Adeno-associated virus-mediated L1 expression promotes functional recovery after spinal cord injury
1W. M. Keck Center for Collaborative Neuroscience, Department of Cell Biology and Neuroscience, Rutgers the State University of New Jersey, Piscataway, NJ, USA, 2Zentrum fuer Molekulare Neurobiologie Hamburg, Universitaetsklinikum Hamburg-Eppendorf, Universitaet Hamburg, Hamburg, Germany, 3Laboratory for Reinnervation Processes, Department of Neurophysiology, Nencki Institute of Experimental Biology, Polish Academy of Sciences, Warsaw, Poland and 4DFG Research Center Molecular Physiology of the Brain at Department of Neurology, University of Goettingen, Goettingen, Germany
Correspondence to: Dr Melitta Schachner, W. M. Keck Center for Collaborative Neuroscience, Department of Cell Biology and Neuroscience, Rutgers the State University of New Jersey, Piscataway, NJ 08854, USA E-mail: schachner{at}biology.rutgers.edu
 |
Summary |
|---|
 Top Summary IntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Paucity of permissive molecules and abundance of inhibitory molecules in the injured spinal cord of adult mammals prevent axons from successful regeneration and, thus, contribute to the failure of functional recovery. Using an adeno-associated viral (AAV) vector, we expressed the regeneration-promoting cell adhesion molecule L1 in both neurons and glia in the lesioned spinal cord of adult mice. Exogenous L1, detectable already 1 week after thoracic spinal cord compression and immediate vector injection, was expressed at high levels up to 5 weeks, the longest time-period studied. Dissemination of L1-transduced cells throughout the spinal cord was wide, spanning over more than 10 mm rostral and 10 mm caudal to the lesion scar. L1 was not detectable in the fibronectin-positive lesion core. L1 overexpression led to improved stepping abilities and muscle coordination during ground locomotion over a 5-week observation period. Superior functional improvement was associated with enhanced reinnervation of the lumbar spinal cord by 5-HT axons. Corticospinal tract axons did not regrow beyond the lesion scar but extended distally into closer proximity to the injury site in AAV-L1-treated compared with control mice. The expression of the neurite outgrowth-inhibitory chondroitin sulphate proteoglycan NG2 was decreased in AAV-L1-treated spinal cords, along with reduction of the reactive astroglial marker GFAP. In vitro experiments confirmed that L1 inhibits astrocyte proliferation, migration, process extension and GFAP expression. Analyses of intracellular signalling indicated that exogenous L1 activates diverse cascades in neurons and glia. Thus, AAV-mediated L1 overexpression appears to be a potent means to favourably modify the local environment in the injured spinal cord and promote regeneration. Our study demonstrates a clinically feasible approach of promising potential.
Key Words: adeno-associated virus; axonal regeneration; L1; locomotor recovery; spinal cord injury
Abbreviations: AAV, adeno-associated viral vector; GFP, green fluorescent protein
.
Received November 16, 2006. Revised February 16, 2007. Accepted February 16, 2007.
|
Introduction |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Functional recovery after spinal cord lesion in humans is an important goal in restorative medicine. Strategies used to enhance recovery include neutralization of inhibitory cues in the adult central nervous system of mammals that prevent regeneration (Filbin, 2003
; McGee and Strittmatter, 2003; Silver and Miller, 2004; Buchli and Schwab, 2005; Thuret et al., 2006; Yiu and He, 2006) and interventions promoting axon regeneration, neuronal survival and synaptic plasticity (Loers and Schachner, 2007). The neural recognition molecule L1 favours axonal growth in an inhibitory environment (Castellani et al., 2002; Roonprapunt et al., 2003; Xu et al., 2004; Chen et al., 2005b; Zhang et al., 2005). Inhibition of L1.1 function in the successfully regenerating spinal cord of zebrafish leads to impairment of axonal regrowth and recovery of swimming after injury (Becker et al., 2004). L1 promotes neurite outgrowth in a homophilic (i.e. self-binding) manner and such homophilic interactions not only promote neurite outgrowth and neuronal migration, but also neuronal survival (Lindner et al., 1983; Rathjen and Schachner, 1984; Moos et al., 1988; Lemmon et al., 1989; Appel et al., 1993; Beggs et al., 1997; Kamiguchi and Yoshihara, 2001; Kleene et al., 2001; Dong et al., 2002, 2003). L1 is involved in modification of synaptic efficacy as well as in learning and memory (Luthi et al., 1994; Fransen et al., 1998; Tiunova et al., 1998; Pradel et al., 2000; Law et al., 2003; Venero et al., 2004). L1 also promotes myelination in the central and peripheral nervous systems (Seilheimer et al., 1989; Wood et al., 1990; Haney et al., 1999; Barbin et al., 2004), and thus could promote remyelination after spinal cord injury.
Application of a fusion protein containing the extracellular domain of L1 and human Fc promotes locomotor recovery in adult rats after contusion spinal cord injury (Roonprapunt et al., 2003). Furthermore, retinal ganglion cell axons regrow in an L1 conducive environment after optic nerve transection (Xu et al., 2004). Embryonic stem cells overexpressing L1 support regrowth of corticospinal tract (CST) axons and survive better compared to non-transfected stem cells in the injured spinal cord of adult mice (Chen et al., 2005b). As a next step towards therapeutic application of L1 after spinal cord injury in humans, we expressed full-length murine L1 in the lesioned mouse spinal cord via an adeno-associated virus (AAV) which has emerged as a powerful transgene delivery vehicle (Flotte, 2004; Shevtsova et al., 2005). The recombinant virus, already used in clinical trials (Mandel and Burger, 2004; Mandel et al., 2006), was applied locally after lesion to exploit a feasible therapeutic approach. We selected an AAV serotype 5 (AAV-5) vector which expresses transgenes from the murine cytomegalovirus (mCMV) immediate early promoter (Shevtsova et al., 2005), a cell-type non-specific promoter, with the intention to target gene expression to different cell types and thus enable extensive homophilic and heterophilic L1 interactions. We observed enhanced functional recovery, positive effects on damaged 5-HT and CST axons, reduced expression of the neurite outgrowth-inhibitory chondroitin sulphate proteoglycan (CSPG) NG2 and reduced astrogliosis as indicated by glial fibrillary acidic protein (GFAP). Furthermore, we found that exogenous L1 mediates changes in signal transduction pathways influencing expression of growth-related molecules in regrowing axons and/or cells residing in the lesioned spinal cord. The combined results encourage further use of L1 in lesion paradigms, not only relating to spinal cord injury, but also in other types of central nervous system (CNS) trauma.
|
Material and methods |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
AAV vector production
An AAV-5 vector was constructed to express L1 in the spinal cord. The genome of the recombinant viral vector consisted of the short version (530 bp) of the murine cytomegalovirus immediately early promoter (mCMV promoter) (Bett et al., 1994
), the cDNA for murine L1 NH-terminally tagged with the FLAG-epitope and the bovine growth hormone polyadenylation site. As a control, an AAV-5 vector expressing enhanced green fluorescent protein (GFP) was constructed. It is noteworthy that the AAV-GFP construct contains a woodchuck hepatitis post-transcriptional control element (WPRE) which stabilizes mRNA, resulting in a 2–10-fold higher rate of protein expression. This element is not contained in the AAV-L1 virus due to vector size restrictions. Vectors were propagated in 293 cells using pDP5 as helper plasmid (Grimm et al., 2003), and purified essentially as described by Malik et al. (2005) except that Q-FF anion-exchange columns (Amersham) were used for FPLC. After dialysis, genome titres were determined by quantitative PCR, purity by SDS-gel electrophoresis and infectious titres by transduction of cultured primary brain cells as described by Kügler et al. (2003). The genome particles: transducing units ratio ranged from 25 : 1 to 35 : 1.
Animal surgery
All surgical procedures and post-operative care were approved by the local authorities following the guidelines of the European Community and the National Institutes of Health (USA). Female C57BL/6J mice, 3 months old, were deeply anaesthetized and operated as described by Chen et al. (2005b). In brief, laminectomy was performed at the T7–T9 level and the spinal cord was lesioned using a mouse spinal cord compression device (Curtis et al., 1993). The viral constructs AAV-L1 or AAV-GFP (3 x 107 transducing units in 1 µl) were injected into the lesion site. The animals were subjected to evaluation of motor function at certain time points, and sacrificed for histology or biochemical measurements after the last evaluation for motor function.
For anterograde labelling of CST axons, six animals from each group were re-operated 10 days before sacrifice as described previously (Chen et al., 2005b). Fluororuby (Invitrogen, Carlsbad, CA, USA) was injected into the sensorimotor cortex, and animals were sacrificed and parasagittal spinal cord sections were collected consecutively. To quantify axonal regrowth, the rostral border of the lesion site demarcated by highest levels of GFAP immunostaining (Chen et al., 2005a), was chosen as the reference point for measurement (Fig. 1). The distance between the rostral border and the furthest detectable axon tips of the CST axons in all consecutive sections was measured. For measurement of regrowth of 5-HT immunoreactive axons from the brainstem, consecutive cross-sections of the spinal cord were taken starting 5 mm rostral to the GFAP demarcated boundary and ending 5 mm caudal to the lesion site.
|
Evaluation of motor functions
Functional recovery was analysed in AAV-L1- and AAV-GFP-treated animals (n = 15 and 14, respectively). The recovery of ground locomotion was evaluated using a mouse rating scale, the Basso mouse score (BMS) (Engesser-Cesar et al., 2005
). In addition, motor recovery was analysed using a novel approach for numerical assessment of different aspects of motor behaviour (Apostolova et al., 2006). This method includes evaluation of four parameters in three different tests: beam walking (foot-stepping angle, rump-height index), voluntary movements without body-weight support (extension–flexion ratio) and inclined ladder climbing. A left- and a right-side view of each animal during two consecutive walking trials were captured prior to the operation and 1, 2, 3 and 5 weeks after the operation with a high-speed video camera and analysed with the affiliated analysis software (SIMI Motion, SIMI Reality Motion Systems GmbH, Unterschleissheim, Germany). In addition to analyse absolute values for the functional parameters, we calculated and analysed relative estimates of functional recovery, recovery indices (Apostolova et al., 2006). The recovery index is an individual animal estimate for any given parameter described earlier and is calculated in percent as:
 |
Western blot analysis
Spinal cord tissue was taken in tissue pieces from four AAV-L1 and four AAV-GFP-injected mice with a length of 5 mm and with the lesion site in the centre, unless specified otherwise. Total protein was extracted by homogenizing and sonicating the tissue in RIPA buffer (Sigma). After centrifugation of the homogenate at 1000 g and 4°C the protein concentration in the supernatant was determined by BCA (Pierce Biotechnology, Rockford, IL, USA). The supernatant was denatured by boiling for 5 min in SDS sample buffer. Twenty micrograms of total protein were then subjected to 4–12% gradient SDS–PAGE. Gels were transferred to polyvinylidene difluoride membrane and probed with the following antibodies: L1 monoclonal antibody 555 (1 : 5000 diluted, Appel et al., 1995), Numb (1 : 50, an antibody developed by Dr Catherine Saner and obtained from the Developmental Studies Hybridroma Bank, University of Iowa, Iowa City, IA, USA), GFAP (1 : 15 000), myelin basic protein (MBP) (1 : 2000), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (1 : 5000, Chemicon, Temecula, CA, USA), Iba-1 (1 : 1000, Wako, Richmond, VA, USA), Rac1 (1 : 1000, Upstate, Lake Placid, NY, USA), phosphoinositide-3 kinase (PI3K) (1 : 1000), phosphorylated extracellular signal-regulated kinase (ERK) 42/44 (1 : 500), total ERK 42/44 (1 : 500), phosphorylated cAMP response element-binding protein (CREB) (1 : 1000), total CREB (1 : 1000), phosphorylated protein kinase A (PKA) (1 : 500) and total PKA (1 : 500) (all from Cell Signalling Technology, Danvers, MA, USA). Secondary mouse, rat or rabbit antibodies conjugated to horseradish peroxidase with ECL illuminescence intensification (Pierce Biotechnology) were used for detection. The grey value of each band was measured and normalized to the grey value of the corresponding GAPDH band.
GTP-RhoA pull-down assay
GTP-RhoA levels in spinal cord homogenates of four AAV-L1- and four AAV-GFP-treated mice were detected using the Rhotekin Rho binding domain reagent (Upstate, Lake Placid, NY, USA) and mouse-anti-RhoA (Santa Cruz Biotechnology, Santa Cruz, CA, USA) according to the manufacturer's instructions.
Immunohistochemistry and immunocytochemistry
For immunohistochemistry, spinal cords taken from mice that had been perfused with 4% formaldehyde in 0.1 M phosphate-buffered saline (PBS), pH 7.3, were used. Cross-sections (from four AAV-L1 and AAV-GFP animals) and parasagittal sections (from five animals for each group) were subjected to antigen retrieval with 10 mM sodium citrate (pH 9.0, 80°C, 30 min). Normal goat serum (5% in PBS, pH 7.3) with 0.2% Triton-X 100 was used for blocking the sections for 1 h at room temperature. Monoclonal antibodies to L1 (diluted 1 : 1000 in the blocking serum solution, Appel et al., 1995), serotonin (1 : 100), NeuN (neuron-specific nuclear protein, 1 : 1000), MBP (1 : 200) (Chemicon, Temecula, CA, USA) and polyclonal antibody to fibronectin (1 : 500, Sigma, Saint Louis, MI, USA), GFAP (1 : 500, DakoCytomation Denmark, Glostrup, Denmark) and Iba-1 (1 : 500) were applied to the sections overnight at 4°C followed by incubation with Cy2 or Cy3 conjugated anti-mouse, rat or rabbit IgG antibodies. For cultured cells, coverslips were briefly washed twice with PBS, pH 7.2, followed by fixation of the cells with 4% formaldehyde for 10 min at room temperature. After washing with PBS, the coverslips were blocked with 5% normal goat serum in PBS containing 0.2% Triton-X 100 for 30 min at room temperature followed by incubation with primary antibodies at 4°C overnight and then with secondary antibodies for 30 min at room temperature. All sections and coverslips were finally counterstained with DAPI and, after washing, mounted with Aqua-Poly/mount (Polysciences, Warrington, PA, USA). The slides were then observed by fluorescence (Axiophot 2, Zeiss, Jena, Germany) and/or confocal (LSM510mega, Zeiss) microscopy. For quantitative morphological evaluation, the NIH software ImageJ was used.
Cyclic AMP measurement
Cyclic AMP levels of AAV-L1- and AAV-GFP-infected spinal cord homogenates from four mice each were quantified using the cyclic AMP ELISA kit according to the manufacturer's instructions (R&D Systems, Minneapolis, MN, USA).
Primary astrocyte cultures and in vitro infection
Purified astrocyte cultures were prepared from whole brains of neonatal (P0) C57BL/6J mice. Brains were removed, meninges were peeled off and the tissue was cut into 1 mm3 pieces and dissociated by sequentially passing the pieces through 200, 150 and 30 µm filter meshes (VWR GmbH, Darmstadt, Germany). The tissue pieces were then washed three times with pre-warmed DMEM supplemented with 10% foetal calf serum, 0.1 mg/ml streptomycin and 10 U/ml penicillin (Invitrogen, Carlsbad, CA, USA) and plated at a density of 106 cells/well on poly-L-lysine-coated 12-well-plates. When cells had grown to 
50% confluence, AAV-L1 and AAV-GFP were added into the medium (3 x 107 transducing units in 500 µl). Ten days after viral transduction, cells were digested with 0.02% trypsin/EDTA (Invitrogen) and subcultured into poly-L-lysine-coated 12-well plates, 96-well plates or coverslips for further studies. Transduction efficiency was 50% after one passage.
Astrocyte proliferation assay
Astrocytes transduced with AAV-L1 or AAV-GFP were plated at a density of 20 000 cells/well in 200 µl DMEM supplemented with 10% foetal calf serum on poly-L-lysine-coated 96-well plates and grown for 3 days. Thereafter, the medium was daily changed for 3 days without addition of supplements or supplemented with either mitogen-activated protein kinase (MAPK) inhibitors (U0126 10 µM; PD98059 10 µM, Sigma, Saint Louis, MI, USA) or anti-fibroblast growth factor receptor 3 (FGFr3) neutralization antibody (R&D Systems, Minneapolis, MN, USA). This antibody neutralizes the signal transduction of mouse FGFr3 alpha (IIIb and IIIc) in the presence of all FGF isoforms. Twenty-four hours after the last treatment of the cells, a proliferation assay using the CyQUANT cell proliferation assay kit was performed according to the manufacturer's instructions (Invitrogen). Each value in the presented data was obtained from six experiments performed in duplicate.
Astrocyte migration assay
Astrocytes transduced with either AAV-L1 or AAV-GFP were plated in DMEM supplemented with 10% foetal calf serum on poly-L-lysine-coated 24 well-plates and grown to confluence. The medium was then removed, and the monolayer was scratched with a sterile 20–200 µl plastic pipette tip. The cells were washed twice with pre-warmed medium and incubated for 4 or 12 h and then fixed with 4% formaldehyde and stained for L1 and GFAP. All nuclei were stained with DAPI.
Measurement of GFAP expression in astrocyte cultures
Four types of astrocyte were subjected to evaluation of GFAP expression: (i) AAV-L1 transduced astrocytes; (ii) AAV-GFP transduced astrocytes; (iii) non-transduced astrocytes cultured on L1-Fc and (iv) non-transduced astrocytes cultured on human Fc. AAV-L1 and AAV-GFP transduced astrocytes (1 x 106) were plated into 12-well plates coated with poly-L-lysine and maintained for 10 days in culture. Astrocytes (1 x 106) were then subcultured into 12-well plates coated with poly-L-lysine. Non-transduced astrocytes were subcultured into 12-well plates coated with either mouse L1-Fc (10 µg/ml, Loers et al., 2005) or human Fc (10 µg/ml, Dianova, Hamburg, Germany). The mouse L1-Fc contains the extracellular domain of mouse L1 fused with human Fc. Gel filtration of the L1-Fc fusion protein preparation showed that L1-Fc migrates as a monomer. Cells were cultured in DMEM to confluence, and then lysed and subjected to Western blotting for measurement of GFAP and NG2 expression levels.
Data acquisition and statistical analysis
All behaviour, morphological and biochemical tests were carried out and analysed in a blind manner. Two-sided Student's t-test for independent samples and one-way analysis of variance (ANOVA) for repeated measurements with Tukey post hoc tests were used to compare two and more than two groups, respectively. Significance threshold level was 0.05. Throughout the text and in the figures data are presented as means with standard errors of mean (SEM).
|
Results |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
AAV-mediated expression of exogenous genes in the lesioned spinal cord
We first studied whether AAV-5 stably transduces neurons and glia by injecting an AAV-5 vector encoding green fluorescent protein (GFP) into the lesion site immediately after spinal cord compression. Expression of GFP was detectable already 1 week after infection, the earliest time-point studied, increasing thereafter to reach peak levels at 3 to 4 weeks (data not shown). At 5 weeks, strong GFP signal was found in the spinal cord extending into the cervical and lumbar spinal cord over distances of at least 10 mm away from the centre of the lesion site (Fig. 2A–C). In the lesion scar, which is fibronectin-positive and GFAP-negative (Fig. 1, Chen et al., 2005a
), GFP was barely detectable (data not shown). Quantitative analysis revealed that 48% of all GFP-positive (AAV-GFP transduced) cells were NeuN-positive neurons (1112 NeuN+/GFP+ out of 2323 GFP+ cells analysed, Fig. 2H–J). As indicated by the size and location of the GFP-positive neurons, transduction was achieved in various cell types, including motoneurons and interneurons in the ventral horn. Of all GFP-positive cells, 29% were GFAP-positive astrocytes (663 GFAP+/GFP+/2307 GFP+ cells, Fig. 2K–M) and 3.2% were MBP-positive oligodendrocytes (63 MBP+/GFP+/1968 GFP+ cells, Fig. 2N–P). Thus, the AAV-5 vector is capable of efficiently and stably, for at least 5 weeks, transducing neurons and glia in the injured spinal cord.
|
We then investigated whether the full-length L1 cDNA of the AAV-L1 vector is expressed in the spinal cord. We could detect, in agreement with previous results (Runyan et al., 2005
), L1 expression only in non-myelinated axons in the superficial dorsal horn of intact spinal cords (data not shown). Five weeks after spinal cord injury and AAV-L1 injection, L1-positive cells were observed throughout the grey matter (Fig. 2D–G). Exogenous expression of L1 was observed in similar cell types as GFP expression. Compared with AAV-GFP, exogenous L1 expression was similar in the cord rostral and caudal to the lesion site but numbers of transduced cells were lower. The latter observation is explainable by the fact that L1 expression is weaker than GFP expression because AAV-GFP, but not AAV-L1, contains the WPRE post-transcription control element which stabilizes the mRNA, resulting in a 2–10-fold higher level of protein expression in comparison with L1 expression (Shevtsova et al., 2005). Quantitative Western blot analysis confirmed the immunohistochemical results on exogenous L1 expression. At 1 week after injury, L1 expression was reduced in both AAV-L1 and AAV-GFP-injected spinal cords compared with intact spinal cords (Fig. 2Q and R). L1 expression was significantly increased in AAV-L1-treated mice compared with AAV-GFP treated animals 2 to 5 weeks after injury (Fig. 2Q and R) indicating transgene L1 overexpression.
AAV-L1 improves recovery of motor functions
Spinal cord compression caused severe disabilities in both AAV-L1 and AAV-GFP-treated mice as indicated by the Basso Mouse Scale score (BMS, Engesser-Cesar et al., 2005) at 1 week after injury (0.30 ± 0.24 and 0.11 ± 0.16, respectively). Between 1 and 5 weeks after injury, the mean score values improved more in AAV-L1-treated than in AAV-GFP-treated mice to reach values of 5.1 ± 0.83 and 1.7 ± 0.68 in the two groups, respectively, at 5 weeks. The time course of BMS recovery is shown in Fig. 3A using recovery indices (see Material and methods section). Analysis of variance for repeated measurements with subsequent Tukey post hoc tests, both using indices (Fig. 3A) and score values (not shown) revealed better recovery at 3 and 5 weeks in the AAV-L1 group. In addition to the BMS, we analysed the plantar stepping ability of the animals using the foot-stepping angle (FSA). Since the mean values prior to and at 7 days after injury were similar in the two animal groups, like those for the BMS score and all other parameters described later, we show all data using recovery indices only. Analysis of the FSA also revealed, in agreement with the BMS scores (Fig. 3A), enhanced recovery in AAV-L1 compared to AAV-GFP-treated mice at 3 and 5 weeks (Fig. 3B). These results clearly indicate, based on use of two independent measures, that AAV-L1 application improves the abilities for ground locomotion after spinal cord injury. Good agreement between rating scale scores and the foot-stepping angle has also been observed previously (Apostolova et al., 2006).
|
We also evaluated more complex motor functions than plantar stepping, for example the rump-height index (RHI), a measure of the ability to support body weight during ground locomotion. In AAV-GFP-treated mice the degree of recovery at 5 weeks was close to zero, despite improved values seen at 2 and 3 weeks (Fig. 3C). Most likely, the transient increase at 2–3 weeks is due to spasticity rather than to functional improvement (Apostolova et al., 2006
for effects of spasticity on the RHI). In contrast to the control group, the index values in AAV-L1-treated mice recovered significantly, compared both to zero and to the AAV-GFP group, within the 5-week observation period (Fig. 3C). The animals’ ability to perform voluntary movements without body-weight support, estimated by the extension–flexion ratio (EFR), was not significantly affected by the kind of treatment, AAV-L1 or AAV-GFP application (Fig. 3D). The same conclusion was reached for numbers of correct steps made by the animals during inclined ladder climbing (Fig. 3E). The ladder-climbing test allows estimation of the ability to perform precise movements requiring highest degree of supraspinal control as compared to the types of movement assessed by the other tests. From the values of the parameters shown in Fig. 3A–E we calculated group mean values (Fig. 3F) and overall recovery indices for each animal (Fig. 3G). This analysis revealed an overall better outcome in mice treated with AAV-L1 compared to AAV-GFP application.
AAV-L1 promotes regeneration of 5-HT and CST axons
To investigate the morphological basis and cellular mechanisms which may contribute to the motor recovery, we first studied 5-HT descending axons in the injured spinal cord. All five mice injected with AAV-L1 had numerous 5-HT fibres growing into the lesion site. As evaluated in series of parasagittal sections, the fibres had reached the caudal border of the lesion site, extended beyond the border and entered the caudal part of the spinal cord (Fig. 4A, C, E and G). In contrast, with AAV-GFP, the 5-HT fibres were only occasionally observed in the lesion and caudal to it (H). This result was confirmed in consecutive cross-sections of the caudal part of the spinal cords. Six hundred micrometres caudal to the lesion site, robust 5-HT-positive fibre bundles were found in white and grey matter in the AAV-L1-treated spinal cords (K). In contrast, only very few fibres were observed in either white or grey matter at the same level in the AAV-GFP-treated mice (M). These findings indicate the L1-enriched environment in the lesioned spinal cord promotes regeneration of non-myelinated 5-HT axons by supporting regrowth or sprouting of spared axons.
|
|
After injury, the proximal segments of the myelinated CST axons retract back from the lesion site (Houle and Jin, 2001
). This ‘dying back’ allows detection of treatment-related effects such as enhanced axonal regrowth towards the lesion site or reduced retraction. We anterogradely labelled CST axons with rhodamine-conjugated dextran Fluororuby (B) and measured the distance between the tips of the labelled axons and the GFAP-positive rostral border of the lesion site (D). This distance was significantly shorter, by 800 µm, in AAV-L1 infected spinal cords compared to AAV-GFP infected spinal cords. Post-traumatic retraction of myelinated CNS axons occurs within minutes after injury (acute axonal degeneration, Kerschensteiner et al., 2005), a time frame during which exogenous L1 is apparently not expressed. We can, therefore, assume that, exogenous L1 expression supports, similar to 5-HT fibres, the regenerative response of myelinated axons which is normally very limited possibly as a result of missing guiding cues (Kerschensteiner et al., 2005).
AAV-L1 alters cellular responses to injury
The injured spinal cord is affected by reactive astrogliosis and progressive accumulation of CSPGs considered to be major neurite outgrowth inhibitory molecules (Fitch and Silver, 1997). We analysed homogenates from 500-µm long spinal cord segments containing the lesion site by Western blot analysis using antibodies against GFAP and NG2. As opposed to L1 expression, which was elevated in the AAV-L1-treated mice compared with the AAV-GFP-treated animals (Fig. 6A), there was a marked decrease in the levels of GFAP (Fig. 6B) and NG2 expression (Fig. 6C). This indicates that L1 overexpression limits astrogliosis and downregulates expression of growth-inhibitory molecules.
|
We also studied the expression of MBP to monitor treatment-related alterations in de-/remyelination in the injured spinal cord. We found that one of the MBP isoforms, the larger 21.5 kDa isomer, was significantly upregulated in the AAV-L1 group (Fig. 6D). No apparent differences between the groups were found for the other MBP isomers (18.5, 17.0 and 14.0 kDa, left panel in Fig. 6).
To determine whether L1 expression also alters the microglial reaction at the lesion site, we used the microglial marker Iba-1 by immunohistochemistry and Western blot analysis. We found, in contrast to the other molecular markers, no differences between AAV-L1 and AAV-GFP-treated mice (data not shown).
Finally, we analysed the expression of the cell fate determinant factor Numb which was found to improve neurite outgrowth by promoting L1 endocytosis at growth cones (Nishimura et al., 2003), suggesting that the L1 and Numb signalling pathways may affect each other in regulating neurite outgrowth. The Western blot analysis revealed that Numb expression was higher in AAV-L1-treated compared with AAV-GFP transduced spinal cords (Fig. 6E).
L1 affects signal transduction mechanisms
To investigate the molecular mechanisms underlying reduced astrogliosis, we studied pathways implicated in L1-mediated signal transduction. We found that AAV-L1 increased the phosphorylated : total value ratio of ERK1/2 when compared with the AAV-GFP group (Fig. 7A), while the total protein form of ERK1/2 was not affected. In contrast, phosphorylation of PKA was similar in the two groups. However, the total level of PKA expression was elevated in the AAV-L1 group compared with the AAV-GFP group, causing a decrease in the phosphorylated : total value ratio in the AAV-L1 group (Fig. 7B). Total PI3K levels was also elevated in the AAV-L1 group compared with the AAV-GFP group (Fig. 7C). We did not detect phosphorylated PI3K in the AAV-L1 or AAV-GFP groups. Previous observations had suggested that GFAP levels in glial cells are affected by the cAMP response element binding (CREB) protein through interacting with transcription factors, which interact with the AP-1 DNA binding sequence in the GFAP promoter (Pennypacker et al., 1996). We analysed spinal cord homogenates by Western blotting using antibodies against phospho-CREB and total-CREB. The phosphorylated : total value ratio of CREB was significantly higher in the AAV-L1 group than in the AAV-GFP group (Fig. 7D) indicating that exogenous L1 expression increases CREB activation.
|
L1-triggered activation of the ERK1/2 pathway has been shown to induce cell motility-associated gene products, among them the small GTPases Rac-1 (Silletti et al., 2004
), which also stimulates neurite outgrowth in vivo (Yip et al., 1998; Schmid et al., 2000; Causeret et al., 2004). We observed that the small GTPase Rac1 expression levels were considerably higher in AAV-L1 transduced compared with AAV-GFP transduced spinal cords (B), suggesting that RhoA activation is decreased by overexpression of L1. The elevation of the CREB activation suggests that cyclic AMP might be involved in the L1 pathway after the AAV transduction. We indeed found higher cAMP level in the AAV-L1 than the AAV-GFP-treated groups (Fig. 8C).
|
AAV-L1 inhibits astrocyte proliferation, migration and process extension in vitro
To further explore the mechanisms underlying reduced astrogliosis in vivo, we investigated proliferation and migration of AAV-L1 and AAV-GFP transduced astrocytes in vitro (Fig. 9A). The level of astrocyte proliferation was significantly reduced by a factor of 2 after AAV-L1 compared with AAV-GFP transduction (Fig. 9B). The MAPK inhibitors U0126 and PD98059 did not significantly change the extent of proliferation of AAV-L1 and AAV-GFP transduced astrocytes as compared with the corresponding untreated cells (Fig. 9B). This result suggests that L1 does not involve the MAPK pathway to inhibit the proliferation of astrocytes.
|
We investigated whether overexpression of L1 activates the FGFr3 pathway to reduce astrocyte proliferation and to downregulate GFAP expression (Pringle et al., 2003
; Loers et al., 2005). After treatment with FGFr3 neutralization antibody, the proliferation of both AAV-L1 and AAV-GFP transduced astrocytes was increased compared to non-treated astrocyte cultures (Fig. 9B). Thus, L1 may not regulate astrocyte proliferation through the FGFr3. To investigate migration of astrocytes, we used an in vitro scratch assay in confluent AAV-L1 and AAV-GFP transduced cultures. Four hours after the scratch, more AAV-GFP transduced astrocytes migrated into the denuded area than AAV-L1 transduced astrocytes. Moreover, AAV-GFP transduced astrocytes at the edges of the scratch had extended processes into the scratched area, while their cell bodies had remained at the edge (Fig. 9C). AAV-L1 transduced astrocytes barely extended processes into the free space (Fig. 9C). Twelve hours later, the initially cell-free area (about 300 µm) was largely closed by the AAV-GFP transduced astrocytes, while the space in the AAV-L1 transduced cultures remained empty. These observations show that L1 reduces astrocyte migration in vitro.
AAV-L1 changes GFAP but not NG2 expression levels in astrocyte cultures
L1 mediates neurite outgrowth through homophilic or heterophilic interactions (Appel et al., 1993; Beggs et al., 1997; Kamiguchi and Yoshihara, 2001). Our in vivo experiments showed that L1 overexpression reduced GFAP expression in the injured spinal cord. To explore whether this reduction is due to homophilic and/or heterophilic interaction of L1, we measured GFAP levels in cultured astrocytes after transduction with AAV-L1 or treatment with substrate-coated L1-Fc. AAV-L1 transduced astrocytes showed lower levels of GFAP expression than AAV-GFP transduced astrocytes, and L1-Fc treated and Fc alone treated astrocytes (Fig. 10). GFAP levels in astrocytes cultured on L1-Fc substrate were not different in the L1-Fc and Fc-treated cultures. These results suggest that L1 reduces GFAP levels by homophilic, but not heterophilic interactions.
|
Since the NG2 levels were also reduced after AAV-L1 transduction of injured spinal cords in vivo, we investigated whether AAV-L1 transduction or L1-Fc treatment influences NG2 expression in cultured astrocytes as measured by Western blot analysis. In contrast to the findings on GFAP expression levels, neither AAV-L1 nor L1-Fc treatment affected NG2 expression (data not shown).
|
Discussion |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Beneficial effects of L1 overexpression on spinal cord regeneration
During spinal cord development, L1 is abundantly expressed by growth cones and axons and becomes downregulated to low levels in the spinal cord of adult mice. Considering the fact that L1 is not upregulated after spinal cord injury in adulthood and that L1 is beneficial for axonal regrowth and neuronal survival (Roonprapunt et al., 2003
; Xu et al., 2004; Chen et al., 2005b), we have overexpressed L1, more prominently in neurons, then in astrocytes and oligodendrocytes, in the lesioned spinal cord of adult mice using an AAV-5 vector. As opposed to control treatment, L1 overexpression in the local microenvironment significantly improved recovery of locomotion which was associated with enhanced axonal regeneration, reduced astrogliosis and attenuated expression of growth-inhibiting molecules such as NG2. Analysis of astrocyte cultures and signal transduction pathways revealed diverse L1-specific effects that can be interpreted as an overall shift of the inhospitable environment in the injured spinal cord to a more hospitable one.
Possible mechanisms underlying improved functional recovery
Using a set of tests to evaluate different motor abilities, we found significant overall improvements in AAV-L1-treated mice compared to control animals. The enhanced overall motor performance of AAV-L1-treated mice at 5 weeks after lesion was mainly due to recovery of walking as indicated by three different measurements: the BMS, the FSA and the RHI. In adult mammals, the abilities to stand and walk after spinal cord injury barely recover spontaneously despite preservation of circuitries distal to the injury, which are capable of initiating and controlling rhythmic and coordinated movements (Edgerton et al., 2004; Fouad and Pearson, 2004). Initiation of activity in locomotor pattern generating centres in the injured spinal cord are achieved by pharmacological treatments activating aminergic transmitter systems (Fouad and Pearson, 2004). Among the transmitters effectively influencing locomotion in rodents is 5-HT. The 5-HT system in the spinal cord affects rhythmic efferent discharges in hind limb nerves (Viala and Buser, 1969), participates in initiation and modulation of locomotor patterns during walking (Barbeau and Rossignol, 1991), and enhances learning-related plasticity in the injured spinal cord (Crown and Grau, 2005). The degree of 5-HT innervation of the injured spinal cord correlates with motoneuron excitability as estimated by analyses of H-reflex responses (Lee et al., 2005). Other researches have also shown a positive correlation between regrowth of 5-HT axons, which originate in the medullary raphe nuclei and the reticular formation (Shapiro, 1997), and degree of functional recovery (Ribotta et al., 2000; Pearse et al., 2004; Fouad et al., 2005). 5-HT fibres provide diffuse innervation and thus a more vigorous axonal regrowth, even if not specifically targeted, is beneficial, in contrast to other projections that require precise re-establishment of synaptic connections. We observed robust 5-HT axonal regeneration in the lumbar spinal cords of AAV-L1-, but not AAV-GFP-treated mice, which could well explain the enhanced improvement of walking.
The finding that CST axons in AAV-L1-treated mice reach much closer proximity to the lesion scar, by some 800 µm compared with control animals, at 5 weeks can be interpreted as an indication for enhanced axonal regeneration, even if no regrowth beyond the lesion scar was seen. Also, the finding of a better serotonergic innervation, which could result from axonal regrowth across the lesion site or sprouting of spared axons, is a clear sign of enhanced regenerative potential when L1 is abundantly expressed in the local environment. Improved axonal regeneration under such conditions would be possible, via heterophilic interactions, even if L1 is not expressed in the regrowing axons. In addition to local axonal regrowth, L1 may also positively influence lesion-induced alterations in synaptic connectivity. The importance of such alterations in the injured spinal cord has recently been indicated by studies on enhanced structural plasticity of synaptic inputs to motoneurons in mice deficient in the growth-inhibiting molecule tenascin-R correlating with better recovery of the animals’ walking abilities (Apostolova et al., 2006).
Exogenous L1 affects signal transduction mechanisms
We observed elevated levels of cyclic AMP, phosphorylated CREB and MAPK in AAV-L1-treated spinal cords. Local application of cyclic AMP leads to enhanced axonal regrowth and functional recovery after spinal cord injury (Lu et al., 2004; Pearse et al., 2004; Spencer and Filbin, 2004). Also, activated CREB promotes regeneration of dorsal root ganglion axons in the injured spinal cord (Gao et al., 2004). Activated MAPK regulates neurite outgrowth and regeneration by inhibiting phosphodiesterase 4, the enzyme that hydrolyses cAMP (Gao et al., 2003; Chierzi et al., 2005). Since the Western blot analysis does not allow to distinguish effects on neuronal from glial cells, here we cannot determine whether changes in signal transduction occur in projecting or local neurons, interneurons or glial cells. However, we can speculate that CREB can regulate astrogliosis by binding to the AP-1 DNA binding sequence in the GFAP gene, thereby regulating GFAP expression in astrocytes and thus reactive astrogliosis (Pennypacker et al., 1996).
L1-triggered activation of the ERK1/2 pathway has been shown to induce cell motility-associated gene products, among them the small GTPases Rac-1 (Silletti et al., 2004), which also stimulates neurite outgrowth in vivo (Yip et al., 1998; Schmid et al., 2000; Causeret et al., 2004). In this study we found that Rac1 levels were elevated in L1 overexpressing spinal cords, suggesting that L1 induces expression of Rac1, which is beneficial to axonal regrowth. The small GTPase RhoA is another regulator of the actin cytoskeleton in neurites and its activation results in growth cone collapse, neurite retraction and neurite growth inhibition (Lehmann et al., 1999; Wahl et al., 2000). Inhibitory molecules, such as Nogo-A, myelin associated glycoprotein, ephrins, semaphorins and CSPGs can activate the RhoA pathway, and hence block regrowth of axons (Monnier et al., 2003; Mueller et al., 2005). Studies in EphA4-deficient mice have shown that RhoA activation leads to worse regeneration and enhanced astrogliosis after spinal cord injury (Goldshmit et al., 2004). In this study, we found that GTP-RhoA levels were decreased in L1 overexpressing spinal cords, while the total RhoA levels were not changed, indicating that also under our experimental conditions reduced RhoA activation might be beneficial for axonal regrowth.
Numb is an important cell fate determinant which was originally identified as an antagonist of Notch signalling (Uemura et al., 1989). Although Numb is barely detectable in adult rodent spinal cords, it is highly elevated after spinal cord injury (Chen et al., 2005a). Numb is associated with L1 and functions in endocytosis of L1 in growth cones to promote axonal growth (Nishimura et al., 2003). We here report that L1 overexpression is able to elevate Numb levels in the injured spinal cord, providing further evidence for a relationship between Numb and L1.
L1 overexpression reduces astroglial proliferation
Astrocytes in the vicinity of the fibronectin immunoreactive lesion core are inhibitory to growing axons and in vitro experiments have shown that L1-modified astrocytes provide a more permissive growth substrate (Adcock et al., 2004). Here we observed that expression of exogenous L1 in vivo decreases the expression of GFAP, an intermediate filament protein indicative of astrogial proliferation (Pekny et al., 1999; Pekny and Pekna, 2004). Although it is not known whether decreased GFAP expression is linked to promoting axonal regrowth (Wang et al., 1997; Xu et al., 1999; Menet et al., 2000, 2001), our observation strongly suggests a less severe reactive astrogliosis, and hence an ameliorated micro-environment at the lesion site. This notion was further supported by in vitro analyses showing that L1 overexpression reduces astrocyte proliferation, an effect apparently not dependent on MAPK and FGFr3 signalling, which is considered to be important for astrocyte development and GFAP expression (Pringle et al., 2003). Furthermore, L1 expression reduces migration and process formation of astrocytes, phenomena that may be related to reduced GFAP expression.
L1 mediates neurite outgrowth through homophilic or heterophilic interactions (Appel et al., 1993; Beggs et al., 1997; Kamiguchi and Yoshihara, 2001). Our in vivo experiments showed that L1 overexpression reduced GFAP expression in the injured spinal cord. To explore whether this reduction is due to homophilic and/or heterophilic interaction of L1, we measured GFAP levels in cultured astrocytes after transduction with AAV-L1 or treatment with substrate-coated L1-Fc. As reported previously, cultured astrocytes do not express L1 as confirmed by Western blotting and immunocytochemistry. When the astrocytes are cultured on an L1-Fc coated substrate, L1 predominantly affects astrocytes by heterophilic binding. However, when AAV-L1 transduced astrocytes express L1 at their cell surface, L1 can bind to L1 molecules on adjacent cells via homophilic (L1 binds to another L1 molecule) or heterophilic (L1 binds to a non-L1 molecule) interactions, or both. The finding that L1 overexpression decreases GFAP expression, but L1-Fc does not do so, suggests that L1 regulates GFAP expression by homophilic binding. Overall, we can conclude that the positive effects of exogenous L1 on spinal cord regeneration are likely related to L1 expression and signal transduction in astrocytes.
Another observation indicating a more friendly environment in injured AAV-L1-treated spinal cords is downregulation of NG2. This molecule inhibits axonal growth via RhoA/ROCK signalling (Monnier et al., 2003) and our observation that RhoA activation is decreased in L1 overexpressing spinal cords is consistent with the downregulation of NG2. NG2 can be expressed in a variety of cell types including synaptocytes/polydendocytes, oligodendrocyte precursors, macrophages and astrocytes after spinal cord injury (Fidler et al., 1999; Berry et al., 2002; Jones et al., 2002). The finding that L1 expression in cultured astrocytes does not lead to changes in NG2 expression indicates that astrocytes are not the major source of NG2 in the injured spinal cord.
L1 overexpression enhances expression of MBP
Since only 3% of all transduced cells were oligodendrocytes, the robust upregulation of the 21.5 kDa isoform of MBP was unexpected. This isoform was reported to accumulate in nuclei of HeLa cells (Staugaitis et al., 1990) and oligodendrocytes (Pedraza et al., 1997). Because of its developmentally early expression, it may play an important role in transcriptional regulation and initiation of myelination. Its expression also correlates with remyelination in multiple sclerosis (Capello et al., 1997). Interestingly, L1 was also found to play a role in the initiation of CNS myelination by regulating axon-oligodendrocyte interactions (Barbin et al., 2004). It remains to be seen whether L1 overexpression influences expression of 21.5 kDa MBP directly or indirectly. Thus, L1 overexpression appears to be beneficial for an important ingredient in recovery after spinal cord injury, namely remyelination. Whether L1 expression in axons, oligodendrocytes or both is required to achieve this effect remains to be elucidated.
Outlook
Here we show that an AAV construct can be used as an efficient vector to introduce a beneficial adhesion molecule into the injured spinal cord. By transducing cells in the spinal cord with the adhesion molecule L1, we observed improved motor functional recovery, enhanced axonal regeneration and amelioration of the local microenvironment. We attribute these effects to the ability of L1 to activate multiple signalling pathways in regrowing neurites and/or surrounding tissue, to limit reactive astrogliosis and to influence intraspinal circuitries. Results from in vitro experiments support the idea that L1 decreases astrocytic proliferation, migration, process extension and GFAP expression, which are indicators of an astrocytic reaction to injury. Thus, L1 transduction using an AAV vector is likely to provide incentives to a therapeutic venue to spinal cord injury in adult mammals, including humans.
|
Footnotes |
|---|
*The last three authors contributed equally to this work.
|
Acknowledgements |
|---|
We thank Dr Gabriele Loers and Ute Bork for help with cultures of astrocytes, and Kathrin Hilke-Steen for typing the manuscript. We are grateful to the New Jersey Commission of Spinal Cord Research (06-2917-SCR-3-0), the Canadian Spinal Research Organization and a BMBF sponsored programme for collaboration in neuroscience between Poland and Germany for support. Melitta Schachner is New Jersey Professor for Spinal Cord Research.
|
References |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Adcock KH, Brown DJ, Shearer MC, Shewan D, Schachner M, Smith GM, et al. (2004) Axon behaviour at Schwann cell–astrocyte boundaries: manipulation of axon signalling pathways and the neural adhesion molecule L1 can enable axons to cross. Eur J Neurosci 20:1425–35.[CrossRef][ISI][Medline]
Apostolova I, Irintchev A, Schachner M. (2006) Tenascin-R restricts posttraumatic remodeling of motoneuron innervation and functional recovery after spinal cord injury in adult mice. J Neurosci 26:7849–59.
Appel F, Holm J, Conscience JF, Schachner M. (1993) Several extracellular domains of the neural cell adhesion molecule L1 are involved in neurite outgrowth and cell body adhesion. J Neurosci 13:4764–75.[Abstract]
Appel F, Holm J, Conscience JF, von Bohlen und Halbach F, Faissner A, James P, et al. (1995) Identification of the border between fibronectin type III homologous repeats 2 and 3 of the neural cell adhesion molecule L1 as a neurite outgrowth promoting and signal transducing domain. J Neurobiol 28:297–312.[CrossRef][ISI][Medline]
Barbin G, Aigrot MS, Charles P, Foucher A, Grumet M, Schachner M, et al. (2004) Axonal cell-adhesion molecule L1 in CNS myelination. Neuron Glia Biol 2:65–72.
Barbeau H and Rossignol S. (1991) Initiation and modulation of the locomotor pattern in the adult chronic spinal cat by noradrenergic, serotonergic and dopaminergic drugs. Brain Res 546:250–60.[CrossRef][ISI][Medline]
Becker CG, Lieberoth BC, Morellini F, Feldner J, Becker T, Schachner M. (2004) L1.1 is involved in spinal cord regeneration in adult zebrafish. J Neurosci 24:7837–42.
Beggs HE, Baragona SC, Hemperly JJ, Maness PF. (1997) NCAM140 interacts with the focal adhesion kinase p125(fak) and the SRC-related tyrosine kinase p59(fyn). J Biol Chem 272:8310–9.
Berry M, Hubbard P, Butt AM. (2002) Cytology and lineage of NG2-positive glia. J Neurocytol 31:457–67.[CrossRef][ISI][Medline]
Bett AJ, Haddara W, Prevec L, Graham FL. (1994) An efficient and flexible system for construction of adenovirus vectors with insertions or deletions in early refion 1 and 3. Proc Natl Acad Sci USA 91:8802–6.
Buchli AD and Schwab ME. (2005) Inhibition of Nogo: a key strategy to increase regeneration, plasticity and functional recovery of the lesioned central nervous system. Ann Med 37:556–67.[CrossRef][ISI][Medline]
Capello E, Voskuhl RR, McFarland HF, Raine CS. (1997) Multiple sclerosis: re-expression of a developmental gene in chronic lesions correlates with remyelination. Ann Neurol 41:797–805.[CrossRef][ISI][Medline]