|Home | About | Journals | Submit | Contact Us | Français|
Increased chondroitin sulfate proteoglycan (CSPG) expression in the vicinity of a spinal cord injury (SCI) is a primary participant in axonal regeneration failure. However, the presence of similar increases of CSPG expression in denervated synaptic targets well away from the primary lesion and the subsequent impact on regenerating axons attempting to approach deafferented neurons has not been studied. Constitutively expressed CSPGs within the extracellular matrix and perineuronal nets of the adult rat dorsal column nuclei (DCN) were characterized using real-time PCR, western blot analysis and immunohistochemistry. We show for the first time that by 2 days and through 3 weeks following SCI, the levels of NG2, neurocan and brevican associated with reactive glia throughout the DCN were dramatically increased throughout the DCN despite being well beyond areas of trauma-induced blood brain barrier breakdown. Importantly, regenerating axons from adult sensory neurons microtransplanted 2 weeks following SCI between the injury site and the DCN were able to regenerate rapidly within white matter (as shown previously by Davies et al., 1999) but were unable to enter the denervated DCN. Application of chondroitinase ABC or neurotrophin 3-expressing lentivirus in the DCN partially overcame this inhibition. When the treatments were combined, entrance by regenerating axons into the DCN was significantly augmented. These results demonstrate both an additional challenge and potential treatment strategy for successful functional pathway reconstruction after SCI.
For regeneration of injured mammalian CNS axons to be meaningful, functional synapses must be restored with target neurons. Increasing evidence suggests that following CNS lesions, several chemical mediators of gliosis that accompany the inflammatory response following breakdown of the blood brain barrier (Fitch and Silver, 1997; Rhodes and Fawcett, 2004; Silver and Miller, 2004) locally stimulate the increased expression of a number of chondroitin sulfate proteoglycans (CSPGs) (Johns et al., 1992; McKeon et al., 1995; Asher et al., 2000; Smith and Strunz, 2005). The increased expression of CSPGs in and around brain or spinal cord injury (SCI) sites is a major contributor to the failure of spontaneous axonal regeneration (McKeon et al., 1991; Fitch and Silver, 1997; Bruce et al., 2000; Jones et al., 2002; Jones et al., 2003; Tang et al., 2003). The mechanism by which these macromolecules inhibit regenerating axons is incompletely understood but may be the result of disrupted adhesive interactions (Grumet et al., 1996; Milev et al., 1996), decreased diffusion of secreted molecules (Gruskin et al., 2003) and altered intracellular signaling (Sivasankaran et al., 2004; Zhou et al., 2006). In the absence of CSPGs (Bradbury et al., 2002; Davies et al., 2004; Grimpe and Silver, 2004) or following the application of neurotrophins (Oudega and Hagg, 1996; Bradbury et al., 1999; Oudega and Hagg, 1999; Lu et al., 2004; Lu et al., 2006; Taylor et al., 2006), injured sensory axons are able to regenerate past SCI sites and through degenerating myelin debris (Bradbury et al., 1999; 2002; Davies et al., 1999; 2004).
Following CNS injury, inflammatory activation of astrocytes and microglia is not solely limited to areas immediately surrounding the sites of trauma but is also present around denervated target neurons (Koshinaga and Whittemore, 1995; Deller and Frotscher, 1997; Aldskogius and Kozlova, 1998; Liu et al., 1998). It is not known whether these activated glial cells intermingled between neurons denervated by SCI (Koshinaga and Whittemore, 1995), might be capable of actually preventing regenerating sensory axons in the dorsal columns from re-entering their proper synaptic targets by further saturating the extracellular matrix with newly secreted CSPGs. In the normal adult CNS, CSPGs can contribute to the formation of the specialized lattice-like perineuronal net that cloaks neuronal cell bodies and is thought to inhibit synaptic plasticity (Golgi, 1893; Svensson et al., 1984; Lander et al., 1997; Bruckner et al., 1998; Takahashi-Iwanaga et al., 1998; Pizzorusso et al., 2002; Corvetti and Rossi, 2005). We recently demonstrated that collateral sprouting by spared primary afferents inside the partially denervated adult rat brainstem cuneate nucleus following cervical SCI is inhibited by CSPGs within the nucleus and is promoted by their local digestion (Massey et al., 2006). Here, using a combination of quantitative real time polymerase chain reaction, immunoblotting and immunohistochemistry we show, for the first time, that cervical SCI results in the significantly increased expression of the CSPGs NG2, neurocan, and brevican in the distant denervated DCN. Furthermore, as is the case at sites of direct CNS injury, these CSPGs present a potent barrier to reinnervation by regenerating axons from microtransplanted adult sensory neurons following SCI that can be overcome by chondroitinase ABC (chABC) application and increased neurotrophin-3 (NT-3) expression within the nucleus. Portions of these results have been previously reported in abstract form (Massey et al., 2005).
All animal manipulation procedures complied with NIH regulations and were approved by the Institutional Animal Care and Use Committee guidelines at both the University of Louisville and Case Western Reserve University School of Medicine. Adult male (275–350 g) Sprague Dawley rats were used for these experiments. Each rat was anesthetized with an intramuscular injection of a mixture of ketamine hydrochloride (50 mg/kg), xylazine (6.5 mg/kg) and acepromazine (2.5 mg/kg) for each surgical procedure.
To anterogradely trace primary afferent axon terminals in the gracile and cuneate nuclei, 5–10 μl of 1% CTB (Sigma-Aldrich, St. Louis, MO), dissolved in 0.25 M Tris HCl, pH 7.4, was injected into the right forelimb median and ulnar nerves and left hindlimb sciatic nerves using a Hamilton syringe with a 30 gauge needle (Webster and Kemplay, 1987; Maslany et al., 1991; Onifer et al., 2005). A nerve crush procedure was not used for the CTB injections, to avoid increased CTB uptake by nociceptive dorsal root neurons (Tong et al., 1999). CTB tracing prior to cervical SCI allowed us to both visualize the degenerating primary afferent terminals (Sayer et al., 2002) and to verify the completeness of the SCI (Onifer et al., 2005).
Rats subjected to SCI were selected at random one week following CTB injections. The dorsal surfaces of the C3-4 spinal cord segments were exposed by laminectomy of the C3 and C4 vertebrae and an incision was made in the underlying dura. A 1.5 mm deep transverse lesion between the C3 and C4 dorsal roots was made with a Vibraknife™ as previously described (Onifer et al., 2005), severing most primary afferents in the dorsal column-DCN pathway.
A 15 mm incision was made over the T2 spinous process in the mid-dorsal line between the scapulae. The dorsal surface of the T1 spinal cord segment was exposed by laminectomy. A pair of ultrafine microforceps was then inserted through the dura into the spinal cord parenchyma to a depth of 1 mm between the median dorsal raphe and the right dorsal root entry zone. The forceps were closed tightly twice for a total of 10 seconds. Great care was taken to avoid additional trauma to the cord or roots and the tips of the forceps were not inserted deeper than 1 mm. The animal recovered for 2 weeks before receiving DRG microtransplants (described below).
Tissues were isolated from randomly chosen normal rats (n=6) and rats 2 days (n=3), 1 week (n=3) or 3 weeks (n=6) following cervical SCI. Each rat was transcardially perfused with 100 ml of cold 0.1 M phosphate buffer, pH 7.4, (PB). The right and left dorsal column nuclei, the right and left visual cortices and the right and left sides of 2mm of the C3-4 spinal cord segments, spanning the SCI site, were carefully dissected, placed immediately in liquid nitrogen and stored at −80°C. The tissue from one side of each rat was used for mRNA analysis and tissue from the opposite side was used for Western blot analysis as described below.
DCN, visual cortices, and spinal cord tissues from one side of each rat were quickly thawed and then homogenized in 1 ml of QIAzol Lysis Reagent (Qiagen, Valencia, CA) using a PolyTron homogenizer. Total RNA was prepared with the RNeasy Lipid Tissue Mini Kit (Qiagen) according to the recommended protocol. The quality and quantity of total RNA was determined by spectrophotometry and agarose gel electrophoresis. There was no detectable difference in 28S and 18S rRNA between normal and injured rats.
One microgram of total RNA was reverse transcribed using the high-capacity cDNA archive kit (Applied Biosystem, Foster City, CA) according to the recommended protocol. Probes used for TaqMan quantitative rtPCR were selected for NG2 (Rn00578849), neurocan (Rn01642355), brevican (Rn01495414), aggrecan (Rn00672903) and glial fibrillary acidic protein (GFAP) (Rn00566603) from the pre-validated Assays-on-Demand (Applied Biosystems, Foster City, CA) library. Individual probes were used in a 40 μl reaction utilizing TaqMan PCR Master Mix and subjected to 40 cycles of real time-PCR (ABI 7300, Applied Biosystem, Foster City, CA). The 18S ribosomal RNA was used as an internal control in the reaction. Tissue samples used for comparing each individual gene were run in the same reaction with a common reaction mixture. Each sample was run in triplicate for both the gene of interest and the 18s control gene. The Comparative CT Method (DDCT Method) was used to calculate changes in gene expression. The mean critical threshold (Ct) was calculated for the gene of interest and 18S genes to determine a relative expression value. Individual rat’s Ct values expressed as fold change from normal were then compared by 1-way ANOVA followed by post-hoc Tukey’s HSD comparison tests using the statistical software SPSS v.13.0 (SPSS, Chicago, IL) and Sigma Plot v.8.0 (Systat, Point Richmond, CA).
The following antibodies were used in this study.: anti-NG2 (clone 7.1, Chemicon), anti-neurocan C-terminal epitope (clone 650.24, which detects both the C-terminal and full-length neurocan isoforms, Chemicon, Temecula CA); anti-neurocan N-terminal epitope (clone 1F6, which labels the N-terminal and full-length neurocan isoforms, Developmental Studies Hybridoma Bank, Iowa City, IA), anti-brevican core protein (rabbit polyclonal antibody B6, which recognizes both the full length and C-terminal brevican isoforms), anti-brevican N-terminal ADAMTS4/5 cleavage product (rabbit polyclonal antibody B50, Viapiano et al., 2003), anti-aggrecan core protein (clone Cat-301, Matthews et al. 2002), anti-GFAP (clone GA5, Sigma-Aldrich) and anti-beta tubulin (Invitrogen, Carlsbad CA). Frozen tissue samples from spinal cord, DCN and visual cortex were thawed on ice and homogenized in 10 volumes of 40 mM TrisHCl, pH 7.6, containing 40 mM sodium acetate and a protease inhibitor cocktail (Complete, Roche Applied Science, Indianapolis, IN). Homogenization was performed on ice by three 5sec bursts of tissue sonication, using a 6.5 mm microprobe at a power of 35 W. Tissue homogenates were subsequently passed through a 30G needle to retain and disrupt unbroken debris. Aliquots of the homogenates at a final protein concentration of 2–3 mg/ml were treated with 0.3 U/ml of protease-free chondroitinase ABC (chABC, Seikagaku, Tokyo, Japan) for 8h at 37°C. Chondroitinase activity was stopped by boiling the samples in the presence of 1X gel-loading buffer. Samples containing 10 μg total protein were electrophoresed on reducing 3–10% SDS-polyacrylamide gels and analyzed by Western blotting. Immunoblots were developed by chemiluminescence and the integrated optical density (O.D.) of each target protein was imaged and quantified using the Gel-Pro Analyzer software (v3.1, Media Cybernetics, Silver Spring MD). Blots were stripped and re-probed for beta tubulin as loading control. O.D. ratios (O.D. target protein/O.D. beta tubulin) for each protein and tissue region were compared by 1-way ANOVA followed by Tukey’s multiple comparison test using SPSS v.13.0 and Sigma Plot v.8.0 statistical software.
Two weeks after thoracic SCI, DRG neurons from adult C57Bl/6N mice that constitutively express green fluorescent protein (GFP) were microtransplanted as previously described (Davies et al., 1999; Grimpe and Silver, 2004). Briefly, on the day of microtransplantation, mice were terminally anesthetized with isoflurane and decapitated. Approximately 35–40 cervical, thoracic, and lumbar DRG were dissected. They were dissociated for 60 minutes in a solution containing dispase (Roche; Indianapolis, IN) and collagenase (Worthington. Lakewood, NJ). Ganglia were triturated in the presence of 25 μL of 5 mg/mL deoxyribonuclease (DNase, Sigma-Aldrich). The resulting single cell suspension was spun at 3000g for three minutes and the cells resuspended in L-15 media (Gibco, Invitrogen), at an approximate density of 1000 neurons/μL. Five μL of the DNase solution was added to the cells, which were kept on ice until transplantation.
Each injured recipient rat was anesthetized and placed into a stereotaxic frame. The dorsal surfaces of the rostral cervical spinal cord and the caudal medulla were exposed. The cell suspension was transplanted through a pipette inserted into the right fasciculus gracilis, 0.2–0.3 mm below the pial surface at about 1 mm caudal to the gracile nucleus, using a Nanoject II delivery system. All transplants used a total volume of 690nl, delivered over ten minutes as previously described (Davies et al., 1997; 1999; Grimpe and Silver, 2004). The pipette was withdrawn five minutes after the final injection. Following transplantation, rats received either no further treatment (n=6), a single injection of chABC (n=4), a single injection of NT-3 lentivirus (n=5) or combined injections of chABC and NT-3 lentivirus (n=6). All animals were sacrificed 10 days following microtransplantation.
A pipette was inserted to a depth of 0.3mm at the lateral junction of the right fasciculus gracilus with the right gracile nucleus gracilis and 690nl of protease-free chABC (20U/mL; Seikagaku America, Falmouth, MA) using a Nanojet II delivery system (Drummond Scientific, Broomall, PA; Massey et al., 2006). A piece of Duragen™ (Integra, Plainsboro, New Jersey) covered the dural defect.
The vector backbone for the self inactivating lentiviral vector was a gift from S. Gerson at Case Western Reserve University (Zufferey et al., 1998). Sequences for cloning of the gene encoding full length murine NT3 (777bp) into this vector were provided by G. Landreth and S. O’Gorman at Case Western Reserve University. All vectors were sequenced after cloning to ensure sequence fidelity.
Virions were prepared by triple transfection of packaging, envelope, and transducing vectors into cultured human 293T cells. Virus-enriched media was collected 48 hours post- transfection. Viral particles were concentrated by passing the media through a Microkos ultrafiltration unit (Spectrum Labs, Rancho Dominguez, CA).
Viral titers of the NT-3 containing vectors were determined indirectly using a reverse transcriptase (RT) assay due to the absence of reporter genes. The titer of the GFP-expressing vector was calculated by FACS sorting and quantification of infected cells. Since both vectors had the same sequence, it was assumed that the level of RT activity would be equivalent and would simply reflect the concentration of virus present. The amount of radionucleotide incorporation present in 10 μl aliquots of the GFP-expressing virus with a known titer was compared to the amount of incorporation present in 10 μl aliquots of the NT-3-expressing vectors with unknown titers. Five separate measurements with different dilutions were taken from each stock solution for the assay. Quantification was provided via scintillation counts over three minutes.
Adult cortical astrocytes were plated onto poly-lysine coated 6-well culture plates. Cells were transduced with lentiviral constructs containing NT3-lentivirus or GFP-lentivirus, both adjusted to 2 × 108 infectious units (IU) per ml. Mock transducions were performed with medium containing 1× polybrine (6ug/ml). Twenty four hours post transduction, culture medium was changed to fresh Eagle medium containing fetal bovine serum (FBS). Four days post transduction, DRG from E15 rats were prepared as described by Davies et al (1994). Four thousand DRG cells were plated onto poly-lysine-coated glass coverslips in 24-well plates. A total of 280ul of conditioned medium from either the transduced or mock transduced astrocytes and 20ul of fresh Eagle medium with serum were added to each well. Each condition was performed at least in triplicate.
Two days after plating, DRG cells were washed and fixed in buffered 4% paraformaldehyde (PFA) for 30 minutes. Fixed cells were blocked at room temperature in PBS containing 5% normal goat serum, 0.1% v/v Triton X-100 and 0.1% w/v bovine serum albumin for 2 hours, before incubation with a mouse monoclonal anti-beta tubulin antibody (Sigma-Aldrich), at a 1:1000 dilution. Slides were incubated with primary antibody overnight, washed three times with PBS and stained with a goat-anti mouse antibody conjugated with the fluorochrome Alexa-594 (Invitrogen, dilution 1:500). The secondary antibody was also incubated overnight, followed by washing and mounting of the coverslips using Citifluor. The surviving number of cells per a well was counted using an Olympus microscope equipped with fluorescence. Results from these experiments were then analyzed by 1-way ANOVA followed by Tukey’s multiple comparison test using SPSS v.13.0 and Sigma Plot v.8.0 statistical software.
Injured rats were anesthetized 12 days after the initial injury and placed in a stereotaxic frame. The dorsal surface of the caudal medulla was exposed and a pipette inserted to a depth of 0.3 mm into the medial zone of the right gracile nucleus. Six hundred and ninety nl of viral supernatant (2 × 108 IU/ml) were injected using a Nanoject II delivery system. A second 690 nl injection was made into the lateral zone of the right gracile nucleus. After each injection, the pipette was allowed to remain in place for 5 minutes. Following surgery, animals were isolated for 48 hours before receiving DRG microtransplants with or without concomitant chABC delivery.
Tissue samples from rats receiving either saline injections (n=3) or injections of NT-3 expressing lentivirus (n=3) ten days earlier were prepared by sonication in lysis buffer (137 mM NaCl, 20 mM Tris-HCl, 1% v/v Nonidet-P40, 10% v/v glycerol and a cocktail of protease inhibitors, pH 7.4), followed by centrifugation and collection of the resulting supernatant. Protein concentration was determined by the Bradford method and equal amounts of each sample (0.2 mg/ml) were analyzed in triplicate by ELISA 96-well plates (Nunc, Rochester, NY) coated with anti-human NT-3 polyclonal antibody (Chemicon) diluted 1:5000 in 0.025 M sodium bicarbonate/0.025 M sodium carbonate, pH = 9.7, overnight at 4° C. Coated plates were blocked for one hour at room temperature, followed by incubation with experimental and control samples overnight at 4° C. NT-3 standards ranging from 4.68 to 300 pg/ml were used to generate a standard curve and run simultaneously with the samples. Following incubation with tissue, the plates were extensively washed with 20 mM Tris-HCl, 150 mM NaCl, 0.05% v/v Tween-20 (TBST). A mouse anti-NT-3 antibody (Chemicon), diluted 1:4000 in blocking buffer, was added to the wells and incubated overnight at 4° C. After washing with TBST, an anti-mouse secondary antibody conjugated to horseradish peroxidase (HRP) diluted 1:1000 in blocking buffer was added to each well and incubated for 2.5 hrs. Following extensive washing, 100 μl of the HRP substrate tetramethylbenzidine was added to each well and allowed to develop. Color development was stopped by adding 100 μl of 1 N HCl to each well, and absorbances were recorded at 450 nm using a microplate reader.
Control, injured and injured/microtransplanted animals were terminally anesthetized by pentobarbital overdose and perfused with oxygenated, heparinized, and calcium-free Tyrodes solution or 0.1M PBS, followed by 4% PFA in 0.1M PB. Cervical spinal cord and brainstem were removed from each rat and cryoprotected at 4°C in PB containing 30% sucrose for 3 to 4 days. Tissues were cryosectioned at 20 μm in the transverse plane and stored at −80°C. Brainstems from rats that received microtransplants were sectioned in the sagittal plane at 50 μm using a Vibratome and transferred to multiwell plates as free floating sections into 0.1M PBS. These sections were mounted onto microscope slides, coverslipped and inspected for fluorescent DRG microtransplants.
The following antibodies and probes were used for histological detection of CSPGs: anti-NG2 (clone 7.1, 1:200 dilution; Chemicon), anti-neurocan (clones 650.24, 1:500 and 1F6, 1:200), and anti-brevican core protein (rabbit polyclonal antibody B756, directed against aminoacids 420–433 of rat brevican, 1:500), and anti-aggrecan (clone Cat-301, 1:200). Additionally, CS chains were detected with Wisteria floribunda (WFA) lectin (1:100, Sigma-Aldrich) as described (Bruckner et. al., 1998; Massey et al., 2006).
To identify the location of the DCN, sections were processed for cytochrome oxidase activity (Wong-Riley, 1979; Crockett et al., 1993). Sections were incubated in a solution of 10% sucrose, 0.1M PBS and dimethyl sulfoxide (Sigma-Aldrich) for 10 minutes. These sections were then reacted with ?% 3,3′-diaminobenzidine tetrahydrochloride (Sigma-Aldrich) and 0.03% cytochrome C (Sigma- Aldrich) at 37°C for 1 to 2 hours before being rinsed two times in 0.1M PB and coverslipped.
Antibodies to beta-tubulin III (BTIII) (1:1000; Covance Research Products, Inc., Berkeley, CA), parvalbumin (1:400; Chemicon), neuronal-specific nuclear protein (NeuN) (1:500; Chemicon) or microtubule-associated protein 2 (MAP2) (1:500; Sigma-Aldrich) were used to observe neurons and their processes (Massey et al., 2006). Antibodies to GFAP (1:2000; Chemicon) and to the iC3b (CD11b) complement receptor (OX42) (1:500; Serotec, Raleigh, NC) were used to visualize astrocytes and microglia, respectively. Goat anti-CTB (1:2000; List Biological Laboratories, Hornby, Ontario, Canada) was used to see CTB-traced primary afferent axon terminals (Massey et al., 2006). Visualization of GFP expressed by microtransplanted mouse DRG neurons was enhanced by reacting sections containing the right DCN with an anti-GFP antibody (1:1000; Chemicon; Davies et al, 1999).
The brainstem and spinal cord sections were incubated with TBS (pH 7.4) containing 10% normal donkey serum (Jackson ImmunoScientific, Inc.) or 5% normal goat serum with 0.1% bovine serum albumin and combinations of the primary antibodies. Sections were then reacted with species-specific donkey IgG (Jackson ImmunoScientific, Inc.) and goat IgG (Invitrogen) secondary antibodies conjugated with either fluorescein isothiocyanate (FITC), Alexa-fluor 488, cyanine 3 (Cy3), Alexa-fluor 594 or 7-amino-4-methyl-3-coumarinylacetic acid (AMCA). Confocal or epifluorecent images of immunoreactivity (IR) from coverslipped sections were obtained using an Olympus Laser Confocal microscope and digitized with an Olympus Optical (Melville, NY) 3 Laser Fluoview 500 or obtained using a Nikon Eclipse E400 (Tokyo, Japan) microscope connected to a SPOT Insight 3.2 color camera.
Separate images of the brainstems from injured rats that received microtransplants were obtained for each fluorochrome and digitally merged using SPOT Advanced software (Nikon). The NeuN immunoreactivity was used to identify neurons in the gracile nuclei in the caudal medulla along with the rest of the neuronal groupings in this area. The ability of GFP-immunoreactive neuritic processes from the transplanted DRG neurons to enter the nucleus gracilis was quantified by measuring the total length of transplant axons within the host gracile nucleus using Metamorph® software (Molecular Devices, Sunnyvale, CA). Statistical comparisons between treatments were performed using SPSS v.13.0 and Sigma Plot v.8.0 statistical software.
Successful treatments developed to repair axonal pathways damaged by SCI ultimately will need to result in regeneration and functional reinnervation of target neurons. Our goal was to investigate whether the gliosis induced by denervation of the adult rat DCN following SCI is accompanied by increased CSPG expression that inhibits anatomical axonal regeneration.
To quantify changes in CSPG gene regulation and expression following denervating SCI, DCN tissue was obtained from normal rats (n=6) and from rats 2 days (n=3), 1 week (n=3) and 3 weeks (n=6) after cervical spinal cord dorsal hemisection, messenger RNA (mRNA) and protein levels were measured in tissue taken from the opposite sides of each rat (Figs. 1 & 2). Tissues from the SCI site and visual cortex of each rat used for the DCN assessments were also taken and analyzed as positive and negative controls, respectively. The changes in CSPG expression at SCI sites observed in our study were consistent (Supplementary Materials, Fig. S1) with those previously reported (Fitch and Silver, 1997; Jones et al., 2002; Jones et al., 2003; Tang et al., 2003). There were no changes in CSPG expression in the visual cortices at any time point (Supplementary Materials, Fig. S2).
To assess NG2 expression, tissue extracts were probed with an antibody (clone 7.1) that recognizes both the membrane bound 300kDa and truncated 275kDa NG2 isoforms (Nishiyama et al., 1995). In tissues from the DCN of uninjured rats, most of the NG2 protein was detected in a single 300kDa band (Fig. 1). Significant increases in total NG2 protein were seen in DCN tissue within 2 days following SCI (Figs. 1, ,2F).2F). This increased NG2 protein (~2 fold) occurred in the presence of only slightly increased mRNA expression (~0.5 fold, Fig. 2E) and returned to normal levels within the first week post injury (Figs. 1, ,2F).2F). Interestingly, nearly all of the increased NG2 protein, both in the DCN and at the SCI site in our control experiments was in the form of a normally undetectable 275Kd isoform (Figs. 1, 2F)2F) that has been described as a shed, soluble cleavage product (Nishiyama et al, 1995).
Increases in the gene regulation and expression of both GFAP and neurocan mRNA were also clearly detected 48 hours post-SCI injury in the denervated DCN (Figs. 2A, G). Subsequently, the levels of both mRNAs dropped by 1 week post injury (Figs. 2A, G), although they remained elevated above control values throughout the time-course of this study. In agreement with previous literature (Matsui et al., 1994), neurocan protein expression in the normal DCN was predominantly found as C-terminal (650.24) and N-terminal (1F6) proteolytic cleavage fragments of the full-length protein (Rauch et al, 1992, 1996), with apparent Mw 150 and 130 kDa respectively (Fig. 1). Little of the unprocessed 245 kDa full-length neurocan protein, containing both the C-terminal and N-terminal eptitopes, was present in the uninjured animals (Fig. 1). Surprisingly, in the denervated DCN we observed a striking increase of the full-length, non-cleaved neurocan protein as early as 2 days post-SCI, which preceded the detectable increase in GFAP protein associated with reactive gliosis (Figs. 1, 2B, H). This also contrasts with the changes in neurocan expression at the SCI, where no significant increase in full-length neurocan was detected until 1 week post injury (supplementary materials, Fig. S1). In addition, the increase in full-length neurocan in the DCN at 2 days post-SCI did not reflect an increase of the levels of the truncated forms (Figs. 1, ,2I),2I), thus suggesting that neo-expression of this CSPG was not followed by its expected proteolytic processing. The increased levels of full-length neurocan peaked at ~25-fold over control values by one week post-SCI and decreased by three weeks post-injury, but remained much higher than control levels (Figs. 1, ,2H).2H). This decrease at 3 weeks post injury coincided with increased levels of the truncated 150-kDa C-terminal neurocan isoform suggesting restoration of at least some proteolytic processing of neurocan (Figs 1, ,2I2I).
Brevican also showed a strong, but delayed, increase in the denervated DCN. (Figs. 1, 2J, K). Beginning at 1 week and persisting to 3 weeks post injury, brevican mRNA was increased and preceded the increase in expression of the full-length brevican protein (B6) observed at 3 weeks (Figs. 1, 2J, K). As was the case for neurocan, the increase in brevican full-length protein was not followed by its normal proteolytic processing. The N-terminal truncated isoform of brevican (B50), normally found in neural tissue, was decreased by ~40% at 3 weeks post-SCI when compared to control values, at the same time that the levels in the full-length protein were increased. Of all the CSPG members of the lectican family analyzed in this study, aggrecan was the only one whose mRNA or protein levels did not increase following SCI. Indeed, while aggrecan mRNA levels remained constant (Fig 2C), there was a small transitory decrease in aggrecan protein levels 1 week post -injury (Figs. 1, ,2D),2D), which disappeared by three weeks after SCI (Figs. 1, ,2D).2D). Collectively, these results demonstrate that there is marked upregulation of several CSPGs in the denervated target of dorsal column axons following SCI despite their remote location from the site of direct trauma.
To demonstrate the anatomical/cellular locations of CSPG expression changes within the normal and denervated DCN, brainstem sections from normal rats (n=7) and from rats 2 days (n=3), 1 week (n=6) and 3 weeks (n=7) following cervical SCI were processed for immunohistochemistry with antibodies against NG2 (7.1), neurocan (650.24 and 1F6), brevican (B756), and aggrecan (Cat-301). Lectin histochemistry with Wisteria floribunda agglutinin (WFA) was also used to detect chondroitin sulfate chains.
NG2 immunohistochemistry revealed small immunoreactive (IR) cell bodies with long thin processes in the normal DCN. None of these cells co-labeled with a marker for microglia (OX-42; Figs. 3A, E) or astrocytes (data not shown). Thus, they are likely O2-A glia (see Peters, 2004 for review). Surprisingly, many of the fine processes extending from the NG2-IR cells were observed to wind through areas containing clusters of CTB traced primary afferent terminals and MAP2-IR dendrites (Fig. 3H). A similar relationship has been described in the cerebellum where synaptic–like connections are formed between neurons and O2A-glia containing calcium-conducting glutamate receptors attached to post-synaptic elements by the intracellular domain of the membrane bound NG2 proteoglycan (Lin and Bergles, 2004; Lin et al., 2005).
At 2 days post SCI, when NG2 protein expression increased (Figs. 1, ,2F),2F), the distribution of NG2-IR increased throughout the DCN (Figs. 3A, B) and was associated with two different populations of cells (Fig. 3F) as defined by OX-42-IR, which labeled small and ramified microglia in the normal DCN (Fig. 3E). One population of cells expressing NG2 but not OX-42 displayed short, irregular, and thickened processes (Fig. 3F). These appeared to lose the intimate relationship with MAP2-IR dendrites observed in the DCN from control animals (Fig. 3H). The second population of cells were OX-42-IR microglia co-labeled by NG2-IR. These cells displayed activated morphologies with short, thickened processes and whose processes and cell bodies often colocalized with NG2-IR visible with high magnification confocal microscopy (Fig. 3F). Similar expression of NG2, has been shown around SCI sites (Jones et al, 2002; Tang et al, 2003) but there have been no previous reports of similar findings in the areas denervated by these injuries.
By 1 week post SCI, when NG2 protein expression decreased (Figs. 1 & 2F), the intensity of NG2-IR was reduced throughout the DCN (Figs. 3B, C) and no longer colocalized with OX-42-IR microglia (Fig. 3G). Both NG2-IR cells and OX-42-IR microglia had also begun to revert back to their respective morphologies observed in the control animals. By 3 weeks post injury, the intensity of the immunoreactivity of both antibodies was similar to that observed in uninjured rat DCN (Figs. 3A, D). These reductions of NG2-IR and OX-42-IR coincided with a significant increase in GFAP protein expression (Figs. 1, ,2B)2B) and activation of GFAP-IR astrocytes that became especially prominent within areas of denervated DCN gray matter (Fig. 4). Importantly, at no time was NG-2-IR found to colocalize with GFAP-positive astrocytes (Fig. 5H). and that NG2 seemed to be exclusively on O2A glia, Therefore while we can not demonstrate this conclusively, the exclusive association with OX-42-IR mircoglia and O2A glia suggest they are the likely source of NG2 upregulation.
From the results of the western blot experiments, we confirmed that the 650.24 and 1F6 antibodies both labeled full-length neurocan and each selectively detected the 150 kDa C-terminal or 130 kDa N-terminal isoforms of neurocan respectively (Fig. 1). In uninjured rats, where full length neurocan expression was low (Fig. 1), the 650.24 neurocan antibody weakly immunolabeled areas evenly distributed throughout the DCN neuropil (Fig. 5A). While the 1F6 neurocan-IR uniformly labeled the gracile nucleus, little labeling was observed within the somatotopic distribution of the CTB-traced forelimb digit primary afferent terminals located within the cluster region of the cuneate nucleus (Fig. 5E). This selective distribution suggests there may different but underappreciated contributions by these two neurocan isoforms however, the true significance of this finding is unknown.
Following SCI, increases in neurocan-IR were detected by both antibodies within the denervated gracile, cuneate and external cuneate nuclei (Figs. 5B–D, F–H) and mirrored the increases in full-length neurocan observed in western blot experiments (Figs. 1, ,2H).2H). At 2 days post injury, increased neurocan-IR was found in distinct cellular profiles intermingled within the increased neuropil immunolabeling throughout the rostral and caudal extent of the DCN (Figs. 5B, F–H). Under high magnification, the most heavily labeled portions of these neurocan-IR profiles co-localized strongly with the cell bodies and processes of GFAP-IR astrocytes but not with NG2-IR cells (Figs. 5F, H). By one week post SCI, these neurocan-IR cellular profiles had disappeared and were replaced by immunolabeling evenly distributed throughout the DCN (Fig. 5C) in locations occupied by denervated β-TIII-IR DCN neurons (Fig. 5E). After 3 weeks post SCI, neurocan-IR had diminished but was still comparatively stronger than that observed in the gracile and cuneate nuclei of uninjured animals (Figs. 5A, D). These data show a good correlation between an increase of neurocan expression detected by immunoblotting and immunohistochemistry. They demonstrate that this increase in areas occupied by denervated DCN neurons is likely due to astroglial expression of this CSPG.
In uninjured control rats, brevican-IR was observed in the DCN and was slightly more intense in areas containing MAP2-IR neurons (Figs. 5I, J). At 3 weeks post SCI, brevican-IR was increased in the DCN. This increase was most apparent within the cuneate and external cuneate nuclei (Figs. 5J–L). In contrast to the diffuse labeling seen in the normal cuneate nuclei and to the increased NG2- and neurocan-IR observed directly associated with activated glia in injured animals, brevican-IR labeling was associated primarily with the surface of MAP2-IR cuneate neurons (Fig. 5K, N). Because brevican can be expressed by both astrocytes (Yamada et al., 1997; Thon et al., 2000; Ogawa et al., 2001) and neurons (Seidenbecher et al., 1998), it is possible that either of these cell types may contribute to its increased expression.
In normal rats, both WFA and aggrecan-IR labeled similar and overlapping areas throughout the DCN (Figs. 6A, D, G, I–L). This immunoreactivity was mostly restricted to the condensed extracellular matrix, known as the “perineuronal net”, around subsets of Map2- or β–TIII-IR neuronal cell bodies and dendrites (Figs. 6D, G, J). The perineuronal nets were often equally labeled by both markers (Fig. 6I), suggesting that most of the WFA-detected chondroitin sulfate glycosaminoglycan side chains were attached to aggrecan core proteins. Importantly, these perineuronal nets were prominent within the highly developed somatotopic areas of the middle to caudal regions of the gracile and cuneate nuclei (Figs. 6A, D, G, L), known to contain the highest concentrations of both primary afferent terminals and thalamically projecting secondary neurons (Maslany et al., 1991; 1992; Crockett et al., 1996; Magnusson et al., 1996; Massey et al., 2006). Neurons labeled by parvalbumin-IR, many of which have been shown to be secondary relay neurons (Crockett et al, 1996; Magnusen et al, 1997), were also often encapsulated by WFA label throughout areas of the gracile and cuneate nuclei (Fig. 6H).
During the period studied following cervical SCI, labeling of the extracellular matrix and DCN perineuronal nets persisted despite the denervation of forelimb and hindlimb primary afferent input. In brainstem sections examined one week post SCI, the small decrease in aggrecan levels detected by western blot (Figs. 1, ,2D)2D) correlated with modest changes in either WFA labeling or aggrecan-IR within the DCN (Figs. 6B, E, M). Similarly, there were no apparent differences in the perineuronal net profiles visible (Figs. 6A, B, D, E, L, M). At 3 weeks post SCI, when aggrecan levels had returned to control values according to our western blotting results (Figs. 1, ,2D),2D), it was difficult to distinguish obvious differences in WFA- or aggrecan labeling between control and injured DCN tissue (Figs. 6A, C, D, F, L, N). In none of the brainstem sections examined from either injured or normal rats did aggrecan-IR colocalize with GFAP-IR astrocytes (Fig. 6K). The significance of these subtle changes in the perineuronal net-restricted CSPG aggrecan are not clear, and could be related to microheterogenous changes in the isoforms of this proteoglycan (Matthews et al., 2002). Additionally, minute shifts in the extracellular matrix that occur may be under appreciated using confocal microscopy alone. However, an important conclusion from these findings is that much of the highly organized matrix that conform the perineuronal nets surrounding specific subsets of gracile and cuneate neurons remains intact following denervation by SCI and appears minimally affected by changes in the amorphous extracellular matrix surrounding neurons and astrocytes.
High concentrations of CSPGs contribute to the regeneration barrier created by the glial scar. Does the increased CSPG expression within the denervated DCN constitute a potential barrier to regenerating axons attempting to reinnervate target neurons? To address this question, GFP-expressing adult DRG neurons were microtransplanted into the gracile fasciculus caudal to the gracile nuclei 2 weeks following complete denervation by a T1 dorsal column SCI (n=6). Axons from DRG microtransplants have been shown to regenerate approximately 1.0 mm per day through both normal and injured dorsal column white matter but not through CSPG-rich SCI lesion sites (Davies et al, 1997; 1999; Grimpe and Silver, 2004). Likewise, when the brainstems of these rats were examined, many regenerating DRG axons elongated bidirectionally from the implantation site within the degenerating dorsal column white matter (Fig. 7A). As axons approached the caudal edge of the CSPG-rich (Figs. 1, ,2,2, ,5,5, ,6)6) gracile nucleus, most turned abruptly and moved ventrally, laterally or haphazardly and failed to penetrate the territory of the nucleus (Figs. 7A–C). Occasional axons originating from the more deeply placed portions of the microtransplants entered some areas of lower brainstem gray matter intermingled among NeuN labeled neurons along small ventrally oriented fiber tracts and, in some cases, passed rostrally completely below, but rarely into, the gracile nuclei (Fig. 7A).
Application of chABC promotes collateral sprouting of spared dorsal column axons over short distances within the partially denervated adult rat cuneate nucleus (Massey et al, 2006) and increases primary afferent regeneration through normally CSPG rich SCI sites (Bradbury et al., 2002). Therefore, we next investigated whether injections of chABC (n=4) increased the ingrowth of regenerating DRG axons into the gracile nucleus following denervation. Injection of chABC into the brainstem resulted in the digestion of CSPG moieties in the gracile nucleus in a manner comparable to our previously reported results (Massey et al, 2006) (data not shown). In contrast to our observations made in non-treated injured rats (336.95±140.6 μm), treatment with chABC moderately, but significantly, increased the ability of regenerating axons to extend through longer portions of the gracile nucleus (1276.2±243.5 μm; p<0.05 by one-way ANOVA and Tukey’s test) (Figs. 7A, D, ,9C).9C). Axons from these animals could be observed penetrating deep into the gracile nucleus intermingled among NeuN-IR neurons, a finding not seen in any of the non-treated injured rats (Figs. 7A, D).
Because the ability of dorsal column axons to cross injury-induced CSPG rich scar tissue is also enhanced by increased NT-3 expression (Oudega and Hagg, 1999; Bradbury et al, 1999; Lu et al., 2004, 2006; Taylor et al, 2006), we injected NT-3 expressing lentivirus alone (n=5) or in combination with chABC (n=6) to improve the regenerative potential of the microtransplanted DRG neurons. The use of lentivirual vectors is an extremely efficient gene delivery system that can stably increase NT-3 expression by either neurons or astrocytes and has been shown to cause only very limited CNS inflammatory responses when compared to saline controls (Zufferey et al, 1998, Abdellatif et al, 2006). To assess the bioactivity and infectivity of the NT-3 expressing vectors, we first cultured E15 rat DRG neurons in conditioned medium from mock-, NT3-lentivirus- and GFP-lentivirus-infected cortical astrocytes. Only conditioned medium from mature astrocytes infected with the NT-3 lentivirus (183±7.2 cells), but not that from either the GFP-carrying lentivirus (89.7±1.8 cells), nor the control transductions (72±9.6 cells) significantly increased the numbers of surviving β-tubulin-IR neurons (one-way ANOVA, p<0.001; post hoc Tukey’s test, p<0.001). Furthermore, neurons receiving NT-3 lentivirus medium also displayed many long processes which contrasted with the sparse, neurite outgrowth observed from neurons receiving media from the control transductions (Figs. 8A & B). NT-3 in vivo expression was then assessed by ELISA assay following matched volume injections of either saline (n=3) or virus (n=3) into the rostral medulla (Fig. 8C). Within 10 days, NT-3 levels in animals receiving virus (32.6±3.6 pg/mg) were significantly increased (Student’s t test, p<0.05) to 3 times greater than those receiving control injections (9.4±4.9 pg/mg). In our experiments, no increases in the inflammatory response nor gross pathological changes were noted in the brainstems of any of the rats receiving virus when compared to those receiving only microtransplants, saline, or chABC alone (data not shown).
After treatment with NT-3 lentivirus alone, there was a moderate but significant increase of axonal regeneration into the gracile nucleus (Figs. 8D, ,9C)9C) when compared to non-treated rats (Fig. 7A) (1118.5±219.8 μm; one-way ANOVA and Tukey’s test, p<0.05). No significant differences in regeneration into the nucleus occurred between the chABC alone and NT-3 alone animals. However, when treatment with the NT-3 lentivrus was combined with chABC, there was a dramatic increase of axons invading into the gracile nucleus. In fact, the paths of most axons appeared relatively unaltered by the nuclear border (Fig. 9A). On higher power examination, GFP labeled axons were observed through nearly all portions of the nucleus (Fig. 9B). Interestingly, the combined treatment not only increased axonal extension (10820.3±2583.5 μm) into the gracile nucleus over non-treated rats (one-way ANOVA, p<0.01; post hoc Tukey’s test, p<0.001) but was increased nearly 10 times that observed with either chABC (post hoc Tukey’s test, p<0.01) or virus (post hoc Tukey’s test, p<0.01) treatments alone (Fig. 9C). In sum, these results show that, while individually effective, there is an important synergistic effect between chABC and NT-3 expression, which greatly promotes axonal regeneration over each treatment alone.
New synaptic contacts must be made for functional improvement after SCI. CSPGs limit both CNS axonal regeneration and functional plasticity (Davies et al, 1997; 1999; McKeon et al, 1991; Fitch and Silver, 1997; Bradbury et al., 2002; Grimpe and Silver, 2004; Pizzorusso et al, 2002; 2006; Barritt et al, 2006; Houle et al, 2006, Massey et al, 2006). Here, we investigated whether denervation of brainstem spinal cord targets induces an increase of CSPG expression and creates a barrier to regenerating axons, thus, limiting reinnervation. Our results show that 1) following SCI and in the absence of direct trauma, expression of the CSPGs NG2, neurocan and brevican is increased and accompanied by altered proteolytic processing in the extracellular matrix surrounding denervated neurons while the perineuronal net associated CSPGs remain intact, 2) increased expression of individual CSPGs coincide with reactive changes in NG2-IR glia, microglia and in astrocytes, and that 3) treatment with chABC and NT-3 synergistically promote robust axonal regeneration by transplanted DRG neurons into these CSPG rich, denervated DCN targets. We conclude from these results that denervation of certain target neurons by SCI results in a reactive gliosis and increased secretion of CSPGs that must be overcome for successful reinnervation.
The increased expression of CSPGs by reactive glia at the sites of direct CNS trauma has been well documented (McKeon et al, 1991; Fitch and Silver, 1997; Bruce et al., 2000; Jones et al., 2002; Jones et al., 2003; Tang et al., 2003). Because most of these studies have restricted their focus to the injury site where the normal vascular integrity is lost, it was thought that the glial response, including increased CSPG expression, is an attempt to re-establish the glia limitans and, therefore, occurs only at sites of blood brain barrier breakdown (Fitch and Silver, 1997; Rhodes and Fawcett, 2004; Silver and Miller, 2004, Rhodes et al, 2006). Other authors have demonstrated reactive changes in both microglia and astrocytes in the DCN following denervation by dorsal column axotomy that takes place without changes in vascular permeability (Liu et al, 1998). In the present study, we show the upregulation of mRNA and protein for several CSPGs in denervated brainstem targets located well outside the area directly traumatized by the C3-C4 spinal cord lesion. Our data suggest that increased CSPG expression by reactive glia can take place without trauma-induced changes in vascular integrity.
The signaling mechanisms that modulate expression of CSPGs and other extracellular components following injury are not completely understood. In our study, increases in NG2 expression occurred within the first 48 hours and were associated with NG2-IR glia and microglia that had adopted activated morphologies. This coincides with the period in which the membranes of the lesioned primary afferent terminals first demonstrate Wallerian changes. Some evidence suggests that increased extracellular potassium and adenosine released from degenerating terminals are sufficient cues to signal mircoglial responses (Khanna et al., 2001; Inoue, 2002). Additionally, following denervation of the dentate gyrus by entorhinal cortex lesions, microglial activation and recruitment, similar to that observed in the DCN in our model and previous studies (Koshinaga and Whittemore, 1995; Liu et al, 1998), depend on CXCR3 receptor activation by neuronal expression of the chemokine CXCL10 (Biber et al., 2001; 2002; Rappert et al., 2002; 2004). At one week post injury, we found that peak neurocan expression coincided with a decrease in activation of NG2-IR glia and microglia and an increase in reactive changes by GFAP-IR astrocytes. Transforming growth factor-β (TGF- β), which is known to be upregulated by microglia and astrocytes at sites of CNS injury and denervation (Morgan et al., 1993; Vincent et al., 1998b) has been shown to induce reactive changes in astrocytes, increase astrocyte neurocan secretion (Asher et al., 2000; Smith and Strunz, 2005) and negatively regulate microglial activation and phagocytic activity (Vincent et al., 1998a; Schilling et al., 2001; Milner and Campbell, 2002; Brionne et al., 2003). Outside the CNS, TGF- β is important in signaling the transition from the inflammatory to the reparative phase where it inhibits the activation of monocytes and induces fibroblasts to proliferate and secrete extracellular matrix components (see Faler et al., 2006 for review). Therefore, TGF- β may participate in similar transitions in the reactive response in the injured CNS by limiting microglial activity and inducing the secretion of CSPGs by astrocytes. Uncovering the precise role of this and other chemical mediators released at sites of deafferentations and their subsequent impact on signal transduction pathways leading to gene regulation, as well as reactive processes of CNS glia and neurons are the topic of ongoing investigations.
In addition to demonstrating increased expression, our results also provide evidence of the heterogeneous processing (e.g., regulated proteolysis) of individual CSPGs after injury. For example, the cleaved 275kDa NG2 isoform accounted for nearly all of the increases NG2 expression both in the denervated DCN and at the SCI in our control experiments. This contrasted with the apparent decreased proteolytic processing of neurocan and brevican. These findings suggest that an altered expression of proteolytic enzymes and their inhibitors, some of which have been shown to take place at sites of direct CNS injury (Muir et al., 2002), accompany the changes in CSPG expression and glial cell activation around deafferented neurons following denervating injuries. The summed physiological consequences of selective expression and processing of individual CSPG isoforms on neuronal interactions with other components of the CNS is difficult to predict and currently poorly understood. For instance, the truncated 275kDa NG2 isoform is thought to undergo a conformational change that exposes two sites inhibitory to axonal regeneration that are not accessible on the normal, membrane-bound, full length NG2 protein (Ughrin et al., 2003; Tan et al., 2005). It is possible that selective increase of the truncated NG2 isoform may contribute to the inhibitory actions of NG2 in and around SCI sites (Tan et al., 2006) which are not observed in the presence of the membrane-bound NG2 alone (Yang et al., 2006). Moreover, the C- and N-terminal cleavage products of full-length neurocan have been reported to have different effects on cell adhesion (Talts et al, 2000) and bind to separate extracelluar CNS targets (Yamaguchi et al, 2000; Rauch et al, 2001; Viapiano and Matthews, 2006). Further understanding of this differential expression of CSPG isoforms throughout the gliotic response may help in identifying specific effects of individual CSPGs after injury and enable a refinement of strategies against specific inhibitory signals from the extracellular matrix (Viapiano and Matthews, 2006).
CSPGs located both in the neural extracellular matrix and in the perineuronal nets contribute to the regulation of axonal outgrowth and synaptic plasticity during normal development (Kalb et al, 1990a,b; Lander et al, 1997; Pizzorusso et al, 2002, 2006; Corvetti and Rossi, 2005). Expression of these molecules coincides with the end of critical period plasticity and in the early postnatal period (Kalb et al, 1990a,b; Lander et al, 1997; Pizzorusso et al, 2002), help guide axonal sorting and plasticity of cortical axons (Steindler and Cooper, 1987; Steindler et al., 1990). In the denervated DCN, the expression of several CSPGs known to inhibit axonal outgrowth (Niederost et al., 1999; Sango et al., 2003; Ughrin et al., 2003) were increased while the normal expression of constitutively expressed CSPGs contained within the perineuronal nets remained largely intact. Because the detailed developmental expression of CSPGs within the brainstem has not been described, it is not known if a similar pattern of expression partitions areas of the DCN as ascending and descending pathways mature. In recent work, we demonstrated that functional collateral sprouting by the uninjured forelimb primary afferents is restricted from denervated portions of the cuneate nuclei by CSPGs (Massey et al, 2006). In the current study, regenerating DRG axons in the white matter compartment were unable to enter the denervated, CSPG rich gracile nuclei without intervention. Thus, these findings strongly suggest that modification of local CSPG expression plays a prominent role in the inhibition of reactive synaptic plasticity, and DCN neuron reinnervation. Similar results have been reported in the superior colliculus (Tropea et al., 2003) and in the hippocampus, where increases in neurocan and brevican have also been described after denervating injuries (Deller and Frotcscher, 1997; Haas et al, 1999; Thon et al, 2000; Mingorance et al, 2006).
Buffering access to deafferented target neurons by the establishment of new barriers may have selective benefits. Increased function by previously silent primary afferents whose terminal fields overlap with those axotomized by partial cervical dorsal rhizotomies has recently been demonstrated to result in the partial return of forelimb use after long periods of recovery (Darian-Smith, 2004; Darian-Smith and Ciferri, 2006). However, sparse collateral sprouting by trigeminal primary afferents into the peripheral portions of the cuneate nucleus following longer periods of survival also occurs and is thought to contribute to the development of phantom sensations (Pons et al., 1991; Florence and Kaas, 1995; Jain et al., 1997; 2000). Thus, changes in CSPG expression may allow surviving afferents with existing synaptic connections within denervated targets to very slowly maximize potential recovery. At the same time, they may minimize the establishment of aberrant and maladaptive sensory pathways by creating an inhibitory border that shields access to open synaptic sites.
Many combinatorial strategies applied directly to injured axonal pathways have been demonstrated to offer increased axonal regeneration and, in some cases, enhance functional recovery when compared to single strategies used in isolation (Lu et al, 2004; Pearse et al, 2004; Steinmetz et al., 2005; Fouad et al., 2005; Houle et al., 2006; Taylor et al., 2006). We have previously demonstrated that injections of chABC into the DCN result in both the digestion of CS glycosaminoglycan side chains and in the reduction of CSPG core proteins (Massey et al., 2006) and therefore much of chABC’s can be attributed to the removal of the barrier created by these molecules. NT-3 through activation of axonal Trk receptors and subsequent intracellular signaling (for review see Huang and Reichardt, 2003) stimulate regeneration despite the presence of CSPGs and other inhibitory molecules (Oudega and Hagg, 1996; Bradbury et al., 1999; Oudega and Hagg, 1999; Lu et al., 2004; Lu et al., 2006; Taylor et al., 2006). In our experiments, combined delivery of chABC and NT-3 dramatically increased DRG axonal extension into denervated target areas over the use of either treatment alone. A synergistic effect between brain derived neurotrophic factor (BDNF) and chABC when applied together has similarly been reported to increase sprouting by retinal ganglion cell axons within denervated midbrain targets (Tropea et al, 2003).
The precise mechanism by which neurotrophin support interacts with CSPG digestion to increase axonal invasion into normally inhospitable environments is not known. Because CSPGs act as a potent barrier to regenerating axons, their digestion with chABC likely potentiates the chemoattractive properties of NT-3. Additionally, some of the inhibitory actions of CSPGs are thought to be mediated by interference of the proper binding of axonal surface molecules and growth promoting extracellular matrix components, including those that activate integrin signaling cascades (McKeon et al, 1995; Grumet et al, 1996; Milev et al, 1996). It has recently been shown that CSPG interference of integrin activation reduces the ability of neurotrophins to induce robust regeneration (Zho et al, 2006). Thus, our combined treatment may have potentiated NT-3 signaling by “de-inhibition” of accompanying integrin signaling. Finally, removal of the negatively charged glycosaminoglycan side chains from the CSPGs by chABC may have increased the availability of NT-3 by increasing its diffusion to approaching axons (Gruskin et al, 2003).
Because of the considerable difficulty that has been encountered attempting to develop treatment strategies that promote successful axonal regeneration of severed axons back to target neurons denervated by SCI, much of the work has justifiably focused on circumventing the barriers erected at the injury site. Our findings demonstrate that additional challenges await approaching axons once they arrive at their desired targets. These findings further suggest that proper reconstruction and functional recovery of severed spinal cord pathways will require that combinatorial strategies be applied both to the sites of axotomy and to their targets to promote and maintain axonal regeneration then to engender reinnervation and synapse formation.
Supplementary Materials, Figure S1. Quantification of real-time PCR and western blot analysis of GFAP (A, B), aggrecan (C, D), NG2 (E, F), neurocan (G–I) and brevican (J, K) expression in the C3-C4 spinal cord of normal rats (Ctr) and rats following C3-C4 SCI. Peak expression of mRNA precedes increases in GFAP, NG2, neurocan and brevican protein expression. This is consistent with previously published studies. *p<0.05; **p<0.01; ***p<0.001.
Supplementary Materials, Figure S2. Images of western blots (A) and quantification of real-time PCR and western blot analysis of GFAP (B, C), neurocan (D, E) and brevican (F, G) expression in the visual cortex of normal rats (Ctr) and rats following SCI. There were no significant changes in mRNA or protein expression in the visual cortex after SCI.
This work was supported by National Institutes of Health NS40411 (S.M.O.), RR15576 (S.M.O.), NS25713 (J.S.), and RR16481 (N.G.F.C.). J.M.M. was supported by a fellowship awarded from the Kentucky National Science Foundation/Experimental Program to Stimulate Competitive Research EPS-9874764 (N.G.F.C.) and by the Summer Research Scholars Program at the School of Medicine in the University of Louisville. The authors extend their appreciation to the Stan Gerson (Case Western Reserve University) and S. O’Gorman (Case Western Reserve University) labs and to Justin Roth for their expertise and technical assistance in development of the lentiviral vectors used in our experiments. We also thank Julie Decker and Polly Fleck for their expert technical assistance and Aaron Puckett and the University of Louisville Research Resources Center veterinarians and staff for their excellent assistance with veterinary care.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.