Rho activation patterns after spinal cord injury and the role of activated Rho in apoptosis in the central nervous system

Growth inhibitory proteins have long been known to inhibit axonal regeneration in the central nervous system (CNS)^*(Schwab et al., 1993). These inhibitory proteins are enriched in myelin, and the three best characterized myelin-derived growth inhibitory proteins include myelin-associated glycoprotein (MAG), Nogo, and oligodendrocyte-myelin glycoprotein (McKerracher and Winton, 2002; Woolf and Bloechlinger, 2002). Recent data indicate that these three inhibitory proteins bind to the same neuronal receptor, the Nogo-66 receptor (NgR), but there are additional inhibitory proteins in the CNS that act through different receptors. More important, both NgR and NgR-independent inhibitory proteins signal to activate Rho, a small intracellular GTPase (Niederost et al., 2002; Winton et al., 2002).

Early experiments demonstrated that lysophosphatidic acid causes neurite retraction and cell rounding by activating Rho (Jalink et al., 1994; Tigyi et al., 1996). The use of C3 transferase to inactivate Rho in primary neurons plated on various types of inhibitory proteins and dominant-negative Rho-expressing PC-12 cells provides direct evidence that the inactivation of Rho results in neurite outgrowth on inhibitory substrates. In vivo experiments in rats and mice have shown that inactivation of Rho or of Rho kinase promotes axon regeneration and functional recovery after spinal cord injury (SCI) in rats and mice (Lehmann et al., 1999; Hara et al., 2000; Dergham et al., 2002). However, it is not known how CNS injury may affect Rho expression and activation.

The mechanism where growth inhibitory proteins may affect Rho signaling are beginning to be understood. It has been shown recently that NgR can activate Rho in a p75 neurotrophin receptor– (p75^NTR) dependent manner (Wang et al., 2002). First, p75^NTR-null mutant mice are not inhibited by MAG, showing a key role of p75^NTR in growth inhibitory signaling by MAG (Yamashita et al., 2002). Also, Rho binds to p75^NTR (Yamashita et al., 1999), and Rho is likely to form part of the membrane raft receptor complex responsible for growth inhibitory signaling (McKerracher and Winton, 2002; Woolf and Bloechlinger, 2002). Although p75^NTR has been implicated in apoptosis after SCI (Casha et al., 2001), it is not known to what extent Rho signaling by p75^NTR participates in apoptotic events after SCI.

Isoforms of Rho exist, and in neurons RhoA is expressed at higher levels than RhoB and RhoC (Lehmann et al., 1999). Therefore, we have focused on RhoA for our studies in neurons. In nonneuronal cells, Rho family GTPases are best characterized for their effects on organization and regulation of the actin cytoskeleton (Ridley, 2001), but they have also been shown to play a role in the regulation of apoptosis (Jimenez et al., 1995; Aznar and Lacal, 2001; Coleman and Olson, 2002). The extent to which Rho may participate in apoptotic pathways in neuronal cells has yet to be determined. In neurons, Rho is activated in response to chemorepulsive molecules (Jin and Strittmatter, 1997; Wahl et al., 2000) and is important in axon guidance during development. In adult neurons, inhibitory substrates (Lehmann et al., 1999; Niederost et al., 2002; Winton et al., 2002) and secreted factors such as TNF (Neumann et al., 2002) can alter Rho activation levels. Levels of Rho expression are altered in malignant disease (Suwa et al., 1998; Fritz et al., 1999; Clark et al., 2000); however, little is known about how traumatic injury and disease in the CNS alter Rho activation states in vivo.

Rho activation can be studied by probing cell homogenates with the Rho-binding domain (RBD) from the Rho-GTP–interacting protein, rhotekin (Reid et al., 1996). We use this pull-down assay to detect a significant increase in active Rho in CNS tissue homogenates after SCI. We show that SCI causes an increase in active Rho without affecting RhoA expression levels. We made use of an in situ pull-down assay (Li et al., 2002) to determine that neurons and glia in the spinal cord show Rho activation. To test the use of a Rho antagonist to reverse Rho activation, we used a cell-permeable form of C3 transferase (C3–05) that has a short transport sequence added to the COOH terminal to help entry into cells (Winton et al., 2002). We show that C3–05 specifically inactivates Rho in vivo and prevents up-regulation of p75^NTR. Treatment of injured spinal cord with C3–05 not only effectively reversed Rho activation but also had cell protective effects.

To examine the effect of growth-inhibitory proteins on Rho activation, we plated PC-12 cells on myelin, MAG, or poly-l-lysine substrates. We measured amounts of GTP-Rho in cell lysates by precipitation with RBD from rhotekin that binds only GTP-bound Rho (Reid et al., 1996). Cells plated on inhibitory substrates had high endogenous Rho-GTP levels compared with poly-l-lysine controls (Fig. 1 A). The activation of Rho in cells plated on myelin or MAG was reversed by treatment with the Rho antagonist C3–05 (Fig. 1 B). The RBD beads incubated without lysate (buffer only) show no active Rho when overexposed GST-RBD is detectable (Fig. 1 C), showing the specificity of the assay for Rho. Treatment of neuronal cells with C3–05 promotes neurite outgrowth on MAG or myelin substrates (Winton et al., 2002). Our results with MAG and myelin are consistent with recent studies showing Rho activation in the presence of Nogo (Niederost et al., 2002) or upon activation of Nogo receptor (Wang et al., 2002).

To investigate RhoA activation states after traumatic CNS injury, we measured active RhoA levels in rodent tissue homogenates by Rho pull-down assay. We studied tissue isolated from regions of traumatically injured spinal cords of both rats and mice because of their different responses to injury. Rats develop an extensive necrotic lesion cavity after SCI, whereas mice do not (Steward et al., 1999). In rats, we examined spinal cord regions after transection or contusion injury. Experiments are shown as paired control and injured CNS samples (Figs. 2 and 3). Homogenates from different animals were not pooled, and each gel lane represents results from one animal. In uninjured CNS tissue, GTP-Rho levels were consistently low (Fig. 2, controls). By contrast, Rho activation is dramatically increased after injury (Fig. 2, A and B), increasing over 10-fold (Fig. 2 C). Expression levels of total RhoA, as detected by Western blots from tissue homogenates used for isolation of GTP-Rho, did not change (Fig. 2, A and B). These results show that Rho is massively activated in CNS tissue of rats and mice after SCI compared with uninjured spinal cord. To examine if the activation of Rho after SCI injury was sustained or transient, we prepared homogenates from transected spinal cord 1.5 h, 24 h, 3 d, and 7 d after lesion. Interestingly, we found that Rho was active as early as 1.5 h after injury. The significant increase in activation observed by 24 h was sustained for at least 7 d (Fig. 2, B and C).

In neuronal-like cells (PC-12), Rho becomes activated when cells are plated on substrates of MAG or myelin (Fig. 1 A). It has also been shown that MAG is not only present in myelin but is released from damaged white matter after injury (Tang et al., 2001). To confirm that growth inhibitory proteins were present in the lesion sites after SCI, we examined MAG expression levels after injury. The same tissue homogenates used for the experiments in Fig. 2 B were probed with a monoclonal antibody raised against MAG. MAG levels remain unchanged; minimal protein degradation was visible in overexposed blots (Fig. 2 D). The sustained activation of Rho after SCI may result, in part, from contact of damaged cells with myelin-derived inhibitory proteins.

To test if we could reverse the increase in Rho activation in injured spinal cord, we made use of the Rho antagonist C3–05 (Winton et al., 2002). We injected C3–05 in a fibrin matrix into the lesion site after spinal cord transection, or C3–05 alone into contused spinal cord, and the lesion sites were removed 24 h later. Treatment with C3–05 inactivated Rho, bringing the RhoA activation levels back to the normal basal state (Fig. 3, A and B). To determine if the reversal of Rho activation was sustained after a single injection of the compound, we examined rats 7 d after transection injury and treatment. Even 7 d after C3–05 treatment, Rho activation still remained at basal levels (Fig. 3, A and B). Next, we asked if C3–05 remained at the lesion site after treatment. Probing the homogenates with a polyclonal antibody raised against C3 (Winton et al., 2002) demonstrated that C3–05 was detected at the lesion site at all of the time points tested (Fig. 3 A). To determine if endogenous cells in the spinal cord were able to take up and retain C3–05 after treatment, we examined sections of rat spinal cord double labeled with an antibody specific for C3 and with cell type–specific markers. We detected intracellular C3 immunoreaction in neurons, astrocytes, and oligodendrocytes after injection of C3–05 (Fig. 3 C), showing that endogenous cells from the spinal cord take up C3–05 in vivo.

Although the location of C3–05 can indicate the potential to suppress Rho activation, it does not permit us to determine which cells have increased Rho activation after SCI. To further examine increased GTP-Rho after SCI, we used a modified in situ pull-down method, omitting cell transfection with recombinant Rho (Li et al., 2002) to detect endogenous Rho activation levels. We incubated sections with GST-RBD, and cells that bound high levels of RBD were detected with an anti-GST antibody. Active Rho was detected in many cells in both the gray (Fig. 4, panels 1 and 2) and white matter (Fig. 4, panel 3) of injured spinal cord. We also found that Rho was activated both rostral (Fig. 4, panel 1) and caudal (Fig. 4, panel 2) to the lesion site. At further distances from the lesion site, staining for active Rho was very faint or absent (Fig. 4, panel 4). Rho-GTP was not detected in uninjured spinal cord (Fig. 5 A, left) or after C3–05 treatment was used to reverse the increase in Rho activation after SCI (Fig. 5 A, middle). To assess the specificity of the technique, we incubated sections with GST without RBD, and no positive cellular active Rho staining is visible (Fig. 5 A, right).

To specifically determine which cell types express active Rho, sections were double labeled with cell type–specific markers. We detected active Rho in neurons, astrocytes, and oligodendrocytes 24 h, 3 d, and 7 d after injury (Fig. 5 B). Therefore, the activation of Rho observed after injury (Figs. 2 and 3) is an endogenous cellular response (Fig. 5 B). The ability to detect GTP-Rho without prior transfection (Li et al., 2002) confirms the high level of endogenous active Rho in neurons and glia after SCI.

Both neurons and glia undergo apoptosis after SCI in rat, which leads to the formation of a large lesion cavity (Liu et al., 1997; Shuman et al., 1997; Grossman et al., 2001). Even though mice do not develop cavitation at the site of SCI, we detected apoptotic neurons, astrocytes, and oligodendrocytes by double staining with cell-specific markers and TUNEL (Fig. 6 A, top), similar to that observed after rat SCI (Fig. 6 A, bottom). Importantly, in both mice and rats treated with C3–05, the number of TUNEL-labeled cells was significantly reduced by ∼50% after SCI (Fig. 6 B). Not only was C3–05 present in neurons, astrocytes, and oligodendrocytes (Fig. 3 C), but most cells containing C3–05 were not TUNEL positive (Fig. 6, C and D). The small number of cells double labeled with C3 and TUNEL (16%) suggests that C3–05 penetrated into some cells that had progressed too far into the apoptotic cascade to be rescued from death. Together our results indicate that inactivation of Rho after SCI protects cells from apoptosis. These findings have clinical relevance because neuroprotective treatments after SCI lead to improved functional recovery (Liu et al., 1997).

It has been shown recently that MAG activates Rho in the presence of p75^NTR (Yamashita et al., 2002) and that MAG interacts with neuronal lipid rafts containing NgR, GT1b, p75^NTR, and Rho (Vinson et al., 2003). We have shown that after SCI, MAG is present at the lesion sites (Fig. 2 D). To determine the mechanism by which Rho is activated after injury, we examined if Rho activation was p75^NTR dependent. We first examined if p75^NTR was present in cells containing active Rho. We found that p75^NTR colocalizes with active Rho in both gray (Fig. 7 A, top) and white matter (Fig. 7 A, bottom) 24 h after SCI. We then examined levels of active Rho in mice lacking the p75^NTR gene (p75^NTR−/−). No change in Rho activation was detected 24 h after SCI (Fig. 7 B). Rho activation was, however, detected in these animals 3 d after injury (Fig. 7 C). These results indicate that Rho is activated through a p75^NTR-dependent mechanism early after SCI, but at later time points p75^NTR-independent activation occurs.

The p75^NTR has been implicated in the regulation of apoptosis after injury in the nervous system (Cheema et al., 1996; Frade and Barde, 1999; Dechant and Barde, 2002). In addition to counting TUNEL cells after SCI (Fig. 6 B), we counted p75^NTR-labeled cells and double labeled cells. Many cells express p75^NTR after SCI (Fig. 8 B), and 74% of these cells were also TUNEL positive. Treatment with C3–05 resulted in a decrease in p75^NTR alone and in p75^NTR and TUNEL (Fig. 8, A and B). These results suggest a correlation between Rho activation, p75^NTR expression, and cell death. To further investigate the involvement of p75^NTR in Rho activation after SCI, we probed the homogenates of transected spinal cord with p75^NTR-specific antibodies using the same homogenates as shown in the bottom panel of Fig. 8 C. There was very low p75^NTR detected in Western blots in the adult spinal cord (Fig. 7 B and Fig. 8 C, controls). Levels of p75^NTR increased as early as 24 h after SCI, and high levels were detected 3 and 7 d postinjury, a finding consistent with previous reports of p75^NTR up-regulation after SCI (Casha et al., 2001; Widenfalk et al., 2001; Beattie et al., 2002). Treatment with C3–05 not only blocked the increase in p75^NTR protein levels after SCI, as detected by Western blot (Fig. 8 C), but also reduced the number of p75^NTR-TUNEL–labeled cells (Fig. 8 B). These results suggest, that early on, Rho activation after SCI is mediated, at least in part, by p75^NTR. These results also indicate that Rho activation is important for p75^NTR up-regulation after SCI.

It is now well established that neurons in the CNS respond to negative growth inhibitory cues and positive growth-promoting signals. Much work has underlined the importance of myelin-derived growth inhibitory proteins expressed by oligodendrocytes and present in white matter (Schwab et al., 1993; Schwab, 2002). The consequence of a growth inhibitory environment in the CNS is that injured neurons fail to regrow their transected axons even though they have an inherent capacity to regenerate. The molecular mechanisms of neuronal growth after injury have been widely studied in order to promote regeneration after SCI (David and Lacroix, 2003). Work from our lab and others has established that inactivation of Rho signaling is sufficient to promote axon growth on growth inhibitory substrates (Jin and Strittmatter, 1997; Lehmann et al., 1999; Dergham et al., 2002; Niederost et al., 2002; Fournier et al., 2003) and to stimulate axon regeneration after SCI (Hara et al., 2000; Dergham et al., 2002; Fournier et al., 2003). We have shown here that Rho is activated when neuronal like cells are plated on myelin and that Rho is significantly activated in neurons and glia in vivo after SCI. We report that administration of C3–05 reverses Rho inactivation in vivo and protects cells from apoptotic cell death. We further show a role for p75^NTR in mediating Rho activation after SCI.

Activation of Rho in CNS tissue after SCI injury likely results from changes in the local inhibitory and inflammatory environment. Both neurons and glial cells show increased Rho activation. There is evidence that the inhibitory environment of the CNS contributes to increased Rho activation after SCI. First, we showed that neuronal Rho is activated by MAG and myelin when PC-12 cells are plated on inhibitory substrates (Fig. 1). Soluble Nogo fusion proteins can also activate neuronal Rho (Niederost et al., 2002; Fournier et al., 2003). Other evidence indicates that Rho is activated by NgR-independent growth inhibitory proteins. Inactivation of Rho promotes neurite growth on chondroitin sulfate proteoglycans (Dergham et al., 2002) that are present at glial scars. Also, collapsin and ephrins, chemoreplusive factors that act through different receptors, respectively, both activate Rho (Jin and Strittmatter, 1997; Wahl et al., 2000). Preliminary evidence from our lab indicates that astrocytes plated on inhibitory substrates show Rho activation (unpublished data). Therefore, inhibitory proteins may activate Rho in both neurons and glial cells by NgR and NgR-independent mechanisms.

The inflammatory environment may contribute to Rho activation after SCI. Reactive astrocytes secrete TNF, and TNF has been shown to activate Rho in neurons expressing TNF receptors (Neumann et al., 2002). Inflammation after injury is considered to cause secondary damage because it progresses with time and causes continued cell death after the primary traumatic insult (Schwartz and Fehlings, 2002; Popovich and Jones, 2003). Our failure to detect Rho activation in p75^NTR-null mutant mice early after SCI (24 h) suggests activation of NgR signaling to p75^NTR is an early event in CNS injury. However, 3 d after injury in these mice Rho activation was observed, a finding that indicates that at later time points Rho activation is p75^NTR independent. Many factors activate Rho independently of p75^NTR such as semaphorins, ephrins, and thrombin that are known to be present after SCI (Donovan et al., 1997; Wahl et al., 2000; De Winter et al., 2002; Shirvan et al., 2002; Swiercz et al., 2002). The p75^NTR dependence of early Rho activation is interesting because thrombin and TNF, both known to activate Rho, are p75^NTR independent and are present early after SCI (Donovan et al., 1997; Citron et al., 2000; Lee et al., 2000). Our experiments with the p75^NTR knockout mice were with whole tissue homogenates and do not address significant changes in individual cell types early after injury. The massive Rho activation we observe in normal mice and rats after SCI likely represents the combined effects of the many different Rho-activating factors. Secondary damage by inflammation may also contribute to activation of Rho, and if this is the case, then Rho may be an important target to prevent secondary inflammatory damage. Our results after treatment with C3–05 show that inactivation of Rho reduces cell death that follows injury. Further, the massive activation of Rho that we observed after injury was sustained for at least 7 d. Therefore, multiple local signals may activate Rho in CNS cells. We speculate that continued presence of growth inhibitory molecules at the site of a CNS lesion contributes to sustained activation of Rho in neurons and glia after SCI.

Rho-GTPases are known regulators of apoptosis in various cell types. In nonneuronal cells, such as epithelial cells (Fiorentini et al., 1998a,b), endothelial cells (Hippenstiel et al., 2002), T cells (Moorman et al., 1996; Gomez et al., 1997), and some fibroblasts (Bobak et al., 1997), inactivation of Rho causes apoptosis through a Bcl-2–dependent mechanism. In other cells, such as PC-12 cells (Mills et al., 1998) and endogenous cells of the spinal cord, as we show here, inactivation of Rho protects cells from apoptosis. In NIH 3T3 fibroblasts, overexpression of active Rho induces cell death upon serum withdrawal (Jimenez et al., 1995). Thus, the cell background is critical in the effect of Rho signaling and cell death. In PC-12 cells, a neural cell line, Rho proteins have been shown to induce Rho-dependent membrane blebbing (Mills et al., 1998), a morphological characteristic of apoptosis. In cultured astrocytes and hippocampal neurons, treatment with thrombin, a protease found after CNS trauma, causes Rho-dependent apoptosis. Treatment with C3 reversed the thrombin-induced apoptosis of astrocytes and neurons by ∼50% (Donovan et al., 1997). In neurons, TNF activates Rho (Neumann et al., 2002), and antibody-mediated blocking of TNF reduces apoptosis after SCI (Lee et al., 2000). These data support our direct evidence that Rho activation contributes to apoptosis after traumatic SCI. Rho activation in neurons alone may not be sufficient to cause cell death, such as when neurons are plated on inhibitory substrates in culture. Our data suggest that in vivo the combination of multiple Rho-activating factors, including the myelin-derived inhibitory factors, contribute to apoptotic signaling cascades.

Our results show that blocking the activation of Rho after SCI prevents an increased synthesis of p75^NTR protein (Fig. 8 C), implicating Rho activation in the transcriptional changes in p75^NTR expression. This result of C3–05 can be explained by a mechanism in which the change in transcriptional factor activation is Rho dependent (Fig. 9). It is known that Rho is involved in the activation of transcription factors in the nucleus that control synthesis of proapoptotic mRNAs such as c-jun and NFκβ, members of the p75^NTR apoptotic cascades (Aznar and Lacal, 2001; Huang and Reichardt, 2001). Therefore, we speculate that treatment with C3–05 to block Rho activation after injury suppresses apoptosis by preventing the synthesis of proapoptotic proteins such as p75^NTR.

The p75^NTR contributes to initial apoptotic cascades that follow injury in the CNS. In oligodendrocytes, an increase in the expression of p75^NTR after SCI leads to apoptotic cell death (Casha et al., 2001; Beattie et al., 2002). In astrocytes, p75^NTR expression is observed after exposure to inflammatory cytokines (Hutton et al., 1992; Semkova and Krieglstein, 1999), also known to activate Rho (Neumann et al., 2002). In neurons, apoptosis after growth factor deprivation is mediated by p75^NTR (Kaplan and Miller, 2000), and reducing p75^NTR levels prevents death of axotomized neurons (Cheema et al., 1996). Reducing p75^NTR levels reduces apoptosis in contused spinal cord (Brandoli et al., 2001). In p75^NTR−/− mice there is a decrease in apoptosis in the spinal cord (Frade and Barde, 1999). We have shown that both neurons and glia have unusually high levels of active Rho after SCI (Fig. 5 B). Moreover, reversal of Rho with C3–05 reduced apoptosis (Fig. 6 B) and prevented p75^NTR up-regulation (Fig. 8, B and C). Together these results indicate that Rho activation induces apoptosis in a p75^NTR-dependent manner early on after SCI and that inactivation of Rho is cell protective.

Rats were anaesthetized under 2–3% isoflurane. For SCI, adult female Long-Evans rats (200–250 g) underwent laminectomy at thoracic level T10-T11 for SCI at T10. Control animals were sham operations with laminectomy only. Dorsal over-hemisections were done at a depth of 1.6 mm. For contusion experiments, the NYU impactor device was used with 10 g at 25 mm. From 24 h to 7 d after SCI, animals received an overdose of chloral hydrate anaesthetic, were perfused with saline, and the spinal cords were removed. Approximately 5 mm surrounding the injured area was isolated and frozen at –80°C. Balb/c female mice (20–22 g) and p75 knockout mice (Lee et al., 1992) (Jackson mice, stock number 002213) were anaesthetized with hypnorm (20 ml/Kg) and diazepam (1 mg/Kg). Dorsal over-hemisections were performed at T8: n = 5 at 24 h and n = 5 at 3 d. After perfusion with saline, 2–3 mm of spinal cord from the lesion site was removed for analysis. After SCI, bladders of all animals were expressed two to three times per day. Rats were given 5 ml of 0.9% saline subcutaneously twice a day for 1 wk and received daily subcutaneous injections of baytril (10 mg/Kg).

To treat rats with a Rho antagonist, 50 μg of C3–05 was injected in a fibrin matrix (Tisseel kit; Baxter) into transected spinal cord as described (Dergham et al., 2002). In mice, 10 μg of C3–05 in fibrin was injected, except for the experiment for TUNEL labeling where 1 μg in fibrin was injected. C3–05 (50 μg) in PBS without fibrin was injected into rat contusion injury sites. All animal procedures followed guidelines from the Canadian Council of Animal Care.

PC-12 cells were grown in DME with 10% horse serum, 5% FBS, 1% penicillin-streptomycin. PC-12 cells were grown on poly-l-lysine (0.1 μg/ml) (Sigma-Aldrich) or myelin (8 μg per well) or MAG- (8 μg) coated 6-well culture dishes. After the cells settled (3–6 h at 37°C), the media was aspirated, and fresh media containing the C3–05 (1 μg/ml) was added to the undifferentiated cultures. The cells were harvested 24 h later, washed with ice cold TBS, and lysed in modified RIPA buffer (50 mM Tris, pH 7.2, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 500 mM NaCl, 10 mM MgCl₂, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 1 mM PMSF). Cell lysates were clarified by centrifugation at 13,000 g for 10 min at 4°C and kept at −80°C.

Purification of GST-RBD was performed as described previously (Ren and Schwartz, 2000). Bacteria expressing GST-RBD in a pGEX vector (a gift from John Collard, Division of Cell Biology, Netherlands Cancer Institute, Amsterdam, Netherlands) were grown in L-broth with 100 μl/ml ampicillin. Overnight cultures were diluted 1:10 into 3,600 ml L-broth and incubated in a shaking bacterial incubator at 37°C for 2 h. Isopropyl-β-d-thiogalactopyranoside (0.5 mM) was then added to the incubating cultures for 2 h. Bacteria were collected by centrifugation at 5,000 g for 15 min. The pellets were resuspended in 40 ml lysis buffer (50 mM Tris, pH 7.5, 1% Triton X, 150 mM NaCl, 5 mM MgCl₂, 1 mM DTT, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 1 mM PMSF). After sonication, the lysates were spun at 14,000 rpm for 30 min at 4°C. The clarified bacterial lysate was then incubated with glutathione agarose beads (0.6 ml wet volume; preswelled with water) (Sigma-Aldrich) for 60 min at 4°C. The coupled beads were then washed six times in wash buffer (50 mM Tris, pH 7.5, 0.5% Triton X-100, 150 mM NaCl, 5 mM MgCl₂, 1 mM DTT, 1 μg /ml aprotinin, 1 μg/ml leupeptin, and 0.1 mM PMSF) and once in wash buffer containing 10% glycerol. Beads were then resuspended in 8 ml of the wash buffer containing 10% glycerol and stored overnight at −80°C. Frozen tissue was homogenized in modified RIPA buffer (50 mM Tris, pH 7.2, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 500 mM NaCl, 10 mM MgCl₂, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 1 mM PMSF). The homogenates and cell lysates were clarified by two 10-min centrifugations at 13,000 g at 4°C. They were then incubated for 50 min at 4°C with GST-RBD (a gift from John Collard, Division of Cell Biology, Netherlands Cancer Institute, Amsterdam, Netherlands) coupled beads (20–30 μg/sample). The beads were then washed four times and eluted in sample buffer. GTP-bound Rho and total Rho present in tissue homogenates were detected by Western blot. The proteins were transferred to nitrocellulose and probed using a monoclonal RhoA antibody (Santa Cruz Biotechnology, Inc.). Bands were visualized with peroxidase-linked secondary antibodies (Promega) and an HRP-based chemiluminescence reaction (Pierce Chemical Co.). C3–05 was detected using a C3-specific polyclonal antibody (Winton et al., 2002). P75^NTR was detected with a polyclonal antibody raised against p75^NTR (Promega). For all blots, 20 μg of protein was loaded into each lane. Blots were scanned for densitometry using an Epson perfection 1200U scanner, transferred to Adobe Photoshop^® 6.0, and the images were the analyzed with the densitometry IQ MAC 1.2 software (Molecular Dynamics). The software measures the pixel density in the band image after background subtraction, and the densitometry value is in arbitrary units. Statistical tests were performed using In Stat (Graph Pad).

For in situ pull-down assays, rat spinal cord cryosections (16-μm thickness, fresh) were postfixed with 4% PFA and incubated with the clarified bacterial lysate, prepared from bacteria expressing GST-RBD or GST alone, as described above, overnight at 4°C. The sections were then washed three times in TBS, blocked in 3% BSA for 1 h at room temperature, and incubated with anti-GST antibody (New England Biolabs, Inc.) and with cell type–specific antibodies (NeuN, GFAP, and MAB328; Chemicon) or with antibody raised against p75^NTR (Promega) overnight at 4°C. Sections were washed in TBS and incubated for 2 h at room temperature with FITC, Texas red, or rhodamine-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories).

Spinal cord samples of 3 and 4 mm spanning the lesion sites of mice and rats, respectively, were dissected. Normal spinal cord was a 4-mm section from sham control cords. All spinal cord pieces were postfixed in 4% PFA, washed, and embedded in paraffin. Transverse sections of 6-μm thickness were cut, deparaffinized in xylene, and rehydrated by ethanol washes. TUNEL labeling was performed using the Fluorescein-FragEL DNA Fragmentation kit (Oncogene Research Products). The sections were costained with Hoechst 33342 (Sigma-Aldrich), and only TUNEL-positive cells that correlated with Hoechst 33342–stained nuclei were counted. To quantitatively examine the numbers of apoptotic cells, TUNEL-positive cells were counted on sections from control, lesion, and C3–05–treated animals. A blinded researcher counted the total number of TUNEL-positive cells located in the entire transverse section. The average number of TUNEL-positive cells per section was calculated from values obtained by counting 40–50 random sections throughout the lesion site of each animal, with three animals examined per group. The TUNEL-positive cells (green) were distinguished from autofluorescent macrophages (red) through the use of a merge red/green filter. Cells labeled with both TUNEL (green) and p75 (red) were counted in a merge red/green filter after verifying colocalization with Hoechst stain. Values were obtained by counting 20 random sections throughout the lesion site of each animal, with three animals examined per group. Immunohistochemistry with cell type–specific antibodies (NeuN, GFAP, and MAB328; Chemicon) or with a polyclonal C3 antibody (Winton et al., 2002) was performed on paraffin sections. After deparaffinization, transverse sections were treated with 2× saline sodium citrate at 80°C for 20 min. Sections were blocked in TBS containing 3% BSA and 2% goat serum and incubated overnight with primary antibody at 4°C followed by a 2-h incubation with FITC or Texas red–conjugated secondary antibodies (Jackson ImmunoResearch Laboratories). All pictures were taken with northern eclipse software and transferred to Adobe Illustrator^® 9.0.

We gratefully acknowledge Rozica Bolovan and Fan Yang (Bioaxone Therapeutic, Inc., Montréal, Québec, Canada), and Charles Essagian, Leanna Loy, and Pauline Dergham (Université de Montréal) for providing spinal cord injured animals for these studies. We thank John Collard (Division of Cell Biology, Netherlands Cancer Institute, Amsterdam, Netherlands) for providing the RBD-GST construct. We further thank Rozica Bolovan for help with cryostat sectioning and Dana Lasko for her comments on the manuscript. L. McKerracher is affiliated with BioAxone Therapeutic Inc.

This work was funded by the Canadian Institute of Health Research and the Natural Sciences and Engineering Research Council.