Intraspinal microinjection of chondroitinase ABC following injury promotes axonal regeneration out of a peripheral nerve graft bridge

Abstract

Chondroitin sulfate proteoglycans (CSPG) within the glial scar formed after central nervous system (CNS) injury are thought to play a crucial role in regenerative failure. We previously showed that delivery of the CSPG-digesting enzyme chondroitinase ABC (ChABC) via an osmotic minipump allowed axonal regeneration and functional recovery in a peripheral nerve graft (PNG)-bridging model. In this study, we sought to overcome the technical limitations associated with minipumps by microinjecting ChABC directly into the distal lesion site in the PN bridging model. Microinjection of ChABC immediately rostral and caudal to an injury site resulted in extensive CSPG digestion. We also demonstrate that this delivery technique is relatively atraumatic and does not result in a noticeable inflammatory response. Importantly, microinjections of ChABC into the lesion site permitted more regenerating axons to exit a PNG and reenter spinal cord tissue than saline injections. These results are similar to our previous findings when ChABC was delivered via a minipump and suggest that microinjecting ChABC is an effective method of delivering the potentially therapeutic enzyme directly to an injury site.

The glial scar formed after adult central nervous system (CNS) injury is both a chemical and physical barrier to axonal regeneration. The upregulation of various axonal growth inhibitors within the glial scar following injury is viewed as a major impediment for regeneration (Silver and Miller, 2004; Fawcett, 2006). One class of reactive astrocyte-associated inhibitors is the chondroitin sulfate proteoglycans (CSPG) (Morgenstern et al., 2002; Carulli et al., 2005). CSPGs function as potent inhibitory extracellular matrix (ECM) molecules and consist of a protein core to which many large, sulfated glycosaminoglycan (GAG) chains are covalently attached. These GAGs confer most of CSPG's inhibitory properties. The presence of CSPG within the lesion penumbra appears to be an important factor preventing the regeneration of axons (Davies et al., 1999; Tom et al., 2004). Use of the enzyme chondroitinase ABC (ChABC), which cleaves the inhibitory GAGs from the protein core of CSPG, has been shown to promote growth past the lesion and some functional recovery (Bradbury et al., 2002; Caggiano et al., 2005; Fouad et al., 2005; Cafferty et al., 2007). More recently, it was demonstrated that treating a lesion site with ChABC via an osmotic minipump permitted axons regenerating in a peripheral nerve graft (PNG) to exit into host spinal cord tissue, where functional synapses were formed (Houle et al., 2006).

While delivering ChABC with a minipump promoted regeneration, there are inherent limitations with the system. For example, the insertion of the pump into the lesion cavity is blind and air bubbles in the catheter will cause failure of ChABC delivery. We hypothesized that microinjecting ChABC directly into the lesion site would efficiently and more easily deliver ChABC to promote regenerating axons to exit a PNG and reenter the spinal cord.

Adult female Sprague-Dawley rats (225-250g, Charles River) were anesthetized with isoflurane. A hemilaminectomy was performed on the fifth cervical vertebra (C5). The dura was cut to expose the right dorsal spinal cord and a dorsal quadrant lesion (DQ) of approximately 1mm³ was made via aspiration. The dura was closed using a 10-0 suture. One μl of saline (n=5) or ChABC (Seikagaku; 20U/mL, n=3) was injected over 10 minutes using a pulled glass pipette inserted directly into the ventral spinal cord tissue immediately rostral and caudal to the C5DQ. The exposed cord was covered with silastic membrane, the overlying musculature was sutured, and the skin was closed using wound clips. The animals were given ampicillin (200mg/kg) and buprenorphine (0.1mg/kg) and placed on a thermal barrier to recover. Two and four days later, the animals were reanesthetized, the C5DQ exposed, and the injections were repeated in the same general areas. Five days after injury, animals were given an overdose of Euthasol and were transcardially perfused with ice-cold 4% paraformaldehyde (PFA) in PB (pH 7.4).

Twenty-five μm sections containing the C5DQ were cut and collected in series of six. Sections were rinsed in PBS, blocked for 1 hour at room temperature with 5% normal goat serum, 1% bovine serum albumin, ±0.1% Triton X-100 in PBS (depending upon the primary antibody used), and then incubated in rabbit anti-glial acidic fibrillary protein (GFAP; Dako), mouse anti-ED1 (Serotec), and mouse anti-2B6 antibody (recognizes the 4-sugar “stub” following ChABC-digestion of CSPG; Seikagaku) overnight at 4°C. The sections were rinsed and incubated in the appropriate pre-absorbed, fluorescently-tagged secondary antibody overnight at 4°C. The sections were rinsed again, mounted on glass slides, coverslipped with VectaShield, and sealed with nail polish.

To quantify ED1 expression, four sections from the center of the lesion per animal were selected. Images were taken and montaged together so that the entire half of the section unilateral to the lesion was visible (Corel Draw). Each montaged image was thresholded (MetaMorph). A template consisting of four regions was laid over each thresholded image. Each region was approximately 1mm × 250μm, and the regions were stacked so that the entire set of all four regions would span an area of 1mm × 1mm. The top-most side of the first region ran tangentially to the ventral edge of the C5DQ. The thresholded pixel area in each region was then calculated (MetaMorph). The pixel areas for each region were analyzed for statistical significance by performing a student's t-tests (Microsoft Excel).

To determine the effect of microinjections of ChABC on regeneration out of a PNG, adult rats (n=12) were anesthetized, the caudal end of the second and the entire third cervical vertebrae (C3) was removed, and the dura was cut to expose the dorsal spinal cord. The entire right half of the C3 spinal cord was aspirated out, resulting in a cavity that was 2-3 mm in length. Predegenerated tibial nerves were isolated from donor rats and ∼2-3mm of the perineurium was reflected from the proximal nerve end. This end was placed into the C3 cavity site and secured in place by suturing the perineurium to the dura. The distal end of the PNG was left free outside of the spinal column along the C4-C8 vertebrae. The dura and overlying musculature were closed using sutures, and the skin was closed using wound clips. Two weeks later, the animals were anesthetized and a C5DQ was made as described above. One μl of saline (n=8) or ChABC (20U/mL, n=4) was injected rostral and caudal to the C5 injury as described above. The injections were repeated two and four days later. After the injection on day 4, the dura above the C5DQ was reopened. The free, distal end of the PNG was trimmed by 1mm and inserted into the C5DQ cavity where it was secured in place by suturing the perineurium to the dura. Silastic membrane was placed over the graft. After the musculature was sutured, the skin was closed using wound clips. After at least 3 weeks, the animals were anesthetized and the cord and graft were exposed. The graft was cut and soaked with 10% biotinylated dextran amine (BDA; 3000MW; Molecular Probes) to label regenerating axons by tracer diffusion though the distal end of the graft. Two days later, the animals were perfused with 4% PFA.

BDA-labeled axons were visualized by incubating sections in Avidin-HRP for 3 hours at room temperature, rinsed in PBS, and then reacted with diaminobenzidine (DAB). Sections were rinsed again, mounted, counterstained with thionin, and coverslipped. To quantify the number of BDA+ axons emerging from the graft and back into spinal cord tissue, four representative DAB-reacted sections from each animal were analyzed. A virtual line 500μm ventral to the PNG/spinal cord interface was made and the number of BDA+ labeled profiles that intersected with the line was counted. The numbers for each treatment group were analyzed for statistical significance by performing a student's t-tests (Microsoft Excel).

We wanted to determine if the microinjection technique was traumatic to the tissue, which would limit its therapeutic potential. Therefore, we examined the inflammatory response at a subacute time point after injury and several injections. Five days after the lesion in the saline-treated spinal cords, ED1+ macrophages/microglia were found primarily in the lesion site (Fig. 1B, C, lesion site denoted by asterisk in Fig. 1A) and tissue just adjacent to it. The phagocytic cells did not appear to migrate much further than the lesion cavity wall that was surrounded by reactive, GFAP+ astrocytes (Fig. 1A, C). Microinjection of ChABC produced a similar response. As in saline-treated sections, ED1+ cells were located primarily in the lesion cavity (Fig 1E, F, lesion site denoted by asterisk in Fig. 1D) and did not extend very far past the cavity wall that was surrounded by many GFAP+ astrocytes (Fig. 1D, F). Based upon densitometric measurements, there was no difference in ED1+ immunoreactivity between saline and ChABC-treated tissue (Fig. 1G).

Figure 1.

Microinjections of saline and ChABC produces no noticeable inflammatory response. A-C) A section from a saline-treated animal 5 days after a C5DQ lesion was made. There were many reactive GFAP+ astrocytes (A,C, green) ipsilateral to the lesion (asterisk). ED1+ cells were found primarily within the lesion cavity with very little staining in ventral tissue. D-F) In sections from animals treated with ChABC, there were many reactive GFAP+ astrocytes (D,F, green) adjacent to the lesion cavity (D, asterisk) and in tissue ventral to the injury, similar to saline-treated tissue. ED1+ cells were found primarily in the lesion site and not in ventral areas (E,F, red), also similar to control, saline-injected animals. G) Saline-treated (white bars) and ChABC-treated sections (black bars) had the most ED1+ cells in the area nearest to the lesion boundary. There was no significant difference in the amount of ED1+ staining with ChABC treatment. HI) Sections from saline- and ChABC-treated animals were stained for the CSPG “stub” antibody 2B6. The inverted fluorescent images are shown. There is very little staining in saline-treated sections (H), similar to what was seen in sections that were incubated in secondary antibody only (not shown). In animals that were treated with ChABC, strong 2B6 staining was found throughout the tissue ipsilateral to the lesion (asterisk) and to some extent contralateral to the lesion (I). The 2B6 staining extended to the lesion wall (I, arrowhead). Scale bar: 1mm.

To determine how effective microinjections were in delivering ChABC, we examined the resulting CSPG digestion pattern using the 2B6 antibody that recognizes the 4-sugar “stub” that remains following ChABC-digestion. Tissue that had been treated with ChABC had abundant 2B6 immunoreactivity within ipsilateral grey and white matter, including the lesion penumbra (Fig. 1I, arrowhead) and some of the contralateral tissue (Fig. 1I). Although there was some immunoreactivity for 2B6 in sections from saline-treated animals even though no ChABC was used (Fig. 1H), the staining pattern was identical to that of control sections incubated with secondary antibodies only (data not shown), suggesting that this pattern is non-specific binding due to the “sticky” nature of the scar region. Furthermore, there is strikingly less 2B6+ staining than in ChABC-treated sections.

We wanted to determine if digestion of scar-associated CSPG with microinjections of ChABC was sufficient to allow for increased regeneration in a PNG bridge model. The axons regenerating through the PNG were labeled with BDA. Very few BDA+ axons emerged from the PNG with saline treatment (Fig. 2A, A'). Significantly more labeled axons were found 500μm beyond the PNG-spinal cord interface following ChABC treatment (Fig. 2B, B', C, p<0.00001).

Figure 2.

Microinjections of ChABC promotes axon regeneration from the PNG into host tissue. A-B) Sections stained for BDA and counterstained with thionin. Tracings of the BDA+ axons are shown in A'-B'. There were abundant BDA+ fibers within the PNG in all of the sections. In the saline-treated animals, very few axons were able to extend past the PNG-spinal cord interface (A, A', C). However significantly more axons were able to extend past the graft-host interface into ventral tissue in ChABC- treated animals (B, B', C). There were significantly more axons seen at 500μm beyond the interface in ChABC-treated animals than saline-treated animals (* = p<0.00001). Scale bar: 500μm.

CSPGs associated with the glial scar appear to play a major role in regenerative failure (Morgenstern et al., 2002; Jones et al., 2003; Fawcett, 2006; Yiu and He, 2006; Galtrey and Fawcett, 2007). When injured axons encounter the inhibitory nature of the scar conferred by molecules such as CSPG they become dystrophic (Tom et al., 2004) and in some instances even turn around, as if deflected by the scar (Davies et al., 1997; Davies et al., 1999). Moreover, digestion of the GAG chains from CSPGs with ChABC (reviewed by (Crespo et al., 2007)) enhances regeneration and sprouting following injury (Bradbury et al., 2002; Caggiano et al., 2005; Ikegami et al., 2005; Barritt et al., 2006), further demonstrating the importance of this family of scar-associated molecules. When CSPGs were digested by infusing ChABC at the interface of a Schwann cell and olfactory ensheathing cell graft and spinal cord or an embryonic tissue transplant, more serotonergic fibers were seen caudal to the injury (Fouad et al., 2005; Kim et al., 2006).

Recently, our group has shown that continuous, local delivery of ChABC with an osmotic minipump in a PNG bridging model allowed for more axons to exit the graft and reenter host spinal cord, where they reestablished synaptic contacts and mediated some functional recovery (Houle et al., 2006). Although we had success with this approach, there were several technical limitations with using an osmotic minipump for ChABC delivery. For instance, there is debate about whether ChABC that is stored within the pump at body temperature remains biologically active over a 5-day infusion period (Curinga et al., 2007; Tester et al., 2007). Air bubbles within the catheter would prevent accurate delivery of enzyme. Additionally, the enzyme was likely diluted by cerebrospinal fluid, so the final active concentration that penetrated the injury site is unknown. Furthermore, while we previously used a cannula to deliver ChABC, other studies used an intrathecal catheter to distribute the enzyme (Bradbury et al., 2002; Caggiano et al., 2005; Fouad et al., 2005; Barritt et al., 2006; Kim et al., 2006), which may produce additional damage to the spinal cord (Jones and Tuszynski, 2001). We hypothesized that microinjections of fresh, concentrated ChABC directly into spinal cord tissue would produce minimal tissue trauma and would achieve similar results to continuous infusion of more diluted ChABC over the same time.

Microinjections of ChABC produced extensive CSPG digestion throughout both grey and white matter. The staining pattern of the CSPG stub antibody 2B6 was broader than that described with intrathecal ChABC delivery (Bradbury et al., 2002; Barritt et al., 2006), indicating that microinjecting ChABC is a very effective and efficient method for delivery. Furthermore, we observed more axonal growth from the graft with ChABC than saline. Many of these regenerating axons co-labeled with the tracer BDA and synapsin (data not shown), suggesting that the fibers may be functionally active, similar to what was seen in our previous study (Houle et al., 2006). ChABC-mediated digestion of perineuronal nets, which has been shown to increase synaptic plasticity (Pizzorusso et al., 2002; Massey et al., 2006; Pizzorusso et al., 2006), would also increase the probability that BDA+ regenerating axons would reestablish synaptic contacts. However, we did not test the ChABC-treated animals behaviorally to determine definitively that these axons do mediate functional recovery. That study will be conducted in the future. The finding that there were virtually no inflammatory cells ventral to the C5DQ lesion cavity in either saline- or ChABC-injected animals suggests that the microinjection technique is relatively atraumatic, despite the repeated injections. Thus, this technique is an excellent alternative to using intrathecal delivery systems to treat injured CNS tissue with ChABC to effectively achieve widespread CSPG digestion and to promote axonal regeneration.