Matrix Metalloproteinase-Mediated Disruption of Tight Junction Proteins in Cerebral Vessels is Reversed by Synthetic Matrix Metalloproteinase Inhibitor in Focal Ischemia in Rat

Abstract

Matrix metalloproteinases (MMPs) disrupt the blood—brain barrier (BBB) during reperfusion. Occludin and claudins are recently described tight junction proteins (TJPs) that form the BBB. We hypothesized that the opening of the BBB was because of the degradation of TJPs by the MMPs. Spontaneously hypertensive rats had a 90 mins middle cerebral artery occlusion with reperfusion for 2, 3, or 24 h. Matrix metalloproteinases were measured by immunohistochemistry and in situ and gel zymography. Real-time polymerase chain reaction (PCR) measured mRNAs of MMP-2 and −9, furin, membrane-type MMP (MT1-MMP), occludin, and claudin-5. There was opening of the BBB in the piriform cortex after 3 h of reperfusion, and an MMP inhibitor, BB-1101 (30 mg/kg), prevented the opening. At 3 h, in situ zymograms showed gelatinase activity. Zymography and PCR showed greater increases in MMP-2 than in MMP-9. There were increased mRNA and immunohistochemistry for MT1-MMP and furin, which activate MMP-2. Claudin-5 and occludin mRNA expression decreased at 2 h in both hemispheres with fragments of both proteins seen on Western blot by 3 h on the ischemic side; treatment with BB-1101 reversed the degradation of the TJPs. Immunohistochemistry at 3 h showed fragmented TJPs within the endothelial cell clefts. By 24 h, in situ zymography showed gelatinase activity and gel zymography showed elevated levels of MMP-9. Disrupted TJPs previously seen in endothelial cells appeared in the surrounding astrocytes. Our results provide direct evidence that MMPs open the BBB by degrading TJPs and that an MMP inhibitor prevents degradation of the TJPs by MMPs.

Keywords

blood—brain barrier opening claudin-5 matrix metalloproteinases middle cerebral artery occlusion occludin rat

Introduction

Tight junction proteins (TJPs) in endothelial cells are a major structural component of the blood—brain barrier (BBB) formed by components of the neurovascular unit (Ballabh et al, 2004; Bazzoni and Dejana, 2004). During ischemia/hypoxia in endothelial cell cultures and in situ models of BBB, TJPs are disrupted (Huber et al, 2001; Mark and Davis, 2002; Witt et al, 2003; Yamagata et al, 2004). The TJPs include zona occludens (ZOs) proteins in the endothelial cells, while in the endothelial cell clefts are found occludin and claudins (Furuse et al, 1993, 1998; Morita et al, 1999a; Nitta et al, 2003). Proteases of the serine and matrix metalloproteinase (MMP) gene families have been shown to degrade proteins of the basal lamina around the cerebral blood vessels (Hamann et al, 1995; Fukuda et al, 2004). An earlier study showed that gelatinase B (MMP-9) degrades ZOs in focal ischemia (Asahi et al, 2001). Less is known about the effect of MMPs on the major proteins in the endothelial cell clefts, namely, occludin and claudin. Studies of cultures of retinal vessels showed that the blood—retinal barrier is altered by MMP-9 through its action on claudin-5 and occludin (Lohmann et al, 2004; Giebel et al, 2005). Increased expression of MMPs by cultured human umbilical vein endothelial cells induced a shift in the molecular size of occludin from 65- to 50-kDa, which was blocked by the MMP inhibitor 1,10-phenanthroline (Wachtel et al, 1999). An increase in the 50-kDa occludin was seen in cultured human corneal epithelial cells that were treated with active MMP-9 in different concentrations (Pflugfelder et al, 2005). These studies in other tissues suggest that MMPs may affect BBB permeability by acting on the TJPs between endothelial cells as well as on the basal lamina proteins in focal ischemia.

Focal ischemia with reperfusion causes a biphasic opening of the BBB (Kuroiwa et al, 1985). Gelatinase A (MMP-2) has been associated with the initial opening, which takes place within several hours (Chang et al, 2003). This early opening is blocked by broad-spectrum MMP inhibitors, which have been used to block the toxicity of recombinant tissue plasminogen activator (Pfefferkorn and Rosenberg, 2003). Therefore, we hypothesized that disruption of the BBB corresponded with degradation of TJPs, which was mediated by MMPs. We also postulated that MMP inhibitors acted by preventing TJP degradation. To test the hypothesis we studied spontaneously hypertensive rats (SHR) that had undergone a middle cerebral artery occlusion (MCAO) with reperfusion for various times. We measured mRNA for MMP-2 and −9, membrane-type MMP (MT1-MMP), furin, and TJPs. Membrane-type MMP activates MMP-2 and furin activates MT1-MMP (Strongin et al, 1995; Sato et al, 1996). Gelatinases were measured by gelatin and in situ zymography and TJPs by immunohistochemistry and Western blots. Since the interactions between endothelial cells and astrocytes are required for induction of BBB properties, the morphologic relationship of TJPs with astrocytes was analyzed (Duffy et al, 2000; Willis et al, 2004; Garcia et al, 2004).

Materials and methods

Middle Cerebral Artery Occlusion with Reperfusion

The study was approved by the University of New Mexico Animal Care Committee and conformed to the National Institutes of Health Guidelines for use of animals in research. Male SHR, weighing 300 to 320 g, were anesthetized with 1.5% halothane in 70% nitrous oxide and 30% oxygen. At 90-mins MCAO was performed by insertion of an intraluminal nylon suture with a bulb on the end as previously described (Rosenberg et al, 2001). Briefly, neck vessels were exposed through a midline incision, and branches of the right external carotid artery were isolated and ligated. A 6-0 silk suture was loosely tied around the external carotid artery stump. A 4-0-monofilament nylon suture was introduced into the external carotid and advanced into the MCA. A silk suture around the stump was tied down onto the thread with the end of the thread protruding slightly. Reperfusion was achieved by slowly pulling the thread back after 90 mins occlusion. For inhibitor studies, the MMP inhibitor, BB-1101 (Generous gift from A Gearing, British Biotechnology, Oxon, UK), was delivered in 30 mg/kg body weight by intraperitoneal injection 10 mins before the MCAO started. Sham animals underwent neck surgery with a suture inserted, but the suture was rapidly removed.

Blood—Brain Barrier Permeability Observed with Evans Blue and Quantified with 14C-Sucrose

Blood—brain Barrier disruption after reperfusion in MCAO was visualized by injection of 2% Evans blue in saline injected intravenously via femoral vein immediately at the onset of reperfusion in the 3 h reperfusion experiments and 2 h before the end of the 24 h reperfusion experiments. Three or 24 h after reperfusion, the animals were killed, and the brains were removed and rapidly frozen in isopentane at −80°C. Frozen brains were cut into 16 to 20-μm-thick sections with a cryostat and mounted for fluorescent microscopy (Olympus BX51, Olympus Optical Co. Ltd, Japan). Leakage of Evans blue appeared as a red fluorescence.

Brain uptake of sucrose was measured in 10 rats that had undergone 90 mins of MCAO with reperfusion for 3 h. Blood—brain barrier permeability was measured by a modification of the brain uptake method (Rosenberg et al, 1998). At 10 mins before death, the rats were anesthetized with 1.5% halothane in 70% N₂O and 30% O₂ and infused intravenously over 30 secs with 10 μCi of ¹⁴C-sucrose (Dupont/New England Nuclear). Samples of blood were drawn from the femoral vein, and the heart was stopped with an intracardiac injection of a saturated potassium chloride solution. The brains were removed and frozen in 2-methylbutane cooled to −80°C. Tissue sections were taken from the infarcted cortex and caudate and similar noninfarcted regions. Brain and blood samples were dissolved in Aquasol (New England Nuclear) and were counted for radioactivity in a liquid scintillation counter (Packard). The ratio of brain sucrose content to that in the blood was calculated.

Gelatin Zymography

For zymography, tissues were taken from a contiguous 5-mm coronal section directly posterior to the one used for BBB measurements. Similar regions of the infarcted and noninfarcted sides were studied. The frozen tissues were homogenized with lysis buffer (50 mmol/L Tris-HCl pH 7.6, 150 mmol/L NaCl, 5 mmol/L CaCl₂, 0.05% Brij-35, 0.02% NaN₃, and 1% Triton X-100). The protein concentration in the homogenate was determined using Bradford reagent (Bio-Rad). Matrix metalloproteinase-2 and −9 in homogenates were concentrated with gelatin-sepharose beads (Gelatin Sepharose 4B, Amersham Biosciences, Piscataway, NJ, USA) and analyzed by gelatin zymography as previously reported with modifications (Planas et al, 2000). Briefly, 0.8 mg protein in 500 μl was incubated with 80 μl Gelatin-Sepharose 4B beads for 1 h at 4°C with gentle rotation. The beads were collected and the MMP-2 and −9 were eluted by incubating with 80 μl elution buffer (10% dimethyl sulfoxide in phosphate-buffered saline (PBS) for 30 mins at 4°C with gentle shaking. Equal amounts of samples (20 μl) were electrophoretically separated on 10% sodium dodecyl sulfate -polyacrylamide gels co-polymerized with 1 mg/ml gelatin (Sigma-Aldrich Chemicals, St Louis, MO, USA) under nonreducing conditions. Gels were washed in 2.5% Triton X-100 and then incubated for 40 h with a developing buffer containing 50 mmol/L Tris, pH 7.6, 5 mmol/L CaCl₂, 0.2 mmol/L NaCl, and 0.02% (w/v) Brij-35 at 37°C before they were stained with 0.125% Coomassie blue R-250 for 30 mins in 10% (v/v) acetic acid and 50% methanol. Gels were destained with a solution containing 10% acetic acid and 10% methanol until clear bands of gelatinolysis appeared on a dark blue background. To confirm that detected activities were zinc-dependent gelatinases, some zymogram gels were incubated with developing solution in the presence of 10 mmol/L ethylenediaminetetraacetic acid (EDTA). Disappearance of lytic bands in these gels confirmed the metal dependence of gelatinolytic activity characteristic of MMPs. The gels were dried and scanned for densitometry (Alpha Imager™ 2200; Alpha Innotech, San Leandro, CA, USA). A mixture of human MMP-9 and −2 (Chemicon International, Inc. Temecula, CA, USA) was used as gelatinase standards.

In Situ Zymography on Brain Sections with Evans Blue

Sections from brain tissues injected with Evans blue as described above were used for in situ zymography after reperfusion (Rivera et al, 2002). The frozen, nonfixed sections were incubated in a humidity chamber for 3 h at 37°C in a reaction buffer containing 40 μg/ml of fluorescein isothiocyanate (FITC)-labeled DQ-gelatin (EnzCheck Collagenase Kit, Molecular Probes, Eugene, OR, USA). Gelatin-FITC is cleaved by gelatinases, yielding peptides whose fluorescence is representative of net proteolytic activity. Sections were rinsed three times in PBS and mounted in Gel/Mount™ (Biomeda Corp. Foster City, CA, USA) for fluorescence microscopy. In each experiment, some sections were preincubated with blocking antibodies against MMP-2 (dilution 1:300) and MMP-9 (dilution 1:300) for 1 h, or 1 mmol/L 1,10-phenanthroline, a potent zinc chelator that inhibits all MMP activities, in the reaction buffer (Rivera et al, 2002; Frederiks and Mook, 2004). Sections incubated without DQ-gelatin were not fluorescent.

Antibodies Used for Immunohistochemistry and Western Blots

Anti-MMP-2 antibody (Chemicon International, Inc. Temecula, CA, USA) is a polyclonal rabbit anti-rat antibody that recognizes rat MMP-2. Anti-MT1-MMP (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) is a goat polyclonal antibody against a peptide mapping near carboxy terminus of MT1-MMP of human origin. Antifurin antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) is a goat polyclonal antibody against a peptide mapping near carboxy terminus of furin of human origin. Antioccludin and anticlaudin-5 antibodies were purchased from Zymed Laboratories (South San Francisco, CA, USA). Antioccludin (0.25 mg/ml) is a rabbit polyclonal mouse antibody specific for the occludin protein (65 kDa) directed against 150 amino acids in the carboxy terminal region of human occludin. Anti-claudin-5 (0. 5 mg/ml) is a mouse monoclonal antibody that reacts specifically with the 22–24 kDa Claudin-5. Rabbit and mouse monoclonal antiglial fibrillary acidic protein (GFAP) antibodies (Sigma-Aldrich Chemicals, St Louis, MO, USA) were used for localizing GFAP in astrocytes. All of the antibodies above are documented to react with rat by immunoblot and immunohistochemistry. Mouse antineuron-specific nuclear protein (NeuN) (Chemicon International, Inc., Temecula, CA, USA) was used as a marker for identifying neurons. Anti-mouse IgG, anti-rabbit IgG, anti-goat, and anti-rabbit IgG horseradish peroxidase-conjugated secondary antibodies (Jackson ImmunoResearch Laoratories, Inc., USA) were used.

Immunohistochemistry

Immunohistochemistry for MT1-MMP and furin was performed in MCAO rats after reperfusion for 3 h. Ten-μm-thick cryosections of PLP (2% paraformaldehyde, 0.1 mol/L sodium periodate, 0.075 mol/L lysine in 100 mmol/L phosphate buffer at pH 7.3) fixed brains were rinsed in PBS and preincubated in cold acetone for 5 mins. Endogenous peroxidase activity was inhibited with 0.1% H₂O₂ in methanol for 10 mins. Nonspecific binding sites were blocked by preincubating tissue for 1 h at room temperature in PBS containing 0.1% Tween-20, 1% bovine serum albumin, and 5% normal serum. Sections were then incubated at 4°C overnight in a PBS solution containing the primary antibodies diluted 1:300 (MT1-MMP) or 1:300 (Furin) in PBS, 0.1% Tween-20, 0.5% bovine serum albumin. Tissue was processed as described above with HRP-labeled donkey anti-goat secondary antibodies for 90 mins at room temperature. Immunostaining was visualized with DAB (DAKO Corporation) and sections were counterstained with NeuroTACS™ Blue counterstain (Trevigen Inc., Galthersburg, MD, USA). Double fluorescent staining of MMP-2 (1:300) with MT1-MMP, NeuN, GFAP (1:500), claudin-5 (1 μg/ml), and occludin (1:500) was performed in sections as mentioned above. Sections were rinsed in PBS, preincubated as indicated above without 0.1% H₂O₂ and then incubated overnight at 4°C with the primary antibodies. Tissue was rinsed and incubated for 2 h at room temperature with secondary antibodies labeled with FITC or Cyanine (CY3), then rinsed and mounted in Prolong Antifade (Molecular Probes, Eugene, OR, USA). Slides were photographed with a fluorescent microscope (Olympus BX51, Olympus Optical Co. Ltd, Japan) and a confocal microscope (Zeiss LSM 510, Carl Zeiss MicroImaging, Thornwood, NY, USA). Sections incubated in the absence of the primary antibody, with rabbit IgGs were not immunofluoresent.

Quantitative Real-Time Polymerase Chain Reaction

Quantitative real-time PCR was performed on total RNA isolated from brain tissues of SHR animals exposed to 90 mins of ischemia with reperfusion for 2 h to quantify the mRNA expression of MMPs, furin, and TJPs. Ten μg RNA was reverse-transcribed with MultiScribe reverse transcriptase and random hexamers (Applied Biosystems, Foster City, CA, USA). Primers for the quantitative realtime PCR are listed in Table 1. Polymerase chain reaction was performed according to the following protocol: 2 mins hold at 50°C, 10 mins hold at 95°C, followed by 15 secs at 95°C, and 1 mins at 60°C (40 cycles). Amplification was quantified using the TaqMan™ probe real-time PCR kit and the SYBR Green kit (Applied Biosystems, Foster City, CA, USA). A linear concentration—response curve was established by diluting pooled samples. Primers and probes for 18S ribosomal RNA (TaqMan Ribosomal RNA Control Reagents, Roche) were also included for quantifying relative gene expression. A GeneAmp 5700 Sequence Detection System (Applied Biosystems) was used to measure the expression of the mRNA levels for MMPs, furin, and TJPs.

Table 1

Primers for real-time PCR

Name/Accession no.	Forward primer	Reverse primer	Length of amplicon (bp)
MMP-2/U65656	5′-GAT CTG CAA GCA AGA CAT TGT CTT-3′	5′-GCC AAA TAA ACC GAT CCT TGA A-3′	77
MMP-3/X02061	5′-TCC CAG GAA AAT AGC TGA GAA CTT-3′	5′-GAA ACC CAA ATG CTT CAA AGA CA-3′	74
MMP-9/U24441	5′-GTA ACC CTG GTC ACC GGA CTT-3′	5′-ATA CGT TCC CGG CTG ATC AG-3′	80
MT1-MMP/X83537	5′-CTG GAC TGT CCA AAA TGA GGA T-3′	5′-AAC TCA TGC ACA GCC ACC AA-3′	63
Furin/X55660	5′-AAC AAT GGT GTC TGT GGT GTA GGT-3′	5′-ACA TAC GCA CCC CTC CAA TC-3′	61
Claudin-5/XM_344058	5′-AAATTCTGGGTCTGGTGCTG-3′	5′-GCCGGTCAAGGTAACAAAGA-3′	284
Occludin/AB016425	5′-TGGGCAGTCGGGTTGACT-3′	5′-GGGCATCATGGTGTTCATTG-3′	60
Rp l/NM_013226	5′-TGT CCT CTA AGA ACC GAA AAG CC-3′	5′-CGT TGG GAT TGG TGA CTC TGA-3′	72

Western Blots of Tight Junction Proteins

Protein samples of brain tissues extracted by RIPA buffer were analyzed for protein expression of occludin and claudin-5, using Western blot. Samples (150 μg) were separated using a 15% Tris-HCl precast gel for polyacrylamide electrophoresis (Bio-Rad Laboratories, Hercules, CA, USA) at 100 V for 75 to 90 mins. Brilliant Blue R staining solution (Sigma-Aldrich Co., St Louis, MO, USA) was used to normalize protein loading. The proteins were transferred to polyvinylidene fluoride membranes with 100 V at 4°C for 60 mins. The membranes were blocked in PBS containing 5% nonfat milk and 0.5% Tween-20 for 4 h and were then incubated overnight at 4°C with primary antibodies against occludin or claudin-5 at a 1:500 dilution. The membranes were washed with PBS containing 0.5% Tween-20 before incubation with the respective secondary antibodies at a 1:5000 dilution for 75 mins at room temperature. Blots were developed using ChemiLucent Detection System Kit (Chemicon International, Inc. Temecula, CA, USA), and protein bands were visualized on X-ray film. Semiquantitation of the protein was performed with the use of Scion image software (Scion, Frederick, MD, USA), and the results are reported as normalized protein value by using the Brilliant Blue R staining on polyvinylidene fluoride for quantifying relative protein expressions.

Statistical Analysis

For all experiments, data were tested for statistical significance with a two-tailed unpaired t-test or a one-way ANOVA for multiple t-tests (Prism, GraphPad Software Incorporated, San Diego, CA, USA). The statistical analysis was run on the normalized data for real-time PCR and Western blot to determine significant differences. The statistical significance was set at P < 0.05. The data are expressed as mean ± standard error of the mean (s.e.m.).

Results

Matrix metalloproteinase-2 Gelatinolytic Activity in Piriform Cortex Corresponds with Early Opening of Blood—Brain Barrier

We used in situ zymography to identify the regions with gelatinolytic activity. After 3 h of reperfusion, there was a marked increase in gelatinase activity in the ischemic hemisphere that was mainly seen in the piriform cortex, corpus callosum, caudate, and lateral cortex near the meninges (Figure 1A). Preincubation of MMP-2 antibody reduced the fluorescence signal, and the metalloproteinase inhibitor, 1,10-phenanthroline, also reduced postischemic gelatinase activity (Figure 1A, inset).

Figure 1

(A) In situ zymography of SHR after 90-min MCAO with 3 h of reperfusion. There is activation of gelatinases in the piriform cortex (arrowhead sites 1 and 2) and lateral cortex near the meninges (arrowhead site 3) on the ischemic side. Activation is also seen in the region of the corpus callosum (arrowhead site 4) and slightly in the caudate (arrowhead site 5). Inset shows the site 1 and 2 of piriform cortex where the gelatinase activity is attenuated by the potent zinc chelator 1,10-phenanthroline (Phen), a wide-spectrum metalloproteinase inhibitor. (B) Matrix metalloproteinase-2 immunostaining in the region of the piriform cortex (arrowhead site 2 in (A)). An increase in the fluorescence is seen in the neurons primarily in the ischemic cortex, but also some staining was seen in the nonischemic side. (C) Evans blue was injected intravenously immediately after reperfusion. Evans blue (red) is seen in a vessel in the piriform cortex (arrow). In situ zymography shows gelatinase activity in the vessel (green). The merged image (yellow) shows the Evans blue in the gelatinase-expressing vessel. Inset shows the double-stained cells that have taken up Evans blue and expressed gelatinase activity.

Matrix metalloproteinase-2 immunostaining showed increased expression in the piriform cortex in the ischemic, but only minimally on the nonischemic side (Figure 1B). The Evans blue was seen grossly in the region of the piriform cortex where the gelatinase activity was prominent. To further confirm that the leaky vessels were associated with the increased gelatinolytic activity, we co-localized the fluorescence from the Evans blue and the in situ zymography. Sites of increased Evans blue signal in the ischemic hemisphere co-localized with the regions of gelatinase activity (Figure 1C).

As in situ zymography cannot distinguish between MMP-2 and −9, gelatin-substrate zymography was used to identify the source of gelatinase activity. Tissue samples from the piriform cortex showed a significant increase in the 72-kDa MMP-2 (P < 0.007), and a smaller, but significant increase in 92-kDa MMP-9 (P < 0.04) compared with the nonischemic side (Figure 2A). A 62-kDa form of the MMP-2 proform was seen in the ischemic tissues, indicating activation of MMP-2. There were no active bands of MMP-9 observed in the zymograms.

Figure 2

(A) A gelatin zymogram from tissues in the piriform cortex (arrowhead sites 1 and 2 in Figure 1A) in the region with the gelatinase activity shows marked expression of MMP-2 (72-kDa) and slight MMP-9 (92-kDa) expression in the ischemic side. In addition, lower molecular weight bands (62-kDa) from active MMP-2 are seen in the ischemic side. The graph is the optical density values of the lysis zones. While both MMP-2 and −9 reached statistical significance, there is a much larger increase in the band from MMP-2 (P < 0.007, n = 5). (B) ¹⁴C-sucrose space with tissue samples from the ischemic piriform cortex. The sucrose space is markedly increased (P < 0.0003, n = 5). A broad spectrum MMP inhibitor, BB-1101, blocked the opening of the BBB. No change in permeability is seen in the nonischemic side.

Matrix Metalloproteinase Inhibitor Reduced the Blood—Brain Barrier Opening

Qualitative evidence for disruption of the BBB particularly in the piriform cortex was obtained with Evans blue. We quantified the BBB injury in the piriform cortex, using the brain uptake of ¹⁴C-sucrose. After intravenous injection of the radiolabel, tissue samples were collected from the piriform cortex. We found a marked increase of the sucrose space in the ischemic hemisphere. The broad-spectrum MMP inhibitor, BB-1101, significantly reduced the BBB opening on the ischemic side (P < 0.0003) without affecting the nonischemic hemisphere, implicating MMPs in the increased vascular permeability (Figure 2B).

mRNA of Matrix Metalloproteinase-2 Activators Increased at 2 h

Matrix metalloproteinase-2 is secreted in a latent form that requires activation at the cell surface by the action of MT1-MMP, which is activated by the proconvertase, furin. We used real-time quantitative PCR to measure mRNA for MMP-2, MT1-MMP, and furin. By 2 h of reperfusion, there was an increase in mRNA for MMP-2 (Figure 3A). In addition, we found an increase in mRNA for MT1-MMP (Figure 3C), and furin (Figure 3D) in the piriform region of the ischemic hemisphere. At this early time point, there was no change seen in the mRNA for MMP-9 (Figure 3B).

Figure 3

Quantitative real-time PCR for mRNA expression of MMP-2, −9, MT1-MMP, and furin in animals with a 90-min MCAO and 2 h reperfusion (2 h RP). (A) MMP-2 mRNA showed a significant increase compared with sham-operated animals on the ischemic side (P < 0.01, n = 6), but not on the nonischemic side. (B) There was no increase for the MMP-9 mRNA. (C) MT1-MMP showed a significant increase (P < 0.02, n = 6). (D) Furin showed an increase on the ischemic side compared with sham-operated animals (P < 0.05, n = 6).

Immunohistochemistry Showed Matrix Metalloproteinase-2 in Astrocytes and Neurons

Immunohistochemistry for MMP-2 was performed to determine the cell types expressing the protein. There were very low levels of MMP-2 expression in nonischemic hemisphere, but increased MMP-2 immunoreactivity was seen in the ischemic hemisphere. To determine the cellular sources involved in increased expression of MMP-2, sequential double fluorescent immunostaining was observed by confocal microscopy. A strong association of MMP-2 immunoreactivity with GFAP staining was observed in the piriform cortex, corpus callosum, caudate, and the lateral cortex near the meninges, suggesting that astrocytes are an important source of MMP-2 (Figure 4A). Matrix metalloproteinase-2 co-localized with GFAP-expressing astrocytes in the region of the end feet next to the piriform cortex blood vessels where gelatinolytic activity was seen with in situ zymography and Evans blue leakage was also prominently observed. NeuN staining showed that gelatinase activity was located in neurons in lateral and piriform cortices (Figure 4B). MT1-MMP immunostaining co-localized with MMP-2 in astrocytes, where it was seen mainly around the periphery of the cells (Figure 4C).

Figure 4

Confocal micrographs showing the immunostaining for MMP-2 and MT1-MMP in glial cells and neurons in the ischemic hemisphere. (A) Matrix metalloproteinase-2 co-localized with astrocytes (GFAP) in the end feet abutting the blood vessels. (B) Matrix metalloproteinase-2 (Cy-3) co-localized with NeuN (FITC). (C) MMP-2 and MT1-MMP are co-localized. Scale bars indicate 10 μm.

DAB immunostaining showed the MT1-MMP around the blood vessels in the ischemic side in regions similar to those that showed staining for furin, while both were absent on the nonischemic side (data not shown).

Reorganization of Claudin-5 and Occludin Occurs Rapidly in Reperfusion Injury

The effect of focal ischemia on TJPs was studied by immunohistochemistry, PCR, and Western blot. We studied the two major proteins involved in the tight junctions of cerebral vascular endothelia, claudin-5 and occludin. Claudin-5 was found in blood vessels in the sham-operated animals, forming straight rows of staining (Figure 5A). Astrocyte end feet around the blood vessels were completely separated from the claudin-5. Claudin-5 immunoreactivity in nonischemic cortex was similar to that seen in sham-operated animals (Figure 5B). Focal ischemia caused a fragmentation of the claudin-5 (Figure 5C). This resulted in a decrease in the intensity of the claudin-5 staining on the ischemic side, which was absent in some vessels, suggesting that the claudin-5 was reorganized or broken down.

Figure 5

Confocal micrographs showing claudin-5 and occludin immunoreactivity in the sham-operated animals and nonischemic hemispheres after 3 h of reperfusion. (A and B) The sham-operated animals and the nonischemic side show that the claudin-5 (Cy-3) in blood vessels is separated from the astrocytes (GFAP-FITC) surrounding them. The merged images show that the claudin-5 and astrocytes are separate. (C) In the ischemic hemisphere, there is fragmentation and degeneration of the claudin-5 immunoreactivity. Co-localization of claudin-5 and GFAP was seen in the ischemic hemisphere. (D) The nonischemic side shows that minimal MMP-2 immunoreactivity was detected in blood vessels where claudin-5 presented straight rows of staining that was similar to that seen in nonischemic cortex in (B). (E) Confocal micrographs showing MMP-2 immunoreactivity surrounding blood vessels co-localized with claudin-5 in the ischemic hemispheres after 3 h of reperfusion. (F and G) Confocal micrographs showing occludin immunoreactivity in the sham-operated animals and nonischemic piriform cortex. The nonischemic side shows the occludin (Cy-3) and the astrocytes (GFAP-FITC) were co-localized around blood vessels. (G) The occludin immunostaining in ischemic piriform cortex was seen with degeneration and reduced expression compared with that in nonischemic piriform cortex. Scale bars indicate 10 μm.

To further confirm that the fragmentation of the claudin-5 was associated with the increased MMP-2 activity, we co-localized the fluorescence from MMP-2 and claudin-5. In the nonischemic hemisphere, a cross-section of a blood vessel immunostained for claudin-5 shows the rod-like linear strands of protein as dots (Figure 5D). Matrix metalloproteinase-2 is absent from the region of a blood vessel in the nonischemic hemisphere, which is seen more clearly in the merged image (Figure 5D). In the ischemic hemisphere, a blood vessel sectioned at an angle shows attenuated strands of claudin-5 and increased MMP-2 immunoreativity surrounding a vessel. The merged image shows the co-localization of the MMP-2 around the claudin-5 (Figure 5E).

Occludin, another integral TJP, anchors the claudins to ZO-1 and ZO-2, and the plasma membranes of adjacent cells. We observed a continuous distribution of occludin immunostaining in the cerebral vessels in the sham-operated animals, which co-localized with GFAP-positive astrocytes (Figure 5F). A similar pattern of occludin immunostaining was seen in nonischemic cortex vessels (Figure 5G). However, in the ischemic piriform cortex, cerebral vessels showed a reduction in the intensity of occludin immunostaining, suggesting degradation of occludin protein (Figure 5H).

Real-time quantitative mRNA analysis at 2 h reperfusion for claudin-5 (Figure 6A) and occludin (Figure 6B) showed a statistically significant drop in the levels of both TJPs in the ischemic and nonischemic hemispheres.

Figure 6

Real-time PCR for the expression of TJP mRNAs. (A) Expression of claudin-5 mRNA after a 2-h reperfusion (2 h RP). There is a reduction in the levels of mRNA in ischemic and nonischemic hemispheres compared with sham-operated animals (P < 0.001, n = 6). (B) Occludin mRNA is similarly reduced in the ischemic and nonischemic hemispheres (P < 0.001, n = 6).

Loss of Claudin-5 and Occludin Protein Reversed by A Matrix Metalloproteinase Inhibitor

Claudin-5 has a molecular weight normally of 22-kDa. When it is degraded, there are lower molecular weight fragments seen. Immunoblots of TJPs after 3 h of reperfusion showed a marked drop in 22-kDa claudin-5 (P < 0.01) in the ischemic hemispheres (Figure 7A). In addition, reperfusion resulted in a significant increase in the 17-kDa claudin-5 fragment in the ischemic hemisphere (P < 0.01) (Figure 7A), suggesting that the claudin-5 molecule underwent degradation as observed in the claudin-5 immunostaining.

Figure 7

Western blots showing degradation of TJPs after a 90-min MCAO followed by 3h reperfusion (3 h RP). (A) The Western blot showed significant reductions (P < 0.01, n = 6) of claudin-5 (22-kDa) at 3 h in the ischemic piriform cortex (PFC). In addition, a new 17-kDa band for claudin-5 was found in ischemic hemisphere at 3h post reperfusion. The 17-kDa protein shows a significant increase (P < 0.01). Treatment with BB-1101 (3 h RP/BB1101) significantly reversed reduction of claudin-5 (22-kDa) (P < 0.02, n = 6). (B) A significant reduction of occludin (65-kDa) at 3h was seen in ischemic hemispheres (P < 0.02, n = 6) and treatment with BB-1101 reversed the reduction of occludin expression (P < 0.02, n = 6). A lower 60-kDa band was induced in both hemispheres (P < 0.003 in nonischemic, P < 0.02 in ischemic). BB-1101 blocked the appearance of the 60 kDa band (P < 0.05, n = 6).

Changes in expression of occludin after 90 mins MCAO with 3 h reperfusion were also demonstrated by Western blot (Figure 7B). Occludin normally has bands at 65- and 63-kDa (Figure 7B), which is indicative of phosphorylated (β-band) and nonphosphorylated (α-band) states, respectively (Witt et al, 2003). A significant reduction in the 65-kDa occludin band was seen by Western blot in the ischemic side (P < 0.002). A concomitant increase in the occludin fragment at a molecular weight of 60-kDa was seen in the blots in both nonischemic and ipsilateral hemispheres at 3h reperfusion (P < 0.003 and 0.02, respectively).

Since BB-1101 blocked the early opening of the BBB, we tested the effect of this broad-spectrum inhibitor on prevention of degradation of TJPs with Western blots. Quantification by densitometry showed that the reversal in the fall in the higher molecular weight band of claudin-5 was statistically significant, providing strong support for a major role of the MMPs in TJP reorganization (Figure 7A). Furthermore, BB-1101 significantly blocked the reduction in expression of occludin at 3 h reperfusion and the appearance of the third band (Figure 7B), indicating that MMPs induced occludin proteolysis after MCAO with reperfusion, but had a more complicated effect on claudin-5.

Matrix Metalloproteinase-9 Protein and mRNA Associated with Blood—Brain Barrier Opening at 24 h

At 24 h mRNA for MMP-9 increased without any changes in the MMP-2 mRNA, which was the reverse of the pattern found at 3 h (Figure 8A and 8B). Gelatin gel zymography confirmed that MMP-9 was significantly increased on the ischemic side, while the MMP-2 was not significantly altered (Figure 8C). The increased MMP-9 was associated with disruption of the BBB on the ischemic side (Figure 8D).

Figure 8

(A) There was no increase in MMP-2 mRNA measured by real-time PCR after 24 h of reperfusion (n = 5). (B) An increase in MMP-9 mRNA was seen (n = 5). (C) MMP-9 protein increased as shown by gelatin zymography at 24 h (n = 5), but not for MMP-2. (D) ¹⁴C-sucrose space was significantly increased in the ischemic side (n = 3). Asterisks indicate P < 0.001 significance.

In the animals studied after 24 h of reperfusion, we observed intense gelatinase activity in the in situ zymograms in the gray matter of the lateral and piriform cortices and lateral caudate (Figure 9A). Preincubation of the tissue sections with antibodies against MMP-2 and −9 dramatically reduced the fluorescence, showing that the gelatinolytic activity was from either MMP-2 or −9 (Figure 9A, insets).

Figure 9

(A) In situ zymography in a 90-min MCAO with 24 h of reperfusion. There is extensive gelatinase activity in the lateral aspect of the caudate and the lateral cortex in ischemic side. Insets show piriform cortex where the gelatinase activity is attenuated by pre-incubation with antibodies (AB) against MMP-2 and MMP-9. (B) Confocal micrographs showing immunoreactivity of claudin-5 with GFAP at brain vessels in nonischemic hemisphere with 24 h reperfusion. Images were taken from penumbral areas in the caudate and lateral cortex. (C) Ischemia caused a loss of claudin-5 (Cy-3) from the blood vessels at 24 h. (D) There was co-localization of the claudin-5 in the GFAP-reacting astrocytes. These photos were taken from penumbral areas in the caudate and lateral cortex. Scale bars indicate 10 μm in (B and C) and 5 μm in (D).

Claudin-5 and Occludin Translocate to Astrocytes at 24 h

A different pattern of claudin-5 and occludin immunoreactivity was seen at 24 h in the ischemic hemisphere. Blood vessels on the nonischemic hemisphere showed the characteristic linear patterns of claudin-5 with clearly marked separations between the GFAP-positive astrocytic foot processes and the claudin-5 immunostaining and a continuous rim or line of immunostaining in the endothelial cells (Figure 9B). However, ischemia caused a loss of the claudin-5 immunoreactivity of blood vessels, and a disrupted appearance of astrocytic foot processes around the endothelial cells (Figure 9C). We observed GFAP-positive immunoreactivity co-localized with a diffuse cytoplasmic claudin-5 immunostaining on the ischemic hemisphere in the lateral and piriform cortices (Figure 9D). Similar results were obtained when sections were double labeled for occludin and GFAP (data not shown).

Western blots showed a significant reduction of 22-kDa claudin-5 in both hemispheres at 24 h (P < 0.05, 0.01, respectively, n = 6; Figure 10A). There was a significant increase in the 17-kDa fragment of claudin-5 that was only present in the ischemic hemisphere (P < 0.002). On the contrary, the decrease in occludin, although present in both hemispheres failed to reach statistical significance, but there was a significant increase in the lower molecular weight fragment (P < 0.0001; Figure 10B).

Figure 10

(A) Western blots of claudin-5 showed a reduction of the higher molecular weight form (22-kDa) in both the ischemic and nonischemic hemispheres compared with sham-operated animals (P < 0.05, 0.01, respectively, n = 6). A significant increase in the 17-kDa claudin-5 was seen in the ischemic hemisphere (P < 0.002). (B) Western blots of occludin at 24 h showed no significant reductions in the 65-kDa form with a significant increase in the 60-kDa form in the ischemic side (P < 0.0001).

Discussion

The early opening in the biphasic pattern of BBB changes seen after reperfusion has been shown to be important in the toxicity of recombinant tissue plasminogen activator, and closing the BBB with inhibitors to MMPs dramatically reduces the toxicity. Our results show that there is a structural change in the proteins forming the tight junctions of the BBB that leads to increased permeability and that a broad-spectrum MMP inhibitor prevents the changes in the TJPs. We found that the early change in permeability was related to an increase in mRNA and protein of MMP-2, and that the major activator of MMP-2, MT1-MMP, and the activator of MT1-MMP, furin, were also increased. More importantly, there was a change in the gelatinolytic activity as measured by in situ zymography in the regions where the TJPs were disrupted. Western blots showed that the TJPs were degraded both at the early time point and in the delayed injury to the BBB, and that the degradation was partially reversed by the MMP inhibitor. The TJPs remained within the endothelial cell clefts during the initial opening and appeared to be attenuated or fragmented. However, a strikingly different pattern of TJPs was seen in the brains of the animals reperfused for 24 h. We observed that the TJPs, which were decreased in the endothelial cells, were co-localized with astrocytes in the region of the vessels. This could explain the reversibility of the early opening and the more prolonged injury associated with the delayed injury to the BBB. Our results suggest that the initial reversible opening of the BBB is related to activation of the constitutively expressed MMP-2, and that the TJPs may be loosened, but remain within the endothelial clefts. However, with lapse of time and increased neuroinflammation resulting in the expression of MMP-9 and possibly other proteases and free radicals, the TJPs are further degraded in the endothelial cell clefts. These results provide a molecular basis for both the transient and long-term alterations in the BBB due to the MMPs in reperfusion injury, and they provide a molecular explanation for the beneficial effects of MMP inhibitors at the early reperfusion times.

Earlier studies provided indirect evidence that MMP-2 played a key role in the initial opening of the BBB in rat brain (Rosenberg et al, 1998), and in the non-human primate brain (Heo et al, 1999; Chang et al, 2003). We combined in situ zymography, which identified the regions where the gelatinases, either MMP-2 or −9, were active, and gel zymography, which showed that the major form was MMP-2. In addition, the presence of mRNA and protein for the activators of MMP-2, namely, MT1-MMP and furin, further supports that MMP-2 was the major enzyme involved in the early TJP disruption. These findings support the studies in the non-human primate (Fukuda et al, 2004). Immunohistochemistry showed that MMP-2 protein was markedly expressed in the ischemic regions in both neurons and astrocytes. We co-localized the MMP-2 and MT1-MMP in the neurons and the astrocytic foot processes. While the MMP-2 in the astrocytic foot processes abutting the cerebral blood vessels most likely contributes to the changes in BBB permeability, the role of the MMP-2 in the neurons remains to be elucidated.

A very different picture emerged at 24 h where the pattern of gelatinase activity in the in situ zymography was much more extensive and the dominant gelatinase expressed in gelatin zymograms was MMP-9. There was expression of MMP-9 mRNA, but not that of MMP-2, which was consistent with the protein expression. Earlier studies had demonstrated expression of MMP-9 at 24 h (Rosenberg et al, 1996; Romanic et al, 1998; Gasche et al, 1999). A striking feature of the later injury to the BBB was the appearance of the TJPs in the astrocytes. Further studies will have to be performed to determine the source of the TJPs in astrocytes.

Tight junction proteins join endothelial cells together to form the interface between blood and brain (Ballabh et al, 2004). In brain, claudin-1 and claudin-5, together with occludin have been described to be present in endothelial TJs forming the BBB and to be both required for formation of the BBB (Morita et al, 1999b; Liebner et al, 2000). Claudins form the primary seal of the tight junction. Claudin-5 is a 22-kDa phosphoprotein with four transmembrane domains. Loss of claudin-1 from cerebral vessels was demonstrated under pathologic conditions such as tumor, stroke, and inflammation (Liebner et al, 2000; Lippoldt et al, 2000). Occludin is a 65-kDa phosphoprotein that assembles with claudin-5 into heteropolymers and intramembranous strands; it is a regulatory protein that can alter paracellular permeability (Tsukita and Furuse, 2000; Hirase et al, 2001; Hawkins and Davis, 2005). Zona occludens protein was shown to be degraded in ischemic injury in the mouse brain, but occludin was found to be unaffected (Asahi et al, 2001). Breakdown of BBB in tissue surrounding brain tumors occurs with concomitant loss of occludin expression (Papadopoulos et al, 2004). In vitro, hypoxia with reoxygenation altered occludin location and molecular weight, which correlated with the observed changes in cerebral microvascular permeability (Mark and Davis, 2002; Witt et al, 2003).

Cerebral blood vessels in the nonischemic piriform cortex were immunostained with claudin-5, which has been described in cerebral endothelial cells. The claudin-5 appeared in linear rows, which has been referred previously to have a ‘zip-locked' appearance (Tsukita et al, 2001). Several possible mechanisms could explain the apparent loosening of the tight junctions. MMPs are released by astrocytic foot processes where they could degrade the basal lamina proteins, which may be providing structural support for the capillary. Another possible explanation is that the attack on the basal lamina exposes the TJPs, which are hydrolyzed by MMPs contained in the blood vessels, such as, MMP-9. The fall in TJP mRNAs in both hemispheres suggests a generalized suppression of synthesis because of the ischemic insult. Another possibility is a decrease in stability of the mRNAs. Further studies will be needed to clarify this issue.

The second opening of the BBB at 24 h is more complicated and results in greater tissue damage. By that time the inflammatory response secondary to ischemia/hypoxia with reperfusion has fully developed, and many additional factors are involved, including cytokines, MMP-3 and −9, and infiltrating leukocytes (Lo et al, 2003). Matrix metalloproteinase-9 is released by several types of cells as well as by infiltrating neutrophils and lymphocytes, making it a highly toxic substance involved in the induction of edema, hemorrhage and cell death. Astrocytes in culture exposed to interleukin-1β cease producing the gap junction protein, C_x43, and begin to produce claudin-1 (Duffy et al, 2000). Claudin promotes the activation of MMP-2, which would allow for amplification to take place, increasing the extent of the injury to the BBB (Miyamori et al, 2001).

An important observation of this study is the ability of a broad spectrum MMP inhibitor to reduce the BBB damage by reducing degradation of the TJPs. A potential role for MMP inhibitors in stroke was recently demonstrated in studies that showed that prevention of the opening of the BBB by MMP inhibitors reduced the toxicity of tissue plasminogen activator in acute stroke (Lapchak et al, 2000; Pfefferkorn and Rosenberg, 2003; Lo et al, 2004). Our results provide a molecular basis for the action of MMP inhibitors.

In conclusion, our results provide direct evidence that changes in the TJPs, claudin-5, and occludin, are occurring through the action of MMP-2 in the early reversible disruption of the BBB, and that MMP-9 is important in the delayed damage. We showed that an MMP inhibitor could block the fragmentation of the TJPs during the early transient opening of the BBB. However, in the later stages, TJPs appear in the astrocytes around the blood vessels. Whether the astrocytes take up the degraded TJPs or synthesize them in response to the injury will require further study.

Footnotes

Acknowledgements

We thank Justin Tibbitts for technical support with the PCR studies.

References

Asahi

Wang

Mori

Sumii

Jung

Moskowitz

Fini

(2001) Effects of matrix metalloproteinase-9 gene knock-out on the proteolysis of blood—brain barrier and white matter components after cerebral ischemia. J Neurosci 21:7724–32

Ballabh

Braun

Nedergaard

(2004) The blood—brain barrier: an overview: structure, regulation, and clinical implications. Neurobiol Dis 16:1–13

Bazzoni

Dejana

(2004) Endothelial cell-to-cell junctions: molecular organization and role in vascular homeostasis. Physiol Rev 84:869–901

Chang

Hosomi

Lucero

Heo

Abumiya

Mazar

del Zoppo

(2003) Activation systems for latent matrix metalloproteinase-2 are upregulated immediately after focal cerebral ischemia. J Cereb Blood Flow Metab 23:1408–19

Duffy

John

Lee

Brosnan

Spray

(2000) Reciprocal regulation of the junctional proteins claudin-1 and connexin43 by interleukin-1beta in primary human fetal astrocytes. J Neurosci 20:RC114

Frederiks

Mook

(2004) Metabolic mapping of proteinase activity with emphasis on in situ zymography of gelatinases: review and protocols. J Histochem Cytochem 52:711–22

Fukuda

Fini

Mabuchi

Koziol

Eggleston

Jr del Zoppo

(2004) Focal cerebral ischemia induces active proteases that degrade microvascular matrix. Stroke 35:998–1004

Furuse

Fujita

Hiiragi

Fujimoto

Tsukita

(1998) Claudin-1 and −2: novel integral membrane proteins localizing at tight junctions with no sequence similarity to occludin. J Cell Biol 141:1539–50

Furuse

Hirase

Itoh

Nagafuchi

Yonemura

Tsukita

(1993) Occludin: a novel integral membrane protein localizing at tight junctions. J Cell Biol 123:1777–88

10.

Garcia

Darland

Massingham

D'Amore

(2004) Endothelial cell—astrocyte interactions and TGF beta are required for induction of blood—neural barrier properties. Brain Res Dev Brain Res 152:25–38

11.

Gasche

Fujimura

Morita-Fujimura

Copin

Kawase

Massengale

Chan

(1999) Early appearance of activated matrix metalloproteinase-9 after focal cerebral ischemia in mice: a possible role in blood—brain barrier dysfunction. J Cereb Blood Flow Metab 19:1020–8

12.

Giebel

Menicucci

McGuire

Das

(2005) Matrix metalloproteinases in early diabetic retinopathy and their role in alteration of the blood—retinal barrier. Lab Invest 85:597–607

13.

Hamann

Okada

Fitridge

del Zoppo

(1995) Microvascular basal lamina antigens disappear during cerebral ischemia and reperfusion. Stroke 26:2120–6

14.

Hawkins

Davis

(2005) The blood—brain barrier/neurovascular unit in health and disease. Pharmacol Rev 57:173–85

15.

Heo

Lucero

Abumiya

Koziol

Copeland

del Zoppo

(1999) Matrix metalloproteinases increase very early during experimental focal cerebral ischemia. J Cereb Blood Flow Metab 19:624–33

16.

Hirase

Kawashima

Wong

Ueyama

Rikitake

Tsukita

Yokoyama

Staddon

(2001) Regulation of tight junction permeability and occludin phosphorylation by Rhoa-p160ROCK-dependent and -independent mechanisms. J Biol Chem 276:10423–31

17.

Huber

Witt

Hom

Egleton

Mark

Davis

(2001) Inflammatory pain alters blood—brain barrier permeability and tight junctional protein expression. Am J Physiol Heart Circ Physiol 280:H1241–8

18.

Kuroiwa

Ting

Martinez

Klatzo

(1985) The biphasic opening of the blood—brain barrier to proteins following temporary middle cerebral artery occlusion. Acta Neuropathol 68:122–9

19.

Lapchak

Chapman

Zivin

(2000) Metalloproteinase inhibition reduces thrombolytic (tissue plasminogen activator)-induced hemorrhage after thromboembolic stroke (in process citation). Stroke 31:3034–40

20.

Liebner

Kniesel

Kalbacher

Wolburg

(2000) Correlation of tight junction morphology with the expression of tight junction proteins in blood—brain barrier endothelial cells. Eur J Cell Biol 79:707–17

21.

Lippoldt

Kniesel

Liebner

Kalbacher

Kirsch

Wolburg

Haller

(2000) Structural alterations of tight junctions are associated with loss of polarity in stroke-prone spontaneously hypertensive rat blood—brain barrier endothelial cells. Brain Res 885:251–61

22.

Broderick

Moskowitz

(2004) tPA and proteolysis in the neurovascular unit. Stroke 35:354–6

23.

Dalkara

Moskowitz

(2003) Mechanisms, challenges and opportunities in stroke. Nat Rev Neurosci 4:399–415

24.

Lohmann

Krischke

Wegener

Galla

(2004) Tyrosine phosphatase inhibition induces loss of blood—brain barrier integrity by matrix metalloproteinase-dependent and -independent pathways. Brain Res 995:184–96

25.

Mark

Davis

(2002) Cerebral microvascular changes in permeability and tight junctions induced by hypoxia—reoxygenation. Am J Physiol Heart Circ Physiol 282:H1485–94

26.

Miyamori

Takino

Kobayashi

Tokai

Itoh

Seiki

Sato

(2001) Claudin promotes activation of promatrix metalloproteinase-2 mediated by membrane-type matrix metalloproteinases. J Biol Chem 276:28204–11

27.

Morita

Furuse

Fujimoto

Tsukita

(1999a) Claudin multigene family encoding four-transmembrane domain protein components of tight junction strands. Proc Natl Acad Sci USA 96:511–6

28.

Morita

Sasaki

Furuse

Tsukita

(1999b) Endothelial claudin: claudin-5/TMVCF constitutes tight junction strands in endothelial cells. J Cell Biol 147:185–94

29.

Nitta

Hata

Gotoh

Seo

Sasaki

Hashimoto

Furuse

Tsukita

(2003) Size-selective loosening of the blood—brain barrier in claudin-5-deficient mice. J Cell Biol 161:653–60

30.

Papadopoulos

Saadoun

Binder

Manley

Krishna

Verkman

(2004) Molecular mechanisms of brain tumor edema. Neurosci 129:1011–20

31.

Pfefferkorn

Rosenberg

(2003) Closure of the blood—brain barrier by matrix metalloproteinase inhibition reduces rtPA-mediated mortality in cerebral ischemia with delayed reperfusion. Stroke 34:2025–30

32.

Pflugfelder

Farley

Luo

Chen

De Paiva

Olmos

Fini

(2005) Matrix metalloproteinase-9 knockout confers resistance to corneal epithelial barrier disruption in experimental dry eye. Am J Pathol 166:61–71

33.

Planas

Sole

Justicia

Farre

(2000) Estimation of gelatinase content in rat brain: effect of focal ischemia. Biochem Biophys Res Commun 278:803–7

34.

Rivera

Ogier

Jourquin

Timsit

Szklarczyk

Miller

Gearing

Kaczmarek

Khrestchatisky

(2002) Gelatinase B and TIMP-1 are regulated in a cell- and time-dependent manner in association with neuronal death and glial reactivity after global forebrain ischemia. Eur J Neurosci 15:19–32

35.

Romanic

White

Arleth

Ohlstein

Barone

(1998) Matrix metalloproteinase expression increases after cerebral focal ischemia in rats: inhibition of matrix metalloproteinase-9 reduces infarct size. Stroke 29:1020–30

36.

Rosenberg

Cunningham

Wallace

Alexander

Estrada

Grossetete

Razhagi

Miller

Gearing

(2001) Immunohistochemistry of matrix metalloproteinases in reperfusion injury to rat brain: activation of MMP-9 linked to stromelysin-1 and microglia in cell cultures. Brain Res 893:104–12

37.

Rosenberg

Estrada

Dencoff

(1998) Matrix metalloproteinases and TIMPs are associated with blood—brain barrier opening after reperfusion in rat brain. Stroke 29:2189–95

38.

Rosenberg

Navratil

Barone

Feuerstein

(1996) Proteolytic cascade enzymes increase in focal cerebral ischemia in rat. J Cereb Blood Flow Metab 16:360–6

39.

Sato

Kinoshita

Takino

Nakayama

Seiki

(1996) Activation of a recombinant membrane type 1-matrix metalloproteinase (MT1-MMP) by furin and its interaction with tissue inhibitor of metalloproteinases (TIMP)-2. FEBS Lett 393:101–4

40.

Strongin

Collier

Bannikov

Marmer

Grant

Goldberg

(1995) Mechanism of cell surface activation of 72-kDa type IV collagenase. Isolation of the activated form of the membrane metalloprotease. J Biol Chem 270:5331–8

41.

Tsukita

Furuse

(2000) Pores in the wall: claudins constitute tight junction strands containing aqueous pores. J Cell Biol 149:13–6

42.

Tsukita

Furuse

Itoh

(2001) Multifunctional strands in tight junctions. Nat Rev Mol Cell Biol 2:285–93

43.

Wachtel

Frei

Ehler

Fontana

Winterhalter

Gloor

(1999) Occludin proteolysis and increased permeability in endothelial cells through tyrosine phosphatase inhibition. J Cell Sci 112:4347–56

44.

Willis

Leach

Clarke

Nolan

Ray

(2004) Reversible disruption of tight junction complexes in the rat blood—brain barrier, following transitory focal astrocyte loss. Glia 48:1–13

45.

Witt

Mark

Hom

Davis

(2003) Effects of hypoxia—reoxygenation on rat blood—brain barrier permeability and tight junctional protein expression. Am J Physiol Heart Circ Physiol 285:H2820–31

46.

Yamagata

Tagami

Takenaga

Yamori

Itoh

(2004) Hypoxia-induced changes in tight junction permeability of brain capillary endothelial cells are associated with IL-1beta and nitric oxide. Neurobiol Dis 17:491–9