Vascular Events After Spinal Cord Injury: Contribution to Secondary Pathogenesis.

Spinal cord injury results in the initial physical disruption of structures in the spinal cord (primary insult) and in the generation of secondary events that collectively injure intact, neighboring tissue. The concept of secondary injury was originally described in 1911 by Allen,[1,2] who reported an improvement of neurological function in the injured spinal cord after myelotomy and removal of the contused tissue. These results suggested that noxious agents in the contused tissue were damaging adjacent, intact spinal cord segments and led to the development of the concept of primary and secondary injuries. Primary injury typically refers to the initial mechanical damage, whereas secondary injury is progressive cell injury that begins in the gray matter and progresses into the white matter.[3]

After traumatic injury, the primary mechanical injury to spinal cord blood vessels results in pronounced sequelae of secondary events, beginning with prominent hemorrhage into different compartments that correspond to the epidural, subdural, subarachnoid, and intramedullar spaces.[4,5] Blood can be toxic to the central nervous system (CNS).[6] In experimental models of traumatic spinal cord injury, the early intraparenchymal distribution of hemorrhage coincides at later time points with the appearance of a cavity, which is partially filled with nonneuronal structures.[4,5] Together, the findings illustrate that the disruption of blood vessels, a consequence of the primary insult, is integral to subsequent pathogenesis.

Secondary pathogenesis after traumatic spinal cord injury is related not only to grossly injured vessels but also to complex responses of the intact spinal cord vasculature. These responses include the disruption of the blood-spinal cord barrier and the generation of an inflammatory response. Both barrier disruption and inflammation disturb the microenvironment and expose neurons to plasma-derived cells and molecules that can be injurious to intact, neighboring tissue.[7]

This review is divided into 5 sections. In the first section, we describe the blood supply to the human spinal cord. It is important to appreciate the vascularity of the spinal cord, because it is known that vulnerability within the gray and white matter is, in part, related to segmental distribution of blood vessels. In addition, the anatomy of the vascular structures provides clues regarding the development of secondary injury along the axis of the spinal cord. In section 2, we characterize the secondary tissue responses to intraparenchymal and subarachnoid hemorrhage, which are direct consequences of the physical shearing or tearing of blood vessels. In section 3, We examine inflammatory responses that are implicated in both cell injury and synaptic plasticity in the injured spinal cord. In section 4, we provide an overview of the blood-spinal cord barrier, the consequences of barrier disruption, and the factors that modulate this abnormal permeability after spinal cord injury. In the final section, we describe the metalloproteinases, which are integral to barrier disruption, inflammation, and angiogenesis in the injured spinal cord.

Blood Supply to the Spinal Cord

The blood supply to the human spinal cord is provided by longitudinally oriented vascular trunks that receive a segmental arterial supply. With few exceptions, there is remarkable overlap of vascular territories.[8]

Extrinsic Arteries

The mature spinal cord receives its blood supply from a variable series of arteries that arise, in most cases, from the topographically closest arteries outside the vertebral column. A segmental spinal artery enters the intervertebral foramen and divides into 3 branches outside the spinal canal at each segmental level of the spinal cord: the anterior and posterior longitudinal spinal canal arteries and the radicular artery (Fig. 1).[8] The radicular artery continues along the nerve root and divides into an anterior radicular artery and a posterior radicular artery. After penetrating the dura mater, the anterior and posterior radicular arteries join 3 major arteries on the surface of the spinal cord: the anterior median longitudinal spinal artery and the right and left posterolateral longitudinal spinal arteries (Fig. 1).[9] Numerous anastomoses exist among these arteries, which collectively form an irregular net of vessels on the pia mater referred to as the "pial plexus" or "vasocorona" (Fig. 2). This anastomotic arrangement is particularly critical in the spinal cord injury, where local blood flow may be impaired by the injury.

[Figures 1-2 ILLUSTRATION OMITTED]

Branching of each radicular artery into anterior and posterior radicular arteries to the spinal cord is the exception in humans.[10] On average, there are 7 or 8 ventral vessels and approximately 15 dorsal vessels.[11] The most substantial vessels in the cervical region are radicular vessels that typically arise from the deep cervical artery and enter the spinal cord between C5 and C7. There is one large anterior radicular artery that serves the thoracolumbar region. This artery, initially described in 1882 by Adamkiewicz,[12] is integral to the blood supply to this region. Thoracolumbar ischemia has been demonstrated in animal models when the anterior radicular artery is occluded.[13]

Intrinsic Arteries

The intrinsic arteries of the spinal cord consist of central arteries, which originate from the anterior median longitudinal spinal artery and traverse the ventral median fissure sulcus, and arterioles, issuing from the pial plexus and the posterolateral longitudinal spinal artery (Fig. 2). The distribution of these central arteries varies along the axis of the cord, with the densest distribution in the cervical region (8-13 arteries per centimeter) and in the thoracolumbar region (2-3 arteries per centimeter).

The intramedullary territories of arterial supply can be roughly divided into 2 zones (Fig. 2). The first zone encompasses approximately two thirds of the cross section of the spinal cord and is supplied by the anterior medial longitudinal spinal artery (including the central artery and the vasocorona). The second zone serves the dorsal white matter and the dorsal horn and is supplied by arteries issuing from the posterolateral longitudinal spinal artery and the posterior part of the vasocorona. An intermediate zone is variably supplied by one system or the other.[14] These intramedullary territories of arterial supply are characterized by different flow directions, involving both centrifugal (central artery) and centripetal (vasocorona) systems. Together, these systems may define a watershed zone.

Concept of Watershed Zones

A watershed zone is a region that does not receive a direct blood supply but rather is dependent on overlapping terminal vascular fields. Regions within a watershed zone, therefore, may be particularly vulnerable if any of the contributing vascular fields are interrupted. Adamkiewicz[12] was the first clinician to differentiate territories of blood supply in the spinal cord. The early studies emphasized the variability in the number of vessels in certain areas of the spinal cord as well as the presence of overlapping vascular territories. The medullary segments with the highest incidence of arterial tributaries correspond to the regions with the highest density of neurons (ie, the cervical and thoracolumbar enlargements). In these areas, the anterior medial longitudinal spinal artery is also of large caliber. In the thoracic region, the vessels are fewer and smaller, and they anastomose with a more diminutive anterior longitudinal artery. These distinct differences in the vascularity in the thoracic cord are thought to contribute to the watershed zone.[15]

Although watershed zones have been implicated in secondary pathogenesis after spinal cord injury, there is reason to challenge this concept. The segmental variability in blood supply may not be predictive of vulnerability but may reflect local metabolic requirements. Neuronal activity depends on an adequate blood supply. In the spinal cord, there is a close correlation between segmental supply and the volume of gray matter.[16] Thus, the segmental, vascular arrangements may be indicative of local metabolic needs that are dictated by neuronal activity.[16]

Venous Drainage of the Spinal Cord

Large anterior and posterior central veins collect blood from both sides of the spinal cord accompanying the entering arteries but do not necessarily drain blood from the same areas as supplied by the nearest artery.[8] The posterior central vein has no arterial counterpart and drains into a posterior median longitudinal vein. Between the anterior and posterior central veins there is a central anastomosis that encircles the central canal. Radial veins join the venous pial plexus around the circumference of the spinal cord.[8] From the anterior and posterior longitudinal veins, the blood drains into anterior and posterior radicular veins that lie parallel and adjacent to the nerve root.

Partial Flow Theory

The blood flow to the spinal cord is not unidirectional, and blood flow can reverse within some vessels. There are ascending and descending blood flow currents in the ascending and descending branches of the radicular arteries forming the anterior median and posterolateral longitudinal spinal arteries.[14,17] This partial flow theory, originally described by Adamkiewicz[12] and verified more recently with spinal angiography,[14] acknowledges that blood flow into each spinal cord segment reflects the partial contribution of both ascending and descending flow currents. In spinal cord injuries, this multidirectional flow may serve to protect the spinal cord when local blood vessels are injured and incapable of supporting surrounding tissue.

Reversal of blood flow has been observed in response to normal physiological stimuli (changes in the posture of the spinal cord) and pathological conditions.[18] This reversing capability may likewise confer added protection in spinal cord injuries by diverting flow to those areas of greatest need.

Experimental Models of Spinal Cord Injury

Experimental models of spinal cord injury include contusions,[4,5,19-22] compression[23-25] ischemia,[13,26] and crush injuries.[27] Each model generates injuries that mimic certain clinical aspects of the mechanical damage and the ensuing posttraumatic ischemia. Researchers have also used either transection[28] or hemisection[29,30] to study injury. Transection and hemisection are less clinically relevant but offer the distinct advantage of consistent reproducibility. The hemisection model offers the ability to compare cellular responses in the ipsilateral and contralateral spinal cord. The common denominator among the experimental models is destruction of both gray and white matter. The lesioned site is characterized by early hemorrhage, breakdown of the blood-spinal cord barrier, neuronal cell loss, and infiltration of monocytes and macrophages and of neutrophils.[29,31,32] Progressive necrosis and active phagocytosis are key processes that act in concert to remodel the cytoarchitecture of the spinal cord. Several months postinjury, the injured spinal cord typically exhibits marked axonal degeneration and a fluid-filled cavity that occupies at least part of the gray matter and adjacent white matter.

Perhaps one of the most commonly used experimental models for studying spinal cord injury is the weight-drop contusion model. This model, originally described in 1911 by Allen,[1] involves dropping a calibrated weight a defined distance onto the exposed dura of the spinal cord. Since its original description, the procedure has been modified in order to better produce a graded, reproducible injury. The model has been instrumental in the development of optimal therapies for treating spinal cord injury, including the use of methylprednisolone (MP), which was shown to be neuroprotective after experimental contusion injuries.[33] These studies provided an impetus for subsequent evaluation of MP in human spinal cord injury, where it was found to improve neurologic function when given within 8 hours after injury.[34]

The contusion model has also offered important insights into the pathogenesis of secondary spinal cord injury.[33] A component of early cell death after spinal cord contusion injury is related to vascular events. The disruption of vessels leads to intraparenchymal hemorrhage; the disruption of the blood-spinal cord barrier, which is closely associated with edema formation; the release of vasoactive molecules that influence the extent of spinal cord perfusion; and the loss of autoregulation.[33] Together, these vascular events result in varying degrees of spinal cord ischemia. Ischemia not only results in the death of cells but also triggers a cascade of secondary events, including excitotoxicity, that lead to further damage to the spinal cord. Given this background, it is not surprising that there has been a considerable research effort directed toward the development of strategies for limiting the extent of ischemic injury.[33] Particular attention has been directed at the penumbral zone surrounding the ischemic, necrotic core. This region exhibits progressive cell death over time and thus is a target for therapeutic intervention. The clinical relevance of the penumbral zone is based, in part, on an important observation that was first identified in an experimental model of spinal cord contusion injury, namely that effective locomotion can occur with only 5% to 10% of the original axonal population.[35] Thus, protecting the penumbral zone could have an important impact on the extent of functional recovery.

Spinal Cord Injury and Hemorrhage

Relationship to Secondary Pathogenesis

Mechanical trauma to the spinal cord causes immediate vasospasm of the superficial vessels and intraparenchymal hemorrhage, which is initially localized in the highly vascularized and most vulnerable central gray matter.[36] The generation of central gray matter injury may be best understood by a consideration of the distribution of the mechanical forces in the traumatized spinal cord. For example, an anteroposterior compressive force on a flexible tube filled with gelatin, a physical model of the contused spinal cord, produces longitudinal stress that is most intense in the center of the tube.[35] Thus, the vessels on the surface of the spinal cord are relatively spared from initial insult, whereas the microvasculature in the gray matter is subjected to stretching and shear stress, a consequence of the difference in compliance between the gray and white matter.[37] The immediate mechanical damage to gray matter microvasculature impairs the microcirculation and impedes perfusion,[38,39] leading to a profound reduction of spinal cord blood[40,41] accompanied by impairment of autoregulation.[40] The impaired blood flow in the traumatized spinal Cord may be further compromised by systemic responses, including posttraumatic hypotension, bradycardia, and decreased cardiac output, that exacerbate the ischemic damage.[42]

Intraparenchymal Hemorrhage

Noble and Wrathall[4,5,21] studied the anatomical characteristics of central hemorrhage after graded contusive injury in the rat. The size of the hemorrhage, which is directly proportional to the severity of the initial impact, is maximal at the injury site and extends into the dorsal columns in the rostral and caudal segments. The central hemorrhage occupies the gray matter and a variable proportion of adjacent white matter at the epicenter of the injury. At distant sites, the hemorrhage is typically found in the most central part of the dorsal column (Fig. 3).

[Figure 3 ILLUSTRATION OMITTED]

There is both indirect and direct evidence that hemorrhages may be damaging to the CNS. Perhaps the most compelling indirect evidence relates to the evolution of a fusiform-shaped cavity that closely approximates the distribution of early intraparenchymal hemorrhage.[4,5] Evidence that tissue damage may be a consequence of this intraparenchymal hemorrhage comes from a study by Bullock and Fujisawa,[43] who demonstrated that injection of blood into the extra-axial space resulted in cell damage.

Several factors may contribute to the tissue necrosis following central hemorrhage. Blood flow is reduced in tissue around hemorrhage, resulting in different degrees of ischemia.[40] Ischemic damage may arise from one or all of the following events: (1) vasogenic edema as a consequence of blood-spinal cord barrier breakdown,[44] (2) direct compression by adjacent tissue,[45] and (3) vasospasm, which occurs as a result of mechanical trauma or due to exposure to red blood cell components, including oxyhemoglobin and endothelin.[36,46,47] Free radicals are rapidly generated upon reperfusion of ischemic tissue or hypoperfused tissue. Activation of phagocytic cells and liberation of catalytic metalions, which are present in high concentrations during the degradation of hemoglobin, is one process that results in the production of free radicals. The CNS is particularly prone to free radical damage.[48] The cell membranes are rich in polysaturated fatty acid chains, which are sensitive to free radical attack. In addition, the CNS has limited antioxidant defense mechanisms. The brain and spinal cord exhibit low levels of catalase activity and only moderate levels of superoxide dismutase and glutathione peroxidase.[49]

Heine Oxygenase-1

The experimentally contused spinal cord generates substantial intraparenchymal hemorrhage. A focus of our laboratory has been to define how neural and glial cells respond to hemorrhage. This research interest has led to us investigate the role of heme oxygenase-1 (HO-1) in the traumatized spinal cord. Heme oxygenase (HO) is the rate-limiting enzyme that is involved in the degradation of heme to biliverdin, carbon monoxide, and iron.[50] This pathway is of particular importance because bile pigments are potent antioxidants and may therefore serve an important function in cellular defense against free radical-mediated cell damage.[51]

There are at least 3 HO enzymes: (1) HO-1, the inducible form found mainly in microglia,[52] (2) HO-2, the constitutive form.[53] and (3) HO-3, whose transcripts have been recently reported in the CNS.[54] Heme oxygenase-1 is induced by a variety of stimuli, including heme.[55-60] In the intact spinal cord, basal expression of HO-1 is restricted to neurons.[52,53] However, we have demonstrated marked induction of HO-1 in astrocytes and in microglia and macrophages after contusion injury.[52] The pattern of induction corresponds to regions exhibiting posttraumatic hemorrhage.[52] Hemorrhage, therefore, may be a potent inducer of HO-1 after spinal cord injury. To begin to test this hypothesis, we have exposed the subarachnoid space, overlying the intact spinal cord, to lysed blood. Preliminary data demonstrate focal induction of HO-1 in glia, a finding that suggests these cells may play an important role in sequestering and metabolizing heme (Fig. 4).

[Figure 4 ILLUSTRATION OMITTED]

The central question, however, is whether glial induction of HO-1 is protective or detrimental to the traumatized spinal cord. Although somewhat controversial,[61] there is growing evidence that HO-1 is protective in CNS injury.[62,63] Panahian et al,[62] in a recent study using transgenic mice that overexpress HO-1, found an attenuation of cell injury after cerebral ischemia. Whether similar protection occurs in the hemorrhagic traumatized spinal cord remains to be determined.

Inflammation

Inflammation: The Patient With Spinal Cord Injury

Inflammation is central to secondary pathogenesis after spinal cord injury. In the clinical setting, there is compelling evidence to establish early and prolonged inflammation in the traumatized spinal cord. This inflammation is exemplified in a clinical retrospective analysis involving 1,917 patients over a period of 10 years.[64] The number of white blood cells was found to be elevated in the cerebrospinal fluid within the first 7 days after spinal cord injury. It is likely that the increased number of leukocytes is indicative of an early immune response. The inflammation is not limited to the acutely traumatized spinal cord, but rather occurs over a period of weeks in regions of white matter undergoing wallerian degeneration. Recent research has shown elevated plasma levels of inflammatory mediators, including cytokines (ie, interleukin-2, interleukin-6), the soluble interleukin-2 receptor, and intercellular adhesion molecule-1 (ICAM-1), in patients with long-standing spinal cord injury.[65] Such findings emphasize prolonged inflammatory responses in human spinal cord injury.

Inflammatory Cells

Microglia, leukocytes (lymphocytes, neutrophils, and monocytes), and astrocytes contribute to the cellular inflammatory response after experimental spinal cord injury.[29,66-71] Our review focuses on those cells that are most closely associated with the vasculature, namely leukocytes and macrophages.

Time Course of the Inflammatory Response

Traumatic spinal cord injury results in both a primary injury and a cascade of secondary processes that collectively lead to additional loss of tissue.[29,66,67,72] Posttraumatic inflammation, characterized, in part, by the accumulation of activated microglia and macrophages, is thought to contribute to secondary pathogenesis.[66,67,72-74]

Infiltration of neutrophils signals an early inflammatory response after spinal cord injury.[29,75,76] Neutrophils infiltrate the traumatized cord within the first hour postinjury, peak at 24 hours postinjury,[29,68,75-77] begin to diminish by 48 hours postinjury,[29] and are negligible by 7 days postinjury.[76] The early appearance of neutrophils most likely reflects the early hemorrhage. Activated macrophages are apparent early after injury, but they plateau between 2 and 4 weeks postinjury in the contused spinal cord.[72] There is a rapid transformation of resident microglia into macrophages.[72] which precedes infiltration of macrophages derived from infiltrating monocyte precursors. B and T lymphocytes appear in the injured spinal cord within hours of the injury and persist up to 7 days postinjury.[70,71] There is marked infiltration of B lymphocytes between 3 and 6 hours postinjury, and both B and T lymphocytes are identified in the spinal cord by 4 days postinjury. There is a marked decline in these cells by 7 days postinjury.[71]

The role of inflammation in spinal cord injury is controversial. Inflammatory cells are associated with delayed neuronal death and demyelination,[74,78-80] and they also may be integral to neural regeneration.[74,76,79,81,82] Infiltrated leukocytes and endogenous glia, acting as macrophages, mediate tissue damage, including myelin vesiculation and lipid peroxidation, through the generation of a variety of toxic molecules (eg, reactive oxygen and nitrosyl radicals, cytokines and chemokines).[80,83-89] These molecules are thought to damage surrounding healthy tissue. This hypothesis is supported by a recent study by Carlson et al[68] that demonstrated a correlation between the axial extent of tissue damage and the numbers of macrophages and microglia. Strategies aimed at blocking neutrophil or macrophage influx and at inhibition of phagocytic and secretory activity of macrophages in the injured spinal cord have resulted in neuroprotection and improved locomotory function.[85]

There is also evidence indicating that inflammation may play a beneficial role in the traumatized spinal cord. Macrophages and microglia promote regeneration of axons by scavenging myelin and neuronal debris[74,76,79,81,82]; by producing the proregenerative cytokine, transforming growth factor-beta ([TGF-.sub.[Beta]])[76]; and by enhancing neurite outgrowth.[90] The enhancement of neurite outgrowth is exemplified by a recent study that demonstrated that the extent of axonal regeneration in the injured rat spinal cord was correlated with the presence of phagocytic cells.[91] In that study, nitrocellulose membranes, which were treated with [TGF-.sub.[Beta]] or coated with microglia, were co-transplanted with fetal spinal cord tissue into the injured spinal cord of an adult rat. Cut dorsal roots were apposed to both sides of the nitrocellulose. Four weeks later, regenerated sensory axons were found to be associated with macrophages. Conversely, this axonal regeneration could be experimentally inhibited by implanting a nitrocellulose strip containing a macrophage inhibitory factor, suggesting a prominent role for phagocytic cells in the regenerative response.[91]

The Blood-Spinal Cord Barrier

Arteries in the subarachnoid space become arterioles as they penetrate into the substance of the spinal cord. The subarachnoid space surrounds the penetrating arteriole and is referred to as the "Virchow-Robin space" (Fig. 5). This space contains not only cerebrospinal fluid but also pia-arachnoid cells, which have the capacity to behave as phagocytes. The arteriole wall is composed of endothelial cells, which are surrounded by a smooth muscle coat. The arteriole terminates at the level of the capillary. The capillary has several distinguishing features; it is smaller in diameter than that of the arteriole, lacks a smooth muscle coat, and is marginally contacted by pericytes. A basal lamina surrounds the endothelial cell and splits to enclose each pericyte. Astrocytic foot processes abut the parenchymal side of the basal lamina (Figs. 6 and 7).

[Figures 5-7 ILLUSTRATION OMITTED]

The blood-brain/spinal cord barrier is located at the level of the capillary and is composed of specialized endothelial cells that regulate and restrict transport of molecules into the CNS. This specialized interface provides a stable microenvironment, which is necessary for normal neuronal function.[92] The morphologic basis of the blood-brain/spinal cord barrier is attributed, in part, to the presence of fight junctions between adjacent endothelial cells that effectively block the intercellular movement of large molecules, including plasma proteins.[93] The restrictive nature of the barrier is also attributed to a complex glycoprotein-rich glycocalyx, which is located on the luminal (humeral) front of the endothelial cell. This glycocalyx is anionic (negative) in charge (Fig. 8) and thus serves as a repulsive interface to circulating plasma proteins that bear a similar charge.[94] The basement membrane maintains a close relationship with the abluminal (parenchymal) side of the endothelial cell. This structure is thought to play a critical role in maintaining the integrity of the barrier, in part, by providing structural support to the endothelial cell wall.[95]

[Figure 8 ILLUSTRATION OMITTED]

Time Course of Barrier Disruption

Disruption of the blood-spinal cord barrier after spinal cord injury is characterized by a transient loss of anionic charged sites along the endothelial glycocalyx[4] and indiscriminate extravasation of plasma proteins[4] (Figs. 8 and 9). The time course for barrier disruption to circulating molecules varies from 4 to 28 days postinjury and is not restricted to the injured site but extends along the axis of the spinal cord.[4,96] A recent study by Popovich et al[96] demonstrated that the blood-spinal cord barrier remains permeable to small molecules such as [14C]-aminoisobutyric acid up to at least 28 days postinjury. Such an extended time course for barrier breakdown has been confirmed in a recent magnetic resonance imaging study after a spinal cord contusion.[97]

[Figure 9 ILLUSTRATION OMITTED]

Importance of Barrier Disruption

Altered barrier permeability exposes the spinal cord to the toxic effects of inflammatory cells.[7] Amino acid neurotransmitters such as glutamate and glycine, when present at high concentrations, also can be toxic to cells.[7] As noted earlier, there is substantial evidence that infiltrating leukocytes can be injurious to neural structures. There is also evidence that barrier disruption, in the absence of an inflammatory response, can be detrimental. For example, barrier breakdown resulting from a transient rise in blood pressure has been shown to produce a pathology consistent with irreversible tissue damage.[98] Some researchers[99,100] have demonstrated that transient opening of the blood-brain barrier consistently induces the stress proteins HSP32 and HSP70. The stress protein HSP70 is an established marker of cell injury,[101] whereas HSP32 has been reported to be an indicator of oxidative stress.[102,103] Taken together, these findings emphasize the importance of barrier breakdown in secondary pathogenesis after spinal cord injury.

Factors That Contribute to Abnormal Barrier Permeability

The traumatized spinal cord is exposed to a wide spectrum of substances, including cytokines[32] and vasoactive peptides that have been implicated in barrier disruption and edema formation.[32,104] In our laboratory, we have been particularly interested in the contribution of the vasoactive peptide endothelin-1 (ET-1) in modulation of the blood-spinal cord barrier after spinal cord injury.

Endothelin-1 is a 21-amino acid peptide that was originally identified by Yanagisawa et al[105] in the supernatant of aortic endothelial cells. Since the characterization of ET-1, 3 additional isoforms have been identified: endothelin-2 (ET-2), endothelin-3 (ET-3), and vasoactive intestinal constrictor polypeptide (VIC). These peptides, encoded by separate genes, consist of 21 amino acid residues joined by 2 disulfide bridges with 6 conserved amino acid residues at the carboxy terminus.[106]

There is evidence that implicates ET-1 in barrier disruption after spinal cord injury. There is an increased expression of this peptide in the injured spinal cord,[107] which correlates with the pattern of blood-spinal cord barrier breakdown.[108] Intrathecal administration of ET-1 in the intact spinal cord results in disruption of the blood-spinal cord barrier.[104] Finally, pharmacologic blockade of ET-1-mediated vasoconstriction attenuates breakdown of the blood-spinal cord barrier after traumatic spinal cord injury.[108]

The biology of ET-1-mediated pathology relates, in part, to the endothelin receptor subtypes. There are at least 3 endothelin receptor subtypes, designated [ET.sub.A], [ET.sub.B1], and [ET.sub.B2], that influence vascular reactivity in the CNS.[106,109] The [ET.sub.A] receptor is localized in vascular smooth muscle and mediates prominent, sustained vasoconstriction. The [ET.sub.B] receptor is subdivided into the [ET.sub.B1] and [ET.sub.B2] types. The [ET.sub.B1] subtype is localized in vascular endothelial cells and generates vasodilation, whereas the [ET.sub.B2] receptor is present in vascular smooth muscle and produces vasoconstriction. Thus, ET-1-mediated vascular changes (ie, barrier disruption, vasoconstriction) in the traumatized spinal cord could be executed through either the [ET.sub.A] or [ET.sub.B2] receptor subtype. The [ET.sub.A] receptor has a particularly close link to vasospasm, secondary to subarachnoid hemorrhage,[46] and prolonged vasoconstriction,[6] concomitant with ischemic damage.[110]

There are 3 probable sources of ET-1 after hemorrhage. The first source may be derived from the humoral compartment. Increased ET-1 levels, which are observed after subarachnoid hemorrhage,[111] could enter the CNS through a disrupted blood-CNS barrier. The second source may be the injured tissue itself, and the third source may be red blood cells. An in vitro study by Tippler et al[112] demonstrated a 3-fold increase in ET-1 immunoreactivity in the lysate of disrupted red blood cells.

The mechanism whereby ET-1 disrupts the blood-spinal cord barrier may be related to its role as a prominent vasoconstrictor. Endothelin-1 has the ability to produce prolonged periods of vasospasm,[6] resulting in ischemic damage to neurons[110] and disruption of the barrier.[104] Most recently it has been reported that intrathecal administration of endothelin generates free radicals.[113] Such a finding suggests that endothelin-mediated barrier disruption may reflect an oxidative insult.

Metalloproteinases and Spinal Cord Injury

Matrix metalloproteinases (MMPs) are a large family of enzymes that mediate extracellular matrix degradation and release of growth factors and cytokines from the matrix. This important family of inflammatory enzymes has become a new area of interest in the study of CNS injury. Some researchers[114,115] have demonstrated the presence of MMPs in disease and injury in the brain and spinal cord. Experimental models suggest that MMP activity is required for inflammatory cell infiltration and may contribute to both abnormal blood-brain barrier permeability and ischemia-induced angiogenesis following injury.

MMP Family

The MMPs are zinc- and calcium-dependent endopeptidases that together can hydrolyze essentially all of the components of the extracellular matrix. These endopeptidases are roughly classified by substrate classes and therefore fall into the categories of collagenases, gelatinases, stromelysins, membrane-type MMPs, and "other MMPs."

MMPs and Inflammation

Matrix metalloproteinase activity is required for the inflammatory cell infiltration that occurs following spinal cord injury and most likely contributes to early barrier disruption. The early inflammatory response involves an initial wave of infiltrating neutrophils, followed by migration of monocytes and macrophages into injured segment. Each of these inflammatory cells expresses MMPs, including MMP-2 (gelatinase A), MMP-8 (neutrophil collagenase), MMP-9 (gelatinase B), MMP-11 (stromelysin-3), and MMP-12 (metalloelastase). Together, these MMPs are thought to participate in infiltration and migration, tissue destruction, extracellular matrix degradation, blood-spinal cord barrier disruption, and edema.[116-122]

Matrix metalloproteinase activity is thought to be required for the infiltration of inflammatory cells because they must degrade the basal lamina that surrounds blood vessels, a structure that is normally a barrier to cell migration. The degradation of this basal lamina leads to the next consequence of MMP activity: increased permeability of the blood-spinal cord barrier. In the CNS, vascular endothelial cells form a tight junction, and, together with astrocyte processes, form a specialized basal lamina. During the infiltration of inflammatory cells and during angiogenesis, this basal lamina becomes degraded and the barrier becomes permeable.[123] Several experimental models of brain injury demonstrate the relationship between MMP activity and blood-brain barrier permeability. For example, injection of proinflammatory cytokines or tumor necrosis factor-[Alpha] induces MMP-9 expression in the brain.[124] The subsequent breakdown of the blood-brain barrier can be directly attributed to MMP activity in that treatment with MMP inhibitors blocks this abnormal barrier permeability.[124,125] A similar relationship has been established in ischemic brain injury. Romanic et al[120] demonstrated that an MMP-9 neutralizing antibody reduced infarct size, thus underscoring the contribution of MMP activity in the ischemic brain.

MMPs and Angiogenesis

Angiogenesis occurs following spinal cord injury in response to localized tissue hypoxia. Membrane type 1 MMP (MMP-1) is found on endothelial cells and is required for endothelial cell migration in angiogenesis.[126] Hypoxic and ischemic conditions occur as a result of the physical damage to blood vessels as well as the hypoperfusion and vasodilation that accompany spinal cord injury. Proangiogenic factors released from the injured site cause rapid increases in vascular permeability.[127-129] Fibrinogen leaking from the microvascular bed is polymerized into a fibrin matrix. Endothelial cells require MMP-1 activity to degrade this matrix and invade the surrounding tissue.[126] Many researchers[130-133] have demonstrated increased MMP-1, -2, and -9 activity at sites of angiogenesis. Vascular endothelial growth factor expression, known for its proangiogenic as well as endothelial cell permeability activities, also strongly correlates with MMP-2 and -9 expression[134,135] and is thought to up-regulate MMP-9 expression.[136] Matrix metalloproteinase proteolysis has 4 consequences: (1) it permits invasion of endothelial cells into the surrounding matrix, (2) it generates extracellular matrix degradation products that are chemotactic for endothelial cells, (3) it activates and releases growth factors localized in the extracellular matrix, and (4) it results in increased permeability of the blood-spinal cord barrier, as discussed earlier. Whether angiogenesis is beneficial or detrimental to the traumatized spinal cord is not clear. Recent data support a role for angiogenesis in promoting neural regeneration,[137] but other data suggest that inhibition of angiogenesis is neuroprotective.[138]

Summary

Spinal cord injury results in injury to both neural and vascular elements. The extent to which the vascular injury contributes to secondary pathogenesis is dependent on not only the initial disruption of blood vessels, leading to prominent intraparenchymal hemorrhage, but also on progressive disruption of the blood-spinal cord barrier coincident with the infiltration of inflammatory cells. Collectively, these events influence both acutely and chronically injured spinal cords and define, in part, the extent of functional recovery.

References

[1] Allen AR. Surgery of experimental lesion of spinal cord equivalent to crush injury of fracture dislocation of spinal column: preliminary report. JAMA. 1911;57:878.

[2] Allen AR. Remarks in histopathological changes in spinal cord due to impact: an experimental study. J Nerv Ment Dis. 1914;41:141-147.

[3] Balentine JD. Pathology of experimental spinal cord trauma, I: the necrotic lesion as a function of vascular injury. Lab Invest. 1978;39:236-253.

[4] Noble LJ, Wrathall JR. Distribution and time course of protein extravasation in the rat spinal cord after contusive injury. Brain Res. 1989;482:57-66.

[5] Noble LJ, Wrathall JR. Correlative analyses of lesion development and functional status after graded spinal cord contusive injures in the rat. Exp Neurol. 1989;103:34-40.

[6] Asano T, Tanishima T, Sasaki T, Sano K. Possible participation of free radical reactions initiated by clot lysis in the pathogenesis of vasospasm after subarachnoid hemorrhage. In: Wilkins RH, ed. Cerebral Arterial Spasm. Baltimore, Md: Williams & Wilkins; 1980:190-201.

[7] Schlosshauer B. The blood-brain barrier: morphology, molecules, and neurothelin. Bioessay. 1993;15:341-346.

[8] Crock HV, Yoshizawa H, eds. The Blood Supply of the Vertebral Column and Spinal Cord in Man. New York, NY: Springer; 1977.

[9] Domisse GF, ed. The Arteries and Veins of the Human Spinal Cord From Birth. New York, NY: Churchill Livingstone Inc; 1975.

[10] Jellinger K. Experimental studies on the question of the vertebral artery supply of the spinal cord. Acta Neuropathol. 1966;6:201-208.

[11] Brockstein B, Johns L, Gewertz BL. Blood supply to the spinal cord: anatomic and physiological correlations. Ann Vasc Surg. 1994;8:394-399.

[12] Adamkiewicz A. Die Blutgefa[Beta]e des menschlichen Ruckenmarks: Die Gefa[Beta]e der Ruckenmarksoberflache. Sitzungsberichte der Akademie der Wissenschaften in Wein--Mathematisch-Naturwissenschaftliche Klasse--Abteilung 3. 1882;84:101-130.

[13] Zivin JA, De Giorlami U. Spinal cord infarction: a highly reproducible stroke model. Stroke. 1980;11:200-202.

[14] Thron AK, ed. Vascular Anatomy of the Spinal Cord. New York, NY: Springer-Verlag New York Inc; 1988.

[15] Suh TH, Alexander L. Vascular system of the human spinal cord. Arch Neurol Psych. 1939;41:659-677.

[16] Gillilan LA. The arterial blood supply of the human spinal cord. J Comp Neurol. 1958;110:75-103.

[17] Di Chiro G, Fried LC. Blood flow currents in spinal cord arteries. Neurology. 1971;21:1088-1096.

[18] Viraswami V. A Study of the Blood Supply of the Spinal Nerve Roots in Man and the Rabbit With an Experimental Analysis of the Collateral Circulation Following Ligation of Arteries [PhD thesis]. London, England: University of London; 1963.

[19] Basso DM, Beattie MS, Bresnahan JC, et al. MASCIS evaluation of open field locomotor scores: effects of experience and teamwork on reliability. Multicenter Animal Spinal Cord Injury Study. J Neurotrauma. 1996;13:343-359.

[20] Beattie MS, Bresnahan JC, Komon J, et al. Endogenous repair after spinal cord contusion injuries in the rat. Exp Neurol. 1997;148:453-463.

[21] Noble LJ, Wrathall JR. Spinal cord contusion in the rat: morphometric analyses of alterations in the spinal cord. Exp Neurol. 1985;88:135-149.

[22] Wrathall JR, Pettegrew RK, Harvey F. Spinal cord contusion in the rat: production of graded reproducible, injury groups. Exp Neurol. 1985;88:108-122.

[23] Guth L, Zhang Z, DiProspero NA, et al. Spinal cord injury in the rat: treatment with lipopolysaccharide and indomethacin enhances cellular repair and locomotor function. Exp Neurol. 1994;126:76-87.

[24] Nacimiento AC, Bartels M, Loew F. Increased functional vulnerability to acute compression injury of a spinal cord segment under barbiturate anesthesia. Surg Neurol. 1986;26:9-12.

[25] Rivlin AS, Tator CH. Effect of duration of acute spinal cord compression in a new acute cord injury model in the rat. Surg Neutrol. 1978;10:38-43.

[26] Kanellopoulus GK, Kato H, Hsu CY, Kouchoukos NT. Spinal cord ischemic injury: development of a new model in the rat. Stroke. 1997;28:2532-2538.

[27] Zhang Z, Krebs CJ, Guth L. Experimental analysis of progressive necrosis after spinal cord trauma in the rat: etiological role of the inflammatory response. Exp Neurol. 1997;143:141-152.

[28] Noble LJ, Ellison JA. Effect of transection of the blood-spinal cord barrier of the rat after isolation from descending sources. Brain Res. 1989;487:299-310.

[29] Dusart I, Schwab ME. Secondary cell death and the inflammatory reaction after dorsal hemisection of the rat spinal cord. Eur J Neurosci. 1994;6:712-724.

[30] Goshgarian HG, Xu XY, Rafolds JA. Neuronal and glial changes in the rat phrenic nucleus occurring within hours after spinal cord injury. J Comp Neurol. 1989;284:519-533.

[31] Bartholdi D, Schwab ME. Methylprednisolone inhibits early inflammatory processes but not ischemic cell death after experimental spinal cord lesion in the rat. Brain Res. 1995;672:177-186.

[32] Bartholdi D, Schwab ME. Expression of pro-inflammatory cytokine and chemokine mRNA upon experimental spinal cord injury in mouse: an in situ hybridization study. Eur J Neurosci. 1997;9:1422-1438.

[33] Amar AP, Levy ML. Pathogenesis and pharmacological strategies for mitigating secondary damage in acute spinal cord injury. Neurosurgery. 1999;44:1027-1039.

[34] Bracken MB, Shepard MJ, Collins WF, et al. A randomized, controlled trial of methylprednisolone or naloxone in the treatment of acute spinal-cord

injury: results of the Second National Acute Spinal Cord Injury Study. N Engl J Med. 1990;322:1405-1411.

[35] Blight AR, DeCrescito V. Morphometric analysis of experimental spinal cord injury in the cat: the relation of injury intensity to survival of myelinated axons. Neuroscience. 1986;19:321-341.

[36] Ducker TB, Assenmacher DR. Microvascular response to experimental spinal cord trauma. Surg Forum. 1969;20:428-430.

[37] Tator CH, Koyanagi I. Vascular mechanisms in the pathophysiolosy of human spinal cord injury. J Neurosurg. 1997;86:483-492.

[38] Dohrmann GJ, Wagner FC Jr, Bucy PC. Transitory traumatic paraplegia: electron microscopy of early alterations in myelinated nerve fibers. J Neurosurg 1972;36:407-415.

[39] Schneider A. Kreislaufstorungen und Gefa[Beta]prozesse des Ruckenmarks. In: Doerr W, Seifert G, eds. Spezielle Pathologische Anatomie. Berlin, Germany: Springer; 1980:511-649.

[40] Senter HJ, Venes JL. Altered blood flow and secondary injury in experimental spinal cord trauma. J Neurosurg. 1978;49:569-578.

[41] Young W, Flamm ES, Demopoulos HB, et al. Effect of naloxone on posttraumatic ischemia in experimental spinal contusion. J Neurosurg. 1981;55:209-219.

[42] Guha A, Tator CH. Acute cardiovascular effects of experimental spinal cord injury. J Trauma. 1988;28:481-490.

[43] Bullock R, Fujisawa H. The role of glutamate antagonists for the treatment of CNS injury. J Neurotrauma. 1992;9(suppl 2):443-473.

[44] Yang GY, Betz AL, Chenevert TL, et al. Experimental intracerebral hemorrhage: relationship between brain edema, blood flow, and blood-brain barrier permeability in rats. J Neurosurg. 1994;81:93-102.

[45] Mendelow AD, Bullock R, Teasdale GM, et al. Intracranial haemorrhage induced at arterial pressure in the rat, part 2: short-term changes in local cerebral blood flow measured by autoradiography. Neurol Res. 1984;6:189-193.

[46] Clozel M, Breu V, Burri K, et al. Pathophysiological role of endothelin revealed by the first orally active endothelin antagonist. Nature. 1993;365:759-761.

[47] Macdonald RL, Weir BK. A review of hemoglobin and the pathogenesis of cerebral vasospasm. Stroke. 1991;22:971-982.

[48] LeBel CP, Bondy SC. Oxygen radicals: common mediators of neurotoxicity. Neurotoxicol Teratol. 1991;13:341-346.

[49] Halliwell B. Reactive oxygen species and the central nervous system. J Neurochem. 1992;59:1609-1623.

[50] Maines MD. Heme oxygenase: function, multiplicity, regulatory mechanism, and clinical applications. FASEB J. 1988;2:2557-2568.

[51] Stocker R. Induction of haem oxygenase as a defence against oxidative stress. Free Radic Res Commun. 1990;9:101-112.

[52] Mautes AE, Kim DH, Sharp FR, et al. Induction of heme oxygenase-1 (HO-1) in the contused spinal cord. Brain Res. 1998;795:17-24.

[53] Ewing JF, Haber SN, Maines MD. Normal and heat-induced patterns of expression of heme oxygenase-1 (HSP32) in rat brain: hyperthermia causes rapid induction of mRNA and protein. J Neurochem. 1992;58:1140-1149.

[54] McCoubrey WK Jr, Huang TJ, Maines MD. Isolation and characterization of a cDNA from the rat brain that encodes hemoprotein oxygenase-3. Eur J Biochem. 1997;247:725-732.

[55] Dwyer BE, Nishimura RN, Lu SY. Differential localization of heme oxygenase and NADPH-diaphorase in spinal cord neurons. Neuroreport. 1995;6:973-976.

[56] Lavrovsky Y, Schwartzmann ML, Abraham NG. Novel regulatory sites of the human heme oxygenase-1 promotor region. Biochem Biophys Res Commun. 1993;196:336-341.

[57] Lee PJ, Jian BH, Chin BY, et al. Hypoxia-inducible factor-1 mediates transcriptional activation of heme oxygenase-1 gene in response to hypoxia. J Biol Chem. 1997;272:5375-5381.

[58] Mitani K, Fujita H, Kappas A, Sassa S. Heme oxygenase is a positive acute-phase reactant in human Hep3B hepatoma cells. Blood. 1992;79:1255-1259.

[59] Muller RM, Taguchi H, Shibahara S. Nucleotide sequence and organization of the rat heme oxygenase gene. J Biol Chem. 1987;262:6795-6802.

[60] Okinaga S, Shibahara S. Identification of a nuclear protein that constitutively recognizes the sequence containing a heat shock element: its binding properties, and possible function modulating heat-shock induction of the rat heme oxygenase gene. Eur J Biochem. 1993;212:167-175.

[61] Panizzon KL, Dwyer BE, Nishimura RN, Wallis RA. Neuroprotection against CA1 injury with metalloporphyrins. Neuroreport. 1996;7:662-666.

[62] Panahian N, Yoskiura M, Maines MD. Overexpression of heme oxygenase-1 is neuroprotective in a model of permanent middle cerebral artery occlusion in transgenic mice. J Neurochem. 1999;72:1187-1203.

[63] Takizawa S, Hirabayashi H, Matsushima K, et al. Induction of heme oxygenase protein protect neurons in cortex and striatum, but not in the hippocampus, against transient forebrain ischemia. J Cereb Blood Flow Metab. 1998;18:559-569.

[64] Travlos A, Anton HA, Wing PC. Cerebrospinal fluid cell count following spinal cord injury. Arch Phys Med Rehabil. 1994;75:293-296.

[65] Segal JL, Gonzales E, Yousefi S, et al. Circulating levels of IL-2R, ICAM-1, and IL-6 in spinal cord injuries. Arch Phys Med Rehabil. 1997;78:44-47.

[66] Blight AR. Macrophages and inflammatory damage in spinal cord injury. J Neurotrauma. 1992;9(suppl 1):S83-S91.

[67] Blight AR, Cohen TI, Saito K, Heyes MP. Quinolinic acid accumulation and functional deficits following experimental spinal cord injury. Brain. 1995;118(pt 3):735-752.

[68] Carlson SL, Parrish ME, Springer JE, et al. Acute inflammatory response in spinal cord following impact injury. Exp Neurol. 1998;151:77-88.

[69] Koshinaga M, Whittemore SR. The temporal and spatial activation of microglia in fiber tracts undergoing anterograde and retrograde degeneration following spinal cord lesion. J Neurotrauma. 1995;12:209-222.

[70] Popovich PG, Wei P, Stokes BT. Cellular inflammatory response after spinal cord injury in Sprague-Dawley and Lewis rats. J Comp Neurol. 1997;377:443-464.

[71] Schnell L, Schneider R, Berman MA, et al. Lymphocyte recruitment following spinal cord injury in mice is altered by prior viral exposure. Eur J Neurosci. 1997;9:1000-1007.

[72] Popovich PG, Reinhard JF Jr, Flanagan EM, Stokes BT. Elevation of the neurotoxin quinolinic acid occurs following spinal cord trauma. Brain Res. 1994;633:348-352.

[73] Bethea JR, Castro M, Keane RW, et al. Traumatic spinal cord injury induces nuclear factor-kappaB activation. J Neurosci. 1998;18: 3251-3260.

[74] Hirschberg DL, Yoles E, Belkin M, Schwartz M. Inflammation after axonal injury has conflicting consequences for recovery of function: rescue of spared axons is impaired but regeneration is supported. J Neuroimmunol. 1994;50:9-16.

[75] Means ED, Anderson DK. Neuronophagia by leukocytes in experimental spinal cord injury. J Neuropathol Exp Neurol. 1983;42:707-719.

[76] Streit WJ, Semple-Rowland SL, Hurley SD, et al. Cytokine mRNA profiles in contused spinal cord and axotomized facial nucleus suggest a beneficial role for inflammation and gliosis. Exp Neurol. 1998;152:74-87.

[77] XuJA, Hsu CY, Liu TH, et al. Leukotriene B4 release and polymorphonuclear cell infiltration in spinal cord injury. J Neurochem. 1990;55:907-912.

[78] Blight AR. Delayed demyelination and macrophage invasion: a candidate for secondary cell damage in spinal cord injury. Cent Nerv Syst Trauma. 1985;2:299-315.

[79] Davis S, Bouchard C, Tsatas O, Giftochristos N. Macrophages can modify the nonpermissive nature of the adult mammalian central nervous system. Neuron. 1990;5:463-469.

[80] Dijkstra CD, Dopp EA, van den Berg TK, Damoiseaux JG. Monoclonal antibodies against rat macrophages. J Immunol Methods. 1994;174:21-23.

[81] Lu X, Richardson PM. Inflammation near the nerve cell body enhances axonal regeneration. J Neurosci. 1991;11:972-978.

[82] Perry VH, Brown M. Macrophages and nerve regeneration. Curr Opin Neurobiol. 1992:2:679-682.

[83] Casmmer W, Bloom B, Norton W, Gordon S. Degradation of basic protein in myelin by neutral proteases secreted by stimulated macrophages: a possible mechanism of inflammatory demyelination. Proc Natl Acad Sci USA. 1978;75:1554-1558.

[84] Colton CA, Gilbert DL. Production of superoxide anions by a CNS macrophage, the microglia. FEBS Lett. 1987;223:284-288.

[85] Giulian D, Robertson C. Inhibition of mononuclear phagocytes reduces ischemic injury in the spinal cord. Ann Neurol. 1990;27:33-42.

[86] Griot C, Burge T, Vandevelde M, Peterhans E. Antibody-induced generation of reactive oxygen radicals by brain macrophages in canine distemper encephalitis: a mechanism for bystander demyelination. Acta Neuropathol (Berl). 1989;78:396-403.

[87] Merrill JE, Ignarro J, Sherman MP, et al. Microglial cytotoxicity of oligodendrocytes is mediated through nitric oxide. J Immunol. 1993;151:2132-2141.

[88] Tonnesen M, Worthen G, Johnston R. Neutrophil emigration, activation, and tissue damage. In: Clark R, Heson P, eds. The Molecular and Cellular Biology of Wound Repair. New York, NY: Plenum Press; 1988:149-184.

[89] Wisniewski H, Blood B. Primary demyelination as a nonspecific consequence of a cell-mediated immune reaction. J Exp Med. 1975;141:346-359.

[90] Rabchevsky AG, Streit WJ. Grafting of cultured microglial cells into the lesioned spinal cord of adult rats enhances neurite outgrowth. J Neurosci Res. 1997;47:34-48.

[91] Prewitt CM, Niesmann IR, Kane CJ, Houle JD. Activated macrophage/microglia cells can promote the regeneration of sensory axons into the injured spinal cord. Exp Neurol. 1997;148:433-433.

[92] Cserr HF, Bundgaard M. The neuronal microenvironment: a comparative view. Ann N Y Acad Sci. 1986;481:1-6.

[93] Reese TS, Karnovsky MJ. Fine structural localization of a blood-brain barrier to exogenous peroxdiase. J Cell Biol. 1967;34:207-217.

[94] Noble LJ, Mautes AE, Hall JJ. Characterization of the microvascular glycocalyx in normal and injured spinal cord in the rat. J Comp Neurol. 1996;376:542-556.

[95] Yurchenko P, Schittney J. Molecular architecture of the basement membrane. FASEB J. 1990;4:1577-1590.

[96] Popovich PG, Horner PJ, Mullin BB, Stokes BT. A quantitative spatial analysis of the blood-spinal cord barrier, I: permeability changes after experimental spinal contusion injury. Exp Neurol. 1996;142:258-275.

[97] Runge VM, Wells JW, Baldwin SA, et al. Evaluation of the temporal evolution of acute spinal cord injury. Invest Radiol. 1997;32:105-110.

[98] Westergaard E. The blood-brain barrier to horseradish peroxidase under normal and experimental conditions. Acta Neuropathol (Berl). 1977;39:181-187.

[99] Fukada K, Panter SS, Sharp F, Noble LJ. Induction of heme oxygenase-1 (HO-1) after traumatic brain injury in the rat. Neurosci Lett. 1995;199:127-130.

[100] Richmon J, Fukada K, Sharp F, Noble LJ. Induction of HSP-70 after hyperosmotic opening of the blood-brain barrier in the rat. Neurosci Lett. 1995;202:1-4.

[101] Gonzalez MF, Shiraishi K, Hisanaga K, et al. Heat shock proteins as markers of neural injury. Mol Brain Res. 1989;6:93-100.

[102] Ferrante RJ, Browne SE, Shinobu LA, et al. Evidence of increased oxidative damage in both sporadic and familial amyotrophic lateral sclerosis. J Neurochem. 1997;69:2064-2074.

[103] Schipper HM, Libermann A, Stopa EG. Neural heme oxygenase-1 expression in idiopathic Parkinson's disease. Exp Neurol. 1998;50:60-68.

[104] Westmark R, Noble LJ, Fukuda K, et al. Intrathecal administration of endothelin-1 in the rat: impact on spinal cord blood flow and the blood spinal cord barrier. Neurosci Lett. 1995;192:173-176.

[105] Yanagisawa M, Kurihara H, Kimura S, et al. A novel potent vasoconstrictor peptide produced by vascular endothelial cells. Nature. 1988;332:411-415.

[106] Stjernquist M. Endothelins: vasoactive peptides and growth factors. Cell Tissue Res. 1998;292:1-9.

[107] Salzman SK, Acosta R, Beck G, et al. Spinal endothelin content is elevated after moderate local trauma in the rat to levels associated with locomotor dysfunction after intrathecal injection. J Neurotrauma. 1996;13:93-101.

[108] McKenzie AL, Hall JJ, Aihara N, et al. Immunolocalization of endothelin in the traumatized spinal cord: relationship to blood-spinal cord barrier breakdown. J Neurotrauma. 1995; 12:257-268.

[109] Zimmermann M, Seifert V. Endothelin and subarachnoid hemorrhage: an overview. Neurosurgery. 1998;43:863-875.

[110] Macrae IM, Robinson MJ, Graham DI, et al. Endothelin-1-induced reduction in cerebral blood flow: dose dependency, time course, and neuropathological consequences. J Cereb Blood Flow Metab. 1993;13:276-284.

[111] Fujimori A, Yanagisawa M, Saito A, et al. Endothelin in plasma and cerebrospinal fluid of patients with subarachnoid hemorrhage [letter]. Lancet. 1990;336(8715):663.

[112] Tippler B, Herbst C, Simmet T. Evidence for the formation of endothelin by lysed blood cells from endogenous precursor. Eur J Pharmacol. 1994;271:131-139.

[113] Armstead WM. Role of endothelin in pial artery vasoconstriction and altered responses to vasopressin after brain injury. J Neurosurg. 1996;85:901-907.

[114] Maeda A, Sobel RA. Matrix metalloproteinases in the normal human central nervous system, microglial nodules, and multiple sclerosis lesions. J Neuropath Exp Neurol. 1996;55:300-309.

[115] Pagenstecher A, Stalder AK, Kincaid CL, et al. Differential expression of matrix metalloproteinase and tissue inhibitor of matrix metalloproteinase genes in the mouse central nervous system in normal and inflammatory states. Am J Pathol. 1998;152:729-741.

[116] Biddison WE, Taub DD, Cruikshank WW, et al. Chemokine and matrix metalloproteinase secretion by myelin proteolipid protein-specific CD8+ T cells: potential roles in inflammation. J Immunol. 1997;158:3046-3053.

[117] Delclaux C, Delacourt C, D'Ortho MP, et al. Role of gelatinase B and elastase in human polymorphonuclear neutrophil migration across basement membrane. Am J Respir Cell Mol Biol. 1996;14:288-295.

[118] Di Girolamo N, Tedla N, Lloyd A, Wakefield D. Expression of matrix metalloproteinases by human plasma cells and B lymphocytes. Eur J Immunol. 1998;28:1773-1784.

[119] Gronski TJ Jr, Martin RL, Kobayashi DK, et al. Hydrolysis of a broad spectrum of extracellular matrix proteins by human macrophage elastase. J Biol Chem. 1997;272:12189-12194.

[120] Romanic AM, White RF, Arleth AJ, et al. Matrix metalloproteinase expression increases after cerebral focal ischemia in rats: inhibition of matrix metalloproteinase-9 reduces infarct size. Stroke. 1998;29: 1020-1030.

[121] Trocme C, Gaudin P, Berthier S, et al. Human B lymphocytes synthesize the 92-kDa gelatinase, matrix metalloproteinase-9. J Biol Chem. 1988;273:20677-20684.

[122] Vu TH, Werb Z. Gelatinase B: structure, regulation, and function. In: Parks WC, Mecham RP, eds. Matrix Metalloproteinases. San Diego, Calif: Academic Press Inc; 1998:115-148.

[123] Rosenberg GA. Matrix metalloproteinases in brain injury. J Neurotrauma. 1995;12:833-842.

[124] Rosenberg GA, Estrada EY, Dencoff JE, Stetler-Stevenson WG. Tumor necrosis factor-alpha-induced gelatinase B causes delayed opening of the blood-brain barrier: an expanded therapeutic window. Brain Res. 1995;703:151-155.

[125] Rosenberg GA, Kornfeld M, Estrada EY, et al. TIMP-2 reduces proteolytic opening of blood-brain barrier by type IV collagenase. Brain Res. 1992;576:203-207.

[126] Hiraoka N, Allen E, Apel IJ, et al. Matrix metalloproteinases regulate neovascularization by acting as pericellular fibrinolysins. Cell. 1998;95:365-377.

[127] Dvorak HF, Detmar M, Claffey KP, et al. Vascular permeability factor/vascular endothelial growth factor: an important mediator of angiogenesis in malignancy and inflammation. Int Arch Allergy Immunol. 1995;107:233-235.

[128] Risau W, Esser S, Engelhardt B. Differentiation of blood-brain barrier endothelial cells. Pathol Biol (Paris). 1998;46:171-175.

[129] Senger DR. Molecular framework for angiogenesis: a complex web of interactions between extravasated plasma proteins and endothelial cell proteins induced by angiogenic cytokines [comment]. Am J Pathol. 1996;149:1-7.

[130] Iwasaka C, Tanaka K, Abe M, Sato Y. Ets-1 regulates angiogenesis by inducing the expression of urokinase-type plasminogen activator and matrix metalloproteinase-1 and the migration of vascular endothelial cells. J Cell Physiol. 1996;169:522-531.

[131] Mignatti P, Rifkin DB. Plasminogen activators and matrix metalloproteinases in angiogenesis. Enzyme Protein. 1996;49:117-137.

[132] Moses MA. The regulation of neovascularization of matrix metalloproteinases and their inhibitors. Stem Cells. 1997;15:180-189.

[133] Zucker S, Conner C, DiMassmo BI, et al. Thrombin induces the activation of progelatinase A in vascular endothelial cells: physiologic regulation of angiogenesis. J Biol Chem. 1995;270:23730-23738.

[134] Dias S, Boyd R, Balkwill F. IL-12 regulates VEGF and MMPs in a murine breast cancer model. Int J Cancer. 1998;78:361-365.

[135] Tolnay E, Wiethege T, Kuhnen C, et al. Expression of type IV collagenase correlates with the expression of vascular endothelial growth factor in primary non-small cell lung cancer. J Cancer Res Clin Oncol. 1997;123:652-658.

[136] Wang H, Keiser JA. Vascular endothelial growth factor upregulates the expression of matrix metalloproteinases in vascular smooth muscle cells: role of flt-1. Circ Res. 1998;83:832-840.

[137] Zhang Z, Guth L. Experimental spinal cord injury: Wallerian degeneration in the dorsal column is followed by revascularization, glial proliferation, and nerve regeneration. Exp Neurol. 1997;147: 159-171.

[138] Wamil AW, Wamil BD, Hellerqvist CG. CM101-mediated recovery of walking ability in adult mice paralyzed by spinal cord injury. Proc Natl Acad Sci USA. 1998;95:13188-13193.

AEM Mautes, PhD, Head, Neurosurgical Research Laboratory, Saarland University Medical School, Homburg, Germany.

MR Weinzierl, MD, Visiting Research Fellow, Department of Neurosurgery, University of California, San Francisco, Calif, and Medical Faculty, Department of Neurosurgery, University of Technology, Aachen, Germany.

F Donovan, PhD, Postdoctoral Fellow, Department of Anatomy, University of California, San Francisco, Calif.

LJ Noble, PhD, Professor, Department of Neurosurgery, University of California, C224, 521 Parnassus Ave, San Francisco, CA 94143-0520 (USA) ([e-mail]). Address all correspondence to Dr Noble.

All authors contributed to writing. Dr Jim Velier assisted in the production of the figures.

This research was supported by grant NS23324 (LJN) and BMBF grant 01K09405/4 (AEM).