Neurovascular Matrix Metalloproteinases and the Blood-Brain Barrier

Abstract

Blood-brain barrier (BBB) leakage and brain edema is a critical part of stroke pathophysiology. In this mini-review, we briefly survey the potential role of matrix metalloproteinases (MMPs) in BBB dysfunction. A large body of data in both experimental models as well as clinical patient populations suggests that MMPs may disrupt BBB permeability and interfere with cell-cell signaling in the neurovascular unit. Hence, ongoing efforts are underway to validate MMPs as potential biomarkers in stroke as well as pursue MMP blockers as therapeutic opportunities. Because BBB perturbations may also occur in neurodegeneration, MMPs and associated neurovascular unit mechanisms may also be potential targets in a broader range of CNS disorders.

THE BLOOD-BRAIN BARRIER AND EDEMA

The blood-brain barrier (BBB) comprises a critical interface between the central nervous system (CNS) and the rest of the body. Its primary function is to help regulate the microenvironment of the CNS. To do so, the BBB functions both as a true barrier as well as a dynamically controlled influx/efflux transport system for handling a wide range of metabolic and signaling factors.

Historically, the BBB was thought to solely mediated by tight junctions between cerebral endothelial cells [1, 2]. However, recent data now suggest that the system is much more complicated. Astrocytes and pericytes are now known to be vital contributors in barrier function as well [3, 4]. Additionally, the BBB is now also considered as part of the “neurovascular unit” – a concept that emphasizes the importance of cell-cell signaling between all cell types residing in neuronal, glial and vascular compartments [5–7]. Hence, BBB leakage in CNS disorders can be thought of as merely one manifestation of neurovascular unit dysfunction [8, 9].

In the context of brain injury and disease, BBB leakage and neurovascular unit dysfunction are most clearly observed clinically as brain edema. Classically, brain edema has been defined as either cytotoxic (swelling and intracellular accumulation of water in astrocytes and neurons) or vasogenic (accumulation of water in extracellular space) in origin [10, 11]. In acutely ill patients, brain edema most commonly leads to central and systemic complications due to elevations in intracranial pressure, regardless of whether the edema is “cytotoxic” or “vasogenic”. Indeed, it is now accepted that these theoretical concepts underlie a more complicated reality where BBB dysfunction and leakage comprise a mix of extracellular and intracellular responses in multiple cell types.

The spatial and temporal evolution of BBB leakage and edema is also complicated, depending on the type of injury or disease. There are many mechanisms at play, including loosening of tight junctions, alterations in transporters, alterations in pinocytosis, degradation of matrix lamina etc. In acute stroke, biphasic patterns of BBB permeability have been described with openings and closings taking place during ischemia-reperfusion [12]. The BBB is not an inert static barrier, intact or breached. But rather, BBB permeability and transport are dynamically regulated – with various degrees of opening and closing depending on the nature of molecular factors or size of tracers involved. Ultimately, the BBB can perhaps be thought of as an interface representing cell-cell signaling at a locus where all components of the neurovascular unit come together.

This mini-review briefly outlines how matrix metalloproteinases (MMPs) may contribute to neurovascular injury after stroke, with a focus on how these potential drug targets may correlate with these known basic mechanisms. Due to the limited scope of this mini-review, the reader is referred to many excellent reviews on MMPs, the BBB and brain edema [13–17], as well as other papers in this special BBB issue of Current Pharmaceutical Design for more in-depth dissections of these important topics.

MATRIX METALLOPROTEINASES AND THE NEUROVASCULAR UNIT

If one accepts the premise that BBB function is part of and requires cell-cell signaling within the neurovascular unit, then the matrix that connects all the components together should be a critical substrate for both function and dysfunction. In this regard, proteases that disrupt neurovascular matrix integrity may comprise a set of mediators and targets for potential drug design.

When stroke or brain injury occurs, MMPs become dysregulated and may be a central cause of tissue damage. MMPs comprise a large family of extracellular zinc endopeptidases [14]. But in the context of stroke, the largest amount of data may exist for the gelatinases MMP-2 and MMP-9.

In animal models of focal and global cerebral ischemia, MMPs are upregulated, and treatment with MMP inhibitors prevent neuronal cell death, decrease infarction and improve outcomes [18–20]. Knockout mice that lack MMP-9 show significantly reduced brain cell death after cerebral ischemia or traumatic brain injury [21–24]. Conversely, transgenic mice that overexpress tissue inhibitors of metalloproteinase (TIMP) have better outcomes [25].

Mechanistically, the data in animal models fit well with the premise that high levels of MMPs can damage neurovascular matrix and cause BBB injury, edema and hemorrhage [13]. Degradation of various basal lamina and tight junction proteins has been correlated with BBB leakage and blockade of MMPs reduce edema [21, 23]. Matrix proteolysis and BBB disruption was reduced in knockout mice lacking MMP-9 [23]. MMP activation and BBB leakage also appears to coincide with the generation of free radicals. And as neurovascular injury continues to evolve, recruitment of cytokines and vascular adhesion molecules add onto the accumulating tissue damage and may even further amplify MMPs and inflammation [26].

Beyond vascular leakage per se, MMP-mediated proteolysis of neurovascular matrix may also interfere with homeostatic signals between different cell types in the neurovascular unit. Resting matrix signaling via integrins is vital for normal cell function. Disruption of extracellular matrix by MMPs can induce anoikis in neurons and cerebral endothelial cells [27, 28] In animal models, degradation of matrix correlates with cell death [29]. In a nonhuman primate model of focal cerebral ischemia, areas where matrix antigens are lost correspond to growing regions of collapsing penumbra and dying cores [30]. The importance of these matrix signals is further confirmed in fibronectin knockout mice in which neuronal apoptosis and brain damage are amplified after cerebral ischemia [31].

From a molecular perspective, matrix coupling may also help sustain trophic coupling between cells of the neurovascular unit. For example, cerebral endothelial cells may be a rich source of trophic factors such as FGF and BDNF, and this type of vascular neuroprotection is a critical defense against mutiple insults such as hypoxia, oxidative stress and perhaps even amyloid-beta [32, 33]. In white matter, an analogous oligovascular unit may exist. Cerebral endothelial cells and astrocytes also produce may trophic factors that sustain and protect oligodendrocyte precursor cells against injury [34, 35]. By interfering with these vital interactions between multiple cell types, MMP-mediated disruption of matrix-trophic coupling in the gray and white matter may significantly contribute to stroke and trauma pathophysiology.

BIPHASIC PROPERTIES OF MATRIX METALLOPROTEINASES AND BRAIN PLASTICITY

MMPs play a deleterious role during acute stroke by augmenting BBB disruption, edema, hemorrhage and brain injury. However, emerging data now suggest that MMPs may play biphasic roles. During the acute stages of stroke, MMPs are deleterious. But during delayed phases of stroke recovery, MMPs may play surprisingly beneficial roles [36, 37].

In part, this duality of MMP phenotype may be related to its original physiologic roles in normal development of brain morphology [38]. In developing brain, these proteases modify extracellular matrix to allow newborn cells to migrate and neurites and axons to extend and connect. Additionally, MMPs may also facilitate the actions of other signaling molecules. For example, MMP-9 may be an “angiogenic switch” by processing and releasing bioactive VEGF to promote vascular growth and/or remodeling [39]. MMP-9 has also been implicated in associative learning in the hippocampus. The broad spectrum MMP inhibitor FN-439 interferes with long term potentiation [40]. MMP-9 knockout mice display deficits in learning and memory [41].

During stroke recovery, the brain attempts to remodel. MMPs may be recruited as part of this endogenous recovery process. So blocking MMPs at the wrong place or wrong time may worsen outcomes. Following focal cerebral ischemia in mice, endogenous neurogenesis is amplified in the subventrcular zone and newborn neuroblasts are diverted form their original rostral migratory stream towards damaged brain [42]. This process requires MMPs, and delayed blockade of MMPs disrupts neuroblast migration [43]. At 2 weeks after focal strokes in rats, a secondary upregulation of MMPs in peri-infarct cortex can be detected in astrocytes and endothelial cells [44]. Late blockade of MMPs in this model system is damaging since MMPs appear to mediate VEGF processing, compensatory angiogenesis and stroke recovery [44]. Hence MMP inhibitors can sometimes lead to beneficial reductions of acute edema, while resulting in impaired long term recovery [45]. Similar biphasic properties of MMPs may also exist in spinal cord injury, where MMP-2 is increased together with reactive gliosis [46]. But genetic deletion of MMP-2 exacerbated white matter damage and decreased motor recovery [47]. Of course, not all MMP-mediated plasticity is guaranteed to be beneficial. How MMPs augment normal or abnormal rewiring in recovering brains after stroke or trauma remains to be fully elucidated.

MATRIX METALLOPROTEINASES IN CLINICAL STROKE

Because MMPs can degrade BBB integrity and function in the neurovascular unit, they have been proposed as potential biomarkers in clinical stroke and brain injury. Plasma levels of MMP-9 are elevated during acute stages of both ischemic and hemorrhagic stroke, and appear to be correlated with poor neurological outcomes [48, 49]. In animal models of embolic stroke, tPA amplifies MMP-9 [50]. Emerging clinical data may be consistent with the experimental literature. Patients with higher plasma levels of MMP-9 may be more susceptible to hemorrhagic transformation following tPA thrombolysis for acute ischemic strokes [51, 52].

In addition to serving as positive protein signals in plasma, MMP responses have also been detected in genetic and brain compartments of stroke patients. After ischemia or brain injury, circulating blood cells show rapid alterations in gene expression. In particular, responses in MMP-9 genes are highly conserved [53, 54]. In the brain parenchyma itself, MMP-9 positive astrocytes colocalize with peri-hematoma edema [55]. After ischemic strokes, MMP-9 positive neutrophils appear to coincide with local disruptions in microvessels [56, 57]. Taken together, these signals are broadly consistent with data and mechanisms derived from experimental models.

Beyond their utility as biomarkers, MMPs have also been pursued as potential therapeutic targets. In animal models of focal cerebral ischemia, MMP inhibitors reduce infarct volumes when administered early during acute ischemic strokes. Consistent with its proposed mechanisms, MMP inhibitors appear to be especially effective in terms of reducing brain edema and hemorrhage. One aspect of this pathophysiology with particular clinical relevance may be the relationship between MMPs and tPA thrombolysis. tPA is known to bind several lipoprotein receptors in cerebral endothelium that can upregulate MMP-9 [58]. Therefore, it is possible that some of the hemorrhagic transformation complications seen in tPA-thrombolysis patients may be caused by an inadvertent increase in MMPs [51, 52]. An obvious question is whether clinically acceptable compounds can be used as MMP inhibitors in stroke thrombolysis? In this regard, minocycline has been proposed as a potential “re-purposed drug” to target MMP-9 [59]. In experimental clot-based models of focal stroke in hypertensive rats, minocycline plus tPA as combination stroke therapy seem to suppress MMP-9, decrease hemorrhagic transformation and widen the therapeutic time window for safe and effective reperfusion [60]. Based in part on these experimental data, clinical trials have been started [61]. Initial findings are promising as minocycline appeared to dampen plasma MMP-9 biomarker levels, as hypothesized [62]. Nevertheless, some caution might be warranted. As discussed earlier, MMP blockade may interfere with endogeneous recovery after brain injury, and long-term use of minocycline worsened outcomes in an amyotrophic lateral sclerosis clinical trial [63].

Although the majority of clinical MMP data has been collected in ischemic strokes, recent efforts extend the role of these proteases to hemorrhagic strokes as well. MMPs are upregulated in subarachnoid hemorrhage patients, and blood and CSF levels of MMP-9 may track vasospasm and clinical outcomes [64]. Mechanistically, MMPs contribute to early brain injury and may also process gelsolin that can further amplify neuroinflammation [65]. In experimental models of subarachnoid hemorrhage, MMP inhibition improves outcomes [66]. Whether these targets work clinically remains to be determined.

NEUROVASCULAR ABNORMALITIES AND NEURODE-GENERATION

The data implicating MMPs in BBB disruption appear to be strongest in stroke and brain injury. However, accumulating data now suggest that similar mechanisms may operate in other CNS disorders. MMP-2 may be upregulated in the brain tissue of human patients with Parkinson's disease [67]. Blockade of MMPs can decrease cell death, neuroinflammaton and functional impairment in some experimental animal models of Parkinson's disease [68]. MMPs are also induced by and can process amyloid [69, 70]. In Alzheimer's disease and cerebral amyloid angiopathy, abnormal activation of MMPs may contribute to the BBB pathophysiology and neurodegeneration [71].

BBB leakage is commonly associated with edema in stroke and brain trauma. But BBB disturbances may also be important in other CNS disorders including Alzheimer's disease, Parkinson's disease, Huntington's disease, amyotrophic lateral sclerosis, multiple sclerosis etc [16]. Furthermore, BBB dysfunction is not always a binary “open or shut” phenomenon. There may be gradations in BBB permeability that reflect subtle perturbations in cell-cell signaling within the entire neurovascular unit. Perturbations in neurovascular unit function may impact a wide spectrum of neurodegenerative disorders.

CONCLUSIONS

The BBB is critically important for brain homeostasis. Any disruption in BBB integrity will impact brain function. Because the mechanisms that underlie BBB homeostasis are complex, identification of drug targets to ameliorate BBB problems in CNS disorders will not be easy. In this mini-review, we focused on MMPs. In the context of stroke, a large body of preclinical data and accumulating clinical findings support a role of MMPs as both biomarker as well as target. However, some caution may also be warranted since MMPs can play biphasic roles after brain injury – deleterious in the acute phase but potentially beneficial in delayed remodeling and recovery.