| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Research Center for Cellular and Molecular Neurobiology, University of Liège, Liège, Belgium
1Correspondence: Research Center for Cellular and Molecular Neurobiology, University of Liège, Tour de Pathologie B36, 1erétage, local 1/4A, CHU Sart-Tilman 4000 Liège, Belgium. E-mail address: rfranzen{at}ulg.ac.be
ABSTRACT
Macrophages (monocytes/microglia) could play a critical role in central nervous system repair. We have previously found a synchronism between the regression of spontaneous axonal regeneration and the deactivation of macrophages 3–4 wk after a compression-injury of rat spinal cord. To explore whether reactivation of endogenous macrophages might be beneficial for spinal cord repair, we have studied the effects of granulocyte-macrophage colony stimulating factor (GM-CSF) in the same paraplegia model and in cell cultures. There was a significant, though transient, improvement of locomotor recovery after a single delayed intraperitoneal injection of 2 µg GM-CSF, which also increased significantly the expression of Cr3 and brain-derived neurotrophic factor (BDNF) by macrophages at the lesion site. At longer survival delays, axonal regeneration was significantly enhanced in GM-CSF-treated rats. In vitro, BV2 microglial cells expressed higher levels of BDNF in the presence of GM-CSF and neurons cocultured with microglial cells activated by GM-CSF generated more neurites, an effect blocked by a BDNF antibody. These experiments suggest that GM-CSF could be an interesting treatment option for spinal cord injury and that its beneficial effects might be mediated by BDNF.—Bouhy, D., Malgrange, B., Multon, S., Poirrier, A. L., Scholtes, F., Schoenen, J., Franzen, R. Delayed GM-CSF treatment stimulates axonal regeneration and functional recovery in paraplegic rats via an increased BDNF expression by endogenous macrophages.
Key Words: spinal cord injury • axonal regrowth • cytokine • neurotrophin
WHEREAS NERVES REGENERATE in the peripheral nervous system (PNS), axonal regeneration in the central nervous system (CNS) rapidly aborts. This seems mainly due to the unfavorable environment through which central axons have to regenerate. The immune system plays a key role in the healing of injured tissues, with the immune cells secreting consecutively toxic and trophic factors. These healing processes are favorable to the regeneration of injured PNS axons but are much less effective after CNS injury.
Boosting or modulating the immune response seems to be a promising strategy for a successful CNS repair. Macrophages (monocytes/microglia) are thus considered to be key actors for a successful axonal regeneration. Several in vivo studies show indeed that grafts of activated macrophages can promote regeneration (1
2
3
4)
, and a phase I/II clinical trial of grafts of activated macrophages is ongoing in spinal cord injured patients (www.proneuron.com).
Besides their capacity to eliminate growth-inhibiting myelin products by phagocytosis (2)
, macrophages are also a source of cytokines and growth factors that actively promote regrowth (4
5
6
7
8
9
10
11
12)
.
Although early inflammatory responses may participate in secondary injury processes (13)
, more delayed inflammatory events may be reparative. Macrophage phenotype varies with postinjury delays (14)
. Indeed, we found that 21 days after a spinal cord compression-injury (SCI) in rats a decrease in the number of activated (OX42-immunoreactive) macrophages that occurred in parallel with the regression of spontaneously regenerating axons (15)
. We hypothesized that this phenotypic shift of macrophages could be causally related to the absence of sustained axonal regeneration in the spinal cord and that reactivating macrophages 3–4 wk after injury might improve axonal regrowth and locomotor recovery.
Growing evidence indicates granulocyte macrophage-colony stimulating factor (GM-CSF) involvement in macrophage activation. Peripheral nerve lesion leads to a major GM-CSF production, which in turn regulates molecular and cellular events such as recruitment and activation of the phagocytic activity of Schwann cells and macrophages (16
17)
. GM-CSF also stimulates the proliferation of neural progenitor cells (18)
, inhibits the apoptosis of neurons and neural progenitor cells (18
19)
, and has neuritogenic effects (20
, 21)
. In addition, GM-CSF is an already widely used cytokine in clinical onco-hematological disorders (see ref 22
for a review). We therefore decided to study the effects of a delayed GM-CSF treatment on axonal regeneration and locomotor recovery in adult paraplegic rats. We wondered if GM-CSF could act by increasing neurotrophic factor release by macrophages and focused our work on brain-derived neurotrophic factor (BDNF), a neurotrophin well known for its favorable effects on axonal regeneration (23)
. We also tested this hypothesis on microglial cell cultures. Our results indicate that a single delayed dose of GM-CSF induces locomotor improvement and axonal regrowth in severely injured rats and suggest that these effects could be mediated by an increased production of BDNF by activated macrophages.
MATERIALS AND METHODS
Surgical procedures and GM-CSF treatment
Eighty-three adult female Wistar rats (250 g) were used in this study. The experiments were performed in accordance with the rules and regulations of the Ethical Committee for animal research of the Belgian National Fund for Scientific Research. All rats underwent a "closed" spinal cord compression-injury at the T8-T9 level with a subdural inflatable micro balloon as described previously (24)
. Briefly, after anesthesia with intraperitoneal injection of xylazine (10 mg/kg) and ketamine (80 mg/kg), the skin is opened and paravertebral muscles are detached. A single metameric laminectomy is performed at thoracic level T10. After laminectomy, an inflatable microballoon mounted on a catheter (Ingenor, model GV15, Paris) is inserted through a small incision in the subdural space and moved gently to a rostral position approximately two segments above the laminectomy (T8-T9). The balloon is then inflated with 20 or 40 µl distilled water and left in place for 5 min, after which it is deflated and withdrawn, and the surgical wound is closed. After surgery, care was taken to prevent dehydration by intraperitoneal physiological saline injections and infection by intraperitoneal antibiotics (amoxicilline-clavulanic acid) during 1 wk. The bladder was manually expressed every day, until spontaneous micturition resumed. Food and water were provided ad libitum.
Four weeks after the compression-injury, half of the animals (n=42) received a single intraperitoneal injection of 2 µg recombinant rat GM-CSF (R&D Systems) diluted in 1 ml saline and the other half (n=41) a single intraperitoneal injection of 1 ml physiological saline. Animals were killed at various survival delays. Their repartition in the different experimental groups is described in Table 1
.
|
Corticospinal tract anterograde tracing
Rats that underwent a 20 µl compression lesion (n=20) were used for anterograde tracing of the corticospinal tract 6 wk postinjury. The anterograde tracer biotinylated dextran amine (BDA MW 10000, 50 mg/ml solution in PBS, Molecular Probes) was stereotaxically injected into the sensorimotor cortex of both hemispheres using a Hamilton syringe (5 injections/side, 0.5 µl each, Bregma –0.8, –1.3, –1.8, –2.3, –2.8; lateral 2 or 2.5, depth 2 mm). Animals were maintained for 2 more wk then killed. After cryosectioning of the tissue, BDA was revealed histochemically on mounted sagittal sections by using a Vector Elite avidin-biotin complex (ABC) kit (Vector laboratories) followed by development with diaminobenzidine {3,3'-diaminobenzidine (DAB)}.
Behavioral analysis
The motor function in hind-limbs was evaluated using the Basso, Beattie, Bresnahan (BBB) open field locomotor test, in which the scores range from 0 to 21 (25)
. Before testing, bladders were expressed, because spontaneous bladder contraction often accompanies hind-limb activity. Briefly, individual rats were placed in an open field. Two examiners, blinded to treatment, observed the animals for 4 min and scored motricity in both hind-limbs according to the BBB rating scale. Hind-limb movements immediately after contact with experimenters were disregarded. The mean of right and left hind-limb scores was taken into account. The test was carried out 1 day postoperatively, and only rats with a BBB score of maximum 1 were kept for the study. The test was carried out once every week up to the end of the experiment (24 wk). Four weeks postinjury, rats were allocated to the "GM-CSF group" or to the "control group" in such a manner that the mean BBB scores of both experimental groups were comparable.
Immunohistochemistry
After appropriate survival periods, all animals were deeply anesthetized by an intraperitoneal injection of Nembutal (0.7 ml) and perfused with 4% paraformaldehyde in 1.0 M phosphate buffer, pH 7.4. The spinal cords were dissected out over their entire length, postfixed in 4% paraformaldehyde for 24 h at 4°C, and kept for 48 h in 30% sucrose at 4°C for cryoprotection. They were then cut longitudinally or transversally at 20 µm on a cryostat and mounted onto gelatin-coated slides. Transverse sections were hematoxylin-eosin stained to evaluate the severity of the injury. Longitudinal sections, covering the full-length of the lesion site and taken from the dorsal part of the cord, were used for immunostaining. Sections from rats killed 1, 3, 5, or 7 days post-treatment were immunostained with Cr3 Ab (CD11b, clone OX-42; diluted 1:1000; Serotec) and BDNF Ab (1:500; R&D Systems) to assess GM-CSF short-term effects on macrophage activation and trophic factor expression, respectively. Double immunofluorescence staining for Cr3 and BDNF was also performed to characterize BDNF-expressing cells. Sections from 24 wk survival animals were immunostained for 43 kDa growth-associated protein (GAP-43; 1:2000; Chemicon), 200 kDa neurofilaments (clone NE-14, 1:6000, Sigma), and 5-hydroxytryptamine (5-HT; 1:200,000, Diasorin) to visualize regenerating axons and serotonergic profiles, respectively. Immunostaining for glial fibrillary acidic protein (GFAP) was done to delineate the lesion site (1: 10,000, DAKO). Sections were incubated in a 3% H2O2 solution in PBS for 10 min at room temperature (RT) to reduce endogenous peroxydase activity. After 3 PBS rinses, nonspecific binding was prevented by incubating the sections in 3% normal goat serum (NGS) in PBS-Triton 0.1% (PBS-T). The sections were incubated overnight at RT with the specific primary Ab diluted in a 1% NGS solution in PBS-T. They were rinsed with PBS-T and incubated with the corresponding secondary biotinylated Ab (1/1000, Vector) in a 2% rat normal serum solution for 60 min at RT. After three PBS rinses, a 1 h RT incubation with the avidin-biotin-peroxydase complex (Labconsult), diluted 1/1000 in PBS, was performed. The slides were then rinsed in PBS and immunoreactivity was visualized with DAB.
For double immunostaining, tissue was permeabilized with a 0.2% triton solution in Tris-buffered saline (TBS-T) for 15 min at RT and nonspecific binding was prevented by a 30 min incubation in 1.5% BSA solution in TBS. Sections were then incubated overnight at RT with the specific primary antibodies against Cr3 (1/1000) and BDNF (1/100), rinsed 3x with TBS-T and incubated with their respective secondary antibodies diluted in a 1% solution of normal donkey serum in TBS (rhodamine-donkey anti-mouse, 1/500 and FITC-donkey anti-chicken 1/50, Jackson) for 4 h at RT. Sections were then rinsed 3x in TBS, 1x in distilled water, and mounted with Vectashield medium (Labconsult).
Image analysis
Bright field images of longitudinal sections from the dorsal part of the cord were taken using an Olympus AX70 microscope equipped with an Olympus DP50 digital camera. Sections were viewed with a x10 or a x20 objective and captured through the digital camera attached to the microscope with exposure time, brightness, and contrast being held constant. All analyses were done blinded to the treatment the rat had received.
OX42-positive macrophages were manually counted within nonoverlapping successive fields covering the entire lesion, and results were expressed as a density of OX42-positive macrophages per micrometers squared. The results represent the average density of six animals/experimental group.
For the quantification of GAP43, NF, 5HT, and BDNF immunostainings, and corticospinal tract ingrowth, pictures were converted to gray scale. A threshold intensity of gray-colored staining was fixed for each slice. Then, a surface covering the entire lesion area was selected (=total lesion area). Thereafter, immunostaining was quantified within nonoverlapping successive fields covering the total lesion area (=total stained area) using an image analysis program (Image Processing and analysis in Java, Wayne Rasband, in the public domain). Data were expressed as the percentage of total stained area (pixel) per total lesioned area (pixel). The results represent averages from 5 or 7 rats in each group, 10 for the CST tracing.
Cell cultures
BV-2 murine microglia cell line (BV2 cells; a generous gift from Dr. Personett, Mayo Clinic) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% heat-inactivated fetal calf serum (FCS) in 24-well culture plates precoated with polyornithine and laminin. In half of the wells, recombinant rat GM-CSF (25 ng/ml, R&D Systems) was added for 24 h. Freshly dissociated adult dorsal root ganglion neurons (DRG; for method see ref (26)
or P7 cerebellar neurons were isolated and plated in all culture wells. After 16 h of coculture, cells were paraformaldehyde-fixed and neurons were visualized with the anti-ßIII tubulin TUJ-1 Ab (1:1500; BABCO).
The neuritic index of adult DRG neurons cocultured with GM-CSF-activated BV2 microglial cells, or with nonactivated BV2 cells, was determined by the following method: all neurons present in 15 microscopic fields (objective x40)/culture well (5 wells/culture conditions) were counted, and the percentage of those presenting neurites was calculated. Neurites were counted only if they had an obvious attachment to the neuronal soma. An average of 350 neurons/well was quantified. For cerebellar neurons, the mean length of the longest neurite for each TUJ-1 positive neuron was calculated. One-hundred microscopic fields (objective x20) per experimental conditions have been investigated. The longest neurite of every single neuron was measured. An average of 320 neurons/condition was determined.
Western blots
Confluent BV2 cells were maintained for 24 h in culture in DMEM with 10% FCS in the presence or absence of recombinant rat GM-CSF (25 ng/ml, R&D Systems). Cells were then rinsed with ice-cold PBS before being collected in lysis buffer containing protease inhibitors (Cocktail Tablet, Roche), phosphatase inhibitors (500 mM NaF and 100 mM Na3VO4), and 1% Triton in PBS for total protein extraction. Lysates were centrifuged at 14,000 g for 15 min at 4°C, and the protein concentration of each sample was determined using the bicinchoninic acid protein assay reagent kit (Pierce).
Whole cell lysates with equal amounts of proteins were separated on 4–12% NuPAGE, and the resolved proteins were electrotransferred to nitrocellulose (Invitrogen). Membranes were incubated with 5% nonfat milk in buffer containing 0.1% Tween 20 for at least 60 min at room temperature and incubated with a chicken anti-BDNF polyclonal antibody (1 µg/ml; R&D Systems) overnight at 4°C. Membranes were then washed three times with PBS and incubated for 1 h at RT with a biotinylated antichicken Ab (1/1000, Labconsult) followed by the ABC (Vector) for 60 min. Finally, membranes were washed several times and DAB visualized. After several PBS rinses, densitometry of bands was carried out by scanning the blots into the computer and using ImageQuant Software (Amersham Biosciences) to determine the mean pixel density for each band. Each experiment was repeated at least twice; the same results were obtained in all cases.
Statistical analyses
All behavioral and histological measurements were performed in a blinded manner. Statistical comparisons between groups were made by ANOVA with planned comparisons using a statistical analyses program (STATISTICA, StatSoft Benelux). Significance concentration was set at P < 0.05. All values are means ± SE. For the neuritic index, the CST in-growth quantification, and GAP-43, NF, 5-HT, BDNF, or OX42 quantifications in vivo, the Mann and Whitney U test was used with a significance level at P < 0.05.
RESULTS
GM-CSF improves locomotor performance
Four weeks after the compression-lesion at T8–9, the mean BBB score of all injured animals was 0.86 ± 0.2. After allocation to treatment groups, mean scores were 1.1 ± 0.4 in the GM-CSF group and 0.61 ± 0.30 in the control group (Fig. 1
A, arrow). Such scores correspond to absence of any detectable movement in any of the three hind-limb joints (ankle, knee, and hip).
|
As soon as 1 wk after the intraperitoneal injection, GM-CSF treated rats had a significantly higher mean BBB score (1.8±0.4), whereas the control group did not improve at all (0.61±0.3; P<0.05). The difference between the two groups remained obvious over the 10 subsequent wk, reaching the level of statistical significance at post-treatment weeks 4 and 7 (Fig. 1A
). BBB scores between 4 and 6 were only reached by animals belonging to the GM-CSF group and not by those treated with saline (Fig. 1B
). Scores of 4 to 6 indicate that the rats are able to move all three hind-limb joints slightly or extensively, whereas lower scores in the control group reflect at best slight motility at one or two joints. From 10 wk post-treatment, onwards BBB scores tended to plateau in both groups. Over post-treatment weeks 10 to 20 of observation, the mean BBB score did not differ significantly between the two groups although scores in GM-CSF treated rats remained consistently higher (mean BBB of 2.11±0.54 in controls and of 3.11±0.62 in GM-CSF treated rats at 20 wk posttreatment).
GM-CSF enhances fiber regeneration
Injection of the anterograde tracer BDA into the sensorimotor cortex of 20 µl injured rats resulted in the labeling of the corticospinal tract. On longitudinal sagitally cut sections, 8 wk postinjury, BDA-positive corticospinal tract axons were detected rostrally to the lesion site, with regenerating fibers entering the lesion site (Fig. 2
A). The tortuous path of the traced fibers is most consistent with morphological profiles generally seen in regenerating axons. This regenerating profile was confirmed by the GAP-43 immunoreactivity (Fig. 2F
). On the opposite, CST fibers from control animals were much less abundant within the lesion (Fig. 2C
). Quantification of those regenerating profiles within the lesion area shows a significant increased CST regeneration in the GM-CSF-treated group (P<0.05, Mann and Whitney U test; Fig. 2E
).
|
This axonal regeneration in the GM-CSF treated group was confirmed in more severely injured rats (40 µl). Indeed, 24 wk postinjury, numerous longitudinally orientated GAP-43 immunoreactive fibers were detected within the lesion site of GM-CSF-treated rats (Fig. 3, A and B
). The quantitative analysis of those fibers revealed that in GM-CSF treated animals, lesions were densely penetrated by GAP-43-labeled axons, whereas control animals showed relatively few axons (Fig. 3
C; P<0.05). This significant axonal regeneration was confirmed by the NF immunostaining (Fig. 4
A, B). The number of NF-labeled axons in the lesion site of the GM-CSF group was significantly higher compared with the control one (Fig. 4C
, P<0.05).
|
|
We also focused on descending axons from raphe neurons, which project serotonergic (5-HT positive) fibers. In the GM-CSF treated animals, there were numerous 5-HT-positive processes in the lesion site compared with very few in the vehicle-treated animals (Fig. 5
A, B). The quantification through the entire lesion site of serotonergic immunoreactive profiles in both experimental groups is consistent with this qualitative assessment (Fig. 5D
, P<0.05). The GFAP immunostaining of adjacent slices demonstrates that regenerating profiles are localized within the lesion site (Fig. 5C
).
|
We then focused on the effect of GM-CSF treatment on Cr3 (OX-42) and BDNF expression by round shaped macrophages invading the lesion site. As illustrated on Fig. 6
, 5 days post-treatment, the manual counting of macrophages in the lesion site shows a significant increase of OX-42-positive macrophages in GM-CSF-treated animals, compared with control ones at the same survival delay. This increase was transient, as both experimental groups returned to lower expression levels 7 days after the treatment (Fig. 6A
, P<0.05). BDNF expression by macrophages, also assessed at various time points after GM-CSF treatment, was found to be significantly increased 7 days after treatment in the GM-CSF treated group. This was confirmed by the morphometric image analysis (Figs. 6, B-D
, P<0.05). Double immunofluorescence staining for Cr3 and BDNF demonstrates that Cr3-positive macrophages do also express BDNF (Fig. 6E, G
).
|
GM-CSF enhances neurite growth via an increased BDNF expression by microglial cells
We next examined BDNF protein expression in lysates of BV2 cells cultured for 24 h in media containing or not recombinant GM-CSF. As shown in Fig. 7
, the expression of BDNF by BV2 is higher when cells were activated with GM-CSF. Pixel quantification of the different lanes compared with actin expression revealed an increase in BDNF expression by GM-CSF-activated microglial BV2 cells (Fig. 7)
.
|
Adult DRG neurons or P7 cerebellar neurons cocultured for 16 h with GM-CSF preactivated BV2 microglial cells were significantly more likely to bear TUJ-1 immunostained neurites than those cocultured with nonactivated BV2 cells (P<0.01). This increase in neuritogenesis was abolished when an anti-BDNF Ab (1 µg/ml) was added to the cocultures (Fig. 8
).
|
DISCUSSION
This study demonstrates for the first time that a single intraperitoneal injection of 2 µg GM-CSF, a macrophage-activating cytokine, 4 wk after an SCI producing irreversible paraplegia in rats is able to promote locomotor recovery and axonal regrowth. It also shows that GM-CSF activated macrophages express more BDNF in vivo and in vitro and that they increase the neuritic index of cocultured neurons. Taken together, these results suggest that the favorable action of GM-CSF observed in experimental paraplegia could at least partially be mediated by the neurotrophic factor BDNF. We will discuss the behavioral results, the experimental protocol, the immunohistological data, the link with BDNF, as well as the therapeutic perspectives.
Experimental protocol and behavioral results
The behavioral improvement we found after a single delayed intraperitoneal injection of 2 µg GM-CSF is in accordance with another recent study showing a favorable effect of GM-CSF treatment on locomotor recovery (19)
. Our results are even more interesting because they were obtained with rats suffering from irreversible paraplegia and with a single dose of GM-CSF. Thus, the locomotor improvement induced seems to be limited in time. It occurs rapidly within 1 wk after treatment, accrues during the following 7 wk and persists up to 20 wk without further increase. This may suggest that repetitive administrations and/or higher doses of GM-CSF could have a more pronounced beneficial effect. We chose the 2 µg of GM-CSF (8–10 µg/kg) dose because it is close to the subcutaneous doses (5 µg/kg) used in humans for agranulocytosis (27
28)
and to intraperitoneal doses used to study hormonal GM-CSF effects in rodents (29)
. After daily administrations of such doses, adverse effects such as fever, myalgia, malaise, or rash occur in 20–30% of patients but remain moderate (30)
, whereas higher doses are often associated with severe adverse effects (30)
. As shown in other studies (29)
, the intraperitoneal administration is convenient in animals and eventually applicable in the clinic. Since GM-CSF crosses the blood-brain barrier (31)
, it avoids the need for intralesional injections, which add a supplementary injury.
In a recent study conducted simultaneously to ours (19)
, a daily intraperitoneal dose of 20 µg GM-CSF was administered immediately after a clip compression of the rat spinal cord during 5 days. The SCI model used in this study produces a largely reversible paraplegia, as control animals reach a mean BBB score of 15.9 5 wk after injury, compared with 0.86 in our closed-site microballoon compression model. The behavioral results obtained in the study of Ha et al. and in ours are thus difficult to compare, but they show nonetheless some striking similarities. In both studies, the BBB score improves significantly in GM-CSF treated rats as soon as 1 wk after treatment. As in our study, the beneficial effect wanes over time in study of Ha et al. (19)
animals, the BBB score difference in favor of GM-CSF being 3.6 at 1 wk, 4 at 3 wk, and 2 at 5 wk postinjury. The smaller BBB difference we found between GM-CSF and saline-treated rats (±2) is not likely to be due to the lower dosage, but rather to the more pronounced spinal lesion. With rare exceptions (3
, 32)
, higher locomotor scores are generally reported with most other SCI models (e.g., hemisection, weight-drop contusion, photochemical injury, and clip compression), which produce less severe lesions and allow a spontaneous motor recovery up to a BBB score of 8–10, indicating that the animals are able to walk without coordination. Even with such milder lesions, however, the treatment effect usually does not exceed 2 points on the BBB scale (33
34
35
36)
. More refined functional tests like grid-walking or foot print recordings can be used in less severely paralyzed animals. These tests were not applicable in our study because they require animals able to support their weight on the plantar-placed hind-limbs with some degree of coordination, which corresponds to scores as high as 10 on the 21 point nonlinear BBB scale. Despite these shortcomings, we selected for our study the closed-site compression-lesion that we have developed in our laboratory (24)
, because it is relevant to the human situation in which most spinal cord injuries are closed, complete or subcomplete, irreversible, and associated with massive destruction of the spinal tissue (37)
. It is known in rats that the degree of locomotor recovery is correlated with the percentage of remaining fibers in the ventrolateral spinal cord. Even a "complete" transection can result in sparing of nerve fibers in the ventrolateral white matter of which <5% are thought to be sufficient for an unequivocal motor recovery after SCI (38)
. Therefore, the lesion model as well as the spinal level of the injury (39)
have to be taken into account when the results on locomotor recovery are compared between different studies.
As mentioned in the Introduction, we chose to apply the GM-CSF treatment 4 wk postinjury because in our previous study (15)
macrophages at the lesion site tended to loose their OX-42 immunoreactivity and spontaneously regenerating axons disappeared at this delay. Delayed treatment would also be favored by studies showing deleterious effects of early macrophages, and the inflammatory processes they trigger, on CNS repair (13
, 40
41)
. It has been shown that decreasing macrophage activation can improve tissue repair and functional recovery after SCI (42
,13)
and that anti-inflammatory treatment has neuroprotective properties only if applied within the first few hours after the insult (43
44
45)
. By contrast, we have reported a beneficial effect of activated macrophage transplants immediately after lesioning on axonal regeneration after a spinal compression injury (2)
and Ha et al. (19)
report locomotor improvement after GM-CSF administrations immediately after lesioning. These contradictory data are attributable to differences in the injury model used, postinjury delays, and treatment modality (46
, 47)
. It seems clear that activated macrophages release different cytokines and trophic factors at different postinjury stages (11)
. Thus, macrophages can have cytotoxic or protective activity depending on the time and the type of their activation.
Immunohistology and the link between GM-CSF and BDNF
The locomotor improvement occurring as early as 1 wk after GM-CSF administration can hardly be explained by axonal regeneration through the entire lesion site. At the same delay, we show that the expression of BDNF in macrophages is markedly increased within the lesion, a phenomenon occurring just after the higher expression of Cr3 by the same macrophages. BDNF is known to stimulate both locomotor activity after moderate or severe spinal cord lesion in rats by activating the local central pattern generator (23
, 48)
and axonal growth notably via intraneuronal increase of cAMP (49)
. It also promotes the regenerative sprouting of injured serotonergic axons in the adult rat brain (50)
. The increased axonal regeneration we find in GM-CSF treated animals with GAP-43, NF, and 5-HT immunocytochemistry 24 wk after injury could thus be a lasting consequence of the macrophage activation and BDNF release induced by GM-CSF 4 wk postinjury. Indeed, it has been shown that after injury axonal sprouts preferentially enter cavities containing large number of macrophages (51)
and that macrophages and microglia produce local trophic gradients, i.e., of BDNF, which stimulate axonal sprouting toward but not beyond the wound edge (7
,8)
. This is compatible with our corticospinal tract tracing study showing numerous tortuous profiles and a significantly higher number of fibers entering the lesioned area in the GM-CSF treated group compared with the control group, 4 wk after the treatment. Those results, coupled to the GAP-43 immunoreactive fibers detected within the lesion site of the same animals, showing similar tortuous fibers, are mostly consistent with regenerating profiles (52)
.
Our in vitro data showing that BDNF expression by microglial cells is enhanced by GM-CSF are in accordance with other published results (53)
showing that activated microglia produces BDNF. Although it cannot be transposed without reservation to the in vivo situation, the neuritic index increase of adult DRG neurons and P7 cerebellar neurons cocultured with GM-CSF activated microglial cells provides a cellular model that may be relevant to the effect of GM-CSF on SCI and its mediation by BDNF.
Modulation of macrophages as a therapeutic perspective
Previous studies (2
, 3
, 54)
showing beneficial effects of transplants of activated macrophages after spinal cord trauma have led to clinical trials (phase I and phase II; www.proneuron.com). The transplants consist of autologous activated macrophages, which are not easily accessible, and have to be performed within 2 wk after a complete SCI. Our results suggest that a single intraperitoneal injection of GM-CSF activates macrophages to produce BDNF and improves for several weeks functional recovery after a most severe spinal cord lesion. It may therefore represent an interesting atraumatic and repeatable pharmacological alternative to modulate macrophage activity in SCI. Another advantage would be that the treatment can be delayed in time and thus applied in larger patient groups, possibly also in chronic paraplegics. Also, its combination with other treatments could be of beneficial interest, as recently studied in acute complete spinal cord injured patients (55)
.
ACKNOWLEDGMENTS
The authors are grateful to Jeanine Mosen, Murielle Wouters, and Patricia Ernst for expert technical assistance. This work was supported by grants of the National Fund for Scientific Research (FNRS-Belgium), the Léon Frédéricq Foundation (Faculty of Medicine-University of Liège), and the "Günter Verbraeckel Foundation for scientific research on spinal cord injury" (private foundation). F. Scholtes, R. Franzen, and B. Malgrange are research fellow, scientific research worker, and research associate, respectively, of the National Fund for Scientific Research (FNRS-Belgium).
Received for publication May 24, 2005. Accepted for publication February 3, 2006.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
