated that brainstem HIF-1α activity was elevated in the SAH-placebo group and even more elevated in the SAH-DFO group when compared with the SAH-placebo group (Fig. 8). Furthermore, HIF-1α activity was lower in the double saline-injection group than in all other groups.
Effects of DFO Administration on Vasospasm and Cerebral Blood Flow
Intraperitoneal injection of DFO resulted in significantly diminished cerebral vasospasm after SAH (Fig. 9). Physiological parameters were maintained within the normal ranges (Table 1). For laser Doppler flowmetry-measured blood flow, the CBF was significantly reduced in the SAH-placebo group (26.3 ± 3.98 mL/100 g/min) when compared with the double saline-injection group (41.7 ± 7.57 mL/100 g/min; P < 0.01; Fig. 10). The administration of DFO in the SAH-DFO group resulted in a significant increase in CBF (46.3 ± 4.55 mL/100 g/min; P < 0.01).
Our investigation is the first report to examine the time course of HIF-1 expression after experimental SAH and to evaluate the relationship between HIF-1 expression and cerebral vasospasm. This study demonstrated that 1) brainstem HIF-1α protein levels were significantly elevated during the acute (10 min) and chronic (7 day) phases after SAH when compared with the control groups, 2) HIF-1α mRNA levels were similar in the SAH groups and the control groups, 3) changes in VEGF mRNA level paralleled those of HIF-1α protein levels and were significantly higher at 10 minutes and 7 days after SAH when compared with the control group, 4) intraperitoneal administration of DFO (300 mg/kg) on Day 4 in the SAH-DFO group resulted in increased brainstem VEGF mRNA levels and HIF-1α protein expression and activity on Day 7 when compared with the SAH-placebo group, and 5) DFO administration resulted in attenuation of BA vasospasm and an increase in brainstem blood flow on Day 7.
Hypoxia-inducible Factor-1 and Cerebral Vasospasm
HIF-1 is a transcriptional complex that mediates oxygen homeostasis and binds to the HIF-1 DNA binding complex, which is a heterodimer of α and β subunits. In oxygenated cells, the α subunits are unstable and are rapidly destroyed by a mechanism that involves ubiquitination by the von Hippel-Lindau tumor suppressor (pVHL) E3 ligase complex (12). In hypoxic cells, HIF-1α degradation is suppressed, leading to transcriptional activation of target genes. Thus, HIF-1α is essentially regulated not on the mRNA level but on the protein level via protein stabilization.
Several studies have suggested that changes in HIF-1α protein levels may mediate reactive alterations in cellular physiology secondary to cerebral ischemia (4,15), cerebral hemorrhage (14), and SAH (25). Ostrowski et al. (25) first demonstrated the expression of HIF-1 in brain parenchyma after acute cerebral ischemia in a rat endovascular perforation model of SAH. In the rat single-SAH model used in this study, maximal acute vasospasm occurred at 10 minutes after SAH using a cisternal injection method (5,35). Jackowski et al. (13) demonstrated that serial measurements of regional CBF (parietal, occipital, and cerebellar cortical regions) by hydrogen clearance revealed that experimental SAH resulted in an immediate 50% global reduction in cortical flow that persisted for up to 3 hours after SAH in rats. Naveri et al. (23) reported that injection of blood immediately decreased CBF, which stabilized after 10 to 15 minutes at approximately 45% of baseline value in a rat single-SAH model. Another investigator (25) described that one of the factors for early brain injury after SAH was acute cerebral ischemia in a rat experimental SAH model. These reports support the idea that the significant elevation of brainstem HIF-1α protein levels during the acute phase after SAH (10 min) is caused by CBF reduction (ischemia) resulting from cerebral vasospasm. In fact, our data show the correlation between the HIF-1α protein expression and the degree of vasoconstriction 10 minutes after SAH. Thrombin can produce brain injury by direct brain cell toxicity, and Jiang et al. (14) reported that thrombin released from a hematoma formed after intracerebral hemorrhage induced an increase in HIF-1α protein content in surrounding brain tissue. Thus, in addition to ischemia, thrombin released from a subarachnoid hematoma could play a role in the increase in HIF-1, which is especially notable for the acute (10 min after SAH) increase in HIF-1. In a rat single-hemorrhage model, late vasoconstriction occurred in 2 days according to Delgado et al. (5). There was mild vasospasm on Day 2 in our study, and a tendency for HIF-1α protein to increase in rats with SAH, but this increase was not significant. One possible explanation is that cortical CBF regained almost normal values by 48 hours after SAH (13), and HIF-1α protein expression was suppressed in some degree because of the recovery of CBF.
A single SAH in rats produced only acute and transient cerebral vasospasm, whereas the double-hemorrhage method of producing SAH in rats resulted in delayed vasospasm that mimicked human vasospasm (20). The present study demonstrated that brainstem HIF-1α protein levels correlated with vasospasm on Day 7 (delayed vasospasm) in the double-hemorrhage rat models. CBF measurement of the brainstem revealed an approximately 50% reduction 7 days after SAH compared with the saline-injection group. Vatter et al. (37) indicated an enhanced delayed reduction in CBF in the rat double-hemorrhage model. One of the most possible factors leading to an increase in HIF-1α protein and activity levels during the delayed phase after SAH may be vasospasm-induced brainstem ischemia.
The upstream and downstream signal transduction elements after SAH that mediate the increase in HIF-1α and transduce its neuroprotective effects remain unclear. VEGF is one of the genes in control of HIF-1, and it has angiogenetic and neuroprotective effects on central neurons (15). Jin et al. (15) demonstrated that hypoxia-sensitive VEGF signaling could be induced in neurons in global cerebral ischemia via HIF-1 activation. Ostrowski et al. (25) directly proved the close relation between HIF-1α and VEGF protein in the rat brainstem 24 hours after SAH. In our experiment, VEGF mRNA expression was correlated with HIF-1α protein expression at 10 minutes and 7 days after SAH. These data suggest that HIF-1α protein activated VEGF mRNA and that the HIF-1/VEGF pathway existed during the acute and delayed phase after SAH. However, it is unclear how its pathway is related to a response to neuronal damage resulting from SAH in the brainstem, because the neuroprotective effect of VEGF was not directly evaluated in our study.
This experiment also revealed that HIF-1α mRNA was constitutively expressed and was not associated with HIF-1α protein expression. HIF-1α activity was not mediated by altered transcription and was controlled on the protein level in circumstances such as SAH. It is thought that the difference between mRNA and protein expression for HIF-1α is mediated by vasospasm-induced ischemia, which inhibits the ubiquitin-proteosome degradation system (12).
Richard et al. (31) reported that angiotensin II, thrombin, and platelet-derived growth factor could induce an increase in HIF-1α protein levels in vascular smooth muscle cells via reduced nicotinamide adenine dinucleotide phosphate oxidase-mediated reactive oxygen species production under normal oxygen conditions. They concluded that reactive oxygen species upregulated HIF-1 expression by hypoxia-independent (O2 concentration-independent) mechanisms, and that these mechanisms should play a major role in vascular remodeling (31). Angiotensin II, thrombin, platelet-derived growth factor, and reactive oxygen species are also candidates for spasmogens (1,21,36,42). Thus, changes in HIF-1α protein levels within the vascular wall may also play a role in vasoreactivity after SAH.
Effects of Deferoxamine-activated Hypoxia-inducible Factor-1 on the Brainstem after SAH
To evaluate the effects of DFO on the delayed chronic cerebral vasospasm, a rat double-hemorrhage model of SAH was used in this session of the experiments. DFO is an iron chelator that may protect against brain injury by preventing lipid peroxidation and free radical formation. Indeed, DFO attenuated cerebral vasospasm after experimental SAH (7,18,38) and inhibited the production of hydroxyl radicals and progression of brain edema after ischemia in neonatal rats (26). DFO ameliorated brain edema after intracerebral hemorrhage in rats by reducing the oxidative stress caused by the release of iron from extravasated blood (10,22). DFO is known to stabilize HIF-1α and lead to transcriptional activity of its target genes, as does hypoxia or cobalt chloride (19,39). Because the association of the von Hippel-Lindau tumor suppressor with HIF-1 is iron dependent, the pVHL/HIF-1 complex cannot form in the cells treated with DFO or cobalt chloride (19). In fact, the interaction between pVHL and HIF-1 is regulated by prolyl hydroxylase, and the requirement of molecular oxygen and iron for prolyl hydroxylase activity may account for the stabilization of HIF-1α observed under hypoxic conditions or after treatment with DFO (11,12).
Prass et al. (27) recently reported that DFO had neuroprotective effects against focal cerebral ischemia via an increase in the DNA binding of HIF-1α in a middle cerebral artery occlusion model in rats. In this study (27), they proved that intraperitoneal injection of DFO induced tolerance with a maximum effect when given 3 days before ischemia induction at a dose of 300 mg/kg body weight. According to their results, DFO at the same dose was administered intraperitoneally 3 days before Day 7, because cerebral vasospasm is most prominent on Day 7 in a rat double-hemorrhage model of SAH (35). In addition, other studies (3,9,28) demonstrated that hypoxia or thrombin preconditioning activates HIF-1α and is neuroprotective against various brain injuries. In our investigation, intraperitoneal administration of DFO in a rat double-hemorrhage model resulted in an increase in brainstem HIF-1α protein levels and caused an improvement in brainstem blood flow and a reduction in cerebral vasospasm.
One possible explanation for the antivasospastic effect of DFO is that DFO itself has the ability to chelate iron. Free iron from a subarachnoid clot catalyzes the generation of cytotoxic free radicals and induces vasospasm (38). Vollmer et al. (38) demonstrated that the iron-chelating agent DFO ameliorated vasospasm in a rabbit model of SAH and that the mechanism of iron chelation suppressed the generation of free radicals. Moreover, iron chelation might also be of direct benefit in reducing neural parenchymal injury induced by free radicals that occurs as a result of ischemia (40). This direct neuroprotective effect of DFO possibly relates to significant attenuation of the brainstem blood flow reduction that was identified in this experiment.
In our investigation, the change of HIF-1 expression in the vascular wall after DFO administration was not assessed. If DFO induces overexpression of HIF-1 in the vascular wall, it is thought that another possible mechanism for the antivasospastic effect of DFO is HIF-1-induced expression of the genes in its downstream. Grasso et al. (6) reported that systemic administration of human recombinant erythropoietin reduced BA vasoconstriction in a rabbit SAH model and concluded that erythropoietin may exert a direct effect on cerebral arteries via erythropoietin receptors or by acting to induce endothelial release of nitric oxide. Furthermore, Ono et al. (24) reported that overexpression of heme oxygenase-1, the principal enzyme involved in the metabolism of hemoglobin, inhibited arterial contractions induced by hemoglobin and reduced vasospasm after experimental SAH. The expression of these antivasospastic genes, which are activated by HIF-1, may contribute to the significant amelioration of BA vasospasm in a rat SAH model, although the expression of these genes in BA were not evaluated. The DFO administration (in the SAH-DFO groups) significantly elevated VEGF mRNA levels in brainstem compared with the placebo administration (the SAH-placebo groups) in this study. The VEGF expression in cerebral vessels and neurons enhanced angiogenesis during the chronic phase after SAH in rats (16). We may speculate that overexpression of HIF-1-induced VEGF by DFO emphasized angiogenesis in the brainstem and caused improvement in the reduction of brainstem blood flow. Because many investigators have shown the neuroprotective effects of erythropoietin via HIF-1 activation (27,28,41), the relationship of other neuroprotective genes in the control of HIF-1 and erythropoietin to the amelioration of reduction of blood flow should be taken into consideration. We hypothesize that the upregulation of HIF-1α protein in the condition of the brain after SAH is a reaction to neuronal damage, and that overexpression of HIF-1α protein by DFO has a neuroprotective effect on the brainstem, in part via DFO itself as an iron chelator, and in part via HIF-1α and the genes that are downstream.
This study suggests that DFO is a promising agent for treating experimental cerebral vasospasm. DFO is clinically used at present, and its safety has been proved. To be convinced of the efficacy of DFO for cerebral vasospasm, we think that additional investigations are needed to clarify the following: 1) Is the effect of DFO via HIF-1 on vasospasm present in large-animal models of SAH such as dogs and primates? 2) How does HIF-1α protein expression change in vascular walls at several time points after experimental SAH? 3) One gene in the control of HIF-1, such as erythropoietin, VEGF, and heme oxygenase-1, should be focused on, and how the gene displays the neuroprotective and antivasospastic effects on the brain after SAH should be explored. 4) DFO brain concentration and its pharmacokinetic parameters after intraperitoneal injection should be evaluated to determine the adequate dosage of DFO.
We report that HIF-1α protein expression increased in the brainstem during the acute and delayed phases after SAH, and that a DFO-induced additional increase in HIF-1α protein levels and activity exerted significant attenuation of BA vasospasm and a reduction in brainstem blood flow in the rat model of SAH. DFO has potential as a therapeutic agent for cerebral vasospasm caused by aneurysmal SAH via upregulation of HIF-1 activity.
We thank Hideki Wakimoto, Masako Arao, and Mayumi Konishi for technical assistance.
In this study, the authors show that subarachnoid hemorrhage (SAH) increases hypoxia inducible factor-1α (HIF-1α) and vascular endothelial growth factor (VEGF) protein in the brainstem of rats. If these are further increased by deferoxamine (DFO), vasospasm can be attenuated. The authors correctly note that several possible mechanisms could be involved, such as increased HIF-1α protein, stabilization of HIF-1α in various protein complexes, and direct beneficial effects of DFO independent of HIF-1α and VEGF. Approximately 10 years ago, there was interest in iron chelators as a treatment for vasospasm, but this was never pursued clinically due to conflicting preclinical data, possible toxic effects, and, I suppose, the lack of a clinician to drive the study.
There continue to be some methodological flaws in vasospasm studies, including this one, that need to be known to researchers in the field. These include lack of blinding and randomization, the use of β-actin as an internal standard for molecular analysis, and the use of laser Doppler to compare blood flow between groups and/or over time (1). I have done them all wrong myself, but it would be helpful to not repeat them. Also, there seems to be a trend toward showing a smaller and smaller portion of the gel in molecular studies. It is more comforting, at least for the reviewer, to see the whole gel.
R. Loch Macdonald
The authors reported that HIF-1α protein expression increased in the brainstem after SAH and that the administration of DFO further increased the protein in the brainstem and significantly attenuated basilar artery vasospasm and reduction of brainstem blood flow in rat SAH models. The effects of DFO in a double hemorrhage model seem interesting, and further study should be performed. Although the authors examined the expression of HIF-1α and speculated the mechanisms of antivasospastic effect and blood flow recovery by DFO, their conclusion that DFO-induced increase in HIF-1α protein levels and activity exerts significant attenuation of basilar artery vasospasm and of reduction in brainstem blood flow is too strong because they did not prove any direct action of HIF-1α on vasospasm and blood flow in their experiments. Although they showed that DFO increased HIF-1α protein in the brainstem, they did not show the increase of HIF-1α in vessel walls or the direct relationship between HIF-1α and the attenuation of vasospasm and reduction in brainstem blood flow. It seems that DFO attenuated the degree of vasospasm and increased brainstem blood flow by unknown mechanisms.
In this study, the authors use a rodent SAH model using a single and double injection method to look at early (10 min) and late (up to 7 d) vasospasm in rats. They studied transcription factors that may be active in the regulation of these processes. Specifically, they studied HIF-1α, which is a transcription factor with the key ability to regulate a number of other factors, many of which are neuroprotective. These are involved in oxygen hemostasis, development of ischemia, and angiogenesis, and seem to regulate many genes, including erythrocyte and VEGF, as well as glucose transporter 1. The authors show that this factor HIF-1α protein is significantly increased at 10 minutes in the single injection model and 7 days in the double injection model. Importantly, at no time is there an upregulation of messenger ribonucleic acid transcription of HIF-1α during these studies. The authors note that times at which the protein has increased are times when there is increased spasm. Their hypothesis, however, is not that HIF-1α increases spasm, but rather that it is present in reaction to the spasm that is present. They then take the experiment further by using an iron chelator DFO mesylate, which decreases the metabolism HIF-1α and requires oxygen and tissue plasminogen activators to break it down. In these circumstances, the authors note an improvement in vasospasm. They also note that the increased HIF-1α protein does correlate as well with an increase in VEGF.
The authors studied these time points based primarily on the time of clinical presence of spasm in these models, but several large periods of time are not covered. The difference between messenger ribonucleic acid and protein expression for HIF-1α is clearly an intriguing part of this article and is not fully understandable. As HIF-1α could affect so many activities after ischemia, it is not clear whether they are seeing cause or effect in the increase in both VEGF and in spasm. Finally, the discussion should give the reader some idea of where this research will be going beyond observation and how they would enhance their studies to develop a causal relationship which may show treatment benefit.
Robert J. Dempsey
Hishikawa et al. have investigated the relationship between HIF-1α expression and rodent SAH-associated vasospasm. Using both single and double hemorrhage models of SAH, HIF-1α protein levels and basilar artery diameter were assessed at 10 minutes, 6 hours, and on Days 1 and 2. The effects of intraperitoneal DFO were also evaluated. Results showed HIF-1α protein levels to be significantly elevated after SAH and correlated with the degree of cerebral vasospasm. DFO administration caused increased HIF-1α protein expression and attenuated cerebral vasospasm. The authors concluded that DFO may be a promising agent in clinical SAH by ameliorating SAH-associated cerebral vasospasm via induction of HIF-1α expression.
HIF-1α regulates the expression of numerous genes, including nitric oxide (NO) synthase, during hypoxia. In addition, the amount of intracellular oxygen available for other reactions, including further NO synthesis, increases as mitochondrial oxygen consumption is blunted (3). As NO inhibits mitochondrial activity, superoxide is produced to activate HIF-1α (2). Furthermore, high-output NO synthase can stabilize HIF-1α (1). Considering these interactions, the development of HIF-1α related clinical agents may be limited by this target's nonenzymatic function and multipotent regulators.
Ricardo J. Komotar
E. Sander Connolly, Jr.
New York, New York