Sensory Neuronal Protection & Improving Regeneration After

Home , Enteral administration, Epineurial repair

Sensory Neuronal Protection & Improving Regeneration After Peripheral Nerve Injury

Andrew McKay Hart

Printed by: Print & Media, Umeå university 2003:303077

UMEÅ UNIVERSITY MEDICAL DISSERTATIONS New Series No. 679 – ISSN 0346-6612 – ISBN 91-7305-370-8

Sensory Neuronal Protection & Improving Regeneration After Peripheral Nerve Injury

Andrew M cKay Hart From the Department of Surgical & Perioperative Sciences, Section for Hand & Plastic Surgery, and the Department of Integrative Medical Biology, Section for Anatomy

Abstract Peripheral nerve trauma is a common cause of considerable functional morbidity, and healthcare expenditure. Particularly in the ~15% of injuries unsuitable for primary repair, standard clinical management results in inadequate sensory restitution in the majority of cases, despite the rigorous application of complex microsurgical techniques. This can largely be explained by the failure of surgical management to adequately address the neurobiological hurdles to optimal regeneration. Most significant of these is the extensive sensory neuronal death that follows injury, and which is accompanied by a reduction in the regenerative potential of axotomised neurons, and in the supportive capacity of the Schwann cell population if nerve repair is delayed. The present study aimed to accurately delineate the timecourse of neuronal death, in order to identify a therapeutic window during which clinically applicable neuroprotective strategies might be adopted. It then proceeded to investigate means to increase the regenerative capacity of chronically axotomised neurons, and to augment the Schwann cells’ ability to promote that regenerative effort. Unilateral sciatic nerve transection in the rat was the model used, initially assessing neuronal death within the L4&5 dorsal root ganglia by a combination of morphology, TdT uptake nick-end labelling (TUNEL), and statistically unbiased estimation of neuronal loss using the stereological optical disector technique. Having identified 2 weeks, and 2 months post-axotomy as the most biologically relevant timepoints to study, the effect upon neuronal death of systemic treatment with acetyl-L-carnitine (ALCAR 10, or 50mg/kg/day) or N-acetyl-cysteine (NAC 30, or 150mg/kg/day) was determined. A model of secondary nerve repair was then adopted; either 2 or 4 months after unilateral sciatic nerve division, 1cm gap repairs were performed using either reversed isografts, or poly-3-hydroxybutyrate (PHB) conduits containing an alginate-fibronectin hydrogel. Six weeks later nerve regeneration and the Schwann cell population were quantified by digital image analysis of frozen section immunohistochemistry. Sensory neuronal death begins within 24 hours of injury, but takes 1 week to translate into significant neuronal loss. The rate of neuronal death peaks 2 weeks after injury, and neuronal loss is essentially complete by 2 months post-axotomy. Nerve repair is incompletely neuroprotective, but the earlier it is performed the greater the benefit. Two clinically safe pharmaceutical agents, ALCAR & NAC, were found to virtually eliminate sensory neuronal death after peripheral nerve transection. ALCAR also enhanced nerve regeneration independently of its neuroprotective role. Plain PHB conduits were found to be technically simple to use, and supported some regeneration, but were not adequate in themselves. Leukaemia inhibitory factor enhanced nerve regeneration, though cultured autologous Schwann cells (SC’s) were somewhat more effective. Both were relatively more efficacious after a 4 month delay in nerve repair. The most profuse regeneration was found with recombinant glial growth factor (rhGGF-2) in repairs performed 2 months after axotomy, with results that were arguably better than were obtained with nerve grafts. A similar conclusion can be drawn from the result found using both rhGGF-2 and SC’s in PHB conduits 4 months after axotomy. In summary, these findings reinforce the significance of sensory neuronal death in peripheral nerve trauma, and the possibility of its` limitation by early nerve repair. Two agents for the adjuvant therapy of such injuries were identified, that can virtually eliminate neuronal death, and enhance regeneration. Elements in the creation of a bioartificial nerve conduit to replace, or surpass autologous nerve graft for secondary nerve repair are presented.

Keywords: cell death; TUNEL; optical disection; dorsal root ganglion; peripheral nerve; nerve conduit; Schwann cell; glial growth factor; leukaemia inhibitory factor; acetyl-L-carnitine; N- acetyl-cysteine 2

CONTENTS

ABSTRACT ...... 1 ORIGINAL PAPERS...... 3 ABBREVIATIONS ...... 4 STEREOLOGICAL ABBREVIATIONS...... 5 INTRODUCTION...... 6 CLINICAL BACKGROUND ...... 6 NEUROBIOLOGICAL EFFECTS OF PERIPHERAL NERVE INJURY UPON PRIMARY SENSORY NEURONS...... 7 NEUROBIOLOGY OF THE DISTAL NERVE STUMP AFTER A TRANSECTING INJURY . 11 NEUROBIOLOGY OF NERVE REGENERATION AFTER REPAIR ...... 12 METHODOLOGICAL ISSUES IN THE QUANTIFICATION OF NEURONAL DEATH ...... 13 THE NERVE CONDUIT CONCEPT...... 14 AIMS OF THE STUDY ...... 16 MATERIALS AND METHODS...... 17 ANIMAL MODEL & ANAESTHESIA...... 17 CONDUIT MATERIALS & PREPARATION...... 17 Preparation of Cultured Schwann Cells:...... 18 Calcium Alginate Hydrogel:...... 19 NERVE REPAIRS ...... 19 DRUG ADMINISTRATION...... 20 TISSUE PROCESSING ...... 20 LIGHT MICROSCOPIC ANALYSIS...... 21 STATISTICAL ANALYSIS...... 23 RESULTS...... 24 PAPER I...... 24 PAPERS II & IV...... 24 PAPER III ...... 25 PAPERS V & VI...... 25 DISCUSSION...... 28 PERIPHERAL NERVE INJURY INDUCES SIGNIFICANT SENSORY NEURONAL DEATH, BUT A THERAPEUTIC WINDOW EXISTS ...... 28 THE TIMING OF NERVE REPAIR AFFECTS THE MAGNITUDE OF NEURONAL DEATH .. 28 NEURONAL DEATH CAN BE VIRTUALLY ELIMINATED BY CLINICALLY APPLICABLE TREATMENTS...... 29 NEURONAL REGENERATIVE CAPACITY CAN BE AUGMENTED AFTER SECONDARY REPAIR...... 30 A BIOARTIFICIAL NERVE CONDUIT TO RIVAL NERVE GRAFT FOR LATE SECONDARY NERVE REPAIR? ...... 31 CAN THE PROGNOSIS OF PERIPHERAL NERVE INJURY BE IMPROVED? ...... 33 CONCLUSIONS ...... 35 ACKNOWLEDGEMENTS ...... 36 REFERENCES...... 39

ORIGINAL PAPERS

The present thesis is based upon the following papers, referred to in the text by their Roman numerals:

I. Hart A. M., Brannstrom T., Wiberg M., Terenghi G. “Primary sensory neurons and satellite cells after peripheral axotomy in the adult rat: timecourse of cell death & elimination.” Exp. Brain Res. 2002;142(3):308-318 Reprinted with permission from Springer-Verlag.

II. Hart A. M., Wiberg M., Terenghi G. “Systemic acetyl-L-carnitine eliminates sensory neuronal loss after peripheral axotomy: A new clinical approach in the management of peripheral nerve trauma.” Exp. Brain Res. 2002; 145(2): 182-189. Reprinted with permission from Springer-Verlag.

III. Hart A. M., Wiberg M., Terenghi G. “Pharmacological enhancement of peripheral nerve regeneration in the rat by systemic acetyl-L-carnitine treatment.” Neuroscience Letters 2002; 334: 181-185. Reprinted with permission from Elsevier Science.

IV. Hart A. M., Terenghi G., Kellerth J.-O., Wiberg M. “Sensory neuroprotection, mitochondrial preservation, and therapeutic potential of N-acetyl-cysteine after nerve injury.” Manuscript; submitted to Neuroscience.

V. Hart A. M., Wiberg M., Terenghi G. “Exogenous leukaemia inhibitory factor enhances nerve regeneration after late secondary repair using a bioartificial nerve conduit.” BJPS, 2003 in press. Reprinted with permission of the editor.

VI. Hart A. M., Mosahebi A., Terenghi G., Wiberg M. “Interaction of glial growth factor & cultured Schwann cells in regeneration after chronic peripheral nerve injury.” Manuscript; submitted to J Periph Nerve Syst. 4

ABBREVIATIONS

Abbreviation Full Name Abbreviation Full Name Mitochondrial AIF’s Apoptosis Inducing Factors mtDNA Deoxyribonucleotidyl Acid ALCAR acetyl- L-carnitine NAC N-acetyl-cysteine ASPA Animals (Scientific Procedures) Act NGF Nerve Growth Factor BDNF Brain Derived Neurotrophic Factor NGS Normal Goat Serum C.A.S.T. Computer Assisted Stereology Toolbox NHS Normal Horse Serum CGRP Calcitonin Gene Related Polypeptide NO Nitric Oxide Ciliary Derived CNTF NOS Nitric Oxide Synthase Neurotrophic Factor Neuronal (constitutive) Nitric Cy-3 Cyanine-3 nNOS Oxide Synthase Nucleoside-Analogue Reverse DRG Dorsal Root Ganglion NRTI Transcriptase Inhibitor DRG’s Dorsal Root Ganglia NTF Neurotrophic Factor DNA Deoxyribonucleotidyl Acid Pan-Axonal Marker FITC Fluorescein Isothiocyanate PAMNF of Neurofilament Fn Fibronectin PBS Phosphate Buffered Saline GAP-43 Growth Associated Protein-43 PDS Polydioxanone GGF Glial Growth Factor PFA Paraformaldehyde GDNF Glial Derived Neurotrophic Factor PGP9.5 Protein Gene Product 9.5

H33342 Hoechst 33342 PI Propidium Iodide HIV Human Immunodeficiency Virus PHB Poly-3-hydroxybutyrate ICE Interleukin Converting Enzyme PTFE polytetrafluoroethylene IL-1β Interleukin-1β ROS Reactive Oxygen Species iNOS Inducible Nitric Oxide Synthase SC’s Schwann Cells ISEL In-Situ End Labelling SD Standard Deviation JNK c-Jun Kinase SOD Superoxide Dismutase Terminal deoxyribonucleotidyl TdT transferase LIF Leukaemia Inhibitory Factor TNFα Tumour Necrosis Factor α Terminal deoxyribonucleotidyl MAPK Mitogen Activated Protein Kinase TUNEL transferase Uptake Nick-End Labelling MMP Mitochondrial Membrane Potential VIP Vasoactive Intestinal Polypeptide mRNA Messenger Ribonucleotidyl Acid

STEREOLOGICAL ABBREVIATIONS

Abbreviation Full Name Abbreviation Full Name

a Area Nv Numerical Volume

a Mean Area Npar Number of Particles t Thickness s Number of Sections

t Mean Thickness Vdis Disector Volume

Ndis Number of Disectors Vref Reference Volume

INTRODUCTION

No adult with a major nerve transection has ever attained normal sensibility. Luis De Medinaceli & Anthony Seaber

The first prerequisite for axonal regeneration is the survival of the neuron following injury. Samuel Fu & Tessa Gordon

Clinical Background

In Europe approximately 300,000 peripheral nerve injuries occur each year, although local variation depends upon the preponderance of heavy industry and mechanised agriculture. Most are caused by direct mechanical trauma, resulting in transection, crush injury, traction, or avulsion 1-5, although other significant mechanisms include compression, thermal injury, and neurotoxins 2,6,7. Injuries are clinically graded as “neuropraxia”, “axonometsis”, or “neurometsis” as progressively more disruption of the normal neural anatomical structure occurs 1, or into Grades 1-5 if more detailed consideration of the fascicular disruption is made 2,8. Nerve repairs can also be classified by technique and timing, and the temporal classification given by Green 9 has been adopted in this thesis. Primary repairs are those performed within 48 hours of injury, early secondary repairs between 2 days and 6 weeks after injury, and late secondary repairs between 6 weeks and 6 months after injury. Nerves are repaired primarily where possible, however secondary repair may be elected for wounds caused by high velocity projectiles; when tissues are significantly devitalised, contaminated, or ischaemic; in complex reconstructions; and following a diagnostic delay after closed injuries or failed primary repair 6,9,10. Current clinical management remains based upon direct microsurgical nerve repair 11 by either epineurial or fascicular suture 12, with autologous nerve grafting reserved for those injuries that cannot be repaired without undue tension 6,10,13. Unfortunately the considerable advances in surgical technique over recent years 6,10,12-19 have not been matched by similar improvements in outcome 19-21, and currently "all qualities of sensory and motor function are difficult or impossible to regain in adult patients" 22. Cutaneous innervation does not return to normal 23, and sensory outcome remains sufficiently poor 6,10,24 that the “...normal restoration of function after significant nerve injury remain(s) unattainable." 18. This is the case in even the best of scenarios, outcome is even worse in the ~35% of injuries that involve major nerve trunks 8, in complex nerve injuries 6,9, and in the 11-16% of cases 8,25,26 that require nerve grafting 6,9,15,26,27 or a significant delay before repair 6,9,17,28,29. Other factors such as patient age, smoking

7 habits, associated injuries, and co-operation with rehabilitation are also significant 8,26,30- 32. Since adequate proprioceptive feedback is a vital component of normal motor control 33, sensory outcome is also an integral determinant of motor outcome, particularly in fine manipulative work 34. Peripheral nerve injuries can therefore cause extreme disability, loss of employment and dependence upon others. As a result, peripheral nerve trauma is a significant cause of functional morbidity, healthcare expenditure, social disruption and lost employment 10. Microsurgical techniques have reached such a level of technical sophistication that future enhancements are unlikely to provide further improvements in outcome, and so attention has turned towards modulation of peripheral nerve neurobiology 19,21. In order to identify the neurobiological factors that underly the poor sensory outcome of peripheral nerve injuries, it is necessary to consider the ways in which its various components react to injury and repair. Neurobiological Effects of Peripheral Nerve Injury Upon Primary Sensory Neurons

In the healthy, intact peripheral nervous system, quality of sensation is dependent upon the density of sensory receptor organs, the number of innervating neurons, the size and degree of overlap of their receptive fields, and upon the cortical interpretation of the resulting topographically and qualitatively defined sensory input 20,30,35. All these components are affected by transection and repair of a peripheral nerve 6,19,20,22,36, and could only be restored by rapid and specific reinnervation by an adequate number of neurons. It is therefore likely that the single most important determinant of the poor sensory outcome of peripheral nerve trauma is the extensive primary sensory neuronal death that occurs. Axonal injury briefly exposes the intracellular compartment to the extracellular milieu 37, triggering ion fluxes and antidromic electrical activity that initiate pathways for neuronal death and for regeneration; these are subsequently reinforced by neurotoxic cytokines such as TNFα at the site of injury 38,39. Calcium dependent restoration of axolemmal continuity then occurs 40 as part of the process of somatofugal axonal atrophy 9,41-43, and the axotomised neurons display a stereotyped perikaryal injury response pattern, termed chromatolysis 44. This involves Nissl body displacement 6,9,41,45,46, plus nuclear condensation and peripheralisation 45-47, while metabolically active mitochondria accumulate within the cell body 47, central processes retract 48, and the perikaryon shrinks 45,49. Conversely, some neurons develop a grossly dilated, vacuolated morphology that has been ascribed to a regenerative phenotype 50,51.

Profound phenotypic changes occur, as seen by alterations in mRNA transcription 52-57, protein synthesis 58-63, growth factor and cell receptor profiles 41,55,64- 70, axonal transport 55,63,71, the secretion of neuropeptides and neurotrophic factors 41,61,62,64,67,72,73, and in the cells’ electrophysiological properties 74,75. Given the importance of target-derived neurotrophic factors (NTF’s) in neuronal survival and regeneration 18,20,41,76,77 it is of note that after injury NGF, GDNF, neurotrophin-3 (NT- 3), and BDNF are retrogradely transported to subpopulations of cells with differing mean diameters 41,78,79. The gp-130 neurocytokines leukaemia inhibitory factor (LIF) and ciliary neurotrophic factor (CNTF) are taken up by an increased range of motorneurons and, mainly small diameter, sensory neurons 41,79,79-83. Other factors, as yet undiscovered, may possibly play equal, or more significant roles. Since the early 1900’s it has been recognised that experimental nerve injuries lead to primary sensory neuronal death 51,84,85, and more recently this has been confirmed by the demonstration of DNA fragmentation in vivo 86-90. It is now apparent that a considerable proportion of all neurons in the sensory pool will die 20,39,41, with estimates ranging from 7 to 50% 38,49,51,86,91-96 depending upon the exact nature of the experimental model. Similar figures have been found in higher mammals 97-100 and a human post-mortem study 101, and appear to correlate with loss of functional sensory units 35. Neuronal loss appears to be more limited after crush injuries or when the transected nerve is immediately co-apted 102,103, but these findings may merely represent an effect upon horseradish peroxidase transport rather than cell death 103. Hence the effect of the timing of nerve repair upon neuronal death has not been fully investigated, despite its obvious clinical significance. Proximal injuries may be more likely to induce neuronal death than distal ones 20, although the evidence is clearer for motor than sensory neurons, which are far more sensitive to the effects of peripheral axotomy 41. Ygge et al demonstrated a fourfold greater loss of L4-6 DRG neurons after proximal sciatic nerve transection than distal 104, however the sciatic nerve contains afferent fibres from approximately 19,000 neurons at the proximal site 105, but from only ~5,000 distally 102. So the fourfold difference in loss may merely reflect the number of neurons actually axotomised, implying that sensory neuronal death might be critical to the clinical sensory outcome of injuries even as distal as digital nerve transections. The timecourse of sensory neuronal death has not been fully described, and yet is of crucial clinical importance since it defines a therapeutic window during which neuroprotective therapies could be instituted. Groves et al 86 demonstrated DNA fragmentation within neurons 7 days after sciatic nerve division in the rat, and although

9 most authors have not demonstrated any neuronal loss within 7 days of axotomy, Vestergaard et al found a 6% loss at 4 days, rising to 19% after 8 days 49. It therefore remains uncertain exactly how soon after peripheral nerve injury neurons begin to die. Nor is it clear how long the period of cell death lasts. The studies that have employed sufficiently long timepoints have not shown any significant difference between the loss found 1-2 months after injury and that found after 6-12 months 86,93,94, but either used non-stereological counting techniques 94, lacked sufficient timepoints 93, or used small groups 86,94. Despite the initiating effects of retrograde axonal stimuli and neurotoxic cytokines, the principal determinant of how many DRG neurons die after peripheral nerve transection seems to be the loss of target derived neurotrophic support 20,39 that arises secondary to interruption of retrograde axonal transport. Evidence of this comes from the protective effects of exogenous neurotrophic factor administration 19,41. Nerve growth factor (NGF), LIF, GDNF, NT-3, CNTF, and BDNF can reduce primary sensory neuronal loss, but efficacy may depend upon the route of administration, and the neuronal subpopulation under consideration 78, 91, 92, 95, 106-111. Signalling is effected principally through the mitogen activated protein kinase (MAPK), c-Jun kinase (JNK), and PI3 kinase pathways 39,77,112, which lead to transcription of both cell death (via JNK,PI3K) 59 and regenerative (via MAPK) 112 genes. These pathways increase c-Jun cell protein activity 58-60,113-116. A member of the AP-1 superfamily, c-Jun is a gene transcription regulator which appears to be essential, though not specific 58,114,117, for apoptosis 59,114,118, since it is also increased in actively regenerating neurons 116,119. In contrast, another immediate early gene, c-fos, does appear to be specific for apoptosis 114 since its induction coincides with the appearance of apoptotic morphology and the irreversible stages of the cell death cycle 59. Much of our knowledge of the mammalian system comes from the identification of homologous proteins, or cDNA sequences in lower order species. In the nematode caenorhabditis elegans the pro-apoptotic protein ced-3 120 has structural homology with cysteine protease interleukin-1β-converting enzyme (ICE) in mammals 120, whilst the anti-apoptotic ced-9 121 has homology with the mammalian family of Bcl proteins 121,122. Both proteins act downstream of c-Jun 59,117 and the ratio of Bax:Bcl gene products is predictive of entry into the irreversible stages of apoptosis 58,123,124, although isolated Bcl, or Bax levels are not predictive of apoptosis 125,126. Although their mechanism of action remains uncertain, they are postulated to be part of the trophin withdrawal pathway 113,127-130, and may control antioxidant activity 121, or attempted entry into the cell cycle 131,132. Bax/Bcl events occur upstream of the ICE/caspase

10 dependent steps 112, and may include the control of c-fos induction 117, or mitochondrial calcium flux and the release of pro-apoptotic molecules 121,132,133. In a range of conditions mitochondria play a central role in neuronal death 47,133-135 through the failure of oxidative metabolism, the generation of reactive oxygen species, and the release of pro-apoptotic molecules that trigger the caspase cascade 133. Mitochondria accumulate in the perikaryon after axotomy, and oxidative stress 47,136 may be mediated by increased nitric oxide synthase NOS activity 137,138 and mitochondrial NO levels 139-141. Physiological regulation of respiratory ATP production 142 normally involves reversible NO inhibition of complex IV activity in the mitochondrial electron transport chain. However sustained increases after axotomy may permanently inhibit cytochrome C (complex I), disrupt electron transport 142 and cause the generation of reactive oxygen species (ROS) 140. It is postulated that as the bioenergetic dysfunction worsens, mitochondrial membrane potential (MMP) 133,139,143 and calcium dysregulation may occur 134, further increasing the production of ROS 134. Mitochondrial protophores may then open, directly triggering neuronal death by the release of cytochrome-C (which activates caspase-9), and other apoptosis inducing factors (AIF’s) that act downstream of the caspases 133,140,144. Although considerable evidence exists for the role of reactive oxygen species (ROS), lipid peroxidation, and oxidative stress as early 145 mediators of neuronal death 124,136,140,141,146, direct evidence from primary sensory neurons in vivo remains rare. Mn-superoxide dismutase activity increases 30% in the DRG after axotomy, suggesting an attempted homeostatic response 147, but although some antioxidant agents are protective against ROS 121,145,148,149, others are not 149, and it is not generally concluded that the successful agents act purely by a direct antioxidant mechanism 121,148,149. Abortive entry into the cell cycle has been proposed as a means of cell death 131,150-152, and anti-apoptotic agents such as Bcl, acetylcysteine and protein synthesis inhibitors also inhibit entry into the cell cycle 131,132,149,153. Thus axotomy may cause abortive entry into the cell cycle, at a time at which this cannot be sustained 151, with the result that neurons apoptose. The ICE-like proteases, or “caspases” 154,155, are the most likely effector mechanism in DRG neurons 120,156, and may be activated by trophin withdrawal, axotomy, Fas ligand or p75 activation 77, and by mitochondrial disruption 133. The caspases, particularly caspases-3 & 9 157-160, are intrinsic to active neuronal death 117,155,157,161, appear in the pathway downstream of Bax activation and c-fos induction 117, and lead to DNA fragmentation by activated α-spectrin proteases such as DNAse-1 124,133.

After peripheral nerve transection glial cells in the DRG also die 86,88. Although the significance of this is uncertain, in the spinal cord 162 it has been suggested to compound the effects of the initial injury 163. In neonates glial death occurs with a slower time course to that of neuronal death 89, but the timecourse in adults has not been described. Despite the importance of loss of neurotrophic support in stimulating neuronal death, and over 40 years of research since the discovery of NGF 164, growth factors have had no impact upon the clinical management of nerve injuries 18 for variety of reasons. NTF’s act on specific subpopulations of neurons 18,41, may have unpredictable effects 165-167, and there are concerns over potential carcinogenic actions and toxicity 168. Side effects 169-171 and complications 169,172 have also limited their applicability, so there is still a pressing need for an alternative approach using an easily administered, clinically safe therapy. In contrast to the specificity of neurotrophic factors, there appears to be relative homogeneity between neuronal subpopulations, and the preservation of the way that the initiation of neuronal death is signalled and then effected 39,146. Mitochondrial bioenergetic dysfunction, and the ensuing generation of reactive oxygen species and release of pro-apotoptic molecules 133, are therefore attractive targets given their potential role in neuronal death in a range of neurodegenerative and traumatic pathologies 133,141,146,148,173-178. Alternatively, nerve repair might provide a “physiological” cocktail of factors that might address all neuronal subtypes. Neurobiology of the Distal Nerve Stump After a Transecting Injury

Nerve transection is accompanied by haemorrhage, retraction of the nerve trunk due to its intrinsic elasticity, and mushrooming of neural contents from the cut ends. An inflammatory reaction occurs causing fibrin deposition, scarring, and release of neurotoxic cytokines. Intra-neural fibrosis begins within the endoneurial sheaths of the distal nerve stump 28-35 days after transection 9, reducing their calibre 42,179 to only 1% after 2 years in man 9. Distal to the injury axons detach from their myelin sheath 41 and are phagocytosed by macrophages and activated Schwann cells 41,63,70,179,180 in the process of Wallerian degeneration 6,9,21,41, which is completed within one month in the rat sciatic nerve 179. Cells in the nerve trunk, most notably Schwann cells, initially upregulate NGF, LIF, GDNF, GAP-43, and GGF receptor (p185erbB2/neu) mRNA 21,52,63,70,82,181,182. Schwann cells increase NGF, BDNF, NT4 and LIF secretion 41,63,70,82,180, and damaged ones are an important source of CNTF 63,70,83. Those in the distal stump shrink away

12 from the endoneurium to form solid columns, the “Bands of Bungner” 21,179,183,184. After an initial period of proliferation, and upregulation of c-erbB2&4 (parts of the neuregulin receptor complex) 65 under the influence of macrophage derived IL-1 41,63,180, Schwann cells begin to atrophy and die, by apoptosis 88, if denervation is prolonged 21,42,179,183,184. They become unresponsive to GGF due to downregulation of c-erbB2&4 expression after 2-6 months of denervation 65,179,183, although they remain able to myelinate axons 185. Schwann cells are essential for the guidance and support of regenerating axons 20,21,70,186,187 and glial growth factor (GGF) 188 is the principal neuregulin controlling that interaction 20,21,41,70. Prolonged axotomy therefore impairs the distal nerve stump’s capacity to promote nerve regeneration after any subsequent repair due to endoneurial fibrosis, atrophy, and loss of GGF responsiveness of the Schwann cell population 42,65,184,189. There is also some correlation with proximal injuries in long nerves undergoing primary repair, where the distal segment also remains chronically denervated 20. The potential for improving regeneration by therapeutically addressing the atrophic Schwann cell population is therefore clear. Target organs atrophy during the period of denervation, but the sensory organs fare somewhat better than muscle 9. In the skin Merkel's discs disappear by 35 days in the cat 190, Meissner’s corpuscles persist for 6 months before degenerating 3, but although Pacinian corpuscles atrophy, they survive for at least 10 months in the rat 191. Neurobiology of Nerve Regeneration After Repair

Axons begin to sprout 20 after a delay of around 3-14 days 6,9. Nerve fibres grow by sprouting ~50-100 neurites that advance through the repair site only to be pruned down to around 5 neurites per parent axon when the endoneurial tubes of the distal stump are reached 20. Although neurite growth is facilitated by contact guidance from neurite outgrowth promoting factors 192-194, tropic, and trophic influences 10,20,22, it is also dependent upon the neurons inherent regenerative capacity. This is enhanced by adoption of the regenerative phenotype 20, partly in response to “injury factors”, such as LIF 82,195. Schwann cells are essential, providing neurotrophic and tropic guidance, and cellular adhesion molecules 19,21,196. As a result axons preferentially reinnervate the distal stump over neighbouring tissues, and display preferential motor reinnervation in the selection of endoneurial tubes 10,194,197. This has been postulated as a potential benefit of short gap entubulation repairs over direct suture 194, and studies have shown the technique to have certain advantages 198-200, although other authors disagree 21,201.

There is evidence to suggest that after prolonged axotomy the regenerative capacity of the neurons deteriorates, in addition to the deleterious changes within the distal stump. Complete loss of regenerative capacity through denatured muscle grafts occurs after more than 56 days delay 28, and axonal ingrowth into the distal stump is reduced after repair onto the native 65, or freshly divided distal stump 202. This may be a reflection of the loss of elements of the regenerative phenotype 20,65, raising the possibility that outcome could be improved by therapies that induce that phenotype once again. Methodological Issues in the Quantification of Neuronal Death

The interpretation of neuron counts has been complicated by the use of various assumption based counting techniques that are “statistically biased”, meaning that the mean of all possible estimates will tend away from the true value 203. Serial reconstruction counts are the gold standard technique, but are impractical in ganglia which contain tens of thousands of neurons, and so methods for estimation have been developed instead. The simplest of these is profile counting, but since large neurons have profiles in more sections than small neurons, the count is likely to be a biased 204 overestimate 205. To combat this problem various correction factors were developed, such as those of Abercrombie 206 and Königsmark 207, but such estimates necessarily involve assumptions about homogeneity of cell shape and distribution that have been demonstrated to be untrue 205,208. Hence the counts are “biased” in the statistical sense that the mean of all possible estimates tends away from the true value 209. The techniques are also imprecise, since variations of over 40% 204 have been found for individual ganglia 210, and figures ranging from 2,000 to 15,000 have been quoted for the number of neurons in normal rat L5 DRG’s 211. Such techniques are inaccurate when tested against serial reconstruction 205. In order to overcome these problems assumption free, design-based stereological techniques have been developed 208, that are comparatively time-efficient 209. These are becoming the standard tools for quantitative morphological study of biological systems, although this has recently been questioned 212. However, when the techniques are applied correctly, with adequate systematic random sampling, they are time efficient, theoretically sound, and have been empirically validated against serial reconstruction counts of DRG neurons 204,213. Correct systematic random sampling obviates any bias arising due to the non-homogenous distribution of the neurons 203, and by counting within three dimensions 203 every cell has an equal chance of being counted irrespective of its size, shape, or orientation 203,208. Stereological counting techniques,

14 such as the physical and optical disector 204, therefore permit “statistically unbiased” counting, whereby the mean of all possible estimates will tend infinitely toward the true value 209. Counts are therefore accurate 208, as long as certain rules are adhered to 208,214. The optical disector technique ought to be superior to the physical, since it is more time- efficient 215 and avoids underestimates due to tissue sectioning artefacts and lost caps 203,210,215. Although precision is proportional to the amount of sampling performed, biological variation rapidly outweighs the benefit of increased sampling, and so counting remains time-efficient 208. Although neuron counting permits quantification of the total number of neurons that have died, complete involution does not occur for some hours in vitro 145,216,217 and an unspecified time in vivo 218 after the start of the cell death cycle. Evidence of individual cell death, whether morphological or DNA fragmentation, is therefore more appropriate for determining the onset of cell death after axotomy, as well as being more sensitive for detecting low rates of cell death, since the normal biological variation in neuronal numbers may also mask very small losses. As a marker of DNA fragmentation, terminal deoxyribonucleotidyl transferase (TdT) uptake nick-end labelling (TUNEL) is a reliable marker of neuronal death 87-89, particularly when combined with microwave tissue permeabilisation 219 and morphological confirmation of the mode of cell death 220,221. The Nerve Conduit Concept

Since the 1820’s attempts have been made to repair nerves using grafts and tubes of a remarkable range of biological and synthetic materials 15,19,27,222-225. Most recently this has been in an attempt to avoid the problems associated with autologous nerve graft harvest 225, and clinical studies have been limited to the use of empty tubes which hope to concentrate autogenous regeneration promoting factors 224-226. Experimental research has turned towards identifying bioengineered solutions to meet the requirements of regenerating neurons 199,227-229, with the aim of creating a bioartifical construct with a better regenerative profile than nerve autograft. Thus a conduit should have adequate physical strength to permit repair, it should be bioresorbable to avoid subsequent nerve compression, provide longitudinal directional cues for neurites, be selectively porous to promote neovascularisation, and allow incorporation of biologically active elements such as growth factors 199,225,230. Various materials have shown promise 15,222,224,231-235, but few match the potential of the poly-3-hydroxybutyrate conduit 198,232,236,237. PHB is a clinically licensed synthetic material with no risk of infectious agent transmission, unlike bovine fibronectin or some collagens. Unlike

15 silicon and PTFE, it is bioresorbed by a non-inflammatory process into a non-toxic metabolite over a period of 6-30 months, a time period that adequately allows for nerve regeneration even across long gaps. It can provide mechanical strength to the repair, whilst isolating the nerve from extrinsic fibrosis, and is soft, flexible and easy to handle in contrast to other polymers such as polydioxanone, polylactic (PLA), or polyglycolic (PGA) acid. The longitudinal fibre alignment also provides a directional cue to the neurites and Schwann cells, and over time some fibres dislodge into the lumen, further enhancing this function. PHB conduits can potentially be filled with various bioresorbable matrix substitutes, and calcium alginate-fibronectin hydrogel enhances regeneration and permits local delivery of cultured Schwann cells 238,239. Alginate has also been used as a slow release delivery system for administration of exogenous growth factors, is licensed for clinical use in other applications, and being of inert algal origin is safer than matrigel, which is derived from a mouse sarcoma cell line.

AIMS OF THE STUDY

The overall aim of this project was therefore to identify clinically applicable therapeutic strategies to improve the outcome of peripheral nerve injuries by reducing the death of primary sensory neurons, and facilitating subsequent regeneration towards the target tissues. Two of the principle problems of late secondary repair were more specifically addressed in the latter part of the project, namely atrophy of the

Schwann cell population, and the decreased regenerative capacity of chronically axotomised neurons.

The specific aims were:

1) To further investigate the morphology of primary sensory neuronal death

after peripheral nerve transection, and to establish the precise timecourse

in order to identify a potential therapeutic window (Paper I).

2) To establish, in principle, whether the timing of peripheral nerve repair

affected the magnitude of primary sensory neuronal death (Paper I).

3) To identify adjuvant neuroprotective therapies that are clinically

applicable (Papers II & IV).

4) To investigate ways to enhance the regenerative capacity of chronically

axotomised neurons (Papers III & V).

5) To investigate means to augment the support of peripheral nerve

regeneration after late secondary repair by Schwann cells (Paper VI).

MATERIALS AND METHODS

More detailed descriptions of the materials and methods employed in this study are given in the relevant original papers (I-VI), but a general overview is given below. Animal Model & Anaesthesia

The experimental model chosen throughout this project was the rat sciatic nerve model. This provided minimal inter-individual variation by the use of an inbred strain, permitted the preparation of isogeneic nerve grafts and Schwann cell cultures, and facilitates comparison with other published work. All work complied with the requirements of the Animals’ (Scientific Procedures) Act 1986 (United Kingdom), and was cogniscent of the need to optimise welfare issues. All surgical procedures from which the animals recovered were performed under an adequate depth of inhalational anaesthesia (Halothane®, May& Baker Ltd., U.K.), and used a standard post-operative analgesic regimen (0.1ml/100g body weight Temgesic®, Schering-Plough Ltd., U.K.). Surgery was conducted with aseptic techniques in a dedicated operating theatre, with a separate facility for post-operative recovery. All studies involved an initial unilateral sciatic nerve transection in the mid- thigh, and, except in the study of the effect of the timing of nerve repair (Paper I), the nerve stumps were ligated and placed into silicon rubber caps to prevent spontaneous regeneration. Resorbable 4/0 Polydioxanone skin sutures (PDS®II, Ethicon Ltd., U.K.) were employed in all cases. Terminal anaesthesia for perfusion-fixation prior to tissue harvest (Papers I, II, IV) was achieved with 0.5ml/kg body weight SagattalTM (Rhône Mérieux, sodium pentobarbitone 60mg/ml). Harvesting of peripheral nerve repairs was performed after terminalisation by CO2 narcosis and cervical dislocation (Papers III, V & VI). Conduit Materials & Preparation

Poly-3-hydroxybutyrate (PHB): PHB (Papers V & VI) is a biodegrable polymer derived by extraction from bacterial culture 240,241, and can be processed into various forms 242-247. PHB is slowly degraded in vivo by hydrolysis into non-toxic β- hydroxybutyric acid 241. This project used sheets compressed from longitudinally oriented 2-20µm 150kD molecular weight PHB diameter fibres (AstraTech AB, Mölndal, Sweden), which were then formed into 1.4cm long tubes 248 with an internal diameter of 1.8mm. In this format bioresorption will occur between 6 and 24-30 months after implantation 243,245,246.

Preparation of Cultured Schwann Cells: In order to obtain Schwann cells (SC’s) for culture, sciatic nerves were harvested from 20 neonatal rats and digested for 45 minutes using 1% collagenase I (Worthington Biochemicals) and 0.25% trypsin (Gibco). The digestant was then triturated through 21G and 23G needles, filtered (70 µm cell filter, Falcon), and centrifuged at 800 revolutions per minute (rpm) for 5 minutes. The resulting cells were resuspended in a basic medium comprising Dulbecco’s Minimum Eagles Medium plus Glutamax , DMEM (Gibco) plus penicillin 100 iu/ml and streptomycin 100µg/ml (Gibco), and 10% fetal calf serum (Imperial Laboratories). After 24 hours incubation in 2 a 25cm poly-D-lysine (Sigma) coated flask stored at 37°C, 95% humidity, and 5% CO2, the basic medium was then exchanged for one containing 10µM cytosine-β-D- arabinofuranoside (Sigma), and incubated a further 48 hours to stop fibroblast growth. Since the majority of the remaining cells were now SC’s, the medium was then changed to a growth medium that comprised basic medium containing recombinant glial growth factor II 63ng/ml (Max1/2 activity, 4.8 ng/ml, Cambridge NeuroScience), and 10µM forskolin (Calbiochem). After reaching confluence the SC’s were further purified by trypsinisation (2.5% Gibco, at 37°C) and centrifuging the suspension at 800 rpm for 5 minutes. Cells were then suspended in 1ml of medium containing 1:1000 mouse anti Thy 1 (Serotec, MCA04), and incubated at 37°C for 30 minutes, plus a further 15 minutes after the addition of 250µl of rabbit anti mouse complement (Cedarlane). The SC’s (around 120x105) were then washed and grown on, changing the medium every 48 hours and splitting the SC’s (1 in 3) on confluence 249. The purity of SC’s was 98% as assessed separately by S100 staining. At the second passage following purification, the SC’s were transduced using the retroviral vector pMFG lacZ nls 250, which contains a 3.5 kb modified E. coli lacZ marker gene encoding the β-galactosidase protein and a nuclear localisation sequence (nls). The Moloney Murine Leukaemia Virus (MMLV) packaging cell line PT67 (Clontech, USA) was used for transduction. Its genome contains the three retroviral structural genes, but lacks the packaging signal provided in trans as part of the MFG lacZ nls that contains the lacZ gene cloned between the signal and two long terminal repeats (LTR). In this way retrovirus encoding the lacZ gene can transduce SC’s, but the transduced SC’s cannot produce any further retrovirus due to the absence of the retroviral structural genes from their genome. A stable PT67 lacZ nls producing clone (titre approximately 106/ml) was isolated, and a confluent layer was incubated with 10ml of growth medium (high glucose DMEM/ FCS) at 32°C overnight to create a medium that containing retroviral particles. This medium was then filtered (0.45µm

19 filter, Nalgene) and added to a flask of 70% confluent SC’s, together with 8 µg/ml of the polyanion polybrene (Sigma). The transducing medium was removed after 4 hours incubation at 32°C, SC growth medium was added 251, and the flask kept at 37°C overnight. This cycle was repeated three times over 3 days, giving an 83% transduction rate as assessed by 5-bromo-4-chloro-3-indoyl-β-D galactosidase (X-gal) staining to detect β-galactosidase expression. Transduced SC’s were then available for subsequent implantation at the time of nerve repair in experimental animals. Calcium Alginate Hydrogel: The conduits were filled with ~50µl of a solution containing 2% ultrapure, low molecular weight, high mannuronic acid content sodium alginate (Pronova, Norway) and 0.05% bovine fibronectin (Sigma Pharmaceuticals, U.K.), which enhances nerve regeneration through alginate hydrogels 238,252. This solution served as a matrix substitute, and delivery vehicle for either recombinant murine leukaemia inhibitory factor (rhLIF 100ng/ml, Autogen Bioclear, U.K.), recombinant glial growth factor (rhGGF-2 500ng/ml, Cambridge Neuroscience, U.K.), cultured autologous Schwann cells (SC’s, 80x106/ml), or both (rhGGF-2 & SC’s). Sodium alginate is an algal derivative comprising linear co-polymers of 1,4-β-mannuronic and 1,4-α-guluronnic acid units, the relative proportions of which determine its physical properties, such as viscosity. The polymer units can be cross-linked by divalent cations (such as Ca++ ions) to create a stable hydrogel, that is gradually bioresorbed by non-enzymatic hydrolysis as the Ca++ ions leach out, and renally excreted 239. The filled conduits were therefore immediately immersed in sterile 0.1M CaCl2 for 2 minutes to create a stable crosslinked hydrogel 238. Nerve Repairs

Direct Neurorraphy: in Paper I nerve repair was performed by direct suture. Stump orientation was optimised and microsurgical repair achieved using four interrupted 9/0 nylon (Ethilon®, Ethicon Ltd., U.K.) sutures. Gap Repair: in Papers III, V, and VI late secondary nerve repairs were performed, either 2 or 4 months after the original transection. Nerve stumps were trimmed back to “healthy” nerve, as in clinical practice, and 1cm gap, secondary nerve repairs were performed using either freshly harvested, reversed 1cm sciatic nerve isografts, or PHB conduits. Nerve grafts were microsurgically inset using interrupted 10/0 epineurial sutures, while conduit repairs used paired horizontal mattress sutures to inset the nerve stumps 2mm into either end of the conduit, thereby leaving a 1cm gap across which the nerve had to regenerate.

Drug Administration

Acetyl-L-carnitine (ALCAR): ALCAR was administered by intraperitoneal injection to a total “high-dose” of 50mg/kg/day, or a total “low-dose” of 10mg/kg/day (Paper II). The first dose was administered prior to recovery of the animal after sciatic nerve transection, and treatment continued until termination either 2 weeks, or 2 months later. A separate group received “sham treatment” by injection of identical volumes of carrier solution (sterile normal saline). The effect of ALCAR treatment upon regeneration after secondary repair (Paper III) was determined by commencing high-dose (50mg/kg/day) treatment immediately after nerve repair, and continuing it until termination 6 weeks later. N-Acetyl-cysteine (NAC): NAC was initially administered by intravenous injection, and then subsequently by intraperitoneal injection to a total “high-dose” of 150mg/kg/day, or a total “low-dose” of 30mg/kg/day Paper (IV). The first dose was administered prior to recovery of the animal after sciatic nerve transection, and treatment continued until termination. A separate group received “sham treatment” by injection of identical volumes of carrier solution (sterile 5% dextrose solution). Tissue Processing

Perfusion Fixation: Dorsal root ganglia were harvested (Papers I, II, IV) after transcardiac perfusion fixation with paraformaldehyde (PFA). After appropriate survival times (Paper I: 1, 4 days, 1, 2 weeks, 1, 2, 4, 6 months; Papers II & IV: 2 weeks, 2 months) animals were anaesthetised with a terminal dose of sodium pentobarbitone. Transcardiac perfusion with chilled phosphate buffered saline, followed by 4% PFA was then completed. Dorsal Root Ganglia: The L4 & L5 DRG’s from both sides (Papers I, II, IV) were harvested after dorsal laminectomy, post-fixed in 4% PFA at 4oC, equilibrated in 15% sucrose & 0.1% sodium azide in PBS. Frozen blocks were then prepared, and each entire ganglion cut into serial 30µm cryosections mounted onto Vectabond® (Vector, U.K.) coated glass slides. These were then permeabilised by microwave irradiation, and stained by a novel triple staining technique that combined fluorescence TdT uptake nick-end labelling (TUNEL, Promega Apoptosis Detection System Fluorescein), the non-specific nuclear stain Hoechst 33342 (H33342), which allows excellent determination of nuclear morphology 163,220,253, and Propidium Iodide (PI). Although PI exclusion by an intact plasmalemma is used in vitro to demonstrate cell viability, in permeabilised fixed frozen sections the neuronal cytoplasm is exposed, giving PI access to demonstrate cytoplasmic morphology by

21 staining of the intracellular organelles. Using fluorescence microscopy and a range of band specific filters each stain could be visualised separately (TUNEL & morphological assessments), or in combination (optical disection). Peripheral Nerve Repairs: The nerve grafts / conduits (Papers III, V, VI) were harvested in continuity with ~3-4mm of the sciatic nerve, both proximally and distally to the repair, and were post-fixed in Zamboni’s solution. After rinsing, and equilibration into cryoprotectant solution (15% sucrose & 0.1% sodium azide in PBS), frozen blocks were prepared, and a systematic random sample of 15µm cryosections collected onto Vectabond® coated glass slides. These were stained by dual- fluorescence immunohistochemistry using primary antisera against pan-axonal marker of neurofilaments (monoclonal antibody against PAMNF, Affiniti Research Products Ltd.) for nerve fibres, and the ubiquitous Schwann cell marker S-100 (polyclonal antibody, DAKO, Denmark). Light Microscopic Analysis

Morphology: The morphological appearance of cells within DRG’s, and that of regenerating nerve fibres was determined by examination under fluorescence microscopy using a combination of 10X, 20X, and 40X objective lenses. TUNEL: Every DRG section was examined for TUNEL positive cells (Papers I, II, IV), whose type was confirmed by examination of the nuclear and cytoplasmic morphology under a 40X objective lens. False-positive staining was excluded by examination of nuclear morphology. Few positive neurons were present, and the count was therefore expressed as a total number per experimental group, rather than as a mean value per ganglion, as was done for glial cells. The number of TUNEL positive neurons gives a reflection of the number that are actively beginning the cell death process at the time the ganglion was harvested. Hence this measure sensitively detects the onset and rate of cell death, rather than its total magnitude which was determined stereologically. Optical Disection: Stereological techniques (Papers I, II, IV) require strict adherence to a set of systematic random sampling, and counting rules in order to give reliable, statistically non-biased estimates 203,204,209,214. As detailed in the methodology section of Paper I in particular, these strictures were adopted in full. The C.A.S.T.-Grid™ computer system was used for optical disection. This is essentially a high resolution fluorescence microscope with video output, whose slide stage is under the motor control of the C.A.S.T.-Grid software system for horizontal movements. A stage-mounted microcator allows the software to keep track of

22 displacements in the vertical axis during focusing movements. Hence the software can measure areas and volumes within thick tissue sections, and control the systematic random placement of a number (Ndis) of sampling boxes (whose volume, Vdis, is known) within those sections. Counting rules are applied within these boxes, to give the number of neurons present (Npar). Hence the density of neurons is calculated, and multiplied by 209 the volume of the ganglion (Vref , determined using the Cavalieri Principle ) to give the number of neurons using the following equation:

Neuron Count = ( Npar / [Ndis .Vdis]) . Vref The number of neurons present in the axotomised ganglia were subtracted from the number present in the contralateral, non-axotomised ganglia to give an estimation of the total number of neurons (“neuron loss”) that had died as a result of axotomy. In Paper I it was found that variance was less when considering the number of neurons in the L4 + L5 ganglia, rather than in individual L4 or L5 ganglia, hence the combined figure was used to determine the effect of treatment in Papers II & IV. Regeneration Distance: This was determined by examining every stained section from each nerve repair under fluorescence microscopy, using a 10X objective lens, and a 0.1mm graticulated eyepiece (Papers II, V, VI). Regeneration distance was measured from the end of the proximal nerve stump (marked during blocking with a piece of rat liver), to the tip of the most distal nerve fibre. This gave five results per experimental group, the mean (SD) of which were used for statistical comparison. Area of Immunostaining: In every repair examined, the two non-consecutive sections that lay most central to the regeneration front were selected. A digital image of the full width of the nerve graft / conduit was then captured, and the area of immunostaining quantified by image analysis (Image-Pro-Plus™ Version 4.0 software); a process reliant upon intensity thresholding for unichromatic light, deselection of occasional areas of background staining, and digital quantification of the area stained. Comparison between grafts, or between conduits was best facilitated by consideration of the percentage area of staining, since this serves to minimise artefactual differences in the total area of staining that arise when longitudinal sections are cut from slightly different positions across the diameter of a tube. However the total area is more appropriate when comparing nerve grafts and conduits, since conduits are of a much greater diameter than grafts, and the pattern of fibre growth is initially more diffuse within a conduit than a graft. Hence, at the timepoint used in these studies (6 weeks after repair), the percentage area of staining within a conduit will almost inevitably be artefactually lower than in a graft, making the total area of staining a more appropriate measure for comparison.

Statistical Analysis

In brief, comparisons between multiple groups were performed using analysis of variance. Pairwise comparisons were performed using either Tukey’s test, Student’s t-test, or the Mann-Whiney Rank Sum test where data distribution was not normal. In all papers statistical analysis was performed with the aid of SigmaStat™ (Version 2.0) software.

RESULTS

Paper I

This study described the timecourse and light microscopic morphology of neuronal and glial cell death after a peripheral nerve injury, and examined, in principle, whether the timing of nerve repair affects the extent of neuronal death that occurs. At each timepoint neuronal death was detected using a combination of TUNEL, to detect the onset of cell death and thereby imply its rate, and quantification of neuronal loss using the optical disector technique. The morphological findings were in keeping with an active cell death process, and many “apoptotic” features were present. However grossly dilated and vacuolated neurons were also found, which may represent an alternative morphological form of active cell death. No evidence of neuronal death was found in non-axotomised ganglia, and the number of neurons present (~30,000 per L4&L5 ganglion pair) was in agreement with other studies in the literature. TUNEL positive neurons were present within 24 hours of sciatic nerve transection, and numbers peaked 2 weeks after axotomy. Neuronal loss was not detectable until 1 week post-axotomy (14.8% loss, P<0.05), and continued to rise thereafter until reaching a plateau value at ~ 2months post-axotomy. Peak neuronal loss was 39.2% of all L4&L5 primary sensory neurons, although the percentage loss of neurons that were actually axotomised by the experimental lesion was estimated as being closer to 70%. Neuronal death was accompanied by a loss of DRG volume, and by delayed glial cell death, which reached a peak 2 months after axotomy. Evidence of axotomy induced death of contralateral glial cells was also found. The timing of nerve repair was found to affect sensory neuronal death. Immediate nerve repair was more neuroprotective (4.8% loss; TUNEL count 11 cells/group) than repair after a 1 week delay (12.4% loss; TUNEL count 14 cells/group), but neither eliminated neuronal death. Repair was protective when compared to no repair (21.3% loss; TUNEL count 25 cells/group). Papers II & IV

In these papers the effect of systemic treatment with two potentially neuroprotective agents (ALCAR & NAC) were investigated. Two durations of treatment were examined, based upon the timecourse of events detailed in Paper I. Firstly a 2 week timepoint was used since this is the peak rate of neuronal death, and

25 secondly a 2 month timepoint was used since neuronal death is essentially complete by this time after axotomy. Both ALCAR and NAC treatment caused a dose dependent improvement in neuronal morphology 2 weeks after axotomy. The number of TUNEL positive neurons was not affected by sham treatment (no treatment 25 neurons/group; sham saline treatment 33 neurons/group, P=0.937; sham dextrose treatment 19 neurons/group, P=0.405). ALCAR significantly reduced the number of TUNEL positive neurons, as did NAC (low-dose ALCAR 6 neurons/group, P=0.009; high-dose ALCAR 3 neurons/group, P=0.004; low-dose NAC 7 neurons/group, P=0.015; high-dose NAC 2 neurons/group, P=0.002). Sham treatment did not affect neuronal loss (saline 21% loss, P=0.99 cf. no treatment; 5% dextrose 21% loss, P=0.96), unlike ALCAR (low-dose 0% loss, P=0.001; high-dose 2% loss, P=0.006), or NAC (low-dose 1% loss, P=0.019; high-dose 3% loss, P=0.031) which both significantly reduced neuron loss. This neuroprotective effect was preserved 2 months after axotomy (no treatment 35% loss; high-dose ALCAR -4% loss, P<0.001; high-dose NAC -3% loss, P=0.002;). Paper III

In a model of late secondary nerve repair, where the neuroprotective effect of ALCAR is obviated, ALCAR treatment (50mg/kg/day) significantly improved nerve regeneration through reversed 1cm isografts. No difference in regeneration distance could be demonstrated since fibres reached the cut distal end of the nerve sample in both ALCAR treated, and untreated animals. Regeneration was morphologically enhanced however, and quantification of staining for nerve fibres (PAMNF antisera) revealed that ALCAR treatment caused a significant increase in the percentage area of immunostaining for nerve fibres. This was the case within the nerve grafts (no treatment 7.3%, ALCAR 24%; P<0.05) and in the distal nerve stump (no treatment 0.38%%, ALCAR 12%; P<0.05). A comparable effect upon area of staining for Schwann cells (S-100) was demonstrated. Papers V & VI

In these papers the aim was to determine the effect upon regeneration after late secondary nerve repair of locally delivered agents that might either restore the regenerative potential of chronically axotomised neurons (rhLIF), boost proliferation and neurotrophic factor secretion by the endogenous Schwann cell population (rhGGF), or augment that population with GGF responsive Schwann cells (cultured autologous

SC’s). Administration was achieved by incorporation into PHB conduits used for 1cm gap repairs, and comparison was made against plain conduits, and against reversed 1cm isografts. Plain PHB conduits supported some regeneration, though significantly less than nerve grafts did. It was also clear that regeneration through plain conduits worsened when the delay to repair was increased. Regeneration was enhanced by rhLIF which had more effect upon regeneration distance (2 month repair 69% increase cf. plain conduits, P=0.019; 4 month repair 123% increase, P=0.021) and morphology than upon the area of staining for nerve fibres, or S-100. It was also apparent that rhLIF was relatively more effective at boosting regeneration when repair was delayed by 4 months rather than two. The addition of SC’s to the conduits improved their regenerative profile, and the effect was most noticeable upon the regeneration distance (2 month repair 80% increase cf. plain conduits, P=0.007; 4 month repair 150% increase, P<0.001). Although SC’s had relatively little effect upon the area of staining for nerve fibres when repair was delayed by 2 months, they caused a significant increase when it was delayed by 4 months (147% increase in percentage area), as they did upon the area of staining for S-100. When repair was delayed by 2 months, rhGGF-2 caused a dramatic increase in the percentage area of staining for nerve fibres (370% increase, P=0.01), and a smaller, but significant increase when repair was performed 4 months after injury. A similar pattern was seen for S-100 immunostaining. Thus it appeared that although SC’s and rhGGF-2 both had beneficial effects upon regeneration, the effect of rhGGf-2 was somewhat greater in magnitude. However as the delay to repair increased, the effect of rhGGF-2 became relatively less, whilst that of SC’s became relatively more. Hence the combination of rhGGF-2 & SC’s was investigated in repairs performed 4 months after injury. As was hoped, the combination approach gave a better regenerative profile than either one alone. Conduits containing both rhGGF-2 & SC’s exhibited the greatest area of staining for nerve fibres, and the longest regeneration distance. When compared to nerve graft repairs, plain conduits were inadequate. Conduits containing rhLIF had shorter, though comparable, regeneration distances with less immunostaining. Those containing SC’s exhibited a somewhat better regenerative profile, but only really equalled nerve graft in one measure, that of regeneration distance in the group repaired 4 months after axotomy. In the group repaired 2 months after axotomy rhGGF-2 conduits contained around twice the total area of immunostaining for nerve fibres, and S-100, that nerve grafts did, but had marginally smaller areas when

27 repair was delayed by 4 months. Regeneration distances were comparable, though slightly less. However the conduits containing rhGGF-2 & SC’s contained a greater total area of staining for nerve fibres and S-100, a regeneration distance that was at least comparable, and a distal stump morphology that was an improvement over that found with nerve graft.

DISCUSSION

A clear need exists for therapies that enhance sensory restitution after peripheral nerve trauma. Microsurgical techniques allow accurate repair of the nerve trunk, or fascicles, but do not provide endoneurial realignment 254, and may not optimally address the neurobiological requirements of nerve regeneration 19,21. In particular, the death of a large proportion of sensory neurons must be prevented if sensory outcome is ever to approach the pre-morbid level. This study has focused upon delineating the salient features of that sensory neuronal loss, in order to develop therapeutic strategies that are realistically applicable within the clinical arena. Subsequently, certain of the neurobiological problems inherent to secondary nerve repair were addressed. Peripheral nerve injury induces significant sensory neuronal death, but a therapeutic window exists

Sciatic nerve transection induced the death of over a third of the entire sensory neuronal pool (Paper I). Similar results have been demonstrated by other authors 41,49,85,93,94,97,99,100, but the timecourse of neuronal death has not previously been elucidated in such detail. It would appear, from the TUNEL results that neuronal death does indeed begin within 24 hours of injury. The rate of neuronal death then continues to increase for 2 weeks thereafter, subsequently tailing off over the ensuing weeks, although occasional neurons are still dying even 6 months later. In keeping with the time taken to progress through the cell death process, and for a sufficient number of neurons to die to permit detection, loss of sensory neurons is not evident until 1 week after axotomy. Loss then continues to accrue, until reaching a plateau value around 2 months after injury. This implies that although neuronal death can be reduced by intervention during a therapeutic window lasting some weeks, the earlier that one intervenes after injury, the greater the potential neuroprotective benefit. Ideally neuroprotection should be delivered within 24 hours of peripheral nerve injury. The timing of nerve repair affects the magnitude of neuronal death

By restoring axonal contact with the neurotrophic factory that is the acutely denervated distal stump 21,41, it may be possible to provide physiological neurotrophic support for the axotomised sensory neurons, limiting the loss of trophic support, and preventing cell death. The results of such an experiment in Paper I support that contention.

Neuronal death, as assessed both by TUNEL and stereologically, was reduced by nerve repair in a time dependent manner. However even immediate repair (which is not clinically feasible) did not eliminate neuronal loss, and numerous TUNEL positive cells were found in the DRG’s of both repair groups 2 weeks after injury. Further neuronal loss is therefore likely to occur in the subsequent weeks. The inability of repair to eliminate neuronal loss may either reflect the fact that the distal stump does not maximise its production of NTF’s for many days after injury, or that NTF production is insufficient to fully meet the survival needs of all axotomised neurons. This data would suggest that nerve repair should be performed as soon as possible after injury, and not delayed for several days to let the distal environment for regeneration optimise, as has frequently been suggested in clinical practice. The need for an adjuvant, more completely neuroprotective treatment that could be commenced before surgery is also apparent. Neuronal death can be virtually eliminated by clinically applicable treatments

The results of Papers II & IV document, for the first time, that post-traumatic neuronal death can be virtually eliminated in vivo by systemic treatment with clinically safe pharmaceuticals. Using an identical experimental model to Paper I, and the biologically relevant timepoints of 2 weeks, and 2 months after axotomy, neuron loss was reduced from 21% & 35% respectively, to ~0%. Neuronal death was not entirely eliminated, since some TUNEL positive neurons were found, but must have been reduced to a sufficiently low rate that it was masked by normal biological variation in neuron counts. Since neuronal loss is essentially complete by 2 months after axotomy in the absence of treatment (Paper I), ALCAR and NAC can be said to provide permanent neuroprotection when treatment is perpetuated. Two key questions are now obvious. Can treatment subsequently be stopped without neuronal death recommencing, or must it be continued until repair has been performed, and target reinnervation is complete? Equally, does a clinically realistic delay between injury and starting treatment (i.e. the average taken to reach healthcare services) still permit adequate neuroprotection? Since Paper I demonstrated that neuronal death was only beginning 24 hours after injury, it is reasonable to assume that a delay in starting treatment of 6-12 hours may not be overly prejudicial. Further longterm studies will be required to address the first question. The mechanism of action was not investigated in these papers, but both agents potentially target aspects of mitochondrial bioenergetic dysfunction. NAC is neuroprotective against a range of insults in-vitro 148,149,255,256 and has some additional

30 inherent antioxidant and cell-cycle controlling properties 149,256, but is likely to exert much of its effect as a glutathione substrate 256-259. Glutathione is the principal renewable free radical scavenger within neurons 260, and also protects mitochondria from NO 142,261. ALCAR is a physiological peptide integral to high-energy substrate oxidative metabolism within mitochondria 262,263, that also has some inherent antioxidant properties 264, and ability to enhance the NGF binding capacity of neuronal populations 265-267. Both agents may therefore target the peri-mitochondrial events in the cell death pathway sufficiently to prevent involution, and early electron microscopical investigations in our laboratory suggest that mitochondrial morphology is indeed protected by treatment. It remains to be seen whether caspase activation is prevented, or whether the agents act by an alternative, and as yet unknown mechanism. If the mechanism is mitochondrial protection, then they may have applicability to a diverse range of neurodegenerative conditions from senile atrophy, to Alzheimer’s disease, and amyotrophic lateral sclerosis. Indeed during the course of this study independent workers have documented a role for ALCAR in treating aspects of senile degeneration, including memory loss 268. Although both agents have an established clinical safety profile, and were significantly neuroprotective, ALCAR was chosen for further investigation since more evidence of a role in nerve regeneration existed for it than for NAC. One study of ALCAR after primary repair exists 269, and suggests a favourable effect. ALCAR also reverses symptoms, and stimulates nerve regeneration in nucleoside analogue reverse transcriptase inhibitor (NRTI) associated distal symmetrical polyneuropathy (DSP) in HIV disease, a condition that also has a mitochondrial bioenergetic aetiology 270. Additionally, a convincing dose response effect upon sensory neuronal survival has now been demonstrated for ALCAR (unpublished data). Neuronal regenerative capacity can be augmented after secondary repair

Having found a way to potentially keep neurons alive, the study turned to ways to enhance regenerative capacity. As discussed previously, this depends upon both the regenerative capacity of individual neurons, and the environment through which the axons must regrow. Papers III & V principally addressed the regenerative capacity of the neurons, while Paper VI addressed the regenerative environment by modulation of the Schwann cell population. ALCAR promotes the transport of acyl moieties across the inner mitochondrial membrane, and by favouring maintenance of aerobic glycolytic pathways 262,263,271 in

31 preference to the normal shift into the less efficient pentose phosphate pathway 272 should increase neuronal metabolic capacity during regeneration. Its effect to enhance NGF binding by p75 may also enhance the neurotropic drive to axonal extension. In order to determine whether this could result in enhanced regeneration indepently of its neuroprotective effect, ALCAR treatment, and nerve repair was delayed until 2 months after axotomy. Nerve graft was chosen for the repair to mimic the current clinical situation, and a significant increase in regeneration was found, particularly when the area of staining for nerve fibres within the distal stump was evaluated (Paper III). The reduction in neuronal regenerative potential found after prolonged axotomy may be in part due to loss of the phenotype induced shortly after axotomy. This phenotype may be the result of retrogradely transported “injury factors”, of which LIF has received recent attention (Paper V). Targeted local delivery of LIF resulted in a significant increase in regeneration distance through a PHB conduit, and a less dramatic effect upon immunostaining for nerve fibres was also found. In keeping with the hypotheses that the regenerative phenotype is lost over a period of some months, and that LIF may restore that phenotype 195, the treatment effect was relatively greater when repair was performed four months after axotomy, rather than two. However LIF is not known to have any significant effect upon the Schwann cell population, which would explain why the regenerative profile of the LIF-conduits was not as good as nerve graft. If LIF were to reach clinical use, it should be as part of a nerve repair system that incorporates elements designed to enhance the Schwann cell response. Nevertheless, these experiments suggest that it is possible to augment the regenerative capacity of chronically axotomised neurons, and that these measures can be applied as part of a secondary nerve repair. A bioartificial nerve conduit to rival nerve graft for late secondary nerve repair?

The atrophic endogenous Schwann cell population was addressed in Paper VI. Schwann cells (SC’s) are recognised as essential to the support of peripheral nerve regeneration 21,196, and their comparatively limited migratory potential has been postulated as one reason why regeneration through long acellular conduits is so much worse than through long nerve grafts. Regeneration is impaired after secondary nerve repair, as compared to primary. Yet regeneration of freshly transected nerves into chronically denervated distal stumps is also impaired 184,273, largely due to the SC population being less able to respond to, and to support, ingrowing axons 65,189. Hence, even if one is able to provide adequate neuroprotection, and to restore the chronically

32 axotomised neurons’ regenerative potential, their growth into the chronically denervated distal stump will be still be impaired, unless the SC response can be augmented. Local administration of cultured autologous SC’s theoretically provides a fresh population of fully GGF responsive cells, which are able to interact with, and support regenerating axons. When suspended in alginate-fibronectin hydrogel and delivered within PHB conduits, SC’s significantly enhanced nerve regeneration when compared to plain conduits 238,274. Their effect was relatively greater when the delay to nerve repair was prolonged, and this would be in keeping with the hypothesis that the endogenous population were becoming increasing atrophic and unable to support axonal growth. However SC conduits did not match nerve grafts (NG’s), which are effectively another form of SC transplant albeit with a different matrix. This may suggest that cultured cells are relatively less active than those within NG’s, but no experimental data exist to expand this concept. Perhaps the concentration of SC’s, although optimal for primary repair was insufficient for secondary repair where the number of endogenous cells is less. The alginate hydrogel may have reduced SC motility, rendering them less able to proliferate and interact with axons; it is possible that future design enhancements to introduce porosity could address this potential problem. Longterm survival of SC’s in GGF free media is not good in vitro, and may have also been the case within the implanted conduits, although viable transplanted cells were still present at harvest, and similar problems should exist for the SC’s within nerve grafts. In reality it is likely that all these factors contribute to some extent, but do not detract from the demonstrable ability of cultured SC’s to enhance regeneration. An alternative approach to the use of cultured SC’s is to attempt to boost the endogenous population by administering exogenous GGF. This should cause SC proliferation, enhance NTF secretion, stimulate migration, and has improved regeneration after early repair 228. Although the endogenous SC’s will have virtually lost expression of c-erbB2&4 by 4 months after axotomy, and their GGF responsiveness might be questionable, a recent in vitro study has confirmed that they can be activated by supraphysiological concentrations of rhGGF 275. In this study (Paper VI) we demonstrate that rhGGF-2 within a bioartificial conduit does stimulate the endogenous Schwann cell population, although the effect was considerably less when repair was performed 4 months after axotomy, rather than 2 (sixfold increase in S-100 staining vs threefold increase, respectively). This Schwann cell activation was associated with a dramatic increase in nerve regeneration, such that 2 months after axotomy the rhGGF-2 conduits rivalled nerve grafts in their regenerative profile. Clearly the efficacy of

33 rhGGF-2 decreased when the duration of distal stump denervation increased, and this presumably reflects the reducing responsiveness of the distal stump Schwann cells. It would be interesting to investigate the effect of providing GGF to the distal stump during the period of denervation, in an attempt to prevent atrophy, and loss of c- erbB2&4 expression. In an attempt to circumvent the reduced GGF responsiveness of the endogenous Schwann cells, rhGGF-2 was combined with cultured SC’s in the final experimental group. This approach improves functional reinnervation after early nerve repair 228, but has not been investigated in the setting of secondary nerve repair, nor compared against nerve graft. As described in Paper VI, rhGGF-2+SC’s resulted in a major improvement in the regenerative profile of the conduits. Regeneration distance was greater than through conduits containing either rhGGF or SC’s alone, and the percentage area of staining for nerve fibres approximately doubled. Indeed the rhGGF- 2+SC conduits were at least comparable to, if not better than nerve grafts, the total area of staining for nerve fibres being greater within the conduits. Thus it was possible to create a bioartificial nerve graft that matched nerve graft in this experimental model of late secondary peripheral nerve repair, using components that are already either licensed for other clinical applications (PHB, alginate), or are potentially available for clinical trial. It is currently possible to extract autologous fibronectin from patients sera, and to culture human Schwann cells, whilst recombinant GGF-2 is already available. An intermediate step might be to continue to use nerve autograft, but to coat it with alginate containing rhGGF-2. In order to confirm the functional benefit of such a conduit it will first be necessary to undertake longer term animal studies of target organ reinnervation and electrophysiological function however. Can the prognosis of peripheral nerve injury be improved?

Currently less than ~30-60% of patients with peripheral nerve transections achieve a sensory outcome that is termed “good, or excellent” 8. However no patients return to normal sensation, and true functional restitution remains unachievable with current clinical practice. As discussed earlier, this is in large part due to loss of a large proportion of innervating sensory neurons, and the findings of Papers I, II & IV would suggest that neuronal loss can be addressed. Nerve repair should be performed as early as is clinically practicable in order to minimise cell death, and the potential for the future elimination of neuronal loss has been demonstrated. Since both agents are

34 already in clinical use for other applications, the progression to clinical trials could be rapid. Having kept the neurons alive, one must then make them regenerate better. Although this study did not directly address primary nerve repair, neuroprotection would be of benefit in this setting. RhGGF might also enhance the regenerative drive of the endogenous Schwann cells, and has subsequently been demonstrated to enhance their ability to penetrate longer conduits 276. ALCAR might also enhance the neurons’ regenerative capacity, as it did in the setting of secondary nerve repair (Paper III) just as LIF did (Paper V). Even without these measures, the results of Paper VI suggest that rhGGF-2 can dramatically enhance regeneration in the first months after axotomy, whilst cultured Schwann cells become an increasingly beneficial adjunct as the period of denervation increases. Lastly, by combining rhGGF-2 with SC’s, a bioartificial nerve conduit that rivals nerve graft for late secondary repair has finally been identified. However to derive the most benefit it would be wise to address neuroprotection, the Schwann cell population, and neuronal regenerative capacity simultaneously; this awaits further experimental investigation. Only then may regeneration of the peripheral nervous system truly be enhanced to the extent that we discover whether central plasticity will become the limiting factor, or the final link, in achieving sensory restitution after peripheral nerve trauma.

CONCLUSIONS

The results of the present thesis demonstrate that:

1) Peripheral nerve transection results in death of nearly 40% of the entire sensory neuronal pool. Death begins within 24 hours of axotomy, but does not translate into significant neuronal loss until 1 week post-axotomy. Death reaches a peak rate after 2 weeks, and then subsides, such that neuronal loss is essentially complete by 2 months after injury. This is accompanied by glial cell death and loss of volume in the dorsal root ganglia.

2) Peripheral nerve repair reduces the number of primary sensory neurons that die; the sooner nerve repair is performed the greater the neuroprotective benefit.

3) Primary sensory neuronal death after peripheral nerve transection can be virtually eliminated by systemic treatment with either N-acetyl-cysteine, or acetyl-L- carnitine.

4) The regenerative capacity of chronically axotomised neurons can be increased by systemic treatment with acetyl-L-carnitine, or by targeted local administration of leukaemia inhibitory factor.

5) The endogenous Schwann cell population can be induced to proliferate and improve nerve regeneration after secondary nerve repair, by targeted local administration of recombinant glial growth factor.

6) Cultured isogeneic Schwann cells enhance peripheral nerve regeneration after secondary nerve repair using a bioartificial nerve conduit. That effect is further augmented by the incorporation of recombinant glial growth factor.

These findings prepare the way for clinically applicable strategies to improve the outcome of peripheral nerve surgery by reducing sensory neuronal death, enhancing nerve regeneration, and ultimately developing a nerve conduit to replace nerve graft for gap repairs.

ACKNOWLEDGEMENTS

Science is a conspiracy of brains against ignorance Freeman Dyson

It is at the moment that they are working the least, that minds achieve the most. Leonardo da Vinci All work presented in this thesis was carried out in the Departments of Surgical & Perioperative Science (Section for Hand & Plastic Surgery), Integrative Medical Biology (Section for Anatomy), and Medical Biosciences (Pathology) at Umeå University, and in the Blond-McIndoe Laboratories, UCL, London, U.K. Without the flexibility, opportunities, and support offered by these centres nothing would have been possible. It has been my great pleasure to have such opportunities, and to have begun to work with a diverse group of colleagues, but to finish with a close group of friends. In particular I wish to express my gratitude to: Prof. Mikael Wiberg, my PhD supervisor, for introducing me to academia in Umeå University, and for rescuing my sanity during long spells of optical disection in Umea, and city life in the U.K. with many trips to the outdoor extravaganza that is Docksta. Thanks for giving up so much of your time and energy in supporting this work, and for your many introductions to the richness of Swedish life, although I’m still not sure about the sustrami. In London, Prof. Giorgio Terenghi has always provided insight, good-humoured encouragement and tempered restraint at appropriate times; I’m very grateful for your trust in letting me take the unusual step of running my project in two countries. Mikael and Giorgio’s complementary talents, trust, and friendship enabled me to work effectively, despite ferrying my work across the Baltic. Professor Jan-Olof Kellerth has provided invaluable advice at important times, and some stimulating discussions that never failed to restore my enthusiasm during some of the hardest times of the project. He and Prof. Wiberg have also brought together a research group that has never failed to offer me support, friendship, and excellent coffee. Lev and Luda Novikov, and Gunnel Folkesson for their coffee break conversations, interest in my progress, and friendship. Thanks also for undertaking the EM work, I never knew what an iceberg would lie under the tip of that initial request. Annette, and then Maria for your help when it was needed, and offers of help even when it wasn’t. Anna-Lena Tallander without whose efforts I would never have managed to arrange any of the defence, parties, or graduation. Izabella Blastyck for still speaking to me despite my inculcating her into stereology and laboratory work, and for introducing me to many things Polish.

Christina Ljungberg, and Johan, for welcoming me into your homes so many times with generous hospitality, and encouragement. Your friendship has been one of the lasting pleasures of my time in Umeå, and as a Scot, it was comforting to find uisge beatha held in such regard. Without Thomas Brannström, Department of Pathology, I would never have come to understand the perfection of stereology, and without his selfless technical help, occasional re-wiring and fire-fighting skills, I would never have completed my work. In London it was my great pleasure to share a “charmingly compact & bijou” office space (and a seemingly even more compact & bijou set of hard drives) with Stuart James, Ash Mosahebi, Pari Mohanna, Magda Simon, and Richard Young. They helped to show me how to live in the U.K.’s other capital city, and provided a mixture of enthusiasm, discussion, poor chat, good humour, and friendship that has helped in no small part to the completion of this research project. Thanks to Andy Wilson for picking up my batton and running with it, hope the Vitamin D levels get back up to scratch soon. Sorry I’m missing your wedding for this Stu, but I hope that there will still be enough skirt-wearing ginger-haired inebriates to make it authentically Scottish. To Anna, Johanna & Rebecca Wiberg for frequently accepting me into their home with such unfailing hospitality, for claiming to like summer in Scotland, and for getting Nicky started at skiing. To Canniesburn Hospital for guiding me in Plastic Surgery, particularly to Mr. Rob Ratcliffe for first showing me how not to “scratch around like a headless chicken” whilst operating, and to Mr. Arup Ray, Mr. David Soutar, and Mr. Rob Morris for saving me from neurosurgery and pointing me towards the Blond-McIndoe in the first place. Subsequently to Mr. Ian Taggart, Mr. John Scott, and particularly to Mr. Stuart Watson, for tolerance, intensive teaching and friendly psychotherapy during my return to clinical surgery, for really teaching me how to operate, and for giving me the microsurgery and trauma bug. I would also like to thank the Postgraduate Dean for bending some rules to let me finish my PhD. Lastly I wish to thank my parents, family and friends for their usual quiet encouragement, stability, tolerance and friendly slagging, albeit from what has become more of a distance than is ideal. I hope that I can spend more time with you now, and that I may yet live up to my dad’s enviable example of caring clinical practice, and clinically directed research, though he’d doubtless be mortified to hear that. Thanks most of all to my wife, Nicky, who never realised that Coire Mhic Fhearchair would be the easy bit. Despite becoming a “research widow” at many times during the course of this project, she has continued to care, and that tolerance gave me the drive to step back from work when it mattered, and so achieve most in the end.

This work was funded by the Swedish Medical Research Council (Grant 02286), County of Vasterbottens, Ane’rs Foundation, AstraTec AB (Sweden), European Community Grant A-1335-00, and the East Grinstead Medical Research Trust (Registered Charity No. 258154). I would also like to thank the Stanley-Thomas Johnson Foundation for their generous financial support, and the Medical & Odontological Faculty of Umeå University. The unqualified donation of anhydrous L- acetyl-carnitine by Sigma Tau Farmacia, Italy is also greatly appreciated as is the unqualified donation of supplies of N-acetyl-cysteine by the liver transplant group the Royal Free Hospital in London.

References

1. Seddon HJ. War injuries of the peripheral nerves in wounds of the extremities. British Journal of Surgery (War Surg Suppl) 1948; 2: -325. 2. Seddon HJ. Medical Research Council Special Report Series No. 282: Peripheral nerve injuries. London: HMSO, 1954. 3. Dellon AL. Evaluation of sensibility and re-education of sensation in the hand. Baltimore: Williams & Wilkins Company, 1981. 4. Bryan WJ, Schauder K, Tullos HS. The axillary nerve and its relationship to common sports medicine shoulder procedures. Am J Sports Med. 1986; 14: 113-116. 5. Katzman BM, Bozentka DJ. Peripheral nerve injuries secondary to missiles. Hand Clin. 1999; 15: 233-44, viii. 6. Birch R, Bonney G, Wynn Parry CB. Surgical Disorders of the Peripheral Nerves. Edinburgh: Churchill-Livingstone, 1998. 7. Hart AM, Wiberg M. Nerve Compression or mechanical neuropathy: neuropathology. Current Orthopaedics 2001; 15: 245-248. 8. Allan CH. Functional results of primary nerve repair. Hand Clin. 2000; 16: 67- 72. 9. Green B. Repair and grafting of the peripheral nerve. In: Green, ed. Plastic Surgery. 1998; 630-695. 10. Dagum AB. Peripheral nerve regeneration, repair, and grafting. J Hand Ther. 1998; 11: 111-117. 11. Trumble TE, McCallister WV. Repair of peripheral nerve defects in the upper extremity. Hand Clin. 2000; 16(1): 37-52. 12. Diao E, Vannuyen T. Techniques for primary nerve repair. Hand Clin. 2000; 16: 53-66. 13. Millesi H. Techniques for nerve grafting. Hand Clin. 2000; 16: 73-91. 14. Lundborg G. Nerve injury and repair. Churchill Livingstone, Edinburgh, 1988. 15. Doolabi VB, Hertl MC, Mackinnon SE. The role of nerve conduits in nerve repair: a review. Reviews in the Neurosciences 1996; 7: 47-84. 16. Millesi H. Nerve Grafting. Clin Plast Surg. 1964; 11: 105-113. 17. Terris DJ., Willard E.F. Current issues in nerve repair. Arch Otolarygol Head Neck Surg. 1993; 119: 725-731.

18. Terzis JK, Sun DD, Thanos PK. Historical and basic science review: past, present, and future of nerve repair. J Reconstr Microsurg. 1997; 13: 215-225. 19. Lundborg G. A 25-year perspective of peripheral nerve surgery: evolving neuroscientific concepts and clinical significance. J Hand Surg. 2000; 25A: 391-414. 20. Fu SY, Gordon T. The cellular and molecular basis of peripheral nerve regeneration. Mol Neurobiol. 1997; 14: 67-116. 21. Hall S. Nerve repair: a neurobiologists view. J Hand Surg. 2001; 26B: 129-136. 22. Lundborg G, Dahlin L, Danielsen N, Zhao Q. Trophism, tropism, and specificity in nerve regeneration. J Reconstr Microsurg. 1994; 10: 345-354. 23. Wiberg M, Hazari A, Ljungberg C, et al. Sensory recovery after hand replantation: a clinical, morphological and neurophysiological study in man. Scand J Plast Reconstr Surg Hand Surg 2002; 36: 1-11. 24. Glickman LT, Mackinnon SE. Sensory recovery following digital replantation. Microsurgery 1990; 11: 236-242. 25. de Medinaceli L, Prayon M, Merle M. Percentage of nerve injuries in which primary repair can be achieved by end-to-end approximation: review of 2,181 nerve lesions. Microsurgery 1993; 14: 244-246. 26. Vanderhooft E. Functional outcomes of nerve grafts for the upper and lower extremities. Hand Clin. 2000; 16: 93-104. 27. Lundborg G. Neurotropism, frozen muscle grafts and other conduits. J Hand Surg. 1991; 16B: 473-476. 28. Gattuso JM, Glasby MA, Gschmeissner SE, Norris RW. A comparison of immediate and delayed repair of peripheral nerves using freeze thawed autologous muscle grafts- in the rat. BJPS. 1989; 42: 306-313. 29. Mackinnon SE. New directions in peripheral nerve surgery. Ann Surg. 1989; 22: 272. 30. Lundborg G. Brain plasticity and hand surgery: an overview. J Hand Surg. 2000; 25B: 242-252. 31. Rosen B, Lundborg G, Dahlin LB, Holmberg J, Karlson B. Nerve repair: correlation of restitution of functional sensibility with specific cognitive capacities. J Hand Surg. 1994; 19B: 452-458.

32. Le TB, Aszmann O, Chen YG, Royall RM, Brushart TM. Effects of pathway and neuronal aging on the specificity of motor axon regeneration. Exp Neurol. 2001; 167: 126-132. 33. Westling G, Johansson RS. Factors influencing the force control during precision grip. Exp Brain Res. 1984; 53: 277-284. 34. Burstedt MKO, Schenker M, Wiberg M, Johansson R. Precision grip function in patients after hand replantation. 2003; manuscript. 35. Kuypers PDL, van Egeraat JM, Godschalk M, Hovius SER. Loss of Viable neuronal units in the proximal stump as possible cause for poor function recovery following nerve reconstructions. Exp Neurol. 1995; 132: 77- 81. 36. Liss AG. Peripheral nerve injury and its effects on the central nervous system. A study of transection of sensory nerves in cats and monkeys. Acta Universitatis Upsaliensis, 1995. 37. de ML, Merle M. Applying "cell surgery" to nerve repair: a preliminary report on the first ten human cases. J Hand Surg. 1991; 16B: 499-504. 38. Ekstrom PA, Hedlund G, Karlsson J, Andersson G. The immune modulator Linomide prevents neuronal death in injured peripheral nerves of the mouse. Neuroreport. 1998; 9: 1337-1341. 39. Ambron RT, Walters ET. Priming events and retrograde injury signals. A new perspective on the cellular and molecular biology of nerve regeneration. Mol Neurobiol. 1996; 13: 61-79. 40. Howard MJ, David G, Barrett JN. Resealing of transected myelinated mammalian axons in vivo: evidence for involvement of calpain. Neuroscience 1999; 93: 807-815. 41. Terenghi G. Peripheral nerve regeneration and neurotrophic factors. J Anat. 1999; 194: 1-14. 42. Roytta M, Salonen V. Long-term endoneurial changes after nerve transection. Acta Neuropathol 1988; 76: 35-45. 43. Gold BG, Spencer PS. Neurotrophic Function in Normal Nerve and in Peripheral Neuropathies. In: Gorio A., ed. Neuroregeneration. New York: Raven Press Ltd., 1993; 101-122. 44. Bowe CM, Hildebrand C, Kocsis JD, Waxman SG. Morphological and physiological properties of neurons after long-term axonal regeneration: observations on chronic and delayed sequelae of peripheral nerve injury. J Neurol Sci 1989; 91: 259-292.

45. Sterman AB, Delannoy MR. Cell body responses to axonal injury: traumatic axotomy versus toxic neuropathy. Exp Neurol. 1985; 89: 408-419. 46. Wells MR, Vaidya U. Morphological alterations in dorsal root ganglion neurons after peripheral axon injury: association with changes in metabolism. Exp Neurol. 1989; 104: 32-38. 47. Al-Abdulla NA, Martin LJ. Apoptosis of retrogradely degenerating neurons occurs in association with the accumulation of perikaryal mitochondria and oxidative damage to the nucleus. Am J Pathol. 1998; 153: 447-456. 48. Liss AG, af Ekenstam FW, Wiberg M. Changes in the spinal terminal pattern of the superficial radial nerve after a peripheral nerve injury. Scand J Plast Reconstr Surg Hand Surg. 1995; 29: 000-000. 49. Vestergaard S, Tandrup T, Jakobsen J. Effect of permanent axotomy on number and volume of dorsal root ganglion cell bodies. J Comp Neurol. 1997; 388: 307-312. 50. Groves MJ, Giometto B, Scaravilli F. Axotomy-induced vacuolation of primary sensory neurons and effect of administered neurotrophic factors: a morphometric, immunocytochemical and ultrastructural study. Prim Sensory Neuron 1997; 2: 111-127. 51. Ranson SW. Alterations in the spinal ganglion cells following neurotomy. J Comp Neurol Psychol 1909; 19: 125-153. 52. Sun Y, Zigmond RE. Leukaemia inhibitory factor induced in the sciatic nerve after axotomy is involved in the induction of galanin in sensory neurons. Eur J Neurosci. 1996; 8: 2213-2220. 53. Sebert ME, Shooter EM. Expression of mRNA for neurotrophic factors and their receptors in the rat dorsal root ganglion and sciatic nerve following nerve injury. J Neurosci Res. 1993; 36: 357-367. 54. Krekoski CA, Parhad IM, Clark AW. Attenuation and recovery of nerve growth factor receptor mRNA in dorsal root ganglion neurons following axotomy. J Neurosci Res. 1996; 43: 1-11. 55. Tonra JR, Curtis R, Wong V, et al. Axotomy upregulates the anterograde transport and expression of brain- derived neurotrophic factor by sensory neurons. J Neurosci. 1998; 18: 4374-4383. 56. Jiang YQ, Pickett J, Oblinger MM. Long-term effects of axotomy on beta-tubulin and NF gene expression in rat DRG neurons. J Neural Transplant Plast. 1994; 5: 103-114.

57. Noguchi K, Senba E, Morita Y, Sato M, Tohyama M. Prepro-VIP and preprotachykinin mRNAs in the rat dorsal root ganglion cells following peripheral axotomy. Brain Res Mol Brain Res. 1989; 6: 327-330. 58. Gillardon F, Klimaschewski L, Wickert H, et al. Expression pattern of candidate cell death effector proteins Bax, Bcl- 2, Bcl-X, and c-Jun in sensory and motor neurons following sciatic nerve transection in the rat. Brain Res. 1996; 739: 244-250. 59. Ham J, Babij C, Whitfield J, et al. A c-Jun dominant negative mutant protects sympathetic neurons against programmed cell death. Neuron 1995; 14: 927-939. 60. Kenney AM, Kocsis JD. Temporal variability of jun family transcription factor levels in peripherally or centrally transected adult rat dorsal root ganglia. Brain Res Mol Brain Res. 1997; 52: 53-61. 61. Woolf CJ, Reynolds ML, Molander C, O'Brien C, Lindsay RM, Benowitz LI. The growth-associated protein GAP-43 appears in dorsal root ganglion cells and in the dorsal horn of the rat spinal cord following peripheral nerve injury. Neuroscience 1990; 34: 465-478. 62. Sommervaille T, Reynolds ML, Woolf CJ. Time-dependent differences in the increase in GAP-43 expression in dorsal root ganglion cells after peripheral axotomy. Neuroscience 1991; 45: 213-220. 63. Seniuk NA. Neurotrophic factors: role in peripheral neuron survival and axonal repair. J Reconstr Microsurg 1992; 8: 399-404. 64. Sterne GD. Nt-3 modulates NPY expression in primary sensory neurons following peripheral nerve injury. J Anat. 1998; 193: 273-281. 65. Li H. Effects of delayed re-innervation on the expression of c-erbB receptors by chronically denervated rat Schwann cells in vivo. Glia 1997; 20: 333- 347. 66. Lee SE, Shen H, Taglialatela G, Chung JM, Chung K. Expression of nerve growth factor in the dorsal root ganglion after peripheral nerve injury. Brain Res. 1998; 796: 99-106. 67. Kashiba H, Hyon B, Senba E. Glial cell line-derived neurotrophic factor and nerve growth factor receptor mRNAs are expressed in distinct subgroups of dorsal root ganglion neurons and are differentially regulated by peripheral axotomy in the rat. Neurosci Lett. 1998; 252: 107-110.

68. Levi-Montalcini R, Dal TR, della VF, Skaper SD, Leon A. Update of the NGF saga. J Neurol Sci 1995; 130: 119-127. 69. Ji RR, Zhang Q, Zhang X, Piehl F, Reilly T, Pettersson RF, Hokfelt T. Prominent expression of bFGF in dorsal root ganglia after axotomy. Eur J Neurosci. 1995; 7: 2458-2468. 70. Reynolds ML, Woolf CJ. Reciprocal Schwann cell-axon interactions. Curr Opin Neurobiol. 1993; 3: 683-693. 71. Oblinger MM, Lasek RJ. Axotomy-induced alterations in the synthesis and transport of neurofilaments and microtubules in dorsal root ganglion cells. J Neurosci. 1988; 8: 1747-1758. 72. Inaishi Y, Kashihara Y, Sakaguchi M, Nawa H, Kuno M. Cooperative regulation of calcitonin gene-related peptide levels in rat sensory neurons via their central and peripheral processes. J Neurosci. 1992; 12: 518-524. 73. Dijkhuizen PA, Hermens WT, Teunis MA, Verhaagen J. Adenoviral vector- directed expression of neurotrophin-3 in rat dorsal root ganglion explants results in a robust neurite outgrowth response. J Neurobiol. 1997; 33: 172-184. 74. Xu ZQ, Zhang X, Grillner S, Hokfelt T. Electrophysiological studies on rat dorsal root ganglion neurons after peripheral axotomy: changes in responses to neuropeptides. Proc Natl Acad Sci U.S.A. 1997; 94: 13262-13266. 75. Czeh G, Kudo N, Kuno M. Membrane properties and conduction velocity in sensory neurones following central or peripheral axotomy. J Physiol (Lond.) 1977; 270: 165-180. 76. Yin Q, Kemp GJ, Frostick SP. Neurotrophins, neurones, and peripheral nerve regeneration. J Hand Surg 1998; 23B: 433-437. 77. Casaccia-Bonnefil P, Gu C, Chao MV. Neurotrophins in cell survival/death decisions. Adv Exp Med Biol. 1999; 468: 275-282. 78. Matheson CR, Carnahan J, Urich JL, et al. Glial cell line-derived neurotrophic factor (GDNF) is a neurotrophic factor for sensory neurons: comparison with the effects of the neurotrophins. J Neurobiol. 1997; 32: 22-32. 79. DiStefano PS, Currie A.R. Receptor mediated retrograde axonal transport of neurotrophic factors is increased after peripheral nerve injury. Prog Brain Res. 1994; 103: 35-42.

80. Hendry IA, Murphy M, Hilton DJ, Nicola NA, Bartlett PF. Binding and retrograde transport of leukemia inhibitory factor by the sensory nervous system. J Neurosci. 1992; 12: 3427-3434. 81. Thompson SW, Vernallis AB, Heath JK, Priestley JV. Leukaemia inhibitory factor is retrogradely transported by a distinct population of adult rat sensory neurons: co-localization with trkA and other neurochemical markers. Eur J Neurosci. 1997; 9: 1244-1251. 82. Curtis R, Scherer SS, Somogyi R, et al. Retrograde axonal transport of LIF is increased by peripheral nerve injury: correlation with increased LIF expression in distal nerve. Neuron 1994; 12: 191-204. 83. Curtis R, Adryan KM, Zhu Y, et al. Retrograde axonal transport of ciliary neurotrophic factor is increased by peripheral nerve injury. Nature 1993; 365: 253-255. 84. Cajal SRY. Degeneration & Regeneration of the Nerve Centres. In: May RM, ed. Degeneration & Regeneration of the Nervous System. 1.London: Oxford University Press, 1928; 397-463. 85. Ranson SW. Retrograde degeneration in the spinal nerves. J Comp Neurol. 1906; 16: 265-293. 86. Groves MJ, Christopherson T, Giometto B, Scaravilli F. Axotomy-induced apoptosis in adult rat primary sensory neurons. J Neurocytol. 1997; 26: 615-624. 87. Oliveira AL, Risling M, Deckner M, Lindholm T, Langone F, Cullheim S. Neonatal sciatic nerve transection induces TUNEL labeling of neurons in the rat spinal cord and DRG. Neuroreport. 1997; 8: 2837-2840. 88. Ekstrom PA. Neurones and glial cells of the mouse sciatic nerve undergo apoptosis after injury in vivo and in vitro. Neuroreport. 1995; 6: 1029- 1032. 89. Whiteside G, Doyle CA, Hunt SP, Munglani R. Differential time course of neuronal and glial apoptosis in neonatal rat dorsal root ganglia after sciatic nerve axotomy. Eur J Neurosci. (Suppl.) 1998; 10: 3400-3408. 90. Hou XE, Lundmark K, Dahlstrom AB. Cellular reactions to axotomy in rat superior cervical ganglia includes apoptotic cell death. J Neurocytol. 1998; 27: 441-451. 91. Rich KM, Luszczynski JR, Osborne PA, Johnson EMJ. Nerve growth factor protects adult sensory neurons from cell death and atrophy caused by nerve injury. J Neurocytol. 1987; 16: 261-268.

92. Cheema SS, Richards L, Murphy M, Bartlett PF. Leukemia inhibitory factor prevents the death of axotomised sensory neurons in the dorsal root ganglia of the neonatal rat. J Neurosci Res. 1994; 37: 213-218. 93. Rich KM, Disch S.P., Eichler ME. The influence of regeneration and nerve growth factor on the neuronal cell body reaction to injury. J Neurocytol. 1989; 18: 569-576. 94. Himes BT, Tessler A. Death of some dorsal root ganglion neurons and plasticity of others following sciatic nerve section in adult and neonatal rats. J.Comp.Neurol. 1989; 284: 215-230. 95. Ljungberg C, Novikov L, Kellerth J-O, Wiberg M. The neurotrophins NGF and NT-3 reduce sensory neuronal loss in adult rat after peripheral nerve lesion. Neurosci Lett. 1999; 262: 29-32. 96. Arvidsson J, Ygge J, Grant G. Cell loss in lumbar dorsal root ganglia and transganglionic degeneration after sciatic nerve resection in the rat. Brain Res. 1986; 373: 15-21. 97. Risling M, Aldskogius H, Hildebrand C, Remahl S. Effects of sciatic nerve resection on L7 spinal roots and dorsal root ganglia in adult cats. Exp Neurol. 1983; 82: 568-580. 98. Tessler A, Himes BT, Krieger NR, Murray M, Goldberger ME. Sciatic nerve transection produces death of dorsal root ganglion cells and reversible loss of substance P in spinal cord. Brain Res. 1985; 332: 209-218. 99. Liss AG, af Ekenstam FW, Wiberg M. Cell loss in sensory ganglia after peripheral nerve injury: an anatomical tracer study using lectin-coupled horse-radish peroxidase in cats. Scand J Plast Reconstr Surg Hand Surg. 1994; 28: 177-187. 100. Liss AG, af Ekenstam FW, Wiberg M. Loss of neurons in the dorsal root ganglia after transection of a peripheral sensory nerve: an anatomical study in monkeys. Scand J Plast Reconstr Surg Hand Surg. 1996; 30: 1-6. 101. Suzuki H, Oyanagi K, Takahashi H, Kono M, Yokoyama M, Ikuta F. A quantitative pathological investigation of the cervical cord, roots and ganglia after long-term amputation of the unilateral upper arm. Acta Neuropathol 1993; 85: 666-673. 102. Swett JE, Hong CZ, Miller PG. Most dorsal root ganglion neurons of the adult rat survive nerve crush injury. Somatosens Mot Res. 1995; 12: 177-189. 103. Peyronnard JM, Charron L, Lavoie J, Messier JP, Bergouignan FX. A comparative study of the effects of chronic axotomy, crush lesion and

re-anastomosis of the rat sural nerve on horseradish peroxidase labelling of primary sensory neurons. Brain Res. 1988; 443: 295-309. 104. Ygge J. Neuronal loss in lumbar dorsal root ganglia after proximal compared to distal sciatic nerve resection: a quantitative study in the rat. Brain Res. 1989; 478: 193-195. 105. Schmalbruch H. The number of neurons in dorsal root ganglia L4-L6 of the rat. Anat Rec. 1987; 219: 315-322. 106. Otto D, Unsicker K, Grothe C. Pharmacological effects of nerve growth factor and fibroblast growth factor applied to the transectioned sciatic nerve on neuron death in adult rat dorsal root ganglia. Neurosci Lett. 1987; 83: 156-160. 107. Wiberg M, Ljungberg C, O'Byrne A, Brown R, Whitworth I, Liss A, Terenghi G. Primary sensory neuron survival following targeted administration of NGF to an injured nerve. Scand J Plast Reconst Surg Hand Surg. 1999; 33: 387-392. 108. Bennett DL, Michael GJ, Ramachandran N, et al. A distinct subgroup of small DRG cells express GDNF receptor components and GDNF is protective for these neurons after nerve injury. J Neurosci. 1998; 18: 3059-3072. 109. Groves MJ, An SF, Giometto B, Scaravilli F. Inhibition of sensory neuron apoptosis and prevention of loss by NT-3 administration following axotomy. Exp Neurol. 1999; 155: 284-294. 110. Lo AC, Li L, Oppenheim RW, Prevette D, Houenou LJ. Ciliary neurotrophic factor promotes the survival of spinal sensory neurons following axotomy but not during the period of programmed cell death. Exp Neurol. 1995; 134: 49-55. 111. Edstrom A, Ekstrom PA, Tonge D. Axonal outgrowth and neuronal apoptosis in cultured adult mouse dorsal root ganglion preparations: effects of neurotrophins, of inhibition of neurotrophin actions and of prior axotomy. Neuroscience 1996; 75: 1165-1174. 112. Nunez G, del Peso L. Linking extracellular survival signals and the apoptotic machinery. Curr Opin Neurobiol. 1998; 8: 613-618. 113. Vogelbaum MA, Tong JX, Rich KM. Developmental regulation of apoptosis in dorsal root ganglion neurons. J Neurosci. 1998; 18: 8928-8935. 114. Estus S, Zaks WJ, Freeman RS, Gruda M, Bravo R, Johnson EM Jr. Altered gene expression in neurons during programmed cell death: identification of c-

jun as necessary for neuronal apoptosis. J Cell Biol. 1994; 127: 1717- 1727. 115. De Felipe C, Hunt SP. The differential control of c-jun expression in regenerating sensory neurons and their associated glial cells. J Neurosci. 1994; 14: 2911-2923. 116. Kenney AM, Kocsis JD. Peripheral axotomy induces long-term c-Jun amino- terminal kinase-1 activation and activator protein-1 binding activity by c-Jun and junD in adult rat dorsal root ganglia In vivo. J Neurosci. 1998; 18: 1318-1328. 117. Deckwerth TL, Easton RM, Knudson CM, Korsmeyer SJ, Johnson EMJ. Placement of the BCL2 family member BAX in the death pathway of sympathetic neurons activated by trophic factor deprivation. Exp Neurol. 1998; 152: 150-162. 118. Park DS, Stefanis L, Yan CYI, Farinelli SE, Greene LA. Ordering the cell death pathway. Differential effects of BCL2, an interleukin-1-converting enzyme family protease inhibitor, and other survival agents on JNK activation in serum/nerve growth factor-deprived PC12 cells. J Biol Chem. 1996; 271: 21898-21905. 119. Jenkins R, McMahon SB, Bond AB, Hunt SP. Expression of c-Jun as a response to dorsal root and peripheral nerve section in damaged and adjacent intact primary sensory neurons in the rat. Eur J Neurosci. 1993; 5: 751- 759. 120. Gagliardini V, Fernandez PA, Lee RK, Drexler HC, Rotello RJ, Fishman MC, Yuan J. Prevention of vertebrate neuronal death by the crmA gene [see comments] [published erratum appears in Science 1994 Jun 3;264(5164):1388]. Science 1994; 263: 826-828. 121. Davies AM. The Bcl-2 family of proteins, and the regulation of neuronal survival. Trends Neurosci. 1995; 18: 355-358. 122. Merry DE, Korsmeyer SJ. Bcl-2 gene family in the nervous system. Annu Rev Neurosci. 1997; 20:245-67: 245-267. 123. Middleton G, Nunez G, Davies AM. Bax promotes neuronal survival and antagonises the survival effects of neurotrophic factors. Development 1996; 122: 695-701. 124. Stewart B. Mechanisms of apoptosis: Integration of Genetic, Biochemical, and Cellular Indicators. Journal of the National Cancer Institute 1994; 86: 1286-1295.

125. Kruman II, Mattson MP. Pivotal role of mitochondrial calcium uptake in neural cell apoptosis and necrosis. J Neurochem. 1999; 72: 529-540. 126. Tong JX, Eichler ME, Rich KM. Intracellular calcium levels influence apoptosis in mature sensory neurons after trophic factor deprivation. Exp Neurol. 1996; 138: 45-52. 127. Figueirido HF. Skin-derived NGF modulates message levels of apoptosis-related genes in developing terminal ganglia. Society for Neuroscience 1998; 24. 128. Bergeron L, Yuan J. Sealing one's fate: control of cell death in neurons. Curr Opin Neurobiol. 1998; 8: 55-63. 129. Allsopp TE, Wyatt S, Paterson HF, Davies AM. The proto-oncogene bcl-2 can selectively rescue neurotrophic factor- dependent neurons from apoptosis. Cell 1993; 73: 295-307. 130. Batistatou A, Merry DE, Korsmeyer SJ, Greene LA. Bcl-2 affects survival but not neuronal differentiation of PC12 cells. J Neurosci. 1993; 13: 4422- 4428. 131. Borner C. Diminished cell proliferation associated with the death-protective activity of Bcl-2. J Biol Chem. 1996; 271: 12695-12698. 132. Hockenbery DM. bcl-2, a novel regulator of cell death. Bioessays 1995; 17: 631- 638. 133. Budd SL, Nicholls DG. Mitochondria in the life and death of neurons. Essays Biochem 1998; 33: 43-52. 134. Nicholls DG, Budd SL. Mitochondria and neuronal survival. Physiol Rev. 2000; 80: 315-360. 135. Beal MF. Energetics in the pathogenesis of neurodegenerative diseases. Trends Neurosci. 2000; 23: 298-304. 136. Martin LJ, Kaiser A, Price AC. Motor neuron degeneration after sciatic nerve avulsion in adult rat evolves with oxidative stress and is apoptosis. J Neurobiol 1999; 40: 185-201. 137. Novikov L, Novikova L, Kellerth J-O. Brain-derived neurotrophic factor promotes survival and blocks nitric oxide synthase expression in adult rat spinal motoneurons after ventral avulsion. Neurosci Lett. 1995; 200: 45-48.

138. He J, Hirata K, Kuraoka A, Kawabuchi M. An improved method for avulsion of lumbar nerve roots as an experimental model of nitric oxide-mediated neuronal degeneration. Brain Res Brain Res Protoc. 2000; 5: 223-230. 139. Brorson JR, Schumacker PT, Zhang H. Nitric oxide acutely inhibits neuronal energy production. The Committees on Neurobiology and Cell Physiology. J Neurosci 1999; 19: 147-158. 140. Heales SJ, Bolanos JP, Stewart VC, et al. Nitric oxide, mitochondria and neurological disease. Biochim Biophys Acta 1999; 1410: 215-228. 141. Estevez AG, Spear N, Manuel SM, et al. Nitric oxide and superoxide contribute to motor neuron apoptosis induced by trophic factor deprivation. J Neurosci 1998; 18: 923-931. 142. Moncada S. Nitric oxide and cell respiration: physiology and pathology. Verh K Acad Geneeskd Belg. 2000; 62: 171-179. 143. Keller JN, Kindy MS, Holtsberg FW, et al. Mitochondrial manganese superoxide dismutase prevents neural apoptosis and reduces ischemic brain injury: suppression of peroxynitrite production, lipid peroxidation, and mitochondrial dysfunction. J Neurosci 1998; 18: 687-697. 144. Castilho RF, Ward MW, Nicholls DG. Oxidative stress, mitochondrial function, and acute glutamate excitotoxicity in cultured cerebellar granule cells. J Neurochem. 1999; 72: 1394-1401. 145. Greenlund LJ, Deckwerth TL, Johnson EMJ. Superoxide dismutase delays neuronal apoptosis: a role for reactive oxygen species in programmed neuronal death. Neuron 1995; 14: 303-315. 146. Koliatsos VE, Ratan RR. Cell Death and Diseases of the Nervous System. Totowa, New Jersey: Humana Press, 1999. 147. Rosenfeld J, Cook S, James R. Expression of superoxide dismutase following axotomy. Exp Neurol. 1997; 147: 37-47. 148. Mayer M, Noble M. N-acetyl-L-cysteine is a pluripotent protector against cell death and enhancer of trophic factor-mediated cell survival in vitro. Proc Natl Acad Sci U.S.A. 1994; 91: 7496-7500. 149. Ferrari G, Yan CY, Greene LA. N-acetylcysteine (D- and L-stereoisomers) prevents apoptotic death of neuronal cells. J Neurosci. 1995; 15: 2857- 2866. 150. Freeman RS. The cell cycle and neuronal death. In: Koliatsos VE, Ratan RR, eds. Cell Death and Diseases of the Nervous System. Totowa, New Jersey: Humana Press Inc., 1999; 103-122.

151. ElShamy WM, Fridvall LK, Ernfors P. Growth arrest failure, G1 restriction point override, and S phase death of sensory precursor cells in the absence of neurotrophin-3. Neuron 1998; 21: 1003-1015. 152. Friedlander RM, Gagliardini V, Rotello RJ, Yuan J. Functional role of interleukin 1 beta (IL-1 beta) in IL-1 beta- converting enzyme-mediated apoptosis. J Exp Med. 1996; 184: 717-724. 153. Martin DP, Schmidt RE, DiStefano PS, et al. Inhibitors of protein synthesis and RNA synthesis prevent neuronal death caused by nerve growth factor deprivation. J Cell Biol. 1988; 106: 829-844. 154. Boonman Z, Isacson O. Apoptosis in neuronal development and transplantation: role of caspases and trophic factors. Exp Neurol. 1999; 156: 1-15. 155. Nicholson DW, Thornberry NA. Caspases: killer proteases. Trends Biochem Sci. 1997; 22: 299-306. 156. Bredesen D.E. Keeping neurons alive: the molecular control of apoptosis ( Part I). The Neuroscientist 1996; 2: 181-190. 157. Kouroku Y, Urase K, Fujita E, et al. Detection of activated Caspase-3 by a cleavage site-directed antiserum during naturally occurring DRG neurons apoptosis. Biochem Biophys Res Commun. 1998; 247: 780- 784. 158. Cregan SP, MacLaurin JG, Craig CG, et al. Bax-dependent caspase-3 activation is a key determinant in p53-induced apoptosis in neurons. J Neurosci 1999; 19: 7860-7869. 159. Krajewski S, Krajewska M, Ellerby LM, et al. Release of caspase-9 from mitochondria during neuronal apoptosis and cerebral ischemia. Proc Natl Acad Sci U.S.A. 1999; 96: 5752-5757. 160. Yakovlev AG, Knoblach SM, Fan L, et al. Activation of CPP32-like caspases contributes to neuronal apoptosis and neurological dysfunction after traumatic brain injury. J Neurosci. 1997; 17: 7415-7424. 161. Yuan J. Transducing signals of life and death. Curr Opin Cell Biol. 1997; 9: 247- 251. 162. Liu XZ, Xu XM, Hu R, et al. Neuronal and glial apoptosis after traumatic spinal cord injury. J Neurosci. 1997; 17: 5395-5406. 163. Crowe MJ, Bresnahan JC, Shuman SL, Masters JN, Beattie MS. Apoptosis and delayed degeneration after spinal cord in jury in rats and monkeys. Nature Medicine 1997; 3: 73-76.

164. Cohen S, Levi-Montalcini R. A nerve growth-stimulating factor isolated from snake venom. Proc Natl Acad Sci U.S.A. 1956; 42: 571-574. 165. Gold BG. Axonal regeneration of sensory nerves is delayed by continuous intrathecal infusion of nerve growth factor. Neuroscience 1997; 76: 1153-1158. 166. Novikova LN, Novikov LN, Kellerth J-O. Survival effects of BDNF and NT-3 on axotomized rubrospinal neurons depend on the temporal pattern of neurotrophin administration. Eur J Neurosci. 2000; 12: 776-780. 167. Novikova LN, Novikov LN, Kellerth J-O. BDNF abolishes the survival effect of NT-3 in axotomised Clarke's nucleus of adult rats. J Comp Neurol. 2000; 428: 671-680. 168. Soilu-Hanninen M, Ekert P, Bucci T, et al. Nerve growth factor signaling through p75 induces apoptosis in Schwann cells via a Bcl-2-independent pathway. J Neurosci. 1999; 19: 4828-4838. 169. McArthur JC, Yiannoutsos C, Simpson D, et al. A phase II trial of nerve growth factor for sensory neuropathy associated with HIV infection. Neurology 2000; 54: 1080-1088. 170. Metcalf D, Gearing DP. Fatal syndrome in mice engrafted with cells producing high levels of the leukemia inhibitory factor. Proc Natl Acad Sci U.S.A 1989; 86: 5948-5952. 171. Martin D, Merkel E, Tucker KK, et al. Cachectic effect of ciliary neurotrophic factor on innervated skeletal muscle. Am J Physiol. 1996; 271: R1422- R1428. 172. Eriksdotter JM, Nordberg A, Amberla K, et al. Intracerebroventricular infusion of nerve growth factor in three patients with Alzheimer's disease. Dement Geriatr Cogn Disord. 1998; 9: 246-257. 173. Gadaleta MN, Cormio A, Pesce V, Lezza AM, Cantatore P. Aging and mitochondria. Biochimie 1998; 80: 863-870. 174. Borthwick GM, Johnson MA, Ince PG, Shaw PJ, Turnbull DM. Mitochondrial enzyme activity in amyotrophic lateral sclerosis: implications for the role of mitochondria in neuronal cell death. Ann Neurol. 1999; 46: 787- 790. 175. Fiskum G, Murphy AN, Beal MF. Mitochondria in neurodegeneration: acute ischemia and chronic neurodegenerative diseases. J Cereb Blood Flow Metab. 1999; 19: 351-369.

176. Budd SL. Mechanisms of neuronal damage in brain hypoxia/ischemia: focus on the role of mitochondrial calcium accumulation. Pharmacol Ther. 1998; 80: 203-229. 177. Kong J, Xu Z. Massive mitochondrial degeneration in motor neurons triggers the onset of amyotrophic lateral sclerosis in mice expressing a mutant SOD1. J Neurosci 1998; 18: 3241-3250. 178. Wu Y, Li Y, Liu H, Wu W. Induction of nitric oxide synthase and motoneuron death in newborn and early postnatal rats following spinal root avulsion. Neurosci Lett. 1995; 194: 109-112. 179. Miyamoto Y., Higaki T., Sugita T., Ikuta Y., Tsuge K. Morphological reaction of cellular elements and the endoneurium following nerve section. Peripheral Nerve Repair & Regeneration 1986; 3: 7-18. 180. Griffin J.W., George R., Ho T. Macrophage systems in peripheral nerve repair. A review. J Neuropathol Exp Neurol 1993; 52: 553-560. 181. Hammarberg H, Piehl F, Cullheim S, et al. GDNF mRNA in Schwann cells and DRG satellite cells after chronic sciatic nerve injury. Neuroreport. 1996; 7: 857-860. 182. Curtis R. Growth-associated protein-43 (GAP-43) is expressed by glial cells of the central and peripheral nervous system. Ann NY Acad Sci. 1993; 679: 407-411. 183. Terenghi G, Calder JS, Birch R, Hall SM. A morphological study of Schwann cells and axonal regeneration in chronically transected human peripheral nerves. J Hand Surg 1998; 4B: 1-5. 184. Holmes W., Young J.Z. Nerve regeneration after immediate and delayed suture. J Anat. 1942; 77: 63-96. 185. Olawale ARS, Gordon T. Effects of short- and long-term Schwann cell denervation on peripheral nerve regeneration, myelination, and size. Glia 2000; 32: 234-246. 186. Frostick SP, Yin Q, Kemp GJ. Schwann cells, neurotrophic factors, and peripheral nerve regeneration. Microsurgery 1998; 18: 397-405. 187. Hall SM. The effect of inhibiting Schwann cell mitosis on the re-innervation of acellular autografts in the peripheral nervous system of the mouse. Neuropathol Appl Neurobiol. 1986; 12: 401-414. 188. Lemke GE, Brockes JP. Identification and purification of glial growth factor. J Neurosci. 1984; 4: 75-83.

189. Höke A, Gordon T, Zochodne DW, Sulaiman OAR. A decline in glial cell-line- derived neurotrophic factor expression is associated with impaired regeneration after long-term Schwann cell denervation. Exp Neurol. 2002; 173: 77-85. 190. Burgess P.R., English K.B., Horsch K.W. Patterning in the regeneration of cutaneous receptors. J Physiol. 1974; 236: 57. 191. Zelena J. Survival of Pacinian corpuscles after denervation in adult rats. Cell Tissue Res. 1982; 224: 673-683. 192. Tonge DA, Golding JP, Edbladh M, et al. Effects of extracellular matrix components on axonal outgrowth from peripheral nerves of adult animals in vitro. Exp Neurol. 1997; 146: 81-90. 193. Tonge DA, Aaronson OS, Golding JP, Jaggers D. Cellular migration and axonal outgrowth from adult mammalian peripheral nerves in vitro. J Neurobiol. 1996; 29: 151-164. 194. Lundborg G, Dahlin L, Danielsen N, Zhao Q. Trophism, tropism, and specificity in nerve regeneration. J Reconstr Microsurg. 1994; 10: 345-354. 195. Murphy M, Dutton R, Koblar S, Cheema S, Bartlett P. Cytokines which signal through the LIF receptor and their actions in the nervous system. Prog Neurobiol 1997; 52: 355-378. 196. Hall SM. The effect of inhibiting Schwann cell mitosis on the re-innervation of acellular autografts in the peripheral nervous system of the mouse. Neuropathol Appl Neurobiol. 1986; 12: 401-414. 197. Brushart TM, Gerber J, Kessens P, Chen YG, Royall RM. Contributions of pathway and neuron to preferential motor reinnervation. J Neurosci 1998; 18: 8674-8681. 198. Ljungberg C, Johansson-Ruden G, Bostrom KJ, Novikov L, Wiberg M. Neuronal survival using a resorbable synthetic conduit as an alternative to primary nerve repair. Microsurgery. 1999; 19: 259-264. 199. Dahlin L, Lundborg G. Use of tubes in peripheral nerve repair. Neurosurg Clin North Am. 2001; 12: 341-352. 200. Lundborg G, Rosen B, Dahlin LB, Holmberg J, Karlson B. Functional sensibility of the hand after nerve repair. Lancet 1993; 342: 1300. 201. Brushart TM, Mathur V, Sood R, Koschorke GM. Joseph H. Boyes Award. Dispersion of regenerating axons across enclosed neural gaps. J Hand Surg. 1995; 20A: 557-564.

202. Fu SY, Gordon T. Contributing factors to poor functional recovery after delayed nerve repair: prolonged axotomy. J Neurosci 1998; 15: 3876-3855. 203. West MJ. New stereological methods for counting neurons. Neurobiology of Aging 1993; 14: 275-285. 204. Coggleshall RE. A consideration of neural counting methods. TINS. 1992; 15: 9-13. 205. Coggeshall RE, Pover C.M., Fitxgerald M. Dorsal root ganglion cell death and surviving cell numbers in relation to the development of sensory innervation in the rat hindlimb. Dev Brain Res 1994; 82: 193-212. 206. Abercrombie M. Estimation of nuclear populations from microtome sections. Anat Rec. 1948; 94: 239-247. 207. Königsmark BW. Methods for the counting of neurons. In: Nauta WJ, Ebbesson SOE, eds. ed.Contemporary research methods for the counting of neurons. New York: Springer-Verlag, 1970; 315-340. 208. West MJ. Stereological methods for estimating the total number of neurons and synapses: issues of precision and bias. Trends Neurosci. 1999; 22: 51- 61. 209. Gundersen JG, Bendtsen TF, Korbo L, et al. Some new, simple and efficient stereological methods and their use in pathological research and diagnosis. APMIS 1988; 96: 379-394. 210. Pover C.M., Coggeshall RE. Verification of the disector method for counting neurons, with comments on the empirical method. Anat Rec 1991; 231: 573-578. 211. Tandrup T. A method for producing unbiased and efficient estimation of number and mean volume of specified neuron subtypes in rat dorsal root ganglion. J Comp Neurol. 1993; 329: 269-276. 212. Benes FM, Lange N. Two-dimensional versus three-dimensional cell counting: a practical perspective. Trends Neurosci. 2001; 24: 11-17. 213. Coggeshall RE, La FR, Klein CM. Calibration of methods for determining numbers of dorsal root ganglion cells [published erratum appears in J Neurosci Methods 1991 Dec;40(2- 3):87-90]. J Neurosci Methods 1990; 35: 187-194. 214. Harding A.J., Halliday G.M., Cullen K. Practical considerations for the use of the optical disector in estimating neuronal number. J Neurosci Methods 1994; 51: 83-89.

215. Gundersen HJG, Bagger P, Bendtsen TF, et al. The new stereological tools: Disector, fractionator, nucleator and point sampled intercepts and their use in pathological research and diagnosis. APMIS 1988; 96: 857-881. 216. Barres B.A., Jacobson M.D., Schmid R., Sendtner M., Raff MC. Does oligodendrocyte survival depend upon axons? Curr Biol. 1993; 3: 497. 217. Wyllie A.H. Glucocorticoid-induced thymocyte apoptosis is associated with endogenous endonuclease activition. Nature 2000; 284: 555. 218. Bilbao F.De., Dubois-Dauphin M. Time-course of axotomy-induced apoptotic cell death in facial motorneurons of neonatal wild type and bcl-2 transgenic mice. Neuroscience 1996; 71: 1111-1119. 219. Negoescu A, Lorimier P., Labat-Moleur F., et al. TUNEL: Improvement and Evaluation of the Method for in situ apoptotic cell identification. Biochemica 1997; 2: 12-17. 220. Whiteside G, Cougnon N, Hunt SP, Munglani R. An improved method for detection of apoptosis in tissue sections and cell culture, using the TUNEL technique combined with Hoechst stain. Brain Res Brain Res Protoc. 1998; 2: 160-164. 221. Charriaut-Marlangue C, Ben-Ari Y. A cautionary note on the use of the TUNEL stain to determine apoptosis. Neuroreport. 1995; 7: 61-64. 222. Whitworth I., Brown R.A., Dore C.J., Green C.J., Terenghi G. Oriented mats of fibronectin as a conduit material for use in peripheral nerve repair. J Hand Surg 1995; 20B: 429-436. 223. Fields RD, Le Beau J.M., Longo F.M., Ellisman M.H. Nerve regeneration through artificial tubular implants. Neurobiology 1989; 33: 87-134. 224. Meek MF, Coert JH. Clinical use of nerve conduits in peripheral nerve repair. J Reconstr Microsurg. 2002; 18: 97-109. 225. Weber RA, Dellon AL. Synthetic nerve conduits: indications and technique. Sem Neurosurg. 2001; 12: 65-79. 226. Lundborg G., Rosen B., Dahlin L., et al. Tubular versus conventional repair of median nerve in the human forearm: early results from a prospective randomised clinical study. J Hand Surg. 1997; 22A: 99-106. 227. Bellmkonda R.V., Valentini R.F. Fabrication and Implantation of Gel-Filled nerve guidance channels. In: Morgan J.R., Yarmush M.L., eds. ed.Methods in Molecular Medicine, Volume 18: Tissue Engineering Methods and Protocols. Humana Press Inc., 1998; 101-119.

228. Bryan DJ, Holway AH, Wang KK, et al. Influence of glial growth factor and Schwann cells in a bioresorbable guidance channel on peripheral nerve regeneration. Tissue Eng 2000; 6: 129-138. 229. Evans GRD, Brandt K, Katz S, et al. Bioactive pol(L-lactic acid) conduits seeded with Schwann cells for peripheral nerve regeneration. Biomaterials 2002; 23: 841-848. 230. Evans GRD. Challenges to nerve regeneration. Sem Surg Oncol. 2000; 2000: 312-318. 231. Hadlock T, Sundback C, Koka R, et al. A novel, biodegradable polymer conduit delivers neurotrophins and promotes nerve regeneration. Laryngoscope 1999; 109: 1412-1416. 232. Hazari A, Wiberg M, Johansson-Ruden G, Green C, Terenghi G. A resorbable nerve conduit as an alternative to nerve autograft in nerve gap repair. BJPS. 1999; 52: 653-657. 233. Evans GRD, Brandt K, Lu L, et al. In vivo evaluation of poly(L-lactic acid) porous conduits for peripheral nerve regeneration. Biomaterials 1999; 20: 1109-1115. 234. Sufan W, Suzuki Y, Tanihara M, et al. Sciatic nerve regeneration through alginate with tubulation or nontubulation repair in cat. J Neurotrauma 2001; 18: 329-338. 235. Zhang F, Blain B, Beck J, Zhang J, Chen Z, Chen Z-W, Lineweaver WC. Autogenous venous graft with one-stage prepared Schwann cells as a conduit for repair of long segmental defects. J Reconstr Microsurg. 2002; 18: 295-2002. 236. Hazari A, Johansson-Ruden G, Junemo-Bostrom K, Ljungberg C, Terenghi G, Green C, Wiberg M. A new resorbable wrap-around implant as an alternative nerve repair technique. J Hand Surg. 1999; 24B: 291-295. 237. Young RC, Wiberg M, Terenghi G. Poly-3-hydroxybutyrate (PHB): a resorbable conduit for long gap repair in peripheral nerves. BJPS. 2002; 55: 235- 240. 238. Mosahebi A, Wiberg M, Terenghi G. Addition of fibronectin to alginate matrix improves peripheral nerve regeneration in tissue engineered conduits. Tissue Eng. 2002; 239. Mosahebi A, Simon M, Wiberg M, Terenghi G. A novel use of alginate hydrogel as Schwann cell matrix. Manuscript.

240. Anderson AJ, Dawes EA. Occurrence, metabolism, metabolic role, and industrial uses of bacterial polyhydroxyalkanoates. Microbiol Rev. 1990; 54: 450- 472. 241. Holmes P. Biologically produced (R)-3-hydroxy-alkanoate polymers and copolymers. In: Bessett DC, ed. Developments in crystalline polymers. London: Elsevier, 1998; 1-65. 242. Doyle C, Tanner ET, Bonfield W. In vitro and in vivo evaluation of polyhydroxybutyrate and of polyhydroxybutyrate reinforced with hydroxyapatite. Biomaterials 1991; 12: 841-847. 243. Malm T, Bowald S, Bylock A, Busch C. Prevention of postoperative pericardial adhesions by closure of the pericardium with absorbable polymer patches. An experimental study. J Thorac Cardiovasc Surg. 1992; 104: 600-607. 244. Malm T, Bowald S, Bylock A, Busch C, Saldeen T. Enlargement of the right ventricular outflow tract and the pulmonary artery with a new biodegradable patch in transannular position. Eur Surg Res 1994; 26: 298-308. 245. Malm T, Bowald S, Bylock A, Saldeen T, Busch C. Regeneration of pericardial tissue on absorbable polymer patches implanted into the pericardial sac. An immunohistochemical, ultrastructural and biochemical study in the sheep. Scand J Thorac Cardiovasc Surg. 1992; 26: 15-21. 246. Malm T, Bowald S, Karacagil S, Bylock A, Busch C. A new biodegradable patch for closure of atrial septal defect. An experimental study. Scand J Thorac Cardiovasc Surg. 1992; 26: 9-14. 247. Gogolewski S, Jovanovic M, Perren SM, Dillon JG, Hughes MK. Tissue response and in vivo degradation of selected polyhydroxyacids: polylactides (PLA), poly(3-hydroxybutyrate) (PHB), and poly(3- hydroxybutyrate-co-3-hydroxyvalerate) (PHB/VA). J Biomed Mater Res 1993; 27: 1135-1148. 248. Hazari A, Wiberg M, Johansson-Ruden G, Green C, Terenghi G. A resorbable nerve conduit as an alternative to nerve autograft in nerve gap repair. BJPS. 1999; 52: 653-657. 249. Brockes JP, Fields KL, Raff MC. Studies on cultured rat Schwann cells. I. Establishment of purified populations from cultures of peripheral nerve. Brain Res. 1979; 165: 105-118.

250. Ferry N, Duplessis O, Houssin D, Danos O, Heard JM. Retroviral-mediated gene transfer into hepatocytes in vivo. Proc.Natl.Acad.Sci U.S.A. 1991; 88: 8377-8381. 251. Kotani H, Newton PB, Zhang S, et al. Improved methods of retroviral vector transduction and production for gene therapy. Hum Gene Ther. 1994; 5: 19-28. 252. Tham SK, Dowsing B., Finkelstein D., et al. Leukaemia Inhibitory Factor enhances the regeneration of transected rat sciatic nerve and the function of reinnervated muscle. J Neurosci Res. 1997; 47: 208-215. 253. Whiteside G, Munglani R. TUNEL, hoechst and immunohistochemistry triple- labelling: an improved method for detection of apoptosis in tissue sections-an update. Brain Res Brain Res.Protoc. 1998; 3: 52-53. 254. de Medinaceli L, Seaber AV. Experimental nerve reconnection: importance of initial repair. Microsurgery 1989; 10: 56-70. 255. Seaton TA, Cooper JM, Schapira AH. Free radical scavengers protect dopaminergic cell lines from apoptosis induced by complex I inhibitors. Brain Res. 1997; 777: 110-118. 256. Yan CY, Ferrari G, Greene LA. N-acetylcysteine-promoted survival of PC12 cells is glutathione- independent but transcription-dependent. J Biol Chem. 1995; 270: 26827-26832. 257. Wagner R, Heckman HM, Myers RR. Wallerian degeneration and hyperalgesia after peripheral nerve injury are glutathione-dependent. Pain 1998; 77: 173-179. 258. Dringen R, Hamprecht B. N-acetylcysteine, but not methionine or 2- oxothiazolidine-4- carboxylate, serves as cysteine donor for the synthesis of glutathione in cultured neurons derived from embryonal rat brain. Neurosci Lett 1999; 259: 79-82. 259. Dringen R, Pfeiffer B, Hamprecht B. Synthesis of the antioxidant glutathione in neurons: supply by astrocytes of CysGly as precursor for neuronal glutathione. J Neurosci 1999; 19: 562-569. 260. Cooper AJ, Kristal BS. Multiple roles of glutathione in the central nervous system. Biol Chem. 1997; 378: 793-802. 261. Lizasoain I, Moro MA, Knowles RG, Darley-Usmar V, Moncada S. Nitric oxide and peroxynitrite exert distinct effects on mitochondrial respiration which are differentially blocked by glutathione or glucose. Biochem J. 1996; 314 (Pt 3): 877-880.

262. Colucci WJ, Gandour RD. Carnitine acyltransferase: a review of its biology, enzymology and bioorganic chemistry. Bioorg Chem. 1988; 16: 307- 334. 263. Bremer J. The role of carnitine in intracellular metabolism. J Clin.Chem Clin Biochem. 1990; 28: 297-301. 264. Tesco G, Latorraca S, Piersanti P, et al. Protection from oxygen radical damage in human diploid fibroblasts by acetyl-L-carnitine. Dementia 1992; 3: 58-60. 265. Manfridi A, Forloni GL, Arrigoni-Martelli E, Mancia M. Culture of dorsal root ganglion neurons from aged rats: effects of acetyl-L-carnitine and NGF. Int J Dev Neurosci. 1992; 10: 321-329. 266. Taglialatela G, Angelucci L., Ramacci M.T., et al. Acetyl-L-carnitine enhances the response of PC12 cells to nerve growth factor. Brain Res Dev Brain Res. 1991; 59: 221-230. 267. Angelucci L, Ramacci MT, Taglialatela G, et al. Nerve growth factor binding in aged rat central nervous system: effect of acetyl-L-carnitine. J Neurosci Res. 1988; 20: 491-496. 268. Liu L, Head E, Gharib AM, et al. Memory loss in old rats is associated with brain mitochondrial decay and RNA/DNA oxidation: partial reversal by feeding acetyl-L-carnitine and/or/ R-α-lipoic acid. Proc Natl Acad Sci USA. 2002; 99: 2356-2361. 269. Fernandez E, Pallini PR, Gangitano C, et al. Studies on the degenerative and regenerative phenomena occurring after transection and repair of the sciatic nerve in rats: effects of acetyl-L-carnitine. Int J Clin Pharm Res. 1990; 10: 85-99. 270. Hart AM, Terenghi G, Johnson MA, Youle M. L-acetyl-carnitine: a pathogenesis based treatment for HIV-Associated distal symmetrical polyneuropathy. Manuscript. 271. Jessen KR, Mirsky R, Morgan L. Myelinated, but not unmyelinated axons, reversibly down-regulate N-CAM in Schwann cells. J Neurocytol. 1987; 16: 681-688. 272. Singer P, Mehler S. Glucose and leucine uptake in the hypoglossal nucleus after hypoglossal nerve transection with and without prevented regeneration in the Sprague-Dawley rat. Neurosci Lett. 1986; 67: 73-77. 273. Fu S.Y., Gordon T. Contributing factors to poor functional recovery after delayed nerve repair: prolonged denervation. J Neurosci. 1995; 15: 3886-3895.

274. Mosahebi A, Fuller P, Wiberg M, Terenghi G. Effect of allogeneic Schwann cell transplantation on peripheral nerve regeneration. Exp Neurol. 2002; 173: 213-223. 275. Li H, Wigley C, Hall SM. Chronically denervated rat Schwann cells respond to GGF in vitro. Glia 1998; 24: 290-303. 276. Mohanna P-N, Young RC, Wiberg M, Terenghi G. A composite PHB-GGF conduit for long nerve gap repairs. Exp Neurol. 2003; in process citation.

III

Exp Brain Res (2002) 142:308Ð318 DOI 10.1007/s00221-001-0929-0

RESEARCH ARTICLE

Andrew McKay Hart á Thomas Brannstrom Mikael Wiberg á Giorgio Terenghi Primary sensory neurons and satellite cells after peripheral axotomy in the adult rat Timecourse of cell death and elimination

Received: 6 April 2001 / Accepted: 26 September 2001 / Published online: 18 December 2001 © Springer-Verlag 2001

Abstract The timecourse of cell death in adult dorsal Neuron loss was greater in L5 than L4 axotomised gan- root ganglia after peripheral axotomy has not been fully glia, began at 1 week (15%, P=0.045) post-axotomy, characterised. It is not clear whether neuronal death be- reached 35% at 2 months (P<0.001) and was not signifi- gins within 1 week of axotomy or continues beyond cantly greater at 4 months or 6 months. Volume of axo- 2 months after axotomy. Similarly, neither the time- tomised ganglia fell to 19% of control by 6 months course of satellite cell death in the adult, nor the effect of (P<0.001). In animals that underwent nerve repair, both nerve repair has been described. L4 and L5 dorsal root the number of TUNEL-positive neurons and neuron loss ganglia were harvested at 1Ð14 days, 1Ð6 months after were reduced. Immediate repair was more protective sciatic nerve division in the adult rat, in accordance with than repair after a 1-week delay. Thus TUNEL positivity the Animals (Scientific Procedures) Act 1986. In sepa- precedes actual neuron loss, reflecting the time taken to rate groups the nerve was repaired either immediately or complete cell death and elimination. Neuronal death be- following a 1-week delay, and the ganglia were harvest- gins within 1 day of peripheral axotomy, the majority oc- ed 2 weeks after the initial transection. Microwave perm- curs within the first 2 months, and limited death is still eabilisation and triple staining enabled combined occurring at 6 months. Neuronal death is modulated by TUNEL staining, morphological examination and neuron peripheral nerve repair and by its timing after axotomy. counting by the stereological optical dissector technique. Secondary satellite cell death also occurs, peaking TUNEL-positive neurons, exhibiting a range of morph- 2 months after axotomy. These results provide a logical ologies, were seen at all timepoints (peak 25 cells/group framework for future research into neuronal and satellite 2 weeks after axotomy) in axotomised ganglia only. cell death within the dorsal root ganglia and provide fur- TUNEL-positive satellite cell numbers peaked 2 months ther insight into the process of axotomy induced neuro- after axotomy and were more numerous in axotomised nal death. than control ganglia. L4 control ganglia contained 13,983 (SD 568) neurons and L5, 16,285 (SD 1,313). Keywords TUNEL á Stereology á Optical dissection á Apoptosis á Dorsal root ganglion á Rat A. McKay Hart á G. Terenghi (✉) Blond-McIndoe Laboratories, Royal Free and University College Medical School, Abbreviations a Area aø Mean area á C.A.S.T.-Grid University Department of Surgery, Royal Free Campus, Computer-assisted stereology toolbox á DRG Dorsal root Rowland Hill Street, London NW3 2PF, UK ganglia á GDNF Glial-derived neurotrophic factor e-mail: [email protected] Tel.: +44-20-77940500 ext. 3944/8438, Fax: +44-20-74314528 H33342 Hoechst 33342 nuclear stain á ISEL In situ end labelling á LIF Leukaemia inhibitory factor á Ndis Total T. Brannstrom number of optical dissector bricks á NGF Nerve growth Department of Medical Biosciences Ð Pathology, Umea University, 901 87 Umea, Sweden factor á Npar Estimated number of particles á NT3 Neurotrophin-3 á N Numerical volume á A. McKay Hart á M. Wiberg V Department of Surgical and Perioperative Science, PFA Paraformaldehyde á PBS Phosphate-buffered saline á Section for Hand and Plastic Surgery, University Hospital, PI Propidium iodide á s Total number of sections á Umea, Sweden t Section thickness á øt Mean section thickness á M. Wiberg TUNEL TdT (terminal deoxyribonucleotidyl transferase) Department of Integrative Medical Biology, Section for Anatomy, uptake nick-end labelling á Vdis Volume of optical Umea University, Umea, Sweden dissector brick á Vref Reference volume 309 Introduction bers may mask loss occurring at very low rates. Also, complete involution does not occur for some hours in The extensive primary sensory neuronal death found vitro (Barres et al. 1993; Greenlund et al. 1995; Wyllie after experimentally induced peripheral nerve lesions 2000), and an unspecified time in vivo (Bilbao and (Himes and Tessler 1989; Liss et al. 1994, 1996; Ranson Dubois-Dauphin 1996), after the start of cell death. Evi- 1906; Rich et al. 1989; Risling et al. 1983; Terenghi dence of individual cell death, whether morphological 1999; Vestergaard et al. 1997) is likely to be a major or DNA fragmentation, is therefore more appropriate for factor in the poor clinical outcome of peripheral nerve determining when cells first begin to die after axotomy trauma. A variety of stimuli may initiate neuronal death, and is more suitable for detecting low rates of cell although loss of target-derived neurotrophic support ap- death. DNA fragmentation can be demonstrated by the pears to be the most significant (Ambron and Walters use of staining techniques such as terminal deoxyribo- 1996; Ekstrom et al. 1998), as demonstrated by the pro- nucleotidyl transferase (TdT) uptake nick-end labelling tective effect of exogenous neurotrophic factors such as (Gavrieli et al. 1992) which rely upon enzymatically NGF (Ljunberg et al. 1999; Otto et al. 1987; Rich et al. mediated polymerisation of labelled nucleotides onto 1987), NT3 (Groves et al. 1999; Ljungberg et al. 1999), the free ends of DNA strand breaks in fixed tissues. GDNF (Bennett et al. 1998: Leclere et al. 1998) and LIF Peripheral axotomy induces satellite cell death in neo- (Cheema et al. 1994). nates (Whiteside et al. 1998b) and adults (Groves et al. Between 7% and 50% of primary sensory neurons die 1997a), but the timecourse is not yet known. It may have after peripheral nerve injury (Himes and Tessler 1989; similar significance to the delayed apoptosis of oligoden- Liss et al. 1994; Ranson 1906; Rich et al. 1989; Risling drocytes found after spinal cord (Crowe et al. 1997) or et al. 1983; Terenghi 1999; Vestergaard et al. 1997; Ygge optic nerve injury induced (Barres et al. 1993) neuronal 1989), the range of experimental models and techniques death, which has been suggested to compound the direct employed (Ygge 1989) accounting for the wide variation effects of injury. in estimates (Rich et al. 1989). Until recently variation This study therefore aimed to determine the time- was increased by the use of “statistically biased”, as- course of primary sensory neuronal and satellite cell sumption-based counting techniques (West 1993), death after a defined peripheral nerve injury using wherein the mean of all possible estimates tends away TUNEL staining (for DNA fragmentation), light micros- from the true value. Stereological counting techniques copy of nuclear and cellular morphology and neuron (such as the physical, and optical dissector) now permit counting by the optical dissector technique. “statistically unbiased” counting, meaning that estimates will tend infinitely towards the true value. The optical dissector technique is superior to the physical, since it is Materials and methods more time-efficient and avoids potential underestimates due to tissue-sectioning artefacts (Pover and Coggleshall All work was performed in accordance with the Principles of lab- 1991; West 1993). oratory animal care (NIH publication No. 86Ð23, revised 1985) and with the terms of the Animals (Scientific Procedures) Act The precise onset of neuronal death after peripheral 1986 (project No. 70/4210). The experimental design was cogni- nerve injury remains uncertain. DNA fragmentation oc- scent of the need to minimise the number of experimental animals curs at 7 days post-axotomy (Groves et al. 1997a), but and to limit suffering. there is a lack of data for earlier timepoints and, whilst Under halothane anaesthesia, young adult male Sprague-Daw- ley rats underwent either left- or right-sided unilateral sciatic significant neuron loss has generally not been found nerve division at the upper border of quadratus femoris. Spontane- within 1 week of axotomy (Himes and Tessler 1989; ous regeneration was prevented by excising a 3-mm segment of Rich et al. 1989), Vestergaard et al. have demonstrated distal nerve stump, ligating both nerve stumps (6/0 Prolene), se- a 6% loss at 4 days, rising to 19% at 8 days (Vestergaard curing them into silicone caps (3 mm long, internal diameter 1.0 mm) with single 9/0 Ethilon epineurial sutures and burying the et al. 1997). proximal stump deep to quadratus femoris. Equally, it remains unclear exactly how long after After survival periods of 1 day, 4 days, 1 week, 2 weeks, axotomy neuronal death continues, as only three relevant 1 month, 2 months, 4 months and 6 months (n=6 per timepoint), longterm studies on the rat (Groves et al. 1997a; Himes ipsilateral axotomised and contralateral control L4 and L5 ganglia and Tessler 1989; Rich et al. 1989) have been published. were harvested. Terminal anaesthesia with intraperitoneal pentobarbitone (240 mg/kg) was employed and pre-fixation achieved Although no significant increase in neuron loss occurred by transcardiac perfusion with 200 ml PBS then 4% PFA beyond 1Ð2 months post-axotomy, these studies princi- (150 ml/100 g body weight), both at 4¡C. Ganglia were post-fixed pally addressed the mechanism of neuronal death in 4% PFA for 6 h at room temperature. After three rinses in PBS (Groves et al. 1997a), the effects of different injury types containing 15% sucrose and 0.1% sodium azide, ganglia were blocked in OCT and stored at Ð40¡C. (Rich et al. 1989) or employed non-stereological count- In two further groups (n=6), the sciatic nerve was divided and ing techniques (Himes and Tessler 1989). Additionally, then repaired either immediately or after a 1-week delay, using neither the optical dissector technique, nor staining for four 9/0 Ethilon epineurial sutures. The L4 and L5 ganglia were DNA fragmentation were employed throughout the harvested at a timepoint 2 weeks after the initial nerve division. Each entire ganglion was cut into serial 30-µm cryosections entire timecourse studied. onto Vectabond-coated (Vector UK) glass slides and dried over- Neuron loss quantifies the extent of neuronal death; night. Sections were permeabilised by microwave irradiation however, the normal biological variation in neuron num- (478 W for 1.5 min with a 1.5-l water load) and rapid cooling in 310 iced PBS. TUNEL (Promega Apoptosis Detection System, Fluo- under a ×4 objective. The thickness (t) of every section examined rescein) was then performed in accordance with the recommended was also directly measured, using a stage-mounted microcator and protocol (Promega 1999). The protocol was adapted to allow si- a ×60 oil-immersion objective, by focusing down from the surface multaneous staining with H33342 and PI. H33342 is a non-specif- of the section to the focal plane at its underside. Results were then ic nuclear stain that allows excellent determination of nuclear used for calculation of mean section thickness (øt), mean surface morphology (Crowe et al. 1997; Whiteside and Munglani 1998; area (øa) and hence of ganglion volume (Vref). Whiteside et al. 1998a), whilst PI highlights cytoplasmic morphol- The density (or NV) of neurons within the ganglion was then ogy. Each stain can be visualised separately using fluorescence determined by using a series of optical dissector bricks. These are microscopy and band-specific filters. cuboid, computer-generated geometrical constructs of a known Changing between fluorescent filters easily permitted differen- volume which are placed within the same tissue sections selected tiation of neurons from satellite cells on the basis of their size, rel- for volume estimation. Systematic randomisation is performed ative distribution and their larger, paler nuclei. The reliability of in the x- and y-axes by movements (x steps and y steps) from a these features was previously confirmed in separate experiments, random start point. Using an objective lens with a narrow focal where staining with PI and H33342 was combined with immuno- plane (less than 1 µm), the dissector bricks are generated by focus- staining using antisera to Pan-Axonal, marker of neurofilaments ing down through a predetermined distance within the thickness (Affiniti UK) and the satellite cell marker S-100 (DAKO, UK). (z-axis) of the tissue section. During optical dissection, use of an Olympus WU filter allowed In practice, optical dissection and systematic random place- simultaneous visualisation of both PI and H33342. ment of optical bricks within each section were performed by the All sections were examined for the presence of TUNEL-posi- C.A.S.T.-Grid system, which controlled a microscope stage mo- tive cells using fluorescence microscopy and a ×20 objective lens torised in the x-y plane. Manual movement in the z-axis (focusing (Olympus B60 microscope). Cells were confirmed to be neurons down through the section) was tracked by the stage-mounted mi- × 3 or satellite cells by further examination under a 40 objective, and crocator. The dissector volume (Vdis) was 81,702 µm , x steps and the morphology of each TUNEL-positive neuron was examined by y steps were 300 µm and the sections were found to be 28.0 µm switching filters to visualise the nuclear and cytoplasmic staining. (SD 1.85 µm) thick. In every case the serial sections immediately before and after the The leading edge of the nucleus was used as the unique point section containing the TUNEL-positive neuron were inspected to of the neuron for cap counting; standard stereological counting ensure that each positive neuron was only counted once. Absolute frames (Gundersen et al. 1988b) and optical dissector counting numbers of TUNEL-positive neurons were therefore obtained for rules were employed (Gundersen et al. 1988a). These rules deter- each ganglion. mine that only particles lying entirely within the volume of the At each timepoint the number of TUNEL-positive cells was bricks or crossing any of three mutually perpendicular faces are expressed either for individual ganglia (L4 or L5) or as a com- counted. A “cap” is a particle, or neuron in this case, which meets bined (L4+L5) count. For neurons the total number per experi- the counting rules employed. Neurons which cross any of the op- mental group was used, whilst for satellite cells a mean figure per posite three faces are not counted (“profiles”). The mean cap animal was calculated. count was 148 (SD 27.5) per ganglion, in keeping with the re- Neuron counts were performed using an Olympus B50 micro- quirement to count between 100 and 200. scope with parfocalised ×60 oil-immersion and ×4 objectives. The The cap count thereby obtained is divided by the total volume number of surviving L4 and L5 DRG neurons was estimated using within which counting was performed (i.e. individual dissector the stereological technique of optical dissection (Gundersen et al. brick volume multiplied by the total number employed) to give the 1988a; Harding et al. 1994; West 1993). density (NV) of neurons within the ganglion: With this technique one initially estimates the ganglion's volume by using the Cavalieri principle (Gundersen et al. 1988b), (2) which states that the volume of a structure approximates very closely to its mean cross-sectional surface area multiplied by its where: N is numerical volume of neurons; V is volume of each height. Mean surface area of the section (øa) is calculated from V dis optical disector brick; and Ndis is number of optical disector bricks measurements of the surface area (a) of ten or more serial sections used. in a systematic random sample from the ganglion. The “height” of Finally, the numerical volume (N ) of neurons is multiplied by the ganglion equals the total number of tissue sections cut from V the volume of the ganglion (Vref) to give the estimated number of the ganglion (s) multiplied by the mean thickness of those sections neurons within the ganglion: (øt), and this latter figure was 28.0 µm (SD 1.85 µm). The volume of the DRG (Vref) is therefore calculated thus: (3)

(1) where Npar is the estimated number of neurons within the ganglion. Neuron counts were expressed for individual ganglia (L4 or ø where Vref is volume of DRG, øa= mean area of section, t = mean L5) and as a combined (L4+L5) count for each side (axotomised section thickness and s is the total number of sections. or contralateral control). Neuron loss was calculated by subtract- In stereology every cell has an equal chance of being counted, ing the number of neurons in axotomised ganglia from that in their obviating any bias due to variations in cell size or distribution contralateral controls. Loss was then expressed as a percentage of within the ganglion. Central to this tenet are the counting rules as- the neuron count in the control ganglia. sociated with the optical dissector bricks and the process of sys- Statistical analysis of results was performed using the Sigma- tematic random sampling at all levels of the process after tissue stat 2.0 software package. Comparison between groups was per- sectioning. Sampling is achieved by starting from a randomly cho- formed using ANOVA, Students t-test or the Mann-Whitney rank sen start point and then systematically moving through the gangli- sum test when data were not normally distributed. P<0.05 was on or tissue section, in a series of predetermined steps. adopted for rejection of the null hypothesis. In practice, a systematic random sample of the serial sections from each ganglion was obtained for volume estimation as follows. The start section (1stÐ3rd of the serial sections) was selected using a random number table and was examined along with every Results 3rd subsequent section from the ganglion. Where this randomisation procedure resulted in fewer than the required minimum of 10 Morphology sections being available, the procedure was altered to select every 2nd section (start section 1st or 2nd). The neuron-containing area (a) of each section was measured TUNEL-positive neurons displayed a range of morph- using C.A.S.T.-Grid (version 1) software by tracing its margins ologies (Fig. 1aÐc, e), although all had abnormal nuclei 311 Fig. 1 Morphology of primary sensory neurons after peripheral axotomy. Positive TUNEL staining shows as green fluorescence, cytoplasmic staining by propidium iodide shows as red, and nuclear staining by Hoechst 33342 shows as blue. TUNEL-positive neurons (solid horizontal arrows) showed a range of cytoplasmic morphologies (aÐc, e). Typical features included pyknosis (a), granulation (a), vacuolation (b) and membrane blebbing (b). However, some cells were essentially normal in appearance (c). Numerous enlarged and grossly vacuolated neurons were found at all timepoints studied. Generally such cells were TUNEL-negative (outline arrow, d), although two were TUNEL-positive cells, one of which is shown (e) alongside a TUNEL-positive satellite cell (vertical arrow). Scale bar 20 µm

(H33342 staining) exhibiting pyknosis, chromatin con- (Fig. 1d) were found at every timepoint, with a peak in- densation or even nuclear fragmentation. Whilst many cidence 1Ð2 months after axotomy. These cells were had cytoplasmic features typical of apoptosis (nuclear found almost exclusively in axotomised ganglia and had eccentricity, pyknosis, granulation, vacuolation and eccentric, elliptical nuclei, with a degree of chromatin membrane irregularities; Fig. 1a, b), some exhibited an condensation. Two such cells were TUNEL-positive, one apparently normal cytoplasmic morphology (Fig. 1c). of which is shown in Fig. 1e. Satellite cells with abnormal, intensely stained, irregular, condensed or ring-shaped nuclei were frequently seen. Whilst many were TUNEL-positive, a substantial minor- TUNEL staining ity were TUNEL-negative. Apart from the classic pattern of chromatolysis and TUNEL-positive cells were found at all depths within nuclear eccentricity, other morphologically abnormal but tissue sections. The number found at each timepoint after TUNEL-negative neurons were seen. “Ghost cells”, axotomy is given in Table 1. TUNEL-positive neurons which are thought to represent the site of fully involuted were only found in axotomised ganglia and were present apoptotic neurons, were occasionally found, as were at all timepoints. The cumulative (L4+L5) number of morphologically apoptotic neurons (exhibiting pyknosis, TUNEL-positive neurons at 1 day after axotomy was cytoplasmic granulation, vacuolation and membrane 2 cells/group and first became significantly different irregularities). Grossly dilated and vacuolated cells from control ganglia after 1 week (15 cells/group; 312 Table 1 Timecourse of positive TUNEL staining in neurons and satellite cells after peripheral axotomy

Timepoint DRG neurons (total count/group) Satellite cells (mean count/ganglion)

Control Axotomised ganglia Control ganglia Axotomised ganglia Axotomised Ð ganglia control ganglia L4+L5 L4 L5 L4+L5 L4 L5 L4+L5 L4 L5 L4+L5 L4+L5

1 day 0 1 1 2 5.5 4.0 10.3 5.5 4.8 10.0 4 days 0 5 3 8 2.2 4.3 6.5 7.0 6.3 13.0 6.8 1 week 0 2 13* 15* 0.8* 7.2 8.0 4.7 12 17.0* 8.7 2 weeks 0 13* 12* 25** 3.5* 7.8 11.0 12 11 22.0* 11.0 1 month 0 4 2 6 3.0 2.7 5.7 4 10 14.0* 8.3 2 months 0 2 3 5 20.0 11.0 31.0 17 34 51.0 20.0 4 months 0 1 0 1 7.5 5.2 13.0 13 9.7 22.0* 9.7 6 months 0 2 0 2 5 2.3 7.3 8.2 9.8 18.0 11.0

Data based upon experimental groups of six animals at each timepoint *P<0.05; **P<0.001 (compared with control)

Table 2 Effect of timing of nerve repair upon post-axotomy neuron death

No nerve repair Delayed nerve repair Immediate nerve repair

Mean neuron loss (L4+L5) 6,324** 2,450 1,236 TUNEL-positive neuron count/group (L4+L5) 25** 14* 11*

Data based upon experimental groups of six animals at each timepoint *P<0.05; **P<0.001 (compared with control)

P<0.05). Numbers peaked at 25 cells/group after was 13,983 (SD 568) and 16,285 (SD 1,313) for L5. No 2 weeks (P<0.001), falling thereafter to 2 cells/group at statistically significant differences were found between 6 months. control ganglia at different timepoints or between sides TUNEL-positive satellite cells were present at all for L4 (right 13,648, SD 1,633; left 14,187, SD 1,633) timepoints and were more numerous in axotomised gan- or L5 (right 16,574, SD 2,853; left 16,112, SD 1,864) glia than in control ganglia. Subtracting the mean cumu- ganglia. lative (L4+L5) count on the control side from that of the In axotomised ganglia the magnitude of neuron loss axotomised side gave a reflection of the number of satel- was greater in L5 (peak loss 7,878 neurons, 43.6%) than lite cells that were TUNEL-positive due the direct effect L4 (peak loss 5,024 neurons, 34.6%); however, the time- of axotomy. This count was 6.8 cells/animal at 4 days, course of neuron loss was comparable. No neuron loss peaked at 20.2 cells at 2 months and was 10.7 cells at had occurred at 4 days post-axotomy, indeed counts were 6 months after axotomy (Table 1). marginally higher in the axotomised ganglia than in the The timing of peripheral nerve repair affected the contralateral controls (hence the calculation of “nega- number of TUNEL-positive neurons, as displayed in tive” loss in Table 3). Neuron loss (L4+L5) in axotomi- Table 2. After immediate repair of the sciatic nerve, sed ganglia was first evident at 1 week post-axotomy cumulative (L4+L5) TUNEL-positive neuron counts (15% when compared with control ganglia; P=0.045), were lower (11 cells/group; P=0.002 compared with reaching 35% at 2 months (P<0.001). Despite the control ganglia) than with repair after a 1-week delay fact that loss peaked 4 months after axotomy (39.2%; (14 cells/group; P=0.016). The count was still higher Table 3), the percentage loss found at 4 months and when the nerve was not repaired (25 cells/group; 6 months was not statistically different from that found P<0.001 compared with control, and P<0.05 compared 2 months after axotomy. with immediate repair). The difference between the Two weeks after sciatic nerve transection, the extent counts obtained after immediate and delayed repair was of neuron loss was related to the timing of nerve repair, not statistically significant. as is shown in Table 2. Cumulative (L4+L5) neuron loss was highest when the nerve was not repaired (21.3%, 6,325 neurons), whereas loss was lower following repair Stereology after a 1-week delay (12.4%, 2,450 neurons), and this effect was even more marked after immediate repair The number of neurons surviving after axotomy is given (4.8%, 1,236 neurons). However, the effect upon neuron in Table 3, along with the calculated percentage neuron loss was not statistically significant. losses. When all experimental groups were considered The change in the volume (Vref) of the neuron-con- together, the mean neuron count in control L4 ganglia taining region of ganglia which occurred both as a result 313 Table 3 Effect of peripheral axotomy upon primary sensory neuron counts

Time- Control ganglia Axotomised ganglia Percentage neuron loss point L4 L5 L4+L5 L4 L5 L4+L5 L4 L5 L4+L5

Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD

4 days 14,233 2,810 15,972 669 30,205 1,122 15,138 4,829 16,413 1,377 31,551 3,322 Ð6.0% Ð3.0% Ð4.5% 1 week 13,517 2,064 14,966 2,193 28,483 1,447 11,005 452 13,256 939 24,261 2,352 18.6%* 14.3% 14.8%* 2 weeks 13,949 1,193 15,692 2,198 29,641 926 11,239 1,305 12,077 1,500 23,317 876 19.0%* 23.0%* 21.3%** 1 month 14,658 1,953 14,647 948 29,305 474 11,349 2,497 11,520 1,727 22,869 1,295 22.6%** 25.6%** 22.0%** 2 months 13,948 1,007 17,597 2,524 31,545 999 9,751 1,230 10,709 1,082 20,460 873 30.1%** 39.1%** 35.1%** 4 months 14,536 801 18,055 2,279 32,591 791 9,511 655 10,315 621 19,826 237 34.6%** 43.6%** 39.2%** 6 months 13,039 1,392 17,066 2,289 30,105 843 9,370 519 10,341 1,601 19,711 711 28.1%** 39.4%** 34.5%**

Data based upon experimental groups of six animals at each timepoint *P<0.05; **P<0.001 (compared with control)

Table 4 Effect of age and peripheral axotomy upon neuron containing volume of dorsal root ganglia

Time- Age-related increase (from 4 days post-axotomy) in neuron-containing Neuron-containing volume of axotomised point volume of ganglia ganglia (% of control)

Control L4 Axotomised L4 Control L5 Axotomised L5 L4 L5 L4+L5

4 days 111% 105% 108.0% 1 week 19.3%* 3.0% 35.9%* 3.7% 95% 80%* 87.9%* 2 weeks 30.1%* Ð3.2% 22.7% Ð2.7% 82%* 82%* 82.6%* 1 month 15.0% Ð11.2% Ð2.3% Ð26.7%* 85%* 79%* 82.2%* 2 months 32.2%* Ð4.6% 0.0% Ð3.1%* 80%* 69%* 73.4%** 4 months 63.4%** 17.3% 69.3%** 16.4%* 80%* 72%* 75.5%** 6 months 68.4%* 28.0%* 68.3%** 28.0%* 82%* 80%* 80.8%**

Data based upon experimental groups of six animals at each timepoint *P<0.05; **P<0.001 (compared with control) of the increasing age of animals over the study period tors by the proximal stump, thereby clarifying the role of and owing to the effect of peripheral axotomy is shown withdrawal of target organ and distal stump-derived neu- in Table 4. Axotomy caused a significant reduction in rotrophins in determining the magnitude of neuronal Vref (L4+L5) at 1 week post-axotomy (87.9% of control; death in vivo. P=0.04), which then fell to 81% of control by 6 months TUNEL is a technique for detecting DNA fragmenta- (P<0.001). An age-related increase of 68.4% in L4 and tion (the earliest irreversible stage of neuronal death) 68.3% in L5 occurred in the neuron-containing volume that was initially proposed as a specific marker for apop- of control ganglia, whilst axotomised L4 and L5 ganglia tosis (Gavrieli et al. 1992). However, positive staining both increased by only 28.0%. has subsequently been found in the S-phase of mitosis (Sanders and Wride 1996) and in necrosis (Charriaut- Marlangue and Ben-Ari 1995; Gold et al. 1994), so the Discussion results presented herein should be interpreted as refer- ring simply to cell death, and not specifically to apopto- Methodology sis. Methodological concerns have been expressed previously over the sensitivity of TUNEL and ISEL The experimental model was standardised as far as pos- staining, owing to inadequate tissue permeabilisation by sible, since factors such as age and the proximity of the traditional enzymatic methods such as digestion with nerve transection to the ganglia affect the DRG neuron proteinase-K (Groves et al. 1997a; Negoescu et al. response to injury (Kerezoudi et al. 1995; Snider et al. 1997). Microwave irradiation of tissue sections increases 1992) and the magnitude of DRG neuron loss (Ygge the efficacy of permeabilisation sufficiently to give a 1989). All animals therefore underwent sciatic nerve sensitivity of more than 90% with no reduction in speci- division at the same age, the level of transection in the ficity (Negoescu et al. 1997), and we therefore optimised mid-thigh was anatomically standardised, and several a microwave permeabilisation protocol. Despite using measures were taken to prevent spontaneous nerve re- thick (30 µm) tissue sections, penetration of the TUNEL generation (which would restore distal trophic support). reagents was not impaired, since TUNEL-positive cells These measures also created significant physical barriers were found at all depths within tissue sections and not to the uptake of distal stump-derived neurotrophic fac- just in the surface layer. The specificity of TUNEL stain- 314 ing was confirmed by the fact that all TUNEL-positive Vestergaard et al. 1997), although smaller losses have cells exhibited abnormal nuclear morphology. been reported (Groves et al. 1997a; Himes and Tessler Optical dissection avoids both the statistical bias of 1989; Rich et al. 1989). Variations may be due to differ- assumption-based techniques (Hedreen 1998b) and the ent animal models, counting techniques or levels of axo- potential underestimates inherent to the physical dissec- tomy. Also, the procedure used to prevent spontaneous tor technique, due to lost caps (Hedreen 1998a; Pover regeneration in this study may have excluded trophic and Coggleshall 1991; West 1993). It has therefore been support from the distal stump more completely than in proposed as the ideal technique for estimation of neuron other studies, resulting in greater neuronal death. One numbers, although this has been questioned recently on may estimate that, of the neurons that were actually axo- the grounds that the anisotropic arrangement of neurons tomised, 56Ð75% were lost, since only 54Ð70% of neu- within ganglia and the effects of section shrinkage may rons in L4 and L5 DRGs contribute axons to the sciatic result in inaccuracies (Benes and Lange 2001). However, nerve at the mid-thigh (Devor et al. 1985; Himes and when the technique is correctly applied, with adequate Tessler 1989; Swett et al. 1991). That neuron loss was systematic random sampling, it is efficient, theoretically greater in L5 than L4 ganglia reflects the greater contri- sound, and it has been empirically validated against seri- bution of L5 neurons to the sciatic nerve at the mid-thigh al reconstruction counts of DRG neurons (Coggleshall (Devor et al. 1985). 1992; Coggleshall et al. 1990). Volume estimation by the Cavalieri principle (Gundersen et al. 1988b) requires the use of at least ten sections, whose thickness must be Timecourse of neuronal death known accurately. Accordingly, section thickness was measured directly from the sections after staining, rather Primary sensory neuronal death was found to commence than assumed to be equal to that at which the sections within 24 h of sciatic nerve transection, as shown by the were cut (tissue shrinkage during staining may invalidate presence of TUNEL-positive neurons at 1 day post-axo- such an assumption; Harding et al. 1994). Adequate tomy. Whilst previous studies have given the exact onset numbers of caps were counted to meet the requirement of neuron loss as being at 4 (Vestergaard et al. 1997) to of 100Ð200 per ganglion (West 1999); and at 28% and 10 days (Himes and Tessler 1989), we demonstrate that 19% respectively, the x and y dimensions of the dissector statistically significant neuron loss in fact begins 7 days frame employed were more than 4% of the x and y step after sciatic nerve transection in the mid-thigh. The delay distances (Harding et al. 1994). between the start of neuronal death and that of neuron loss reflects both the time taken for individual neurons to progress from DNA fragmentation to complete involu- Neuron counts tion, and for the rate of death to increase sufficiently to overcome the masking effect of the normal biological The mean neuron counts in control ganglia (L4 13,983, variation in neuron numbers. SD 568; L5 16,285, SD 1,313; L4+L5 30,268, SD 2,548) In vitro, neurons take a few hours to progress from were in agreement with other results in the literature ob- DNA fragmentation (with the potential for positive tained by optical (Bergman and Ulfhake 1998; Tandrup TUNEL staining) to eventual involution (Chun and 1993, 1995) and physical (Groves et al. 1997a) dissector Blaschkie 1997; Deckwerth and Johnson 1993); in vivo, methods. There was no difference between the counts a delay of days is found in the case of axotomised neona- obtained from the left or right sides, negating concerns tal facial motoneurones (Bilbao and Dubois-Dauphin over other experimental models that employ lesions on 1996). In this study, at most 6 days passed between posi- one side only. tive TUNEL staining and the occurrence of significant L4+L5 counts have been calculated because this com- neuron loss. Although neuronal involution in the first prises 98.4% of the entire sciatic nerve pool (Swett et al. few days may have occurred at too slow a rate to trans- 1991). Furthermore, the variance in counts from individ- late into significant neuron loss, the full course of senso- ual control ganglia is greater than that found for com- ry neuronal death in vivo, from beginning (DNA frag- bined counts (coefficient of variation L4 12%, L5 14%, mentation) to end (neuron loss), appears to last some L4+L5 8%). This suggests that biological constancy is days. maintained in the total number of neurons in the sciatic TUNEL staining also exhibits an earlier peak (at nerve root pool, rather than in the number present in 2 weeks post-axotomy) than neuron loss (at 4 months either ganglion alone, hence the combined counts are post-axotomy). ISEL staining (a marker specific for sin- more sensitive indicators of the effect of axotomy than gle- rather than double-strand DNA breaks) similarly has those for individual ganglia alone. been found to peak at 3 weeks post-axotomy (Groves et al. 1997a), although the count is only a single neuron more than at 2 weeks. The peak incidence of DNA frag- Neuron loss mentation, and so of neuronal death, therefore occurs some 2Ð3 weeks after peripheral nerve transection. Neu- At 39.2%, the peak neuron loss was broadly in keeping ron loss was found to occur predominantly during the with results in the literature (Liss et al. 1994, 1996; first 2 months after injury, since the losses found at 315 4 months and 6 months were not significantly different to creasing (Groves et al. 1997a). Alternatively, some neu- that found 2 months after axotomy. This is broadly in ronal death may occur without any potentially stainable keeping with the literature, although no previous study DNA strand breakage, as has been demonstrated in other has involved sufficiently long survival times (Vestergaard cell types (Clarke 1999). et al. 1997) and been as detailed (Groves et al. 1997a; Himes and Tessler 1989). The discrepancy between the timecourse of TUNEL Mechanism of neuronal death staining and that of neuron loss reflects a difference in what each technique detects. As a marker of DNA frag- Loss of distal trophic support is postulated as the major mentation, TUNEL staining detects the beginning of a cause of neuron loss after nerve transection, and surgical cell's death rather than the subsequent elimination re- repair is thought to restore that support. Hence, one gains flected by neuron loss. TUNEL positivity therefore pre- information about the magnitude of loss attributable to cedes loss, with the number of positive neurons at any trophin withdrawal by harvesting the ganglia at the same timepoint reflecting the rate of cell death, whilst neuron timepoint after axotomy having repaired the nerve at dif- loss quantifies the total number of cells that have died up ferent intervals, and an effect was found. until then. The number of TUNEL-positive neurons present It was unclear for exactly how long neuronal death 2 weeks after axotomy was reduced by nerve repair, with continues after a peripheral nerve injury. As discussed the effect of immediate repair being significant (P<0.05). previously, neuron loss is a relatively crude measure of The same trend was found for neuron loss, which was this, and various studies have not found any statistically 21.3% of control (L4+L5) in the absence of a nerve re- significant increase in neuron loss beyond 1Ð2 months pair, 12.4% when repair was delayed by 1 week and was after axotomy (Himes and Tessler 1989; Vestergaard minimal (4.8%) when the nerve was repaired immediate- et al. 1997). Indeed there is a trend (not statistically sig- ly after transection. One week after transection neuron nificant) for neuron loss to be less at 6 months post-axo- loss is 14.8% (Table 3), suggesting that virtually no fur- tomy than at 2 months, although the presence of morpho- ther neuron loss occurred in the week following delayed logically apoptotic and TUNEL-positive neurons indi- nerve repair and providing further evidence of the bene- cates that neuronal death does continue for at least ficial effect upon neuron survival of contact with the dis- 6 months after axotomy (Groves et al. 1997a; current tal stump (with accompanying neurotrophic support). study). Death is likely to be due to axotomy rather than The lack of a statistically significant difference in neuron old age, since control ganglia showed no evidence of loss between groups with or without nerve repair is most neuronal death, and since L4 neuron counts at 3 months probably due to the fact that 2 weeks was not long and 3 years of age in male Sprague-Dawley rats are not enough to develop a difference of sufficient magnitude statistically different (Bergman and Ulfhake 1998). Spe- for statistical differentiation. These results give further cies differences may exist, because neuron loss contin- support to the hypothesis that loss of distal-stump neuro- ued for 18 months post-axotomy in the cat (Carlson et al. trophic support is the major factor determining the mag- 1979), although this result was obtained using assump- nitude of neuronal death after peripheral axotomy. tion-based counting techniques and may not be reliable. Although the accurate morphological consideration of It is the large discrepancy between the small number the mode of death is outside the scope of this paper, of TUNEL-positive neurons found and the large neuron since further electron-microscopic morphological study loss that remains to be explained adequately. Although would be required, the light-microscopic morphological greater than in other studies, the number of TUNEL-pos- findings were not all classically apoptotic, and these lim- itive neurons was still low, despite the use of microwave ited findings may add to the current debate. TUNEL- irradiation to enhance permeabilisation and sensitivity. It positive neurons exhibiting no abnormalities of cytoplas- is not clear from the literature for how long a dying cell mic morphology were found and may be apoptotic cells will stain positively with TUNEL in vivo, although esti- caught just prior to the development of visible morpho- mates range from as little as 2Ð5 min (Negoescu et al. logical features (Migheli et al. 1995; Sanders and Wride 1997). In Gavrieli's original description, cells are de- 1996). However, others did display cytoplasmic granula- scribed as being TUNEL-positive before they are mor- tion, vacuolation, pyknosis and membrane irregularities, phologically apoptotic (Gavrieli et al. 1992), and it is in keeping with apoptosis. likely that the DNA fragments are exposed for only a The grossly dilated and vacuolated cells that were few hours before being masked by the condensation into predominantly found in axotomised ganglia have been nucleosomes (Wyllie 2000) that precedes the more ad- described previously (Ranson 1909) and demonstrated to vanced morphological features of apoptosis. This brief be large, GAP-43-positive myelinated neurons contain- time window for staining is therefore the most likely ex- ing proteinaceous material (Groves et al. 1997b). Since planation for the discrepancy between the small number none of the neurons were ISEL-positive, the features of TUNEL-positive neurons and the large number lost, were ascribed to regeneration rather than cell death. as previously has been suggested by the peak in ISEL However, our finding that two such neurons were staining of neurons 3 weeks after axotomy, when num- TUNEL-positive may mean that the previous failure to bers of morphologically apoptotic neurons were still in- demonstrate DNA fragmentation was a result of the scar- 316 city of such neurons, and the small percentage of dying relationship. In control ganglia, TUNEL positivity may neurons detected by DNA fragment labelling. Further- represent physiological turnover in the potentially labile more, the number of such cells present closely matches satellite cell population, although the marked increase the timecourse of neuron loss and is reduced by NT3 ad- 2 months after axotomy suggests a direct effect of con- ministration (Groves et al. 1997b), although that might tralateral axotomy, perhaps mediated by degeneration of be expected to increase regeneration (Terenghi 1999). A neurites in crossed tracts. These facts imply that adult recent review of morphological electron-microscopic primary sensory neuronal death induces delayed second- studies (Clarke 1999) has cast doubt upon the classic ary death of satellite cells, perhaps mediated by trophic apoptotic description of adult neuronal death, and the interdependence mechanisms or neurotoxic cytokines. type 3B (“cytoplasmic”) cell death morphological fea- This is of particular significance because the delayed oli- tures are similar to those found here. Apoptosis has godendrocyte death found after optic nerve (Barres et al. already been discounted as the mechanism in other forms 1993) and spinal cord injury (Crowe et al. 1997), and of adult neuronal death (Colbourne et al. 1999), and it thought to be due to axonal degeneration, has been im- may be the case therefore that the dilated vacuolated plicated in demyelination of surviving axons and expan- morphology is another pattern of death in myelinated sion of the cord lesion. primary sensory neurons after axotomy.

Conclusions Effect of axotomy upon ganglion volume The results presented herein therefore offer a logical The 27% rise in control L4 dorsal root ganglion volume time-structure for future in vivo experiments on primary between 2 and 6 months of age matches the previously sensory neuronal death and for testing any potential ther- described 28% increase in volume from 3 months to apies to prevent neuronal death. The secondary satellite 3 years of age (Bergman and Ulfhake 1998), since gan- cell death described remains to be fully characterised, as glion volume reached its maximum by 4 months of age. do the precise mechanisms of neuronal death and the A 68% increase in volume was found in both L4 and L5 features that differentiate the neurons that survive from control ganglia over the first 6 months of life; however, those that die. in axotomised ganglia volume increased by only half as much (28%), presumably reflecting their reduced cellu- Acknowledgements This work was funded by the Swedish Medi- larity. Axotomy caused a reduction in volume of both L4 cal Research Council (grant 02286), County of Vasterbottens, Ane’rs Foundation and the East Grinstead Medical Research Trust and L5 ganglia when compared with controls. This com- (registered charity No. 258154). menced 1 week after axotomy and plateaued after 2 months. Loss of volume was less than neuron loss, but showed a similar timecourse and may reflect both the References reduction in volume of surviving neurons (Vestergaard et al. 1997) and the actual loss of cells. Ganglion volume Ambron RT, Walters ET (1996) Priming events and retrograde also decreased relative to that found before neuron loss injury signals. A new perspective on the cellular and molec- commenced at 4 days post-axotomy, and the normal age- ular biology of nerve regeneration. Mol Neurobiol 13:61Ð 79 related increase in ganglion volume did not begin until Barres BA, Jacobson MD, Schmid R, Sendtner M, Raff MC after the main period of neuronal death had finished (i.e. (1993) Does oligodendrocyte survival depend upon axons? after 2 months post-axotomy). This implies that as neu- Curr Biol 3:489Ð497 rons die the spatial relationship of surviving cells is Benes FB, Lange N (2001) Two-dimensional versus three-dimensional cell counting: a practical perspective. Trends Neurosci altered, which may have implications for local neuron- 24:11Ð17 glia interactions and for local concentrations of trophic Bennett DL, Michael GJ, Ramachandran N, Munson JB, Averill S, factors. Yan Q, McMahon SB, Priestley JV (1998) A distinct subgroup of small DRG cells express GDNF receptor components and GDNF is protective for these neurons after nerve injury. J Neurosci 18:3059Ð3072 Axotomy-induced satellite cell death Bergman E, Ulfhake B (1998) Loss of primary sensory neurons in the very old rat: neuron number estimates using the dissector Although ISEL-positive satellite cells have been noted method and confocal optical sectioning. J Comp Neurol previously in control and axotomised ganglia, the time- 396:211Ð222 course of cell death has not been reported (Groves Bilbao FD, Dubois-Dauphin M (1996) Time-course of axotomy- induced apoptotic cell death in facial motorneurons of neona- 1997a). In this adult study, TUNEL-positive satellite tal wild type and bcl-2 transgenic mice. Neuroscience cells showed a definite temporal relationship to sciatic 71:1111Ð1119 nerve division, with a peak 2 months after axotomy, and Carlson J, Lais AC, Dyck PJ (1979) Axonal atrophy from perma- were consistently found within axotomised ganglia. The nent peripheral axotomy in adult cat. J Neuropathol Exp Neurol 38:579Ð585 neuron peak preceded the satellite cell peak by 6 weeks, Charriaut-Marlangue C, Ben-Ari Y (1995) A cautionary note on in contrast to 3 days in neonates (Whiteside et al. the use of the TUNEL stain to determine apoptosis. Neuro- 1998b), therefore implying a difference in the precise report 7:61Ð64 317 Cheema SS, Richards L, Murphy M, Bartlett PF (1994) Leukemia Hedreen JC (1998a) Lost caps in histological counting methods. inhibitory factor prevents the death of axotomised sensory Anat Rec 250:366Ð372 neurons in the dorsal root ganglia of the neonatal rat. J Neuro- Hedreen JC (1998b) What was wrong with the Abercrombie and sci Res 37:213Ð218 empirical cell counting methods? A review. Anat Rec Chun J, Blaschkie AJ (1997) Identification of neural programmed 250:373Ð380 cell death through the detection of DNA fragmentation in situ Himes BT, Tessler A (1989) Death of some dorsal root ganglion and by PCR. In: Current protocols in neuroscience. Wiley, neurons and plasticity of others following sciatic nerve section London, pp 3.8.1Ð3.8.19 in adult and neonatal rats. J Comp Neurol 284:215Ð230 Clarke PGH (1999) Apoptosis versus necrosis: how valid a dichot- Kerezoudi E, King RH, Muddle JR, O'Neill JA, Thomas PK omy for neurons. In: Koliatsos VE, Ratan RR (eds) Cell (1995) Influence of age on the late retrograde effects of sciatic death and diseases of the nervous system. Humana, Totowa, nerve section in the rat. J Anat 187:27Ð35 NJ, pp 3Ð28 Leclere P, Ekstrom P, Edstrom A, Priestley J, Averill S, Tonge DA Coggleshall RE (1992) A consideration of neural counting (1998) Effects of glial cell line-derived neurotrophic factor on methods. Trends Neurosci 15:9Ð13 axonal growth and apoptosis in adult mammalian sensory neu- Coggleshall RE, La Forte R, Klein CM (1990) Calibration of rons in vitro. Neuroscience 82:545Ð558 methods for determining numbers of dorsal root ganglion cells Liss AG, af Ekenstam FW, Wiberg M (1994) Cell loss in sensory (erratum in J Neurosci Methods Dec 1991; 40:87Ð90). J Neu- ganglia after peripheral nerve injury: an anatomical tracer rosci Methods 35:187Ð194 study using lectin-coupled horse-radish peroxidase in cats. Colbourne F, Sutherland GR, Auer RN (1999) Electron-micro- Scand J Plast Reconst Hand Surg 28:177Ð187 scopic evidence against apoptosis as the mechanism of neuro- Liss AG, af Ekenstam FW, Wiberg M (1996) Loss of neurons in nal death in global ischemia. J Neurosci 19:4200Ð4210 the dorsal root ganglia after transection of a peripheral sensory Crowe MJ, Bresnahan JC, Shuman SL, Masters JN, Beattie MS nerve: an anatomical study in monkeys. Scand J Plast Re- (1997) Apoptosis and delayed degeneration after spinal cord constr Hand Surg 30:1Ð6 in jury in rats and monkeys. Nat Med 3:73Ð76 Ljungberg C, Novikov L, Kellerth J-O, Wiberg M (1999) The neu- Deckwerth TL, Johnson EMJ (1993) Temporal analysis of events rotrophins NGF and NT-3 reduce sensory neuronal loss in adult associated with programmed cell death (apoptosis) of sympa- rat after peripheral nerve lesion. Neurosci Lett 262:29Ð32 thetic neurons deprived of nerve growth factor. J Cell Biol Migheli A, Attanasio A, Schiffer D (1995) Ultrastructural detec- 123:1207Ð1222 tion of DNA strand breaks in apoptotic neural cells by in situ Devor M, Govrin-Lippmann R, Frank I, Raber P (1985) Prolifera- end-labelling techniques. J Pathol 176:27Ð35 tion of primary sensory neurons in adult rat dorsal root gangli- Negoescu A, Lorimier P, Labat-Moleur F, Azoti L, Robert C, on and the kinetics of retrograde cell loss after sciatic nerve Guillermet C, Brambilla E (1997) TUNEL: Improvement and section. Somatosens Res 3:139Ð167 evaluation of the method for in situ apoptotic cell identifica- Ekstrom PA, Hedlund G, Karlsson J, Andersson G (1998) The tion. Biochemica 2:12Ð17 immune modulator Linomide prevents neuronal death in in- Otto D, Unsicker K, Grothe C (1987) Pharmacological effects of jured peripheral nerves of the mouse. Neuroreport 9:1337Ð nerve growth factor and fibroblast growth factor applied to the 1341 transectioned sciatic nerve on neuron death in adult rat dorsal Gavrieli Y, Sherman Y, Ben-Sasson SA (1992) Identification of root ganglia. Neurosci Lett 83:156Ð160 programmed cell death in situ via specific labeling of nuclear Pover CM, Coggleshall RE (1991) Verification of the dissector DNA fragmentation. J Cell Biol 119:493Ð501 method for counting neurons, with comments on the empirical Gold R, Schmied M, Giegerich G, Breitschoff H, Hartung HP, method. Anat Rec 231:573Ð578 Toyka KV, Lassman H (1994) Differentiation between cellular Promega (1999) Apoptosis detection system, fluorescein. In: In- apoptosis and necrosis by the combined use of in situ and nick structions for use of product G3250. (Technical Bulletin 235) translation techniques. Lab Invest 71:219Ð225 Promega Incorporated Greenlund LJ, Deckwerth TL, Johnson EMJ (1995) Superoxide Ranson SW (1906) Retrograde degeneration in the spinal nerves. dismutase delays neuronal apoptosis: a role for reactive oxy- J Comp Neurol 16:265Ð293 gen species in programmed neuronal death. Neuron 14:303Ð Ranson SW (1909) Alterations in the spinal ganglion cells follow- 315 ing neurotomy. J Comp Neurol Psychol 19:125Ð153 Groves MJ, Christopherson T, Giometto B, Scaravilli F (1997a) Rich KM, Luszczynski JR, Osborne PA, Johnson EM Jr (1987) Axotomy-induced apoptosis in adult rat primary sensory neu- Nerve growth factor protects adult sensory neurons from cell rons. J Neurocytol 26:615Ð624 death and atrophy caused by nerve injury. J Neurocytol Groves MJ, Giometto B, Scaravilli F (1997b) Axotomy-induced 16:261Ð268 vacuolation of primary sensory neurons and effect of adminis- Rich KM, Disch SP, Eichler ME (1989) The influence of regenera- tered neurotrophic factors: a morphometric, immunocyto- tion and nerve growth factor on the neuronal cell body reac- chemical and ultrastructural study. Prim Sensory Neuron tion to injury. J Neurocytol 18:569Ð576 2:111Ð127 Risling M, Aldskogius H, Hildebrand C, Remahl S (1983) Effects Groves MJ, An SF, Giometto B, Scaravilli F (1999) Inhibition of of sciatic nerve resection on L7 spinal roots and dorsal root sensory neuron apoptosis and prevention of loss by NT-3 ganglia in adult cats. Exp Neurol 82:568Ð580 administration following axotomy. Exp Neurol 155:284Ð294 Sanders EJ, Wride MA (1996) Ultrastructural identification of Gundersen HJG, Bagger P, Bendtsen TF, Evans SM, Korbo L, apoptotic nuclei using the TUNEL technique. Histochem J Marcussen N, Moller A, Nielsen K, Nyengaard JR, Pakken- 28:275Ð281 berg B, Sorensen FB, Vesterby A, West MJ (1988a) The new Snider WD, Elliott JL, Yan Q (1992) Axotomy induced neuronal stereological tools: dissector, fractionator, nucleator and point death during development. J Neurobiol 23:1231Ð1246 sampled intercepts and their use in pathological research and Swett JE, Torigoe Y, Elie VR, Bourassa CM, Miller PG (1991) Sen- diagnosis. APMIS 96:857Ð881 sory neurons of the rat sciatic nerve. Exp Neurol 114: 82Ð103 Gundersen JG, Bendtsen TF, Korbo L, Marcussen N, Moller A, Tandrup T (1993) A method for producing unbiased and efficient Nielsen K, Nyengaard JR, Pakkenberg B, Sorensen FB, estimation of number and mean volume of specified neuron Vesterby A, West MJ (1988b) Some new, simple and efficient subtypes in rat dorsal root ganglion. J Comp Neurol 329: stereological methods and their use in pathological research 269Ð276 and diagnosis. APMIS 96:379Ð394 Tandrup T (1995) Are the neurons in the dorsal root ganglion Harding AJ, Halliday GM, Cullen K (1994) Practical consider- pseudounipolar? A comparison of the number of neurons and ations for the use of the optical dissector in estimating neuro- number of myelinated and unmyelinated fibres in the dorsal nal number. J Neurosci Methods 51:83Ð89 root. J Comp Neurol 357:341Ð347 318 Terenghi G (1999) Peripheral nerve regeneration and neurotrophic Whiteside G, Cougnon N, Hunt SP, Munglani R (1998a) An im- factors. J Anat 194:1Ð14 proved method for detection of apoptosis in tissue sections Vestergaard S, Tandrup T, Jakobsen J (1997) Effect of permanent and cell culture, using the TUNEL technique combined with axotomy on number and volume of dorsal root ganglion cell Hoechst stain. Brain Res Brain Res Protoc 2:160Ð164 bodies. J Comp Neurol 388:307Ð312 Whiteside G, Doyle CA, Hunt SP, Munglani R (1998b) Differen- West MJ (1993) New stereological methods for counting neurons. tial timecourse of neuronal and glial apoptosis in neonatal rat Neurobiol Ageing 14:275Ð285 dorsal root ganglia after sciatic nerve axotomy. Eur J Neurosci West MJ (1999) Stereological methods for estimating the total [Suppl] 10:3400Ð3408 number of neurons and synapses: issues of precision and bias. Wyllie AH (2000) Glucocorticoid-induced thymocyte apoptosis is Trends Neurosci 22:51Ð61 associated with endogenous endonuclease activition. Nature Whiteside G, Munglani R (1998) TUNEL, hoechst and immuno- 284:555 histochemistry triple-labelling: an improved method for detec- Ygge J (1989) Neuronal loss in lumbar dorsal root ganglia after tion of apoptosis in tissue sections-an update. Brain Res Brain proximal compared with distal sciatic nerve resection: a quan- Res Protoc 3:52Ð53 titative study in the rat. Brain Res 478:193Ð195 Exp Brain Res (2002) 145:182Ð189 DOI 10.1007/s00221-002-1100-2

RESEARCH ARTICLE

Andrew McKay Hart á Mikael Wiberg á Mike Youle Giorgio Terenghi Systemic acetyl-L-carnitine eliminates sensory neuronal loss after peripheral axotomy: a new clinical approach in the management of peripheral nerve trauma

Received: 24 November 2001 / Accepted: 7 March 2002 / Published online: 4 May 2002 © Springer-Verlag 2002

Abstract Several hundred thousand peripheral nerve in- ALCAR 2%, P<0.013). Two months after axotomy the juries occur each year in Europe alone. Largely due to neuroprotective effect of high-dose ALCAR treatment the death of around 40% of primary sensory neurons, was preserved for both TUNEL counts (no treatment sensory outcome remains disappointingly poor despite five/group; high-dose ALCAR one/group) and neuron considerable advances in surgical technique; yet no clini- loss (no treatment 35%; high-dose ALCAR Ð4%, cal therapies currently exist to prevent this neuronal P<0.001). These results provide further evidence for the death. Acetyl-L-carnitine (ALCAR) is a physiological role of mitochondrial bioenergetic dysfunction in post- peptide with roles in mitochondrial bioenergetic func- traumatic sensory neuronal death, and also suggest that tion, which may also increase binding of nerve growth acetyl-L-carnitine may be the first agent suitable for clin- factor by sensory neurons. Following unilateral sciatic ical use in the prevention of neuronal death after periph- nerve transection, adult rats received either one of two eral nerve trauma. doses of ALCAR or sham, or no treatment. Either 2 weeks or 2 months later, L4 and L5 dorsal root ganglia Keywords TUNEL á Stereology á Nerve injury á were harvested bilaterally, in accordance with the Ani- Pharmacotherapy á Dorsal root ganglion mal (Scientific Procedures) Act 1986. Neuronal death was quantified with a combination of TUNEL [TdT (terminal deoxyribonucleotidyl transferase) uptake nick end Introduction labelling] and neuron counts obtained using the optical disector technique. Sham treatment had no effect upon In Europe the overall incidence of peripheral nerve trau- neuronal death. ALCAR treatment caused a large reduc- ma is 1/1,000, giving an estimated 300,000 cases/year, tion in the number of TUNEL-positive neurons 2 weeks although locally incidences are higher in regions with a after axotomy (sham treatment 33/group; low-dose greater preponderance of heavy industry. Even with opti- ALCAR 6/group, P=0.132; high-dose ALCAR 3/group, mal surgical repair, cutaneous innervation will remain re- P<0.05), and almost eliminated neuron loss (sham treat- duced (Wiberg et al. 2001), normal sensation is seldom ment 21%; low-dose ALCAR 0%, P=0.007; high-dose reattained (Glickman and Mackinnon 1990), and clinically the sensory outcome remains very poor (Dagum A.M. Hart á G. Terenghi (✉) 1998; Glickman and Mackinnon 1990). This also im- Blond-McIndoe Centre, pacts adversely upon motor function, particularly fine Royal Free & University College Medical School, manipulative work (Burstedt et al. 2001), since adequate University Department of Surgery, Royal Free Campus, sensory feedback is a vital component of normal proprio- Rowland Hill Street, London NW3 2PF, UK e-mail: [email protected] ception and therefore of motor control (Westling and Tel.: +44-20-77943944, Fax: +44-20-74314528 Johansson 1984). Peripheral nerve trauma is therefore a major cause of morbidity (Dagum 1998), and carries a A.M. Hart á M. Wiberg Department of Surgical & Perioperative Science, high cost in healthcare, lost employment, and social dis- Umeå University, Umeå, Sweden ruption. Experimentally induced peripheral nerve injuries re- M. Wiberg Department of Integrative Medical Biology, Umeå University, sult in the death of around 40% of primary sensory neu- Umeå, Sweden rons (Hart et al. 2002; Terenghi 1999; Vestergaard et al. M. Youle 1997), and postmortem studies have suggested that a Royal Free Centre for HIV Medicine, Royal Free Hospital, similar magnitude of loss occurs clinically (Suzuki et al. London, UK 1993). Although various factors are implicated (Fu and 183 Gordon 1997; Lundborg et al. 1994), it is likely that the 1998), changes which may be mediated by increased NO death of a large proportion of the sensory neuronal pool synthesis (Brorson et al. 1999; Estevez et al. 1998). The is the single most important determinant of the poor sen- preservation of these mechanisms within different neu- sory outcome of peripheral nerve injuries, since quality ron subpopulations means that a therapeutic agent acting of sensation depends upon the number of innervating at the signalling, or effector, level of the cell death pro- neurons and the size of their sensory fields (Fu and cess, particularly one addressing mitochondrial dysfunc- Gordon 1997; Terenghi 1999). Indeed, “the first prereq- tion (Budd and Nicholls 1998), ought to be protective for uisite for axonal regeneration is the survival of the neu- a wide range of neuron subtypes. One agent which may ron following injury” (Fu and Gordon 1997). fulfil such a role is acetyl-L-carnitine (ALCAR). Antidromic electrical activity and neurotoxic cyto- ALCAR is the acetyl ester of L-carnitine, and is the kines from the site of injury trigger pathways both for re- predominant acylcarnitine in human tissues where both generation and for cell death within the injured neurons peptides have physiological roles in facilitating the trans- (Ambron and Walters 1996). However, the magnitude of port of long-chain free fatty acid molecules across inner neuronal death is principally determined by the resulting mitochondrial membranes (Bremer 1990). In addition, loss of target derived neurotrophic support, as suggested ALCAR is an acetyl-group donor in oxidative glycolysis by the protective effects of exogenous neurotrophic fac- (Colucci and Gandour 1988), has antioxidant activity tor administration after experimental nerve injuries (Fu (Tesco et al. 1992), and in cultured neurons it increases and Gordon 1997; Ljungberg et al. 1999; Terenghi the expression and affinity of neurotrophin receptors for 1999). Yet despite over 30 years of research, neurotro- nerve growth factor (NGF) (Angelucci et al. 1988; phic factors have failed to find a mainstream clinical ap- Manfridi et al. 1992; Taglialatela et al. 1991). In vitro plication for a variety of reasons. Neurotrophic factors ALCAR is neuroprotective (Manfridi et al. 1992; act on specific subpopulations of primary sensory neu- Taglialatela et al. 1991; Virmani et al. 1995) and pre- rons (Terenghi 1999) and so to prevent death in all sub- vents mitochondrial uncoupling (Virmani et al. 1995), groups a complex cocktail would be required, although whilst in vivo it increases nerve regeneration after pe- exogenous neurotrophic factors may interact to negate ripheral nerve lesions (Fernandez et al. 1989). ALCAR is their own efficacy (Novikova et al. 2000a). Furthermore, currently showing promising results in clinical trials for efficacy may vary unpredictably depending upon the ex- diabetic neuropathy (De Grandis et al. 1995) and HIV- act timing of administration (Novikova et al. 2000b), and associated peripheral neuropathy, where regeneration of there are concerns over potential carcinogenic actions neurites into target tissues has been demonstrated (Hart and toxicity in other cell types (Soilu-Hanninen et al. et al. 2000). 1999), and difficulties with systemic administration exist Experimentally, neuronal death can be detected by due to side effects (Giammusso et al. 1996) and compli- seeking evidence of individual cell death using tech- cations at injection sites (McArthur et al. 2000). As a re- niques such as TdT (terminal deoxyribonucleotidyl trans- sult most experimental studies of local administration ferase) uptake nick end labelling (TUNEL) (Gavrieli et have employed continuous intrathecal administration al. 1992). More recently, stereological techniques (such (Ljungberg et al. 1999), even though this may not be as the optical disector) have provided systems for obtain- clinically applicable (Eriksdotter et al. 1998). There is ing statistically unbiased neuron counts, meaning that therefore a need for an easily administered, clinically counts are not assumption based, and that the mean of all safe therapy which will prevent sensory neuron death possible estimates tends infinitely toward the true value within a range of subpopulations. (West 1999). By then comparing the number of surviving In contrast to the specificity of neurotrophic factors, neurons present in axotomised ganglia to that in unoper- there appears to be relative homogeneity between neuron ated controls, neuron loss can effectively be quantified. subpopulations in the way that the initiation of cell death This study therefore investigated the effect of system- is signalled, and then effected (Ambron and Walters ic ALCAR treatment upon primary sensory neuron death 1996; Koliatsos and Ratan 1999). Signalling is effected after peripheral nerve injury, by using a combination of principally through the mitogen activated protein kinase TUNEL staining and neuron counting by the optical di- (MAPK), phosphatidyl inositol-3-kinase (PI3K) and sector technique. c-jun kinase (JNK) pathways (Ambron and Walters 1996; Casaccia-Bonnefil et al. 1999), leading to transcription of both cell death and regenerative genes. Ulti- Materials and methods mately the ratio of these gene products, for example the ratio of pro-apoptotic Bax to anti-apoptotic Bcl gene All work was performed in accordance with the terms of the Ani- mals (Scientific Procedures) Act 1986. Under halothane (May & products (Gillardon et al. 1996), appears to determine Baker Ltd., UK) anaesthesia, groups of young adult male Sprague- the fate of the cell, and intracellular calcium levels be- Dawley rats underwent unilateral sciatic nerve division at the up- come predictive of cell death (Tong et al. 1996). Finally, per border of quadratus femoris (n=6/group). Both nerve stumps mitochondria play an essential role (Al-Abdulla and were ligated, and secured into silicon caps with single 9/0 Ethilon Martin 1998) through the failure of oxidative metabo- sutures in order to prevent spontaneous regeneration. Parenteral systemic therapy with acetyl-L-carnitine (ALCAR, lism, the generation of free oxygen species, and the re- SigmaTau Pharmaceuticals, Italy) in sterile normal saline was lease of pro-apoptotic molecules (Budd and Nicholls commenced in the immediate postoperative period; two groups re- 184 ceived 50 mg/kg/day (“high-dose treatment”) and one group received 10 mg/kg/day (“low-dose treatment”). A further group of Results animals received an equivalent volume of normal saline (“sham treatment”), and all animals were dosed until termination. In two No adverse effects upon animal behaviour were found, other groups of animals no treatment was given (“no treatment”). and weight gain during the study period was similar in Survival periods of either 2 weeks (all groups) or 2 months (one all groups at both 2 weeks (P=0.198) and 2 months “high-dose treatment” and one “no treatment” group) were employed. Using terminal anaesthesia (pentobarbitone 240 mg/kg), (P=0.343). animals were perfusion fixated with 4% paraformaldehyde, and the ipsilateral axotomised, and contralateral control L4 and L5 dorsal root ganglia harvested. Ganglia were postfixed in 4% paraformal- Morphology dehyde prior to equilibration in cryoprotectant solution (phosphate buffered saline, PBS, containing 15% sucrose and 0.1% sodium azide), and were then stored at Ð40¡C. Prominent morphological changes were evident in the Each entire ganglion was cut into serial 30-µm cryosections, axotomised ganglia of untreated animals (Fig. 1b), with which were then permeabilised by microwave irradiation (478 W the majority of neurons exhibiting nuclear eccentricity, for 1.5 min with a 1.5-l water load) and rapid cooling in iced PBS. and chromatolysis. Neurons displaying features of apop- TUNEL (Promega Apoptosis Detection System Fluorescein, Promega, Madison, USA) was then performed, and sections were tosis (cytoplasmic granulation, vacuolation, membrane counterstained with the non-specific nuclear stain Hoechst 33342 irregularities) were seen, as were numerous enlarged, (H33342, Sigma, Poole, UK) and with propidium iodide (PI, Sig- and grossly vacuolated neurons. Morphological findings ma, Poole, UK), which highlighted cytoplasmic morphology. Dur- were unchanged by sham treatment with normal saline; ing fluorescence microscopy each stain was visualised separately using band specific filters, whilst during optical disection the use however, low-dose ALCAR caused a qualitative im- of an Olympus WU filter allowed simultaneous visualisation of provement in neuronal morphology (Fig. 1c), which was both PI and H33342. even more pronounced after high-dose ALCAR treat- All sections were examined for the presence of TUNEL posi- ment (Fig. 1d). tive cells using fluorescence microscopy and a ×20 objective lens (Olympus BX60 microscope). Positive cells were confirmed to be neurons by further examination under a ×40 objective, when their nuclear and cytoplasmic morphology was also examined. In every TUNEL staining case the serial sections immediately before and after the section containing the TUNEL positive neuron were inspected to ensure The effect of ALCAR treatment upon the number of that each positive neuron was only counted once. Counts were ex- TUNEL positive neurons, and the neuron loss found pressed as the total number of positive neurons in the L4 and L5 dorsal root ganglia (DRG) of each experimental group of six ani- 2 weeks after axotomy, is summarised in Table 1, and mals. Table 2 gives the equivalent findings 2 months after axo- The number of surviving L4 and L5 DRG neurons was estimat- tomy. TUNEL positive neurons were only found in axo- ed by the stereological technique of optical disection (Gundersen et tomised ganglia, and a range of cytoplasmic morphologies al. 1988) performed using a CAST (computer assisted stereology toolbox)-Grid (version 1) system and an Olympus BX50 micro- were found in untreated, sham treated, and ALCAR scope, as has been described previously (Hart et al. 2002). The treated animals. All exhibited nuclear morphological fea- Cavalieri principle (Gundersen et al. 1988), which states that vol- tures of cell death (chromatin condensation, and nuclear ume approximates very closely to mean cross-sectional surface ar- irregularity, or fragmentation) similar to those described ea multiplied by height, was used to calculate ganglion volume from direct measurements in a systematic random sample of ten or in a previous paper (Hart et al. 2002). Two weeks after more serial sections from the ganglion. Next, the density, or “nu- sciatic nerve transection the total number of TUNEL merical volume” (Nv), of neurons within the ganglion was deter- positive neurons present in the axotomised L4 and L5 mined from a series of CAST-Grid generated optical disector ganglia of untreated animals was 25 neurons/group of bricks, placed in a systematic random pattern within the same tis- 6 animals. Although sham treatment had no significant sue sections used for volume estimation. The estimated number of effect upon the number of TUNEL positive neurons neurons in a ganglion then equalled its volume multiplied by Nv, and was statistically unbiased since the use of systematic random (33 neurons/group, P=0.937), a significant reduction in sampling, stereological counting frames (Gundersen et al. 1988), the number of TUNEL positive neurons was found after and standard optical disector counting rules (Gundersen et al. ALCAR treatment at both the low (6 neurons/group, 1988) ensured that every neuron had had an equal chance of being counted, irrespective of size, or distribution within the ganglion. P=0.009) and high (3 neurons/group, P=0.004) doses Neuron counts were expressed as combined (L4 plus L5) used. A reduction in TUNEL positive neurons was also counts for each side (axotomised, or contralateral control). Neuron found with high-dose ALCAR 2 months after axotomy loss was calculated by subtracting the neuron count in a pair of (Table 2). axotomised ganglia from that of their contralateral controls, and mean loss was then expressed as a percentage of the mean neuron count in the control ganglia. Statistical analysis of results was performed using the Sigmastat 2.0 software package. Comparison be- Neuronal loss tween multiple groups was performed using ANOVA, and pairwise comparisons were made using Student’s t-test (neuron counts Two weeks after axotomy the mean number of neurons and neuronal loss), or the Mann-Whitney rank sum test when data was not normally distributed (TUNEL results), as assessed by the present in the unoperated contralateral control ganglia Kolmogorov-Smirnov test. (L4+5) of untreated animals was 29,641 (SD 2,268). This was similar to the equivalent count in sham treated (30,257, SD 3,261; P=0.61) or low-dose ALCAR treated (30,167, SD 1,877; P=0.67) animals, but lower than that 185

Fig. 1AÐD Dorsal root ganglia 2 weeks after sciatic nerve tran- in the high-dose ALCAR treated group (33,646, SD 875; section stained with the nuclear dye Hoechst 33342 (blue fluores- P<0.05). However, by 2 months after axotomy the con- cence) and the cytoplasmic dye propidium iodide (red fluorescence). Control ganglia (A) contain neurons with normal cytoplas- tralateral control ganglia of untreated animals (31,545, mic morphology, surrounded by numerous smaller satellite cells, SD 2,446) once again contained a similar number of neu- in which nuclear staining predominates. Two weeks after sciatic rons to those in the high-dose ALCAR treated group nerve transection (B) gross morphological changes are evident, (30,560, SD 4,294; P=0.36). with most neurons displaying nuclear eccentricity, membrane irregularities, and cytoplasmic granulation or vacuolation (hollow In the animals receiving no treatment, a loss of 21% arrow). Apoptotic neurons are also evident (solid arrows). Mor- of all L4 and L5 neurons (axotomised mean 23,317, SD phology is somewhat improved after low-dose ALCAR treatment, 2,146; unoperated control mean 29,641, SD 2,267) oc- although nuclear eccentricity and membrane irregularity remain curred during the first 2 weeks after axotomy. This rose frequent (C). After high-dose ALCAR treatment, neuronal mor- to 35% after 2 months (axotomised mean 20,460, SD phology is significantly more normal (D). Scale bar 40 µm

Table 1 Effect of acetyl-L-carnitine upon neuron death 2 weeks counts are stated as the mean (SD) count per experimental group. after axotomy. TUNEL positive neuron counts are expressed as Neuron loss is calculated as the mean neuron count in the axo- the total number of positive neurons present in the axotomised or tomised ganglia subtracted from that in the control ganglia, ex- contralateral control ganglia of each experimental group (n=6 ani- pressed as a percentage of the control value mals/group). Axotomised and contralateral control ganglia neuron

Treatment group Control ganglia (L4+5) Axotomised ganglia (L4+5) Percentage neuron mean (SD) mean (SD) loss (L4+5)

TUNEL Neuron count TUNEL Neuron count count count

No treatment 0 29,641 (2,268) 25 23,317 (2,146) 21% Sham treatment 0 30,257 (3,261) 33 23,958 (2,510) 21% Low dose ALCAR (10 mg/kg/day) 0 30,167 (1,877) 6* 30,270** (2,100) 0%** High dose ALCAR (50 mg/kg/day) 0 33,646* (875) 3* 33,077** (2,410) 2%*

* P<0.05, **P<0.001 (compared to no treatment) 186

Table 2 Effect of acetyl-L-carnitine upon neuron death 2 months counts are stated as the mean (SD) count per experimental group. after axotomy. TUNEL positive neuron counts are expressed as Neuron loss is calculated as the mean neuron count in the axo- the total number of positive neurons present in the axotomised or tomised ganglia subtracted from that in the control ganglia, ex- contralateral control ganglia of each experimental group (n=6 ani- pressed as a percentage of the control value mals/group). Axotomised and contralateral control ganglia neuron

Treatment group Control ganglia (L4+5) Axotomised ganglia (L4+5) Percentage neuron mean (SD) mean (SD) loss (L4+5)

TUNEL Neuron count TUNEL Neuron count count count

No treatment 0 31,545 (2,446) 5 20,460 (2,138) 35% High dose ALCAR (50 mg/kg/day) 0 30,560 (4,294) 1 31,642* (3,775) Ð4%**

*P<0.05, **P<0.001 (compared to no treatment)

2,138; control mean 31,545, SD 2,446). At 2 weeks after counts were used since this comprises 98.4% of the en- axotomy neuron loss was not affected by sham treatment tire sciatic nerve pool (Swett et al. 1991). Also, the (21% loss, P=0.99), but was found to have been almost greater variance present in counts from individual con- completely eliminated by ALCAR treatment (low-dose trol ganglia (L4 or L5) means that individual counts are ALCAR 0% loss, P=0.001; high-dose ALCAR 2% loss, less sensitive as indicators of the effect of axotomy (Hart P=0.006). This effect was preserved at 2 months after et al. 2002). axotomy (high-dose ALCAR Ð4% loss, P<0.001). As a marker of DNA fragmentation TUNEL staining detects the beginning of an individual neuron’s death and therefore precedes the cell’s subsequent elimination, Discussion which will be reflected by neuron loss. TUNEL staining is therefore a more sensitive marker of cell death, since Methodology very small neuron losses may be entirely masked by the normal biological variability in neuron numbers. Using As a marker of DNA fragmentation, TUNEL has been the same rat sciatic nerve model, we have previously demonstrated to be a reliable marker of cell death, demonstrated that, in the absence of any intervention, whether by apoptosis or by necrosis, particularly when numbers of TUNEL positive neurons reach a peak combined with microwave tissue permeabilisation 2 weeks after peripheral axotomy. Whilst significant (Negoescu et al. 1997). In addition, Hoechst 33342 nu- neuron loss has begun by this timepoint, the majority of clear staining has been found to provide useful morpho- neuron loss is not completed until 2 months after injury logical confirmation of cell death (Crowe et al. 1997). (Hart et al. 2002). Hence, the 2-week timepoint reveals The typically small number of TUNEL positive neurons the effect of ALCAR treatment upon the peak of neuron found has previously been ascribed to the short time pe- death after peripheral nerve injury, whilst the 2-month riod during cell death in which cells are potentially timepoint better reveals whether or not any cytoprotec- TUNEL positive (Groves et al. 1997; Hart et al. 2002). tive effect is a lasting one, and what the magnitude of Despite recent criticism (Benes and Lange 2001), op- that effect might be. tical disection is an ideal technique for estimation of ALCAR levels were not measured in this study since neuron numbers (Hart et al. 2002). Indeed it avoids the both parenteral and enteral administration of ALCAR statistical bias of assumption based techniques (West have previously been demonstrated to increase plasma 1999) and the effects of tissue sectioning artefacts, yet is and CNS levels (Parnetti et al. 1992). Also, in diabetic time efficient (Gundersen et al. 1988; Pover and Cogge- rats a dose of 50 mg/kg/day by intraperitoneal injection shall 1991). The disector technique has been validated was sufficient to normalise sciatic nerve ALCAR content for counting DRG neurons (Pover and Coggeshall 1991) (Stevens et al. 1996). although precision and accuracy are dependent upon adherence to certain rules (West 1999). Accordingly, ten or more sections whose thickness was measured directly ALCAR treatment reduces neuron death and eliminates were used for volume estimation by the Cavalieri princi- neuron loss after peripheral nerve transection ple (Gundersen et al. 1988), and adequate numbers of caps (158/ganglion, SD 25.8) were counted to meet the Systemic treatment with ALCAR (50 mg/kg/day) requirement of 100Ð200/ganglion (West 1999). The brought about a highly significant reduction in both the mean neuron counts obtained from control ganglia number of TUNEL positive neurons and the neuron loss (30,969, SD 2,858) were in agreement with other results found 2 weeks after peripheral axotomy. Neuron death in the literature obtained by stereological methods was prevented, rather than just delayed, since this pro- (Bergman and Ulfhake 1998; Tandrup 1993), verifying tective effect was found to be preserved 2 months after the accuracy of the technique used. Combined L4+5 axotomy, by which time neuron loss has effectively fin- 187 ished in the absence of treatment (Hart et al. 2002). Even effect of increased cellular metabolic capacity. ALCAR at the reduced dose of 10 mg/kg/day ALCAR therapy also facilitates the transport of cytosolic long-chain free was still significantly effective in preventing neuronal fatty acids across mitochondrial membranes and pro- death at 2 weeks; however, morphological changes were vides acetyl groups for maintenance of high energy sub- more evident, and the reduction in the number of strate aerobic glycolysis (Bremer 1990; Colucci and TUNEL positive neurons was less, demonstrating a dose Gandour 1988), in preference to the normal shift into response effect on neuronal survival. The low-dose re- the less efficient pentose phosphate shunt (Singer and gime was not therefore used in the 2-month groups, al- Mehler 1986). ALCAR may therefore prevent failure of though future studies are planned to quantify the long- mitochondrial oxidative metabolism, an event that is in- term neuroprotective effect of low-dose ALCAR treat- tegral to the cell death pathway (Budd and Nicholls ment, since the morphological and TUNEL results 1998). This is at a time when energy requirements are in- 2 weeks after axotomy imply that long-term neuron loss creased in support of the regenerating neurites (Singer might be greater than with high-dose treatment. and Mehler 1986), and this effect has been demonstrated Since TUNEL positive neurons were present in in vitro (Virmani et al. 1995). Finally, the antioxidant role ALCAR treated ganglia, neuronal death cannot have of ALCAR may be of benefit in preventing cell damage been completely eliminated, although it was reduced suf- by the reactive oxygen species (Tesco et al. 1992), whose ficiently that when all three treatment groups are consid- generation is thought to be of major importance in the ef- ered together no neuron loss was found to have occurred fector mechanism of cell death (Estevez et al. 1998). (control count 31,457, SD 3,041; operated count 31,663; Experimentally, ALCAR enhances sensory neuron SD 2,931). This represents salvage of around 6,000 neu- survival in vitro (Manfridi et al. 1992; Taglialatela et al. rons/animal at 2 weeks, and over 10,000 neurons at 1991) and has been demonstrated to increase motor 2 months after axotomy, which is a third of the total sen- nerve regeneration after peripheral axotomy (Fernandez sory neuron pool for the sciatic nerve. The lack of any et al. 1989). It has been postulated that in neonates this apparent dose response in neuron loss after 2 weeks of may be due to recruitment of ectopic motoneurones, or ALCAR treatment may probably be due to the small dif- to increased motoneuronal survival (Fernandez et al. ference in the number of dying neurons (TUNEL results) 1990). Clinically, ALCAR has an excellent safety pro- between dose levels, and to the fact that 2 weeks is a rel- file, and is under investigation as a treatment for diabetic atively early timepoint for minor differences in rates of neuropathy, where enhanced pudendal nerve function has neuronal death to translate into differences in neuron loss. been demonstrated (Giamusso et al. 1996). In a group of The reduction in calculated neuron loss was not mere- 426 neuropathic patients, significant increases in con- ly an artefact of changes in the number of neurons in duction velocity have been demonstrated in the motor control ganglia, since ALCAR treatment did not appear nerves of those with mononeuropathy, and in the sensory to have any effect upon the number of neurons in control nerves of those with either mono- or polyneuropathies ganglia. The minor increase found with high-dose (De Grandis et al. 1995). In HIV-associated peripheral ALCAR at 2 weeks after axotomy is likely to be an arte- neuropathy ALCAR treatment can improve neuropathic fact of the unusually low standard deviation in this pain symptoms (Scarpini et al. 1995) and cause signifi- group, rather than a real effect of treatment, since the cant regeneration of cutaneous nerve fibres (particularly mean lies well within the normal distribution of counts of small sensory neurons) into the sweat glands, dermis, for untreated and sham treated animals. Finally, it is and epidermis (Hart et al. 2000). This is in contrast to the evident that the reduction in neuron death was due to lack of convincing evidence for a therapeutic effect of ALCAR rather than the act of drug administration, since L-carnitine in a range of conditions, and to the recent 2 weeks after axotomy sham injections of sterile normal failure of systemic recombinant nerve growth factor saline had had no effect upon either neuron loss or the (rhNGF) in diabetic neuropathy (Apfel et al. 2000). number of TUNEL positive neurons (Table 1). Evidence would therefore suggest that when administered either orally or parenterally, ALCAR is a clinically safe drug that can enhance the regenerative capacity of Mechanism of action injured neurons in vivo, and the results presented herein are the first demonstration that ALCAR can also be sig- The precise mechanism by which ALCAR prevents pri- nificantly neuroprotective after peripheral nerve trauma. mary sensory neuron death after peripheral axotomy re- This suggests that ALCAR may indeed be the first effec- mains to be determined, although it is likely that its active neuroprotective agent that is clinically suitable for tivity is due to attenuation of both the initiator and effec- the adjuvant pharmacotherapy of major peripheral nerve tor mechanisms of cell death. trauma such as brachial plexus injuries. ALCAR may attenuate the loss of distal neurotrophic Further work needs to be undertaken to clarify the re- support by enhancing neurotrophin responsiveness, since quired duration of treatment with ALCAR, and its pre- it can increase neuronal NGF binding capacity (Angelucci cise mechanism of action, but if the postulated mito- et al. 1988) and potentiate the effects of NGF (Manfridi chondrial mechanism is confirmed then ALCAR may be et al. 1992; Taglialatela et al. 1991). However, it is not equally effective in other forms of peripheral and central known if this is a direct effect of ALCAR, or an indirect nervous system trauma. 188 Acknowledgements The authors are grateful for the considerable Fernandez E, Pallini PR, Gangitano C, Del Fa A, Olivieri- support of Dr. Thomas Brannstrom of the Department of Medical Sangiacomo C, Sbriccoli A, Ricoy J, Rossi GF (1990) Studies Biosciences Ð Pathology, Umeå University, and for the unqualified on the degenerative and regenerative phenomena occurring af- donation of acetyl-L-carnitine (anhydrous powder) by Sigma Tau ter transection and repair of the sciatic nerve in rats: effects of Pharmaceuticals, Italy. This work was funded by the Swedish acetyl-L-carnitine. Int J Clin Pharm Res 10:85Ð99 Medical Research Council (grant 02286), County of Vasterbottens, Fu SY, Gordon T (1997) The cellular and molecular basis of pe- Ane’rs Foundation, and the East Grinstead Medical Research ripheral nerve regeneration. Mol Neurobiol 14:67Ð116 Trust (Registered Charity No. 258154). Gavrieli Y, Sherman Y, Ben-Sasson SA (1992) Identification of programmed cell death in situ via specific labeling of nuclear DNA fragmentation. J Cell Biol 119:493Ð501 Giammusso B, Morgia G, Spampinato A, Motta M (1996) Im- References proved pallesthetic sensitivity of pudendal nerve in impotent diabetic patients treated with acetyl-L-carnitine. Acta Urol Ital Al-Abdulla NA, Martin LJ (1998) Apoptosis of retrogradely de- 10:185Ð187 generating neurons occurs in association with the accumula- Gillardon F, Klimaschewski L, Wickert H, Krajewski S, Reed JC, tion of perikaryal mitochondria and oxidative damage to the Zimmermann M (1996) Expression pattern of candidate cell nucleus. Am J Pathol 153:447Ð456 death effector proteins Bax, Bcl- 2, Bcl-X, and c-Jun in senso- Ambron RT, Walters ET (1996) Priming events and retrograde in- ry and motor neurons following sciatic nerve transection in the jury signals. A new perspective on the cellular and molecular rat. Brain Res 739:244Ð250 biology of nerve regeneration. Mol Neurobiol 13:61Ð79 Glickman LT, Mackinnon SE (1990) Sensory recovery following Angelucci L, Ramacci MT, Taglialatela G, Hulsebosch C, Morgan digital replantation. Microsurgery 11:236Ð242 B, Werrbach-Perez K, Perez-Polo JR (1988) Nerve growth Groves MJ, Christopherson T, Giometto B, Scaravilli F (1997) factor binding in aged rat central nervous system: effect of Axotomy-induced apoptosis in adult rat primary sensory neu- acetyl-L-carnitine. J Neurosci Res 20:491Ð496 rons. J Neurocytol 26:615Ð624 Apfel SC, Schwartz S, Adornato BT, Freeman R, Biton V, Rendell Gundersen HJG, Bagger P, Bendtsen TF, Evans SM, Korbo L, M, Vinik A, Giuliani M, Stevens JC, Barbano R, Dyck PJ Marcussen N, Moller A, Nielsen K, Nyengaard JR, Pakkenberg (2000) Efficacy and safety of recombinant human nerve B, Sorensen FB, Vesterby A, West MJ (1988a) The new ste- growth factor in patients with diabetic polyneuropathy: a ran- reological tools: disector, fractionator, nucleator and point domized controlled trial. JAMA 284:2215Ð2221 sampled intercepts and their use in pathological research and Benes MB, Lange N (2001) Two-dimensional versus three-dimen- diagnosis. APMIS 96:857Ð881 sional cell counting: a practical approach. Trends Neurosci 24: Gundersen JG, Bendtsen TF, Korbo L, Marcussen N, Moller A, 11Ð17 Nielsen K, Nyengaard JR, Pakkenberg B, Sorensen FB, Bergman E, Ulfhake B (1998) Loss of primary sensory neurons in Vesterby A, West MJ (1988b) Some new, simple and efficient the very old rat: neuron number estimates using the disector stereological methods and their use in pathological research method and confocal optical sectioning. J Comp Neurol 396: and diagnosis. APMIS 96:379Ð394 211Ð222 Hart AM, Terenghi G, Johnson MA, Youle M (2000) Immuno- Bremer J (1990) The role of carnitine in intracellular metabolism. histochemical quantification of cutaneous innervation in J Clin Chem Clin Biochem 28:297Ð301 HIV-associated peripheral neuropathy: a study of acetyl-L- Brorson JR, Schumacker PT, Zhang H (1999) Nitric oxide acutely carnitine therapy. Abstract 36. 3rd International Workshop on inhibits neuronal energy production. The Committees on Neu- Salvage for HIV Infection, Chicago 2000. Antivir Ther robiology and Cell Physiology. J Neurosci 19:147Ð158 5(Suppl 2):32 Budd SL, Nicholls DG (1998) Mitochondria in the life and death Hart AM, Brannstrom T, Wiberg M, Terenghi G (2002) Primary of neurons. Essays Biochem 33:43Ð52 sensory neurons and satellite cells after peripheral axotomy in Burstedt MKO, Schenker M, Wiberg M, Johansson R (2001) Pre- the adult rat: timecourse of cell death and elimination. Exp cision grip function in patients after hand replantation. Scand J Brain Res 142:308Ð318 Plast Recon Hand Surg (in press) Koliatsos VE, Ratan RR (1999) Cell death and diseases of the ner- Casaccia-Bonnefil P, Gu C, Chao MV (1999) Neurotrophins in vous system. Humana, Totowa, NJ cell survival/death decisions. Adv Exp Med Biol 468:275Ð282 Ljungberg C, Novikov L, Kellerth JO, Ebendal T, Wiberg M Colucci WJ, Gandour RD (1988) Carnitine acyltransferase: a re- (1999) The neurotrophins NGF and NT-3 reduce sensory neu- view of its biology, enzymology and bioorganic chemistry. ronal loss in adult rat after peripheral nerve lesion. Neurosci Bioorg Chem 16:307Ð334 Lett 262:29Ð32 Crowe MJ, Bresnahan JC, Shuman SL, Masters JN, Beattie MS Lundborg G, Dahlin L, Danielsen N, Zhao Q (1994) Trophism, (1997) Apoptosis and delayed degeneration after spinal cord tropism, and specificity in nerve regeneration. J Reconstr injury in rats and monkeys. Nature Med 3:73Ð76 Microsurg 10:345Ð354 Dagum AB (1998) Peripheral nerve regeneration, repair, and Manfridi A, Forloni GL, Arrigoni-Martelli E, Mancia M (1992) grafting. J Hand Ther 11:111Ð117 Culture of dorsal root ganglion neurons from aged rats: ef- De Grandis D, Santoro L, Di Benedetto P (1995) L-Acetylcarnitine fects of acetyl-L-carnitine and NGF. Int J Dev Neurosci 10: in the treatment of patients with peripheral neuropathies. Clin 321Ð329 Drug Invest 10:317Ð322 McArthur JC, Yiannoutsos C, Simpson D, Adornato BT, Singer Eriksdotter JM, Nordberg A, Amberla K, Backman L, Ebendal T, EJ, Hollander H, Marra CM, Rubin M, Cohen B, Tucker T, Meyerson B, Olson L, Seiger, Shigeta M, Theodorsson E, Navia BA, Schifitto G, Katzenstein DA, Rask CA, Zaborski L, Viitanen M, Winblad B, Wahlund LO (1998) Intracerebroven- Smith SE, Shriver S, Millar L, Clifford DB, AIDS Clinical tricular infusion of nerve growth factor in three patients with Trials Group Team 291 (2000) A phase II trial of nerve growth Alzheimer’s disease. Dement Geriatr Cogn Disord 9:246Ð257 factor for sensory neuropathy associated with HIV infection. Estevez AG, Spear N, Manuel SM, Radi R, Henderson CE, Neurology 54:1080Ð1088 Barbeito L, Beckman JS (1998) Nitric oxide and superoxide Negoescu A, Lorimier P, Labat-Moleur F, Azoti L, Robert C, contribute to motor neuron apoptosis induced by trophic factor Guillermet C, Brambilla E (1997) TUNEL: improvement and deprivation. J Neurosci 18:923Ð931 evaluation of the method for in situ apoptotic cell identifica- Fernandez E, Pallini PR, Gangitano C, Aurora DF, Sangiacomo tion. Biochemica 2:12Ð17 CO, Sbriccoli A, Ricoy J, Rossi GF (1989) Effects of L-carni- Novikova LN, Novikov LN, Kellerth J-O (2000a) BDNF abolish- tine, L-acetylcarnitine and gangliosides on the regeneration of es the survival effect of NT-3 in axotomised Clarke’s nucleus the transected sciatic nerve in rats. Neurol Res 11:57Ð62 of adult rats. J Comp Neurol 428:671Ð680 189 Novikova LN, Novikov LN, Kellerth J-O (2000b) Survival effects Taglialatela G, Angelucci L, Ramacci MT, Werrbach-Perez K, of BDNF and NT-3 on axotomized rubrospinal neurons de- Jackson GR, Perez-Polo JR (1991) Acetyl-L-carnitine enhanc- pend on the temporal pattern of neurotrophin administration. es the response of PC12 cells to nerve growth factor. Brain Eur J Neurosci 12:776Ð780 Res Dev Brain Res 59:221Ð230 Parnetti L, Gaiti A, Mecocci P, Cadini D, Senin U (1992) Pharma- Tandrup T (1993) A method for producing unbiased and efficient cokinetics of IV and oral acetyl-L-carnitine in a multiple dose estimation of number and mean volume of specified neuron regimen in patients with senile dementia of Alzheimer type subtypes in rat dorsal root ganglion. J Comp Neurol 329:269Ð [published erratum appears in Eur J Clin Pharmacol 1993; 276 44(6):604]. Eur J Clin Pharmacol 42:89Ð93 Terenghi G (1999) Peripheral nerve regeneration and neurotrophic Pover CM, Coggeshall RE (1991) Verification of the disector factors. J Anat 194:1Ð14 method for counting neurons, with comments on the empirical Tesco G, Latorraca S, Piersanti P, Piacentini S, Amaducci L, Sorbi method. Anat Rec 231:573Ð578 S (1992) Protection from oxygen radical damage in human Scarpini E, Sacilotto G, Baron P, Cusini M, Scarlato G (1997) diploid fibroblasts by acetyl-L-carnitine. Dementia 3:58Ð60 Effect of acetyl-L-carnitine in the treatment of painful periph- Tong JX, Eichler ME, Rich KM (1996) Intracellular calcium lev- eral neuropathies in HIV+ patients. J Peripher Nerv Syst els influence apoptosis in mature sensory neurons after trophic 2:250Ð252 factor deprivation. Exp Neurol 138:45Ð52 Singer P, Mehler S (1986) Glucose and leucine uptake in the hy- Vestergaard S, Tandrup T, Jakobsen J (1997) Effect of permanent poglossal nucleus after hypoglossal nerve transection with and axotomy on number and volume of dorsal root ganglion cell without prevented regeneration in the Sprague-Dawley rat. bodies. J Comp Neurol 388:307Ð312 Neurosci Lett 67:73Ð77 Virmani MA, Biselli R, Spadoni A, Rossi S, Corsico N, Calvani Soilu-Hanninen M, Ekert P, Bucci T, Syroid D, Bartlett PF, M, Fattorossi A, De SC, Arrigoni-Martelli E (1995) Protective Kilpatrick TJ (1999) Nerve growth factor signaling through actions of L-carnitine and acetyl-L-carnitine on the neurotoxic- p75 induces apoptosis in Schwann cells via a Bcl-2-indepen- ity evoked by mitochondrial uncoupling or inhibitors. Pharma- dent pathway. J Neurosci 19:4828Ð4838 col Res 32:383Ð389 Stevens MJ, Lattimer SA, Feldman EL, Helton ED, Millington West MJ (1999) Stereological methods for estimating the total DS, Sima AA, Greene DA (1996) Acetyl-L-carnitine deficien- number of neurons and synapses: issues of precision and bias. cy as a cause of altered nerve myo-inositol content, Na, Trends Neurosci 22:51Ð61 K-ATPase activity, and motor conduction velocity in the strep- Westling G, Johansson RS (1984) Factors influencing the force tozotocin-diabetic rat. Metabolism 45:865Ð872 control during precision grip. Exp Brain Res 53:277Ð284 Suzuki H, Oyanagi K, Takahashi H, Kono M, Yokoyama M, Ikuta Wiberg M, Hazari A, Ljungberg C, Petterson K, Nordh E, Kvast- F (1993) A quantitative pathological investigation of the cervi- Rabben O, Terenghi G (2002) Sensory recovery after hand re- cal cord, roots and ganglia after long-term amputation of the plantation: a clinical, morphological and neurophysiological unilateral upper arm. Acta Neuropathol 85:666Ð673 study in man. Scand J Plast Recon Hand Surg (in press) Swett JE, Torigoe Y, Elie VR, Bourassa CM, Miller PG (1991) Sen- sory neurons of the rat sciatic nerve. Exp Neurol 114:82Ð103 Neuroscience Letters 334 (2002) 181–185 www.elsevier.com/locate/neulet

Pharmacological enhancement of peripheral nerve regeneration in the rat by systemic acetyl-L-carnitine treatment

Andrew McKay Harta,b, Mikael Wibergb,c, Giorgio Terenghic,*

aBlond-McIndoe Centre, Royal Free and University College Medical School, University Department of Surgery, Royal Free Campus, Rowland Hill Street, London NW3 2PF, UK bDepartment of Surgical and Perioperative Science, Section for Hand and Plastic Surgery, University Hospital, Umea, Sweden cDepartment of Integrative Medical Biology, Section for Anatomy, Umea University, Umea, Sweden Received 24 June 2002; received in revised form 16 August 2002; accepted 16 August 2002

Abstract Peripheral nerve trauma remains a major cause of morbidity, largely due to the death of ,40% of innervating sensory neurons, and to slow regeneration after repair. Acetyl-L-carnitine (ALCAR) is a physiological peptide that virtually eliminates sensory neuronal death, and may improve regeneration after primary nerve repair. This study determines the effect of ALCAR upon regeneration after secondary nerve repair, thereby isolating its effect upon neuronal regenerative capacity. Two months after unilateral sciatic nerve division 1 cm nerve graft repairs were performed (n ¼ 5), and treatment with 50 mg/kg/day ALCAR was commenced for 6 weeks until harvest. Regeneration area and distance were determined by quantitative immunohistochemistry. ALCAR treatment significant increased immunostaining for both nerve fibres (total area 264% increase, P , 0:001; percentage area 229% increase, P , 0:001), and Schwann cells (total area 111% increase, P , 0:05; percentage area 86% increase, P , 0:05), when compared to no treatment. Regeneration into the distal stump was greatly enhanced (total area 2242% increase, P ¼ 0:008; percentage area 3034% increase, P ¼ 0:008). ALCAR significantly enhances the regenerative capacity of neurons that survive peripheral nerve trauma, in addition to its known neuroprotective effects. q 2002 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Peripheral nerve; Regeneration; Acetyl-L-carnitine; Mitochondria; Nerve repair; Quantitative immunohistochemistry

Peripheral nerve trauma, with an estimated incidence of chronic nerve injuries due to ongoing neuronal death and ,1/1000 population, remains a major cause of morbidity deleterious events in the distal nerve stump, hence these and social disruption [4], since “all qualities of sensory injuries are associated with the worst clinical outcome and motor function are difficult or impossible to regain in [7,13,19]. adult patients” [11]. Although motorneurons initially Although the technique, and timing of repair may be survive peripheral nerve injury, the death of around 40% critical [8,13,19], the greatest potential improvements in of all innervating primary sensory [8] ensures that even outcome would result from the elimination of sensory after optimal microsurgical nerve repair normal innervation neuronal death, and an increase in the rate of neurite regen- density, and so quality of sensation, can never be reattained. eration. The administration of exogenous neurotrophic The regenerating neurites of those neurons which do survive factors has given promising experimental results with this must bridge the repair site, enter into functionally (i.e. motor regard [18], yet none have been suitable for clinical trials. or sensory) and somatotopically appropriate endoneurial Acetyl-L-carnitine (ALCAR) is the predominant acyl-carni- tubes in the distal nerve stump [11], before regenerating tine in the body, has essential physiological roles in mito- to the target sensory or motor organs prior to the occurrence chondrial oxidative energy metabolism, and is clinically of significant denervation atrophy. All these challenges are safe when administered orally or parenterally [5]. In vitro compounded in the case of the late secondary repair of studies have suggested that exogenous ALCAR treatment may enhance neurite growth, and this result was confirmed in a single animal study [6]. * Corresponding author. Tel.: 144-20-7794-0500 ext. 3944; fax: 144-20-472-6711 That effect was initially presumed to be the result of an E-mail address: [email protected] (G. Terenghi). increase in the regenerative capacity of individual neurons,

0304-3940/02/$ - see front matter q 2002 Elsevier Science Ireland Ltd. All rights reserved. PII: S0304-3940(02)00982-5 182 A. McKay Hart et al. / Neuroscience Letters 334 (2002) 181–185 secondary to increased mitochondrial oxidative capacity quantified by computerised image analysis (Image-Pro- and nerve growth factor (NGF) responsiveness [1,17]. Pluse Version 4.0 software) using a standard process However it has recently been demonstrated that ALCAR based upon intensity thresholding for unichromatic light, treatment virtually eliminates sensory neuronal death after with the opportunity to deselect occasional areas of back- peripheral axotomy in vivo [9]. As a result it is now unclear ground staining prior to digital quantification of area. The whether the reported improvement in regeneration was due total area of immunostaining for that section was summated, to an increase in the regenerative potential of individual along with the total area of the nerve graft present in the neurons, or to a reduction in cell death having resulted in captured images. In order to facilitate comparison between an increased number of regenerating neurons. By using a nerve grafts of different diameters, the area of immunostain- model in which sciatic nerve repair was delayed until after ing was also expressed as a percentage of the area of the neuronal death was complete, this study isolated the effect nerve graft examined. Total area of immunostaining was of ALCAR treatment upon individual neurons, whose expressed in mm2, and percentage area was quoted directly. regenerative capacity was determined by quantitative All quantification was performed by one observer, and care immunohistochemistry. was taken to ensure that similar settings were employed for Under halothane (May & Baker Ltd.) inhalational anaes- both experimental groups during the image capturing and thesia, two groups of young adult male Sprague–Dawley thresholding process. A total of ten measurements of the rats (n ¼ 5) underwent unilateral sciatic nerve division at area (absolute or percentage) of immunostaining was there- the upper border of quadratus femoris. Two months later fore recorded for each experimental group, and the mean nerve repair was performed using a freshly harvested, value used for statistical examination of the effect of treat- syngeneic, reversed, 1 cm sciatic nerve isograft. One ment. group received ALCAR (50 mg/kg/day) by daily intraper- Statistical analysis of results was performed by paired t- itoneal injection from the time of nerve repair until termina- test, using SigmaState (Version 2.0) software, after tion 6 weeks after repair, whilst the control group was not confirming the normality of data distribution by the Kolmo- treated. Using a lethal dose (30 mg/kg) of intraperitoneal gorov–Smirnov test. sodium pentobarbitone (Sagattale, Rhoˆne Me´rieux Ltd.), In both experimental groups some axons crossed the full the repairs were harvested in continuity with ,3mm length of the repair in all cases, and reached the distal cut lengths of proximal and distal nerve stump, and were end of the distal nerve stump (ALCAR treatment group 12.9 fixed in Zamboni’s solution overnight at 4 8C, prior to mm from the proximal anastomosis, SD 0.90; control group rinsing into a cryoprotectant solution of 15% sucrose in 13.3 mm, SD 1.53). As a result no effect of ALCAR treat- phosphate buffered saline. Longitudinal 15 mm cryosections ment upon mean regeneration distance could be demon- were collected onto Vectabond (Vector Laboratories Inc.) strated. Visible gastrocnemius contraction in response to coated glass slides, and stained by fluorescence immunohis- dividing the nerve proximal to the repair was however tochemistry using primary antisera to pan-axonal marker of noted in one animal that had received treatment with neurofilament (PAMNF; monoclonal antibody, Affiniti ALCAR. Wet muscle weights for gastrocnemius were not Research Products Ltd.), and the Schwann cell marker S- significantly different between the treated group (mean 13% 100 (polyclonal antibody, DAKO Ltd.). PAMNF staining of contralateral control, SD 4.4%) and the untreated controls was visualised by a secondary antiserum conjugated to (mean 11% of contralateral control, SD 2.8%), suggesting Cy-3 (Cy-3 horse anti-mouse, Amersham Pharmacia that significant motor regeneration had not occurred by the Biotech), whilst a fluorescein isothyocyanate (goat anti- time of harvest in either group. rabbit, Vector Laboratories Inc.) conjugated secondary anti- Within the nerve grafts the total area of immunostaining serum was used for the S-100 staining. Non-specific tissue for PAMNF, and for S-100, was significantly greater in binding of the secondary antibodies was prevented by pre- ALCAR treated nerve repairs than in controls, as was the incubation of tissue sections with normal goat (Sigma- case when the mean percentage area of immunostaining was Aldrich UK) or normal horse (Sigma-Aldrich UK) sera, considered (Table 1). The total area of nerve graft present in and non-specific hydrophobic attachment was minimised the quantified sections was slightly higher in the ALCAR by co-incubation with normal rat serum (DAKO, Denmark). treated group (967046 mm2, SD 279641) than in the controls Regeneration distance was measured from the end of the (774571 mm2, SD 146813), but the difference was not statis- proximal nerve stump to the tip of the most distal nerve fibre tically significant (P ¼ 0:21). in each repair, using a 10 £ objective lens and a graticulated In the distal nerve stump, 1 mm beyond the distal anasto- eyepiece (0.1 mm graticulations). In two non-consecutive mosis, the effect of ALCAR treatment was more sections from the centre of each repair the area of immu- pronounced (Table 2). The area of immunostaining for nostaining was determined by capturing a band of images nerve fibres was significantly increased by ALCAR treat- across the full width of the graft (at a position 1 mm distal to ment (total area of immunostaining 2242% increase, the proximal anastomosis), using a high resolution fluid (P ¼ 0:008); percentage area 3034%, P ¼ 0:008), whilst cooled digital camera (Diagnostic Instruments, USA) and the area of staining for Schwann cells exhibited a more a20£ objective lens. The area of immunostaining was then modest increase (total area of staining 722% increase, A. McKay Hart et al. / Neuroscience Letters 334 (2002) 181–185 183

Table 1 Effect of ALCAR treatment upon nerve regeneration into nerve graftsa

Treatment group Total area of immunostaining (mm2) mean Percentage area of immunostaining (SD) mean (SD)

PAMNF S100 PAMNF S100

No treatment 58259 (25405) 112295 (29902) 7.3% (2.3) 14% (2.1) ALCAR (50 mg/kg/day) 212245 (41861)** 236641 (41841)** 24% (9.5)* 26% (10)*

a The mean area of immunostaining for nerve ﬁbres (PAMNF), or Schwann cells (S-100) within reversed sciatic nerve grafts harvested 6 weeks after a late secondary repair of a 1 cm gap in the rat sciatic nerve (2 month interval between transection and repair). Percentage area of immunostaining equals the total area expressed as percentage of the total area of graft within which staining was measured. *P , 0:05, **P , 0:001 compared to no treatment.

P ¼ 0:056; percentage area 816% increase, P ¼ 0:095). The ably because only ,3 mm of nerve was harvested distal to total area of distal stump present in the quantified sections the repair, and so any difference in the distance which fibres was slightly higher in the ALCAR treated group (477024 travelled beyond that point would not be detected. Such a mm2, SD 216515) than in the controls (401642 mm2,SD difference would be detected by studies using earlier time- 124590), but the difference was not statistically significant points, or by quantifying target organ reinnervation at later (P ¼ 0:49). timepoints. ALCAR did however appear to increase the rate Repairs were performed 2 months after nerve transection at which fibres crossed the distal anastomosis, and entered since sensory neuronal loss has stabilised by this time [8], the distal stump, since those of ALCAR treated repairs hence any effect of ALCAR treatment should represent an contained far more staining for PAMNF than did controls. effect upon the regenerative capacity of individual neurons, An increase in myelinated fibre counts would also demon- and not merely an increase in the size of the regenerative strate enhanced regeneration within the distal stump, and pool secondary to neuroprotection. Such a delay made end- this has previously been found after ALCAR treatment to-end epineurial repair unfeasible due to retraction of the [6]. Myelinated counts were not repeated in this study, nerve stumps, so nerve grafts were used to bridge the result- which sought merely to demonstrate whether any increase ing gap in accordance with clinical practice. Although a in regeneration would be independent of ALCAR’s neuro- relatively short (1 cm) gap was used, the aim of this study protective role, rather than to fully characterise that regen- was not to investigate regeneration across a long gap, but eration. merely to quantify the effect of ALCAR upon regeneration Treatment was shown to increase the amount of nerve across a known bridgeable gap. Further studies in larger regeneration, since within the nerve grafts the total area of animal species will be required to determine if treatment regenerating nerve fibres, as demonstrated by immunostain- has any effect upon the maximum length of gap across ing with antisera against PAMNF, was significantly greater which nerves can adequately regenerate. in the ALCAR treated group than in the untreated controls. PAMNF is a constitutive component of all mammalian This was matched by a comparable increase in the total area nerve fibres, and so the area of immunostaining with anti- of immunostaining for Schwann cells. Artifactual variations sera against it gives a reliable determination of the area of in the total area of immunostaining may arise if there are nerve fibres present. Similarly S-100 is a cytosolic compo- differences in the area of nerve graft in which staining is nent of all Schwann cells, and so it provides a means to measured, either because of actual differences in the quantify the Schwann cell response to regenerating neurites. diameters of the nerve grafts used, or of variability in the Although no increase in regeneration distance could be exact position within the graft at which the section was cut. demonstrated as a result of ALCAR treatment, this is prob- Since the nerve is cylindrical, longitudinal sections cut at

Table 2 Effect of ALCAR treatment upon nerve regeneration into the distal nerve stumpa

Treatment group Total area of immunostaining (mm2) Percentage area of immunostaining mean (SD) mean (SD)

PAMNF S100 PAMNF S100

No treatment 1711 (1599) 12541 (9786) 0.38% (0.26) 3.3% (1.9) ALCAR (50 mg/kg/day) 40066 (20619)* 103121 (21583)* 12%* (11) 30% (35)

a The mean area of immunostaining for nerve fibres (PAMNF), or Schwann cells (S-100) within the distal nerve stump 6 weeks after a late secondary repair of a 1 cm gap in the rat sciatic nerve (2 month interval between transection and repair) using a reversed sciatic nerve graft. Percentage area of immunostaining equals the total area expressed as a percentage of the total area of nerve within which staining was measured. *P , 0:05, compared to no treatment. 184 A. McKay Hart et al. / Neuroscience Letters 334 (2002) 181–185 slightly different positions through the graft will inevitably pentose-phosphate pathway [15], regeneration is enhanced have different diameters. Hence the area of immunostaining due to an increase in high energy substrate metabolism fuel- was also expressed as a percentage of the area of the graft to ling the biosynthetic pathways and axonal transport compensate for this effect, and a significant increase as a systems. The effect of ALCAR treatment to increase neuro- result of ALCAR treatment was still found for nerve fibres, nal NGF binding capacity [1], and responsiveness [17] may and for Schwann cells. Neurite ingrowth causes Schwann also be significant, given NGF’s beneficial effect upon cell proliferation [18], and this would explain why ALCAR peripheral nerve regeneration [18]. treatment also caused an increase in the percentage area of ALCAR may therefore improve peripheral nerve regen- S-100 immunostaining, particularly with regard to the eration independently of its effect to increase the number of atrophic population within the distal nerve stump. Hence neurons surviving peripheral axotomy. The regenerative ALCAR does appear to increase the capacity of surviving capacity of individual neurons may be enhanced either by neurons to regenerate after peripheral nerve repair indepen- a direct effect to support axonal growth, or by a more facil- dently of its neuroprotective effect, and also to greatly itative effect whereby ALCAR treatment boosts the regen- enhance the ability of regenerating neurites to bridge the erative response provoked by physiological prompts such as second anastomotic site and enter the distal stump. This is target derived neurotrophins. of particular clinical relevance given the potential failure of By commencing treatment immediately after nerve injury neurites to exit nerve grafts and enter the distal nerve it should be possible to limit neuronal loss, and so combine following nerve graft repairs [13]. an increase in the number of regenerating neurons, with an ALCAR was administered at 50 mg/kg/day since this is improvement in their regenerative capacity. As a result, comparable to the dose used clinically for the treatment of nerve regeneration could potentially be greatly enhanced. peripheral neuropathies, and when given by intraperitoneal The results of this study also suggest that ALCAR treatment injection was sufficient to normalise sciatic nerve ALCAR would be of clinical benefit even after late secondary nerve content in diabetic rats [16]. ALCAR levels were not repairs where neuroprotection has ceased to be of impor- measured in this study since both parenteral and enteral tance, and that treatment might best be continued until target administration have previously been demonstrated to organ reinnervation has occurred after peripheral nerve increase plasma and CNS levels [12,14]. Parenteral admin- repair. If ALCAR has the same effect upon central nervous istration was chosen in order to eliminate variability due to system neurons, then it may partly explain its effect to gastrointestinal absorption. Sham injections of saline have promote memory and ‘rejuvenate’ the ageing brain [10], previously been reported to have no effect upon neuronal and may also make it worthy of investigation in models of death [9], or regeneration [6], hence the control group did CNS trauma. not receive any injections in order to optimise animal welfare. All significant loss of L4 & L5 DRG sensory neurons [1] Angelucci, L., Ramacci, M.T., Taglialatela, G., Hulsebosch, occurs within the first 2 months after sciatic nerve transec- C., Morgan, B., Werrbach-Perez, K. and Perez-Polo, J.R., Nerve growth factor binding in aged rat central nervous tion in this experimental model [8], and so regeneration system: effect of acetyl-L-carnitine, J. Neurosci. Res., 20 after subsequent nerve repair relies upon a relatively stable (1988) 491–496. pool of surviving neurons. Any effect of ALCAR treatment [2] Bremer, J., The role of carnitine in intracellular metabolism, upon nerve regeneration can therefore be assumed to result J. Clin. Chem. Clin. Biochem., 28 (1990) 297–301. primarily from an effect upon the regenerative capacity of [3] Colucci, W.J. and Gandour, R.D., Carnitine acyltransferase: a review of its biology, enzymology and bioorganic chem- individual neurons, rather than upon increased neuronal istry, Bioorg. Chem., 16 (1988) 307–334. survival. Although late secondary nerve repair involves [4] Dagum, A.B., Peripheral nerve regeneration, repair, and limited trimming of the proximal nerve stump [13], this is grafting, J. Hand Ther., 11 (1998) 111–117. unlikely to cause any significant neuronal death since the [5] DeGrandis, D., Santoro, L. and DiBenedetto, P., L-acetylcar- neurons which were susceptible to axotomy induced death nitine in the treatment of patients with peripheral neuropathies, Clin. Drug Invest., 10 (1995) 317–322. will have died already. Furthermore, such trimming does not [6] Fernandez, E., Pallini, P.R., Gangitano, C., Aurora, D.F., involve the obligatory loss of distal neurotrophic support Sangiacomo, C.O., Sbriccoli, A., Ricoy, J. and Rossi, G.F., that occurs after transection of an intact nerve, and which Effects of L-carnitineL-acetylcarnitine and gangliosides on is likely to determine the magnitude of neuronal death [8]. the regeneration of the transected sciatic nerve in rats, In addition, the nerve was then repaired immediately and Neurolog. Res., 11 (1989) 57–62. [7] Green, B., Repair and grafting of the peripheral nerve, In B. this may further limit neuronal death [8]. Green (Ed.), Plastic Surgery, 1998, pp. 630–695. The precise mechanism by which ALCAR improved [8] Hart, A.M., Brannstrom, T., Wiberg, M. and Terenghi, G., nerve regeneration after late secondary repair is not yet Primary sensory neurons and satellite cells after peripheral certain. Neuronal energy requirements are increased during axotomy in the adult rat: timecourse of cell death and elim- regeneration [15], and it is postulated that by improving free ination, Exp. Brain Res., 142 (2002) 308–318. [9] Hart, A.M., Wiberg, M., Youle, M. and Terenghi, G., fatty acid transport into mitochondria, and by supporting Systemic acetyl-L-carnitine eliminates sensory neuronal aerobic glycolysis [2,3,20] in preference to the less efficient loss after peripheral axotomy: A new clinical approach in A. McKay Hart et al. / Neuroscience Letters 334 (2002) 181–185 185

the management of peripheral nerve trauma, Exp. Brain hypoglossal nucleus after hypoglossal nerve transection Res., 145 (2002) 182–289. with and without prevented regeneration in the Sprague– [10] Liu, L., Head, E., Gharib, A.M., Yuan, W., Ingersoll, R.T., Dawley rat, Neurosci. Lett., 67 (1986) 73–77. Cotman, C.W. and Ames, B.N., Memory loss in old rats is [16] Stevens, M.J., Lattimer, S.A., Feldman, E.L., Helton, E.D., associated with brain mitochondrial decay and RNA/DNA Millington, D.S., Sima, A.A. and Greene, D.A., Acetyl-L- oxidation: partial reversal by feeding acetyl-L-carnitine and/ carnitine deﬁciency as a cause of altered nerve myo-inositol or/ R-a-lipoic acid, Proc. Natl. Acad. Sci. USA, 99 (2002) content Na,K-ATPase activity, and motor conduction velo- 2356–2361. city in the streptozotocin-diabetic rat, Metabolism, 45 [11] Lundborg, G., Dahlin, L., Danielsen, N. and Zhao, Q., Troph- (1996) 865–872. ism, tropism, and speciﬁcity in nerve regeneration, J. [17] Taglialatela, G., Angelucci, L., Ramacci, M.T., Werrbach- Reconstr. Microsurg., 10 (1994) 345–354. Perez, K., Jackson, G.R. and Perez-Polo, J.R., Acetyl-L-carni- [12] Marzo, A., Arrigoni, M.E., Urso, R., Rocchetti, M., Rizza, V. tine enhances the response of PC12 cells to nerve growth and Kelly, J.G., Metabolism and disposition of intrave- factor, Brain Res. Dev. Brain Res., 59 (1991) 221–230. nously administered acetyl-L- carnitine in healthy volun- [18] Terenghi, G., Peripheral nerve regeneration and neuro- teers, Eur. J. Clin. Pharmacol., 37 (1989) 59–63. trophic factors, J. Anat., 194 (1999) 1–14. [13] Millesi, H., Techniques for nerve grafting, Hand Clin., 16 [19] Vanderhooft, E., Functional outcomes of nerve grafts for (2000) 73–91. the upper and lower extremities, Hand Clin., 16 (2000) 93– [14] Parnetti, L., Gaiti, A., Mecocci, P., Cadini, D. and Senin, U., 104. Pharmacokinetics of IV and oral acetyl-L-carnitine in a [20] Virmani, M.A., Biselli, R., Spadoni, A., Rossi, S., Corsico, N., multiple dose regimen in patients with senile dementia of Calvani, M., Fattorossi, A., De, S.C. and Arrigoni-Martelli, E., Alzheimer type published erratum appears in Eur. J. Clin. Protective actions of L-carnitine and acetyl-L-carnitine on Pharmacol., 44(6) (1993) 604, Eur. J. Clin. Pharmacol., 42 the neurotoxicity evoked by mitochondrial uncoupling or (1992) 89–93. inhibitors, Pharmacol. Res., 32 (1995) 383–389. [15] Singer, P. and Mehler, S., Glucose and leucine uptake in the Hart, et al. ▪ Neuroprotection by NAC Submitted

NEUROSCIENCE

Sensory neuroprotection, mitochondrial preservation, and therapeutic potential of N-acetyl-cysteine after nerve injury.

1,2,3 1 3 2,3 Andrew McKay Hart , Giorgio Terenghi , Jan-Olof Kellerth , Mikael Wiberg ,

1Blond-McIndoe Research Laboratories, University of Manchester, U.K.; and 2Department of Surgical & Perioperative Science, Section for Hand & Plastic Surgery, University Hospital, Umeå, Sweden; and 3Department of Integrative Medical Biology, Section for Anatomy, Umea University, Umeå, Sweden

SUMMARY. Neuronal death is a major factor in many neuropathologies, particularly traumatic, and yet no clinical neuroprotective therapies are currently available, although antioxidants and mitochondrial protection appear to be fruitful avenues of research. The simplest system involving neuronal death is that of the dorsal root ganglion after peripheral nerve trauma, where the death of ~40% of primary sensory neurons is a major factor in the overwhelmingly poor clinical outcome of the several million nerve injuries that each year occur worldwide. N-acetyl-cysteine is a glutathione substrate which is neuroprotective in a variety of in vitro models of neuronal death, and which may enhance mitochondrial protection. Using TdT uptake nick-end labelling (TUNEL), optical disection, and morphological studies, the effect of systemic N-acetyl-cysteine treatment upon L4&5 primary sensory neuronal death after sciatic nerve transection was investigated. N-acetyl-cysteine (150mg/kg/day) almost totally eliminated the extensive neuronal loss found in controls at both 2 weeks (no treatment 21% loss, N-acetyl-cysteine 3%, P=0.03) and 2 months post-axotomy (no treatment 35% loss, N-acetyl-cysteine 3 %, P=0.002). Glial cell death was reduced (mean number TUNEL positive cells 2 months post-axotomy: no treatment 51/ganglion pair, NAC 16/ganglion pair), and mitochondrial architecture was preserved. The effects were less profound when a lower dose was examined (30mg/kg/day), although significant neuroprotection still occurred. This provides evidence of the importance of mitochondrial dysregulation in axotomy-induced neuronal death in the peripheral nervous system, and suggests that N-acetyl-cysteine merits investigation in central nervous system trauma. N-acetyl-cysteine is already in widespread clinical use for applications outside the nervous system; it therefore has immediate clinical potential in the prevention of primary sensory neuronal death, and has therapeutic potential in other neuropathological systems.

Keywords: TUNEL, Stereology, Nerve Injury, Dorsal Root Ganglion, Cell Death, Apoptosis, Neuroprotection.

investigation. In contrast, the dorsal root Introduction ganglia (DRG) provide a relatively isolated system for investigation, but one in which Neuronal death is a major factor in neuronal death remains a significant clinical many neuropathologies, particularly as relates issue since injury to peripheral sensory, or to nervous system trauma, yet the number of mixed nerves is the commonest form of nervous interacting systems within the central nervous system trauma. It is a significant cause of system (CNS) complicates experimental functional morbidity (Dagum, 1998; Bruyns et Hart, et al. ▪ Neuroprotection by NAC 2 al., 2003), despite considerable advances in administered, clinically safe therapy, which will techniques for surgical repair, since cutaneous prevent the death of a range of neuronal innervation does not return to normal (Wiberg subpopulations, in response to a variety of et al., 2002), and the clinical sensory outcome insults. remains very poor (Lundborg, 2000). This also In contrast to the specificity of impairs fine motor function, since sensory neurotrophic factors (Terenghi, 1999), there feedback is a vital component of the normal appears to be relative homogeneity in the way control loop (M. K. O. Burstedt, M. Schenker, that the initiation of neuronal death is signalled M.Wiberg, and R. S. Johansson, unpublished and then effected in different subpopulations observations). (Ambron and Walters, 1996; Koliatsos and Although various factors are Ratan, 1999). A therapeutic agent acting at this implicated in the poor sensory outcome level of the cell death process ought, therefore, (Lundborg et al., 1994; Fu and Gordon, 1997; to be protective for a wide range of neuronal Lundborg, 2000), the single most important subpopulations against a variety of stimuli. factor is probably the death of ~40% of relevent Signalling leads to transcription of both cell primary sensory neurons (Hart et al., 2002a) death and regenerative genes, and ultimately the since quality of sensation relates to the number ratio of these gene products, for example the of innervating neurons and the size of their ratio of pro-apoptotic Bax to anti-apoptotic Bcl sensory fields. Similarities exist between the gene products (Gillardon et al., 1996), mechanisms underlying axotomy induced determines the fate of the cell. Abortive entry neuronal death within the peripheral nervous into the cell cycle has been implicated system (PNS) and CNS (Koliatsos and Ratan, (Freeman, 1999) in neuronal death, as has 1999), with the result that any treatment which mitochondrial dysregulation. Possibly in protects primary sensory neurons may also have response to increased nitric oxide (NO) therapeutic implications for the management of synthesis (Estevez et al., 1998; Heales et al., traumatic brain, spinal cord, or brachial plexus 1999; Brorson et al., 1999), mitochondria may injury. play an essential role (Al-Abdulla and Martin, Axotomy initiates neuronal death by 1998; Budd and Nicholls, 1998) through the various mechanisms, although the main factor failure of oxidative metabolism, and the determining the magnitude of death that occurs subsequent generation of free oxygen species appears to be the loss of target derived and release of pro-apoptotic molecules (Budd neurotrophic support (Fu and Gordon, 1997). and Nicholls, 1998). The importance of reactive Accordingly, exogenous neurotrophic factor oxygen species to the events culminating in administration is partially neuroprotective after neuronal death (Koliatsos and Ratan, 1999) experimental PNS and CNS lesions (Terenghi, therefore makes antioxidants potentially 1999), and neurotrophic factors have been a attractive as clinically applicable leading therapeutic target in the prevention of neuroprotective agents for stroke or trauma neuronal death. Yet they have failed to find a (Chabrier et al., 1999). One such agent, which mainstream clinical application for several also affects cell cycle control, is N-acetyl- reasons. Firstly, neurotrophic factors act on cysteine (NAC). specific subpopulations of neurons (Terenghi, NAC is currently in clinical use as a 1999), so a complex cocktail would be mucolytic agent in respiratory disease, and as a necessary to completely eliminate neuronal treatment for paracetamol (acetominophen) death, and yet exogenous neurotrophic factors poisoning where it acts to maintain glutathione may interact to negate their own efficacy levels in hepatocytes. In neurons glutathione (Novikova et al., 2000). Furthermore, local acts a free radical scavenger (Cooper and (McArthur et al., 2000) and systemic (Martin et Kristal, 1997), helps maintain mitochondrial al., 1996) complications after systemic oxidative metabolism by protecting the administration have led most workers to use cytochrome oxidase complex I from NO continuous intra-thecal administration, even mediated damage (Moncada, 2000), and though this may not be clinically applicable improves neuronal survival in response to a (Eriksdotter et al., 1998). There is therefore a range of insults in vitro (Ratan et al., 1994; need for an alternative approach, using an easily Heales et al., 1999). By acting as a cysteine Hart, et al. ▪ Neuroprotection by NAC 3

donor, NAC maintains intracellular glutathione group of animals received an equivalent volume levels and is neuroprotective for a range of of 5% dextrose solution (“sham treatment”), neuronal cell types against a variety of stimuli and all animals were dosed until termination. in vitro (Mayer and Noble, 1994; Ferrari et al., Two other groups of animals received no post- 1995; Yan et al., 1995; Seaton et al., 1997). operative interventions (“no treatment”). NAC may therefore reduce neuronal death by Survival periods of either 2 weeks (all blocking attempted entry into the cell cycle, by groups), or 2 months (one “high-dose improving free radical surveillance, or by treatment”, and one “no treatment” group) were preserving mitochondrial function. Such employed. Under terminal anaesthesia mitochondrial protection is of particular interest (pentabarbitone 240mg/kg) animals were given the recent demonstration of the perfusion fixed with 4% paraformaldehyde, and neuroprotective effects of acetyl-L-carnitine the ipsilateral axotomised and contralateral (Bigini et al., 2002; Hart et al., 2002b), a control L4 & L5 DRG were harvested. Ganglia physiological peptide whose precise mechanism were post-fixed in 4% paraformaldehyde prior of action is unclear, but which is likely to act at to equilibration in cryoprotectant solution the mitochondrial level. (phosphate buffered saline containing 15% This study therefore investigated the sucrose and 0.1% sodium azide), and storage at effect upon cell death within dorsal root ganglia -40ºC in frozen blocks of OCT™ mounting (DRG) of systemic NAC treatment after medium. peripheral nerve injury, with the aim of Each entire ganglion was cut into quantifying its neuroprotective potential, and of serial 30µm cryosections, which were then seeking further evidence for the role of permeabilised by microwave irradiation mitochondrial dysregulation in axotomy- (478Watts for 1.5 minutes with a 1.5l water induced neuronal death. Mitochondrial load) and rapid cooling in iced phosphate morphology was evaluated by transmission buffered saline (PBS). A previously published electron micrography, and the effect of triple staining protocol (Hart et al., 2002a) was treatment upon cell survival was quantified employed to permit the detection of cell death using a combination of TUNEL staining and by TUNEL (Promega Apoptosis Detection neuron counting by the optical disector System Fluorescein), and by light microscopic technique. examination of nuclear and cytoplasmic morphology, stained by Hoechst 33342 Experimental Procedures (H33342) and propidium iodide (PI) respectively. During fluorescence microscopy All work was performed in each stain was visualised separately using band accordance with the terms of the Animals specific filters, whilst the use of an Olympus (Scientific Procedures) Act 1986, and the WU filter during optical disection allowed experimental design was cogniscent of the need simultaneous visualisation of both PI and to minimise the number of experimental H33342. animals, and to optimise welfare. Under All sections were examined for the halothane anaesthesia groups (n=6) of young presence of TUNEL positive cells using adult male Sprague-Dawley rats underwent fluorescence microscopy and a 20x objective unilateral sciatic nerve division at the upper lens (Olympus BX60 microscope). Positive border of quadratus femoris. Both nerve cells were confirmed to be neurons, or glia, by stumps were then ligated, and secured into further examination under a 40x objective, silicon caps with single 9/0 Ethilon™ sutures when their nuclear and cytoplasmic morphology (Ethicon Ltd., U.K.) in order to prevent was also examined. The immediately spontaneous regeneration. Parenteral systemic antecedent and subsequent serial sections were therapy with N-acetyl-cysteine (NAC, inspected to ensure that each TUNEL positive “Parvolex”, Medeva Pharmaceuticals U.K.) in neuron was only counted once, and the total sterile 5% dextrose solution (Maco Pharma, number was expressed as a combined count U.K.) was commenced post-operatively; two (L4+L5) per experimental group of 6 animals. groups received 150mg/kg/day (“high-dose The more numerous TUNEL positive glia were treatment”), and one group received expressed as a mean count per L4+L5 ganglion 30mg/kg/day (“low-dose treatment”). A further Hart, et al. ▪ Neuroprotection by NAC 4

pair. Since positive glial cells were present in of C.A.S.T.-Grid (Version 1) software, of all ganglia, the number that were positive due to optical disectors within the tissue sections cut the ipsilateral effects of axotomy was estimated from the ganglion. Counting rules were used to by subtracting the count in axotomised ganglia ensure that every cell within a ganglion has an from that of their contralateral controls. equal chance of being counted, obviating any The number of surviving L4 & L5 bias due to variations in cell size, or distribution DRG neurons was estimated using the within the ganglion. Finally, the number of statistically unbiased, stereological technique of neurons within the ganglion (Npar) is estimated optical disection (West, 1993; Harding et al., by multiplying the numerical volume (NV) of 1994), the protocol for which has previously neurons by the volume of the ganglion (Vref ): described (Hart et al., 2002a). Essentially this process involved estimation of a ganglion’s Npar = NV . Vref volume (Vref) using the Cavalieri Principle, and direct measurements in a systematic random Neuron counts were expressed as a sample of ten or more sections from each combined (L4 plus L5) count for each side ganglion to permit calculation of the mean (axotomised, or contralateral control). Neuron thickness and cross-sectional area. The density loss was then calculated by subtracting the (or “numerical volume”, Nv) of neurons within neuron count for a pair of axotomised ganglia that ganglion was then determined by from that of their contralateral controls, and was systematic random placement, under the control expressed as a percentage of the neuron count

A B

C D

Figure 1: Effect of N-acetyl-cysteine (NAC) treatment upon primary sensory neuronal morphology after peripheral nerve transection.Dorsal root ganglia stained with the nuclear dye Hoechst 33342 (blue fluorescence) and the cytoplasmic dye propidium iodide (red fluorescence). Non-axotomised ganglia [A] contain neurons with normal cytoplasmic morphology, surrounded by numerous smaller satellite cells, in which nuclear staining predominates. Two weeks after sciatic nerve transection gross morphological changes are evident in the axotomised ganglia of animals sham-treated with 5% dextrose solution [B], with most neurons displaying nuclear eccentricity, membrane irregularities, and cytoplasmic granulation or vacuolation (hollow arrow). Apoptotic neurons are also evident (solid arrow). Morphology is somewhat improved after low-dose treatment (N-acetyl-cysteine 30mg/kg/day) [C], although nuclear eccentricity and membrane irregularity remains frequent, and some dilated vacuolated cells are present (arrowhead). After high-dose treatment (N-acetyl-cysteine 150mg/kg/day) neuronal morphology is significantly more normal [D]. Scale bar equals 40µm. Hart, et al. ▪ Neuroprotection by NAC 5 in the control ganglia. Mann-Whitney Rank Sum test since data was In a separate experiment, a further not always normally distributed (as assessed by four animals underwent unilateral sciatic nerve the Kolmogorov-Smirnov Test). For all other axotomy; two then received high-dose treatment data comparison between multiple groups was with NAC (150mg/kg/day), and two received performed using ANOVA, and pairwise no treatment. Two weeks later the DRG were comparisons using Students t-test. harvested as described above, except that only 50ml of PBS was used, and perfusion fixation Results was achieved with 500ml of 3% glutaraldehyde & 1% paraformaldehyde in PBS. DRG were No adverse effects upon animal post-fixed in the same solution, and transferred behaviour were found, and NAC treatment did to 1% osmium tetroxide in 0.1M phosphate not significantly affect weight gain during the buffer at 37oC prior to dehydration in acetone study period (2 weeks, p=0.87; 2 months and embedding in Vestopal. Blocks were p=0.30). trimmed on a Pyramitome (LKB, Sweden), and Morphology sectioned with a 2128 Ultratome (LKB). Serial When compared to normal ganglia 1µm sections were cut and examined by light (Figure 1A), prominent morphological changes microscopy to identify representative neurons. were present in the axotomised ganglia of Ultrathin (60-70nm) sections were then untreated animals, with the majority of neurons collected onto formvar-coated one-hole copper exhibiting nuclear eccentricity, and grids, stained with uranyl acetate and lead chromatolysis (Figure 1B). Neurons displaying citrate, and examined using a JEOL 100CX features of apoptosis (cytoplasmic granulation, electron microscope. vacuolation, membrane irregularities) were Statistical analysis of results was seen, as were numerous enlarged, and grossly performed using the Sigmastat™ 2.0 software vacuolated neurons. Although the package, using a probability level of p<0.05 for morphological findings were unchanged by rejection of null hypotheses. Comparison of sham treatment with 5% dextrose solution, a TUNEL counts for neurons was made by the qualitative improvement in neuronal

A B

Figure 2: Effect of N-acetyl-cysteine upon mitochondrial morphology. Representative electron micrographs of C mitochondria within primary sensory neurons of the L5 dorsal root ganglion. Two weeks after unilateral sciatic nerve division many neurons within ipsilateral, axotomised ganglia of control animals (that received no treatment) exhibit morphological features of cell death. These neurons contain abnormal mitochondria [A] that are pathologically dilated, exhibit loss of cristal architecture, or have ruptured. In contrast, comparable ganglia in animals treated with 150mg/kg/day N-acetyl- cysteine (NAC) contain neurons whose mitochondrial structure is predominantly intact [B], although some swelling, and occasional cristal disruption is still evident. This morphology compares favourably with that of mitochondria in the contralateral ganglia, whose neurons have not been axotomised [C]. Scale bar equals 500nm. Hart, et al. ▪ Neuroprotection by NAC 6 morphology was found after low-dose NAC positive neurons found in axotomised ganglia 2 treatment (Figure 1C). Neuronal morphology months after axotomy was also reduced by was significantly more normal in ganglia from high-dose NAC treatment (Table 2). animals that received high-dose treatment with TUNEL positive glia (Table 3) were NAC (Figure 1D). significantly more numerous in the axotomised Electron micrography (Figure 2) ganglia than in the contralateral non-axotomised revealed that in untreated or sham treated ganglia of the sham treatment (P=0.002) and no animals, the primary sensory neurons within treatment groups (P=0.05), but the difference axotomised ganglia which displayed the was minimal in both groups treated with high- abnormal morphological features described dose NAC. When compared to no treatment, the above typically exhibited varying degrees of increase attributable to axotomy found two disruption to their mitochondrial ultrastructure. weeks after axotomy was unaffected by sham This ranged from swelling, and occasional treatment or low dose NAC. However, it was disruption of the cristae within the neurons halved by high-dose NAC, which also obviated exhibiting mild morphological changes in the effect of axotomy 2 months later (mean keeping with the chromatolysis reaction, to increase: no treatment 20.2 cells/group, SD severe disruption of mitochondrial architecture 24.7; high dose NAC mean 0.3 cells/group, SD with vacuolation, complete disruption of the 10.1). Although treatment had no effect upon cristae, and rupture of the mitochondrial the number of TUNEL positive glia present in membrane in those neurons displaying even non-axotomised ganglia two weeks after early morphological features of cell death axotomy, NAC nearly halved the number (Figure 2A). Such mitochondrial disruption present two months after axotomy. was not a feature of neurons of animals treated Neuron Loss with 150mg/kg/day NAC (Figure 2B). Indeed Two weeks after axotomy the mean mitochondrial morphology was found to be number of neurons present in the non- qualitatively similar to that of intact axotomised, contralateral control L4+5 ganglia contralateral neurons (Figure 2C). of untreated animals was 29641 (SD 2268), and TUNEL Staining was similar in animals that received sham The effect of NAC treatment upon the treatment with 5% dextrose solution (29845 number of TUNEL positive neurons, and the neurons, SD 1553; p=0.86), or treatment with neuron loss found 2 weeks after axotomy is NAC (low-dose 30844 neurons, SD 3839, summarised in Table 1, whilst Table 2 gives the p=0.58; high dose 31761 neurons, SD 2804, equivalent findings 2 months after axotomy. p=0.18). Two months after axotomy there was Table 3 summarises the effect of treatment upon still no difference between the number of the mean number of TUNEL positive glial cells. neurons present in the non-axotomised ganglia TUNEL positive neurons were only of untreated animals (31545 neurons, SD 2446), found in axotomised ganglia, and all exhibited and those that received NAC treatment (high morphological features of cell death, both dose, 30034 neurons, SD 4539; p=0.52). nuclear (chromatin condensation and nuclear Two weeks after axotomy 21% irregularity, or fragmentation), and cytoplasmic. (p=0.002 vs. control ganglia) of all neurons in Two weeks after sciatic nerve transection the the axotomised L4&5 ganglia had died in the total number of TUNEL positive neurons group receiving sham treatment (Table 1), as present in the axotomised L4&5 ganglia of also occurred in those that received no untreated animals was 25 per group of 6 treatment (21% loss, p=0.96 vs. sham animals, and was not significantly altered by treatment). This loss of neurons was almost sham treatment (19 /group, p=0.405). A large completely eliminated by 2 weeks of treatment reduction in the number of TUNEL positive with NAC (low-dose NAC 1% loss, p=0.019; neurons was found after treatment with NAC at high-dose NAC 3% loss, p=0.031), and the both the low (7 /group), and high (2 /group) same magnitude of neuroprotective effect was doses used, as compared to either sham still present 2 months after axotomy (no treatment (low-dose p=0.065; high-dose treatment 35% loss; high-dose NAC 3% loss, p=0.003), or no treatment (low-dose p=0.015; p=0.002) (Table 2). high-dose p=0.002). The number of TUNEL Hart, et al. ▪ Neuroprotection by NAC 7

Control Ganglia Axotomised Ganglia Table 1: Effect of NAC treatment upon neuron (L4+5) (L4+5) Percentage death 2 weeks after axotomy. TUNEL positive Treatment Mean (SD) Mean (SD) Neuron neuron counts are expressed as the total number Group Loss TUNEL Neuron TUNEL Neuron of positive neurons present in the axotomised, (L4+5) or contralateral control ganglia of each Count Count Count Count experimental group. Axotomised, and contralateral control ganglia neuron counts are No 29641 23317 0 25 21% Treatment (2268) (2146) stated as the mean (standard deviation) count per experimental group. Neuron loss is calculated as the mean neuron count in the Sham 29845 23625 0 19 21% axotomised ganglia subtracted from that in the Treatment (1553) (1899) control ganglia, expressed as a percentage of the control value. Low Dose *P Value <0.05 (compared to sham treatment). 30844 30441* NAC (30mg/ 0 7 1%* (1567) (2321) kg/ day)

High Dose 31761 30699* NAC(150mg 0 2* 3%* / kg/day) (2804) (4942)

Table 2: Effect of NAC treatment upon neuron Control Ganglia Axotomised Ganglia (L4+5) (L4+5) death 2 months after axotomy. The numbers of TUNEL positive neuron are expressed as the Mean (SD) Mean (SD) Neuron Treatment Loss total number of positive neurons present in the Group (L4+5) axotomised, or contralateral control ganglia of TUNEL Neuron TUNEL Neuron each experimental group. Axotomised, and Count Count Count Count contralateral control ganglia neuron counts are stated as the mean (standard deviation) count per experimental group. Neuron loss is No 31545 20460 calculated as the mean neuron count in the 0 5 35% Treatment (2446) (2138) axotomised ganglia subtracted from that in the control ganglia, expressed as a percentage of the control value. High Dose *P Value <0.05 (compared to no treatment). NAC 30034 29243* 0 3 3%* (150mg/kg/ (4539) (4100) day)

Table 3: Effect of NAC treatment upon Increase Control Axotomised Attributable TUNEL positive glial cells, numbers of which Ganglia Ganglia Treatment Group to Axotomy are expressed as the mean (standard deviation) (L4+5) (L4+5) (L4+5) number of positive cells present each Mean (SD) Mean (SD) Mean (SD) axotomised, or contralateral control L4+L5 ganglion pair in each experimental group. The 11.3 22.2* 10.8 No Treatment increase directly attributable to axotomy is (7.1) (9.7) (13.2) calculated as the mean count in the axotomised Sham 5 17.3* 12.3 L4+L5 ganglion pair minus that in the non- Treatment (2.5) (6.2) (5.6) axotomised control ganglion pair. * P<0.05 compared to no treatment. Low Dose 9.5 21.8 12.3 NAC

2 Weeks (9.6) (9.6) (13.7) (30mg/kg/ day) Post-Axotomy Post-Axotomy High Dose NAC 9.5 15.0 4.4 (150mg/kg/ (8.3) (15.5) (15.7) day)

51* No Treatment 30.8 20.2 (25.2) (39.7) (24.6)

High Dose

2 Months NAC 15.7 16.0 0.3 Post-Axotomy Post-Axotomy (150mg/kg/ (11.0) (5.7) (10.1) day)

Hart, et al. ▪ Neuroprotection by NAC 8

Discussion complete after 2 months (Hart et al., 2002a). Hence in this study, the 2 week timepoint TUNEL has proven to be a reliable reveals the effect of NAC treatment at the time marker of cell death within DRG, despite the when the rate of neuron death is maximal, potential for cells in the S-phase of mitosis to whilst the 2 month timepoint reveals the stain positive (Sanders and Wride, 1996), when magnitude and permanence of that effect. The nuclear morphology is used to confirm cell effect upon glial cell death is also clarified, death (Crowe et al., 1997; Hart et al., 2002a). since numbers of TUNEL positive glia also The typically small number of TUNEL positive peak 2 months after axotomy (Hart et al., neurons that are found has previously been 2002a). ascribed to the short time period during cell Systemic treatment with NAC death in which cells are potentially TUNEL (150mg/kg/day) brought about a significant positive (Hart et al., 2002a). Optical disection reduction in both the number of TUNEL is an ideal technique for estimation of neuron positive neurons, and the neuron loss found 2 numbers (Hart et al., 2002a), since it avoids the weeks after peripheral axotomy. Neuronal statistical bias of assumption based techniques death was prevented, rather than just delayed, (West, 1999) and the effects of tissue sectioning since this protective effect was found to be artefacts (Hedreen, 1998), yet is time-efficient. preserved 2 months after axotomy, by which The technique has been validated for counting time neuron loss has effectively finished in the DRG neurons, although precision and accuracy absence of treatment (Hart et al., 2002a). are dependent upon adherence to certain rules. Treatment with a reduced dose of NAC Accordingly, ten or more sections whose (30mg/kg/day) was still significantly effective thickness was measured directly were used for in preventing neuronal death, although a dose volume estimation by the Cavalieri Principle response was evident in the number of TUNEL (Gundersen et al., 1988), and adequate numbers positive neurons and glial cells, and in the of caps were counted (mean 152 per ganglion, morphological findings. The loss found in the SD26) to meet the requirement of 100-200 per low-dose group 2 weeks after axotomy was ganglion (West, 1999). The mean neuron counts slightly less than in the high-dose group, but the obtained from control ganglia (30618, SD 2969) difference lies comfortably within the were in agreement with other results in the variability of the counting technique, and literature obtained by stereological methods probably does not reflect a greater protective (Tandrup, 1993; Bergman and Ulfhake, 1998; effect since the TUNEL count was higher with Hart et al., 2002a) verifying the accuracy of the low-dose NAC. As previously discussed, technique used. Combined L4+5 counts were TUNEL is a more sensitive marker of cell death used since this comprises almost the entire at 2 weeks post-axotomy and it is likely that if sciatic nerve pool, and since combined counts both treatment doses were given for 2 months, exhibit less variance than those from individual then the neuron loss would be greater in the ganglia (L4 or L5) (Hart et al., 2002a) making low-dose group. them more sensitive as indicators of the effect Neuronal death cannot have been of axotomy. completely eliminated by NAC treatment since By highlighting DNA fragmentation, TUNEL positive neurons were still present, TUNEL demonstrates the very early stages of a however NAC was sufficiently protective that 2 neuron’s death, whereas the calculated neuronal weeks after axotomy the overall neuronal loss loss reflects the cell’s subsequent complete in both treatment groups considered together involution. Very small losses may be entirely was only 2% (control count 31323, SD 3238; masked by the normal biological variability in operated count 30581; SD 3793). This neuron numbers, therefore TUNEL is the more neuroprotective effect persisted 2 months after sensitive marker, although it is not such a good axotomy, when loss was only 3%, and quantitative tool as optical disection. Using the represents salvage of around 5,500 neurons per same experimental model, we have previously animal after 2 weeks, and of more than 10,000 demonstrated that in the absence of any neurons (over a third of the total sensory intervention numbers of TUNEL positive neuronal pool) 2 months after axotomy. The neurons peak 2 weeks after peripheral axotomy, massive reduction in the calculated neuron loss and that the neuronal loss is essentially Hart, et al. ▪ Neuroprotection by NAC 9 was not merely an artefact of changes in the maintenance of intra-neuronal glutathione non-axotomised ganglia, since NAC treatment levels (Dringen et al., 1999). In vivo treatment did not affect the number of neurons in these with NAC might therefore increase glial internal control ganglia. Equally, it is evident glutathione production, potentially protecting that the neuroprotective effect is due to NAC glia directly, and also indirectly via and not merely to the act of intraperitoneal neuroprotection secondary to increased injection, since vehicle solution (sterile 5% neuronal glutathione levels. dextrose) had no effect upon the number of There are a variety of mechanisms by TUNEL positive neurons, or upon neuronal which N-acetyl-cysteine may potentially exert loss. its neuroprotective effect, but it is necessary to TUNEL has been utilised to infer from other experimental models of demonstrate satellite cell death after neuronal neuronal death since few studies on primary injury in the CNS, which was then implicated in sensory neurons exist. Axotomy causes a extension of the lesion (Crowe et al., 1997). In loss of distal neurotrophic support that may the dorsal root ganglion, ISEL and TUNEL initiate neuronal death via an effect upon positive glial cells have previously been cellular redox balance, NO levels, or oxidative reported in both axotomised and contralateral stress. Deprivation of certain neurotrophic control ganglia after unilateral peripheral factors in vitro may induce similar events axotomy (Groves et al., 1997; Hart et al., (Pettmann and Henderson, 1998), whilst 2002a). Positive cells are more numerous in exogenous neurotrophic factors may counteract axotomised ganglia and there appears to be a these events by increasing neuronal cysteine temporal relationship, with a peak found two uptake and glutathione synthesis (Mattson et al., months after axotomy (Hart et al., 2002a). 1995; Pan and Perez-Polo, 1996). NAC may Although little else is known about the therefore directly mimic some of the effects of underlying processes, the putative roles of glial neurotrophic support (Yan and Greene, 1998), cells in neuronal death as sources for but also modulates neurotrophic factor neuroprotective glutathione precursors or as signalling pathways within neurons (Kamata et mediators of neuronal death (Barker et al., al., 1996). In keeping with this, NAC prevents 1996; Dringen et al., 1999) raises intriguing apoptosis in trophic factor deprived PC12 cells questions regarding NAC’s efficacy in reducing by an effect upstream of c-Jun kinase (JNK) the number of TUNEL positive glia. Although activation (Park et al., 1996). there is likely to be a basal turnover of glia, Axotomy induced retrograde giving a low “background” number of TUNEL degeneration occurs in association with positive cells, most glial death within perikaryal accumulation of mitochondria and axotomised ganglia probably occurs secondary oxidative stress (Al-Abdulla and Martin, 1998), to associated neurons either dying or retracting whilst spinal root avulsion sufficient to induce their projections. The latter has been proposed death of 30% of motorneurons is accompanied in the CNS, and degeneration of crossed by oxidative stress (Martin et al., 1999) and secondary fibres may account for the delayed NOS activation (Novikov et al., 1995). The finding of increased glial cell death in mitochondrial electron transport chain may then contralateral DRG. Protection of these fibres be disrupted due to reversible inhibition of by NAC treatment might then account for the complex IV (cytochrome C) by transient finding of a reduced number of TUNEL increases in NO, or permanent inhibition of positive glia in the contralateral ganglia two complexes I & II by a sustained increase months after axotomy. (Brown et al., 1995; Brorson et al., 1999). It cannot be said whether the effect of Redox imbalance then favours generation of NAC is upon the glia, or is secondary to a reactive oxygen species (ROS) that disturb neuroprotective effect, but the differential effect mitochondrial membrane potential, disrupting of low and high-dose treatment found 2 weeks calcium homeostasis, and potentially leading to after axotomy may suggest an element of both. cell death (Keller et al., 1998; Heales et al., Also, in vitro work suggests the importance of 1999; Brorson et al., 1999). Mitochondrial glutathione release by glia, for whom NAC is glutathione depletion potentiates the effect, an effective substrate (Han et al., 1997), in the which is countered in vitro by antioxidants Hart, et al. ▪ Neuroprotection by NAC 10

(Seaton et al., 1997) and glutathione (Lizasoain The demonstrated neuroprotective et al., 1996). NAC protects mtDNA from effect of NAC may therefore be due to a damage due to increased NO or oxidative stress combination of its antioxidant, and reductant mimicking the anti-apoptotic species Bcl-2 properties, the promotion of glutathione (Myers et al., 1995; Deng et al., 1999). synthesis, and prevention of abortive entry into Depletion of glutathione, particularly the cell cycle by injured neurons. mitochondrial (Cooper and Kristal, 1997; This study has therefore demonstrated Wullner et al., 1999), increases the that NAC is highly neuroprotective in an in susceptibility of neurons to a variety of toxic vivo model of peripheral axotomy in the stimuli including neurotrophin withdrawal and relatively isolated neuronal system of the DRG. oxidative stress (Ratan et al., 1994; Drukarch et Subsequent studies to define NAC’s precise al., 1997; Wullner et al., 1999). Since mechanism of action may help to elucidate the glutathione is one of the principal defences axotomy-induced events culminating in against ROS and oxidative stress within neuronal death, but mitochondrial dysfunction neurons, increasing levels may be protective is clearly involved, and can be limited by NAC (Ratan et al., 1994; Cooper and Kristal, 1997). treatment. Production of NO and reactive However glutathione does not cross the blood- oxygen species, glutathione depletion, and brain barrier (BBB), and neurons cannot abortive cell cycling are further implicated. A synthesise thiol substrates, being dependent pressing clinical need exists to reduce sensory during oxidative stress upon the more resistant neuronal death after peripheral nerve trauma, astroglia (Dringen and Hamprecht, 1999) for and the results of this study strongly suggest the supply of peptide precursors (Dringen et al., that NAC may be a suitable therapeutic agent, 1999) for glutathione synthesis (Sagara et al., and additionally raise the possibility of a role in 1993; Barker et al., 1996; Wang and Cynader, central nervous system trauma. 2000) Cysteine is normally the rate limiting precursor (Pan and Perez-Polo, 1996; Wang and Cynader, 2000), but can readily be taken up as Reference List NAC (Dringen and Hamprecht, 1999). Al-Abdulla, N. A. and Martin, L. J., 1998. Accordingly, neuronal glutathione levels are Apoptosis of retrogradely degenerating neurons increased by NAC in vitro (Yan et al., 1995; occurs in association with the accumulation of Kamata et al., 1996; Wagner et al., 1998; perikaryal mitochondria and oxidative damage Dringen and Hamprecht, 1999), and the effect is to the nucleus. Am. J. Pathol. 153, 447-456. enhanced by astroglial co-culture (Dringen et Ambron, R. T. and Walters, E. T., 1996. al., 1999). Given the key role that glia play in Priming events and retrograde injury signals. A neuronal glutathione synthesis and survival, it is new perspective on the cellular and molecular of interest that NAC may also protects glia biology of nerve regeneration. Mol. Neurobiol. (Mayer and Noble, 1994; Barker et al., 1996; 13, 61-79. Han et al., 1997). Barker, J. E., Bolanos, J. P., Land, J. M., Clark, At high doses NAC has direct J. B., Heales, S. J., 1996. Glutathione protects reductant (Yan et al., 1995; Kamata et al., 1996) astrocytes from peroxynitrite-mediated and antioxidant effects, and can block lipid mitochondrial damage: implications for peroxidation by peroxynitrate (ONOO-) (Han et neuronal/astrocytic trafficking and al., 1997). However the neuroprotective effect neurodegeneration. Dev. Neurosci 18, 391-396. of NAC (Yan and Greene, 1998) is not fully Bergman, E. and Ulfhake, B., 1998. Loss of explained by its role as an antioxidant, and primary sensory neurons in the very old rat: glutathione substrate (Ferrari et al., 1995; Yan neuron number estimates using the disector et al., 1995). There is evidence that inco- method and confocal optical sectioning. J. ordinated cell cycling may play a role in Comp. Neurol. 396, 211-222. primary sensory neuronal death (ElShamy et al., Bigini, P., Larini, S., Pasquali, C., Muzio, V., 1998), hence regulation of entry into the cell Mennini, T., 2002. Acetyl-L-carnitine shows cycle and of DNA synthesis (Ferrari et al., neuroprotective and neurotrophic activity in 1995) may partly (Yan and Greene, 1998) primary culture of rat embryo motoneurons. explain NAC’s efficacy. Neurosci. Lett. 329, 334-338. Hart, et al. ▪ Neuroprotection by NAC 11

Brorson, J. R., Schumacker, P. T., Zhang, H., precursor for neuronal glutathione. J. Neurosci 1999. Nitric oxide acutely inhibits neuronal 19, 562-569. energy production. The Committees on Drukarch, B., Schepens, E., Jongenelen, C. A., Neurobiology and Cell Physiology. J Neurosci Stoof, J. C., Langeveld, C. H., 1997. Astrocyte- 19, 147-158. mediated enhancement of neuronal survival is Brown, G. C., Bolanos, J. P., Heales, S. J., abolished by glutathione deficiency. Brain Res. Clark, J. B., 1995. Nitric oxide produced by 770, 123-130. activated astrocytes rapidly and reversibly ElShamy, W. M., Fridvall, L. K., Ernfors, P., inhibits cellular respiration. Neurosci. Lett. 193, 1998. Growth arrest failure, G1 restriction point 201-204. override, and S phase death of sensory Bruyns, C. N., Jaquet, J. B., Schreuders, T. A., precursor cells in the absence of neurotrophin-3. Kalmijn, S., Kuypers, P. D., Hovius, S. E., Neuron 21, 1003-1015. 2003. Predictors for return to work in patients Eriksdotter, J. M., Nordberg, A., Amberla, K., with median and ulnar nerve injuries. J. Hand Backman, L., Ebendal, T., Meyerson, B., Olson, Surg. [Am. ] 28, 28-34. L., Seiger, Shigeta, M., Theodorsson, E., Budd, S. L. and Nicholls, D. G., 1998. Viitanen, M., Winblad, B., Wahlund, L. O., Mitochondria in the life and death of neurons. 1998. Intracerebroventricular infusion of nerve Essays. Biochem. 33, 43-52. growth factor in three patients with Alzheimer's Chabrier, P. E., Auguet, M., Spinnewyn, B., disease. Dement. Geriatr. Cogn Disord. 9, 246- Auvin, S., Cornet, S., Demerle-Pallardy, C., 257. Guilmard-Favre, C., Marin, J. G., Pignol, B., Estevez, A. G., Spear, N., Manuel, S. M., Radi, Gillard-Roubert, V., Roussillot-Charnet, C., R., Henderson, C. E., Barbeito, L., Beckman, J. Schulz, J., Viossat, I., Bigg, D., Moncada, S., S., 1998. Nitric oxide and superoxide contribute 1999. BN 80933, a dual inhibitor of neuronal to motor neuron apoptosis induced by trophic nitric oxide synthase and lipid peroxidation: a factor deprivation. J. Neurosci 18, 923-931. promising neuroprotective strategy [see Ferrari, G., Yan, C. Y., Greene, L. A., 1995. N- comments]. Proc. Natl. Acad. Sci. U. S. A 96, acetylcysteine (D- and L-stereoisomers) 10824-10829. prevents apoptotic death of neuronal cells. J. Cooper, A. J. and Kristal, B. S., 1997. Multiple Neurosci. 15, 2857-2866. roles of glutathione in the central nervous Freeman, R. S. 1999. The cell cycle and system. Biol. Chem. 378, 793-802. neuronal death. In: Koliatsos, V. E. and Ratan, Crowe, M. J., Bresnahan, J. C., Shuman, S. L., R. R. (Ed's), Cell Death and Diseases of the Masters, J. N., Beattie, M. S., 1997. Apoptosis Nervous System Humana Press Inc., Totowa, and delayed degeneration after spinal cord in New Jersey, pp. 103-122. jury in rats and monkeys. Nature Medicine 3, Fu, S. Y. and Gordon, T., 1997. The cellular 73-76. and molecular basis of peripheral nerve Dagum, A. B., 1998. Peripheral nerve regeneration. Mol. Neurobiol. 14, 67-116. regeneration, repair, and grafting. J. Hand Ther. Gillardon, F., Klimaschewski, L., Wickert, H., 11, 111-117. Krajewski, S., Reed, J. C., Zimmermann, M., Deng, G., Su, J. H., Ivins, K. J., Van, H. B., 1996. Expression pattern of candidate cell death Cotman, C. W., 1999. Bcl-2 facilitates recovery effector proteins Bax, Bcl- 2, Bcl-X, and c-Jun from DNA damage after oxidative stress. Exp. in sensory and motor neurons following sciatic Neurol. 159, 309-318. nerve transection in the rat. Brain Res. 739, Dringen, R. and Hamprecht, B., 1999. N- 244-250. acetylcysteine, but not methionine or 2- Groves, M. J., Christopherson, T., Giometto, B., oxothiazolidine-4- carboxylate, serves as Scaravilli, F., 1997. Axotomy-induced cysteine donor for the synthesis of glutathione apoptosis in adult rat primary sensory neurons. in cultured neurons derived from embryonal rat J. Neurocytol. 26, 615-624. brain. Neurosci Lett 259, 79-82. Gundersen, J. G., Bendtsen, T. F., Korbo, L., Dringen, R., Pfeiffer, B., Hamprecht, B., 1999. Marcussen, N., Moller, A., Nielsen, K., Synthesis of the antioxidant glutathione in Nyengaard, J. R., Pakkenberg, B., Sorensen, F. neurons: supply by astrocytes of CysGly as B., Vesterby, A., West, M. J., 1988. Some new, simple and efficient stereological methods and Hart, et al. ▪ Neuroprotection by NAC 12 their use in pathological research and diagnosis. differentially blocked by glutathione or glucose. APMIS 96, 379-394. Biochem. J. 314 ( Pt 3), 877-880. Han, D., Sen, C. K., Roy, S., Kobayashi, M. S., Lundborg, G., 2000. A 25-year perspective of Tritschler, H. J., Packer, L., 1997. Protection peripheral nerve surgery: evolving against glutamate-induced cytotoxicity in C6 neuroscientific concepts and clinical glial cells by thiol antioxidants. Am. J. Physiol significance. J. Hand Surg. 25A, 391-414. 273, R1771-R1778. Lundborg, G., Dahlin, L., Danielsen, N., Zhao, Harding A.J., Halliday G.M., Cullen K., 1994. Q., 1994. Trophism, tropism, and specificity in Practical considerations for the use of the nerve regeneration. J. Reconstr. Microsurg. 10, optical disector in estimating neuronal number. 345-354. J. Neurosci. Methods 51, 83-89. Martin, D., Merkel, E., Tucker, K. K., Hart, A. M., Brannstrom, T., Wiberg, M., McManaman, J. L., Albert, D., Relton, J., Terenghi, G., 2002a. Primary sensory neurons Russell, D. A., 1996. Cachectic effect of ciliary and satellite cells after peripheral axotomy in neurotrophic factor on innervated skeletal the adult rat: timecourse of cell death and muscle. Am. J. Physiol. 271, R1422-R1428. elimination. Exp. Brain Res. 142, 308-318. Martin, L. J., Kaiser, A., Price, A. C., 1999. Hart, A. M., Wiberg, M., Youle, M., Terenghi, Motor neuron degeneration after sciatic nerve G., 2002b. Systemic acetyl-L-carnitine avulsion in adult rat evolves with oxidative eliminates sensory neuronal loss after peripheral stress and is apoptosis. J. Neurobiol. 40, 185- axotomy: A new clinical approach in the 201. management of peripheral nerve trauma. Exp. Mattson, M. P., Barger, S. W., Begley, J. G., Brain Res. 145, 182-189. Mark, R. J., 1995. Calcium, free radicals, and Heales, S. J., Bolanos, J. P., Stewart, V. C., excitotoxic neuronal death in primary cell Brookes, P. S., Land, J. M., Clark, J. B., 1999. culture. Methods Cell Biol. 46:187-216, 187- Nitric oxide, mitochondria and neurological 216. disease. Biochim. Biophys. Acta 1410, 215- Mayer, M. and Noble, M., 1994. N-acetyl-L- 228. cysteine is a pluripotent protector against cell Hedreen, J. C., 1998. Lost caps in histological death and enhancer of trophic factor-mediated counting methods. Anat. Rec. 250, 366-372. cell survival in vitro. Proc. Natl. Acad. Sci. Kamata, H., Tanaka, C., Yagisawa, H., U.S.A. 91, 7496-7500. Matsuda, S., Gotoh, Y., Nishida, E., Hirata, H., McArthur, J. C., Yiannoutsos, C., Simpson, D., 1996. Suppression of nerve growth factor- Adornato, B. T., Singer, E. J., Hollander, H., induced neuronal differentiation of PC12 cells. Marra, C. M., Rubin, M., Cohen, B., Tucker, T., N-acetylcysteine uncouples the signal Navia, B. A., Schifitto, G., Katzenstein, D. A., transduction from ras to the mitogen-activated Rask, C. A., Zaborski, L., Smith, S. E., Shriver, protein kinase cascade. J Biol. Chem. 271, S., Millar, L., Clifford, D. B., AIDS Clinical 33018-33025. Trials Group Team 291, 2000. A phase II trial Keller, J. N., Kindy, M. S., Holtsberg, F. W., St, of nerve growth factor for sensory neuropathy C. D., Yen, H. C., Germeyer, A., Steiner, S. M., associated with HIV infection. Neurology 54, Bruce-Keller, A. J., Hutchins, J. B., Mattson, 1080-1088. M. P., 1998. Mitochondrial manganese Moncada, S., 2000. Nitric oxide and cell superoxide dismutase prevents neural apoptosis respiration: physiology and pathology. Verh. K. and reduces ischemic brain injury: suppression Acad. Geneeskd. Belg. 62, 171-179. of peroxynitrite production, lipid peroxidation, Myers, K. M., Fiskum, G., Liu, Y., Simmens, S. and mitochondrial dysfunction. J Neurosci. 18, J., Bredesen, D. E., Murphy, A. N., 1995. Bcl-2 687-697. protects neural cells from cyanide/aglycemia- Koliatsos, V. E. and Ratan, R. R., 1999. Cell induced lipid oxidation, mitochondrial injury, Death and Diseases of the Nervous System. and loss of viability. J. Neurochem. 65, 2432- Humana Press, Totowa, New Jersey. 2440. Lizasoain, I., Moro, M. A., Knowles, R. G., Novikov, L., Novikova, L., Kellerth, J.-O., Darley-Usmar, V., Moncada, S., 1996. Nitric 1995. Brain-derived neurotrophic factor oxide and peroxynitrite exert distinct effects on promotes survival and blocks nitric oxide mitochondrial respiration which are synthase expression in adult rat spinal Hart, et al. ▪ Neuroprotection by NAC 13 motoneurons after ventral avulsion. Neurosci. West, M. J., 1993. New stereological methods Lett. 200, 45-48. for counting neurons. Neurobiology of Aging Novikova, L. N., Novikov, L. N., Kellerth, J.- 14, 275-285. O., 2000. BDNF abolishes the survival effect of West, M. J., 1999. Stereological methods for NT-3 in axotomised Clarke's nucleus of adult estimating the total number of neurons and rats. J. Comp. Neurol. 428, 671-680. synapses: issues of precision and bias. Trends. Pan, Z. and Perez-Polo, R., 1996. Increased Neurosci. 22, 51-61. uptake of L-cysteine and L-cystine by nerve Wiberg, M., Hazari, A., Ljungberg, C., growth factor in rat pheochromocytoma cells. Petterson, K., Nordh, E., Kvast-Rabben, O., Brain Res. 740, 21-26. Terenghi, G., 2002. Sensory recovery after hand Park, D. S., Stefanis, L., Yan, C. Y. I., Farinelli, replantation: a clinical, morphological and S. E., Greene, L. A., 1996. Ordering the cell neurophysiological study in man. Scand. J. death pathway. Differential effects of BCL2, an Plast. Reconstr. Surg. Hand Surg. 36, 1-11. interleukin-1-converting enzyme family Wullner, U., Seyfried, J., Groscurth, P., protease inhibitor, and other survival agents on Beinroth, S., Winter, S., Gleichmann, M., JNK activation in serum/nerve growth factor- Heneka, M., Loschmann, P., Schulz, J. B., deprived PC12 cells. J. Biol. Chem. 271, Weller, M., Klockgether, T., 1999. Glutathione 21898-21905. depletion and neuronal cell death: the role of Pettmann, B. and Henderson, C. E., 1998. reactive oxygen intermediates and Neuronal cell death. Neuron 20, 633-647. mitochondrial function. Brain Res. 826, 53-62. Ratan, R. R., Murphy, T. H., Baraban, J. M., Yan, C. Y., Ferrari, G., Greene, L. A., 1995. N- 1994. Oxidative stress induces apoptosis in acetylcysteine-promoted survival of PC12 cells embryonic cortical neurons. J. Neurochem. 62, is glutathione- independent but transcription- 376-379. dependent. J. Biol. Chem. 270, 26827-26832. Sagara, J. I., Miura, K., Bannai, S., 1993. Yan, C. I. and Greene, L. A., 1998. Prevention Maintenance of neuronal glutathione by glial of PC12 cell death by N-acetylcysteine requires cells. J Neurochem. 61, 1672-1676. activation of the Ras pathway. J. Neurosci. 18, Sanders, E. J. and Wride, M. A., 1996. 4042-4049. Ultrastructural identification of apoptotic nuclei using the TUNEL technique. Histochem. J. 28, 275-281. Acknowledgements Seaton, T. A., Cooper, J. M., Schapira, A. H., The authors are grateful for the 1997. Free radical scavengers protect considerable technical support of Dr. Thomas dopaminergic cell lines from apoptosis induced Brannstrom of the Department of Medical by complex I inhibitors. Brain Res. 777, 110- Biosciences – Pathology, Umeå University, to 118. Mr. Andrew Wilson (Research Fellow at the Tandrup, T., 1993. A method for producing Blond-McIndoe Centre), and to Gunnel unbiased and efficient estimation of number and Folkesson and Dr. Lev Novikov of the mean volume of specified neuron subtypes in Department of Integrative Medical Biology, rat dorsal root ganglion. J. Comp. Neurol. 329, Section for Anatomy, Umeå University for the 269-276. majority of the work involved in preparing the Terenghi, G., 1999. Peripheral nerve transmission electron micrographs. regeneration and neurotrophic factors. J. Anat. This work was funded by the Swedish 194, 1-14. Medical Research Council (Grant 02286), Wagner, R., Heckman, H. M., Myers, R. R., County of Vasterbottens, Ane’rs Foundation, 1998. Wallerian degeneration and hyperalgesia and the East Grinstead Medical Research Trust after peripheral nerve injury are glutathione- (Registered Charity No. 258154). dependent. Pain 77, 173-179. Wang, X. F. and Cynader, M. S., 2000. Astrocytes provide cysteine to neurons by The Authors releasing glutathione. J. Neurochem. 74, 1434- 1,2,3 1442. Andrew McKay Hart BSc MD (Hons), MRCS AFRCS Surgical Research Fellow, The Department of Surgical & Perioperative Hart, et al. ▪ Neuroprotection by NAC 14

Science, Umeå University & The Blond- Prof. Mikael Wiberg 2,3 PhD MD, Professor, McIndoe Centre. Depts of Anatomy and Hand & Plastic Surgery

Prof. Jan-Olof Kellerth 3 PhD, Professor, Umeå University Dept of Anatomy, Umeå University.

Correspondence to Giorgio Terenghi, PhD, Prof. Giorgio Terenghi 1 PhD FRCPath, FRCPath Director & Head of Nerve Regeneration Research, The Blond-McIndoe Centre. Paper submitted 15 March 2003.

Hart, et al. ▪ Leukaemia inhibitory factor in late secondary repair Accepted

BRITISH JOURNAL OF PLASTIC SURGERY

Exogenous leukaemia inhibitory factor enhances nerve regeneration after late secondary repair using a bioartificial nerve conduit.

Andrew McKay Hart1,2,3, Prof. Mikael Wiberg2,3, Prof. Giorgio Terenghi1

SUMMARY. The clinical outcome of peripheral nerve injuries remains disappointing, even in the ideal situation of a primary repair performed with optimal microsurgical technique. Yet primary repair is appropriate for only ~85% of injuries, and outcome is worse following secondary nerve repair, partly due to the reduced regenerative potential of chronically axotomised neurons. Leukaemia inhibitory factor (LIF) is a gp-130 neurocytokine that is thought to act as an “injury factor”, triggering the early injury phenotype within neurons and potentially boosting their regenerative potential after secondary nerve repair. Two or four months after sciatic nerve axotomy in the rat, 1cm gap repairs were performed using either nerve isografts, or poly-3- hydroxybutyrate conduits containing a calcium alginate and fibronectin hydrogel. Regeneration was determined by quantitative immunohistochemistry 6 weeks after repair, and the effect of incorporating recombinant LIF (100ng/ml) into the conduits assessed. LIF increased the regeneration distance in repairs performed after both 2 (69%, p=0.019) and 4 month delays (123%, p=0.021), and was statistically comparable to nerve graft. The total area of axonal immunostaining increased by 21% (p>0.05) and 63% (p>0.05) respectively. Percentage immunostaining area was not increased in the 2 month group, but increased by 93% in the repairs performed 4 months after axotomy. Exogenous LIF therefore has a potential role in promoting peripheral nerve regeneration after secondary repair, and can be effectively delivered within poly- 3-hydroxybutyrate bioartificial conduits used for nerve repair.

Keywords: peripheral nerve regeneration, peripheral nerve repair, leukaemia inhibitory factor, nerve conduit, growth factors.

Peripheral nerve injuries are amongst period of denervation atrophy in target muscles the commonest injuries encountered in trauma and sensory organs, while endoneurial fibrosis surgery, with an estimated annual incidence of 7,8 combines with Schwann cell atrophy and 1/1000 population. Yet despite major technical phenotypic changes 7,9,10 to impair the distal advances in surgical repair outcome remains nerve stump’s ability to support neurite disappointingly poor 1, particularly in the 11- ingrowth after repair 10,11. Due to nerve stump 16% of injuries unsuitable for primary retraction the repair usually involves bridging a neurorraphy 2-4. gap, yet fibres must still reach type specific, Several factors account for the fact and topographically appropriate endoneurial that outcome is worst 4-6 after late secondary tubes in order to have any chance of nerve repair, many of which cannot be functionally useful regeneration 12. Within the improved by further advances in microsurgical dorsal root ganglia up to 40% of all primary technique. The delay in repair prolongs the sensory neurons will die 13, massively reducing Hart, et al. ▪ Leukaemia inhibitory factor in late secondary repair 2 the pool of neurons available for regeneration, harvested 1cm syngeneic reversed sciatic nerve and leading to a ~70% loss of CNS endings 14. isografts, or bioartificial nerve conduits. The Furthermore, both the axotomised conduits were 1.4cm long tubes made from motorneurones and the surviving sensory compressed sheets of longitudinally oriented neurons begin to exhibit loss of their early poly-3-hydroxybutyrate (PHB) fibres (Astra injury phenotype, rendering them less able to Tech, Mölndal, Sweden), filled with ~75µl of a regenerate after any subsequent repair 5,15. hydrogel that comprised 2% ultrapure, low Although promising strategies are molecular weight, high mannuronic acid under development to address many of these content, calcium alginate (Pronova, Sweden) problems, restoration of the early injury, or and 0.05% bovine fibronectin (Sigma regenerative, phenotype has not been described. Pharmaceuticals, U.K.). This gel served as a In the immediate phase after axotomy induction matrix substitute, and as a delivery vehicle for of this neuronal phenotype may result from the recombinant murine leukaemia inhibitory factor action of retrogradely transported “injury (rhLIF 100ng/ml, Autogen Bioclear, U.K.) in factors” that undergo receptor mediated uptake, the two treatment groups. The nerve stumps and act synergistically with other neurotrophic were trimmed back to “healthy” nerve, and factors. Such injury factors are expressed at low inset 2mm into either end of the conduit using levels in normal peripheral nerve but are paired horizontal mattress sutures to leave a rapidly upregulated after injury, thereby 1cm gap across which the nerve had to mimicking the timecourse of the injury regenerate. phenotype in axotomised neurons. Increasing Six weeks later the repairs were evidence exists for the gp-130 neuroregulatory harvested in continuity with ~3mm of both the cytokines, principally leukaemia inhibitory proximal and distal nerve, fixed in Zamboni’s factor (LIF), fulfilling this role after peripheral solution at 4oC, and equilibrated in nerve injury 16. cryoprotectant solution (15% sucrose in If the increased regenerative capacity phosphate buffered saline). Blocks were then of acutely injured neurons is due to injury prepared by rapid freezing into OCT™ factors, then local administration ought to mounting medium, from which a systematic enhance the regenerative capacity of random sample of longitudinal, 15µm chronically axotomised neurons. This study cryosections was collected. These were stained therefore sought to determine whether nerve by fluorescence immunohistochenistry to show regeneration after late secondary repair would nerve fibres by using primary antisera against be enhanced by exogenous LIF, delivered pan-axonal marker of neurofilament within a bioartificial nerve conduit used for gap (monoclonal antibody against PAMNF, Affiniti repair. Research Products Ltd.), and a secondary horse anti-mouse antisera conjugated to Cy-3 Methods (Amersham Pharmacia Biotech). The Schwann cell marker S-100 (polyclonal antibody, All work was conducted in keeping DAKO, Denmark) was also used, in with the terms of the Animals (Scientific combination with a goat anti-rabbit antisera Procedures) Act 1986, and the experimental conjugated to fluorescein isothyocyanate design was cogniscent of the need to optimise (FITC, Vector Laboratories Inc., U.S.A.). welfare. PAMNF staining therefore appeared as red Under inhalational anaesthesia fluorescence, and S-100 appeared as green. (Halothane, May & Baker Ltd., U.K.), six Non-specific tissue binding of the secondary groups of young adult male Sprague-Dawley antibodies was limited by pre-incubation of rats (n=5) underwent unilateral sciatic nerve tissue sections with normal goat (Sigma- division at the upper border of quadratus Aldrich U.K.) or normal horse (Sigma-Aldrich femoris in the mid-thigh. The nerve stumps U.K.) sera, and non-specific hydrophobic were ligated with 6/0 Prolene, and placed attachment was minimised by co-incubation within silicon rubber caps to prevent with normal rat serum (DAKO, Denmark). spontaneous regeneration. Either 2 or 4 months Two outcome measures were later, the animals were anaesthetised, and 1cm quantified. The “regeneration distance” was gap repairs were performed using either freshly Hart, et al. ▪ Leukaemia inhibitory factor in late secondary repair 3 measured from the end of the proximal nerve Results stump to the tip of the most distal nerve fibre in each repair, by examining all sections under a Six weeks after implantation the 10x objective lens with a 0.1mm graticulated nerve conduits exhibited minimal fibrosis, and eyepiece. Next, using a high resolution fluid good neovascularisation along their full length; cooled digital camera (Diagnostic Instruments, the nerve grafts were equally healthy in U.S.A.) and a 20X objective lens, a band of appearance. At the microscopic level there was images was captured across the full width of a qualitative improvement in nerve fibre two non-consecutive sections that lay most morphology within the conduits that contained central to the growth cone in each repair. The rhLIF, as compared to the plain conduits. area of immunostaining within this band of In the nerves repaired 2 months after images was then quantified by image analysis axotomy fibres were found to have regenerated (Image-Pro-Plus™ Version 4.0 software); a into the distal nerve stump when nerve grafts process reliant upon intensity thresholding for were employed (Figure 1), but had proceeded a unichromatic light, deselection of occasional significantly shorter distance through the plain areas of background staining, and digital PHB conduits (P<0.001). The addition of rhLIF quantification of the area stained. The total area resulted in a 69% increase in regeneration of immunostaining was summated for each distance (p=0.019), and was comparable to that section, as was the total area of the nerve repair found within nerve grafts (Figure 1). A similar present in the captured images in order to pattern was found when the area of express immunostaining as a percentage of the immunostaining for PAMNF was examined area of the nerve repair examined, thereby (Table 1). Nerve graft repairs contained the facilitating comparison between conduits of highest total area of staining, and although the different diameters. The “total area of area within plain PHB conduits was 56% less than this, the addition of rhLIF was associated immunostaining” was expressed in µm2, and with a 21% increase. The highest percentage the “percentage area” quoted directly. A total of area of immunostaining for PAMNF was also ten measurements of the area of found within nerve grafts, reflecting their immunostaining (both absolute and percentage) comparatively small diameter, while plain PHB was therefore recorded for each antiserum in conduits an 87% lower percentage area each experimental group, and the mean value (P<0.001). Conduits containing rhLIF had a used for statistical examination of the effect of similar percentage area of staining to plain treatment. PHB conduits. Statistical analysis of results was When nerve repair was delayed until performed by paired t-test, using SigmaStat™ 4 months after axotomy fibres still reached the (Version 2.0) software, after confirming the distal nerve stump when nerve graft was used, normality of data distribution by the but the regeneration distance within the plain Kolmogorov-Smirnov test.

Regeneration 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Distance (mm) 13.3mm Nerve Graft (1.5) ** 12.4mm (2.4) ** Plain Conduit 6.0mm (3.0) 4.9mm (2.6) 10.2mm (1.0) * rhLIF Conduit 10.8mm(3.9) * Figure 1: Nerve regeneration distance 6 weeks following sciatic nerve repair Sciatic nerve repairs were performed either 2 months (solid arrows), or 4 months (dashed arrows) after nerve transection. Regeneration distance was quantified 6 weeks after repair, by measuring the distance (mm) from the end of the proximal nerve stump, to the tip of the most distal nerve fibre, in longitudinal sections stained by fluorescence immunohistochemistry with primary antiserum to pan-axonal marker of neurofilament (PAMNF). Figures given are the mean (SD) for each experimental group (n=5). *P<0.05, **P≤0.001 (compared to repair using plain conduits). Hart, et al. ▪ Leukaemia inhibitory factor in late secondary repair 4

PHB conduits was even worse (Figure 1). addition of rhLIF. However in the group Despite this, the effect of rhLIF was more repaired after a 4 month delay the rhLIF pronounced than after a two month delay in conduits contained somewhat more staining nerve repair, and resulted in a 123% increase (18% increase in percentage area of (p=0.021) in immunostaining cf. plain conduits). regeneration distance to a figure that was not significantly different (P>0.05) to that achieved Discussion with nerve graft. Similarly, the total area of immunostaining for PAMNF within nerve Further major advances in the grafts was 71% greater (Table 1) than was treatment of peripheral nerve injuries are more present within plain PHB conduits, although likely to result from modulating the response of the addition of rhLIF resulted in a 63% neurons and their supporting cells to injury and increase. Again, the percentage area of regeneration, than from further enhancement of immunostaining for PAMNF was greatest when microsurgical technique. This approach may nerve graft was used, the figure for plain PHB involve novel adjuvant therapies to prevent 17 conduits being considerably lower. Percentage neuronal death , techniques such as short gap area of staining for PAMNF increased by 93% repair to facilitate the selection of appropriate 18 with the addition of rhLIF to the conduits, endoneurial tubes by regenerating neurites , when repair was delayed for four months rather and modulation of the normal regenerative than two. responses of both neurons and their supporting 19 Quantification of immunostaining for cells . In experimental models of primary S-100 revealed that nerve graft repairs nerve repair promising results have been found 19 contained a greater total area than plain PHB with the use of exogenous growth factors , conduits (Table 2), which had 75% and 68% and bioengineered conduits are under lower values in the groups repaired 2 and 4 development to replace autologous nerve graft, months after axotomy respectively. The some incorporating cultured Schwann cells 20,21 percentage area of immunostaining was 93% . Although a significant minority of nerve lower in the plain conduits than in equivalent injuries are not suitable for primary neurorraphy 4, and their outcome is nerve graft repairs in both these groups. In the 3 group whose sciatic nerve was repaired after a compromised accordingly , the specific 2 month delay the area of S-100 staining in the therapeutic problems inherent to secondary plain PHB conduits was not increased by the repair have rarely been clearly addressed

Repair After 2 Month Repair After 4 Month Repair After 2 Month Repair After 4 Month Delay Delay Delay Delay Mean (SD) Mean (SD) Mean (SD) Mean (SD) Type of Type of Total Total Repair % Area % Area Repair Total Total Area of Area of of of Area of % Area Area of % Area PAMNF PAMNF PAMNF PAMNF S-100 of S-100 S-100 of S-100 Staining Staining Staining Staining Staining Staining Staining Staining (µm2) (µm2) (µm2) (µm2)

Nerve 58259 7.3 ** 56890 * 7.1 ** Nerve 112295 * 14** 72955 * 9.7 ** Graft (25405) (2.3) (24546) (1.9) Graft (29902) (2.1) (21562) (3.7)

Plain 25471 0.93 16488 0.52 Plain 28290 1.1 23593 0.75 Conduit (22634) (0.81) (12957) (0.36) Conduit (26305) (1.1) (14755) (0.39)

rhLIF 30744 0.82 26802 1.0 rhLIF 24586 0.64 23745 0.89 Conduit (13047) (0.49) (21594) (0.78) Conduit (7084) (0.30) (15873) (0.55)

Table 1: Area of immunostaining against PAMNF within Table 2: Sciatic nerve repairs were performed either 2, or sciatic nerve repairs. Sciatic nerve repairs were performed 4 months after nerve transection. Area of immunstaining either 2, or 4 months after nerve transection. Area of against S-100 was quantified 6 weeks after repair, by immunstaining against PAMNF was quantified 6 weeks digital image analysis of a band of images across the full after repair, by digital image analysis of a band of images width of the repair at a position 1mm distal to the end of across the full width of the repair at a position 1mm distal to the proximal nerve stump. This was expressed either as the end of the proximal nerve stump. This was expressed the mean (SD) total area of immunostaining (µm2) for either as the mean (SD) total area of immunostaining (µm2) each experimental group, or as a percentage of the entire for each experimental group, or as a percentage of the entire area of the band of images that was captured (“percentage area of the band of images that was captured (“percentage area of staining”). area of staining”). *P<0.05, **P≤0.001 (compared to repair using plain *P<0.05, **P≤0.001 (compared to repair using plain conduits). conduits). Hart, et al. ▪ Leukaemia inhibitory factor in late secondary repair 5 experimentally. The reduction in regenerative Endoneurial fibrosis causes a potential may be due to loss of the early injury, mechanical impedance to neurite ingrowth or regenerative phenotype, which is within chronically denervated distal nerve characterised by chromatolysis, and stumps, and begins within 28-35 days of nerve upregulation of markers such as CGRP, transection 8. It is significant by 3-6 months cytoskeletal proteins, GAP-43 and galanin 15. after injury 7,22, endoneurial tube diameter is Such phenotypic changes may result from the reduced to 26% of control one year after action of a range of retrogradely transported transection and to only 1% after two years 8,22, “injury” factors, such as leukaemia inhibitory and yet there are no therapies available to factor (LIF). prevent it. Further functional inhibition of LIF is a neurotrophic gp-130 regeneration results from the profound atrophy neurocytokine 15,16,26 that is expressed at and loss of neuregulin responsiveness that extremely low levels in normal peripheral nerve occurs within the distal stump Schwann cell and muscle, but which undergoes rapid population 9,10,10, although the use of cultured upregulation and increased specific receptor autologous Schwann cells and exogenous glial mediated retrograde transport after injury 16,27- growth factor holds promise in addressing these 29. Although very few intact adult changes 23. motorneurones, and only around ~24% of However, irrespective of the state of uninjured primary sensory neurons the distal nerve stump, there is evidence to (predominantly small, nociceptive) exhibit the suggest that chronically axotomised neurons LIF-receptor, peripheral nerve injury induces display decreased regenerative capacity. marked upregulation by a majority of neurons Although early studies suggested only a minor 16,30, rendering them sensitive to LIF. loss of regenerative capacity 24, this has been Exogenous LIF stimulates phenotypic changes questioned more recently in studies where the akin to axotomy in intact primary sensory distal stump effect is obviated by suturing onto neurons 26,31-33, and after axotomy is fresh distal stumps, or by using conduits. The neuroprotective for both sensory and capacity of motorneurons to regenerate into motorneurons when applied to the proximal fresh distal stumps decreases with prolonged stump in neonates 34,35. It meets the criteria for axotomy 25, and expression by sensory neurons an injury factor, given that it’s timecourse of of the regenerative neuropeptide GAP-43 is lost expression is intimately related to injury, it acts 2 months after axotomy 10. Furthermore, by specific, receptor mediated uptake and complete loss of regenerative capability retrograde transport, and it has trophic and through denatured muscle grafts has been tropic actions 16,28. Furthermore, LIF enhances demonstrated after more than 56 days delay, neurite outgrowth in vitro, and pre-treatment with any delay impairing myelination 5. Thus, with antibodies against the gp-130 in the rat, it is evident that regenerative neurocytokine receptor moiety inhibit the capacity diminishes with delays of greater than response of sensory neurones to axotomy 26. around 60 days. Hence in this study nerve As a therapeutic agent, LIF also has repair was delayed for a minimum of two the potential benefit over traditional months, and as expected the regeneration neurotrophic factors, such as nerve growth distance and area of staining within plain PHB factor, that it acts upon both motor and sensory conduits and nerve grafts, were reduced when neurons after injury, and is synergistic with there was a 4 month delay prior to repair, as other neurotrophic factors, such as nerve compared to a 2 month delay. A 1cm gap growth factor 16. In primary nerve repair LIF repair was used since the purpose of this study resulted in significant increases in nerve was not to examine the effect of rhLIF upon the regeneration through a silicone tube, and in length of gap across which nerve can subsequent motor function 36,37. LIF is also regenerate, but to examine its effects upon myotrophic, and enhances motor reinnervation regeneration across a standard, bridgeable by axon sprouts 16,29. distance with the extra biological hurdle of a Administration of rhLIF within a significant delay prior to nerve repair. nerve conduit has several advantages over the systemic route. Although LIF acts at the Hart, et al. ▪ Leukaemia inhibitory factor in late secondary repair 6 neuronal cell body, receptor mediated uptake postulated that this is a more sensitive measure and retrograde transport ensures that rhLIF of functional regeneration than the area of administered at the site of injury will reach the staining, which may be increased by the cell body in a physiological manner. The tropic extension of multiple random neurites rather role of LIF is used to advantage by delivery at than functional axonal elongation 36. In the repair site, and systemic complications can keeping with this, rhLIF has been found to be avoided due to the ability to use lower doses. improve axonal morphometrics and function Furthermore, nerve conduits are after primary repair, but with reduced axonal useful in themselves, given the potential donor counts 36,37. The effect of rhLIF upon morbidity associated with the use of autologous regeneration distance and area of staining for nerve graft, and are showing promise in clinical nerve fibres was relatively greater when nerve trials 38. This may reflect the greater freedom repair was delayed by 4 months (distance 123% that regenerating neurites have in determining increased, total area 63% & percentage area which endoneurial tubes they enter, thereby 93% increased), rather than 2 months (distance potentially improving type, or topographical 69% increased, total area 21% increased, specificity. PHB conduits have been shown to percentage area not increased). This is in be effective in supporting nerve regeneration keeping with the concept that the regenerative over gaps of 1-4cm 20,39, whilst calcium capacity of the neurons deteriorates as the alginate has been shown to give sustained period of axotomy increases, and that rhLIF release of bioavailable rhLIF for several tends to restore that regenerative potential by months in vivo 40. A similar concentration of triggering the early injury phenotype. rhLIF was therefore employed in this study, Although conduits containing rhLIF and fibronectin was added since it enhances did not support regeneration quite as well as neurite outgrowth, and has been found to be nerve graft, it is possible that by using a higher advantageous when combined with rhLIF 37,41. concentration of rhLIF a greater effect upon It is clear that, in common with other regeneration might be obtained, and other non-bioactive synthetic conduits, the plain PHB studies using saline as a delivery vehicle have conduits did not adequately support peripheral employed concentrations in the order of 0.5- nerve regeneration when compared to the 0.9µg/µl, to a total dose of ~10µg 34,36,37. It is clinical gold standard of nerve graft. This also possible that regeneration could be further presumably reflects the lack of any Schwann enhanced by varying the chemical, or physical cell element to the conduit, in contrast to the characteristics of the alginate. However rhLIF endogenous population within a nerve graft, need not be the sole additive in a bioartificial and this difference is heightened in the setting conduit, and further changes in conduit design of delayed repair given the deleterious changes may be determined by the need to incorporate occurring within the recipient nerve’s Schwann features that enhance Schwann cell function. cell population. This was confirmed by the This study therefore demonstrates significantly lower areas of staining for S-100 that PHB conduits filled with a calcium found within the conduits as compared to the alginate and fibronectin hydrogel matrix are an nerve grafts. effective delivery system for rhLIF, which Despite this, the addition of rhLIF appears to have more effect upon regeneration resulted in a significant improvement in the distance, than the area of staining for nerve regenerative profile of the conduits. Although fibres. LIF becomes relatively more effective the area of staining for nerve fibres showed as the delay between injury and repair only modest increases after the addition of increases, and it is evident that rhLIF has rhLIF, regeneration distance increased considerable potential as one component in the significantly, and was not significantly worse development of a bioartificial nerve conduit to than that obtained with nerve graft. The replace autologous nerve grafts for late regeneration distance, particularly combined secondary nerve repairs. with the observed morphological improvement, reflects not merely the extension of multiple neurites, but stimulation of directed axonal growth into the distal nerve stump. It has been Hart, et al. ▪ Leukaemia inhibitory factor in late secondary repair 7

Acknowledgements 10. Li H, Terenghi G, Hall SM. Effects of delayed re-innervation on the expression This work was funded by the of c-erbB receptors by chronically Swedish Medical Research Council (Grant denervated rat Schwann cells in vivo. 02286), County of Vasterbottens, Ane’rs Glia 1997; 20: 333-347. Foundation, and the East Grinstead Medical 11. Fu SY, Gordon T. Contributing factors to Research Trust (Registered Charity No. poor functional recovery after delayed 258154). The authors would like to thank nerve repair: prolonged denervation. J AstraTec for the unqualified donation of Neurosci 1995; 15: 3886-3895. supplies of poly-3-hydroxybutyrate sheets, and 12. Lundborg G, Dahlin L, Danielsen N, Zhao the Stanley Thomas Johnson Foundation for Q. Trophism, tropism, and specificity in their generous financial support. nerve regeneration. J Reconstr Microsurg 1994; 10: 345-354. References 13. Hart AM, Brannstrom T, Wiberg M, Terenghi G. Primary sensory neurons 1. Lundborg G, Dahlin L, Danielsen N, Zhao and satellite cells after peripheral Q. Trophism, tropism, and specificity in axotomy in the adult rat: timecourse of nerve regeneration. J Reconstr cell death and elimination. Exp Brain Microsurg 1994; 10: 345-354. Res 2002; 142: 308-318. 2. de Medinaceli L, Prayon M, Merle M. 14. Liss AG. Peripheral nerve injury and its Percentage of nerve injuries in which effects on the central nervous system. A primary repair can be achieved by end- study of transection of sensory nerves in to-end approximation: review of 2,181 cats and monkeys. Acta Universitatis nerve lesions. Microsurgery 1993; 14: Upsaliensis 1995. 244-246. 15. Fu SY, Gordon T. The cellular and 3. Vanderhooft E. Functional outcomes of molecular basis of peripheral nerve nerve grafts for the upper and lower regeneration. Mol Neurobiol 1997; 14: extremities. Hand Clin. 2000; 16: 93- 67-116. 104. 16. Murphy M, Dutton R, Koblar S, Cheema 4. Allan CH. Functional results of primary S, Bartlett P. Cytokines which signal nerve repair. Hand Clin. 2000; 16: 67- through the LIF receptor and their 72. actions in the nervous system. Prog 5. Gattuso JM, Glasby MA, Gschmeissner Neurobiol 1997; 52: 355-378. SE, Norris RW. A comparison of 17. Hart AM, Wiberg M, Youle M, Terenghi immediate and delayed repair of G. Systemic acetyl-L-carnitine peripheral nerves using freeze thawed eliminates sensory neuronal loss after autologous muscle grafts- in the rat. Br peripheral axotomy: A new clinical J Plast Surg 1989; 42: 306-313. approach in the management of 6. Birch R., Bonney G., Wynn Parry C.B. peripheral nerve trauma. Exp Brain Res Surgical Disorders of the Peripheral 2002; in press. Nerves. Edinburgh: Churchill- 18. Lundborg G, Rosen B, Dahlin L, et al. Livingstone, 1998. Tubular versus conventional repair of 7. Roytta M, Salonen V. Long-term median nerve in the human forearm: endoneurial changes after nerve early results from a prospective transection. Acta Neuropathol 1988; 76: randomised clinical study. J Hand Surg. 35-45. 1997; 22A: 99-106. 8. Green B. Repair and grafting of the 19. Terenghi G. Peripheral nerve regeneration peripheral nerve. In: Green, ed. Plastic and neurotrophic factors. J Anat 1999; Surgery. 1998; 630-695. 194: 1-14. 9. Terenghi G, Calder JS, Birch R, Hall SM. 20. Hazari A, Wiberg M, Johansson-Ruden G, A morphological study of Schwann cells Green C, Terenghi G. A resorbable and axonal regeneration in chronically nerve conduit as an alternative to nerve transected human peripheral nerves. J Hand Surg 1998; 4B: 1-5. Hart, et al. ▪ Leukaemia inhibitory factor in late secondary repair 8

autograft in nerve gap repair. Br J Plast 30. Thompson SW, Vernallis AB, Heath JK, Surg 1999; 52: 653-657. Priestley JV. Leukaemia inhibitory 21. Mosahebi A, Woodward B, Wiberg M, factor is retrogradely transported by a Martin R, Terenghi G. Retroviral distinct population of adult rat sensory labeling of Schwann cells: in vitro neurons: co-localization with trkA and characterisation and in vivo other neurochemical markers. transplantation to improve peripheral Eur.J.Neurosci 1997; 9: 1244-1251. nerve regeneration. Glia 2001; 34: 8-17. 31. Corness J, Stevens B, Fields RD, Hokfelt 22. Miyamoto Y, Higaki T, Sugita T, Ikuta Y, T. NGF and LIF both regulate galanin Tsuge K. Morphological reaction of gene expression in primary DRG cellular elements and the endoneurium cultures. Neuroreport. 1998; 9: 1533- following nerve section. Peripheral 1536. Nerve Repair & Regeneration 1986; 3: 32. Sun Y, Zigmond RE. Leukaemia 7-18. inhibitory factor induced in the sciatic 23. Hart AM, Mosahebi A, Wiberg M, nerve after axotomy is involved in the Terenghi G. Glial growth factor & induction of galanin in sensory neurons. Schwann cells improved nerve Eur J Neurosci 1996; 8: 2213-2220. regeneration following delayed repair. J 33. Ozturk G, Tonge G. Effects of leukaemia Periph Nerv Sys 2001; 6: 146. inhibitory factor on galanin expression 24. Holmes W, Young JZ. Nerve regeneration and on axonal growth in adult dorsal after immediate and delayed suture. J root ganglion neurons in vitro. Exp Anat 1942; 77: 63-96. Neurol 2001; 169: 376-385. 25. Fu SY, Gordon T. Contributing factors to 34. Cheema SS, Richards LJ, Murphy M, poor functional recovery after delayed Bartlett PF. Leukaemia inhibitory factor nerve repair: prolonged axotomy. J rescues motoneurones from axotomy- Neurosci 1998; 15: 3876-3855. induced cell death. Neuroreport 1994; 5: 26. Thompson SW, Priestley JV, Southall A. 989-992. gp130 cytokines, leukemia inhibitory 35. Cheema SS, Richards L, Murphy M, factor and interleukin-6, induce Bartlett PF. Leukemia inhibitory factor neuropeptide expression in intact adult prevents the death of axotomised rat sensory neurons in vivo: time-course, sensory neurons in the dorsal root specificity and comparison with sciatic ganglia of the neonatal rat. J Neurosci nerve axotomy. Neuroscience 1998; 84: Res 1994; 37: 213-218. 1247-1255. 36. Donato R, Cheema S, Finkelstein D, 27. Kurek JB, Austin L, Cheema SS, Bartlett Bartlett P, Morrison W. Role of PF, Murphy M. Up-regulation of leukaemia inhibitory factor (LIF) in rat leukaemia inhibitory factor and peripheral nerve regeneration. Ann interleukin-6 in transected sciatic nerve Acad Med Singapore 1995; 24: 94-100. and muscle following denervation. 37. Tham SK, Dowsing B., Finkelstein D., et Neuromuscul Disord 1996; 6: 105-114. al. Leukaemia Inhibitory Factor 28. Curtis R, Scherer SS, Somogyi R, et al. enhances the regeneration of transected Retrograde axonal transport of LIF is rat sciatic nerve and the function of increased by peripheral nerve injury: reinnervated muscle. J Neurosci Res correlation with increased LIF 1997; 47: 208-215. expression in distal nerve. Neuron 1994; 38. Weber RA, Dellon AL. Synthetic nerve 12: 191-204. conduits: indications and technique. 29. Kami K, Morikawa Y, Kawai Y, Senba E. Sem Neurosurg 2001; 12: 65-79. Leukaemia inhibitory factor, glial cell 39. Young RC, Wiberg M, Terenghi G. Poly- line-derived neurotrophic factor, and 3-hydroxybutyrate (PHB): a resorbable their receptor expressions following conduit for long gap repair in peripheral muscle crush injury. Muscle & Nerve nerves. Br J Plast Surg 2002; 55: 235- 1999; 22: 1576-1586. 240. Hart, et al. ▪ Leukaemia inhibitory factor in late secondary repair 9

40. Austin L, Bower JJ, Kurek JB, Muldoon Surgical & Perioperative Science, Umeå CM. controlled release of leukaemia University. inhibitory factor (LIF) to tissues. Growth Factors 1997; 15: 61-68. Prof. Mikael Wiberg2,3 PhD MD, Professor, 41. Mosahebi A, Wiberg M, Terenghi G. Depts of Anatomy and Hand & Plastic Surgery Addition of fibronectin to alginate Umeå University matrix improves peripheral nerve Prof. Giorgio Terenghi1 PhD FRCPath, regeneration in tissue engineered Director & Head of Nerve Regeneration conduits. Tissue Eng. 2003; in press. Research, The Blond-McIndoe Centre.

Correspondence to Giorgio Terenghi, PhD, The Authors FRCPath Andrew McKay Hart1,2,3 BSc MD (Hons), MRCS AFRCS Surgical Research Fellow, The Paper received 8 July 2002 Blond-McIndoe Centre & The Department of Accepted 5 February 2003, no revisions

Hart, et al. ▪ GGF & Schwann cells in late secondary repair Submitted

JOURNAL OF THE PERIPHERAL NERVOUS SYSTEM

Interaction of glial growth factor and cultured Schwann cells in regeneration after chronic nerve injury.

1,2,3 1,2 2,3 1 Andrew McKay Hart , Afshin Mosahebi , Mikael Wiberg , Giorgio Terenghi

SUMMARY. Schwann cells (SC) atrophy and exhibit reduced glial growth factor (GGF) responsiveness following prolonged axotomy. Resultant endoneurial fibrosis and impaired support of axonal regrowth is probably a major factor in the poor clinical outcome of secondary nerve repair. The effect of boosting the endogenous Schwann cell population with recombinant GGF-2 (rhGGF-2), or augmenting it with cultured SC was therefore determined in a rat model. Two or four months after unilateral sciatic nerve transection, 1cm gap repairs were performed using either nerve isografts (NG), poly-3-hydroxybutyrate (PHB) conduits filled with calcium alginate-fibronectin hydrogel, or similar conduits containing either rhGGF-2 (500ng/ml), SC (80x106/ml), or both (SC+rhGGF-2). Six weeks later regeneration was assessed morphologically, and by quantitative immunohistochemistry. Nerve regeneration distance was significantly increased through conduits containing SC or GGF when repair was delayed by 2 months (SC 80% increase, rhGGF-2 87%; P=0.007) or by 4 months (SC 150%, rhGGF-2 135%; P<0.001), and was greatest with SC+rhGGF-2, even when compared to NG. Area of staining for nerve fibres was dramatically increased by rhGGF-2, but less so by SC. Total area stained was greater within SC+rhGGF-2 conduits than within NG. SC had relatively more effect when nerve repair was delayed by 4 months than by 2, while the reverse was true for rhGGF-2. PHB conduits are therefore technically simple, but effective delivery systems for cultured autologous SC and rhGGF-2. Their potential to enhance regeneration after secondary nerve repair is demonstrated, and when used in combination resulted in a better regenerative profile than nerve graft.

Keywords: peripheral nerve regeneration, peripheral nerve repair, leukaemia inhibitory factor, nerve conduit, growth factors. 2000). Hence there is a need to shift focus Introduction away from surgical technique and towards therapeutic modulation of peripheral nerve Peripheral nerve injuries remain a neurobiology. major cause of functional morbidity, since they Several factors account for the fact are common, involve major nerve trunks in that outcome is worst after late secondary nerve ~35% of cases (Allan, 2000), and their clinical repair (Allan, 2000; Birch,. et al., 1998; outcome remains disappointingly poor (Allan, Gattuso, et al., 1989; Hall, 2001). Death of 2000; Lundborg, 2000a; Bruyns, et al., 2003). primary sensory neurons begins within 24 hours Despite major advances in surgical technique of injury and ~40% will subsequently die (Hart, outcome has not dramatically improved over et al., 2002a), as eventually will ~20% of the last 25 years (Lundborg, 2000a), and motorneurons (Ma et al., 2001). This massively remains worst in the 11-16% of injuries that are reduces the pool of neurons available for unsuitable for primary neurorraphy (Allan, regeneration, and leads to the loss of ~70% of 2000; de Medinaceli, et al., 1993; Vanderhooft, CNS endings and breakdown of cortical Hart, et al. ▪ GGF & Schwann cells in late secondary repair 2 mapping (Liss, 1995; Lundborg, 2000a; design was cogniscent of the need to optimise Lundborg, 2000b). Surviving neurons may then welfare. lose their early injury phenotype, impairing In order to obtain Schwann cells for their capacity to regenerate after subsequent culture, sciatic nerves were harvested from 20 repair (Fu and Gordon, 1997; Gattuso, et al., neonatal rats and digested for 45 minutes using 1989). Due to nerve stump retraction a gap 1% collagenase I (Worthington Biochemicals) usually must be bridged, and currently this and 0.25% trypsin (Gibco). The digestant was indicates the use of autologous nerve graft, with then triturated through 21G and 23G needles, associated donor morbidity problems. Fibres filtered (70 µm cell filter, Falcon), and must then reach type and topographically centrifuged at 800 revolutions per minute (rpm) appropriate endoneurial tubes in order to have for 5 minutes. The resulting cells were any chance of functionally useful regeneration resuspended in a basic medium comprising (Lundborg, et al., 1994), while endoneurial Dulbecco’s Minimum Eagles Medium plus fibrosis (Green, 1998; Roytta and Salonen, Glutamax  (DMEM, Gibco), plus penicillin 1988) combines with Schwann cell atrophy and 100 iu/ml and streptomycin 100µg/ml (Gibco), phenotypic changes (Hall, 2001; Li, et al., and 10% fetal calf serum (FCS, Imperial 1997; Roytta and Salonen, 1988; Terenghi, et Laboratories). After 24 hours incubation in a al., 1998) to impair the distal stumps ability to 25cm2 poly-D-lysine (Sigma) coated flask support that regeneration (Fu and Gordon, stored at 37°C, 95% humidity, and 5% CO2, the 1995; Li, et al., 1997). Meanwhile denervation basic medium was then exchanged for one atrophy within the target motor and sensory containing 10µM cytosine-β-D- organs is exacerbated by the prolonged delay in arabinofuranoside (Sigma), and incubated a reinnervation. further 48 hours to stop fibroblast growth. Since Although the use of exogenous the majority of the remaining cells were now neurotrophic factors to reduce neuronal death, Schwann cells (SC), the medium was then and enhance regeneration after primary repair changed to a growth medium that comprised has been investigated for many years basic medium containing recombinant glial (Lundborg, 2000a; Terenghi, 1999), it is only growth factor II 63ng/ml (Max1/2 activity, 4.8 recently that a clinically applicable treatment ng/ml, Cambridge NeuroScience), and 10µM has been described that prevents sensory forskolin (Calbiochem). After reaching neuronal loss (Hart, et al., 2002d), and that confluence the SC were further purified by secondary repair has been addressed (Hart, et trypsinisation (2.5% Gibco, at 37°C) and al., 2002b; Hart, et al., 2002c). However, centrifuging the suspension at 800 rpm for 5 given that Schwann cells are essential for minutes. Cells were then suspended in 1ml of peripheral nerve regeneration (Hall, 1986; medium containing 1:1000 mouse anti Thy 1 2001), it is essential to address atrophy of the (Serotec, MCA04), and incubated at 37°C for distal stump population in order to optimise 30 minutes, plus a further 15 minutes after the regeneration after secondary nerve repair. addition of 250µl of rabbit anti mouse Using quantitative immunohistochemistry this complement (Cedarlane). The SC (around study therefore aimed to examine two potential 5 therapeutic strategies, both incorporated within 120x10 ) were then washed and grown on, bioartificial nerve conduits suitable for use in changing the medium every 48 hours and gap repair. The first involves augmenting the splitting the SC (1 in 3) on confluence (Brockes, endogenous population with cultured syngeneic et al., 1979). The purity of SC was 98% as Schwann cells (SC), whilst the second aims to assessed separately by S100 staining. boost the endogenous population by the At the second passage following administration of recombinant glial growth purification, the SC were transduced using the factor 2 (rhGGF-2). retroviral vector pMFG lacZ nls (Ferry, et al., 1991), which contains a 3.5 kb modified E. coli lacZ marker gene encoding the β-galactosidase Materials & Methods protein and a nuclear localisation sequence. The All work was conducted in keeping Moloney Murine Leukaemia Virus packaging with the terms of the Animals (Scientific cell line PT67 (Clontech, USA) was used for Procedures) Act 1986, and the experimental transduction. Its genome contains the three Hart, et al. ▪ GGF & Schwann cells in late secondary repair 3 retroviral structural genes, but lacks the Schwann cells (SC, 80x106/ml), or both packaging signal provided in trans as part of the (rhGGF-2 & SC) were added to this solution. MFG lacZ nls that contains the lacZ gene Conduits were immediately immersed in sterile cloned between the signal and two long 0.1M CaCl2 solution for 2 minutes to permit terminal repeats. In this way retrovirus calcium crosslinkage of the alginate, thereby encoding the lacZ gene can transduce SC, but creating a stable hydrogel (Mosahebi, et al., the transduced SC cannot produce any further 2002b) that served as a delivery vehicle, and retrovirus due to the absence of the retroviral matrix substitute. The nerve stumps were structural genes from their genome. A stable trimmed back to “healthy” nerve, and inset PT67 lacZ nls producing clone (titre 2mm into either end of the conduit using paired approximately 106/ml) was isolated, and a horizontal mattress sutures, thereby leaving a confluent layer was incubated with 10ml of 1cm gap across which the nerve had to growth medium (high glucose DMEM/ FCS) at regenerate. 32°C overnight to create a medium that Six weeks later the repairs were containing retroviral particles. This medium was then filtered (0.45µm filter, Nalgene) and added to a flask of 70% confluent SC, together with 8 µg/ml of the polyanion polybrene (Sigma). The transducing medium was removed after 4 hours incubation at 32°C, SC growth medium was added (Kotani, et al., 1994), and the flask kept at 37°C overnight. This cycle was repeated three times over 3 days, giving an 83% transduction rate as assessed by 5-bromo- 4-chloro-3-indoyl-β-D galactosidase (X-gal) staining to detect β-galactosidase expression. Transduced SC were then available for subsequent implantation at the time of nerve repair in experimental animals. Under inhalational anaesthesia (Halothane, May & Baker Ltd., U.K.), six groups of young adult male Sprague-Dawley rats (n=5) underwent unilateral sciatic nerve division at the upper border of quadratus femoris in the mid-thigh. The nerve stumps were ligated with 6/0 Prolene, and placed inside silicon rubber caps to prevent spontaneous regeneration. Either 2 or 4 months later, the Figure 1: Viable cultured Schwann cells within a animals were again anaesthetised, and 1cm gap polyhydroxybutyrate nerve conduit. Section through a poly- 3-hydroxybutyrate (PHB) nerve conduit containing cultured repairs performed using either freshly harvested syngeneic Schwann cells, six weeks after a secondary nerve reversed, 1cm sciatic nerve isografts, or repair (1cm gap). bioartificial nerve conduits. The conduits were A. When stained by fluorescence immunohistochemistry 1.4cm long tubes made from compressed sheets against S-100, viable Schwann cells were evident within the alginate-fibronectin hydrogel, as shown in this of longitudinally oriented poly-3- representative photomicrograph, although cells not in hydroxybutyrate (PHB) fibres, and were filled contact with nerve fibres did not display the normal spindle- with ~50µl of a solution containing 2% shaped morphology (scale bar 20µm). ultrapure, low molecular weight, high mannuronic acid content sodium alginate (Pronova, Sweden) and 0.05% bovine fibronectin (Sigma Pharmaceuticals, U.K.). In the treatment groups either recombinant glial growth factor (rhGGF-2 500ng/ml, Cambridge Neuroscience, U.K.), cultured autologous Hart, et al. ▪ GGF & Schwann cells in late secondary repair 4

harvested in continuity with ~3mm of both the therefore appeared as red fluorescence, and S- proximal and distal nerve, fixed in Zamboni’s 100 appeared as green. In order to confirm the solution at 4oC, and equilibrated in presence of implanted cultured Schwann cells, cryoprotectant solution (15% sucrose in separate sections were stained for the presence phosphate buffered saline). Blocks were then of the lacZ gene product X-gal (Mosahebi, et prepared by rapid freezing into OCT™ al., 2001). mounting medium, from which a systematic Every section from each repair was random sample of longitudinal, 15µm examined under a 10x objective lens with a cryosections was collected for dual stain 0.1mm graticulated eyepiece in order to fluorescence immunohistochemistry. Non- measure the “regeneration distance” from the specific hydrophobic attachment of primary end of the proximal nerve stump to the tip of antisera against pan-axonal marker of the most distal nerve fibre. Next, the two non- neurofilament (monoclonal antibody against consecutive sections that lay most central to the PAMNF, Affiniti Research Products Ltd.), and growth cone in each repair were selected, and a the Schwann cell marker S-100 (polyclonal band of images was captured across their full antibody, DAKO Ltd.) was minimised by co- width using a high resolution fluid cooled incubation with normal rat serum (DAKO, digital camera (Diagnostic Instruments, U.S.A.) Denmark). After pre-incubation of tissue with a 20X objective lens. The area of sections with normal goat (Sigma-Aldrich U.K.) immunostaining within this band of images was and normal horse (Sigma-Aldrich U.K.) sera, determined by an image analysis process bound primary antibody was detected by (Image-Pro-Plus™ Version 4.0 software) reliant incubation with secondary horse anti-mouse upon intensity thresholding for unichromatic antiserum conjugated to Cy-3 (Amersham light, deselection of occasional areas of Pharmacia Biotech), and goat anti-rabbit background staining, and digital quantification antiserum conjugated to fluorescein of the area stained. The “total area of isothyocyanate (FITC, Vector Laboratories immunostaining” (quoted in µm2) was noted for Inc.). PAMNF staining of nerve fibres each section, as was the total area of the nerve

Regeneration Distance (mm) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Plain Conduit 6.0mm(3.0)

4.9mm(2.6) 13.3mm Nerve Graft (1.5) ** 12.4mm (2.4) **

SC Conduit 10.8mm (0.27) * 12.1mm(0.55) **

rhGGF-2 11.3mm (1.3) * Conduit 11.4mm(0.81) ** 14.0mm( rhGGF / SC 0.70) ** Conduit

Figure 2: Nerve Regeneration Distance 6 Weeks Following Sciatic Nerve Repair. Either 2 months (solid arrows), or 4 months (dashed arrows) after nerve transection sciatic nerve repairs (1cm gap) were performed using either reversed syngeneic nerve isograft (“Nerve Graft”), or polyhydroxybutyrate tubes containing calcium alginate / fibronectin gel (“Plain Conduit”) with the addition of 500ng/ml rhGGF-2 (“rhGGF-2 Conduit”), cultured syngeneic Schwann cells (“SC Conduit”), or both (“rhGGF-2/SC Conduit”). Regeneration distance was quantified 6 weeks after repair, by measuring the distance (mm) from the end of the proximal nerve stump, to the tip of the most distal nerve fibre, using longitudinal sections stained by fluorescence immunohistochemistry with primary antiserum to pan-axonal marker of neurofilament (PAMNF). Figures given are the mean (SD) for each experimental group (n=5). *P<0.05, **P≤0.001 (compared to repair using plain conduits). Hart, et al. ▪ GGF & Schwann cells in late secondary repair 5

repair present in the captured images, in order conduits (Figure 1). The conduits that to express immunostaining as a percentage of contained rhGGF-2 and SC had the most the area of the nerve repair examined. This profuse outgrowth from the proximal stump, “percentage area” facilitates comparison with numerous morphologically normal between conduits of slightly different regenerating fibres entering the distal nerve diameters, while total area allows better stump. comparison between conduits and nerve grafts When nerve repair was performed 2 due to their markedly different diameters and months after axotomy, fibres had regenerated a distribution of fibres. A total of ten mean of 6mm through the plain PHB conduits measurements of the area of immunostaining by the time of harvest (Figure 2). The addition (both absolute and percentage) was therefore of SC caused an 80% increase in regeneration recorded for each antiserum in each distance (P=0.007), such that fibres reached experimental group, giving a mean and standard into the distal nerve stump. The effect of deviation for statistical examination of the rhGGF-2 was slightly more pronounced (87% effects of treatment. increase; P=0.007), and although a longer Statistical analysis of results was regeneration distance was found with nerve performed by paired t-test, using SigmaStat™ graft, it was not significantly longer than that (Version 2.0) software, after confirming the found in rhGGF-2 conduits (P>0.05). By normality of data distribution by the comparison, in the nerves repaired 4 months Kolmogorov-Smirnov test. after axotomy, regeneration was impaired. Fibres reached a mean of 4.9mm into the plain Results PHB conduits. The addition of rhGGF-2 led to a 135% increase in regeneration distance At the time of harvest there was good (P<0.001), and SC were slightly more effective neovascularisation of the conduits, with (150% increase; P<0.001). Regeneration had minimal fibrosis. Nerve grafts were equally proceeded marginally further within nerve graft healthy in appearance, but somewhat more repairs, but remained comparable to that in adherent to the muscle bed. Morphological either the rhGGF-2 (P=0.41) or SC conduits examination of the nerve grafts revealed many (P=0.81). The longest regeneration distance fine neurites showing poorly directed growth at occurred through the conduits that contained the proximal anastomosis, and multiple both rhGGF-2 and SC. regenerating axons passing down the graft to A similar pattern was found when the reach the distal stumps. Growth across the distal area of immunostaining for PAMNF was anastomosis appeared to be somewhat impaired, examined (Table 1). Of the groups repaired 2 particularly after the 4 month delay, although months after axotomy, the plain conduits fibres were clearly present in the distal stump. contained the least staining. The addition of SC Plain PHB conduits contained outgrowth that led to a 35% increase, whilst rhGGF-2 was was predominantly of fine, poorly directed much more effective, causing a 422% neurites, with comparatively few regenerating increase(P=0.007) to a total area that was over axons. Nerve fibre morphology within the twice that present within nerve graft repairs conduits that contained rhGGF-2 was much (P=0.056). Plain conduits supported even improved, with a wide regeneration front smaller total areas of staining when repair was containing a convincing, directed cone of delayed until 4 months after axotomy, but the axonal growth between residual masses of 100% increase due to the addition of SC was alginate hydrogel, although numerous fine relatively greater than in the 2 month group, neurites were also present. There appeared to despite the absolute area being somewhat lower. be no block to entry into the distal nerve stump, Conduits containing rhGGF-2 exhibited a still although fewer fibres were present in the distal greater total area of staining, although the 167% stump than in the nerve graft repairs. Conduits increase was less than had been attributable to containing SC had fewer fibres, but these rhGGF-2 in the groups repaired 2 months after exhibited healthy axonal morphology, with axotomy. Nerve grafts contained a slightly directed growth into the distal stump. Viable greater total area of PAMNF staining, but this lacZ positive SC were evident within the was still less than that present in the rhGGF-2 & alginate hydrogel throughout the length of the Hart, et al. ▪ GGF & Schwann cells in late secondary repair 6

SC conduits, which contained an area that was S-100 (Table 2) reveals that when nerve repair 27% higher than in the nerve graft repairs. was performed 2 months after axotomy the Plain conduits contained the lowest percentage plain conduits contained the least staining for area of immunostaining for PAMNF. SC were Schwann cells. The addition of SC increased relatively more effective at increasing the the area of staining (45% greater total area, 14% percentage area of staining when repair was greater percentage area), but by much less than performed four months after axotomy (147% rhGGF-2 did (594% greater total area, P=0.002; increase), rather than two (12% increase). 497% greater percentage area, P=0.004). Although rhGGF-2 caused greater increases Conduits containing rhGGF-2 exhibited a 75% than SC alone, it was relatively less effective greater total area of staining for S-100 when repair was delayed by four months (230% (P=0.057) than nerve graft repairs. In all increase; P=0.035) than by two (370% increase; groups the total area of S-100 immunostaining P=0.01). Of all the conduits, those containing was less when the delay between axotomy and rhGGF-2 & SC exhibited the highest percentage repair was four months rather than two, area of immunostaining for PAMNF; twice that although the change within the conduits found in the SC conduits, but less than that containing SC was minimal. Indeed when the found in the nerve graft repairs in keeping with percentage area of immunostaining is their smaller cross-sectional area. considered (Table 2), it is apparent that the Quantification of immunostaining for reduction was greatest for the conduits containing rhGGF-2, and that in those Repair After 2 Month Repair After 4 Month Delay Delay containing SC the percentage area was actually Mean (SD) Mean (SD) not reduced. It is also apparent that conduits Type of Total Total % Area Repair Area of Area of Area of containing rhGGF-2 contained a greater total of PAMNF PAMNF PAMNF PAMNF area of staining for S-100 than nerve graft Staining Staining Staining Staining (µm2) (µm2) (P=0.021), as did those containing rhGGF-2 & Plain 25471 0.93 16488 0.52 SC (P=0.012), which exhibited the highest total Conduit (22634) (0.81) (12957) (0.36) area of S-100 staining of all the groups.

Nerve 58259 7.3 ** 56890 * 7.1 ** Graft (25405) (2.3) (24546) (1.9) Discussion SC 34271 1.0 33022 1.3 Conduit (19291) (0.44) (12220) (0.73) The current clinical management of peripheral nerve trauma yields less than optimal rhGGF-2 132887* 4.36* 44033* 1.7* conduit (62240) (2.1) (26600) (0.99) results, despite the increasing sophistication of rhGGF-2 surgical techniques, and post-operative 72048* 2.6* / SC (29717) (0.98) regimens (Lundborg, 2000a; Lundborg, 2000b). conduit Future improvements are therefore most likely Table 1: Area of immunostaining against PAMNF within to arise from the prevention of neuronal death sciatic nerve repairsEither 2 or 4 months after nerve (Hart, et al., 2002d), and manipulation of the transection sciatic nerve, 1cm gap repairs were performed neurobiological events integral to peripheral using either reversed syngeneic nerve isograft (“Nerve Graft”), or polyhydroxybutyrate tubes containing calcium nerve regeneration (Hall, 2001; Lundborg, alginate / fibronectin gel (“Plain Conduit”); this gel 2000a). Yet, to date, as the neurobiology has contained either 500ng/ml rhGGF-2 (“rhGGF-2 Conduit”), been elucidated there has been a failure to cultured syngeneic Schwann cells (“SC Conduit”), or both (“rhGGF-2/SC Conduit”) in the treatment groups. Repairs translate increased understanding into improved were harvested six weeks later, and longitudinal outcome (Lundborg, 2000a). In particular the cryosections stained by fluorescence immunohistochemistry administration of exogenous neurotrophic with primary antiserum to pan-axonal marker of factors (NTF’s), although experimentally neurofilament (PAMNF). The area of immunostaining was quantified using image analysis of a band of digital images promising (Terenghi, 1999), has failed to captured across the full width of the repair at a position progress through clinical trials (Lundborg, 1mm beyond the end of the proximal nerve stump. The area 2000a). It may therefore be necessary to step of immunostaining is expressed either as a mean value per repair (µm2), or as a percentage of the total area of further away from the neurons themselves, in graft/conduit present within the band of images. Figures our attempt to find clinically appropriate ways given are the mean (SD) for each experimental group (n=5). to enhance regeneration. *P<0.05, **P≤0.001 (compared to repair using plain Schwann cells are essential for peripheral nerve conduits). regeneration (Fu and Gordon, 1997; Hall, Hart, et al. ▪ GGF & Schwann cells in late secondary repair 7

1986; Hall, 1997) and in effect become more important in the case of secondary nerve indigenous neurotrophic factories after repair, since the initial period of Schwann cell peripheral nerve injury, producing neurotrophic, proliferation that accompanies nerve injury and tropic agents such as CNTF, LIF, NGF, (Terenghi, 1999) is followed by entry into a BDNF, NT-4, GAP-43, and GDNF (Fu and phase of profound atrophy and cell death Gordon, 1997; Hall, 1997; Hall, 2001; (Ekstrom, 1995; Mirsky and Jessen, 1999; Hammarberg, et al., 1996; Reynolds and Woolf, Miyamoto, et al., 1986; Terenghi, et al., 1998). 1993; Terenghi, 1999). By enhancing this Between 2 and 6 months after axotomy function it may be possible to indirectly Schwann cells downregulate expression of promote axonal regeneration in a more GDNF mRNA (Höke, et al., 2002), and the physiological manner than by directly neuregulin receptor moeities c-erbB2 &B4 administering individual NTF’s, while at the (Hall, 2001; Li, et al., 1997), and upregulate same time enhancing the distal nerve stump’s expression of the inhibitory glycoprotein MAG ability to receive and promote axonal (Einheber, et al., 1993), making them less able regeneration. Key to the normal, mutually to support neurite ingrowth after repair. beneficial, axon-Schwann cell interaction Numbers of Schwann cells decrease during regeneration is the action of glial growth significantly within 10 weeks, and they are factors (GGF) released by regenerating entirely absent from many endoneurial tubes by axons(Fu and Gordon, 1997; Li, et al., 1997; 30 weeks (Roytta and Salonen, 1988; Terenghi, Reynolds and Woolf, 1993). Glial growth et al., 1998). Within 28-35 days of nerve factors are a group of differentially spliced transection Schwann cell atrophy leads to transcripts of a single neuregulin gene endoneurial fibrosis (Green, 1998), reducing the (Marchionni, et al., 1993) that act through the diameter of endoneurial tubes to 26% of normal c-erbB2, 3 & 4 receptor moeities; GGF-2 is the after one year (Miyamoto, et al., 1986), and to most characterised, and stimulates Schwann cell only 1% after two years (Green, 1998). Thus by migration, proliferation, and NTF secretion as addressing the atrophic Schwann cell local concentrations increase (Li, et al., 1997; population it may be possible not only to Mahanthappa, et al., 1996). indirectly boost the regenerative capacity of the Enhancing the function of the neurons, but also to enhance their progress by endogenous Schwann cell population is even restoring the normal Schwann cell axonal

Repair After 2 Month Table 2: Area of immunostaining against S-100 Repair After 4 Month Delay Delay Mean (SD) within sciatic nerve repairs. Either 2 or 4 months Mean (SD) after nerve transection sciatic nerve, 1cm gap repairs Percentage Area of S- were performed using either reversed syngeneic 100 Staining Type of Total Total nerve isograft (“Nerve Graft”), or Repair Area of % Area Area of Change polyhydroxybutyrate tubes containing calcium S-100 of S-100 S-100 cf. alginate / fibronectin gel (“Plain Conduit”); this gel repair Staining Staining Staining % Area contained either 500ng/ml rhGGF2 (“rhGGF (µm2) (µm2) after 2 month Conduit”), cultured syngeneic Schwann cells (“SC delay Conduit”), or both (“rhGGF/SC Conduit”) in the treatment groups. Repairs were harvested six weeks Plain 28290 1.1 23593 0.75 32% later, and longitudinal cryosections stained by Conduit (26305) (1.1) (14755) (0.39) decrease fluorescence immunohistochemistry with primary antiserum to the Schwann cell marker S-100. The 112295 Nerve 14** 72955 * 9.7 ** 31% area of immunostaining was quantified using image * Graft (2.1) (21562) (3.7) decrease (29902) analysis of a band of digital images captured across the full width of the repair at a position 1mm beyond SC 40886 1.2 39091* 1.5* 21% the end of the proximal nerve stump. The area of Conduit (22106) (0.40) (20214) (1.1) increase immunostaining is expressed either as a mean value per repair (µm2), or as a percentage of the total area of graft/conduit present within the band of images. rhGGF-2 196303* 6.5* 66622* 2.6* 60% Figures given are the mean (SD) for each conduit (78908) (2.8) (30050) (1.1) decrease experimental group (n=5). The right-handmost column demonstrates change in the percentage area rhGGF-2 107093* 2.3* / SC of staining for S-100 for each type of repair when (55523) (1.1) conduit performed 4 months after injury, as compared to 2 months after injury, and reveals a reduction the staining for S-100 with all types of repair except SC conduits. *P<0.05, **P≤0.001 (compared to repair using plain conduits). Hart, et al. ▪ GGF & Schwann cells in late secondary repair 8

interaction, and by limiting further endoneurial denervation atrophy of mechanoreceptors fibrosis. (Kopp, et al., 1997). Local delivery is Two therapeutic approaches to this preferable to systemic, since it reduces the problem were examined. Firstly, to augment doses involved and produces a more targeted the endogenous population with cultured action. Since alginate gel permits sustained autologous Schwann cells, and secondly to release of growth factors (Austin, et al., 1997) attempt to boost the endogenous Schwann cell similar PHB conduits to those employed for SC population by local administration of delivery could be used. A dose of 500ng/ml recombinant glial growth factor 2 (rhGGF-2). was used since this optimises NTF secretion by In both cases local delivery was achieved using Schwann cells and their stimulation of neurite a bioartifical nerve conduit, since in clinical extension in vitro, and is supramaximal for practice these injuries generally require a gap stimulating migration and proliferation repair. There is also some evidence that even (Mahanthappa, et al., 1996), even in where direct repair is feasible, a short gap repair chronically denervated SC (Li, et al., 1998). may enhance outcome by facilitating selection The purpose of this study was not to of appropriate endoneurial tubes by investigate the maximum distance across which regenerating neurites (Lundborg, 2000a), fibres could be made to regenerate, but rather to although this remains controversial (Hall, determine the effects of rhGGF-2 and SC upon 2001). regeneration across a known, bridgeable gap, In primary nerve repair empty poly-3- with the extra biological hurdle of a clinically hydroxybutyrate (PHB) conduits support relevant, and biologically significant delay in regeneration across 1-4cm gaps (Hazari, et al., repair. In the rat, it is evident that regenerative 1999; Young, et al., 2002), and when filled with capacity diminishes with delays of greater than an alginate-fibronectin hydrogel constitute an around 1-2 months, and the Schwann cell and effective delivery system for cultured Schwann endoneurial changes described above are well cells (Mosahebi, et al., 2001; Mosahebi, et al., established by 4 months (Fu and Gordon, 1997; 2002b). Cultured Schwann cells maintain their Gattuso, et al., 1989; Li, et al., 1997; Olawale function after retroviral labelling with the lacZ and Gordon, 2000). Hence 1cm gap repairs gene (Mosahebi, et al., 2001), and both were performed at these timepoints after injury. allogeneic (Mosahebi, et al., 2002a) and Nerve isograft repairs were included in the syngeneic (Bryan, et al., 1996; Mosahebi, et al., study to represent the current standard of 2001; Zhang, et al., 2002) cells enhance clinical practice. regeneration after primary gap repair, where an It is clear that when compared to optimal concentration of 80x106/ml has been nerve graft, the plain conduits were unable to established (Mosahebi, et al., 2001). The one support adequate nerve regeneration (either study that has failed to show any enhancement quantitatively or qualitatively) when repair was of regeneration (Evans, et al., 2002), used a low delayed until 2 or 4 months after injury. This is concentration of SC (1x104-6/ml) in a collagen not unexpected, since unlike nerve grafts the gel that may have impeded axonal regeneration plain conduits are acellular, and the ability of and SC migration. The clinical use of cultured the endogenous Schwann cells to proliferate SC is even more attractive in secondary repair and migrate is impaired. However some given the potential for atrophy of the regeneration did occur, and it has been a endogenous population, and since one requires common finding in the primary repair literature 4-6 weeks to raise sufficient numbers of that it begins more slowly in synthetic conduits autologous cells in culture to prepare an than in nerve grafts. It is also important to note autologous conduit. There would also be the that regeneration (morphology, distance, and possibility for genetic manipulation to enhance area) was clearly worse when the repair was their function (Jones, et al., 2001; Joung, et al., delayed by 4 months rather than 2. This effect 2000). is most noticeable when considering the area of The administration of exogenous S-100 staining, since in both types of repair the GGF also improves nerve regeneration after percentage area fell by one third, further primary gap repair (Bryan, et al., 1996), or highlighting the need to therapeutically address crush injury (Chen, et al., 1998), and reduces Schwann cell atrophy. Hart, et al. ▪ GGF & Schwann cells in late secondary repair 9

Quantification of Jessen, 1999). A direct interaction seems most immunohistochemical staining for nerve fibres likely in the 4 month delayed repairs. gives a reflection of the amount of axonal When compared to plain conduits, regeneration which occurs, whilst the rhGGF-2 caused a much greater increase in S- regeneration distance at a given timepoint after 100 staining than cultured SC did, particularly repair better reflects the rate at which the when repair was delayed by 2 months. RhGGF- leading fibres regenerate. The percentage area 2 conduits exhibited a 75% greater total area of of immunostaining for nerve fibres (PAMNF) staining than the nerve grafts at this time. When was relatively unaffected by the addition of SC repair was delayed by four months rather than to the conduits used for repair after a 2 month two, rhGGF-2 became significantly less delay, but when the delay was 4 months the effective at increasing the area of S-100 figure was doubled by SC. Regeneration staining, such that nerve grafts then contained distance was significantly increased by SC, more S-100 staining. This reduction in the particularly when repair was delayed by 4 proliferative response to GGF is in keeping with months where the regeneration distance became the reduced expression of GGF receptor comparable to that through nerve grafts. This moieties as the period of denervation progresses ability of SC to enhance regeneration distance beyond 2 months (Li, et al., 1997; Li, et al., more than axonal area has been noted in studies 1998). There would also be fewer endogenous of primary repair (Bryan, et al., 2000; cells present to proliferate, since Schwann cell Mosahebi, et al., 2002b). In keeping with the death (Miyamoto, et al., 1986; Roytta and morphological findings of this study, it has been Salonen, 1988) will have been ongoing for a postulated that this reflects stimulation of further 2 months. As would be expected from directed axonal growth rather than the its effect upon Schwann cells, rhGGF-2 speculative extension of multiple transient significantly enhanced all measures of nerve neurites. regeneration. That it still had a significant Interestingly, when repair was effect when repair was delayed by 4 months performed after a 2 month delay the percentage may suggest that either GGF-receptor area of staining for Schwann cells (S-100) in expression had not been completely lost, or that the SC conduits was not dramatically greater the in-vitro findings of limited GGF than in the plain conduits. This may imply responsiveness in chronically denervated, c- either that those SC not in contact with axons erbB2-4 deficient SC (Li, et al., 1998) are had died, or that staining was being measured applicable in vivo. When repaired after a 2 within an area where the effect of Schwann cell month delay the regeneration distance through proliferation in response to axonal ingrowth rhGGF-2 conduits was comparable to nerve predominated. That numerous viable SC were graft, but the total area of immunostaining for present more distally within the alginate of only nerve fibres was 128% greater. Percentage area those conduits containing cultured SC (Figure of PAMNF staining was 370% greater than in 1) suggests the latter, as would the reported the plain conduits, and the morphological viability of cultured SC for at least 6 weeks findings were greatly improved. When repair within similar conduits (Mosahebi, et al., was delayed by 4 months the effect was 2002b). Furthermore, in the repairs performed quantitatively less, although the morphological after a 4 month delay, when the endogenous improvements persisted, the total area of population would be more atrophied and less staining in rhGGF-2 conduits no longer able to proliferate in response to neurite exceeded that in nerve grafts, but the ingrowth, S-100 staining was significantly regeneration distance remained comparable. greater in the conduits containing SC. Since These dramatic effects upon nerve regeneration viable SC were present in the conduits at are most likely due to stimulation of harvest, the improved regeneration may reflect proliferation, migration, and neurotrophic factor either direct interaction with regenerating secretion by Schwann cells, thereby indirectly axons, or an effect to boost endogenous stimulating axonal growth. Schwann cell function through the paracrine Although rhGGF-2 still caused a mechanisms that exist (Hall, 2001; Mirsky and dramatic improvement in regeneration when repair was delayed by 4 months, its efficacy Hart, et al. ▪ GGF & Schwann cells in late secondary repair 10 was somewhat reduced as compared the regenerating axons to stimulate them to situation after a 2 month delay when rhGGF-2 proliferate and secrete NTF’s that in turn conduits were comparable to nerve graft. This promote regeneration. By combining rhGGF-2 is most likely due to the fact that as the period and SC in the same conduit, not only is the between injury and repair increases, the survival of the cultured SC likely to be responsiveness of endogenous Schwann cells enhanced, but their pro-regenerative function is decreases, as the results for S-100 staining boosted. This is in addition to the recruitment show. The most obvious way around this of endogenous Schwann cells, which might problem is to combine rhGGF-2 with a otherwise play a lesser role due to consumption population of fresh, GGF-responsive cultured of endogenous GGF by the transplanted SC. SC, and this was done in the final experimental Although studies quantifying target group. organ reinnervation after longer regeneration Only one published study has times, and others investigating longer gap described the combined use of cultured repairs will be necessary, it is clear that both autologous SC and rhGGF, and it failed to show cultured syngeneic Schwann cells and benefit by combining them within the same exogenous rhGGF-2 can have significant and conduit after primary repair (Bryan, et al., beneficial effects upon regeneration after 2000) unlike our demonstration of an enhanced secondary nerve repair. This study also regenerative profile, even compared to nerve demonstrates that PHB conduits filled with graft. However the concentration of SC was calcium alginate-Fn hydrogel represent a sub-optimal (~1x106/ml), and SC merely lined a technically simple, yet effective delivery smooth walled conduit containing no matrix system, although it may be possible to further substitute, thereby failing to provide sufficient enhance regeneration by altering the physical or mechanical cues for neurite elongation and chemical structure of the gel. Schwann cell migration. In this study PHB SC appear to have more effect upon fibres, and a fibronectin rich matrix provided regeneration distance than area of staining for guidance, and conduits containing rhGGF-2 & nerve fibres, and become relatively more SC not only supported significantly better effective as the delay between injury and repair regeneration than plain conduits, but were an increases. In contrast, rhGGF-2 causes greater improvement over conduits containing either absolute increases in the area of staining, and SC, or rhGGF-2 alone. Percentage area of may therefore obviate the need for cultured immunostaining for PAMNF was twice that SC’s in short gap, early repairs, but its efficacy found with SC alone, and 1.5 times that found reduces as the delay between injury and repair with rhGGF-2, while the regeneration distance increases. Conduits incorporating rhGGF-2 and morphology was also enhanced, with were comparable to nerve grafts when repair multiple fibres reaching the cut ends of the was performed after a 2 month delay, whilst harvested distal stumps in all repairs. Results conduits containing a combination of rhGGF-2 were actually slightly better than were obtained & SC exhibited a better regenerative profile with nerve graft, since the total area of than nerve graft when repair was performed 4 immunostaining for PAMNF was 27% greater, months after axotomy. The potential for a and for S-100 was 47% greater. bioartificial nerve conduit to replace autologous The improvement in regeneration nerve grafts for late secondary nerve repair has over the use of rhGGF-2 alone is to be been demonstrated, although future expected, since a population of fully GGF modifications may further enhance the responsive Schwann cells is present in the regenerative profile. rhGGF-2 & SC conduits, unlike the absolute reliance upon the atrophic endogenous population when using rhGGF-2 in isolation. Acknowledgements The improvement over the use of cultured SC This work was funded by the Swedish Medical alone is also to be expected, since in culture SC Research Council (Grant 02286), County of remain quiescent, or apoptose, in the absence of Vasterbottens, Ane’rs Foundation, and the East GGF stimulation. When implanted in-vivo they Grinstead Medical Research Trust (Registered predominantly rely upon GGF released by Charity No. 258154). The authors would like to Hart, et al. ▪ GGF & Schwann cells in late secondary repair 11 thank the Stanley-Thomas Johnson Foundation, study). Neurochem Int 33 (4): 341- the Smith Charity, and the Geoffrey Burton 351. Trust for their generous financial support, and de Medinaceli L, Prayon M, Merle M . (1993). AstraTec AB for the unqualified donation of Percentage of nerve injuries in which supplies of poly-3-hydroxybutyrate sheets. The primary repair can be achieved by photomicrograph shown in Figure 1 was end-to-end approximation: review of prepared by Mr. Mel Tohill, Surgical Research 2,181 nerve lesions. Microsurgery 14 Fellow at the Blond-McIndoe Research (4): 244-246. Laboratories. Einheber S, Milner TA, Giancotti F, Salzer JL. (1993). Axonal regulation of Schwann cell integrin expression References suggests a role for α6β4 in Allan CH. (2000). Functional results of primary myelination. J Cell Biol 123 (5): nerve repair. Hand Clin 16 (1): 67- 1223-1236. 72. Ekstrom PA. (1995). Neurones and glial cells of Austin L, Bower JJ, Kurek JB, Muldoon CM. the mouse sciatic nerve undergo (1997). Controlled release of apoptosis after injury in vivo and in leukaemia inhibitory factor (LIF) to vitro. Neuroreport 6 (7): 1029-1032. tissues. Growth Factors 15 (1): 61- Evans GRD, Brandt K, Katz S, Chauvin P, Otto 68. L, Bogle M, Wang B, Meszlenyi RK, Birch R, Bonney G, Wynn Parry CB. (1998). Lu LL, Mikos AG, Patrick Jr. CW. Surgical Disorders of the (2002). Bioactive pol(L-lactic acid) Peripheral Nerves. Churchill- conduits seeded with Schwann cells Livingstone, Edinburgh. for peripheral nerve regeneration. Brockes JP, Fields KL, Raff MC. (1979). Biomaterials 23: 841-848. Studies on cultured rat Schwann cells. Ferry N, Duplessis O, Houssin D, Danos O, I. Establishment of purified Heard JM. (1991). Retroviral- populations from cultures of mediated gene transfer into peripheral nerve. Brain Res 165: 105- hepatocytes in vivo. Proc Natl Acad 118. Sci U.S.A. 88: 8377-8381. Bruyns CN, Jaquet JB, Schreuders TA, Kalmijn Fu SY, Gordon T. (1995). Contributing factors S, Kuypers PD, Hovius SE. (2003). to poor functional recovery after Predictors for return to work in delayed nerve repair: prolonged patients with median and ulnar nerve denervation. J Neurosci 15 (5): 3886- injuries. J Hand Surg 28a (1): 28-34. 3895. Bryan DJ, Holway AH, Wang KK, Silva AE, Fu SY, Gordon T. (1997). The cellular and Trantolo DJ, Wise D, Summerhayes molecular basis of peripheral nerve IC. (2000). Influence of glial growth regeneration. Mol Neurobiol 14 (1- factor and Schwann cells in a 2): 67-116. bioresorbable guidance channel on Gattuso JM, Glasby MA, Gschmeissner SE, peripheral nerve regeneration. Tissue Norris RW. (1989). A comparison of Eng 6 (2): 129-138. immediate and delayed repair of Bryan DJ, Wang KK, Chakalis-Haley DP. peripheral nerves using freeze thawed (1996). Effect of Schwann cells in the autologous muscle grafts- in the rat. enhancement of peripheral-nerve BJPS 42: 306-313. regeneration. J Reconstr Microsurg Green B. (1998). Repair and grafting of the 12 (7): 439-6. peripheral nerve. In: Plastic Surgery. Chen LE, Liu K, Seaber AV, Katragadda S, Green (Ed), pp 630-695. Kirk C, Urbaniak JR. (1998). Hall SM. (1997). Axonal regeneration through Recombinant human glial growth acellular muscle grafts. J Anat 190: factor 2 (rhGGF2) improves 57-71. functional recovery of crushed peripheral nerve (a double-blind Hart, et al. ▪ GGF & Schwann cells in late secondary repair 12

Hall SM. (2001). Nerve repair: a cellular bridges and gene therapy for neurobiologists view. J Hand Surg spinal cord injury. J Physiol 533 (1): 26B (2): 129-136. 83-89. Hall SM. (1986). The effect of inhibiting Joung I, Kim HS, Hong JS, Kwon H, Kwon Schwann cell mitosis on the re- YH. (2000). Effective gene transfer innervation of acellular autografts in into regenerating sciatic nerves by the peripheral nervous system of the adenoviral vectors: potentials for gene mouse. Neuropathol Appl Neurobiol therapy of peripheral nerve injury. 12 (4): 401-414. Mol Cells 10 (5): 540-545. Hammarberg H, Piehl F, Cullheim S, Fjell J, Kopp DM, Trachtenberg JT, Thompson WJ. Hokfelt T, Fried K. (1996). GDNF (1997). Glial Growth Factor rescues mRNA in Schwann cells and DRG schwann cells of mechanoreceptors satellite cells after chronic sciatic from denervation-induced apoptosis. nerve injury. Neuroreport 7 (4): 857- J Neurosci 17 (17): 6697-6706. 860. Kotani H, Newton PB, Zhang S, Chiang YL, Hart AM, Brannstrom T, Wiberg M, Terenghi Otto E, Weaver L, Blaese RM, G. (2002a). Primary sensory neurons Anderson WF, McGarrity GJ. (1994). and satellite cells after peripheral Improved methods of retroviral vector axotomy in the adult rat: timecourse transduction and production for gene of cell death and elimination. Exp therapy. Hum Gene Ther 5: 19-28. Brain Res 142: 308-318. Li H, Terenghi G, Hall SM. (1997). Effects of Hart AM, Wiberg M, Terenghi G. (2002b). delayed re-innervation on the Exogenous leukaemia inhibitory expression of c-erbB receptors by factor enhances nerve regeneration chronically denervated rat Schwann after late secondary repair using a cells in vivo. Glia 20: 333-347. bioartificial nerve conduit. BJPS (in Li H, Wigley C, Hall SM. (1998). Chronically press). denervated rat Schwann cells respond Hart AM, Wiberg M, Terenghi G. (2002c). to GGF in vitro. Glia 24 (3): 290- Pharmacological enhancement of 303. peripheral nerve regeneration by Liss AG. (1995). Peripheral nerve injury and systemic acetyl-L-carnitine treatment. its effects on the central nervous Neurosci Lett 334: 181-185. system. A study of transection of Hart AM, Wiberg M, Youle M, Terenghi G. sensory nerves in cats and (2002d). Systemic acetyl-L-carnitine monkeys. Acta Universitatis eliminates sensory neuronal loss after Upsaliensis. peripheral axotomy: A new clinical Lundborg G. (2000a). A 25-year perspective of approach in the management of peripheral nerve surgery: evolving peripheral nerve trauma. Exp Brain neuroscientific concepts and clinical Res 145 (2): 182-189. significance. J Hand Surg. 25A: 391- Hazari A, Wiberg M, Johansson-Ruden G, 414. Green C, Terenghi G. (1999). A Lundborg G. (2000b). Brain plasticity and hand resorbable nerve conduit as an surgery: an overview. J Hand Surg. alternative to nerve autograft in nerve 25B (3): 242-252. gap repair. BJPS 52 (8): 653-657. Lundborg G, Dahlin L, Danielsen N, Zhao Q. Höke A, Gordon T, Zochodne DW, Sulaiman (1994). Trophism, tropism, and OAR. (2002). A decline in glial cell- specificity in nerve regeneration. J line-derived neurotrophic factor Reconstr Microsurg 10 (5): 345-354. expression is associated with Ma J, Novikov LN, Wiberg M, Kellerth JO. impaired regeneration after long-term (2001). Delayed loss of spinal Schwann cell denervation. Exp motoneurons after peripheral nerve Neurol 173 (1): 77-85. injury in adult rats: a quantitative Jones LL, Oudega M, Bunge MB, Tuszynski morphological study. Exp Brain Res MH. (2001). Neurotrophic factors, 139 (2): 216-223. Hart, et al. ▪ GGF & Schwann cells in late secondary repair 13

Mahanthappa NK, Anton ES, Matthew WD. Terenghi G. (1999). Peripheral nerve (1996). Glial growth factor 2, a regeneration and neurotrophic factors. soluble neuregulin, directly increases J Anat 194 (1): 1-14. Schwann cell motility and indirectly Terenghi G, Calder JS, Birch R, Hall SM. promotes neurite outgrowth. J (1998). A morphological study of Neurosci 16 (15): 4673-4683. Schwann cells and axonal Marchionni MA, Goodearl AD, Chen MS, regeneration in chronically transected Bermingham-McDonogh O, Kirk C, human peripheral nerves. J Hand Hendricks M, Danehy F, Misumi D, Surg 4 (23B): 1-5. Sudhalter J, Kobayashi K. (1993). Vanderhooft E. (2000). Functional outcomes of Glial growth factors are alternatively nerve grafts for the upper and lower spliced erbB2 ligands expressed in the extremities. Hand Clin 16 (1): 93- nervous system. Nature 362 (6418): 104. 312-318. Young RC, Wiberg M, Terenghi G. (2002). Mirsky R, Jessen KR. (1999). The neurobiology Poly-3-hydroxybutyrate (PHB): a of Schwann cells. Brain Pathol 9 (2): resorbable conduit for long gap repair 293-311. in peripheral nerves. BJPS 55 (3): Miyamoto Y, Higaki T, Sugita T, Ikuta Y, 235-240. Tsuge K. (1986). Morphological Zhang F, Blain B, Beck J, Zhang J, Chen Z, reaction of cellular elements and the CHen Z-W, Lineweaver WC. (2002). endoneurium following nerve section. Autogenous venous graft with one- Peripheral Nerve Repair & stage prepared Schwann cells as a Regeneration 3: 7-18. conduit for repair of long segmental Mosahebi A, Fuller P, Wiberg M, Terenghi G. defects. J Reconstr Microsurg 18 (4): (2002a). Effect of allogeneic 295-2002. Schwann cell transplantation on peripheral nerve regeneration. Exp Neurol 173: 213-223. The Authors Mosahebi A, Wiberg M, Terenghi G. (2002b). Andrew McKay Hart1,2,3 BSc MD (Hons), Addition of fibronectin to alginate MRCS AFRCS Surgical Research Fellow, The matrix improves peripheral nerve Blond-McIndoe Centre & The Department of regeneration in tissue engineered Surgical & Perioperative Science, Umeå conduits. Tissue Eng (in process University. citation). Mosahebi A, Woodward B, Wiberg M, Martin Afshin Mosahebi 1,2,3 PhD FRCS Surgical R, Terenghi G. (2001). Retroviral Research Fellow, The Blond-McIndoe Centre & labeling of Schwann cells: in vitro The Department of Surgical & Perioperative characterisation and in vivo Science, Umeå University. transplantation to improve peripheral

nerve regeneration. Glia 34: 8-17. 2,3 Olawale ARS, Gordon T. (2000). Effects of Prof. Mikael Wiberg PhD MD, Professor, short- and long-term Schwann cell Depts of Anatomy and Hand & Plastic Surgery, denervation on peripheral nerve Umeå University regeneration, myelination, and size. 1 Glia 32: 234-246. Prof. Giorgio Terenghi PhD FRCPath, Reynolds M.L., Woolf CJ. (1993). Reciprocal Director & Head of Nerve Regeneration Schwann cell-axon interactions. Curr Research, The Blond-McIndoe Centre. Opin Neurobiol 3: 683-693. Roytta M, Salonen V. (1988). Long-term Correspondence to Giorgio Terenghi, PhD, endoneurial changes after nerve FRCPath transection. Acta Neuropathol 76: 35-45. Paper submitted 5 February 2003