Tissue Plasminogen Activator–Mediated Fibrinolysis Protects against Axonal Degeneration and Demyelination after Sciatic Nerve Injury

The most common reaction of the nervous system after disconnection of axons from the neuronal body is axon lysis and myelin sheath fragmentation. These processes contribute to functional impairment of the nerve. Axonal and myelin damage are observed in a variety of nervous system diseases with diverse etiologies, such as stroke (Sawlani et al. 1997), spongiform encephalopathies (Liberski and Gajdusek 1997), Guillan-Barre syndrome (Trojaborg 1998), insulin-dependent diabetic neuropathy (Said et al. 1992), and multiple sclerosis (Trapp et al. 1998; Arnold 1999). In these diseases, axonal degeneration is considered to be responsible for permanent disability. It has been proposed that proteases such as plasminogen activators (Reich 1978; Tsirka et al. 1995; Seeds et al. 1997) and matrix metalloproteinases (LaFleur et al. 1996) may play a role in nervous system damage.

Plasminogen activators (PAs) are serine proteases, which convert plasminogen to plasmin. Plasmin is a potent serine protease that can degrade a variety of proteins, but its primary substrate in vivo is fibrin, the proteinaceous component of blood clots (Bugge et al. 1996). There are two PAs identified in mammals: tissue-type and urokinase-type. The PA/plasmin system can have deleterious effects in pathological settings via proteolysis of tissue proteins. For example, plasmin-catalyzed degradation of laminin promotes excitotoxic-mediated neurodegeneration in the hippocampus (Chen and Strickland 1997), and plasmin can also degrade myelin basic protein (MBP) in vitro (Cammer et al. 1978; Norton et al. 1990). Conversely, the PA/plasmin system can be beneficial via its fibrinolytic activity, since PA/plasmin-mediated proteolysis plays a critical role in fibrin(ogen) clearance (Bugge et al. 1996).

In experimental nerve injury (Bignami et al. 1982; Sallés et al. 1990), as well as in human neuroinflammatory diseases such as multiple sclerosis, PAs and their inhibitors can be upregulated (Akenami et al. 1997, Akenami et al. 1999; Cuzner and Opdenakker 1999). However, it is not clear if this altered expression is deleterious or a beneficial response of the tissue to the disease. In the current study, we sought to investigate how PAs affect degeneration severity in the peripheral nervous system (PNS) in vivo. To this end, we performed sciatic nerve injury experiments in mice genetically deficient for tPA (Carmeliet et al. 1994), plasminogen (Bugge et al. 1995; Ploplis et al. 1995), and fibrinogen (Suh et al. 1995). Nerve crush in the PNS is a reliable experimental model that allows the study of axonal degeneration (Bridge et al. 1994). Our analysis revealed induction of tPA expression after nerve injury. Axonal and myelin loss were exacerbated in tPA- and plasminogen-deficient mice, whereas fibrinogen depletion was protective. Fibrin deposition was also correlated with nerve damage. These results demonstrate that tPA/plasmin-catalyzed fibrin degradation can protect neurons against axonal injury at least in the PNS.

tPA-deficient mice (tPA^−/−) (Carmeliet et al. 1994) have been backcrossed to C57Bl/6J for nine generations. Urokinase plasminogen activator (uPA)-deficient mice (uPA^−/−) (Carmeliet et al. 1994) also have been backcrossed to the C57Bl/6J background. C57Bl/6 mice were used as wild-type controls. Plg-deficient mice (plg^−/−) (Bugge et al. 1995), which are heterozygous for fib deficiency, fibrinogen-deficient mice (fib^−/−) (Suh et al. 1995), which are heterozygous for plg deficiency; and combined plg-, fibrinogen-deficient mice (plg^−/−fib^−/−) were of mixed genetic background. Heterozygous plg^+/−fib^+/− littermates were employed as controls in all studies. No differences were observed between C57Bl/6J mice and heterozygous plg^+/−fib^+/− mice in the crush injury model. All mice were 8–15-wk old at the start of the experiment. The genotype of all mice was confirmed at the end of the experiments by PCR analysis of genomic DNA extracted from mice tails. All experiments were performed in compliance with institutional guidelines.

Adult mice were anesthetized by intraperitoneal injection of 2.5% avertin (0.18 ml/10 g body weight; Aldrich Chemical Co.). For the crush injury, sciatic nerves were crushed midthigh three times (20 s each) with watchmaker's forceps, the tips of which had been previously cooled in liquid nitrogen (Funakoshi et al. 1998). The crush site was marked with India ink (LaFleur et al. 1996). Skin incisions were closed with sutures.

For in situ zymographies (Sappino et al. 1993), 10-μm unfixed, cryostat sciatic nerve sections were prepared. Sections were covered by a casein overlay containing 1% low melting point agarose and 25 μg/ml plasminogen, and covered with a coverslip. Plasminogen was prepared from human plasma (Deutsch and Mertz 1970). Control experiments were performed with overlay mixtures lacking plasminogen. To distinguish uPA from tPA activity, amiloride (Sigma Chemical Co.), a specific inhibitor of uPA enzymatic activity (Vassalli and Belin 1987), was added in the mixture at 1 mM final concentration. To discriminate between tPA and other fibrinolytic enzymes, tPA-STOP (2,7-bis-(4-amidinobenzylidene)-cycloheptan-1-one dihydrochloride salt; American Diagnostica) at a concentration of 50 μg/ml was used. Sections were incubated at room temperature in a humidified chamber. Conversion of plasminogen into plasmin, which in turn lysed the insoluble casein, resulted in the appearance of lytic zones within 6 h. Zones of plasmin-dependent caseinolysis appeared as black areas when photographed under darkfield illumination.

Dissected nerves were embedded in Tissue-Tek OCT, then immediately frozen on dry ice and stored at −70°C until use. Sections were cut longitudinally on a motor-driven Leica cryostat with a retraction microtome and a steel knife at a cabinet temperature of −20°C.

Immunohistochemical staining was performed on cryostat sections (Akassoglou et al. 1997, Akassoglou et al. 1998). Primary antibodies were as follows: rabbit anti-human tPA (1:1,200; Waller and Schleuning 1985), goat anti–human fibrin(ogen) (1:500; Chemicon International, Inc.), rabbit anti–cow GFAP (1:200; Dako), rabbit anti–mouse laminin (1:1,000; Sigma Chemical Co.), rabbit anti–mouse fibronectin (1:200; Chemicon International, Inc.), rat anti–mouse VCAM-1 (Chemicon International, Inc., 1:50), rabbit anti–human MBP (1:200; DAKO), sheep anti–mouse PA inhibitor 1 (PAI-1) (1:200; American Diagnostica), and rabbit antineuroserpin (1:2,000; provided by D. Lawrence, Department of Vascular Biology, American Red Cross Holland Laboratory, Rockville, MD). Bound antibodies were visualized using either the avidin–biotin-peroxidase complex (Vectastain Elite ABC kit; Vector Laboratories) and 3-amino-9-ethylcarbazole (Sigma Chemical Co.) as a chromogen, or the avidin–biotin-alkaline phosphatase complex (Vectastain Elite ABC kit; Vector Laboratories), using nitroblue tetrazolium/brom-chlor-indolyl phosphate (Vector Laboratories). Staining specificity for the tPA and fibrin(ogen) antibodies was confirmed using tissue from tPA^−/− and fib^−/− mice, respectively. Staining specificity for GFAP antibody was confirmed using rabbit IgG. Incubation without the primary antibody served as a negative control. Histochemical Oil Red O staining, which stains lipids and reveals demyelinated areas (Akassoglou et al. 1997), was performed on cryostat sections. Immunoblot for neuroserpin was performed on sciatic nerve extracts as described in Hastings et al., 1997.

8 d after injury, animals were intracardially perfused under deep anesthesia with ice-cold 2% paraformaldehyde, and 0.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.2, for 1 min followed by ice-cold 3% glutaraldehyde in 0.1 M phosphate buffer for 5 min. The injured sciatic nerve was removed, and the region ∼5 mm below the injury site (crush site marked with India ink) was isolated. The noninjured sciatic nerve (contralateral) served as a control. Nerves were immersion fixed for 24 h in phosphate-buffered 3% glutaraldehyde, postfixed in 2% osmium tetroxide solution, and embedded in resin. Semi-thin sections were cut on an Ultra-cut microtome and stained with toluidine blue. A morphometric grid (100 mm²) was adapted to the microscope, and a minimum of three grids per sample of myelinated axons was counted.

Mice were depleted of fibrinogen as described in Busso et al. 1998. In brief, mice were anesthetized, their backs shaved, and 14-d mini-osmotic pumps (model 2002; Alza Corp.) filled with a solution of 250 U ancrod per ml of 0.1 M NaCl and 0.1 M Tris-HCl, pH 7.5 (Sigma Chemical Co.) were implanted subcutaneously into their backs (one mini-pump per animal). The insertion sites were closed by sutures. The pumps deliver 0.5 μl/h, so the mice received 3 U ancrod/d. In control animals, buffer-filled mini-pumps were implanted. On day 3 of ancrod infusion, sciatic nerve crush or transection was performed. 8 d after the crush, mice were killed, and tissues were prepared for cryostat or semi-thin sections.

8 d after injury, animals were killed, the skeletal muscles from the injured (ipsilateral) and the control (contralateral) site were isolated, and their net weight was calculated. The percentage of atrophy was determined for each animal by subtracting the weight of the muscle from the injured site from the weight of the muscle of the control site. The difference was multiplied by 100 and divided by the muscle weight of the control site.

Plasminogen activator activity (Bignami et al. 1982) and tPA mRNA (Yuguchi et al. 1997) and protein (Neuberger and Cornbrooks 1989) expression are increased in the PNS after injury. To address the major type of PA induced after injury, we performed in situ zymography on sciatic nerves of wild-type, tPA^−/−, and uPA^−/− mice. When compared with control uninjured nerve (Fig. 1 A), wild-type sciatic nerve 8 d after crush showed an induction of proteolytic activity (Fig. 1 B, dark zone around nerve). Wild-type control or injured sciatic nerves incubated with an overlay that lacked plasminogen did not show any proteolytic activity (data not shown), indicating that the activity was plasminogen-dependent. tPA^−/−-injured sciatic nerve did not show appreciable proteolytic activity (Fig. 1 C), whereas uPA^−/−-injured sciatic nerve showed an induction of proteolytic activity (Fig. 1 D) similar to the injured wild-type control (Fig. 1 B). Addition of tPA-STOP, a specific tPA inhibitor, blocked the activity (Fig. 1 E), whereas addition of amiloride, a specific inhibitor of uPA proteolytic activity, had no effect (Fig. 1 F). After 24 h of assay, a lytic zone started to appear in wild-type sciatic nerves in the presence of tPA-STOP as well as in the tPA^−/− crushed sciatic nerves (data not shown), suggesting the presence of low uPA activity. These results show that the proteolytic activity induced after sciatic nerve injury is mainly due to tPA.

Tissue plasminogen activator has been shown to be produced by Schwann cells and the growth cones of peripheral neurons in culture (Krystosek and Seeds 1984; Clark et al. 1991). To identify the cellular localization of tPA protein in vivo, sciatic nerves before and after crush were stained with an antibody against tPA. When compared with the uninjured sciatic nerve (Fig. 2 A), immunoreactivity specific for tPA was increased as early as 2 d after the crush (Fig. 2 B). The lack of staining in the tPA^−/− sciatic nerve with the anti-tPA antibody (Fig. 2 C) showed that the antibody was specific. Staining of parallel sections showed that the tPA immunoreactive cells (Fig. 2 D) had similar morphology to cells positive for GFAP (Fig. 2 E), a Schwann cell marker. In contrast, immunostaining with an antibody against Mac-1, a macrophage-specific marker, showed a different staining pattern than that for tPA (data not shown). These results suggest Schwann cells as a major source of tPA in the sciatic nerve after injury.

To examine the possible role of endogenous inhibitors of tPA, we analyzed the expression of PAI-1 and neuroserpin in the sciatic nerve before and after injury. Immunostaining for PAI-1 revealed that there was no PAI-1 expression before or after injury (data not shown). Neuroserpin, a brain-associated inhibitor of tPA (Osterwalder et al. 1996; Hastings et al. 1997), is constitutively expressed in the sciatic nerve (Lawrence, D., personal communication). Western blot analysis and immunohistochemistry on wild-type and tPA^−/− sciatic nerve extracts did not reveal any significant increase of neuroserpin after injury (data not shown). Overall, these data suggest that elevated tPA activity is due primarily to increased expression of the protein, and not to downregulation of protease inhibitors.

Since tPA is the major PA produced after sciatic nerve injury, we compared lesion formation after sciatic nerve crush in wild-type and tPA^−/− mice. Toluidine blue staining for myelin in semi-thin cross-sections of crushed nerves after 8 d showed a dramatic decrease in myelinated axons in tPA^−/− mice (Fig. 3 D), when compared with wild-type mice (Fig. 3 C). Similarly, light microscopy analysis of the lesion at 8 d with Oil Red O, which stains the lipids associated with the myelin sheath, revealed increased myelin debris in tPA^−/− mice (Fig. 3 F, aggregates of lipid), when compared with wild-type controls (Fig. 3 E). To examine whether the difference in demyelinating damage persisted at later times, we analyzed sciatic nerves 22 d after crush. Since at later times the myelin and axonal debris has been removed, Oil Red O staining is not informative. Therefore, tissue morphology was examined with an antibody against MBP, which stains myelinated axons. In accordance with our observation 8 d after the crush, tPA^−/− mice had decreased immunoreactivity for MBP (Fig. 3 H), when compared with wild-type controls (Fig. 3 G). This result shows that, at later times after crush injury, there are fewer myelinated axons in tPA^−/− mice than in wild-type mice.

To quantitate the effect of tPA on axonal demyelination, we counted myelinated axons in semi-thin sections. This quantification showed (myelinated axons/0.1 mm²) the following: 10.3 ± 2.4 in control, and 2.1 ± 0.3 in tPA^−/− (Fig. 4 D). The decrease in the number of myelinated axons in tPA^−/− mice compared with control mice was statistically significant (P < 0.01).

To assess whether the increase in axonal degeneration in the absence of tPA was due to its proteolytic function, we performed crush injury in mice genetically deficient for plasminogen, which is the primary substrate for tPA. Plg^−/− mice (Fig. 4 A) showed a similar reduction in myelinated axons as tPA^−/− mice (Fig. 3 F). Quantification of myelinated axons/0.1 mm² showed 2.8 ± 0.7 in plg^−/− (Fig. 4 D). The decrease in myelinated axons in plg^−/− mice compared with control mice was statistically significant (P < 0.02), and there was no significant difference between tPA and plg^−/− mice. These results indicate that tPA reduces axonal loss and demyelination in the PNS primarily through its proteolytic effect on plasminogen. These results do not exclude a subtle nonproteolytic effect of tPA, as has been observed in other systems (Kim et al. 1999; Rogove et al. 1999).

Removal of fibrinogen rescues mice from most of the effects of plasminogen deficiency (Bugge et al. 1996). However, the resistance of plasminogen-deficient mice to excitotoxic neuronal degeneration in the hippocampus is not affected by the removal of fibrinogen (Tsirka et al. 1997a). To determine if fibrin(ogen) was playing a role in inflammatory neuronal degeneration, we performed nerve crush in mice with fibrinogen deficiency (fib^−/−) or a double deficiency for plasminogen and fibrinogen. Fib^−/− mice were similar to wild-type mice in myelinated axons (Fig. 4B and Fig. D). The decrease in myelinated axons observed in plg^−/− mice (Fig. 4 A) was alleviated by genetically superimposing fibrinogen deficiency (plg^−/−fib^−/−; Fig. 4C and Fig. D). These results indicate that tPA/plasmin-mediated degradation of fibrin(ogen) protects axons from degeneration and demyelination. Quantification of myelinated axons/0.1 mm² showed the following: 12.0 ± 2.8 in fib^−/−, and 15.2 ± 4.0 in plg^−/−fib^−/− (Fig. 4 D). The increase in the number of myelinated axons in the plg^−/−fib^−/− mice compared with the plg^−/− mice was statistically significant (P < 0.03).

To address the involvement of fibrin(ogen) in axonal degeneration and myelin loss, we performed immunocytochemistry with an antibody against fibrin(ogen). A partial nerve crush at the sciatic nerve revealed that the crushed part of the nerve, which underwent degeneration (Fig. 5 A), had extensive deposition of fibrin(ogen), whereas the immediately adjacent, uninjured region was free of fibrin(ogen) (Fig. 5 B). Absence of staining of a fib^−/− sciatic nerve documented the specificity of the antibody against fibrin(ogen) (not shown). This staining indicates that fibrin(ogen) deposition is spatially correlated with degeneration.

We also examined other extracellular matrix (ECM) proteins for their involvement in sciatic nerve injury. In the mouse hippocampus, tPA-mediated degradation of laminin after injury is implicated in neurodegeneration (Chen and Strickland 1997). To examine the expression pattern of laminin in sciatic nerve injury, we stained sciatic nerves before and after crush with an antibody against laminin. Laminin was expressed in the normal and the tPA^−/− sciatic nerve, and this expression was not significantly altered after injury (data not shown). Another ECM protein, fibronectin, was equally expressed at low levels in wild-type and tPA^−/− sciatic nerves, and not altered by injury. To examine if differences in adhesion molecules could contribute to the exacerbated phenotype observed in tPA^−/− mice, we stained wild-type and tPA^−/− mice for VCAM-1, which is upregulated after injury (Castano et al. 1996). Wild-type and tPA^−/− mice showed similar upregulation of VCAM-1 after crush. These results suggest that laminin, fibronectin, and VCAM-1 do not contribute to the altered peripheral nerve phenotype of tPA^−/− mice.

Administration of ancrod, a Malayan pit viper (Calloselasma rhodostoma) venom protein, leads to consumption of systemic fibrinogen and drastically reduces plasma fibrinogen levels (Bell et al. 1978). Therefore, under these conditions, excessive fibrin deposition should be diminished, which might result in less nerve damage. Ancrod (3 U/d) was administered to tPA^−/− mice for 3 d before crush, and then throughout the experimental period, without any effects on survival. tPA^−/− mice treated with ancrod showed dramatically reduced fibrin(ogen) immunoreactivity (Fig. 6 B) when compared with untreated tPA mice (Fig. 6 A).

Histopathological analysis revealed that depletion of fibrin(ogen) rescued tPA^−/− mice from exacerbated axonal degeneration. tPA^−/− mice treated with ancrod showed more myelinated axons (Fig. 6 D) when compared with untreated tPA^−/− mice (Fig. 6 C). Quantification of myelinated axons/0.1 mm² showed the following: 2.1 ± 0.3 in tPA^−/−, and 5.9 ± 0.6 in tPA^−/− treated with ancrod (Fig. 6 E). The increase in myelinated axons in the tPA^−/− mice treated with ancrod compared with the untreated tPA^−/− mice was statistically significant (P < 0.01).

Although we detected no morphological differences in the uninjured sciatic nerves of the various mouse genotypes, the possibility existed that subtle anatomical changes due to gene disruption might be playing a role in the phenotype. The ancrod experiment is important in eliminating this possibility. Since the tPA^−/− mice used were highly inbred and, therefore, genetically identical, the reversal of the phenotype by fibrinogen depletion indicates that acute fibrin deposition is responsible for exacerbating the nerve damage in the tPA^−/− mice.

To assess a possible functional consequence of fibrin(ogen) depletion, we examined skeletal muscle weight decay after nerve crush. The gastrocnemius muscle atrophies after sciatic nerve injury due to muscle denervation and the percentage of atrophy corresponds to the degree of nerve damage (Funakoshi et al. 1998). Therefore, a direct measure of functional impairment can be obtained by assessing muscle weight (Fig. 7). Muscle mass of wild-type mice dropped 24.5 ± 2.7% 8 d after the sciatic nerve crush compared with the unlesioned, contralateral side. Exacerbation of muscle mass atrophy in the tPA^−/− mice 8 d after crush (40.1 ± 3.8%; P < 0.005 compared with wild-type mice) indicated increased axonal damage in the absence of tPA. 22 d after crush, muscle atrophy was 81% for a tPA^−/− mouse, whereas for a wild-type mouse it was 66%. This preliminary evidence suggests that increased muscle atrophy persists at later time points after injury. After treatment with ancrod, there was less muscle atrophy in tPA^−/− mice (16.6 ± 3.4%). The decrease in muscle mass atrophy in the ancrod-treated tPA^−/− mice compared with the untreated tPA^−/− was statistically significant (P < 0.001). This observation agrees with the histopathological observation that depletion of fibrin(ogen) protects tPA^−/− mice from exacerbated axonal damage. Overall, these results suggest that tPA-mediated fibrinolysis protects motor axons after nerve injury.

The studies reported here demonstrate that upregulation of tPA and plasmin is beneficial in reducing axonal damage; fibrin(ogen) deposition correlates with axonal degeneration; and depletion of fibrin(ogen) rescues mice from exacerbated axonal damage. These results indicate that plasmin-mediated removal of fibrin(ogen) deposits is critical for limiting axonal degeneration and demyelination.

tPA^−/− (Tsirka et al. 1995) and plg^−/− (Tsirka et al. 1997b) mice are resistant to excitotoxin-induced neuronal death in the hippocampus, indicating that a tPA/plasmin proteolytic cascade promotes neuronal degeneration under these conditions. In this model of neuronal death, fib deficiency does not affect the neurodegeneration-resistant phenotype of the plg^−/− mice (Tsirka et al. 1997a), demonstrating that fibrin is not the primary substrate for plasmin in the hippocampus. Further experiments indicated that a major plasmin substrate in the hippocampus is laminin (Chen and Strickland 1997). In contrast, after sciatic nerve crush, exacerbation of axonal degeneration and demyelination observed in tPA^−/− and plg^−/− mice is rescued in the absence of fibrinogen, showing that fibrin(ogen) is the primary plasmin substrate during axonal degeneration and demyelination under inflammatory conditions.

A difference between excitotoxic neuronal death and inflammatory axonal degeneration is the participation of serum proteins in lesion formation. After excitotoxin injection, blood-brain barrier breakdown does not appear to contribute to excitotoxic neuronal damage (Chen et al. 1999). In contrast, under inflammatory conditions, disruption of the blood–brain barrier and fibrin deposition are early expressions of immune effector activity and precede clinical signs and axonal degeneration (Bush et al. 1993). Neuronal death in the PNS is also often associated with enhanced vascular permeability of the blood–nerve barrier (Koh et al. 1993). A preliminary examination of blood–nerve barrier disruption by IgG immunostaining revealed no difference between wild-type and tPA^−/− mice (data not shown).

The present findings, coupled with previous results (Wang et al. 1998; Nagai et al. 1999; Kilic et al. 1999), reveal that a specific proteolytic system (e.g., tPA/plasmin) can be either deleterious or beneficial after different challenges to the nervous system. The significant distinction as to whether plasmin-mediated proteolysis is beneficial or a liability may be the component(s) of the ECM targeted. More specifically, plasmin degradation of fibrin may be advantageous, whereas excessive plasmin-mediated degradation of other matrix components may be associated with pathologic manifestations.

How does fibrin(ogen) deposition contribute to axonal damage? Fibrin(ogen) is necessary for various inflammatory responses in vivo (McRitchie et al. 1991; Tang and Eaton 1993), and deficiency in fibrinolysis can contribute to pathogenic processes (Idell et al. 1989; Bugge et al. 1996; Kitching et al. 1997; Busso et al. 1998; Drew et al. 1998). Fibrin deposition also has been reported in multiple sclerosis, where it has been shown to correlate with demyelinated plaques (Claudio et al. 1995).

Fibrin is deposited after injury in both wild-type and tPA^−/− mice. Fibrin depletion improves the degeneration phenotype of tPA^−/− or plasminogen^−/− mice, but not wild-type mice. Further experiments may reveal subtle quantitative or spatial aspects of fibrin deposition that can explain why tPA^−/− mice have increased damage.

Persistent fibrin(ogen) deposition might affect axonal degeneration by influencing inflammation, with consequent aggravation of the injury. Local matrices may drive the inflammatory response by providing a temporary scaffold for inflammatory cell adhesion and migration and by increasing chemotaxis at sites of inflammation (Tang and Eaton 1993). However, analysis of the inflammatory profile of the sciatic nerve lesions did not reveal any major differences in the macrophage or T cell number between wild-type, tPA^−/−, and systematically defibrinogenated wild-type mice 8 d after injury (Akassoglou, K., and S. Strickland, unpublished observations).

The plasmin/fibrin axis could also contribute to inflammation by upregulation of proinflammatory cytokines. Cytokines regulate the balance between proteases and their inhibitors, and proteolysis can modulate cytokine activity (Cuzner and Opdenakker 1999). Interestingly, fibrin upregulates interleukin 1 (Perez and Roman 1995) and tumor necrosis factor (Formica et al. 1994), two pro-inflammatory cytokines which are involved in inflammatory axonal damage and demyelination in both the peripheral (Creange et al. 1997) and central nervous system (Akassoglou et al. 1998; Wiemann et al. 1998). A more detailed analysis of the inflammatory response and cytokine expression after injury may reveal an immunoregulatory role for fibrin(ogen) in axonal damage and demyelination.

Our results suggest that tPA/plasmin-mediated fibrinolysis reduces axonal damage and, therefore, components of this pathway might be used as a new approach to protect axons from injury. In situations of chronic demyelination, a large percentage of the damage might occur via a fibrin-mediated pathway. In this case, depletion of fibrin(ogen) or prevention of fibrin(ogen) deposition might protect against demyelination. This approach would be especially valuable if it could be accomplished locally at the site of inflammation. Thus, the identification of fibrin(ogen) deposition as a new contributing factor to axonal degeneration may yield additional strategies to prevent damage in the nervous system.

We are very grateful to Fernando J. Sallés, Stella Tsirka, Zu-Lin Chen (all from Department of Pharmacology, State University of New York at Stony Brook, Stony Brook, NY), Nathalie Busso, and Alex So (Centre Hospitalier Universitaire Vaudois, Laboratoire de Rhumatologie, Lausanne, Switzerland) for stimulating discussion. We thank Zu-Lin Chen and Justin Indyk (Department of Pharmacology, State University of New York at Stony Brook, Stony Brook, NY) for preparation of human plasminogen, Daniel Lawrence for neuroserpin antibody, and Greg Rudomen (University Microscopy and Imaging Center, State University of New York at Stony Brook, Stony Brook, NY) for technical assistance in the preparation of the semi-thin sections.

This work was supported by National Institutes of Health grants to S. Strickland and a Human Frontier Science Program postdoctoral fellowship to K. Akassoglou.