Survival, regeneration and functional recovery of motoneurons after delayed reimplantation of avulsed spinal root in adult rat

Experimental Neurology 192 (2005) 89 – 99 www.elsevier.com/locate/yexnr

Survival, regeneration and functional recovery of motoneurons after delayed reimplantation of avulsed spinal root in adult rat Huai-Yu Gua, Hong Chaia, Jian-Yi Zhanga, Zhi-Bin Yaob, Li-Hua Zhoub, Wai-Man Wonga, Iain C. Brucec, Wu-Tian Wua,* a

Department of Anatomy, Faculty of Medicine, The University of Hong Kong, 21 Sassoon Road, Hong Kong SAR, China b Department of Anatomy, Sun Yat-Sen University of Medical Sciences, Guangzhou, China c Department of Physiology, Faculty of Medicine, The University of Hong Kong, Pokfulam, Hong Kong SAR, China Received 29 June 2004; revised 11 October 2004; accepted 20 October 2004 Available online 19 January 2005

Abstract We have established that extensive reinnervation and functional recovery follow immediate reimplantation of avulsed ventral roots in adult rats. In the present study, we examined the consequences of reimplantation delayed for 2 weeks after avulsion of the C6 spinal root. Twelve and 20 weeks after delayed reimplantation, 57% and 53% of the motoneurons in the injured spinal segment survived. More than 80% of surviving motoneurons regenerated axons into the reimplanted spinal root. Cholinesterase–silver staining revealed axon terminals on endplates in the denervated muscles. The biceps muscles in reimplanted animals had atrophied less than those in animals with avulsion only, as indicated by muscle wet weight and histological appearance. After electrical stimulation of the motor cortex or the C6 spinal root, typical EMG signals were recorded in biceps of reimplanted animals. The latency of the muscle potential at 20 weeks was similar to that of shamoperated controls. Behavioral recovery was demonstrated by a grooming test and ipsilateral forepaw movements were well coordinated in both voluntary and automatic activities. These results demonstrate that ventral root reimplantation can protect severed motoneurons, enable the severed motoneurons to regenerate axons, and enhance the recovery of forelimb function even when it is delayed for 2 weeks after avulsion. D 2004 Elsevier Inc. All rights reserved. Keywords: Spinal motoneurons; Root avulsion; Ventral root reimplantation; Regeneration; Reinnervation; Functional recovery; Adult rat

Introduction A common and difficult problem in severe brachial plexus injury involves avulsion from the spinal cord; that is, the roots are torn from the spinal cord at the transition between the central and peripheral nervous systems. This type of nerve injury is generally not treated because it is considered a type of central nervous system injury and thus not amenable to surgery (Narakas, 1984). Clinically, spinal root avulsion occurs when crushing and traction forces are applied to the soft tissues. Under these conditions, it is

* Corresponding author. Fax: +852 2817 0857. E-mail address: [email protected] (W.-T. Wu). 0014-4886/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.expneurol.2004.10.019

difficult to determine the exact location of the nerve injury, so delayed nerve repair has been advocated. Typically, secondary nerve repair within 2–3 weeks is advised, or longer when a neuroma has formed and demarcation is obvious (Watchmaker et al., 1996). A short time lag between the accident and the surgery is recognized as a significant factor for a successful outcome (Carlstedt et al., 2000). Following delayed nerve repair, poor intrinsic and extrinsic muscle recovery and only protective sensation are obtained, regardless of whether the secondary repair is made early (within 2 or 3 weeks) or late (6 months or later) (Chuang et al., 2001). One means of promoting the regeneration of damaged CNS is through the implantation of a peripheral nerve graft in the immediate environment of the injured neurons

90

H.-Y. Gu et al. / Experimental Neurology 192 (2005) 89–99

Thirty-eight adult male Sprague–Dawley rats (200–250 g) were divided into three groups: (1) sham-operated controls (6 rats); (2) C6 ventral root avulsion only (14 rats); and (3) C6 ventral root avulsion with delayed reimplantation (18 rats). In the latter group, 8 animals were sacrificed at 12 weeks and 10 animals were sacrificed 20 weeks after reimplantation. All procedures were approved by the Committee for the Use of Live Animals in Teaching and Research at the University of Hong Kong.

Briefly, 2 weeks after avulsion, the avulsed roots were dissected free and the ventral root was carefully reimplanted into the ventrolateral aspect of spinal segment C6. With a fine glass probe, a small slit was made in the pia mater and care was taken not to injure the spinal white matter. The dorsal root was anchored to the inner surface of dura mater by 10-0 silk to assist attachment of the ventral root to the ventrolateral side of the spinal cord. The dura was closed and the defect covered by a small piece of gelfoam. A piece of muscle was placed over the hemilaminectomy site. The muscles, subcutaneous tissues and skin were closed in separate layers. Next, the brachial plexus was explored through a hockey-stick incision. The pectoralis major and minor muscles were retracted laterally and the components of the brachial plexus were then dissected between the anterior and middle scalene muscles, exposing the spinal nerves; the C5 and C7 nerves were ligated and cut. To avoid direct reinnervation, the proximal and distal extremities of the cut spinal nerves were ligated and about 5 mm of spinal nerve was removed. In addition, the branch between the medial and lateral fasciculi of the brachial plexus was ligated and cut. Muscles, fascia and skin were then sutured successively in layers. The animals were allowed to recover from the anesthetic under supervision and were taken to a recovery room where they were checked regularly for several hours. The rats resumed drinking and eating within 1 day and had an uneventful recovery without visible functional disorder other than flaccid paralysis of the right forelimb. Animals were allowed to survive for 12 or 20 weeks after surgery.

Surgical procedures

Grooming test

All animals were anesthetized with ketamine (80 mg/kg, i.m.) and xylazine (8 mg/kg, i.m.) and all procedures were carried out under sterile conditions.

One week following surgery, all animals were videotaped and the behavior of the right (lesioned side) forelimb was noted, especially when the animal ate or climbed. From the beginning of the 12 postoperative weeks, behavioral analysis was carried out by weekly observation of each animal’s response in the Terzis grooming test, an established method of evaluating forelimb behavior in the rat (Bertelli and Mira, 1993). The test consists of pouring water over the animal’s head to elicit grooming movements of the forepaws toward the head. In normal grooming, the animal raises both forelimbs, licks them, reaches up to the area behind the ears with a smooth motion and finally lowers the limbs down to the snout. The movements proceed repetitively for a few seconds. Pouring water over the snout elicits an abbreviated grooming response to wipe away the wetness; this response was recorded with a digital camera. A short time later, when the animal is more relaxed, a prolonged series of grooming movements are carried out. Because the movements are always bilateral and follow the same pattern, this test provided a reliable method to assess functional recovery of the experimental side relative to the normal side. Terzis grooming test video records were graded by an independent

(Stichel and Muller, 1998) or after a short time lag (Carlstedt and Noren, 1995). However, such implantation has a number of limitations. Apart from the inevitable mechanical damage caused by the physical insertion of the nerves into the spinal cord, most favorable grafts attract only a small proportion of the host fibers (Sellin et al., 1980) and abnormal connections may be formed (Cheng et al., 1996). Recent studies show that a new technique of immediate reimplantation of the avulsed ventral root not only improves the survival of injured motoneurons and enhances axon regeneration, but also does not cause additional injury to the spinal segment in the process (Chai et al., 2000; Gu et al., 2004). In the present study, we investigated the effects of delaying this new method of ventral root reimplantation on the regeneration and functional recovery of motoneurons.

Materials and methods Animals

Sham-operated controls The dura was left intact, and the C5 and C7 spinal nerves were cut and ligated. Avulsion only After removing the right vertebral lamina of C5, the dura was opened and the ventral root and dorsal root with the ganglion of C6 were selectively avulsed from the spinal cord by traction with a fine hook under a surgical microscope. Traction was exerted almost parallel to the natural course of the root. The site was checked visually to confirm complete avulsion. Delayed reimplantation The avulsed ventral roots of C6 were dissected free, fixed to the muscle surface and labeled with a 10-0 suture. Reimplantation of ventral root was made following the procedure described in a previous study (Chai et al., 2000).

H.-Y. Gu et al. / Experimental Neurology 192 (2005) 89–99

observer in a blind fashion based on the criteria illustrated in the following figure.

91

The corresponding electrical responses were recorded from biceps brachii. EMG was amplified (bandpass 0.2– 20 kHz; AC preamplifier, Grass Instruments) and stored on a digital recorder. Sixty successive responses were averaged with respect to each stimulus pulse. During each experiment, the C5 spinal nerve was also stimulated to confirm that the right C6 spinal nerve was the only efferent pathway. Gross anatomical observation After the animals were killed by anesthetic overdose and perfused, the reimplantation site was photographed in situ and any abnormalities were noted. Retrograde axonal tracing

Terzis grooming test functional grading. 0, no response; 1, flexion at elbow, not reaching the snout; 2, flexion reaching the snout; 3, reaching below the eyes; 4, reaching to the eyes; 5, reaching to the ears and beyond (Bertelli and Mira, 1993). Electrophysiological measurements At the end of the 12- or 20-week survival period, all animals were anesthetized with ketamine (80 mg/kg, i.m.) and xylazine (8 mg/kg, i.m.) for electromyographic (EMG) recording. Care was taken to avoid stimulating adjacent normal roots through current spread by keeping the stimulation area filled with paraffin oil and by using partially shielded stimulating electrodes. Also, surrounding tissues were stimulated directly as a control. Biceps brachii was partially exposed and a bipolar electrode was introduced into the center of the muscle belly under a surgical microscope, while another electrode was placed in the subcutaneous tissue to serve as ground. In order to stimulate the motor cortex, the head of the animal was fixed in a stereotaxic frame. A left frontal craniotomy extending from 2 mm posterior to 2 mm anterior to bregma and from 1 to 5 mm lateral to the medial suture exposed the motor cortex that projects to the right C6 spinal motoneurons. The forelimb cortex was identified according to stereotaxic coordinates (Paxinas and Watson, 1986) and a current train of 30 monophasic pulses (1-ms duration, 50–70 AA, at 0.5 Hz; S88 Stimulator; Grass Instruments) was delivered through bipolar, insulated stainless-steel electrodes (tip diameter 200 Am, interelectrode distance 4 mm). For peripheral stimulation, the right C6 spinal nerve was dissected free and immersed in a pool of paraffin oil limited by the surrounding muscles and skin flaps. The nerve was electrically stimulated by a train of monophasic pulses (1-ms duration, 30–50 AA, at 1 Hz) passed through two insulated stainless-steel electrodes (tip diameter 200 Am, interelectrode distance 2 mm).

Three days before the end of each survival period, six animals from each group were anesthetized, and 0.5 Al of 3% Fluoro Gold was injected into the motor branch of biceps brachii via a micropipette with a tip diameter of 25 Am (micropipette puller from Sutter Instrument Co.). The Fluoro Gold solution was slowly injected under the epineurium for about 10 s and the injection site was clamped by a microforceps for 1 s to maximize the number of fibers labeled. Three days later, the rats were deeply anesthetized and perfused with 0.9% saline solution, followed by 4% paraformaldehyde in 0.1 M phosphate buffer. The cervical spinal cord was removed and placed in 30% sucrose for 24 h. Longitudinal or transverse serial frozen sections, 40-Am thick, were cut and mounted on slides, protected by cover slip, and examined under a fluorescence microscope (excitation wavelength: 330– 480 nm, barrier filter: 450 nm). Only labeled neurons with visible nuclei were counted. No corrections for split cell counts were made. The distribution of fluorescence-labeled motoneurons was also recorded by video camera (Neurolucida). After fluorescent microscopy, sections were counterstained with neutral red following the procedures described previously (Wu, 1993). Motoneuron counts and musculocutaneous nerve measurements The number of surviving motoneurons on the lesioned side of the spinal cord was counted in neutral red stained sections by the methods described previously (Clarke and Oppenheim, 1995) and compared with the number on the unoperated side. Only those nucleolated profiles apparently belonging to motoneurons were counted to avoid duplication. Before perfusion, the C6 spinal root and the right musculocutaneous nerve of experimental animals were removed. The specimens were fixed in a paraformaldehyde/glutaraldehyde mixture (1.5%/1% in 0.1 M PB, pH 7.4) for 4 h, and post-fixed in 1% osmium tetroxide for 2 h. The

92

H.-Y. Gu et al. / Experimental Neurology 192 (2005) 89–99

blocks were dehydrated in ascending ethanols, passed through propylene oxide and embedded in Epon. Crosssections (1 Am) were cut and stained with Toluidine Blue for stereological examination through a Nikon microscope with a motorized stage coupled to a color video-camera and a computer running Neurolucida 3.0 (MicroBrightField, Inc.). Computerized histological measurements were made of the number of fibers per unit area and the average diameter of regenerated axons. Histological examination and wet weight of reinnervated biceps brachii To determine the effects of reinnervation, both biceps were removed from all rats and weighed at sacrifice. The muscle tissue was then fixed in 4% paraformaldehyde for 24 h and stored in 30% sucrose for 24 h. Transverse serial frozen sections, 6-Am thick, were cut on a cryostat for routine HE stain. Enzyme histochemical examination of endplates Biceps brachii was frozen to 1608C in isopentane cooled in liquid nitrogen. Longitudinal serial sections, 60-Am thick, were cut in a cryostat, then stained for cholinesterase at 48C for 20 min in a solution containing 1.7 mM acetylthiocholine iodide, 5 mM sodium citrate, 30 mM sodium maleate, 3 mM cupric sulfate and 0.5 mM potassium ferricyanide. After two rinses in water, dehydration in ascending ethanols, and rehydration to water, the sections were fixed at room temperature with 10% formalin. Following treatment with pyridine at 378C for 15 min, the sections were stained in 1% silver nitrate at 378C for 30 min. After rinsing four times in distilled water, they were developed in a solution containing 90 mM hydroquinone and 400 mM sodium sulfite. The stained sections were then dehydrated in ascending ethanols, cleared in xylene and mounted. For fluorescence studies of endplates, sections were washed three times with 0.1 M PBS (pH 7.4), then incubated in Texas Red-X conjugated a-bungarotoxin at 378C for 2 h. The sections were washed four times and examined under a fluorescence microscope. Statistical analysis Morphological results of the retrograde labeling studies and fiber counts from the reimplanted rats were compared with data obtained from the sham-operated controls and the group with avulsion only. The results of the electrophysiological studies and muscle weights were compared with those for the contralateral side. Results are presented as mean F SD. Analysis of variance was used to compare results among groups. The results were accepted as significant if P values b 0.05.

Results Gross anatomical findings At autopsy, all reimplanted ventral roots were found to be connected with the spinal cord, and were embedded in scar tissue; they appeared to penetrate into and fuse with the cord (Figs. 1A,B). Minimal inflammatory or fibrotic reaction was found at the reimplantation site. The musculocutaneous nerve in reimplanted animals was usually of normal dimensions with a white, opaque appearance, but in animals with avulsion only, this nerve was typically thin and translucent and the biceps showed marked atrophy. Survival and retrograde labeling of spinal motoneurons In animals with avulsion only, about 12% of motoneurons survived for 12 weeks and 11% survived for 20 weeks, relative to the normal side (Table 1; Fig. 1C). In contrast, delayed reimplantation of the ventral root significantly enhanced motoneuron survival. About 57% and 53% of the motoneurons survived 12 and 20 weeks following spinal root avulsion, respectively (Table 1). Furthermore, more than 80% of surviving motoneurons regenerated their axons into the reimplanted root as shown by retrograde labeling of Fluoro Gold (Figs. 1D,E). These regenerated motoneurons showed normal morphological features characterized by large cell bodies, centrally localized nuclei and clear Nissl staining. Some regenerated motoneurons showed typical hypertrophy: larger somata and/or more intense Nissl staining (data not shown). Furthermore, a distortion of the ventral horn extending towards the implant was noted in all reimplanted animals. In the sham-operated animals, about twice as many motoneurons were retrogradely labeled by exposing the motor branch of biceps to aqueous Fluoro Gold than in animals with ventral root reimplantation (Table 1). Fluoro Gold-labeled neurons were found mainly in Rexed’s lamina IX, with a dominant concentration in the C6 segment. On the operated side, labeled cells occurred throughout the ventral horn, mainly confined to the neuron pool normally responsible for innervation of biceps (Rexed’s lamina IX). Most labeled neurons were in the size range of a-motoneurons, and the axons and dendrites were labeled as well. The axons of the labeled cells could be traced in the white matter into the reimplanted ventral root. Despite extensive search, very few labeled neurons were found in the C5 and C7 segments, nor did adjacent spinal nerves or roots show any Fluoro Gold labeling. Some fibers and numerous longitudinally arranged cell columns, probably Schwann cells, were labeled where the reimplanted ventral root penetrated the spinal cord. Axonal regeneration and muscle reinnervation Semithin transverse section of musculocutaneous nerve from a normal animal revealed many myelinated axons

H.-Y. Gu et al. / Experimental Neurology 192 (2005) 89–99

93

Table 1 Recovery following reimplantation Sham-operated Percent surviving motoneurons No. of retrolabeled motoneurons MCNb axon density (axons/mm2) MCN axon area (Am2) No. of biceps endplates Biceps weight (mg) EMG latency (cortex) (ms) EMG latency (root) (ms) Terzis score a b

100 89 10,248 76 476 522 1.76 0.51 4.8

F F F F F F F F

9 1926 28 61 43 0.15 0.07 0.3

12-week reimplant 57 51 9928 36 373 408 2.93 0.9 3.5

F F F F F F F F F

20-week reimplant a

5 [12 F 2 ] 5 1261 15 47 36 [154 F 14a] 0.16 0.06 0.5

53 F 7 [11 F 1a] 47 F 4 8417 F 1128 52 F 17 392 F 33 398 F 33 [148 F 11a] 2.25 F 0.23 0.56 F 0.05 4.1 F 0.7 [1.6 F 0.5a]

Avulsion only group. MCN = musculocutaneous nerve.

with different sizes. These axons were well organized and myelinated with a thick myelin sheath (Fig. 1F). In the experimental animals, the reconnected musculocutaneous nerves revealed numerous regenerated axon profiles with varying degrees of myelination at 12 and 20 weeks after

surgery (Fig. 1G,H) whereas no axon was found in the control animals that underwent avulsion only (Fig. 1I). The size of regenerated axons was significantly smaller than that of the normal controls (Fig. 1F). While never reaching control values, the mean cross-sectional area and

Fig. 1. (A, B) The reimplanted ventral root (A, dorsal; B, ventral view) was firmly attached to the spinal cord, and was embedded in scar tissue. (C) Crosssection of the C6 spinal cord from a rat 20 weeks following ventral root avulsion; note the lack of large somata. (D, E) Cross section of the C6 spinal segment from a rat 20 weeks following delayed ventral root reimplantation. Most surviving motoneurons were retrolabeled with Fluoro Gold. The section was first examined under a fluorescent microscope for Fluoro Gold labeled neurons (D), then the same section was counterstained with neutral red (E). Scale bar = 200 Am in C–E. (F–I) Photomicrographs of cross sections of the musculocutaneous nerve in animals undergoing sham operation (F), 12 weeks after delayed ventral root reimplantation (G), 20 weeks after delayed ventral root reimplantation (H), and spinal root avulsion only (I). Scale bar = 5 Am in F–I.

94

H.-Y. Gu et al. / Experimental Neurology 192 (2005) 89–99

axon density of the musculocutaneous nerve and the average motor end-plate count in biceps showed significant recovery 12 and 20 weeks after delayed ventral root reimplantation (Table 1). In animals with avulsion only, no muscle contraction was observed, even after strong stimulation (2 mA) of the dissected C6 root. In these animals, 12 weeks after operation, the weight of the biceps had decreased considerably to only

29% of the weight of normal innervated muscle. Fiber atrophy was pronounced, with widespread cell infiltration and fibrotic tissue proliferation between fascicles, as well as signs of necrosis (Fig. 2C). Similar stimulation produced a muscle twitch in all reimplanted rats, showing that motor axons had reconnected with at least some of the muscle fibers. In these animals, 12 or 20 weeks after the operation, muscle weight was maintained at about 80% of normal

Fig. 2. (A–C) Photomicrographs showing sections of biceps from a sham-operated animal (A), 20 weeks after delayed ventral root reimplantation (B), and 20 weeks after spinal root avulsion only (C). There was no sign of chronic denervation in the reimplanted animals and the muscle fibers regained normal morphology, while marked fiber atrophy, necrosis and inflammatory cell infiltration was evident in muscles from animals with avulsion only. (D–F) The combined silver-cholinesterase method revealed both filaments of nerve terminals and acetylcholinesterase at the endplates. Photomicrographs of endplates in biceps from a sham-operated rat (D), 20 weeks after delayed ventral root reimplantation (E), and 20 weeks after spinal root avulsion only (F). (G–I) Texas RedX-conjugated a-bungarotoxin staining revealed details of the motor endplates in biceps from a sham-operated rat (G), 12 weeks (H) and 20 weeks (I) after delayed ventral root reimplantation. The regenerated endplates showed the typical appearance of continuous and ramified pretzel-like synaptic gutters and subneural lamellae. Scale bars = 50 Am.

H.-Y. Gu et al. / Experimental Neurology 192 (2005) 89–99

95

(Table 1). Signs of chronic denervation, deposits and fiber atrophy were absent; the mosaic pattern was present and the topographic reorganization of muscle fibers was fully consistent with reinnervation of the muscle (Fig. 2B) and similar to what was seen in normal controls (Fig. 2A). By combining silver and cholinesterase-staining methods, we observed motor endplates contacted by ingrowing nerve fibers in control (Fig. 2D) and reimplanted animals (Fig. 2E) but not in animals with avulsion only (Fig. 2F). With Texas Red-X conjugated a-bungarotoxin fluorescence, we consistently observed that the endplates of sham-operated animals were characterized by deep synaptic gutters, plentiful junctional folds, and an AChR distribution that was continuous, bright and tightly confined to the gutters (Fig. 2G). A few endplates in the reimplanted animals had shallow gutters and fewer folds after 12 weeks (Fig. 2H). However, after 20 weeks, most of the regenerated endplates in the reimplanted animals showed topographic characteristics similar to the endplates of sham-operated animals (Fig. 2I).

In contrast, EMG was recorded from biceps brachii in all reimplanted animals. EMG signals recorded from the operated side 12 or 20 weeks after surgery did not show any denervation potentials. Only if root stimulation evoked a muscle potential, while tissue stimulation failed to do so, was the conclusion drawn that the root innervated the muscle. Twelve weeks after reimplantation, the latency was almost double that in sham-operated controls. At the subsequent survival time, 20 weeks, the latency in reimplanted rats had decreased towards, but remained significantly different from, the sham-operated rats (P b 0.05, Table 1; Figs. 3A–C). Stimulation of the forelimb cortex also evoked action potentials in biceps of reimplanted rats. The action potential latency 12 weeks after the surgery was much longer than in the sham-operated group. The latency 20 weeks after reimplantation dropped significantly, again approaching, but still significantly different from shamoperated controls (P b 0.05; Table 1; Figs. 3D–F).

Electromyographic (EMG) recordings from reinnervated muscles No EMG signals were detectable in any of the animals with avulsion alone, nor did they show any signs of functional recovery; EMG exploration disclosed complete denervation of biceps 20 weeks after surgery.

Grooming test All reimplanted animals survived with little apparent neurological deficit other than a partial paralysis of the right forelimb, as assessed by the Terzis grooming test. Onset of function was first discernible at 8–9 weeks (60 F 7 days)

Fig. 3. In sham-operated (A) and reimplanted rats after 12 (B) and 20 weeks (C), stimulation of the right C6 spinal nerve evoked EMG potentials in biceps. Stimulation of the motor cortex also evoked EMG potentials in biceps of sham-operated rats (D), 12 (E) and 20 weeks (F) after undergoing ventral root reimplantation. Arrows indicate stimulus artifacts.

96

H.-Y. Gu et al. / Experimental Neurology 192 (2005) 89–99

after reimplantation. All 18 animals in this group demonstrated a return of function on evaluation at 12 weeks. Functional improvement continued thereafter, three of the 20-week animals almost reaching a normal score. The average Terzis grades for the reimplantation group after 12 and 20 weeks were in the range of 3–4 out of 5, while the group with avulsion alone only achieved 1.6 (Table 1).

Discussion While motor defects that result from peripheral lesions can be reversed by surgical approaches such as direct coaptation of the nerve stumps or with a peripheral nerve graft, muscle reconnection is more difficult to achieve when denervation originates from a central injury or neurodegenerative disease (Glasby and Hems, 1995; Rowland, 1991). Severe spinal nerve root injury involving avulsion or rupture leads to motoneuron loss and muscle denervation (Hoffmann et al., 1990). In this case, the avulsed spinal root can be implanted at the lesion site in an attempt to prevent motoneuron loss and to provide a bridge between CNS and PNS. The direct reimplantation of avulsed roots into the spinal cord has been tested experimentally (Chai et al., 2000; Hoffmann et al., 1996; Moissonnier et al., 1998) and clinically (Carlstedt, 1995). Extending the work of Carlstedt (1993) and our previous studies in rats (Chai et al., 2000; Gu et al., 2004), we investigated the reinnervation of biceps brachii in a simpler model. Avulsion of the C6 roots with ligation and section of the roots at C5 and C7 was followed by reimplantation of the avulsed root 2 weeks after the injury. This rat model allowed us to focus on C6 ventral root involvement in reinnervation with the minimum of surgical damage to the spinal cord. This procedure led to complete paralysis of the ipsilateral biceps. Indeed, animals with avulsion only remained unable to flex the ipsilateral elbow during the whole postoperative period, while reimplanted animals showed significant improvement in this function. We demonstrated that an avulsed ventral root can be directly reconnected to the spinal cord after a 2-week delay, and furthermore, the repair can rescue motoneurons and permit axonal elongation from spinal neurons to the original target muscle. The numbers of myelinated fibers and motoneurons retrogradely labeled from the biceps branch of the musculocutaneous nerve approached that of the normal nerve. Reinnervated muscle recovered responsiveness to electrical stimulation of cortex and root, and animals showed behavioral recovery. In addition, reinnervation of cholinergic endplates and muscle weight recovery were demonstrated. Survival and axon regeneration of spinal motoneurons after ventral root reimplantation As indicated by intracellular staining (Cullheim et al., 1989), reestablishment of anatomical links between the

motoneuron pool and the implanted root results from neurite growth through a CNS environment. These neurites exhibit all the characteristics of CNS fibers and are equipped with oligodendroglia-derived myelin sheaths interrupted by nodes of Ranvier filled with tufts of microvilli-like extensions from surrounding astrocytic processes (Cullheim et al., 1989). In the present study, most of the labeled spinal neurons were located around the implantation site. These retrogradely labeled neurons were not necessarily motoneurons, since many types of neurons may extend their axons into peripheral nerve bridges (Benfey and Aguayo, 1982; Horvat et al., 1988; Richardson et al., 1984). Our findings suggest motoneuron regrowth into the ventral root, because of the reconnection with biceps brachii muscle demonstrated by electrical stimulation of the motor cortex and the C6 root, as well as by muscle histochemistry. In addition, large retrolabeled neurons were detected in Rexed’s lamina IX. However, as documented by our retrograde axonal tracing studies, the number of neurons involved in the reinnervation of the reconnected biceps was reduced relative to sham-operated animals. Extensive collateralization and terminal sprouting of the regenerating axons may explain the reinnervation of most of the muscle fibers observed in the reconnected biceps (Bertelli et al., 1995; Gordon et al., 1993; Hallin et al., 1999). Massive motoneuron loss occurs after spinal root avulsion in adult rats (Koliatsos et al., 1994; Wu, 1993), and the number of motoneurons gradually decreases further during the subsequent 2–8 weeks (Watabe et al., 2000). Motoneuron loss may be the direct result of the death of the injured neurons. However, it is also possible that loss of motoneurons is due to the atrophy of injured neurons since it has been reported that rubrospinal neurons do not die after axotomy, but, rather, they undergo massive atrophy that can be reversed by neurotrophic factors even 1 year after injury (Kwon et al., 2002). However, this is not the case in spinal motoneurons since immediately treatment with neurotrophic factors prevents the loss of motoneurons due to root avulsion (Wu et al., 2003) but delayed treatment (4 weeks after avulsion) with neurotrophic factors does not reverse the survival rate of motoneurons (data not shown), which indicates that injured motoneurons have indeed died. Neurotrophic factors support the survival, growth and differentiation of neurons and regulate synaptic plasticity during development and in the mature nervous system. After spinal root avulsion, replacement of trophic factors produced by the cellular components (e.g., Schwann cells) of peripheral nerves and target muscles have been used in animal studies to protect neurons from axotomy-induced cell death (Chai et al., 1999; Li et al., 1995; Lindsay, 1995; Wu et al., 2003). Spinal root avulsion is associated with leptomeningeal or dural tears, which may initiate a sequence of events, including acute inflammation, chronic inflammation and the formation of granulation tissue in the vicinity of the injury. The expression of neurotrophic factors is strongly induced as part of an inflammatory response mediated by

H.-Y. Gu et al. / Experimental Neurology 192 (2005) 89–99

macrophage infiltration, activation and cytokine release (Korsching, 1993). The results from the present report suggest that delayed reimplantation of ventral root may provide a sound support for axonal regrowth from injured, but surviving, motoneurons both by inducing trophic factors and mechanical bridging. This successful attempt indicated that implantation of the ventral root directly into the spinal cord is not essential. Intracellular injection experiments have shown that the motoneurons in the ventral horn possess dendrites reaching the subpial surface (Ruigrok et al., 1985), which may explain why the present superficially implanted grafts were well reinnervated. It is possible that daughter axons emerging from parent axon stumps reinnervate the replanted ventral root. Although myelin-associated proteins in the white matter strongly inhibit neurite growth (Caroni and Schwab, 1988a,b), motoneurons can regenerate through up to 900 Am of central neural tissue before entering a peripheral nerve implant (Cullheim et al., 1989). On the other hand, avulsion usually leaves tufts of the most proximal parts of roots attached to the spinal cord surface. Regrowth of axons in cats is promoted by the pial matrix, which is known to sustain neurite growth (Risling et al., 1985, 1993). In the pia mater close to the avulsed root, regenerating myelinated fibers emerge from the rootlet tufts (Hallin et al., 1999). In addition, motor axons from the root remnants can elongate along the leptomeninges to the distal stumps of the avulsed roots (Hallin et al., 1999). After spinal root avulsion, increased numbers of GFAP-positive cells show the activation of astrocytes. Avulsed axons in the white matter can regrow toward the pial area in a CNS environment filled with astrocytes (Hoffmann et al., 1996). The elongation of the injured motor axons mainly occurs along the adjacent spinal cord surface and the ventral roots (Novikov et al., 1995). Similarly, abundant myelinated fibers were followed from the avulsion site to the implanted ventral roots in the present study. Rebuilding the motor innervation of biceps brachii A key finding of this study was motor reinnervation of the musculocutaneous nerve and biceps brachii after delayed ventral root reimplantation. After avulsion of C6 and section of C5 and C7, fibers in the corresponding motor branch of biceps were bound to degenerate. Consequently, axons found in the motor branch of biceps were regenerating axons. The ability to evoke a muscle action potential when stimulating the C6 spinal root and contralateral cortex demonstrates that descending cortical neurons had directly and/or indirectly established connections with the surviving, regenerating motoneurons, whose axons had established functional connections with the muscle (Kline et al., 1986). Muscle reinnervation was confirmed anatomically by the large number of retrogradely labeled neurons in the spinal cord and by the large number of myelinated fibers in the musculocutaneous nerve. Furthermore, the combined silver

97

and cholinesterase staining methods and Texas Red-X conjugates of a-bungarotoxin fluorescent studies showed that the regenerated endplates had the appropriate topographic characteristics. After nerve transection, muscle fails to regain its normal mass and strength, even when a nerve graft is interposed between the nerve stumps (Frykman and Gramyk, 1991). In the present report, in spite of 2 weeks delay in reimplantation, reinnervation was remarkably successful. Therefore, in delayed ventral root reimplantation, the ability to improve the quality of motor recovery remains, and is comparable with the good recovery demonstrated after immediate ventral root reimplantation (Gu et al., 2004; Hallin et al., 1999; Hoffmann et al., 1993, 1996; Moissonnier et al., 1998). Biceps weight recovery was about 80%. The extent of axonal growth, measured as the distance between labeled cell bodies and muscle endplates, was at least 40 mm. This capacity of spinal neurons to regenerate over several centimeters has been observed by several authors (Bertelli and Mira, 1994; David and Aguayo, 1981; Duchossoy et al., 2001). A single motor unit can expand to approximately five times its original size, resulting in the ability to compensate for up to 80% motoneuron loss (Weiss et al., 1983). In this study, 53% of the motoneurons survived 20 weeks after ventral root reimplantation. Therefore, the remaining motoneurons were able to fully reinnervate biceps. Clinical relevance At present, patients who suffer from spinal root avulsion either receive no treatment, in which case function of the affected muscles is permanently lost, or undergo repair by the coaptation of the brachial plexus with anterior cervical nerves (Brunelli, 1987), the spinal accessory nerve (Allieu et al., 1987), the intercostal nerve (Nagano et al., 1989) or the contralateral C7 spinal nerve (Gu et al., 1992). In addition, in patients with total root avulsion, shoulder function can be restored by shoulder arthrodesis and hand function can be restored by free muscle transplantation in combination with nerve transfer (Nagano, 1998). A new procedure to restore finger flexion and extension, as well as elbow flexion, has been developed through a double free-muscle transfer and multiple nerve transfers using the spinal accessory and intercostal nerves (Doi et al., 1995). With the above surgical therapy, there is often some restoration of useful function to the proximal musculature, but restoration of function to the hand is rarely observed. Carlstedt recently repaired a spinal root avulsion at C6 to T1 by C6 ventral root implantation and sural nerve graft implantation through small slits in the ventrolateral spinal cord 1 month after trauma. Voluntary activity in proximal arm muscles was first detected electromyographically 9 months after surgery and the deltoid, biceps and triceps muscles regained considerable functional power (Carlstedt et al., 1995). Even although the efficacy of this procedure has

98

H.-Y. Gu et al. / Experimental Neurology 192 (2005) 89–99

not yet been established, it is a possible option if there is no suitable donor for nerve transfer. Ventral root reimplantation has been addressed in a long series of animalexperiments (Carlstedt, 1993; Carlstedt et al., 1986; Chai et al., 2000; Cullheim et al., 1989; Gu et al., 2004; Moissonnier et al., 1998; Risling et al., 1983), which show that axons from spinal cord motoneurons can grow into ventral roots and peripheral nerves. After avulsion, the ventral roots retract and it is difficult to trace them for the reimplantation procedure 2 weeks later. This retraction also reduces the length of root available for reconnection to the cord. The superficial ventral root implantation of this study reduces this problem, as well as obviating the local damage (Horvat et al., 1989) associated with direct implantation into the cord. The demonstration that delayed reimplantation of avulsed spinal ventral root can be used to guide regenerating central neurons to reinnervate muscle raises the possibility of an alternative treatment for patients with avulsion injuries of the ventral roots, especially those supplying the brachial plexus. However, a number of problems remain to be solved before delayed ventral root reimplantation can be routinely performed. What is the most suitable timing of ventral root reimplantation after trauma? What is the mechanism of axon regeneration through the denervated ventral root after a period of delay? Further investigation should compare the morphological and functional outcomes of ventral root reimplantation after different delays.

Conclusion In conclusion, the present data clearly show that an avulsed ventral root can be induced to reconnect the CNS to the PNS 2 weeks later. In our experimental context, the delayed ventral root reimplantation was effective in promoting motoneuron survival, axon regeneration and behavioral recovery. These results have important clinical implications for the treatment of avulsion injuries to the brachial plexus.

Acknowledgments This study was supported by grants from the University of Hong Kong, the Research Grants Council of Hong Kong, and the National Key Basic Research Programme of China (2003CB515303).

References Allieu, Y., Privat, J.M., Bonnel, F., 1987. Paralysis in root avulsion of the brachial plexus: neurotization by the spinal accessory nerve. In: Terzis, J.K. (Ed.), Microreconstruction of Nerve Injuries, pp. 415 – 423. Benfey, M., Aguayo, A.J., 1982. Extensive elongation of axons from rat brain into peripheral nerve grafts. Nature 296, 150 – 152.

Bertelli, J.A., Mira, J.C., 1993. Behavioral evaluating methods in the objective clinical assessment of motor function after experimental brachial plexus reconstruction in the rat. J. Neurosci. Methods 46, 203 – 208. Bertelli, J.A., Mira, J.C., 1994. Brachial plexus repair by peripheral nerve grafts directly into the spinal. cord in rats. Behavioral and anatomical evidence of functional recovery. J. Neurosurg. 81, 107 – 114. Bertelli, J.A., Taleb, M., Saadi, A., Mira, J.C., Pecot-Dechavassine, M., 1995. The rat brachial plexus and its terminal branches: an experimental model for the study of peripheral nerve regeneration. Microsurgery 16, 77 – 85. Brunelli, G., 1987. Neurotization of avulsed roots of the brachial plexus by means of anterior nerves of the cervical plexus. In: Terzis, J.K. (Ed.), Microreconstruction of Nerve Injuries. Elsevier, Amsterdam, pp. 435 – 445. Carlstedt, T., 1993. Functional recovery after ventral root avulsion and implantation in the spinal cord. Clin. Neurol. Neurosurg. 95, S109 – S111 (Suppl.). Carlstedt, T.P., 1995. Spinal nerve root injuries in brachial plexus lesions: basic science and clinical application of new surgical strategies. A review. Microsurgery 16, 13 – 16. Carlstedt, T., Noren, G., 1995. Repair of ruptured spinal nerve roots in a brachial plexus lesion. J. Neurosurg. 82, 661 – 663. Carlstedt, T., Linda, H., Cullheim, S., Risling, M., 1986. Reinnervation of hind limb muscles after ventral root avulsion and implantation in the lumbar spinal cord of the adult rat. Acta Physiol. Scand. 128, 645 – 646. Carlstedt, T., Grane, P., Hallin, R.G., Noren, G., 1995. Return of function after spinal cord implantation of avulsed spinal nerve roots. Lancet 346, 1323 – 1325. Carlstedt, T., Anand, P., Hallin, R., 2000. Spinal nerve root repair and replantation of avulsed ventral root into the spinal cord after brachial plexus injury. J. Neurosurg. 93, 237 – 247 (Suppl.). Caroni, P., Schwab, M.E., 1988a. Antibody against myelin-associated inhibitor of neurite growth neutralizes nonpermissive substrate properties of CNS white matter. Neuron 1, 85 – 96. Caroni, P., Schwab, M.E., 1988b. Two membrane protein fractions from rat central myelin with inhibitory properties for neurite growth and fibroblast spreading. J. Cell Biol. 106, 1281 – 1288. Chai, H., Wu, W., So, K.F., Prevette, D.M., Oppenheim, R.W., 1999. Longterm effects of a single dose of brain derived neurotrophic factor on motoneuron survival following spinal root avulsion in the adult rat. Neurosci. Lett. 274, 147 – 150. Chai, H., Wu, W., So, K.F., Yip, H.K., 2000. Survival and regeneration of motoneurons in adult rats by reimplantation of ventral root following spinal root avulsion. NeuroReport 11, 1249 – 1252. Cheng, H., Cao, Y., Olson, L., 1996. Spinal cord repair in adult paraplegic rats: partial restoration of hind limb function. Science 273, 510 – 513. Chuang, D.C., Lai, J.B., Cheng, S.L., Jain, V., Lin, C.H., Chen, H.C., 2001. Traction avulsion amputation of the major upper limb: a proposed new classification, guidelines for acute management, and strategies for secondary reconstruction. Plast. Reconstr. Surg. 108, 1624 – 1638. Clarke, P.G., Oppenheim, R.W., 1995. Neuron death in vertebrate development: in vitro methods. Methods Cell Biol. 46, 277 – 321. Cullheim, S., Carlstedt, T., Linda, H., Risling, M., Ulfhake, B., 1989. Motoneurons reinnervate skeletal muscle after ventral root implantation into the spinal cord of the cat. Neuroscience 29, 725 – 733. David, S., Aguayo, A.J., 1981. Axonal elongation into peripheral nervous system bbridgesQ after central nervous system injury in adult rats. Science 214, 931 – 933. Doi, K., Sakai, K., Kuwata, N., Ihara, K., Kawai, S., 1995. Double free muscle transfer to restore prehension following complete brachial plexus avulsion. J. Hand Surg. [Am.] 20, 408 – 414. Duchossoy, Y., Kassar-Duchossoy, L., Orsal, D., Stettler, O., Horvat, J.C., 2001. Reinnervation of the biceps brachii muscle following cotransplantation of fetal spinal cord and autologous peripheral nerve into

H.-Y. Gu et al. / Experimental Neurology 192 (2005) 89–99 the injured cervical spinal cord of the adult rat. Exp. Neurol. 167, 329 – 340. Frykman, G.K., Gramyk, K., 1991. Results of nerve grafting. In: Gelberman, R.H. (Ed.), Operative Nerve Repair and Reconstruction, pp. 553 – 567. Glasby, M.A., Hems, T.E., 1995. Repairing spinal roots after brachial plexus injuries. Paraplegia 33, 359 – 361. Gordon, T., Yang, J.F., Ayer, K., Stein, R.B., Tyreman, N., 1993. Recovery potential of muscle after partial denervation: a comparison between rats and humans. Brain Res. Bull. 30, 477 – 482. Gu, Y.D., Zhang, G.M., Chen, D.S., Yan, J.G., Cheng, X.M., Chen, L., 1992. Seventh cervical nerve root transfer from the contralateral healthy side for treatment of branchial plexus root avulsion. J. Hand Surg. [Br] 17 (5), 518 – 521. Gu, H.Y., Chai, H., Zhang, J.Y., Yao, Z.B., Zhou, L.H., Wong, W.M., Bruce, I., Wu, W.T., 2004. Survival, regeneration and functional recovery of motoneurons in adult rats by reimplantation of ventral root following spinal root avulsion. Eur. J. Neurosci. 19, 2123 – 2131. Hallin, R.G., Carlstedt, T., Nilsson-Remahl, I., Risling, M., 1999. Spinal cord implantation of avulsed ventral roots in primates; correlation between restored motor function and morphology. Exp. Brain Res. 124, 304 – 310. Hoffmann, C.F., Thomeer, R.T., Marani, E., 1990. Ventral root avulsions of the cat spinal cord at the brachial plexus level (cervical 7). Eur. J. Morphol. 28, 418 – 429. Hoffmann, C.F., Thomeer, R.T., Marani, E., 1993. Reimplantation of ventral rootlets into the cervical spinal cord after their avulsion: an anterior surgical approach. Clin. Neurol. Neurosurg. 95, S112 – S118 (Suppl.). Hoffmann, C.F., Marani, E., van Dijk, J.G., Kamp, W.vD., Thomeer, R.T., 1996. Reinnervation of avulsed and reimplanted ventral rootlets in the cervical spinal cord of the cat. J. Neurosurg. 84, 234 – 243. Horvat, J.C., Pecot-Dechavassine, M., Mira, J.C., 1988. Functional reinnervation of a denervated skeletal muscle of the adult rat by axons regenerating from the spinal cord through a peripheral nervous system graft. Prog. Brain Res. 78, 219 – 224. Horvat, J.C., Pecot-Dechavassine, M., Mira, J.C., Davarpanah, Y., 1989. Formation of functional endplates by spinal axons regenerating through a peripheral nerve graft. A study in the adult rat. Brain Res. Bull. 22, 103 – 114. Kline, D.G., Hackett, E.R., Happel, L.H., 1986. Surgery for lesions of the brachial plexus. Arch. Neurol. 43, 170 – 181. Koliatsos, V.E., Price, W.L., Pardo, C.A., Price, D.L., 1994. Ventral root avulsion: an experimental model of death of adult motor neurons. J. Comp. Neurol. 342, 35 – 44. Korsching, S., 1993. The neurotrophic factor concept: a reexamination. J. Neurosci. 13, 2739 – 2748. Kwon, B.K., Liu, J., Masserer, C., Kobayashi, N.R., McGraw, J., Oschipok, L., Tetzlaff, W., 2002. Survival and regeneration of rubrospinal neurons 1 year after spinal cord injury. Proc. Natl. Acad. Sci. U.S.A. 99 (5), 3246 – 3251. Li, L., Wu, W., Lin, L.F., Lei, M., Oppenheim, R.W., Houenou, L.J., 1995. Rescue of adult mouse motoneurons from injury-induced cell death by glial cell line-derived neurotrophic factor. Proc. Natl. Acad. Sci. U. S. A. 92, 9771 – 9775. Lindsay, R.M., 1995. Neuron saving schemes. Nature 373, 289 – 290.

99

Moissonnier, P., Duchossoy, Y., Lavieille, S., Horvat, J.C., 1998. Lateral approach of the dog brachial plexus for ventral root reimplantation. Spinal Cord 36, 391 – 398. Nagano, A., 1998. Treatment of brachial plexus injury. J. Orthop. Sci. 3, 71 – 80. Nagano, A., Tsuyama, N., Ochiai, N., Hara, T., Takahashi, M., 1989. Direct nerve crossing with the intercostal nerve to treat avulsion injuries of the brachial plexus. J. Hand Surg. [Am.] 14, 980 – 985. Narakas, A.O., 1984. Thoughts on neurotization or nerve transfers in irreparable nerve lesions. Clin. Plast. Surg. 11, 153 – 159. Novikov, L., Novikova, L., Kellerth, J.O., 1995. Brain-derived neurotrophic factor promotes survival and blocks nitric oxide synthase expression in adult rat spinal motoneurons after ventral root avulsion. Neurosci. Lett. 200, 45 – 48. Paxinas, G., Watson, C., 1986. The Rat Brain in Stereotaxic Coordinates, 2nd ed. Academic Press, New York. Richardson, P.M., Issa, V.M., Aguayo, A.J., 1984. Regeneration of long spinal axons in the rat. J. Neurocytol. 13, 165 – 182. Risling, M., Cullheim, S., Hildebrand, C., 1983. Reinnervation of the ventral root L7 from ventral horn neurons following intramedullary axotomy in adult cats. Brain Res. 280, 15 – 23. Risling, M., Dalsgaard, C.J., Terenius, L., 1985. Neuropeptide Y-like immunoreactivity in the lumbosacral pia mater in normal cats and after sciatic neuroma formation. Brain Res. 358, 372 – 375. Risling, M., Fried, K., Linda, H., Carlstedt, T., Cullheim, S., 1993. Regrowth of motor axons following spinal cord lesions: distribution of laminin and collagen in the CNS scar tissue. Brain Res. Bull. 30, 405 – 414. Rowland, L.P., 1991. In: Rowland, L.P. (Ed.), Amyotrophic Lateral Sclerosis and Other Motor Neuron Diseases. Raven Press, New York. Ruigrok, T.J., Crowe, A., ten Donkelaar, H.J., 1985. Dendrite distribution of identified motoneurons in the lumbar spinal cord of the turtle Pseudemys scripta elegans. J. Comp. Neurol. 238, 275 – 285. Sellin, L.C., Libelius, R., Lundquist, I., Tagerud, S., Thesleff, S., 1980. Membrane and biochemical alterations after denervation and during reinnervation of mouse skeletal muscle. Acta Physiol. Scand. 110, 181 – 186. Stichel, C.C., Muller, H.W., 1998. Experimental strategies to promote axonal regeneration after traumatic central nervous system injury. Prog. Neurobiol. 56, 119 – 148. Watabe, K., Ohashi, T., Sakamoto, T., Kawazoe, Y., Takeshima, T., Oyanagi, K., Inoue, K., Eto, Y., Kim, S.U., 2000. Rescue of lesioned adult rat spinal motoneurons by adenoviral gene transfer of glial cell line-derived neurotrophic factor. J. Neurosci. Res. 60, 511 – 519. Watchmaker, G.P., Mackinnon, S.E., 1996. Nerve injury and repair. In: Peimer, C.A. (Ed.), Surgery of the Hand and Upper Extremity. McGrawHill, New York, pp. 1260. Weiss, G.H., Feeney, D.M., Caveness, W.F., Dillon, D., Kistler, J.P., Mohr, J.P., Rish, B.L., 1983. Prognostic factors for the occurrence of posttraumatic epilepsy. Arch. Neurol. 40, 7 – 10. Wu, W., 1993. Expression of nitric-oxide synthase (NOS) in injured CNS neurons as shown by NADPH diaphorase histochemistry. Exp. Neurol. 120, 153 – 159. Wu, W., Li, L., Yick, L.W., Chai, H., Xie, Y., Yang, Y., Prevette, D.M., Oppenheim, R.W., 2003. GDNF and BDNF alter the expression of neuronal NOS, c-Jun, and p75 and prevent motoneuron death following spinal root avulsion in adult rats. J. Neurotrauma 20, 603 – 612.