Oxidative Stress Increases Blood–Brain Barrier Permeability and Induces Alterations in Occludin during Hypoxia–Reoxygenation

Abstract

The blood–brain barrier (BBB) has a critical role in central nervous system homeostasis. Intercellular tight junction (TJ) protein complexes of the brain microvasculature limit paracellular diffusion of substances from the blood into the brain. Hypoxia and reoxygenation (HR) is a central component to numerous disease states and pathologic conditions. We have previously shown that HR can influence the permeability of the BBB as well as the critical TJ protein occludin. During HR, free radicals are produced, which may lead to oxidative stress. Using the free radical scavenger tempol (200 mg/kg, intraperitoneal), we show that oxidative stress produced during HR (6% O₂ for 1 h, followed by room air for 20 min) mediates an increase in BBB permeability in vivo using in situ brain perfusion. We also show that these changes are associated with alterations in the structure and localization of occludin. Our data indicate that oxidative stress is associated with movement of occludin away from the TJ. Furthermore, subcellular fractionation of cerebral microvessels reveals alterations in occludin oligomeric assemblies in TJ associated with plasma membrane lipid rafts. Our data suggest that pharmacological inhibition of disease states with an HR component may help preserve BBB functional integrity.

Introduction

The blood–brain barrier (BBB) is a physical and metabolic barrier comprised of brain microvascular endothelial cells that restrict the passage of substances from the blood to the brain and help maintain brain homeostasis. The BBB expresses a high number of ion channels and transporters, has a low rate of pinocytosis, and forms intercellular tight junction (TJ) protein complexes that limit paracellular permeability (Hawkins and Davis, 2005). It has previously been shown that disruption of the BBB is associated with hypoxia and reoxygenation (HR) (reviewed in Kaur and Ling, 2008). Hypoxia and reoxygenation leads to an increase in BBB permeability (Witt et al, 2003), which may result in neurotoxic substances leaking from the blood into the brain and contributing to vasogenic edema. Cerebral HR is a central component of many disorders or conditions affecting cognition. These include stroke (Kalaria and Ballard, 2001), cardiac arrest (Lim et al, 2004), postoperative cognitive dysfunction (Caza et al, 2008), acute respiratory distress syndrome (Hopkins et al, 2006), obstructive sleep apnea (El-Ad and Lavie, 2005), high-altitude cerebral edema, and acute mountain sickness (Hackett, 1999).

Much of the cellular damage caused by hypoxic insult is thought to occur during the subsequent reoxygenation phase. Hypoxia and reoxygenation is associated with an increased production of reactive oxygen species (ROS) (reviewed in Wong and Crack, 2008). These ROS contribute to brain injury by reacting with proteins, lipids, and nucleic acids, as well as by activating a number of redox-sensitive signaling pathways. A number of different laboratories have shown that production of ROS can affect BBB permeability by a variety of mechanisms, including modulation of TJ proteins (reviewed in Pun et al, 2009).

Previously, we have shown that HR causes an increase in BBB paracellular permeability in vivo (Witt et al, 2003; Witt et al, 2008). This increase in BBB permeability correlates with alterations in the oligomeric assembly of the critical TJ protein occludin (McCaffrey et al, 2009). Occludin is a transmembrane protein that exists as oligomers in lipid raft regions at the plasma membrane of the BBB (McCaffrey et al, 2007). Multiple domains of occludin have previously been shown to regulate the diffusion of solutes across the TJ (Balda et al, 2000). Occludin oligomers are held together covalently by disulfide bonds (McCaffrey et al, 2009; Walter et al, 2009). These bonds can be disrupted by pathologic conditions such as HR and inflammatory pain, leading to structural changes of the oligomers (McCaffrey et al, 2008; McCaffrey et al, 2009). Changes in the structure of occludin oligomers could potentially influence the ability of the TJ complexes to limit paracellular diffusion of molecules from the blood into the brain. This is the first study to look at the effects of oxidative stress on occludin oligomeric assembly at the BBB TJs in vivo.

To examine the role that oxidative stress has in disrupting the functional integrity of the BBB, we administered the stable, membrane-permeable, water-soluble nitroxide, tempol to rats before the induction of HR. Tempol shows superoxide dismutase-like activity toward the superoxide anion (O₂^.−), as well as reactivity with the hydroxyl radical (OH.) (Saito et al, 2003), nitrogen dioxide (NO₂^.), and the carbonate radical (CO₃^.−) (Augusto et al, 2008). Tempol readily crosses the BBB (Zhelev et al, 2009), and has previously been shown to provide neuroprotection as a free radical scavenger in several models of brain injury and ischemia (Kwon et al, 2003; Deng-Bryant et al, 2008; Rak et al, 2000; Cuzzocrea et al, 2000).

In this study, we looked at the effects of ROS produced during HR in a model that has previously been shown to induce alterations in BBB permeability and the structural integrity of the TJ protein occludin. We use the in situ brain perfusion method to assess functional BBB permeability after HR in animals pretreated with tempol to examine the role of ROS in BBB disruption. In addition, we look at the effect that ROS production has on occludin localization using confocal microscopy. Finally, we use the method of density gradient fractionation of cerebral microvessels followed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) under reducing and nonreducing conditions to examine the role that ROS has in HR-induced changes in the structural integrity of occludin.

Materials and methods

Radioisotopes, Reagents, and Chemicals

The [¹⁴C]-sucrose was purchased from American Radio-labeled Chemicals (St Louis, MO, USA). Tempol was purchased from MP Biomedicals (Solon, OH, USA). Ts-2 tissue solubilizer was purchased from Research Products International (Mt. Prospect, IL, USA). Optiphase Supermix scintillation cocktail was purchased from Perkin Elmer (Shelton, CT, USA). Ethylenediaminetetraacetic acid (EDTA)-free complete protease inhibitors were purchased from Roche (Indianapolis, IN, USA). OptiPrep was purchased from Accurate Chemical (Westbury, NY, USA). The Coomassie Plus Better Bradford Assay Kit was purchased from Thermo Scientific (Rockford, IL, USA). Rabbit anti-occludin directed toward the C-terminus was purchased from Zymed (San Francisco, CA, USA) and mouse anti-platelet-endothelial cell adhesion molecule 1 was purchased from AbD Serotec (Raleigh, NC, USA). Mouse anti-heat shock protein 70 (anti-HSP-70) was purchased from R&D Systems (Minneapolis, MN, USA). Secondary antibodies for western blotting and immunofluorescence were purchased from Amersham (Pittsburgh, PA, USA) and Invitrogen (Carlsbad, CA, USA), respectively. Western Lightning enhanced chemiluminescence reagent was purchased from Perkin Elmer (Shelton, CT, USA). TO-PRO-3 and ProLong Gold were purchased from Invitrogen. The XT sample buffer, XT reducing agent and criterion gels were purchased from Bio-Rad (Hercules, CA, USA). All other antibodies or reagents, unless noted, were purchased from Sigma-Aldrich (St Louis, MO, USA).

Hypoxic Treatment

All experimental protocols were approved by the Institutional Animal Care and Use Committee at the University of Arizona. Adult female Sprague–Dawley rats (Harlan, Indianapolis, IN, USA) weighing 230 to 280 g were housed under standard 12:12 h light–dark conditions, and received food and water ad libitum. Rats were exposed to 6% O₂ for 1 h, followed by 20 mins of reoxygenation. At 10 mins before hypoxic treatment, rats were administered an i.p. injection of saline or tempol (200 mg/kg body weight).

In Situ Brain Perfusion

The in situ brain perfusion technique was adapted from previously published methods (Zlokovic et al, 1986; Takasato et al, 1984; Martel et al, 1997), and was performed as previously described (Witt et al, 2003). After the hypoxic insult (6% O₂, 1 h), rats were anesthetized with a 1.0 mL/kg i.p. injection of an anesthetic cocktail consisting of ketamine (78.3 mg/mL), xylazine (3.1 mg/mL), and acepromazine (0.6 mg/mL) and allowed to reoxygenate for 20 mins in room air (21% O₂) before administering the paracellular permeability marker ¹⁴C-sucrose (molecular weight (MW)= 342). Surgery was performed during the reoxygenation phase. Once anesthetized, the rats were heparinized (10,000 U/kg), and the carotid arteries were isolated and cannulated with silicon tubing. The perfusion medium contained albumin-bound Evans blue dye in a mammalian Ringer solution (117.0 mmol/L NaCl, 4.7 mmol/L KCl, 0.8 mmol/L MgSO₄ · H₂O, 24.88 mmol/L NaHCO₃, 1.2 mmol/L KH₂PO₄, 2.5 mmol/L CaCl₂ · 6H₂O, 10 mmol/L D-glucose, 39 g/L dextran (MW = 70,000), and 1 g/L bovine serum albumin), which was bubbled with 95%O₂/5%CO₂ gas mix. The Ringer solution was filtered, heated (37°C), and de-bubbled before arterial infusion at 3.1 mL/min. The paracellular permeability marker ¹⁴C-sucrose was infused into the perfusate inflow at 0.5 mL/min for 10 mins, followed by a 1 min washout containing Ringer with no radioactive tracer to clear the vascular space of radioactivity.

At the end of perfusion, the rat was killed and the brain was removed. The meninges and choroid plexuses were removed and the brain was divided and placed into preweighed vials. Tissue solubilizer (1 mL) was added to each vial and allowed to dissolve the brain sample for 48 h. After solubilization, 100 μL of 30% glacial acetic acid was added to each sample to quench chemiluminescence. Optiphase Supermix scintillation cocktail (2.5 mL) was added to each sample. Triplicate samples of 100 μL aliquots of the perfusion medium were treated in the same manner as the perfused brain samples. All samples were then measured for disintegrations per minute (dpm) (1450 LSC and Luminescence Counter; Perkin Elmer, Waltham, MA, USA). The ratio of the concentration of ¹⁴C-sucrose in tissue (C_brain; in dpm/g) was compared with perfusate (C_perfusate; in dpm/mL) and expressed as a percent ratio R_brain = (C_brain/C_perfusate) × 100%. We have previously used capillary depletion analysis in this model to show that the increase in ¹⁴C-sucrose after HR was not because of increased trapping within the microvascular endothelial cells (Witt et al, 2003).

Microvessel Enrichment and Fractionation

After hypoxic or control treatment, rats were anesthetized with a 1.0 mL/kg i.p. injection of an anesthetic cocktail consisting of ketamine (78.3 mg/mL), xylazine (3.1 mg/mL), and acepromazine (0.6 mg/mL). They were then killed after 20 mins of reoxygenation and the brains were extracted. The cerebellum, meninges, and choroid plexuses were removed and the cerebral hemispheres were homogenized in 4mL of microvessel isolation buffer A (103 mmol/L NaCl, 4.7 mmol/L KCl, 2.5 mmol/L CaCl₂, 1.2 mmol/L KH₂PO₄, 1.2 mmol/L MgSO₄, 15 mmol/L 4-(2-hydroxyethyl) 1-piperazineethanesulfonic acid (HEPES), 2.5 mmol/L NaHCO₃, 10 mmol/L D-glucose, and 1 mmol/L sodium pyruvate (pH 7.4)). Ice-cold buffer A (4 mL) with 26% dextran was added to the homogenates, which were vortexed and centrifuged at 5,800 g for 10 mins at 4°C and repeated. Pellets were resuspended in buffer A supplemented with protease inhibitors (Roche EDTA-free complete protease inhibitor, 2 mmol/L phenylmethylsulfonyl fluoride, 1 mmol/L Na₃VO₄, 1 mmol/L NaF, and 1 mmol/L sodium pyrophosphate) and then passed through a 100 μm filter. The filtrate was then passed through a 40 μm filter. The microvessels remaining on the filter were collected in buffer A and centrifuged at 1,500 g for 10 mins. A small aliquot of these microvessels was placed on a glass slide and immunostained as below. The pellet from the remaining microvessels was then resuspended in 4mL of buffer B (20 mmol/L Tris-HCl, 250 mmol/L sucrose, 1 mmol/L CaCl₂, and 1 mmol/L MgCl₂ (pH 7.4)) with protease inhibitors and drawn through a 21-G needle 20 ×. The protein concentration of each sample was then determined using the Coomassie Plus Better Bradford Assay Kit, and equal amounts of microvessel homogenates were mixed with 50% OptiPrep in buffer B. This suspension was layered beneath a discontinuous 0/5/10/15/20% Optiprep gradient (prepared in buffer B without ions) and then centrifuged for 90 mins at 52,000 g at 4°C. From the top of the gradient, 1mL fractions were collected using a Biocomp Gradient Station ip (Fredericton, NB, Canada). Each fraction was then assayed for refractive index (density) and protein content using the Coomassie Plus Better Bradford Assay Kit.

Western Blot Analysis

Equal volume aliquots of each gradient fraction were mixed with equal volumes of 2 × perfluorooctanoic acid (PFO) extraction buffer (100 mmol/L Tris, 20% glycerol, and 4% PFO (pH 8.0)) and allowed to incubate at 25°C for 30 min. Samples were then centrifuged at 11,000 g for 10 mins and supernatants containing PFO-solubilized material were added to 4 × XT sample buffer with or without XT reducing agent containing tris(2-carboxyethyl) phosphine hydrochloride (TCEP). Each fraction was then incubated at 70°C for 10 mins before performing SDS-PAGE on 10% Bis-Tris criterion gels. Sets of gels (reducing and nonreducing, for each treatment condition) were then transferred to the same polyvinylidene fluoride membranes, blocked in 5% milk, and incubated overnight in primary antibody (1:1,000 dilution) solution. Blots were then washed, incubated in HRP-conjugated secondary antibody solution, and detected using enhanced chemiluminescence reagent. Each blot was stained with Coomassie and the optical density of each band was normalized to the total protein in the lane according to previously published methods (Aldridge et al, 2008). Densitometry was performed using ImageJ software (Wayne Rasband, Research Services Branch, National Institutes of Mental Health, Bethesda, MD, USA).

Immunofluorescence

All slides from control and treated rats were collected and processed in parallel. Primary antibody was omitted on some slides as a negative control. Fluorescent immunostaining for occludin was performed on microvessel preparations enriched as above. Briefly, microvessels were heat-fixed on glass slides at 95°C for 10 min, followed by fixation in ice-cold acetone for 10 mins. The slides were blocked in 2% goat serum and 1% bovine serum albumin before incubation in rabbit anti-occludin primary antibody (1:250) and the endothelial-specific marker mouse anti-platelet-endothelial cell adhesion molecule 1 (1:1,000) overnight in the cold; slides were incubated with the appropriate Alexafluor-conjugated secondary antibodies.

For HSP-70 immunostaining, microvessels were placed on a glass slide and allowed to air dry before being fixed in 3.7% formaldehyde for 10 min. The microvessels were permeabilized in 0.1% Triton X-100, blocked as above, and incubated overnight in the cold in mouse anti-HSP 70 (1:400) and in the endothelial-specific marker rabbit anti-von Willebrand factor (1:5,000). The slides were then incubated with the appropriate Alexafluor-conjugated secondary antibodies.

For hypoxia-inducible factor 1α (Hif-1α) immunostaining, microvessels were placed on a glass slide and allowed to air dry before being fixed in ice-cold acetone for 10 mins. The slides were blocked as above and incubated in mouse anti-Hif-1α (1:200) along with rabbit anti-von Willebrand factor (1:5,000) overnight in the cold. Slides were then incubated with the appropriate Alexafluor-conjugated secondary antibodies. The nucleus of the microvessels was then stained with the nuclear-specific dye TO-PRO-3 (10 μmol/L) for 10 min. All slides were mounted in ProLong Gold anti-fade reagent before applying the coverslip.

Confocal Microscopy

All slides were imaged on a Zeiss LSM 510 meta-NLO confocal microscope (Zeiss, Oberkochen, Germany), with filters appropriately set to avoid bleed-through. Only microvessels positive for the endothelial-specific markers platelet-endothelial cell adhesion molecule 1 or von Willebrand factor were used for analysis. Semiquantitative analysis of the mean fluorescent intensity of HSP-70 was performed according to previously published methods (Hawkins et al, 2004). Microvessels were randomly chosen from rats in each treatment group and the mean fluorescence intensity was analyzed using Zeiss LSM Image Browser software (Zeiss), with the data expressed as percentage of control. To investigate Hif-1α translocation to the nucleus, the Mander's colocalization coefficient for Hif-1α and TO-PRO-3 was determined using the Zeiss LSM Image Browser software in randomly chosen microvessels in each treatment group.

Statistical Analysis

All data analyses were performed using SigmaPlot software (Systat Software, Inc., San Jose, CA, USA). To determine statistical significance between treatment groups, data were analyzed using one-way analysis of variance (ANOVA), followed by Bonferroni's t-test. Data are presented as means ± s.e. A value of P < 0.05 was accepted as statistically significant.

Results

Hif-1α Expression and Translocation to the Nucleus

Hypoxia-inducible factor 1α is an oxygen-sensitive protein that is continuously synthesized and degraded under normoxic (Nx) conditions. Under hypoxic conditions, degradation of Hif-1α decreases and translocation to the nucleus occurs, wherein Hif-1α heterodimerizes with Hif-1β and binds to hypoxia-responsive elements of target genes. Our confocal imaging data (Figure 1) on our enriched microvessel preparations show that HR causes an increase in Hif-1α colocalization with the nuclear dye TO-PRO-3, suggesting that Hif-1α translocation to the nucleus is because of hypoxia. We measured the ratio of nuclear Hif-1α to total Hif-1α using the Mander's colocalization coefficient (Table 1). A value of 0 indicates no colocalization of Hif-1 with the nucleus, whereas a value of 1 indicates that all pixels colocalize. We obtained a Mander's colocalization coefficient of 0.142 ± 0.019 for saline-treated Nx (Nx + S) and 0.119 ± 0.015 for tempol-treated Nx (Nx + T) rats. Saline-treated HR (HR + S) rats showed a statistically significant increase in Hif-1α nuclear colocalization with a Mander's colocalization coefficient of 0.211 ± 0.22 than did Nx rats. Tempol-treated HR (HR + T) rats had a Mander's colocalization coefficient of 0.154 ± 0.012. Images are representative of six rats per treatment group (n = 6).

Figure 1

Hypoxia and reoxygenation (HR) induces hypoxia-inducible factor 1α (Hif-1α) translocation to the nucleus. Immunofluorescence of Hif-1α (green) and the nuclear dye TO-PRO-3 (blue) in brain microvessels of normoxic (Nx) plus saline (A), Nx plus tempol (B), HR plus saline (C), and HR plus tempol (D)-treated rats. Rats exposed to HR show an increase in translocation of Hif-1α to the nucleus (arrow). Immunofluorescence of the endothelial-specific marker von Willebrand factor (red) is shown in the inset.

Table 1

Mander's colocalization coefficients for Hif-1α and TO-PRO-3

Nx+S	Nx+T	HR+S	HR+T
0.142 ± 0.019	0.119 ± 0.015	0.211 ± 0.022*,#	0.154 ± 0.012

ANOVA, analysis of variance; Hif, hypoxia-inducible factor; HR, hypoxia–reoxygenation; Nx, normoxia; S, saline; T, tempol.

Cerebral microvessels from animals exposed to Nx or HR with S or T were analyzed for colocalization of Hif-1α with the nuclear dye TO-PRO-3. HR+S rats show a significant increase in Hif-1α nuclear colocalization compared with Nx rats. Data are representative of 50–60 randomly chosen microvessels from each treatment group. Values are means ± s.e. One-way repeated measures ANOVA and the post hoc Bonferonni's t-test were used to establish significance.

*

P < 0.05 versus Nx+S

#

P < 0.01 versus Nx+T.

HSP-70 Expression

Upregulation of HSP-70 is associated with oxidative stress (Ronaldson and Bendayan, 2008; Amadio et al, 2008). We looked at levels of HSP-70 in the enriched microvessel preparations of all treatment groups using confocal microscopy (Figure 2A). Images are representative of six rats per treatment group (n = 6). Our data show a significant increase in HSP-70 (1.5-fold) mean fluorescence intensity resulting from HR. Tempol (HR +T) attenuated this increase (1.2-fold), indicating reduced levels of cellular stress because of scavenging of free radicals during HR (Figure 2B).

Figure 2

Oxidative stress induces expression of heat shock protein (HSP-70). Immunofluorescence of HSP-70 (green) in brain microvessels of normoxic (Nx) plus saline (Nx + S) (A), Nx plus tempol (Nx + T) (B), hypoxia–reoxygenation (HR) plus S (HR + S) (C), and HR + T (D)-treated rats. Rats exposed to HR show an increase in HSP-70 immunofluorescence. Tempol attenuated this increase. Immunofluorescence of the endothelial-specific marker von Willebrand factor (red) is shown in the inset. (E) Mean fluorescence intensity of HSP-70 in all treatment groups expressed as a percent control. *P < 0.05 versus HR + T and Nx + T. P < 0.01 versus Nx + S (n = 28 to 33 microvesels per treatment group).

In Situ Brain Perfusion

The role of oxidative stress in altering BBB permeability during HR was investigated using the in situ brain perfusion technique. This method maintains a constant cerebral flow rate of perfusion fluid and eliminates potential confounding variables associated with peripheral blood flow, metabolism, and clearance. ¹⁴C-sucrose was used as a marker of paracellular permeability because of its minimal BBB penetration under physiologic conditions and its inherent metabolic stability in this system. The ¹⁴C-sucrose was not subject to any metabolism during the perfusion, as verified by high performance liquid chromatography analysis (data not shown). In Nx + S rats, the ratio brain (Rbr (%)) was 1.60 ± 0.09% (Figure 3). After HR, the Rbr was 2.43 ± 0.25%, which was significantly increased (P = 0.016) by 51.9% compared with control. With administration of the free radical scavenger tempol 10 mins before HR, the Rbr was 1.61 ± 0.15%, which was significantly lower than the Rbr from the HR + S rats (P = 0.014). This suggests that ROS are involved in the increase in BBB permeability associated with HR. The Nx + T rats showed a nonsignificant increase in Rbr of 1.9 2 ± 0.15% compared with Nx + S rats (Figure 3).

Figure 3

Effects of hypoxia and reoxygenation (HR) and tempol treatment on BBB permeability to ¹⁴C-sucrose. After treatment, normoxia (Nx) plus saline (S) (Nx + S), Nx plus tempol (T) (Nx + T), HR + S, and HR + T rats were anesthetized and subjected to in situ brain perfusion for 10 mins with ¹⁴C-sucrose. The amount of radioactivity in the brain versus the perfusate was expressed as Rbr%. Results are expressed as mean ± s.e. *P < 0.05.

Occludin Microvascular Distribution

To examine the effects of HR and oxidative stress on the cellular localization of the TJ protein occludin, enriched microvessel preparations were placed on glass slides and stained for occludin. Slides were then examined under a laser scanning confocal microscope. Images are representative of five rats per treatment group (n = 5). In microvessels isolated from Nx rats, there was continuous staining of occludin along the margins of intercellular contact (Figure 4). After HR treatment, an increase in intracellular punctate staining of occludin was observed, indicating a change in localization of occludin away from the TJ complex. Rats treated with tempol before HR showed continuous staining of occludin along the TJ, similar to what was seen in the Nx rats. The number of microvessels exhibiting punctate staining in each treatment group was quantitated and is presented in Table 2. Microvessels in Nx + S rats showed punctate staining in 14.95 ± 2.11% of microvessels. Tempol-treated HR and Nx + T rats exhibited punctate staining in 15.94 ± 0.42% and 16.60 ± 2.09% rats, respectively. In HR + S rats, 28.74 ± 3.21% of microvessels showed punctate staining, a statistically significant increase compared with the other treatment groups (P < 0.01). These results indicate that oxidative stress induces a change in occludin intracellular localization.

Figure 4

Oxidative stress induces changes in occludin localization. Immunofluorescence of occludin (green) and the endothelial-specific marker platelet-endothelial cell adhesion molecule 1 (PECAM-1) (red inset) in brain microvessels of normoxic (Nx) plus saline (A), Nx plus tempol (B), hypoxia–reoxygenation (HR) plus saline (C), and HR plus tempol (D). HR induces an increase in punctate staining of occludin (arrows), suggesting the movement of occludin away from the tight junction. Tempol prevents this increase, indicating that oxidative stress is involved in altering the cellular localization of occludin.

Table 2

Percentage of cerebral microvessels exhibiting punctate staining of occludin

Nx+S	Nx+T	HR+S	HR+T
14.95 ± 2.11%	16.60 ± 2.09%	28.74 ± 3.21%*	15.94 ± 0.42%

ANOVA, analysis of variance; HR, hypoxia–reoxygenation; Nx, normoxia; S, saline; T, tempol.

Microvessels were analyzed for punctate staining in Nx+S, Nx+T, HR+S, and HR+T. Values are means ± s.e.; n = 500 microvessels per treatment group. One-way repeated measures ANOVA and the post hoc Bonferonni's t-test were used to establish significance.

*

P < 0.01 compared with N+S, N+T, and HR+T.

Density Gradient Fractionation

Density gradient centrifugation and fractionation of intact microvessels were performed to examine the distribution of occludin within plasma membrane microdomains. The protein concentration and refractive indices from intact microvessels subjected to density gradient fraction are presented as averages of three separate experiments, each containing three rats per treatment group (Figure 5A). Our data show that the protein concentration and refractive index from each fraction is similar for each treatment group. This shows that our analysis was performed on membrane microdomains that are similar across all treatment groups. We have previously shown that the TJs exist primarily in fractions 7 and 8 (McCaffrey et al, 2007). Using SDS-PAGE, we analyzed western blots of fractions 7 and 8 under reducing and nonreducing conditions to investigate the band patterns of PFO-soluble occludin oligomers (Figure 5B). Perfluorooctanoic acid has previously been shown to preserve the oligomeric structure of membrane proteins by stabilizing internal noncovalent interactions between component protein subunits (Ramjeesingh et al, 1999). During HR, a significant increase in the optical density of occludin oligomers under nonreducing conditions compared with Nx controls (4.9-fold in fraction 7 and 11.0-fold in fraction 8) is observed (Figure 5C). Occludin oligomers exhibit a broad, intense staining pattern resulting from HR. This suggests that extensive conformational changes in occludin occur during HR. Administration of tempol before HR served to significantly decrease the extent of oligomeric disassembly and to attenuate the increase in optical density of nonreduced occludin oligomeric isoforms compared with Nx + S rats (2.1-fold in fraction 7 and 4.1-fold in fraction 8). Comparison of occludin banding patterns on nonreducing with reducing blots reveals that the reducing agent TCEP increases the detection of occludin oligomeric isoforms in Nx rats, but not in HR rats. This suggests that HR causes conformational changes of occludin oligomeric assemblies, exposing more of the structure to the antibody.

Figure 5

Density gradient fractionation of cerebral microvessels. (A) Profiles of protein concentrations and refractive indices (density) of cerebral microvessels after fractionation in normoxia (Nx) plus saline (S) (Nx + S), Nx plus tempol (T) (Nx + T), hypoxia–reoxygenation (HR) + S (HR + S), and HR + T. (B) Representative sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and western blot of occludin oligomers in fractions 7 and 8 under nonreducing (NR) and reducing (R) conditions. Fractions 7 and 8 have previously been shown to be associated with the tight junction in lipid rafts at the plasma membrane. (C) Western blot analysis of occludin oligomers in fractions 7 and 8 under reducing and nonreducing conditions. Data are presented as mean ± s.e. and are representative of three separate experiments with three rats pooled in each treatment group. *P < 0.001 versus Nx + S and Nx + T; P < 0.01 versus HR + T in fraction 7. *P < 0.01 versus Nx + S and Nx + T; P < 0.05 versus HR + T in fraction 8.

Discussion

The BBB has a critical role in maintaining central nervous system (CNS) homeostasis. Inability of the BBB to perform its normal physiologic functions is thought to contribute to numerous disease states affecting the CNS, many of which have an HR component (Hawkins and Davis, 2005; Zlokovic, 2008). Cerebral HR occurs during stroke and microvascular infarcts, vascular dementia, cardiac arrest, postoperative cognitive dysfunction, acute respiratory distress syndrome, obstructive sleep apnea, and high-altitude cerebral edema.

We have previously shown that HR causes an increase in paracellular permeability at the BBB (Witt et al, 2003; Witt et al, 2008). We have also shown changes in occludin protein localization and oligomeric assemblies that are associated with HR (McCaffrey et al, 2009). Oligomers of occludin are key components of the TJ protein complexes that are critical for restricting parcellular diffusion of solutes from the blood into the brain. Little is known about the mechanisms that lead to changes in occludin oligomeric assembly at the BBB after HR. Hypoxia and reoxygenation is associated with the production of ROS, which can lead to oxidative stress and BBB disruption (Pun et al, 2009). Recent evidence indicates that the oligomerization of occludin is redox sensitive (Walter et al, 2009). These observations led us to investigate the influence of oxidative stress on BBB permeability and occludin structural integrity during HR. To examine this possibility, we used the free radical scavenger tempol to pharmacologically inhibit production of ROS.

To confirm that the brains of rats exposed to our HR treatment are indeed hypoxic, we stained isolated microvessels for Hif-1α. Hypoxia-inducible factor-1α is a transcription factor that is constitutively degraded under Nx conditions. Hypoxia causes Hif-1α to stabilize, which allows it to translocate to the nucleus, dimerize with Hif-1β, and regulate gene transcription. Our confocal microscopy data show increased colocalization of Hif-1α with the nuclear dye TO-PRO-3 in HR-treated rats, which indicates that Hif-1α is stabilized and does translocate to the nucleus in response to HR. Hypoxia and reoxygenation does not cause a significant increase in rats pretreated with tempol. This may be because of an increase in ROS production during HR. Reactive oxygen species have previously been shown to induce Hif-1α nuclear translocation independent of hypoxia (Haddad and Land, 2001). Because tempol acts as a ROS scavenger, our data suggest that the translocation of Hif-1α to the nucleus was attenuated because of a decrease in ROS production in HR + T rats.

As a marker of cellular stress, we also stained microvessels for HSP-70. Heat shock protein 70 has previously been shown to be upregulated because of stressors such as hypoxia and/or oxidative stress. Our data show prominent HSP-70 immunoreactivity in the brain microvasculature during HR; however, pretreatment with tempol attenuated the HR-associated increase in HSP-70 expression. This suggests a reduction in cellular stress in the brains of rats administered with tempol.

Using the in situ brain perfusion technique, we examined the role of oxidative stress on BBB permeability after an acute exposure to HR in vivo. This technique has advantages over other techniques of perfusion or injection of solutes such as the brain uptake index technique (Zlokovic et al, 1985). The in situ brain perfusion technique is much more sensitive than techniques using single-pass extraction. It also avoids confounding physiologic variables such as differences in blood pressure and peripheral metabolism, and allows one to measure permeability of substances directly at the BBB (Martel et al, 1997). Using ¹⁴C-sucrose as a marker of paracellular permeability (Bhattacharjee et al, 2001), we found that administration of tempol 10 min before HR treatment prevented the increase in BBB permeability associated with HR. This suggests that oxidative stress mediates the increase in BBB permeability seen in our model of acute HR. There was also a slight but nonsignificant increase in BBB permeability in Nx + T. Tempol has been shown to increase bioavailability of nitric oxide (NO) that reacts with O₂^.- under physiologic conditions to produce peroxynitrite (ONOO⁻) (Zollner et al, 1997). Elevated levels of NO have previously been shown to increase BBB permeability in vitro and in vivo (Mark et al, 2004; Boje and Lakhman, 2000). Therefore, the high dose of tempol used in this study might have scavenged O₂^.- and reduced the number of free radicals available to interact with NO. This pharmacological effect of tempol may lead to an increase in BBB permeability mediated by NO.

Paracellular permeability at the BBB is regulated, in part, by the TJs. Our data showing that tempol prevents the increase in BBB permeability to ¹⁴C-sucrose after HR treatment suggest that oxidative stress may also affect the molecular structure of the TJs. Previous work by our laboratory has indicated that the TJ protein occludin is affected by HR (McCaffrey et al, 2009), and its ability to oligomerize is redox sensitive (Walter et al, 2009). These observations led us to investigate the effects of oxidative stress on occludin at both the cellular and molecular level using confocal microscopy and density gradient subcellular fractionation followed by western blot analysis.

Confocal microscopy of isolated cerebral microvessels immunostained for occludin revealed continuous distribution of occludin along the margins of the TJ. Hypoxia and reoxygenation induced intracellular punctate staining of occludin in regions of the microvessel that were not associated with TJs. In contrast, microvessels from HR + T rats show staining of occludin at the TJ in a pattern similar to Nx + S rats, suggesting that tempol may prevent HR-associated changes in occludin localization. Taken together, our data imply that oxidative stress may induce changes in the cellular localization of occludin at the BBB.

We have previously used a detergent-free density gradient centrifugation method to examine the distribution of occludin in plasma membrane microdomains of cerebral microvessels. These studies have revealed valuable insight into the structure of occludin at the BBB. Density gradient fractionation of intact microvessels has shown that occludin exists in the TJ-associated lipid raft regions corresponding to fractions 7 and 8, in which it colocalizes with the lipid raft marker caveolin-1 and the TJ marker Rab-13 (McCaffrey et al, 2007). We have also used PFO (which stabilizes the quaternary structure of protein complexes) to solubilize occludin oligomeric assemblies (Ramjeesingh et al, 1999). Extracts of PFO-soluble occludin subjected to nonreducing PFO-PAGE have revealed that in TJ-associated fractions 7 and 8, occludin exists almost exclusively as high MW (> 250 kDa) oligomers (McCaffrey et al, 2009). It is only when occludin is solubilized in SDS and/or subjected to reducing conditions that lower MW isoforms are detected. These data strongly suggest that occludin exists as high MW oligomeric assemblies at the TJs of the BBB in vivo. We have also shown through the use of hydrophobic and hydrophilic reducing agents that these occludin oligomeric assemblies are held together by disulfide bonds (McCaffrey et al, 2007). It has been shown that the C-terminus of the occludin molecule is involved in forming these disulfide bonds and the formation of these bonds is redox sensitive (Walter et al, 2009). These observations led us to investigate the effects of oxidative stress on the structural integrity of high MW occludin oligomers.

Using a C-terminal antibody to occludin, we subjected equal amounts of protein in fractions 7 and 8 to SDS-PAGE under nonreducing and reducing conditions. Our data show very faint banding patterns of occludin oligomeric assemblies located primarily in fraction 7 of Nx rats. Under reducing conditions with the hydrophilic reducing agent TCEP, these bands become much more intense. This suggests that under nonreducing conditions, the antigen is less accessible to antibody, presumably because it is being masked. Reducing conditions cleave these disulfide bonds and allow the antibody greater access to the antigen. Remarkably, oligomeric assemblies of occludin can still be detected after occludin has been subjected to SDS-PAGE under hydrophilic reducing conditions. This suggests an inner core of hydrophobic disulfide bonds holding the oligomeric assemblies together. In contrast to what was observed in Nx rats, HR rats show a prominent band of occludin oligomers under nonreducing conditions. These data suggest that the antibody binds to nonreduced occludin more readily in rats subjected to HR. This indicates that HR may induce conformational changes in occludin oligomers, possibly through the cleavage of disulfide bonds. In addition, occludin redistributes from fraction 7 to the higher density fraction 8 in the HR rats. Taken together, these data suggest that HR affects both the localization and the structural integrity of occludin oligomeric assemblies. Tempol-treated HR rats show significantly less dense bands of occludin oligomers under nonreducing conditions compared with HR + S rats. This suggests that oxidative stress is involved in the conformational changes of occludin oligomeric assemblies, and free radical scavengers such as tempol can preserve the structural integrity of occludin at the BBB during an HR insult.

In summary, our data suggest that ROS produced during HR may cause disruption of the BBB, leading to an increase in permeability to ¹⁴C-sucrose. In addition, ROS may cause changes in localization and structure of the TJ protein occludin. Both the structure and the localization of occludin are critically important to its function in the maintenance of the BBB's ability to restrict permeability to solutes in the systemic circulation. Changes in occludin localization and structure during HR can be pharmacologically inhibited by pretreating rats with the stable, BBB-permeable molecule tempol. Hypoxia and reoxygenation and/or oxidative stress are important components of several disease states (i.e., cardiac arrest, stroke, high-altitude cerebral edema) that exhibit a breakdown in the BBB and subsequent neurotoxicity and cognitive impairment. It is possible that ROS produced during the HR phase of these pathologic conditions induces changes in occludin structure and localization, leading to a breakdown in the ability of the BBB to protect the CNS. Treatment of these conditions with antioxidant therapy may preserve the functional integrity of the BBB, which may ultimately decrease the amount of CNS damage associated with several pathologic conditions.