Injury-induced gelatinase and thrombin-like activities in regenerating and nonregenerating nervous systems IGOR FRIEDMANN, ANAT FABER-ELMAN, ETI YOLES, AND MICHAL SCHWARTZ1 Department of Neurobiology, The Weizmann Institute of Science, 76100 Rehovot, Israel ABSTRACT It is now widely accepted that injured nerves, like any other injured tissue, need assistance from their extracellular milieu in order to heal. We compared the postinjury activities of thrombin and gelatinases, two types of proteolytic activities known to be critically involved in tissue healing, in nonregenerative (rat optic nerve) and regenerative (fish optic nerve and rat sciatic nerve) neural tissue. Unlike gelatinases, whose induction pattern was comparable in all three nerves, thrombin-like activity differed clearly between regenerating and nonregenerating nervous systems. Postinjury levels of this latter activity seem to dictate whether it will display beneficial or detrimental effects on the capacity of the tissue for repair. The results of this study further highlight the fact that tissue repair and nerve regeneration are closely linked and that substances that are not unique to the nervous system, but participate in wound healing in general, are also crucial for regeneration or its failure in the nervous system.— Friedmann, I., Faber-Elman, A., Yoles, E., Schwartz, M. Injury-induced gelatinase and thrombin-like activities in regenerating and nonregenerating nervous systems. FASEB J. 13, 533–543 (1999) Key Words: thrombin z matrix metalloproteases z CNS z PNS Neurons in the mammalian central nervous system (CNS)2 are unable to regenerate after axonal injury, whereas both the mammalian peripheral nervous system (PNS) and the CNS of lower vertebrates are capable of regeneration (1). PNS axons fail to elongate more than 1 mm into optic nerve grafts (2). In contrast, neurons from most regions of the mammalian CNS can elongate their axons into PNS grafts, presumably aided by the supportive cellular environment of the peripheral nerve (3). It therefore seems that CNS neurons possess, at least in part, the potential to regenerate, but lack the supportive and permissive environment that characterizes regenerative systems after injury. Our research group recently proposed that the processes of tissue repair and regeneration in the nervous system have features in common with those of any recovering tissue (4, 5). Injury of any tissue triggers a complex cascade of events that modifies the environment and induce a 0892-6638/99/0013-0533/$02.25 © FASEB process of tissue repair. In effecting these changes, molecular components such as proteases play a critical role (6 – 8). Failure of the CNS to regenerate after injury might be the result of impairment of the tissue repair mechanism at one or more stages of the healing cascade, including those involving protease participation. One of the proteases known to influence the nervous system is thrombin, a key enzyme in blood coagulation (9). Thrombin acts as a potent mitogen that modulates the morphology of astrocytes (10, 11) and can induce the secretion of nerve growth factor (NGF) by astrocytes (12). It inactivates acidic fibroblast growth factor, known to promote neuritic outgrowth and astrocyte proliferation (13–15). It inhibits outgrowth from neuritic neuronal cells due to interaction with the thrombin receptor PAR-1 (16, 17), which is proteolytically activated by thrombin (7). Thrombin at high concentrations was shown to induce apoptotic cell death in astrocytes and neurons cultured under normal conditions, whereas at moderate concentrations it protected those cells from a variety of metabolic insults (18 –20). Both effects were shown to be due to the activation of the PAR-1 receptor. The processes of neuroprotection and apoptotic cell death display similar characteristics in the signal transduction pathways (19, 21, 22), and it was suggested that different concentrations of thrombin might result in different activation levels of the same pathway (21). Another group of proteases involved in wound healing is the family of matrix metalloproteases (MMPs, collagenases). Two of these, gelatinase A (MMP-2, 72 kDa) and gelatinase B (MMP-9, 92 kDa), belong to the subfamily of gelatinases. MMPs are largely responsible for the degradation of extracellular matrix (ECM) components such as collagen and pro1 Correspondence: E-mail: bnschwar@weizmann.weizmann.ac.il Abbreviations: CNS, central nervous system; DMEM, Dulbecco’s modified Eagle’s medium; ECM, extracellular matrix; IL, interleukin; MMP, matrix metalloprotease; MMP-2, gelatinase A; MMP-9, gelatinase B; NGF, nerve growth factor; OD, optical density; PA, plasminogen activator; PN-I, glial-derived protease nexin I; PNS, peripheral nervous system; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; TIMPs, tissue inhibitors of matrix metalloprotease; TNF, tumor necrosis factor. 2 533 teoglycans in several normal and pathological processes, including tissue remodeling (23), and MMP genes are among the most abundant of those expressed by cells in inflammatory lesions (24). Remodeling of the ECM by proteolytic activity is known to be crucial for growth-cone motility (8). Whereas thrombin is known to produce the matrix component fibrin (9, 25, 26), MMPs are responsible for matrix degradation. Fibrin fibronectin-containing matrix promotes the growth of cultured septal and hippocampal neurons from newborn rats (27). The MMPs are activated by the fibrin-degrading plasmin, which is itself activated by plasminogen activators (PA) (24). Cultured neurons from the PNS release at least two proteases capable of degrading ECM components, a calcium-dependent MMP and a PA (28). In addition, NGF-induced expression of gelatinase A by dorsal root ganglionic neurons occurs in correlation with the ability to degrade ECM and to extend neurites (29), and NGF production in astrocytes is induced not only by thrombin but also by gelatinases (12). In contrast to the above findings, which point to beneficial effects of MMPs in the nervous system, several studies hint at the involvement of MMPs in processes detrimental to the brain. Intracerebral injection of bacterial collagenase as well as of tumor necrosis factor (TNF) -a induces gelatinase B production, which causes delayed opening of the blood–brain barrier (30, 31). Patients with inflammatory neurological disorders, including multiple sclerosis, show increased gelatinase B in the cerebrospinal fluid (32). Gelatinase B is also present in the cerebrospinal fluid of mice with experimental allergic encephalomyelitis, the animal model of multiple sclerosis, where it cleaves myelin basic protein (33). To investigate the relationship of gelatinase and thrombin activities to regenerative and degenerative processes in the nervous system in vivo, we compared their activities in response to axonal injury in two regenerative nervous systems (rat sciatic nerve and fish optic nerve) with a nonregenerative nervous system (rat optic nerve). We show that the induction of gelatinase activity shortly after injury is comparable in the three nerve types. In contrast, the postinjury induction of thrombin-like activity in the nonregenerative optic nerve differs clearly from that in both regenerative nerves in a way that could explain its detrimental and beneficial effects, respectively, on regeneration. Animals were used according to the regulations formulated by IACUC (Institutional Animal Care and Use Committee). Crush injury Rats were anesthetized with 10 mg/kg xylazine (Vitamed, Israel) and 50 mg/kg ketamine (Fort Dodge Laboratories, Fort Dodge, Iowa). For the optic nerve crush, a lateral canthotomy was performed in the right eye under a binocular operating microscope as described previously (34). The conjunctiva was incised laterally to the cornea, the retractor bulbi muscle was separated, and the optic nerve was exposed. The dura was left intact. For the sciatic nerve crush, a small incision was made in the thigh to expose the nerve. The exposed optic or sciatic nerve was crushed by calibrated forceps for 30 s. After the sciatic nerve crush, the skin was sutured. The rats were allowed to recover. Fish were deeply anesthetized with 0.05% 3-aminobenzoic acid ethyl ester (Sigma, St. Louis, Mo.). As in the rat, the conjunctiva was incised laterally to the cornea and the exposed optic nerve was crushed intraorbitally with forceps for 30 s. The fish were then returned to their tanks. Preparation of conditioned media At specified times after crush injury, the animals were again anesthetized and their crushed nerves were dissected out, cleaned with Kimwipes, and washed in Dulbecco’s modified Eagle’s medium (DMEM) without phenol red. Each of the three types of nerves was pooled and incubated separately for 1.5 h at room temperature in DMEM without phenol red. The nerves were then removed and the three resulting media, now termed ‘conditioned media’, were collected, centrifuged at 15000 3 g for 15 min in an Eppendorf centrifuge at 4°C, and stored at 270°C. Before use, the conditioned media were kept on ice to limit proteolysis. Protein concentrations were determined by Bradford analysis (Bio-Rad, Hercules, Calif.). For some experiments, nerves were separated into proximal and distal segments before incubation. As controls, conditioned media were similarly prepared from uninjured nerves. Preparation of nerve extracts As in preparation of the conditioned media, the animals were reanesthetized at specified times after crush injury. The crushed nerves were dissected from their sheaths and some were separated, as before, into proximal and distal segments. The pooled nerves or nerve segments were then homogenized and left overnight at 4°C in extraction buffer containing 0.5% Triton X-100, 0.1 M Tris-HCl, pH 8.0. After centrifugation at 15,000 3 g for 15 min at 4°C, the protein concentrations of the supernatants were determined by Bradford analysis (Bio-Rad). Nerve extracts were stored at 270°C and prior to the activity assays were kept on ice to limit proteolysis. As controls, nerve extracts were similarly prepared from uninjured nerves. Thrombin activity assay MATERIALS AND METHODS Animals Carp (Cyprinus carpio) were purchased from Tnuva, Israel. Sprague-Dawley rats, 8 wk old, were purchased from the Weizmann Institute of Science Experimental Animal Center. 534 Vol. 13 March 1999 An assay for thrombin activity was designed according to a previously described procedure (35). Samples of the conditioned media, each containing 5 mg of total protein, were incubated at 37°C in a flat-bottomed, 96-well plate with 50 mM Tris-HCl, 5 mM EDTA, pH 8.4, and 150 mM of N-p-TosylGly-Pro-Arg-p-nitroanilidine (Chromozym-TH; Sigma), a specific substrate for thrombin, in DMEM without phenol red in a final volume of 200 ml. In some experiments, the thrombin The FASEB Journal FRIEDMANN ET AL. inhibitor hirudin (Sigma) or thromstop (American Diagnostica, Greenwich, Conn.) was added. Optical density (OD) at 405 nm was measured at the times specified. Wells containing all of the above except for the conditioned media were used as blanks, and signals in the tested samples were expressed as the difference between the measured OD and the OD recorded at the beginning of the assay. Only those measurements obtained when enzymatic activity was in the linear range are shown. In some cases, pure thrombin from bovine plasma (Sigma) was assayed in order to quantify the signals of the assay. Thrombin activation by ecarin Samples of the conditioned media, each containing 5 mg of total protein, were incubated for 2 h at 37°C in a flatbottomed, 96-well plate with 50 mM Tris-HCl, 5 mM EDTA, pH 8.4, in DMEM without phenol red, in a final volume of 100 ml with 0.1 U or 1 U of ecarin (American Diagnostica), a specific prothrombin activator derived from snake venom. Controls were treated in the same way, but without ecarin. Thrombin-like activity was then assayed as described above. Most of the existing prothrombin was activated by 0.1 U of ecarin, as increasing the concentration to 1 U was found to have no further effect on the signal. Gelatin zymography Conditioned media or nerve extracts containing equal amounts of protein were electrophoresed through an 8% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel (Bio-Rad) that was polymerized in the presence of 1 mg/ml gelatin. To produce a linear polyacrylamide gradient gel, a 7% and a 15% gel solution— each containing 1 mg/ml gelatin—were mixed by a gradient mixer and poured with a peristaltic pump at a speed of 2 ml/min, as described (36). The gels were washed once for 30 min in 2.5% Triton X-100 to remove the SDS, and once for 30 min in the reaction buffer (50 mM Tris-HCl, 200 mM NaCl, 10 mM CaCl2, pH 7.5). The gels were then incubated in fresh reaction buffer at 37°C for 3 days. Gelatinolytic activity was visualized by staining of the gels with 0.5% Coomassie brilliant blue. Statistics The significance of differences was calculated using the Welch alternate t test. This test assumes Gaussian populations, but does not assume that the compared populations have equal standard deviations, and is therefore more conservative than the conventional unpaired Student’s t test. RESULTS Thrombin-like activity in rat optic and sciatic nerves Thrombin activities were assayed in conditioned media prepared from uninjured rat optic and sciatic nerves and from nerves excised 1, 4, and 7 days after the injury (Fig. 1A). Samples of conditioned media were then incubated for 1 h with the chromogenic substrate Chromozym-TH and their OD values were recorded. Conditioned media derived from optic Figure 1. Thrombin-like activities in rat optic and sciatic nerves. Nerves were subjected to crush injury, the rats were killed at the indicated time points after the injury, and conditioned media were collected. A) Thrombin-like activities in rat optic nerve-conditioned medium (ONCM, white) and sciatic nerve-conditioned medium (SNCM, black). Activity-generated elevations in OD at 405 nm after 1 h of thrombin activity assay are shown. The bar graph shows the mean 6sem of OD values obtained from four different preparations. The increase in thrombin-like activity on day 1 after injury is extremely significant in the optic nerve (***P,0.0001) and very significant in the sciatic nerve (P,0.001) when compared to that obtained in uninjured nerves. Insert: Thrombin-like activity in rat sciatic nerveconditioned medium. Activity-generated elevations in OD at 405 nm after 10 h of thrombin activity assay are shown. Each bar graph shows the mean 6sem of OD values obtained from four different preparations. The increase in thrombin-like activity on day 1 after injury is very significant compared to that obtained in uninjured nerves (**P,0.001). B) Thrombin-like activitiy in rat optic nerve-conditioned medium (ONCM). Activity-generated elevations in OD at 405 nm after 1 h of thrombin activity assay are shown. The graph shows the mean 6sem of values obtained from different preparations, whose numbers (n) are indicated at each point. The increase in thrombin-like activity on day 1 after the injury is extremely significant (***P,0.0001) compared to that obtained in uninjured nerves. GELATINASE AND THROMBIN-LIKE ACTIVITIES IN NERVOUS SYSTEMS 535 of the synthetic thrombin inhibitor thromstop (Fig. 2) or hirudin (not shown). At the different time points examined, similar dose-dependent patterns of inhibition were observed in the optic and the sciatic nerves: the presence of thromstop at a concentration of 0.1 mM inhibited activity by approximately 60%, and at a concentration of 1 mM by about 90%. Thrombin-like activity is potentially associated with retrograde degeneration Figure 2. Thrombin-like activity in the presence of the specific thrombin inhibitor thromstop in rat optic (gray circles) and sciatic (black squares) nerve. Rats were killed on the day after injury and conditioned media were collected. Thrombin-like activity was assayed in the presence of different concentrations of thromstop. Mean values 6sd of representative examples are shown. Calculation of the inhibition was based on OD elevation at 405 nm after 1 h of thrombin activity assay in the optic nerve and 10 h of the assay in the sciatic nerve. nerves excised 1 day after the injury showed a marked increase in thrombin-like activity relative to the uninjured optic nerves (P,0.0001) and corresponded in intensity to the activity evoked by approximately 9 3 1023 U of pure bovine thrombin. The kinetics are shown in more detail in Fig. 1B. An increase in thrombin-like activity levels could be detected as early as 1.5 h after injury. On day 1 postinjury, thrombin levels were about 40-fold higher than those in control nerves and then gradually declined, reaching baseline levels by day 4. The sciatic nerve also showed an increase in thrombin-like activity 1 day after the injury, but this elevation was approximately 1/20th that of the optic nerve (Fig. 1A) and could be simulated by 0.5 3 1023 U of pure bovine thrombin. Because these signals were very low, we extended the incubation time with Chromozym-TH from 1 to 10 h (Fig. 1A, insert). In both cases, the thrombin-like activity in the sciatic nerve on day 1 postinjury was approximately threefold higher than in the uninjured sciatic nerve (P,0.001) and had returned to baseline by 4 days after injury. Thus, both the sciatic and the optic nerve showed an increase in thrombin-like activity 1 day after injury, but the increase in the optic nerve was approximately 20-fold higher than in the sciatic nerve. Specific inhibition of thrombin-like activity in the rat To further verify the specificity of the measured thrombin-like activity, we analyzed it in the presence 536 Vol. 13 March 1999 To find out whether the elevated thrombin-like activity in the injured rat optic nerve on day 1 postinjury is associated with retrograde or anterograde degeneration, some of the nerves were divided into a distal segment (between the injury site and the optic chiasma) and a proximal segment (including the site of injury and the part of the nerve between the injury site and the optic disk). Conditioned media from the pooled proximal segments and the pooled distal segments were collected separately and their thrombin activities were assayed. In all cases examined, thrombin activity was found mainly in the proximal segments (Fig. 3). Increase of thrombin-like activity by prothrombin activation One possible source for the observed activity could be the activation of prothrombin, known to be Figure 3. Thrombin-like activity in different parts of the rat optic nerve on day 1 after the injury. Nerves were subjected to crush injury, the rats were killed 1 day later, and each optic nerve was separated into a proximal segment (which included the site of injury) and a distal segment. OD values at 405 nm after 0.5 h of thrombin activity assay in conditioned media are shown. The bar graph shows the mean 6sem of OD values obtained from three different preparations. The FASEB Journal FRIEDMANN ET AL. expressed in the nervous system (37). To compare the amount of prothrombin (a reservoir for thrombin-like activity) in the rat optic or sciatic nerves before and after injury, we measured prothrombin levels in the various conditioned media. After the addition of ecarin, which converts prothrombin to thrombin, thrombin-like activity was assayed as before. In the uninjured optic nerve (Fig. 4A), incubation with 0.1 U of ecarin did not evoke a thrombin activity signal. In the injured optic nerve, however, the high signal observed 1 day after injury was further increased approximately threefold by ecarin, suggesting that the amount of thrombin-like activity in the nerve after the injury had corresponded to about 30% of the potentially available activity. In the sciatic nerve (Fig. 4B), incubation with ecarin increased the signal of thrombin-like activity by more than 10-fold in both injured and uninjured nerves. Thus, regardless of injury, it appears that less than 10% of the potentially available thrombin-like activity in the sciatic nerve exists in the active form. Thrombin-like activity in fish optic nerve The kinetics of thrombin-like activity was studied in a similar manner in the regenerating fish optic nerve. Figure 5A shows the thrombin-like activity measured in fish optic nerve on different days after injury as well as in uninjured nerves. Samples of conditioned media were incubated with the chromogenic substrate Chromozym-TH and the OD values were measured after 4 h. Injury of the fish optic nerve resulted in an increase in thrombin-like activity with kinetics that differed from those observed in the rat. The activity started to increase on day 1, reached a broad peak that lasted from day 4 to day 7 after the injury, and then declined. The observed increase was significant when compared to that in uninjured nerves (P,0.05). Specific inhibition of thrombin-like activity in fish optic nerve Thrombin-like activity in the fish optic nerve was inhibited in a dose-dependent manner by the synthetic inhibitor thromstop (Fig. 5B) or hirudin (not shown), demonstrating its specificity. Much larger amounts of both inhibitors were needed to achieve inhibition in the fish nerve than in either of the rat nerves: for example, to achieve an inhibition of about 60% in the fish nerve it was necessary to use 100 mM thromstop, a concentration 1000-fold higher than that needed in the rat. The inhibition pattern in the fish optic nerve was identical at several examined time points. Figure 4. Prothrombin levels and thrombin-like activities in rat optic (A) and sciatic (B) nerves. Nerves were subjected to crush injury, the rats were killed 1 day later, and conditioned media were collected from injured and uninjured nerves. For conversion of prothrombin to thrombin, 0.1 U ecarin was added prior to the thrombin activity assay. Representative examples (mean 6sd) of OD values at 405 nm after 0.5 h of the assay are shown. Activities of gelatinases in rat optic and sciatic nerves To examine whether the observed up-regulation of thrombin-like activity in the injured optic nerve of the rat is unique to thrombin or merely part of a general, nonspecific up-regulation of protease activity after injury (which would imply a general difference in protease regulation between regenerative and nonregenerative nervous systems), we also measured the postinjury activities of gelatinases. As mentioned earlier, gelatinases are known to participate in processes that are related to both regeneration (27–29) and degeneration (30 –33) in the nervous system. Conditioned media and nerve extracts from rat optic and sciatic nerves were collected at specified times after injury, as well as from uninjured nerves, and gel zymography was performed. Figure 6 shows representative gels of gelatinase activities in rat optic (Fig. 6A) and sciatic (Fig 6B) nerves. In both cases, gelatinase A (72 kDa) was found to be constitutively expressed and did not increase after injury. In contrast, gelatinase B (92 kDa) could not be detected in either of the nerves before injury, but was up-regulated in both 1 day after injury, after which its activity declined. In the rat optic nerve, this up-regulation was detectable as early as 6 h after the injury (Fig. 6A, lane 2). In some optic nerve preparations, an additional band below 200 kDa was visible (Fig. 7A, lane 1), probably representing a complex formed between a gelatinase and an inhibitor. Apart from the existence of this complex, the kinetics of the rat sciatic and optic nerves were similar. Identical results were obtained when nerve extracts were used instead of conditioned media (not shown). GELATINASE AND THROMBIN-LIKE ACTIVITIES IN NERVOUS SYSTEMS 537 Figure 5. A) Thrombin-like activity in fish optic nerve. Nerves were subjected to crush injury, the fish were killed on the indicated days after injury, and conditioned media were collected. Activity-generated elevations in OD at 405 nm after 4 h of thrombin activity assay are shown. The bar graph shows the mean values 6sem of OD values obtained from different preparations, whose numbers (n) are indicated above each bar. The increase in thrombinlike activity on days 4 and 7 after the injury is significant (*P,0.05) compared to that obtained in uninjured nerves. B) Thrombin-like activity in the fish optic nerve in the presence of the specific thrombin inhibitor thromstop. Fish were killed on day 7 after injury, conditioned medium was collected, and thrombin activity assay was performed in the presence of different amounts of thromstop. Mean OD values 6sd of a representative example are shown. Calculation of the inhibition was based on OD elevation at 405 nm after 4 h of thrombin activity assay. Gelatinase activity is potentially associated with retrograde degeneration Having found that thrombin-like activity in the rat optic nerve is potentially associated with retrograde degeneration (Fig. 3), we were interested in establishing which parts of the injured optic and sciatic nerves are responsible for the observed changes in gelatinase activity. Accordingly, both nerves were again divided into a proximal segment, which included the site of injury, and a distal segment. Nerve extracts were prepared from each segment separately. In both optic and sciatic nerves, induction of gelatinase B was found to occur only in the proximal segment (Fig. 7). Identical results were obtained when conditioned media were used (not shown). Gelatinase activities in the fish optic nerve Figure 6. Gelatinase activities in rat optic (A) and sciatic (B) nerve. Nerves were subjected to crush injury, the rats were killed at the indicated times after the injury, and conditioned media were collected. Gel zymography in 8% gels was performed with 10 mg of total protein. MMP-2, gelatinase A; MMP-9, gelatinase B. 538 Vol. 13 March 1999 As with thrombin, we studied the kinetics of gelatinase activities in response to axonal injury in the regenerating fish optic nerve. As before, conditioned media and nerve extracts were collected at specified times after the injury, as well as from uninjured nerves, and gel zymography was performed. Figure 8 shows representative gels of gelatinase activities in the fish optic nerve. Two major bands were found at 72 and 92 kDa. Since they were identical in size and exhibited similar postinjury behavior to that of the gelatinase activity observed in the rat, they were assumed to be homologous to rat gelatinases A and B, respectively. As in the rat, the 72 kDa gelatinase was found to be constitutively expressed and unaffected by the injury, whereas the 92 kDa gelatinase was up-regulated. Several additional phenomena, not present in the rat, were observed in the fish. 1) Up-regulation of gelatinase activity after injury remained high for much longer than in the rat; for example, baseline levels of the 92 kDa gelatinase were not reached even by day 7. 2) In uninjured nerves, the ratio between the gelatinases at 72 and 92 kDa differed from that obtained in the rat, indicat- The FASEB Journal FRIEDMANN ET AL. ing a greater abundance of the 92 kDa gelatinase in the fish. 3) Additional bands at about 20, 35, 45, 55, and 80 kDa were up-regulated upon injury. These bands might be a result of up-regulation of additional gelatinases or of proteolysis of the known gelatinases where the cleaved products were still active. 4) Less protein was needed in the fish conditioned medium in order to achieve a band intensity similar to that of the rat conditioned medium, indicating that the activity in the former medium is higher, either because gelatinases are more abundant or because the activities of the fish gelatinases are higher than those of the rat. Identical results were obtained when nerve extracts were used instead of conditioned media (not shown). Figure 8. Gelatinase activities in fish optic nerve. Nerves were subjected to crush injury, the fish were killed on the indicated days after injury, and conditioned media were collected. A) Gel zymography in a 7–15% linear gradient gel was performed with 10 mg of total protein. B) Gel zymography in 8% gel was performed with 2.5 mg of total protein. DISCUSSION Figure 7. Gelatinase activities in different parts of rat optic (A) and sciatic (B) nerve. Nerves were subjected to crush injury, the rats were killed on the indicated days after injury, and nerve extracts were collected. Gel zymography in 8% gels was performed with 30 mg of total protein. P, proximal segment; D, distal segment; MMP-2, gelatinase A; MMP-9, gelatinase B. The neuronal response to injury, as manifested by two types of proteolytic activities, was compared in the nonregenerative rat optic nerve and two regenerative nerves: the rat sciatic nerve and the fish optic nerve. In the rat optic nerve, an extremely significant increase in thrombin-like activity (approximately 40fold compared to that in the uninjured nerve) was observed 1 day after injury. The increase was transient and started to decline 2 days after injury, returning to baseline levels on about day 4. In the rat sciatic nerve, a transient threefold increase in thrombin-like activity was seen 1 day after injury. Although this increase was about 20-fold smaller than in the optic nerve, it was still very significant. In both the optic and the sciatic nerves of rat, the thrombin-like activity was inhibited in a dose-dependent manner by two specific inhibitors, thromstop and hirudin, confirming the specificity of the observed signal. High levels of thrombin activity were previously shown to inhibit neurite outgrowth (16) and to induce apoptotic cell death in neurons and astrocytes (20 –22) in vitro. The high levels of thrombinlike activity in the rat optic nerve observed 1 day after injury in our study suggest that the degenerative processes seen after CNS injury might be related to thrombin activity and that thrombin-like activity might contribute to the nonregenerative nature of the optic nerve environment. The marked increase GELATINASE AND THROMBIN-LIKE ACTIVITIES IN NERVOUS SYSTEMS 539 in activity could also override the beneficial effects of thrombin that are expected (from in vitro data) to occur in the CNS, such as the stimulation of NGF secretion by astrocytes (12). In contrast, a slightly increased thrombin-like activity, such as that seen in the rat sciatic nerve, might not only be below the toxicity threshold but might even be beneficial for the injured nerves. In the PNS, the injured nerve is rapidly invaded by macrophages, which are crucial for regeneration in the nervous system (1, 38 – 40). Thrombin is a potent chemoattractant for human monocytes (41) and stimulates the production of monocyte chemotactic proteins (42– 44). In the mammalian CNS, where macrophage recruitment is impaired (38) because of an immune brain barrier (45), these potential effects of thrombin could be minimized or excluded. In addition, levels of thrombin-like activity in the sciatic nerve could be within the range at which neurons are rescued from cell death resulting from metabolic insults, which likely to be present after nerve injury (20, 21). Although it cannot be ruled out, the possibility that the observed thrombin-like activity results from a protein other than thrombin seems very unlikely, an assumption supported both by the cleavage of the specific thrombin substrate Chromozym-TH and the inhibition by two specific thrombin inhibitors. Thrombin activity results from the cleavage of prothrombin, which is known to be expressed in the nervous system (37). The experiments with ecarin showed that activation of the available prothrombin would increase thrombin activity by more than 10fold in the injured sciatic nerve, but by only about 3-fold in the injured optic nerve. This suggests that prothrombin activation is under tighter control in the sciatic nerve than in the optic nerve. A prothrombin activator independent of the blood coagulation cascade was recently described, and its involvement after injury was suggested (46, 47). Although not detected in uninjured brain, it was found to be active in a cell line derived from mammalian CNS (46). In addition, prothrombin activation has been shown to occur on neuronal surfaces (48). Our data suggest that different mechanisms of prothrombin activation in the optic and sciatic nerves could account for the high levels of thrombin-like activity observed after injury in the rat optic nerve. In a recent study of thrombin and prothrombin elevation in rat sciatic nerve (49), thrombin-generated activity increased significantly from day 1 after nerve crush, reaching a peak on day 3 and returning to basal levels on day 6. Prothrombin could not be detected in uninjured nerves, but the ratio of prothrombin-to-thrombin activity after injury was comparable to the ratio obtained in our study, and led those authors to suggest a tight control of thrombin activation in the sciatic nerve. The different kinetics 540 Vol. 13 March 1999 obtained in that study (reflecting a relatively delayed increase in thrombin activity) and the inability to detect prothrombin in the uninjured sciatic nerve might be attributable to the methodology, as nerve extracts and not conditioned media were used. Differences in thrombin-like activities between the rat optic and sciatic nerves could also result from changes in the levels of thrombin inhibitors. A known thrombin inhibitor in the nervous system is the glial-derived protease nexin I (PN-I) (50), which promotes neurite outgrowth (51, 52) and could therefore reverse the negative effects of thrombinlike activity. In the sciatic nerve, PN-I is up-regulated after injury, starting on about day 3 and peaking on day 7 (49, 53). The behavior of PN-I shortly after injury in the optic nerve has not been studied. Nevertheless, our finding of high thrombin-like activity in the optic nerve 1 day after injury may reflect the paucity or absence of thrombin inhibitors, such as PN-I, at this time. In the fish, the kinetics of thrombin-like activity in the optic nerve differed from those in the rat. As in the rat, the increase in activity started on day 1 but, unlike in the rat, continued and reached a plateau lasting from day 4 to day 7 after injury. As similar peaking in the fish nerve was also observed in the activities of the gelatinases, this elevation pattern may represent a general response of this regenerative nerve to injury. The kinetics observed in the fish optic nerve point to the establishment of a different extracellular milieu, which is generated by the proteases and could account, at least in part, for the regeneration-supportive properties of the neuronal environment. As in the rat, thrombin-like activity in the fish optic nerve was specifically inhibited, in a dose-dependent manner, by the inhibitors thromstop and hirudin. Much larger amounts of these inhibitors were needed in the fish than in the rat, indicating the presence of other substances in the fish conditioned medium that influence the interaction between the inhibitors and the enzyme. Studies in our laboratory recently demonstrated the presence of factor XIIIa in the nervous tissues of rat and fish, as well as a correlation between the postinjury appearance and activation of the enzyme and the regenerative ability of the tissue (54). Thrombin is crucial for the regulation of factor XIIIa in vivo, and the differential activities of thrombin might be responsible for the differences in the observed behavior of factor XIIIa. We found that the thrombin-like activity in the rat optic nerve was associated mainly with the part of the nerve adjacent to and including the site of injury, i.e., the proximal segment. This suggests that the observed injury-induced activity is associated with retrograde degeneration and, accordingly, is evoked either by changes emanating from the neuronal cell The FASEB Journal FRIEDMANN ET AL. bodies or from a non-neuronal cell response at the site of injury. As mentioned earlier, prothrombin mRNA was shown to be expressed by cells of the nervous system (37). In addition, thrombin has been detected in brain and astroglial cell cultures (55). Nevertheless, thrombin synthesis could also occur elsewhere, for example, in platelets that adhere to the site of injury. The cellular source responsible for thrombin activity in the nervous system, and particularly after injury, is not yet known. In contrast to the differences in thrombin-like activity, we found that the activation pattern of gelatinase A or B was similar in all three nerve types. Gelatinase B was up-regulated after crush injury, whereas gelatinase A was not affected. In the rat, up-regulation of gelatinase B (which did not occur in the uninjured nerve) peaked about 1 day after optic or sciatic nerve injury and decreased on about day 4. The fish showed a different proportion between the two gelatinases in the uninjured nerve, as well as a relatively delayed peak of the activity. As already mentioned, this delayed peaking (which is not exclusive to gelatinases) may influence the regenerative properties of the fish optic nerve. Gelatinases are known to participate in processes that are related to both regeneration (27–29) and degeneration (30 –33) in the nervous system. This study shows not only that gelatinases are present in regenerative and nonregenerative white matter but that, surprisingly, the induction patterns of gelatinases A and B in all three nerve types examined are similar. One possible explanation is that rather than being directly related to degeneration or regeneration, the mechanisms of activation or inhibition of gelatinases in degenerating and regenerating systems might differ. Gelatinase activity can be inhibited by tissue inhibitors of matrix metalloproteases (TIMPs). TIMP-1 was shown to be induced in the sciatic nerve on days 1 and 4 postinjury, and immunostaining demonstrated its colocalization with Schwann cells and macrophages (56). Another possibility is that not only total protease activity, but also the ratio of ECM-degrading to ECM-producing proteases, might play a role in determining the permissiveness of the extracellular milieu to regeneration. This is in line with our finding that the nonregenerative rat optic nerve differed significantly from the regenerative rat sciatic nerve with respect to the amounts of fibrin-producing thrombin, but not of ECM-degrading gelatinases. As with thrombin-like activity, gelatinase activities in the rat optic and sciatic nerve were limited to the proximal segment, indicating that the observed activities are associated not with anterograde degeneration, but with an active process occurring at the site of injury or in the proximal part of the nerve. Our findings in the sciatic nerve are in line with the observations of La Fleur et al. (56). The cellular source of the gelatinases in vivo remains to be established. However, the inflammation-related cytokines interleukin (IL) -1a, IL-1b, and TNF-a are potent inducers of gelatinases in cultured rat astrocytes (57), and activated microglia release gelatinase B in vitro (58). Gelatinase activity has also been reported in Schwann cells and glioma cells (59, 60). In summary, a comparison of the general kinetics of the induced thrombin-like and gelatinase activities in rat and fish showed that activity in the rat peaks early and declines between days 1 and 4 postinjury, whereas in the fish the activity lasts for at least a week. A comparison of activity levels in the optic and sciatic nerves of the rat showed significant differences in thrombin-like activity but similarity in the activities of gelatinases. The induction pattern of the 72 kDa and 92 kDa gelatinases was similar in fish as well. Both the different kinetics and the differential activity levels might affect the nerve’s ability to regenerate. This study further highlights the fact that tissue repair and regeneration are closely linked and that substances that generally participate in wound healing establish an extracellular milieu that can lead either to regeneration or its failure. We thank Shirley Smith for editorial assistance and Haya Avital for help with graphics. M.S. holds the Maurice and Ilse Katz Professorial Chair in Neurobiology. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. GELATINASE AND THROMBIN-LIKE ACTIVITIES IN NERVOUS SYSTEMS Lotan, M., and Schwartz, M. (1994) Cross talk between the immune system and the nervous system in response to injury: implications for regeneration. FASEB J. 8, 1026 –1033 Weinberg, E. L., and Spencer, P. S. (1979) Studies on the control of myelinogenesis. 3. Signalling of oligodendrocyte myelination by regenerating peripheral axons. Brain Res. 162, 273–279 Richardson, P. M., McGuinness, U. M., and Aguayo, A. J. (1980) Axons from CNS neurons regenerate into PNS grafts. Nature (London) 284, 264 –265 Faber Elman, A., Miskin, R., and Schwartz, M. (1995) Components of the plasminogen activator system in astrocytes are modulated by tumor necrosis factor-alpha and interleukin-1 beta through similar signal transduction pathways. J. Neurochem. 65, 1524 –1535 Faber Elman, A., Solomon, A., Abraham, J. A., Marikovsky, M., and Schwartz, M. (1996) Involvement of wound-associated factors in rat brain astrocyte migratory response to axonal injury: in vitro simulation. J. Clin. Invest. 97, 162–171 Raghow, R. (1994) The role of extracellular matrix in postinflammatory wound healing and fibrosis. FASEB J. 8, 823– 831 Coughlin, S. R., Vu, T. K., Hung, D. T., and Wheaton, V. I. (1992) Characterization of a functional thrombin receptor. Issues and opportunities. J. Clin. Invest. 89, 351–355 Monard, D. (1988) Cell-derived proteases and protease inhibitors as regulators of neurite outgrowth. Trends Neurosci. 11, 541–544 Halkier, T. (1991) Mechanisms in Blood Coagulation, Fibrinolysis and the Complement System. Cambridge University Press, Cambridge Cavanaugh, K. P., Gurwitz, D., Cunningham, D. D., and Bradshaw, R. A. (1990) Reciprocal modulation of astrocyte stellation by thrombin and protease nexin-1. J. Neurochem. 54, 1735– 1743 541 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 542 Perraud, F., Besnard, F., Sensenbrenner, M., and Labourdette, G. (1987) Thrombin is a potent mitogen for rat astroblasts but not for oligodendroblasts and neuroblasts in primary culture. Int. J. Dev. Neurosci. 5, 181–188 Neveu, I., Jehan, F., Jandrot Perrus, M., Wion, D., and Brachet, P. (1993) Enhancement of the synthesis and secretion of nerve growth factor in primary cultures of glial cells by proteases: a possible involvement of thrombin. J. Neurochem. 60, 858 – 867 Lipton, S. A., Wagner, J. A., Madison, R. D., and D’Amore, P. A. (1988) Acidic fibroblast growth factor enhances regeneration of processes by postnatal mammalian retinal ganglion cells in culture. Proc. Natl. Acad. Sci. USA 85, 2388 –2392 Lobb, R. R. (1988) Thrombin inactivates acidic fibroblast growth factor but not basic fibroblast growth factor. Biochemistry 27, 2572–2578 Loret, C., Sensenbrenner, M., and Labourdette, G. (1989) Differential phenotypic expression induced in cultured rat astroblasts by acidic fibroblast growth factor, epidermal growth factor, and thrombin. J. Biol. Chem. 264, 8319 – 8327 Gurwitz, D., and Cunningham, D. D. (1988) Thrombin modulates and reverses neuroblastoma neurite outgrowth. Proc. Natl. Acad. Sci. USA 85, 3440 –3444 Suidan, H. S., Stone, S. R., Hemmings, B. A., and Monard, D. (1992) Thrombin causes neurite retraction in neuronal cells through activation of cell surface receptors. Neuron 8, 363–375 Debeir, T., Benavides, J., and Vige, X. (1996) Dual effects of thrombin and a 14-amino acid peptide agonist of the thrombin receptor on septal cholinergic neurons. Brain Res 708, 159 –166 Donovan, F. M., Pike, C. J., Cotman, C. W., and Cunningham, D. D. (1997) Thrombin induces apoptosis in cultured neurons and astrocytes via a pathway requiring tyrosine kinase and RhoA activities. J. Neurosci. 17, 5316 –5326 Vaughan, P. J., Pike, C. J., Cotman, C. W., and Cunningham, D. D. (1995) Thrombin receptor activation protects neurons and astrocytes from cell death produced by environmental insults. J. Neurosci. 15, 5389 –5401 Donovan, F. M., and Cunningham, D. D. (1998) Signaling pathways involved in thrombin-induced cell protection. J. Biol. Chem. 273, 12746 –12752 Smirnova, I. V., Zhang, S. X., Citron, B. A., Arnold, P. M., and Festoff, B. W. (1998) Thrombin is an extracellular signal that activates intracellular death proteases pathways inducing apoptosis in model motor neurons. J. Neurobiol. 36, 64 – 80 Woessner, J. F., Jr. (1991) Matrix metalloproteinases and their inhibitors in connective tissue remodeling. FASEB J. 5, 2145– 2154 Krane, S. M. (1994) Clinical importance of metalloproteinases and their inhibitors. Ann. N.Y. Acad. Sci. 732, 1–10 Stamatoglou, S. C., Hughes, R. C., and Lindahl, U. (1987) Rat hepatocytes in serum-free primary culture elaborate an extensive extracellular matrix containing fibrin and fibronectin. J. Cell Biol. 105, 2417–2425 Williams, L. R., Longo, F. M., Powell, H. C., Lundborg, G., and Varon, S. (1983) Spatial-temporal progress of peripheral nerve regeneration within a silicone chamber: parameters for a bioassay. J. Comp. Neurol. 218, 460 – 470 Knoops, B., Hubert, I., Hauw, J. J., and van den Bosch de Aguilar, P. (1991) Axonal growth and glial migration from cocultured hippocampal and septal slices into fibrin-fibronectin-containing matrix of peripheral regeneration chambers: a light and electron microscope study. Brain Res. 540, 183–194 Pittman, R. N., and Buettner, H. M. (1989) Degradation of extracellular matrix by neuronal proteases. Dev. Neurosci. 11, 361–375 Muir, D. (1994) Metalloproteinase-dependent neurite outgrowth within a synthetic extracellular matrix is induced by nerve growth factor. Exp. Cell. Res. 210, 243–252 Rosenberg, G. A., Dencoff, J. E., McGuire, P. G., Liotta, L. A., and Stetler Stevenson, W. G. (1994) Injury-induced 92-kilodalton gelatinase and urokinase expression in rat brain. Lab. Invest. 71, 417– 422 Rosenberg, G. A., Estrada, E. Y., Dencoff, J. E., and Stetler Stevenson, W. G. (1995) Tumor necrosis factor-alpha-induced gelatinase B causes delayed opening of the blood– brain barrier: an expanded therapeutic window. Brain Res. 703, 151–155 Gijbels, K., Masure, S., Carton, H., and Opdenakker, G. (1992) Gelatinase in the cerebrospinal fluid of patients with multiple Vol. 13 March 1999 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. sclerosis and other inflammatory neurological disorders. J. Neuroimmunol. 41, 29 –34 Gijbels, K., Proost, P., Masure, S., Carton, H., Billiau, A., and Opdenakker, G. (1993) Gelatinase B is present in the cerebrospinal fluid during experimental autoimmune encephalomyelitis and cleaves myelin basic protein. J. Neurosci. Res. 36, 432– 440 Duvdevani, R., Rosner, M., Belkin, M., Sautter, J., Sabel, B. A., and Schwartz, M. (1990) Graded crush of the rat optic nerve as a brain injury model: Combining electrophysiological and behavioral outcome. Restor. Neurol. Neurosci. 2, 31–38 Lottenberg, R., Hall, J. A., Blinder, M., Binder, E. P., and Jackson, C. M. (1983) The action of thrombin on peptide p-nitroanilide substrates. Substrate selectivity and examination of hydrolysis under different reaction conditions. Biochim. Biophys. Acta 742, 539 –557 Gallagher, S. R. (1997) One dimensional gel electrophoresis of proteins. In: Current Protocols in Molecular Biology (Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A., and Struhl, K., eds) Unit 10.12, Greene Publishing and Wiley-Interscience, New York Dihanich, M., Kaser, M., Reinhard, E., Cunningham, D., and Monard, D. (1991) Prothrombin mRNA is expressed by cells of the nervous system. Neuron 6, 575–581 Perry, V. H., Brown, M. C., and Gordon, S. (1987) The macrophage response to central and peripheral nerve injury. A possible role for macrophages in regeneration. J. Exp. Med. 165, 1218 –1223 Rapalino, O., Lazarov-Spiegler, O., Agranov, E., Velan, G. J., Fraidakis, M., Yoles, E., Solomon, A., Gepstein, R., Katz, A., Belkin, M., Hadani, M., and Schwartz, M. (1998) Implantation of stimulated homologous macrophages results in partial recovery of paraplegic rats. Nature Med. 4, 814 – 821 Lazarov Spiegler, O., Solomon, A. S., Zeev Brann, A. B., Hirschberg, D. L., Lavie, V., and Schwartz, M. (1996) Transplantation of activated macrophages overcomes central nervous system regrowth failure. FASEB J. 10, 1296 –1302 Bar Shavit, R., Kahn, A., Wilner, G. D., and Fenton, J. W. D. (1983) Monocyte chemotaxis: stimulation by specific exosite region in thrombin. Science 220, 728 –731 Sower, L. E., Froelich, C. J., Carney, D. H., Fenton, J. W. N., and Klimpel, G. R. (1995) Thrombin induces IL-6 production in fibroblasts and epithelial cells. Evidence for the involvement of the seven-transmembrane domain (STD) receptor for alphathrombin. J. Immunol. 155, 895–901 Shankar, R., de la Motte, C. A., and DiCorleto, P. E. (1992) Thrombin stimulates PDGF production and monocyte adhesion through distinct intracellular pathways in human endothelial cells. Am. J. Physiol. 262, C199 –C206 Grandaliano, G., Valente, A. J., and Abboud, H. E. (1994) A novel biologic activity of thrombin: stimulation of monocyte chemotactic protein production. J. Exp. Med. 179, 1737–1741 Hirschberg, D. L., and Schwartz, M. (1995) Macrophage recruitment to acutely injured central nervous system is inhibited by a resident factor: a basis for an immune-brain barrier. J. Neuroimmunol. 61, 89 –96 Sekiya, F., Usui, H., Inoue, K., Fukudome, K., and Morita, T. (1994) Activation of prothrombin by a novel membrane-associated protease. An alternative pathway for thrombin generation independent of the coagulation cascade. J. Biol. Chem. 269, 32441–32445 Shikamoto, Y., Shibusawa, S., Okuyama, I., and Morita, T. (1997) Characterization of membrane-associated prothrombin activator in normal and injured murine tissues. FEBS Lett. 412, 526 –530 Grabham, P. W., Monard, D., Gallimore, P. H., and Grand, R. J. A. (1991) Modulation of human neurite outgrowth by serine proteases: a comparison of the interaction of thrombin and prothrombin with glia-derived nexin. Eur. J. Neurosci. 3, 663– 668 Smirnova, I. V., Ma, J. Y., Citron, B. A., Ratzlaff, K. T., Gregory, E. J., Akaaboune, M., and Festoff, B. W. (1996) Neural thrombin and protease nexin I kinetics after murine peripheral nerve injury. J. Neurochem. 67, 2188 –2199 Gloor, S., Odink, K., Guenther, J., Nick, H., and Monard, D. (1986) A glia-derived neurite promoting factor with protease The FASEB Journal FRIEDMANN ET AL. 51. 52. 53. 54. 55. 56. inhibitory activity belongs to the protease nexins. Cell 47, 687– 693 Monard, D., Niday, E., Limat, A., and Solomon, F. (1983) Inhibition of protease activity can lead to neurite extension in neuroblastoma cells. Prog. Brain Res. 58, 359 –364 Zurn, A. D., Nick, H., and Monard, D. (1988) A glia-derived nexin promotes neurite outgrowth in cultured chick sympathetic neurons. Dev. Neurosci. 10, 17–24 Meier, R., Spreyer, P., Ortmann, R., Harel, A., and Monard, D. (1989) Induction of glia-derived nexin after lesion of a peripheral nerve. Nature (London) 342, 548 –550 Monsonego, A., Mizrahi, T., Moalem, G., Bardos, H., Adany, R., and Schwartz, M. (1998) Factor XIIIa as a nerve-associated transglutaminase. FASEB J. 12, 1163–1171 Deschepper, C. F., Bigornia, V., Berens, M. E., and Lapointe, M. C. (1991) Production of thrombin and antithrombin III by brain and astroglial cell cultures. Brain Res. Mol. Brain. Res. 11, 355–358 La Fleur, M., Underwood, J. L., Rappolee, D. A., and Werb, Z. (1996) Basement membrane and repair of injury to peripheral nerve: defining a potential role for macrophages, matrix metal- 57. 58. 59. 60. GELATINASE AND THROMBIN-LIKE ACTIVITIES IN NERVOUS SYSTEMS loproteinases, and tissue inhibitor of metalloproteinases-1. J. Exp. Med. 184, 2311–2326 Gottschall, P. E., and Deb, S. (1996) Regulation of matrix metalloproteinase expressions in astrocytes, microglia and neurons. Neuroimmunomodulation 3, 69 –75 Gottschall, P. E., Yu, X., and Bing, B. (1995) Increased production of gelatinase B (matrix metalloproteinase-9) and interleukin-6 by activated rat microglia in culture. J. Neurosci. Res 42, 335–342 Apodaca, G., Rutka, J. T., Bouhana, K., Berens, M. E., Giblin, J. R., Rosenblum, M. L., McKerrow, J. H., and Banda, M. J. (1990) Expression of metalloproteinases and metalloproteinase inhibitors by fetal astrocytes and glioma cells. Cancer Res. 50, 2322–2329 Muir, D. (1995) Differences in proliferation and invasion by normal, transformed and NF1 Schwann cell cultures are influenced by matrix metalloproteinase expression. Clin. Exp. Metastasis 13, 303–314 Received for publication July 27, 1998. Revised for publication October 29, 1998. 543