Results of recent studies reveal vascular and neuroprotective effects of matrix metalloproteinase-9 (MMP-9) inhibition and MMP-9 gene deletion in experimental stroke. However, the cellular source of MMP-9 produced in the ischemic brain and the mechanistic basis of MMP-9-mediated brain injury require elucidation. In the present study, we used MMP-9−/− mice and chimeric knockouts lacking either MMP-9 in leukocytes or in resident brain cells to test the hypothesis that MMP-9 released from leukocytes recruited to the brain during postischemic reperfusion contributes to this injury phenotype. We also tested the hypothesis that MMP-9 promotes leukocyte recruitment to the ischemic brain and thus is proinflammatory. The extent of blood-brain barrier (BBB) breakdown, the neurological deficit, and the volume of infarction resulting from transient focal stroke were abrogated to a similar extent in MMP-9−/− mice and in chimeras lacking leukocytic MMP-9 but not in chimeras with MMP-9-containing leukocytes. Zymography and Western blot analysis from these chimeras confirmed that the elevated MMP-9 expression in the brain at 24 h of reperfusion is derived largely from leukocytes. MMP-9−/− mice exhibited a reduction in leukocyte-endothelial adherence and a reduction in the number of neutrophils plugging capillaries and infiltrating the ischemic brain during reperfusion; microvessel immunopositivity for collagen IV was also preserved in these animals. These latter results document proinflammatory actions of MMP-9 in the ischemic brain. Overall, our findings implicate leukocytes, most likely neutrophils, as a key cellular source of MMP-9, which, in turn, promotes leukocyte recruitment, causes BBB breakdown secondary to microvascular basal lamina proteolysis, and ultimately contributes to neuronal injury after transient focal stroke.
- endothelial cells
- vascular permeability
the matrix metalloproteinase (MMP) family of zymogen proteases, normally involved in the degradation and remodeling of the extracellular matrix (55), may also participate in the etiology of several central nervous system pathologies (62). Expression of pro- and active MMP-9 is elevated within hours to days after focal stroke in humans (2, 8), nonhuman primates (21), rats (24, 45, 47), and mice (4, 15, 16, 23) and is temporally and spatially correlated with a loss of blood-brain barrier (BBB) integrity (4, 16, 45, 47). Moreover, postischemic BBB breakdown was reduced after MMP inhibition (13, 14) and MMP-9 gene deletion (3, 17). These same experimental interventions were also neuroprotective.
However, the cellular source(s) of MMP-9 responsible for vascular and parenchymal injury after focal stroke and the molecular targets of MMP-9-mediated injury require further elucidation. Immunohistochemical evidence suggests that postischemic MMP-9 expression increases in neurons, glia, and endothelial cells (4, 17, 33, 41, 45, 48) as well as in neutrophils recruited to the ischemic brain (2, 24, 33, 45, 48). Neutrophils contain abundant quantities of pro-MMP-9 in secretory granules (6, 22, 37), and the factors that promote the degranulation and activation of neutrophil MMP-9 from these granules are produced in the brain after ischemia (37, 38, 40). Indeed, inflammatory cells are already implicated in the MMP-9 production that contributes to tumor progression, aneurysm formation, and other pathologies (9, 43).
Thus, to directly test the hypothesis that neutrophil-derived MMP-9 contributes to injury after transient focal stroke, we examined the magnitude of BBB breakdown and infarction in MMP-9−/− mice and two unique groups of chimeric mice, one of which expressed MMP-9 only in circulating immune cells and the other that expressed MMP-9 in all cells except circulating immune cells. We also sought to examine MMP-9-mediated neutrophil recruitment, as indexed by leukocyte-endothelial adherence, neutrophil-capillary plugging, and neutrophil emigration into the ischemic brain in wild-type (WT) and MMP-9−/− mice to test the hypothesis that neutrophil-derived MMP-9 promotes this particular facet of postischemic inflammation. Overall, our results support a mechanism whereby, during reperfusion, MMP-9 released from degranulating neutrophils recruited to the postischemic brain promotes further neutrophil recruitment to these same sites in a positive feedback manner and causes BBB breakdown secondary to collagen IV degradation; as a result, neutrophil-derived MMP-9 contributes significantly to cerebral infarction.
MATERIALS AND METHODS
Experiments were approved by the respective Institutional Animal Care and Use Committees of Washington University and Stanford University. Studies were performed on MMP-9−/− mice and genetically matched 129SvEv WT controls weighing 25–35 g and ranging from 11 to 22 wk old. The MMP-9−/− mice, generated by targeted mutagenesis (58), were bred to 129SvEv mice (Taconic; Germantown, NY) and maintained on a 129SvEv background; 129SvEv mice originally from Taconic were interbred concurrently to serve as WT controls. The MMP-9 knockouts are indistinguishable from littermate WTs by gross appearance, weight, fertility, longevity, and organ histology, although differences in bone growth secondary to an impairment in angiogenesis within skeletal growth plates have been described (58). To generate chimeras, marrow from WT mice was transplanted into MMP-9−/− mice (this group designated hereafter as WT→ MMP-9−/−) or marrow from MMP-9−/− mice was transplanted into WT mice (this group designated hereafter as MMP-9−/−→WT). Four- to six-week-old syngenic donor animals were killed for marrow donation, and 5–8 million cells were injected intravenously into 9- to 10-wk-old recipients 6–7 h after the latter had been irradiated (850 Rads; a lethal dose without rescue). One mouse from each irradiation group was not transplanted to ensure the lethality of the irradiation; this sentinal animal died within 7–14 days. Seven weeks later, complete blood counts were obtained in every animal to verify hematopoietic cell reconstitution. One week later, the animals were subjected to transient focal stroke.
Mice were anesthetized by nose cone with 1.0–1.5% halothane in 70% N2O-30% O2. Rectal temperature was maintained at 37 ± 0.5°C via a thermoregulated system throughout ischemia and 30 min of reperfusion. Transient left middle cerebral artery (MCA) occlusion (MCAO) was performed as described previously (15). In brief, after a midline incision on the neck, the left common carotid artery was ligated, and a 6-0 monofilament nylon suture with a heat-blunted tip was advanced anterogradely from the external carotid artery stump into the internal carotid to lodge in the circle of Willis, blocking the ostium of the MCA. Animals were then withdrawn from anesthesia and placed in an incubator (34°C). After 2 h of ischemia, animals were briefly reanesthetized for removal of the intraluminal suture, reperfusion was confirmed by laser Doppler (see Cerebral blood flow and MCA vascular territory), and the mice were then recovered again in an incubator for 30 min. Additional subgroups (n = 4–6) of mice were instrumented for monitoring arterial blood gases and blood chemistries. Permanent MCAO was induced in two groups of MMP-9−/− and WT mice by two different methods. In one approach, the MCA was occluded by permanent placement of an intraluminal suture using the extracranial surgical approach described above for transient MCAO, and, in the second approach, the more distal MCA was accessed by a craniotomy and directly cauterized (32).
Cerebral blood flow and MCA vascular territory.
Transcranial measurements of cortical cerebral blood flow (CBF) were made in the MMP-9 knockout and corresponding WT mice and the chimeric animals during and after MCAO using a laser Doppler flow device (Transonic BLF21; Ithaca, NY). In brief, the skin overlying the calvarium was retracted, and relative changes in CBF were measured through the skull in the core region of the MCA territory (2 mm posterior to the bregma and 6 mm lateral to midline). In the animals subjected to transient MCAO, measurements were made before ischemia (after 15 min of surgical plane anesthesia), 5 min after the start of MCAO, at the end of the 2-h MCAO period, and at 10 min of reperfusion. The surgical procedure was considered successful if suture placement resulted in a sustained 80% or greater reduction in relative CBF from baseline and early reperfusion CBF was >70% of baseline. Animals not meeting both of these criteria were omitted from the study. In the animals subjected to permanent MCAO, CBF was measured before ischemia, 5 min after MCAO, and again 2 h after MCAO to confirm the permanency of the occlusion. Again, a decrement in MCA territory perfusion of >80% of baseline was the criterion used to gauge the effectiveness of the MCAO.
The area of the MCA vascular watershed was quantified in both WT and MMP-9 knockout mice to underscore the validity of infarct comparisons between the two genotypes secondary to documenting anatomic similarities in the cortical area supplied by the MCA. Nonischemic animals from each group were sequentially transcardiac perfused (at 5 ml/min) with 20 ml heparinized saline (with 0.1mM adenosine), 20 ml of 4% paraformaldehyde, 20 ml of 0.1 M PBS, and finally 20 ml of a 1% solution [in PBS containing 1% gelatin (300 bloom)] of 10-μm-diameter fluorescent polymer microspheres (Duke Scientific; Palo Alto, CA). The size of these microspheres precludes their entry into capillaries, resulting in their lodging within the arterioles and thereby labeling the arterial tree. The animals were placed on ice for 20 min, and the animal's head was then removed and immersion fixed in 10% buffered formalin for 1 day before brain removal and examination of the angioarchitecture by fluorescence microscopy. The cortical surface area of the MCA territory was quantified in the dorsal and lateral views using image analysis (ImagePro Plus, Media Cybernetics; Silver Spring, MD), wherein the two areas (dorsal and lateral) defined by the boundaries of the termination of the most distal, fluorescently labeled cortical arterioles and the main branches or trunk of the MCA were measured and summed to obtain a total cortical surface area for the MCA territory.
Serial neurobehavioral examinations were made in all animals subjected to transient left MCAO by a blinded observer. In animals subjected to transient MCAO or permanent MCAO by the intraluminal filament method, scores were obtained at 30 min of occlusion and at 1 and 24 h of reperfusion or continued ischemia, respectively. Each mouse was assigned a score of 0–4, wherein 0 is no observable neurological deficit, 1 is failure to extend the right forepaw, 2 is circling to the right, 3 is falling/rolling to the right, and 4 is inability to walk spontaneously and/or maintain upright posture. Animals from the aforementioned groups that exhibited a deficit score of 1.0 or less at 1 or 24 h of reperfusion or exhibited seizure activity were omitted from the study. Neither WT nor MMP-9 knockout animals subjected to permanent occlusion of the more distal segment of the MCA by cauterization exhibited neurological deficits.
Intravital epifluorescence microscopy.
Leukocyte dynamics in the cortical microcirculation of the MCA territory were observed by intravital epifluorescence videomicroscopy 24 h after a 1-h period of transient MCAO in subsets of WT and MMP-9−/− mice equipped with closed cranial windows (1). The number of leukocytes adherent to venular endothelium was quantified by off-line videotape analysis (1).
Tissue neutrophil content was quantified spectrophotometrically by measuring the activity of myeloperoxidase (MPO), an enzyme exclusive to neutrophils, as described previously (23, 61), with some modifications. In brief, after transcardiac perfusion, hemisphere samples were washed in aqueous buffer, suspended in 0.5% hexadecyltrimethylammonium bromide, and repeatedly frozen, thawed, and sonicated. After centrifugation, the supernatant fraction (100 μl) was combined with 50 μl of 0.5 mg/ml o-dianisidine and 50 μl of 0.001% of hydrogen peroxide, and a kinetic spectrophotometric assay was run at 450 nm in a 96-well plate. Optical density values were converted to numbers of neutrophils per hemisphere by normalization against a standard curve of MPO extracted from a purified preparation of peritoneal neutrophils obtained 5–6 h after 3% thioglycolate installation in WT mice.
Quantitative measurements of the loss of BBB integrity to albumin-bound Evans blue were obtained at 8 h of reperfusion after transient MCAO. In brief, at 15 min of reperfusion after MCAO, animals received an intravenous injection of Evans blue dye (4 ml/kg of a 2% filtered solution in saline). After 8 h of reperfusion, animals were transcardially perfused (5 ml/min) with 100 ml of heparinized (10 U/ml) 0.9% normal saline. Cerebral hemispheres were homogenized in 300 μl N,N-dimethylformamide, incubated for 18 h in a water bath at 55°C, and centrifuged. Supernatant Evans blue concentrations were determined by conventional spectrophotometric methods using standard curves derived from control brain homogenates (17).
At 24 h after transient or permanent MCAO, animals were anesthetized and transcardially perfused, and infarct areas and edema-corrected infarct volumes were quantified from 2,3-triphenyltetrazolium chloride-stained sections as described previously (32).
MMP purification for zymography and immunoblotting.
Brains were snap frozen at 8 or 24 h of reperfusion after transient MCAO. Neutrophils from the same animal groups were isolated by differential centrifugation of arterial blood samples. Pro- and active MMP-9 and MMP-2 expression in these extracts was determined by gelatin zymography, as described earlier (16, 26, 63), and quantified. Western blots were also performed on brains and isolated neutrophils as described previously (26) using a rabbit anti-rat MMP-9 polyclonal antibody (1/2,000, AB19016, Chemicon International) and a human pro-MMP-9 standard (CC079, Chemicon).
Brains from transcardially perfused mice were frozen, and 20-μm coronal sections were fixed, washed, blocked, and then incubated with an anti-collagen IV (1:50, pro10760; Research Diagnostics) rabbit polyclonal antibody; a swine anti-rabbit IgG FITC secondary antibody (1:20, Dako) was used. The number of fluorescent capillaries from matched brain regions was determined at ×10 magnification by a blinded observer on randomly chosen fields taken from ten 20-μm-thick sections obtained every 200 μm throughout the MCA territory.
For neutrophil immunostaining, coronal sections were blocked and incubated with a monoclonal rat anti-mouse neutrophil antibody (1:50, Cedarlane), followed by a biotinylated rabbit anti-rat IgG antibody (1:100, Vector Laboratories; Burlingame, CA; BA-4001). Chromophore was revealed by the ABC solution followed by a 5- to 10-min incubation with diaminobenzidine.
All values reported are expressed as means ± SE. ANOVA on ranks with the Dunn correction for multiple comparisons or Mann-Whitney nonparametric tests were used to compare differences between groups, with P < 0.05 considered significant.
Hemodynamics, hemocytometry, and angioarchitecture.
No physiologically relevant differences in resting arterial blood pressure or blood gases were noted between WT and MMP-9−/− mice or between the two chimeras (Table 1) before, during, or after ischemia. The extent of blood flow reduction upon suture placement and the degree of postischemic reperfusion were not different among the different animal groups (Table 2). Relative measures of cortical CBF obtained at 24 h of reperfusion by laser Doppler flowmetry at five different medial and five different lateral locations coincident with the location of the cranial window were ∼40% below those measured in the same respective locations in nonischemic shams; however, no differences in postischemic regional flows were found between animal groups (data not shown).
The mean differential blood counts obtained after 7 wk of reconstitution in WT, MMP-9−/−, and chimeric mouse groups were all within normal ranges for adult mice. Leukocyte counts (per mm3) ranged from 9.7 to 14.0 × 103 leukocytes/mm3, erythrocytes from 9.4 to 10.5 × 106 erythrocytes/mm3, and platelets from 635 to 817 × 103 platelets/mm3.
On gross inspection, the cerebrovascular architecture of the brains of WT and MMP-9-deficient mice appeared identical, with no visible differences at the level of the posterior communicating arteries, as noted by others (3, 4), or in terms of the regional coverage of the primary, secondary, and tertiary branches of the MCA. The area of the cortical territory supplied by the MCA did not differ significantly between WT and MMP-9 knockout mice when measured by filling the arterial system with fluorescent microspheres (n = 5; data not shown). Similarly, the total lengths of the primary, secondary, and tertiary branches of the MCA, stained by india ink, a reflection of microvascular density in the MCA territory, were also not significantly different between the WT and knockout groups (n = 6; data not shown).
MMP-9 and MMP-2 zymography.
Zymography was used to document the time course of postischemic MMP production, to validate MMP-9 deficiency in the knockouts, and to determine whether a compensatory increase in MMP-2 levels occurred in MMP-9−/− mice in response to ischemia-reperfusion. Zymographic analysis of the neutrophil subfraction of blood taken from the same animal groups was also undertaken for comparison with brain tissue to identify the relative contributions of resident brain cells and circulating/infiltrated neutrophils to the elevation in tissue MMP-9 expression after transient MCAO. Increases in pro-MMP-9 were found in ischemic hemispheres of WT mice at 8 and 24 h of reperfusion relative to the contralateral hemisphere (Fig. 1A). Moreover, proteins with molecular masses corresponding to the active enzyme [88 and 77 kDa (16)] were detected at both 8 and 24 h of reperfusion only in ipsilateral ischemic tissue (Fig. 1A). As expected, neither pro-MMP-9 nor active MMP-9 was noted on any of the zymograms from MMP-9−/− mice (Fig. 1B). Although there was an upregulation of pro-MMP-2 levels in response to ischemia-reperfusion in WT mice (Fig. 1A), there was no obvious change in pro-MMP-2 levels in MMP-9−/− mice relative to WT mice, and active MMP-2 could not be detected in any of these animals at these times (Figs. 1, A and B), consistent with the finding that MMP-2 mice subjected to transient focal stroke show no neuroprotective phenotype (5). As with normal WT mice, pro-MMP-9 was robustly elevated at 24 h of reperfusion only in the WT→MMP-9−/− chimeras (Fig. 1B); active MMP-9 was noted at similar levels in these two mouse groups. Conversely, only a very faint expression of pro-MMP-9 could be seen in the zymograms of brains from the MMP-9−/−→WT chimeras (Fig. 1B), indicating that the vast majority of brain MMP-9 expression at 24 h after transient focal stroke in normal mice is derived from leukocytes.
Neutrophil zymography confirmed the uniformity and success of the bone marrow transplants and the two unique phenotypes that resulted (Fig. 1C). Specifically, pro-MMP-9 was absent in the neutrophils from MMP-9−/−→WT chimeras. Conversely, pro-MMP-9 expression was evident at levels similar to WT mice in leukocytes from the WT→MMP-9−/− chimeras; active MMP-9 was not detectable in neutrophils. MMP-2 was also not detectable, confirming previous reports that neutrophils do not synthesize or release this protease (37, 39).
Western blots of brains and isolated neutrophils confirmed the zymography results, indicating that MMP-9 expression in the ischemic brain at 24 h is primarily derived from leukocytes (Fig. 2). In particular, pro-MMP-9 was detected in the ischemic hemispheres of WT animals and WT→MMP-9−/− chimeras, whereas MMP-9 expression was completely lacking in brains from MMP-9−/− animals and MMP-9−/−→WT chimeras. Similarly, when isolated neutrophils from circulating blood are blotted, WT mice and WT→MMP-9−/− chimeras exhibited pro-MMP-9, but in the MMP-9 knockouts and MMP-9−/−→WT chimeras, pro-MMP-9 was completely absent. Collectively, the zymographic and immunoblotting results indicate that pro-MMP-9 expression in brains subjected to transient focal stroke and reperfused for 24 h is largely derived from leukocytes.
Extravasation of albumin-bound Evans blue was used to determine the extent to which MMP-9 and, in turn, leukocyte-derived MMP-9 contributes to postischemic vasogenic edema after transient focal stroke. As shown in Fig. 3, vasogenic edema occurring over the initial 8 h of reperfusion was significantly reduced (75%) in MMP-9−/− mice relative to their respective WT controls. Vasogenic edema in the WT→MMP-9−/− chimeras was similar in magnitude to that measured in WT mice. However, MMP-9−/−→WT chimeras exhibited significant reductions (53%) in vasogenic edema, in parallel with, and not significantly different from, that measured in MMP-9−/− mice. These latter findings implicate leukocyte-derived MMP-9 in contributing to the early loss of BBB integrity after transient MCAO.
Collagen IV immunostaining.
Immunostaining for collagen IV, a primary component of the vascular basal lamina, was performed to document changes in basal lamina integrity (20) and to determine whether a reduction in the level of this MMP-9 substrate underlies the BBB breakdown observed after transient MCAO. As shown in Fig. 4, microvessel immunostaining in the ipsilateral ischemic cortex of WT mice was considerably reduced 24 h after transient MCAO relative to levels in the contralateral nonischemic hemisphere. In MMP-9−/− mice, however, the reduction in immunostaining was considerably less marked. Similar interhemispheric differences in cortical collagen IV immunostaining were observed at 8 h of reperfusion. Quantitative analysis of the number of immunofluorescent capillaries at 24 h of reperfusion confirmed an attenuated loss of collagen IV substrate (16 ± 9% loss) after ischemia in MMP-9−/− mice relative to WT (35 ± 12% loss).
Infarction and neurological deficits.
MMP-9 gene deletion afforded significant neuroprotection in the transient MCAO model (Fig. 5), with infarct volume significantly reduced (34%) in MMP-9−/− mice compared with WT. A similarly robust degree of protection (51%, not significantly different from that found between the WT and MMP-9−/− mice) was afforded in MMP-9−/−→WT chimeras relative to WT→MMP-9−/− chimeras, implicating a significant contribution by leukocyte-derived MMP-9 to the ultimate extent of infarction.
MMP-9−/− mice also exhibited a significantly lower neurological deficit score at 24 h of reperfusion (0.3 ± 0.3) relative to their WT counterparts (2.8 ± 0.3). In a parallel fashion, deficit scores in MMP-9−/−→WT chimeras were significantly reduced at 24 h (1.2 ± 0.1) relative to the WT→MMP-9−/− chimeras (2.0 ± 0.1) as well as at 1 h of reperfusion (1.6 ± 0.1 vs. 2.1 ± 0.1, respectively).
Because our results indicated that MMP-9-mediated vascular and parenchymal injury is leukocyte dependent, we reasoned that little or no protective phenotype would be observed in the setting of permanent focal stroke, when leukocytes do not reenter the ischemic zone upon reperfusion. Indeed, subjecting MMP-9 knockout mice to permanent MCAO, either by placement of an intraluminal suture or by direct coagulation of the MCA, resulted in no reduction in infarct size relative to identically treated WT mice (Fig. 6). In particular, permanent occlusion of the proximal MCA by the intraluminal suture method resulted in infarct volumes in MMP-9 knockouts (110 ± 7 mm3) that did not differ (P = 0.395) from those measured in the matched WT animals (101 ± 7 mm3). The neurological deficit scores exhibited by these animals during and after ischemia also did not differ from one another (data not shown). When permanent MCAO was induced by direct coagulation of the more distal portion of the MCA, there were also no significant differences (P = 0.454) in infarct volumes noted between MMP-9−/− mice (23 ± 2 mm3) and the corresponding WT mice (22 ± 2 mm3); animals with these smaller lesions do not exhibit neurological deficits.
Leukocyte-endothelial adherence and infiltrated neutrophils.
Using intravital microscopy, we next examined whether leukocyte-derived MMP-9 promoted proinflammatory leukocyte-endothelial adherence and leukocyte transmigration into the brain parenchyma after transient MCAO (Fig. 7, A–C). The number of leukocytes adherent to venular endothelium was significantly increased 24 h after transient MCAO in WT mice relative to nonischemic WT controls. In contrast, leukocyte-endothelial adherence in the microcirculation of MMP-9−/− mice subjected to transient MCAO was significantly attenuated, even though levels of regional postischemic blood flow did not differ between these groups.
The extent of neutrophil infiltration into the ischemic brain, as measured by MPO activity, was significantly elevated 24 h after transient MCAO in WT mice; significantly fewer neutrophils infiltrated the ischemic hemisphere during the initial 24 h of reperfusion in MMP-9−/− mice (Fig. 7D). Overall, these results were consonant with what was obvious by neutrophil immunohistochemistry. As shown in Fig. 8, a progressive increase in neutrophil-plugged capillaries and intraparenchymal neutrophils was evident only in the ipsilateral hemisphere of WT mice at 8 and 24 h of reperfusion. In contrast, in MMP-9−/− mice, the extent of capillary plugging and neutrophil infiltration was notably reduced at both time points; in fact, few intraparenchymal neutrophils could be found in brain sections from knockout mice. As expected, neutrophils were not observed within vessels or tissue of the corresponding contralateral brain regions. Collectively, these findings implicate MMP-9 in the promotion of leukocyte recruitment to the reperfused brain after transient focal ischemia.
With the use of MMP-9−/− mice and unique chimeric mice lacking MMP-9 in discrete cell populations, the present study provides causal evidence that, during the initial 8 h of reperfusion after focal stroke, leukocyte-derived MMP-9 is largely responsible for the early loss of BBB integrity. Our studies also support a proinflammatory role for MMP-9 in promoting cerebrovascular leukocyte-endothelial adherence and neutrophil transmigration into the brain parenchyma. Collectively, these effects indicate that MMP-9 released from degranulating neutrophils contributes significantly to the final lesion size after transient MCAO and provide a specific mechanistic basis whereby circulating neutrophils cause vascular and parenchymal injury in focal stroke.
To date, the cellular source(s) of MMP-9 produced in response to transient focal ischemia and the mechanisms responsible for MMP-9 activation have not been characterized. The “tertiary” granules of neutrophils, which contain abundant quantities of pro-MMP-9, are the first to be mobilized and the first to degranulate upon chemotactic stimulation of neutrophils (6, 27). In addition to endothelial adherence, degranulation stimuli include serine proteases, reactive oxygen species (38), endothelin-1 (13), and several inflammatory cytokines and chemokines (37), all of which are elaborated in the ischemic brain. pro-MMP-9 so released can then be activated extracellularly to 82- (30) and/or 67-kDa (51) forms. As with degranulation, cerebral ischemia-reperfusion leads to the production of a many molecules that can activate pro-MMP-9 (34), including oxidants (17), neutrophil elastase (44), cathepsin B, stromelysin-1, plasmin, tissue plasminogen activator (59, 60), and nitric oxide (18). In addition, neutrophils appear to “autoregulate” MMP-9 proteolysis by recruiting other proteases and oxidants (14, 40, 52). MMP-9 also cleaves IL-8, which triggers further release of MMP-9-containing granules (54). Moreover, feed-forward, mutually reinforcing disinhibition mechanisms may be operative (11, 36, 39). Finally, active MMP-9 can become associated with the neutrophil cell surface, where it resists tissue inhibitor of metalloproteinase-1 inhibition (39). Thus cerebral ischemia-reperfusion creates an environment rich in molecules that trigger neutrophil MMP-9 release and regulate its posttranslational activation.
Our zymography and Western blot results from the brains of WT mice are consistent with others (4, 24, 33, 41). While not revealing the cellular source of MMP-9, zymography studies of WT mice indicate that expression of pro-MMP-9 and active MMP-9 protein in the brain increases in a rapid and sustained manner after transient focal ischemia. However, the lack of postischemic cerebral MMP-9 expression in MMP-9−/−→WT chimeras relative to the elevated and nearly identical MMP-9 expression that we found in WT mice and WT→MMP-9−/− chimeras, indicates that endothelial-adherent or infiltrated leukocytes must be the cell primarily responsible for the change in brain tissue MMP-9 expression measured 24 h after transient MCAO. Consistent with this contention is the recent finding that the elevated expression of 95-kDa pro-MMP-9 in the rat brain after transient MCAO was completely prevented by neutropenia (24). Indeed, immunohistochemistry studies in several species have confirmed, by immunochemical and morphological criteria, MMP-9-positive neutrophils both within and at the periphery of the ischemic territory (2, 24, 33, 45, 48). The MMP-9-immunopositive neurons, glia, and endothelial cells also observed in the postischemic brain (4, 17, 33, 41, 45, 48) and, while not necessarily indicative of activated MMP-9, may reflect a delayed elaboration of MMP-9 in these cells by neutrophil-derived MMP-9 or other activation mechanisms; further studies are required to identify these potential intercellular activation events and elucidate their timecourse in more detail.
Postischemic vasogenic edema, as evidenced by elevated cerebrovascular permeability to Evans blue over the initial 8 h of reperfusion, is commonly observed in rodents after transient MCAO (15, 47) and coincides temporally with increases in pro- and active MMP-9 expression (4, 16, 33, 45, 47). That this early increase in permeability was significantly reduced in MMP-9−/− mice is consistent with that reported in the same mutants after 18–20 h of reperfusion (4). Unique to our study, however, was the finding that the extent of vasogenic edema in the chimeric mice was robustly dependent on the cells that were null for MMP-9, such that the chimeric mice lacking MMP-9 in leukocytes exhibited a reduction in Evans blue permeability that was significant relative to their chimeric counterparts with MMP-9-containing leukocytes and not statistically different from that measured in the mice with MMP-9 lacking from all cells. Conversely, in the chimeric mice with MMP-9-containing leukocytes and MMP-9 lacking from all other cell types, Evans blue extravasation was as robust as that measured in WT mice. These results strongly implicate leukocytes as a contributor to MMP-9-mediated postischemic edema early during reperfusion. Our results do not rule out the possibility that alterations in BBB integrity at later time points is mediated by MMP-9 produced by other cell types, nor does it rule out the contribution of MMP-9-independent mechanisms to BBB breakdown.
A time-dependent disruption in the molecular framework of the cerebrovascular basal lamina, which serves as the structural basis of the BBB (52), is thought to underlie the loss of microvascular integrity during postischemic reperfusion (19). A putative role for MMP-9 in the loss of BBB integrity after stroke has been advanced (46) because many of the molecular constituents comprising the cerebrovascular basal lamina and its tight junctions (52) are substrates for active MMP-9 (55). The abrogated loss of collagen IV-positive microvessels in MMP-9−/− mice documented in the present study and the reduced loss of the tight junction protein zona occludens-1 also shown in these mutants (4) begin to illuminate the molecular basis for the MMP-9-mediated BBB disruption after MCAO. Also consonant with these findings is the observation that neutrophil depletion resulted in both downregulation of MMP-9 expression and the prevention of BBB disruption triggered by intracerebral injections of IL-8 (56). Experimental studies implicate MMP-9 in the postischemic hemorrhagic transformation that can occur with ongoing or more severe vascular wall proteolysis (21, 28), although results of a recent study indicated that the cells expressing MMP-9 at different postischemic times varies regionally across the infarct territory and may depend on the severity of the ischemic insult (33). At any rate, the present results documenting the vulnerability of the cerebrovasculature to neutrophil-derived, MMP-9-mediated proteolysis in the initial 8 h after stroke may reflect a more widespread role for MMP-9 as a key mediator of neutrophil-associated BBB breakdown in other neuroinflammatory diseases (62).
Continual inactivation of plasma MMP-9 by α2-macroglobulin suggests that circulating neutrophils must become adherent to cerebrovascular endothelium for MMP-9 contained therein to cause vascular damage. Indeed, neutrophil-endothelial adherence triggers the mobilization and release of MMP-9-containing tertiary granules in vitro (27). In vivo, administration of an ICAM-1 antibody to prevent neutrophil-endothelial adherence in rats subjected to transient MCAO reduced cerebral MMP-9 expression (24). A pro-inflammatory, self-amplifying loop may be operative whereby MMP-9 promotes additional leukocyte-endothelial adherence secondary to releasing chemotactic peptides from matrix components and activating cell surface-bound, proinflammatory chemokines and cytokines (37, 55). Our intravital microscopy documentation that ischemia-induced increases in leukocyte-endothelial adherence and neutrophil capillary plugging were robustly attenuated in MMP-9−/− mice confirms for the first time a stimulatory, proadherent role for leukocyte-derived MMP-9 in the ischemic brain.
Whether leukocytes depend on MMP-mediated proteolysis to diapedese across the cerebrovascular endothelium and basal lamina and whether this transmigration event itself contributes to BBB disruption after transient focal stroke require additional investigation. A gelatinase-mediated penetration of matrix structures is a common mechanism of migration for many motile cells, including neutrophil transmigration by MMP-9 (10, 53); although several in vitro (10, 53) and in vivo (12, 25, 35) investigations provide convincing evidence to support the mediation of neutrophil transmigration by MMP-9, studies in the brain are still lacking. The significant reduction in infiltrated neutrophils in the ischemic hemisphere of MMP-9−/− mice documented herein is consistent with the notion that MMP-9 promotes neutrophil transmigration into the brain. Although neutrophils use elastase and other proteases to degrade matrix, results of our studies in chimeric mice suggest that neutrophil-derived MMP-9 also facilitates transmigration in the ischemic brain.
In the final analysis, the chimeric mice carrying leukocytes that lacked MMP-9 exhibited smaller infarcts and lower postischemic neurological deficit scores after transient MCAO, implicating leukocytes, most likely neutrophils, as the source of the injurious MMP-9. These findings would suggest that, under the condition of permanent focal stroke wherein neutrophils cannot access the ischemic zone, the contribution of neutrophil MMP-9 to vascular and parenchymal injury may be spatially limited to penumbral regions with significant perfusion. Indeed, after permanent MCAO in mice, MMP-9 expression was localized to the penumbra and was absent in the infarct core (16). The lack of a neuroprotective effect of MMP-9 gene deletion that we measured in two different models of permanent MCAO in the present study is consistent with the notion that neutrophils are the cellular source of MMP-9 in focal stroke. That another study (3) could demonstrate a small protective effect of MMP-9 gene deletion in permanent MCAO may be related to strain-dependent differences in vascular distribution and other genetic background-dependent variables (32, 49). At any rate, our results provide a specific mechanistic explanation for why antibodies directed against leukocyte or endothelial adhesion molecules (64) and leukocyte CD18 adhesion molecule deficiency (42) do not result in neuroprotection from permanent focal stroke, i.e., neutrophils, and thus neutrophil-derived MMP-9, cannot gain access to a significant portion of the area at risk of infarction. In addition, the extent of MMP-9 activation by oxidative stress (17) would be enhanced by reperfusion, in parallel with the finding that neuroprotection in CuZn-SOD transgenic mice was only evidenced after transient, but not permanent, MCAO (7).
A potential caveat of our study relates to the possibility that bone marrow-derived cells differentiated into other resident brain cells and contributed to the MMP-9-dependent pathology we observed in WT→MMP-9−/− chimeras. However, a study following a very similar transplantation and stroke protocol (50) documented that, 3 mo after WT mice were transplanted with marrow from green fluorescent protein (GFP) transgenic mice, only a few GFP-positive perivascular macrophages were identified in the resting brain, and all resident microglial cells were GFP negative. During the 24 h that followed transient focal stroke, only GFP-positive neutrophils could be seen in infarct areas; activated microglia remained GFP negative. Thus, during early reperfusion, neutrophils are likely to be the only marrow-derived cell type that infiltrates the ischemic brain and thus the only source of MMP-9 in our WT→MMP-9−/− chimeras.
Another caveat is that this study was performed in mice, and there is evidence from investigations of focal stroke in nonhuman primate studies that the time course of MMP elaboration and the relative contributions of different MMPs to the microvascular pathophysiology may show some species dependence (21). Also, more recent in vitro work indicates that MMPs other than MMP-9 may participate in a direct, or indirect, way in mediating neuronal and vascular injury after focal stroke (48). Thus further studies in cerebral ischemia-induced, MMP-mediated proteolysis is required to elucidate these possibilities.
In summary, experiments in MMP-9−/− mice and chimeric mice with either MMP-9-deficient or MMP-9-containing leukocytes indicate that reperfusion after focal stroke leads to the release and activation of MMP-9 from neutrophils that are recruited to the postischemic brain. This neutrophil-derived MMP-9 exhibits self-amplifying proinflammatory effects that trigger further neutrophil-endothelial adherence, neutrophil plugging of capillaries, and diapedesis into brain parenchyma. Moreover, BBB integrity is compromised secondary to enzymatic degradation of collagen IV and other substrates in the basal lamina. These events, in addition to the possibility of extracellular matrix (31) and myelin basic protein (4) degradation, the potentiation of other proteases (29), and direct neuronal injury (18), are important MMP-9-dependent determinants of the final lesion volume. Therefore, specifically inhibiting neutrophil degranulation or the activation of MMP-9 released by neutrophils may be viable therapeutic approaches to limit secondary vascular and parenchymal injury after transient focal stroke.
This study was supported by National Institute of Neurological Disorders and Stroke Javits Neuroscience Investigator Award NS-21045 (to T. S. Park), NS-14543 (to P. H. Chan), NS-25372 (to P. H. Chan), NS-36147 (to P. H. Chan), and NS-38653 (to P. H. Chan), the Spastic Paralysis Research Foundation of the Illinois-Eastern Iowa District of Kiwanis International (to T. S. Park), and Swiss National Science Foundation Grant 32-61995 (to Y. G. Gasche).
The authors thank Drs. J. Michael Shipley and Robert M. Senior for providing MMP-9−/− and wild-type breeders and Dr. Stuart S. Kaplan for helpful discussion.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 by the American Physiological Society