[1] 15w Co-treatment with riluzole and GDNF is necessary for functional recovery after ventral root avulsion injury

[1] 194w Unilateral avulsion of lumbar ventral roots kills approximately 50% of injured motoneurons within 2 weeks of surgery. Immediate treatment involving surgical reimplantation of the ventral root (VRI) or intrathecal glial cell line-derived neurotrophic factor (GDNF) delivery or intraperitoneal injection of riluzole for 2 weeks ameliorates motoneuron death to 80% of control but combining the different treatment paradigms did not further enhance survival except when GDNF was combined with VRI. At 3 months, all combined treatments provided a neuroprotective effect compared to avulsion only, but the neuroprotective effect of surgical reimplantation alone was not maintained unless combined with riluzole and GDNF treatment. Analysis of regenerating motoneurons using retrograde labelling techniques showed that riluzole, but not GDNF, increased the number of dendrites per labelled motoneuron. However, when functional motor recovery was assessed using the BBB locomotor score and rotarod tests, only VRI animals treated with riluzole and GDNF application showed significantly improved locomotor function in both tests. Our results show that functional recovery appears related to a combination of enhanced dendrite formation, increased motoneuron survival and the neurotrophic actions of GDNF. Thus, combination treatment may offer a new therapeutic strategy for treating patients with avulsion injury.

[1] 181w Road traffic accidents are the most common cause of spinal nerve root avulsion injury in adult humans and result in monoplegia and chronic pain (Berman et al., 1998). The prognosis for recovery is poor and conventional clinical treatment is non-curative but palliative involving nerve transfers to compensate for loss of sensorimotor function. Recently, we established an innovative surgical technique in humans that improved motor deficits after avulsion injury by reimplanting avulsed ventral roots (VRI) into the spinal cord (Carlstedt et al., 1995). This method is based on observations that avulsed motoneurons extend regenerating and new axons into reimplanted roots and nerve grafts (reviewed in Carlstedt and Cullheim, 2000). One third of the patients receiving this treatment showed improved motor function postoperatively (Carlstedt et al., 2000). However, the shortcomings of this technique are a poor prognosis proportionate to the delay of surgery, incomplete and unpredictable sensorimotor recovery. One important reason for this is that the lesion rapidly kills many of the avulsed motoneurons (Koliatsos et al., 1994). Therefore, preventing cell death is of paramount importance if we are to improve motor function.

[2] 96w Neurotrophic factors are important for the development, maintenance and regenerative potential of motoneurons after injury (Airaksinen and Saarma, 2002;Henderson et al., 1994). In particular, exogenous administration of either brain-derived neurotrophic factor (BDNF) or glial cell linederived neurotrophic factor (GDNF) can rescue adult motoneurons after avulsion injury (Chai et al., 1999;Kishino et al., 1997;Li et al., 1995;Novikov et al., 1997). These studies have mostly examined the short-term effects of single neurotrophin treatments on motoneuron survival; the longer-term effects of single treatments, or of combinations of treatments, on cell survival or functional motor recovery have not been assessed.

[3] 94w In addition to trophic deprivation, motoneurons die via glutamate-mediated excitotoxicity (Choi, 1992). RIL, 2amino-6-trifluoromethoxy-benzothiazole (Riluzole) inhibits glutamate release by blocking voltage-gated Na + and Ca 2+ channels and NMDA receptor (Doble, 1996) and is the only clinically proven drug that prolongs the life of patients with amyotrophic lateral sclerosis (ALS), a rapidly progressive and fatal disease in which both upper and lower motoneurons degenerate (Riviere et al., 1998). RIL is also neuroprotective in models of Parkinson's disease, spinal cord injury and ventral root avulsion (Obinuet et al., 2002;Nogradi and Vrbova, 2001;Wahl and Stutzmann, 1999).

[4] 41w Therefore, three different strategies exist to promote motoneuron survival and encourage axon regeneration. This study assesses the effects of ventral root reimplantation, GDNF and RIL administration or combinations of these treatments, on long-term motoneuron survival, dendrite number and motor functional recovery.

[1] 183w L4-6 root avulsion produced a significant loss ( P < 0.001) of motoneurons in the L5 ventral horn with 52 F 5% (n = 7) motoneuron survival, compared to the contralateral side 2 weeks after injury. Treatment by VRI, or GDNF or RIL significantly improved motoneuron survival ( P < 0.01) in the L5 segment to, respectively, 81 F 4% (n = 9), 79 F 7% (n = 5) and 81 F 9% (n = 5) suggesting a neuroprotective effect of each of the three treatments in the short term. As a direct result of this finding, we combined the single treatments in an attempt to increase the survival effect. Post hoc analysis showed that motoneuron survival was only significantly improved further using a combination treatment of VRI with GDNF (107 F 6%, n = 7, P < 0.05), as compared to single treatments. Motoneuron survival in VRI rats treated with RIL was 72 F 2% (n = 4), and VRI rats treated with GDNF and RIL was 96 F 3% (n = 5), which was not significantly different from single therapy treatments.

[2] 114w To assess the longer-term effects of these manipulations on functional recovery, the survival time was extended to 3 months post-lesion and treatments consisted only of VRI or VRI with the different drug combinations. In VRA animals, motoneuron survival had declined to 42 F 3% (n = 5). The motoneuron rescue effect produced by VRI alone was not maintained at 3 months (46 F 8%, n = 10) except when VRI was combined with RIL and GDNF co-treatments (80 F 3%, n = 8), where significantly enhanced motoneuron survival was achieved ( P < 0.05). Post hoc statistical analysis revealed that all drug group combinations produced significant rescue effects compared to VRA alone (Fig. 1).

[3] 88w To examine the functional consequences of the different manipulations that promoted motoneuron rescue at 3 months post-injury, the accelerating rotarod test was used to assess each animal's coordinated walking ability (Fig. 2A). The latency with VRA injury (45 F 2 s) was significantly lower as compared to controls (103 F 9 s, P < 0.05). Motor deficits in the VRA group were only significantly corrected in VRI rats treated with GDNF and RIL (91 F 9 s) or VRI rats treated with GDNF alone (79 F 4 s).

[4] 114w The quality of locomotor movement was assessed using the BBB locomotor rating scale. In this performance test, VRA rats showed severe locomotor deficits with an average score of 3.5 F 2 on a scale of 0 -21, indicating that these rats have some movements for two joints (usually the hip and knee) and occasional limited movement of the ankle. Only VRI rats treated with GDNF and RIL showed a significantly increased BBB score ( P < 0.05) averaging 17.8 F 1.4 (Fig. 2B) compared to all other groups except control ( P = 0.21). This score corresponds to consistent plantar stepping and good hind-limb/fore-limb coordination during gait and toe clearance during forward limb movement.

[5] 22w Thus, in two behavioural tests that examine different aspect of locomotion, only VRI rats treated with GDNF and RIL showed significant improvements.

[6] 189w To try to account for the observed functional recovery in the group of VRI rats treated with GDNF and RIL but not in other groups, we used retrograde labelling of regenerating motoneurons with CTB to label motoneurons that had regenerated axons into the sciatic nerve (Fig. 3). We presumed that the group of VRI rats treated with GDNF and RIL would have more retrogradely labelled motoneur-ons per section compared to the other groups. In control animals, an average of 44 F 5 motoneurons per section were labelled in the L5 spinal cord segment with CTB. In VRA animals, the average was 1 F 1, for VRI 10 F 3, VRI with GDNF 9 F 2, VRI with RIL 10 F 3, VRI with GDNF and RIL 10 F 4. Thus, all treatments resulted in approximately a 10-fold increase in numbers of labelled motoneurons stained although this only represented approximately 25% of retrogradely labelled motoneurons in control. This increase represented approximately a third to a half of the surviving motoneurons at this time in these groups as compared to VRA rats where less than 5% of surviving motoneurons were labelled.

[7] 194w Quantitative analysis of labelled motoneurons soma area showed that there was a 25% neuronal atrophy after avulsion compared to control animals (830 F 32 Am 2 vs. 1091 F 51 Am 2 , P < 0.05). Interestingly, VRI rats or VRI rats treated with GDNF or RIL or both reversed the atrophy to less than 13% of control (985 F 24 Am 2 , 1082 F 42 Am 2 , 1083 F 53 Am 2 and 950 F 30 Am 2 , respectively). When dendrite number was analysed (Fig. 4A), it showed that avulsion injury induced a significant loss of dendrites that could be reversed by VRI ( P < 0.05). Surprisingly, only in VRI-treated rats that had received RIL was there a significant increase in dendrite number (4.9 F 0.5) and this was further augmented by co-administration with GDNF (6.4 F 0.3). Not only were there more dendrites, but more of the dendritic tree was labelled (Figs. 4B and 5); dendrites could be seen traversing long distances in the ventral horn. In favourable sections, labelled axons could be seen emanating from the dendrite and could be followed into the implanted root (Fig. 5).

[8] 84w Linear regression analysis showed that there were significant correlations between motoneuron survival and locomotor behaviour tests: For the BBB, r 2 = 0.709 ( F = 65.8, P < 0.0001) and for the rotarod score, r 2 = 0.334 ( F = 12.5, P = 0.001). There was also a significant correlation between dendrite number and rotarod time: r 2 = 0.266 ( F = 7.6, P = 0.012), but not BBB score r 2 = 0.153 ( F = 3.78, P = 0.065).

[1] 360w Our results confirm that a rapid and extensive motoneuron loss occurs in the ventral horn of the avulsed region (Chai et al., 2000;Kishino et al., 1997;Koliatsos et al., 1994;Li et al., 1995;Novikov et al., 1995Novikov et al., , 1997)). Additionally, we show that any one of three different treatment strategies, surgical root reimplantation, intrathecal GDNF or intraperitoneal RIL injection are equally effective in preserving motoneuron numbers after insult during their period of administration. This suggests that several mechanisms contribute to motoneuron cell death including loss of peripheral target, loss of neurotrophic support and vascular trauma leading to excitotoxicity. RIL is known to be neuroprotective in models of traumatic brain and spinal cord injury, Parkinson's disease and ALS (Nogradi and Vrbova, 2001;Obinu et al., 2002;Schwartz and Fehlings, 2001;Wahl and Stutzmann, 1999), although the exact mechanism by which it exerts its effects is unclear. RIL has a role in reducing the pre-synaptic glutamate release by blocking NMDA-receptor and voltage-gated sodium and calcium channels, thus reducing the level of excitation. RIL also stimulates the production of neurotrophic factors such as GDNF and BDNF in astrocytes that may provide additional benefit (Katoh-Semba et al., 2002;Mizuta et al., 2001;Peluffo et al., 1997). Similarly, GDNF protects cultured motoneurons from glutamate injury (Tang et al., 2001) and the short-term neuroprotective effects of GDNF on motoneurons after avulsion injury are well established (Li et al., 1995;Natsume et al., 2002;Watabe et al., 2000). However, we were unable to preserve this rescue effect at 3 months, using only a single therapy approach as survival mediated by the VRI alone was not significantly different from VRA, suggesting that the neuroprotective effect of the implanted root is only viable for a limited period, consistent with the findings of Chai et al. (2000). However, combination treatments did significantly improve motoneuron survival when VRI was combined with GDNF or RIL or both treatments, suggesting that short-term treatments with these drugs can have long-term neuroprotective effects. This is the first report of such an effect for GDNF. A similar result has been reported for BDNF in which a pellet of gelfoam soaked in BDNF was neuroprotective for up to 6 weeks (Chai et al., 1999).

[2] 296w Previous studies have documented limited functional restoration in proximal limb muscles in animal models (Bertelli and Mira, 1994;Carlstedt et al., 1993;Cullheim et al., 1989;Hallin et al., 1999;Smith and Kodama, 1991) as well as in man (Carlstedt et al., 1995(Carlstedt et al., , 2000) ) using surgical grafting and root reimplantation techniques. In most of these studies, functional restoration took 8 -12 months to become evident. Reinnervation of limb muscles occurred in an arbitrary way from non-specific motoneuron pools (Carlstedt and Cullheim, 2000;Carlstedt et al., 1993), which often caused co-activity in antagonistic muscles (Carlstedt et al., 2000;Hallin et al., 1999). Qualitative functional recovery, as assessed by reflex testing and general movement observation, was reported by Nogradi and Vrbova (2001) following RIL plus reimplantation of the L4 root after single root avulsion, in contrast to our results. It is more likely that this recovery was mediated by compensatory adjustments in reflex pathways (Holmberg and Kellerth, 1996). There are no reports of functional recovery in avulsion injury using only GDNF, although its effects on cell survival are well documented (Airaksinen and Saarma, 2002;Henderson et al., 1994;Li et al., 1995). Using two different locomotor tests, we found that significant functional recovery in both tests was only observed at 3 months in VRI rats that received GDNF and RIL co-treatment. The BBB test and rotarod test assess different aspects of locomotion. The BBB score is based on operational definitions translated into simple and specific observational criteria that can follow recovery progression from complete paralysis to normal locomotion. It is a sensitive and reliable scale designed to assess the hind limb function recovery (Basso et al., 1996). The rotarod test, on the other hand, evaluates the coordinated ability to stay on a rotating drum and the possible changes in weight support management.

[3] 192w A plausible explanation for the observed recovery was that the different treatment combinations induced a variable number of regenerating axons into the implanted root, with the VRI rats treated with GDNF and RIL producing the most robust regeneration. This hypothesis was tested by retrograde tracing and proved to be null as similar numbers of retrogradely labelled motoneurons were observed in all treatment groups. Neither behavioural parameter by regression analysis was found to correlate with the total number of labelled motoneurons. The occasional motoneurons labelled in the L5 segment after avulsion only may represent the intramedullary extension of collateral or ''super-numery'' axons (Havton and Kellerth, 1987) that reach the sciatic nerve via re-growth in intact adjacent ventral roots or along the pial surface (Hallin et al., 1999). However, regression analysis showed that behavioural recovery in both tests was strongly correlated with motoneuron survival and this is reflected in the fact that animals treated with GDNF had higher behavioural scores than those that did not receive GDNF. Thus, one major conclusion that can be drawn from this study is that rescuing substantial numbers of motoneurons from avulsion-induced cell death leads to functional locomotor recovery.

[4] 138w Analysis of labelled motoneurons showed that avulsion caused atrophy in the remaining neurons and that this could be reversed by all the different treatment strategies. Neurotrophins are known to reverse such changes and can cause hypertrophy in surviving motoneurons (Li et al., 1995), and peripheral nerve injury increases the production of neurotrophic factors and their receptors as described following axotomy (Hammarberg et al., 2000). The increased production of neurotrophic factors from the implanted root is likely to be available to the injured motoneurons but only for a limited period. Combining VRI with exogenous GDNF or RIL or both did not yield any further increase in motoneuron soma size compared to VRI alone suggesting that most of the neurotrophic effect may be initially derived from the implanted root and that providing the neuron survives, soma size is near normal.

[5] 220w The most surprising effect of these treatments was that RIL promoted dendrite outgrowth whereas other treatments did not. Naive animals treated with riluzole did not show enhanced dendrite outgrowth. Such an effect has not been previously reported for RIL and is based on retrograde labelling methods. CTB is commonly used as a retrograde tracer for labelling motoneurons both before and after injury and labels the cell body and the proximal dendrites of motoneurons. The increase in dendrite number is unlikely to be artifactual as the intensity of labelling appeared similar across the different treatment groups. Unequal transport of CTB is unlikely because similar numbers of motoneurons were labelled in the different groups. Although the mechanisms of dendrite growth are not well understood, growth factors are known to stimulate growth (Danzer et al., 2002) and our results suggest that RIL is also capable of mediating dendrite growth. The underlying mechanism of this phenomenon deserves further study. Regression analysis revealed that rotarod performance was correlated with motoneuron dendrite number and this appears dependent on treatment with RIL rather than GDNF. Interestingly, it has recently been demonstrated that both RIL and GDNF augment neurotransmitter content without affecting neuronal numbers that consequently improves the motor functions and neural deficits in models of Parkinson's disease (Douhou et al., 2002;Grondin et al., 2002;Obinu et al., 2002).

[6] 166w Our single treatment therapies proved ineffective at rescuing all motoneurons from cell death or restoring functional recovery. Combination therapies proved to be far more effective and offer a potential new treatment strategy for avulsion injury patients. GDNF has been coadministered with insulin like growth factor-1 (IGF-1) and BDNF and shown to be neuroprotective in postnatal models of motoneuron cell death and obstetric plexus lesions (Aszmann et al., 2002;Bilak and Kuncl, 2001;Bilak et al., 2001) and synergistic effects of combined neurotrophic factors have been shown on cultured motoneurons (Zurn et al., 1996). Recently, RIL has been used as part of a threedrug cocktail to ameliorate the neurodegenerative effects seen in a mouse model of ALS (Kriz et al., 2003). Our results show that functional recovery is correlated to enhanced dendrite complexity and increased survival of motoneurons. RIL promotes dendrite outgrowth and neurotrophins such as GDNF promote motoneuron survival and axon regeneration and both are required for functional recovery possibly by increasing neurotransmitter level or synthesis in motoneurons.

[1] 15w Animal procedures were carried out in accordance with the UK Animals (Scientific Procedures) Act 1986.

[2] 104w Female Sprague -Dawley rats (180 -250 g) were anaesthetized with halothane and the L4-6 ventral roots were exposed via a dorsal lumbar hemi-laminectomy. The dura was opened and the left L4-6 ventral roots were identified according to their relation to the lumbar vertebrae L1 -L3. Their attachment with the ventral spinal cord was visualised by gently rotating the cord by holding the denticulate ligament. The ventral roots were torn from the spinal cord by traction with fine forceps, which usually produced muscle twitches in the leg. The animals were then subject to one of the following treatments (see Table 1): 1. Avulsion only (VRA).

[3] 49w Three months after avulsion, coordinated walking ability was assessed using the accelerating rotarod (Ugo Basile, Italy) where rats are required to walk for 5 min against the accelerating motion of a rotating drum. Rats were familiarized with the test, and 2 h later, the latency, time (s) taken for

[4] 91w Table 1 Values for n within specific treatment groups at 2 weeks and 3 months time points 2 weeks 3 months Sham 8 8 VRA 7 5 VRA+GDNF 5 -VRA+RIL 5 -VRI 9 10 VRI+GDNF 7 4 VRI+RIL 4 4 VRI+GDNF+RIL 5 8 Fig. 1. Quantitative analysis of motoneuron cell survival in different treatment groups 3 months after avulsion injury. ANOVA followed by post hoc analysis showed the following interactions: * means significantly different from control, z means significantly different from VRA, § means significantly different from VRI+GDNF+RIL (P < 0.05).

[5] 50w animals to fall off the rotarod, was recorded from five trials per rat. Additionally, for each group, open-field locomotor behaviour was evaluated over a 4-min period using the BBB locomotor rating scale (Basso et al., 1995). All groups were tested by two observers, who were blind as to treatment regime.

[6] 40w Following behavioural assessment, and before sacrifice at 3 months post-injury, rats were re-anaesthetized with halothane and the left sciatic nerve was exposed and injected with 4 Al of 1% choleragenoid tracer (CTB, Sigma) as described by Shortland et al. (1997).

[7] 59w Animals were perfused with Zamboni fixative solution and the L3-S1 spinal cords were removed, post-fixed and cryoprotected in 30% sucrose. Free-floating 50-Am transverse sections (1 in 12) were incubated with goat anti-CTB antibody (List Biological, 1:2000) for 24 h and visualised using donkey anti-goat Cy3 (Stratech Scientific Ltd, UK; 1:600). For motoneuron counts, sections were counterstained using Neutral Red.

[8] 59w CTB-positive and motoneuron numbers were counted directly from slides. Images were captured with a CCD Hamamatsu chilled camera connected to a Leica microscope and soma area was analysed using Scion image analysis software (http://www.scioncorp.com) in a minimum of five sections from the L5 spinal cord per animal per group. Dendrite numbers were counted directly from slides at Â40 magnification.

[9] 139w All data are expressed as means F SEM and were compared using one-way ANOVA. Where significance was detected, the ANOVA was followed by a post hoc (multiple comparisons) test to identify which groups were different. The Games -Howell test was used as the different groups were of unequal size and unequal variance. Statistical signif-Fig. 2. Analysis of locomotor ability using (A) rotarod test (B) BBB locomotor rating scale. Post hoc analysis showed groups that were significantly different from control (*) or from VRI+GDNF (E) or from VRI+GDNF+RIL ( §). Note that VRI+GDNF+RIL was significantly different from the VRA group in both tests, while it was not significantly different from the control in either test. icance was set at P < 0.05. In addition, regression analysis was used to examine the relationship between motoneuron survival, dendrite number and locomotor behaviour.

[1] 99w into the L5 spinal cord as described previously (Carlstedt et al., 1993). 3. Intrathecal administration of GDNF after treatment 1 or 2 using an Alzet 2002 mini-osmotic pump (Alza Corporation, Palo Alto, CA) filled with GDNF (isr-metHuGDNF, USAN name liatermin, non-truncated molecule) 0.85 mg/ml, gift from Amgen Inc. Thousand Oaks, CA, USA). The infusion rate was 12 Ag/day for 2 weeks. 4. Daily intraperitoneal administration of 0.1 ml of RIL (Tocris Cookson Ltd, UK; 4 mg/kg) for 2 weeks after treatments 1 or 2. 5. Combination treatments of animals in group 2 with GDNF, or RIL or both compounds.

[2] 53w These groups survived for either 2 weeks or 3 months. For motoneuron survival analysis, the experimental side was compared to the contralateral side so as to reduce inter-animal variation, and for anatomical analysis of regenerating neurons and behavioural analysis, treated groups were compared to a control group (n = 8) of naive animals.