OUP user menu

Neuroprotection by inhibition of matrix metalloproteinases in a mouse model of intracerebral haemorrhage

Jian Wang, Stella E. Tsirka
DOI: http://dx.doi.org/10.1093/brain/awh489 1622-1633 First published online: 30 March 2005


Intracerebral haemorrhage (ICH) is an acute neurological disorder without effective treatment. Mechanisms of acute brain injury after ICH remain to be clarified. Although a few studies suggested a detrimental role for the gelatinase matrix metalloproteinase (MMP)-9 in ICH, the relationship between MMP-9 activity and acute brain injury after ICH is not determined. In this study, we first examined the expression of gelatinases in vivo using a collagenase-induced mouse model of ICH. Gel zymography revealed that MMP-9 was activated and upregulated after ICH. In situ zymography showed that gelatinase activity was mostly co-localized with neurons and endothelial cells of the blood vessel matrix. Inhibition with a broad-spectrum metalloproteinase inhibitor GM6001 (100 mg/kg) ameliorated dysregulated gelatinase activity, neutrophil infiltration, production of oxidative stress, brain oedema and degenerating neurons. Functional improvement and a decrease in injury volume were also observed. We provide evidence that MMP-9 may play a deleterious role in acute brain injury within the first 3 days after ICH. Blockade of MMP activity during this critical period may have efficacy as a therapeutic strategy for the treatment of acute brain injury after ICH.

  • intracerebral haemorrhage
  • stroke
  • matrix metalloproteinase
  • neutrophil
  • reactive oxygen species
  • mouse
  • BBB = blood–brain barrier
  • ICH = intracerebral haemorrhage
  • MMP = matrix metalloproteinase
  • ROS = reactive oxygen species


Stroke is the second most common cause of death in the world after heart disease and a leading cause of disability. Treatment for intracerebral haemorrhage (ICH) is primarily supportive and outcome remains poor. The pathogenesis of damage after ICH is still under investigation. Cell death occurs that has been attributed to inflammation (Castillo et al., 2002; J. Wang et al., 2003) and apoptosis (Matsushita et al., 2000; Qureshi et al., 2003). When ICH occurs, the blood–brain barrier (BBB) becomes disrupted. As a result, macrophages and leucocytes infiltrate the brain parenchyma and their presence has been proposed to constitute a primary mechanism of cell death. Activated microglia also contribute to the neuronal cell death after ICH (J. Wang et al., 2003; Wang and Tsirka, 2005b).

Matrix metalloproteinses (MMPs) are a family of zinc endopeptidases capable of degrading components of the extracellular matrix (Wang and Tsirka, 2005a). MMPs can cause an increase in capillary permeability and produce brain oedema that is secondary to ischaemic and haemorrhagic brain injury (Rosenberg and Navratil, 1997; Gasche et al., 1999). Elevated MMP expression is thought to contribute to tissue destruction in stroke and neuronal damage (Romanic et al., 1998; Yong et al., 2001; Lo et al., 2002; Rosenberg, 2002; Wang and Lo, 2003). Despite a growing body of evidence on the detrimental effects of MMPs in ischaemic stroke, data on their role in ICH are still limited.

Studies of MMPs in ICH were focused on MMP-9 because of its established link to disruption of the BBB, inflammation and tissue injury after ischaemia. In humans, an increased expression of MMP-9 (Abilleira et al., 2003; Alvarez-Sabin et al., 2004) has been reported. In a rat ICH model, endogenous production of MMP-9 was found in brain tissues (Rosenberg et al., 1994). Treatment with the MMP inhibitors BB-1101 or BB-94 reduced brain oedema and mortality following collagenase- or tissue plasminogen activator (tPA)-induced ICH (Rosenberg and Navratil, 1997; Lapchak et al., 2000; Pfefferkorn and Rosenberg, 2003). Recently, in a rat model of ICH, MMP-12 was found to be upregulated, and suppression of this protease with minocycline morphologically protects and improves functional recovery (Power et al., 2003). However, MMP-12 was markedly enhanced only at day 7 after the haemorrhage. Therefore, it is possible that MMP-12 may not play an important role in the acute brain injury after ICH, but rather participates in the delayed inflammatory events. We have reported that, in a mouse collagenase haemorrhage model, gelatinolytic activity is highly increased on day 1 after collagenase injection (J. Wang et al., 2003). However, the cellular sources of gelatinolytic activity are not clear. In the present study, we examined the temporal expression profiles of gelatinases (MMP-2 and MMP-9) in the collagenase-induced ICH mouse model and studied the relationship between MMP activity and acute brain injury after ICH. We observed that MMP-9 was activated and upregulated after ICH. Inhibition with a broad-spectrum metalloproteinase inhibitor (GM6001) ameliorated dysregulated gelatinase activity, neutrophil infiltration, production of oxidative stress and brain oedema, and decreased the number of degenerating neurons. Functional improvement and decrease in injury volume were also observed.

Material and methods

Animal procedures

C57BL/6 mice were cared for by the Department of Laboratory Animal Research at Stony Brook with access to food and water ad libitum. The experiments were performed in accordance with the National Institutes for Health guidelines for the care and use of laboratory animals and institutional guidelines established by the Institutional Animal Care and Use Committee (IACUC) at Stony Brook. All efforts were made to minimize the numbers of animals used and ensure minimal suffering. All studies described below were conducted in a blind manner.

ICH model

The procedure for inducing ICH in mice has been described previously (Clark et al., 1998; J. Wang et al., 2003). Briefly, male mice (n = 176) weighing 25–35 g were anaesthetized by intraperitoneal injection of avertin (0.5 mg per g of body weight) and injected with collagenase (0.1 U in 500 nl of saline) unilaterally into the caudate putamen using stereotactic coordinates: 0.5 mm posterior and 3.0 mm lateral of bregma, 4 mm in depth. Collagenase was delivered over 5 min. The needle stayed in place for an additional 5 min to prevent reflux. The overall mortality rate was <2%.

Treatment with matrix metalloproteinase inhibitor

GM6001 (Chemicon, Temecula, CA) was used as a broad-spectrum MMP inhibitor. Reported Ki values are as follows: 0.4 nM for MMP-1, 27 nM for MMP-3, 0.5 nM for MMP-2, 0.1 nM for MMP-8 and 0.2 nM for MMP-9. GM6001 was administered intraperitoneally as a suspension of 1, 10 or 100 mg/kg in 4% carboxymethylcellulose (CMC) in saline, as described (Solorzano et al., 1997) to a total volume of 200 µl. Control mice received 4% CMC in saline. Injection of GM6001 or CMC (control) was initiated 2 h after ICH, followed by additional treatment every 24 h for another 2 days. GM6001/CMC-treated mice were sacrificed up to 72 h after the first injection. Since MMP inhibition may have systemic effects, during the procedure and until the mice were ambulatory, their core temperature was maintained at 37°C with a heating pad. It has been reported that blood pressure in both MMP-9 knockout and wild-type mice was similar (∼80 mmHg) before ICH and remained similar during collagenase injection, suggesting that MMP inhibition may not directly affect blood pressure (Tang et al., 2004).

Preparation of tissue extracts

At 24, 48 or 72 h after ICH, mice were anaesthetized deeply and perfused transcardially with ice-cold phosphate-buffered saline (PBS), pH 7.4. Sham-operated control mice were perfused similarly at 24 h. The brains were removed and divided into ipsilateral haemorrhagic and contralateral non-haemorrhagic hemispheres. Brain tissue extracts were prepared as described (Justicia et al., 2000). Protein (15 mg) in 500 µl was incubated with 50 µl of gelatin–Sepharose 4B (Amersham Biosciences, Uppsala, Sweden) under constant shaking at 4°C for 1 h and centrifuged at 2500 r.p.m. for 2 min at 4°C. The pellet was washed three times with 500 µl of washing buffer (the same as the lysis buffer, with the exception of Triton X-100) and each time centrifuged at 2500 r.p.m. for 2 min at 4°C. The gelatinases were eluted with 150 µl of buffer [washing buffer containing 10% dimethylsulphoxide (DMSO)] by incubation at 4°C under constant shaking for 30 min followed by centrifugation.

Gelatin gel zymography

Prepared protein samples (300 µg) were loaded and separated on a 10% Tris-glycine gel with 0.1% gelatin as substrate. A mixture of human MMP-9 and MMP-2 (CC073, Chemicon, Temecula, CA) was used as the gelatinase standard. The upper band at ∼92 kDa is human pro-MMP-9. The size of the mouse pro-MMP-9 is ∼105 kDa; active-MMP-9 is ∼97 kDa (Lee et al., 2004). After separation, the gel was incubated in renaturing buffer (2.5% Triton-X-100) twice for 1 h at room temperature with gentle agitation and then incubated with developing buffer (50 mM Tris–HCl pH 7.5, 200 mM NaCl, 5 mM CaCl2, 0.05% Brij-35, 0.02% NaN3). Then the gel was stained with 0.5% Coomassie blue R-250 for 30 min and destained appropriately to be photographed and analysed densitometrically (ImageQuant 5.1, Molecular Dynamics, Sunnyvale, CA) on coded samples. To compare and normalize band intensities among gelatin zymographic gels, the ratios of specific band intensities between ipsilateral and contralateral sides were calculated.

In situ zymography and double labelling with fluorescent probes

In situ gelatinolytic activity was detected on frozen brain sections 10 µm thick using a commercial kit (EnzChek Gelatinase Assay kit; Molecular Probes, Eugene, OR). Fresh sections were incubated with DQ gelatin conjugate, a fluorogenic substrate, at 37°C for 1 h and washed and fixed in 4% paraformaldehyde in PBS. Cleavage of DQ gelatin by MMPs results in a green fluorescent product (excitation, 495 nm; emission, 515 nm). Some tissue sections were incubated with 1,10-phenanthroline (1 mM in DMSO; Molecular Probes), a non-specific inhibitor of MMP activity.

After the gelatinolytic activity was assessed, tissue sections were fixed and incubated with primary antibodies specific for neurons (NeuN, 1 : 250 dilution; Chemicon, Temecula, CA), astrocytes [glial fibrillary acidic protein (GFAP), 1 : 200; Zymed, San Francisco, CA], neutrophils [myeloperoxidase (MPO), 1 : 100; DAKO, UK], endothelial cells (CD31, 1 : 200; Chemicon) then followed by Cy3 (1 : 1000; Jackson ImmunoResearch Labs)-conjugated secondary antibodies. Microglia/macrophages were detected using tetramethylrhodamine isothiocyanate (TRITC)-conjugated isolectin B4 from Griffonia simplicifolia (IB4, Sigma, L5264; 0.05 mg/ml in PBS, pH 7.4) for 2 h at room temperature. Sections were examined with a Nikon PCM 2000 confocal microscope.

Gelatinolytic activity-positive cells were counted in three fields immediately adjacent to the haematoma over a microscopic field of 0.01 mm2 and expressed as cells/0.01 mm2 areas (large blood vessels were avoided). Positively labelled vessels from sham and the ICH sides were counted manually by judges blinded to the experimental protocol, in three fields immediately adjacent to the haematoma using a magnification of ×200 in three consecutive sections through the injection site.


Free-floating sections were washed in PBS for 20 min, blocked in 5% normal serum, and incubated with MPO or F4/80 antibodies followed by Cy3-conjugated secondary antibody. Stained sections were examined using a Nikon PCM 2000 confocal microscope; the images were captured and analysed by SPOT Advanced image software. Control sections were processed as above, except that primary antibodies were omitted. Control sections lacked specific staining. Infiltrating neutrophils and macrophages were counted in three different fields immediately adjacent to the haematoma in at least three sections per animal over a microscopic field of 0.01 mm2, and averaged and expressed as cells/0.01 mm2. Six mice/group were analysed by an observer blind to the experimental treatment.

In situ detection of O2 production

Production of reactive oxygen species (ROS) after ICH was investigated by in situ detection of oxidized hydroethidine (Wang and Tsirka, 2005b). Hydroethidine (a redox-sensitive probe) is oxidized by superoxide to ethidium (Bindokas et al., 1996), which intercalates within the DNA and the nucleus fluoresces bright red. Hydroethidine (in DMSO; Molecular Probes, Eugene, OR) was diluted to 1 mg/ml in PBS and sonicated. At selected time points after ICH, mice were injected intraperitoneally with 300 µl of hydroethidine. Brains were harvested 2 h later and sectioned at 30 µm. The brain sections were incubated with 2.5 × 10−3 mg/ml Hoechst 33258 (Molecular Probes) in PBS for 20 min in a dark chamber, rinsed with distilled water and coverslipped with Vectashield mounting medium (Vector Labs). Ethidium was visualized on a Nikon PCM 2000 confocal microscope (excitation, 510 nm; emission, 580 nm) and photographed using a digital camera system and double exposure to produce images of ethidium and Hoechst 33258. Fluorescence intensity and expression patterns of ethidium in the peri-ICH area were compared among groups. Ethidium, indicative of the presence of ROS, was quantified: cells with ethidium extending to the cytosol were counted under high magnification in three different sites randomly selected in at least three sections per animal and averaged in the entire field. The percentage of these cells in relation to the total cells stained with Hoechst nuclear staining was then analysed on coded samples. In situ zymography was also performed on some of these sections, as described above.

Haemorrhagic injury analysis

After neurological scoring, mice were killed, and their brains were removed, fixed and dehydrated in 4% paraformaldehyde and 20% sucrose in PBS. Injury volumes were digitally quantified, employing the SPOT Software v3.5.2, on 50 µm coronal sections using a previously reported method of Luxol fast blue/cresyl violet staining (J. Wang et al., 2003). Haemorrhagic injury areas were summed in 5–10 coronal slices at different levels. Volumes in mm3 were calculated by multiplying the 0.5 mm slice thickness by the measured areas. Eight or nine mice/group were analysed by an observer blinded to the experimental treatment.


Luxol fast blue/cresyl violet staining and the Fluoro-Jade B (FJB) staining were performed according to published protocols (Schmued and Hopkins, 2000). Cells permeable to FJB were marked for cell death. FJB can sensitively and selectively detect degenerating neurons. Degenerating neurons were counted by sampling an area 0.1 × 0.1 mm2 immediately adjacent to the haematoma in at least three fields using a magnification of ×400 in at least three sections per animal and expressed as cells/0.01 mm2; areas with large blood vessels were avoided. Eight mice/group were analysed by an observer blinded to the experimental treatment.

Neurological deficit

At 24, 48 or 72 h after collagenase injection, mice were tested and scored blindly for neurological deficits using a modified 28-point neurological scoring system (Clark et al., 1998; J. Wang et al., 2003). The tests included body symmetry, gait, climbing, circling behaviour, front limb symmetry and compulsory circling. Each point was graded from 0 to 4. The maximum deficit score is 24.


All data are presented as mean ± SD. Differences between groups were determined by Student's t test or by one-way analysis of variance (ANOVA) followed by Student's t test and the Bonferroni correction (for comparisons among multiple groups). The association between continuous variables was assessed by using the Pearson correlation coefficient. Statistical significance was set at P < 0.05.


Increased MMP activity after ICH

Collagenase induced ICH increased pro- and active-MMP-9 and pro-MMP-2 levels within ipsilateral brain homogenates (Fig. 1A). Low levels of active MMP-9 were observed in the contralateral or sham-operated brains (Fig. 1A). Pro- and active MMP-9 significantly increased at 24 h (>9-fold both) after collagenase injection, and increased further at 48 and 72 h (>22- and >12-fold, respectively) (Fig. 1C). In contrast, an increase in pro-MMP-2 was first observed at 72 h after collegenase injection (Fig. 1C, n = 8/group, P < 0.001). Very low levels of active MMP-2 were also observed after ICH, but did not change significantly (Fig. 1A and C, n = 8/group, all P < 0.05). These results suggest that MMP-9 plays a more dominant role in the acute injury of ICH.

Fig. 1

ICH increased pro-MMP-9, pro-MMP-2 and active MMP-9 levels. (A) Zymography gel showing elevation of pro- and active MMP-9, and pro-MMP-2 after ICH. Human MMP-9 and MMP-2 were loaded as standards (hST). (B) Silver staining shows protein loading. (C) Densitometric measurements, plotting the ratio of band intensities of the ipsilateral to the contralateral sides, confirm the increase in pro- and active-MMP-9 and pro-MMP-2 induced by ICH (n = 8/group; *P < 0.001). Ips = ipsilateral; Contr = contralateral. Marker: SeeBlue plus2 pre-stained protein standard.

Given that MMPs are involved in ischaemia (Rosenberg, 2002; Lo et al., 2003), we wanted to investigate whether administration of an MMP inhibitor systemically could influence ICH. First we examined whether the systemic MMP inhibitor could reduce the ICH-induced gelatinolytic activity. We observed a marked reduction in pro- and active MMP-9 and pro-MMP-2 levels in ipsilateral homogenates (Fig. 2A and C; n = 8/group, P < 0.01). Since MMP inhibitors should not have any direct effects on MMP expression or activation, we speculate that MMP-9 expression and activation may be mediated by other MMPs which may be the direct target of MMP inhibition.

Fig. 2

GM6001 decreased pro- and active-MMP-9 and pro-MMP-2 activity. GM6001 was administered intraperitoneally 2 h after ICH. (A) GM6001 at 100 mg/kg markedly reduced pro- and active-MMP-9 and pro-MMP-2 activities 72 h later. (B) Silver staining indicates protein loading. (C) Densitometric measurements, plotting the ratio of band intensities of the ipsilateral to the contralateral sides, confirm that GM6001 at 100 mg/kg can markedly reduce MMP activity (n = 8/group, *P < 0.01 versus vehicle).

To determine the cellular source of MMP-9 activity after ICH, we used in situ gelatin zymography on tissue sections (Fig. 3A–F). We focused on MMP-9, as this was the protease whose activity was upregulated quickly in ICH. The gelatinolytic activity increased in blood vessels prominently in the peri-ICH zone (Fig. 3C, D and I) (21.5 ± 4.9 versus 38.5 ± 5.3 positive blood vessels on the contralateral and ICH sides, respectively; n = 6/group, P < 0.001). At 72 h, gelatinolytic activity was increased further (Fig. 3E, F and I) (21.5 ± 4.9 versus 63.5 ± 8.6 positive blood vessels on the contralateral and ICH sides, respectively; n = 6/group, P < 0.001). Fluorescent product was also detected in cells mostly in the peri-ICH area at 24 and 72 h after collagenase injection (Fig. 3C–F). Limited gelatinolytic activity was detected in cells in the peri-injury area at 24 h in PBS-injected mice (Fig. 3A and B). Nearly all fluorogenic cleaving activity was blocked by pre-incubation of tissue sections with the inhibitor 1,10-phenanthroline for 15 min (data not shown), confirming the reaction specificity. After injection of GM6001 (100 mg/kg), gelatinolytic activity decreased in blood vessels (Fig. 3G–I) (positive vessels, 63.5 ± 8.6 versus 26.7 ± 5.4; n = 6/group, P < 0.001) and in cells (Fig. 3G, H and J) (positive cells, 25.5 ± 2.3 versus 6.2 ± 1.3; n = 6/group, P < 0.001) examined 72 h later.

Fig. 3

Increased in situ gelatinolytic activity after ICH. (A–H) Gelatinolytic activity (green) developed after incubation of sections (20 µm thick) with the substrate DQ gelatin. The images in B, D, F and H (scale bar: 20 µm) represent higher magnification of the boxed area in A, C, E and G (scale bar: 150 µm). Limited activity (single arrow) was detected at 24 h in PBS-injected mice (A). Gelatinolytic activity-positive cells (arrowheads) are present in the peri-injury area (B). Gelatinolytic activity-positive cell (scale bar, 10 µm). Robust activity was detected after ICH at 24 (C and D) and 72 h (E and F). D and F show activity in blood vessels (arrows) and cells (arrowheads). (D) Activity in a blood vessel cut in cross-section (scale bar, 10 µm). Decreased gelatinolytic activity was detected at 72 h in GM6001 (100 mg/kg)-treated mice (G and H). (I) Quantification of gelatinolytic activity-positive blood vessels after ICH. Values are the mean ± SD. Asterisks indicate a significant increase in activity-positive blood vessels compared with sham-operated mice (n = 6/group, *P < 0.001). GM6001 (100 mg/kg)-treated mice had fewer labelled vessels than vehicle-treated control mice (n = 6/group, *P < 0.001). (J) GM6001 (100 mg/kg)-treated mice had fewer labelled cells than control mice (n = 6/group, *P < 0.001). Values are means ± SD.

Gelatinolytic activity was mostly found in blood vessels and in cells in the peri-ICH area. We stained serial sections for gelatin-cleaving activity and expression of cell-specific antigens (Fig. 4, n = 5 each). Figure 4A shows that gelatinolytic activity was mostly associated with neurons. Only infrequent co-localization was found with astrocytes (Fig. 4B). Co-localization was not observed with IB4, a microglia/macrophage marker (Fig. 4E), or MPO, a neutrophil marker (Fig. 4C) These data are in agreement with previous reports in a model of transient focal cerebral ischaemia (Maier et al., 2004), which indicated that neutrophils are not the primary source of MMP-9 protein. Most gelatinolytic activity was detected in the walls of blood vessels and partially co-localized with endothelial cells (Fig. 4D).

Fig. 4

Increased gelatinolytic activity within the vascular matrix, neurons and endothelial cells after ICH. (A–E) Double fluorescence of gelatin-cleaving activity (green) and cells identified with specific markers: neurons, NeuN; astrocytes, GFAP; neutrophils, myeloperoxidase (MPO); endothelial cells, CD31; microglia/macrophages, IB4. Cell markers were visualized (red) by Cy3-conjugated secondary antibodies or by TRITC-conjugated IB4. Areas from merged confocal images (merged) are shown at higher magnification. Sections were obtained 24 or 72 h after ICH, as in the experiment described in Fig. 3. Arrows point to areas showing co-localization. Scale bar: 30 µm.

Effect of GM6001 on leucocyte infiltration

Acute inflammation is a normal response to injury. Infiltrating neutrophils can be observed in and around the haematoma as early as 4 h after ICH (data not shown). Neutrophils (MPO+) and mature macrophages (F4/80+, round cells with a smoother surface without processes) were present on days 1 (data not shown) and 3 (Fig. 5A and C) after ICH. In the GM6001 (100 mg/kg)-treated mice, fewer neutrophils were observed (763 ± 65.6 versus 394 ± 103/mm2, n = 6/group, P < 0.001), compared with control mice (Fig. 5B, D and E). GM6001 decreased, but did not significantly affect the number of macrophages present in the brain (337 ± 124 versus 176 ± 128/mm2, n = 6/group, P > 0.05).

Fig. 5

GM6001 decreased leucocyte recruitment after ICH. Infiltrated neutrophils (arrowheads, MPO+, A) and mature macrophages (arrowheads, F4/80+, C) are present near the haematoma on day 3. After GM6001 treatment, neutrophil and macrophage infiltration was inhibited on day 3 (B and D). The numbers of neutrophils and macrophages were quantified in the peri-ICH region on day 3 after GM6001 treatment (E). GM6001 (100 mg/kg)-treated mice had fewer infiltrated neutrophils than control mice (n = 6/group, *P < 0.001). GM6001 did not significantly affect the number of infiltrating macrophages following ICH, but the mean number of macrophages was reduced (n = 6/group, P > 0.05). Values are means ± SD. Scale bar, 20 µm.

Effect of GM6001 on ROS production

The presence of ROS was visualized as ethidium+ particles in the nucleus and cytosol. Our previous results (Wang and Tsirka, 2005b) showed that on day 1 after collagenase injection, the peri-ICH region showed significantly increased ethidium compared with the contralateral side. Here we show that 3 days after collagenase injection, the increased ethidium persisted in the peri-ICH region (Fig. 6B). In the contralateral (uninjected) side, there were only a few ethidium+ cells, and the signal was confined to the perinuclear area (Fig. 6A). In the GM6001-treated mice, decreased numbers of ethidium signals were detected in the cytosol (Fig. 6C). Quantification showed that GM6001 treatment significantly attenuated ethidium+ cells (vehicle treated, 62.2 ± 7.7; GM6001 treated, 31.6 ± 7.5; mean ± SD of percentage of ethidium+ cells, P < 0.001, n = 6/group).

Fig. 6

GM6001 decreased the production of ROS after ICH. (A) A few ethidium signals (red) were detected on the uninjected side and were confined to the perinuclear area (arrowheads) on day 3 after ICH. (B) The peri-ICH region (injected side) showed significantly increased ethidium signals (arrowheads). (C) In the GM6001 (100 mg/kg)-treated mice, fewer ethidium signals in the cytosol were detected (arrowheads). (D) Increased ethidium signals were detected in gelatinolytic activity-positive cells (green). A higher magnification picture is provided in E (arrowheads, perinuclear location of ethidium signials). (F) Increased ethidium signals (arrowheads) were also observed in the endothelial cells of blood vessels, which were gelatinolytic activity positive (green). Scale bar, 20 µm.

One day after ICH, ethidium signals extended into the cytosol and gelatinolytic activity was detectable in the nucleus (Fig. 6D and E, arrows). Some cells within the microvasculature, with morphological characteristics of endothelial cells, displayed strong ethidium+ signals. Gelatinolytic activity was visualized along the wall of the blood vessel (Fig. 6F).

Effect of GM6001 on stroke volume, oedema and neuronal cell death

Since delivery of GM6001 post-injury inhibited the production of ROS, we assessed its effect on injury volume, oedema and the ensuing neuronal death. GM6001 (100 mg/kg) significantly reduced brain injury after ICH on day 3 (19.9 ± 7.3 versus 10.7 ± 5.0 mm3, n = 7–8/group, P < 0.05). Lower concentrations of GM6001 were less protective (n = 8–9/group, all P > 0.05) (Fig. 7B).

Fig. 7

GM6001 reduced injury volume after ICH in mice. (A) Coronal sections were collected 3 days after ICH and stained for Luxol fast blue/Cresyl violet. Left, control mice (left); right, GM6001 (100 mg/kg)-treated mice. (B) GM6001-treated mice had smaller injury volumes than control mice (n = 7–9/group, *P < 0.05). Values are means ± SD. Scale bar, 1 mm.

Brain oedema is an important clinical complication of ICH. Compared with vehicle-treated controls, GM6001 (100 mg/kg) treatment significantly reduced brain oedema in the lesioned hemisphere 3 days after ICH (from 80.63 ± 1.1 to 78.87 ± 0.8%, P = 0.01, n = 6/group, Fig. 8A). These results indicate that MMP-9 contributes to brain injury and oedema formation after ICH.

Fig. 8

GM6001 reduced brain oedema and the number of degenerating neurons after ICH. (A) Brain oedema was expressed as percentage water content (wet − dry/wet) of brain weights. GM6001 treatment reduced brain oedema in the lesioned hemisphere 3 days after ICH, compared with control mice (n = 6/group, *P = 0.01). Values are means ± SD. (B) Fluoro-Jade B histological staining of neurons 3 days after ICH. Evenly distributed, intensely labelled neurons and processes are observed in the peri-ICH region. Scale bar, 50 µm. (C) Quantification of the numbers of degenerating neurons in control and GM6001-treated mice (n = 8/group, *P < 0.001). Values are means ± SD.

GM6001 (100 mg/kg) post-treatment reduced the number of degenerating neurons (evaluated by FJB staining) after ICH on day 3 (from 409.4 ± 61.1 to 216.5 ± 56.5/mm2, n = 8/group, P < 0.001), whereas 1 and 10 mg/kg of body weight dose regimens were less protective (all P > 0.05, n = 8/group) (Fig. 8B).

Effect of GM6001 on ICH-induced neurobehavioural deficits

ICH is usually accompanied by characteristic behavioural deficits. Repeated assessments of the animals were performed on days 1–3 after ICH. GM6001 (100 mg/kg) treatment significantly improved the neurobehavioural score of the animals compared with vehicle-treated control animals, with a trend beginning at day 1, and reaching statistical significance by days 2 and 3 after ICH; the score changed from 8.3 ± 1.1 (n = 18) to 6.3 ± 0.9 (n = 17) on day 2 (P < 0.001) and from 6.6 ± 1.1 (n = 17) to 4.6 ± 0.9 (n = 17) on day 3 (P < 0.001) (Fig. 9).

Fig. 9

GM6001 improved neurological functional outcome after ICH in mice. GM6001 (100 mg/kg) improved neurobehavioural performance after ICH with a trend beginning at day 1 but reaching statistical significance at day 2 and day 3 after ICH (n = 17 for GM6001 group; n = 18 for vehicle group; *P < 0.001 versus vehicle).


MMPs have been implicated in brain injury after ICH (Rosenberg et al., 1994; Rosenberg and Navratil, 1997; Power et al., 2003; J. Wang et al., 2003). MMPs, especially MMP-2 and MMP-9, become upregulated and degrade the neurovascular matrix leading to oedema, haemorrhage and cell death. In mouse systems, a dominant role has been ascribed to MMP-9 after ischaemia. We demonstrated here that MMP-9 may also play a key role in ICH brain injury. However, since our work focuses on the gelatinases, it does not take into account other members of the MMPs (e.g. MMP-3 that could lie upstream of the MMP-9 function), ADAM (a disintegin and metalloproteinase) and ADAMTS (a disintegrin and metalloproteinase thrombospondin) families, that may play important roles in mediating ICH, as they are also upregulated during neuroinflammatory events (Mun-Bryce et al., 2002; Rosenberg, 2002).

The rapid onset of gelatinolytic activity after ICH suggests that blood vessels constitutively express MMPs, and release them from the endothelium as a consequence of ICH (Hanemaaijer et al., 1993; X. Wang et al., 2003). Constitutive MMP-9 was reportedly expressed in hippocampal blood vessels (Vecil et al., 2000). Endothelial cells synthesize and release MMP-9, particularly when exposed to such stimuli as mechanical trauma and interleukin-1α (Vecil et al., 2000; Wang et al., 2002). Expression of MMP-9 by blood vessel matrix and endothelial cells at the periphery of the haematoma appears to be a mechanism by which the cerebral vascular walls become compromised, leading to oedema and leucocyte infiltration. MMPs are secreted in the basal surface of endothelial cells, thereby facilitating the degradation of the basement membrane (Unemori et al., 1990). Proteolysis of extracellular matrix components of the basement membrane is associated with an increase in vascular permeability and loss of vascular integrity (Hamann et al., 1995), which renders the blood vessels permissive to neutrophil extravasation. As neutrophils migrate into the brain tissue, they employ MMP-9 for invasion (Weiss and Peppin, 1986). We observed reduced numbers of neutrophils in and around the haematoma after ICH in GM6001-treated mice, a finding consistent with the role of MMP-9 in the transmigration of neutrophils from the vasculature (Delclaux et al., 1996; D'Haese et al., 2000; Keck et al., 2002). Our findings implicate neutrophils in acute brain injury after ICH. We demonstrated decreased infiltration of neutrophils correlated with improved neurological function in GM6001-treated mice over the first 3 days after ICH, a period coinciding with maximal infiltration of neutrophils into the lesion (Del Bigio et al., 1996).

Neutrophils damage tissue by generating ROS and secreting proteases, including MMPs (Weiss, 1989). MMP-9 has broad substrate specificity that includes extracellular matrix proteins and non-matrix proteins such as α1-antiproteinase, a potent inhibitor of neutrophil elastase (Liu et al., 2000). MMP-9 promotes tissue damage either directly by disrupting the extracellular matrix or indirectly by inactivating proteins, such as α1-antiproteinase (Sires et al., 1994; Liu et al., 2000). Because these α1-antiproteinase–elastase complexes are chemotactic for neutrophils, inhibiting MMP-9 may also diminish additional recruitment of neutrophils.

ROS can be released by injured tissues to trigger activation of the immune response. Besides the vascular endothelium and activated microglia/macrophages, neutrophils are also an important source of ROS (Weiss, 1989; Facchinetti et al., 1998). Treatment with free radical inhibitors improves outcomes in collagenase-induced ICH in rats (Peeling et al., 1998), suggesting that ROS are important mediators in ICH. GM6001 treatment resulted in significant reduction in the production of ROS.

MMP expression can be modulated by various mechanisms. ROS, nitric oxide (NO) and proteases have been implicated in MMP activation. During ICH, proinflammatory cytokines and NO increase within brain tissue (Mayne et al., 2001; Jung et al., 2004). Proteases, e.g. stromelysin-1 and tPA, have been associated with MMP-9 expression (Rosenberg et al., 2001; X. Wang et al., 2003). Available data indicate that ROS regulate MMP activity in vitro (Rajagopalan et al., 1996) and in vivo (Gu et al., 2002). Ethidium signals, which indicate ROS production, are localized with MMP activity in neurons and endothelial cells 1 day after ICH. The production of ROS is reduced after GM6001 treatment, along with vascular damage, suggesting that ROS may activate MMP-9 to mediate acute brain injury after ICH.

A critical finding of our study is that the protection afforded by acute intervention with an MMP inhibitor targets early vascular responses associated with inflammation. The abnormal vascular permeability exposes the brain to the toxic effects of inflammatory cells, and to amino acids such as glutamate and glycine, which, when present at high concentrations, can be toxic (Castillo et al., 2002). Similar neuroprotection has been reported after cerebral infarction and spinal cord injury, produced by systemic blockade of MMP-9 with a neutralizing antibody (Romanic et al., 1998). In this study, administration of the MMP inhibitor was initiated at 2 h after ICH and was maintained for 3 days after ICH. The limited duration of treatment, from 2 h to 3 days, more precisely defined the contribution of MMPs to early pathogenesis and the extent to which this acute inhibition would influence neutrophil infiltration and neurological function. Thus, neuroprotection and the resulting improvement in neurological function may be caused by decreased acute secondary damage. The present data support a deleterious role for MMP-9 in acute brain injury after ICH. It was shown recently that when older MMP-9 knockout mice (20–35 weeks of age) were subjected to ICH, increased mortality, haemorrhage and brain oedema were observed compared with those in their wild-type counterparts (Tang et al., 2004), indicating that MMP-9 has a crucial role in the ICH outcome. The authors discussed, however, that their results contradict some of the previous literature on MMP-9 and suggest that deficiency or long-term inhibition of MMP-9 activity may be compensated by upregulation of other MMPs (MMP-2 and MMP-3) (Tang et al., 2004). MMPs are known to play important roles in angiogenesis, remodelling, cell migration and phagocytosis (Szklarczyk et al., 2002; Parks et al., 2004). It is possible that MMPs may play beneficial roles during the chronic recovery from haemorrhagic damage. Therefore, longer term use of MMP inhibitors may be unwarranted.


We wish to thank members of the Tsirka lab for critical reading of the manuscript. This work was supported by grants from the National Institutes of Health (RO1NS042168, S.E.T.), an American Heart Association-Established Investigator Award (0540107N, S.E.T.) and a fellowship from the American Heart Association (0225701T, J.W.).


View Abstract