Retrograde Reactions ofClarke’s Nucleus Neuronsafter Human SpinalCord InjuryAndreas B. Schmitt, MD,1 Sebastian Breuer, PhD,1

Leyla Polat,1 Katrin Pech,1 Byran Kakulas, MD,2

Seth Love, MD,3 Didier Martin, MD,4–6

Jean Schoenen, MD,5,6 Johannes Noth, MD,1

and Gary A. Brook, PhD1

Successful axon regeneration depends on the expressionof regeneration-associated genes by axotomized neurons.Here, we demonstrate, for the first time to our knowl-edge, the expression of regeneration-associated genes byaxotomized human CNS neurons. In situ hybridizationand immunohistochemistry showed a transient inductionof GAP-43 and c-jun in Clarke’s nucleus neurons caudalto traumatic human spinal cord injury. These resultssupport experimental data that nonregenerating centralnervous system neurons can temporarily upregulateregeneration-associated genes, reflecting a transient re-generative capacity that fails over time.

Ann Neurol 2003;54:534–539

Damaged peripheral nervous system (PNS) axons canundergo regeneration leading to the restoration of ap-propriate synaptic contact and functional recovery. Incontrast, damaged central nervous system (CNS) axonsdo not regenerate but undergo abortive sprouting.1

The failure of any significant axonal regenerationwithin the lesioned CNS is caused by several factors,many of which are not fully understood.2,3 An impor-tant difference between the regenerative capacity ofPNS and CNS neurons is the molecular program ini-tiated by the axotomized populations of neurons. It hasbeen postulated that the failure of axon regeneration in

the CNS is correlated with a failure to upregulateGAP-43.4 GAP-43 is a protein that is concentrated inthe plasma membranes of growth cones and is widelybelieved to be important for axon growth during de-velopment, plasticity, and regeneration.4 The transcrip-tion factor c-jun is responsible for the rapid activationof several genes and has been described as a bipotentialmediator of either neuronal death/survival and regen-eration.5 We have reported previously that c-jun andits possible target, GAP-43, are transiently upregulatedin a large subpopulation of Clarke’s nucleus neuronsafter experimental spinal cord injury of the rat. Many,large diameter projection neurons are known to degen-erate several weeks after axotomy.6 This cell body re-sponse has been interpreted as an initial regenerativeattempt that fails over time. The factors responsible forthe failure to maintain regeneration-associated gene(RAG) expression and the eventual cessation of axonalgrowth are only partially understood but include thepresence of a glial environment that is hostile to axonregeneration.1

Data concerning the differential expression of im-portant RAGs in lesioned human CNS or PNS neu-rons are, at present, not available. Using in situ hybrid-ization and immunohistochemical techniques, we haveinvestigated the expression of GAP-43 and c-jun withinlesioned Clarke’s nucleus neurons in postmortem ma-terial obtained from patients who died after traumaticspinal cord injury. Such studies are important for as-sessing the clinical relevance of much of the experimen-tal data available concerning the molecular control ofaxon regeneration.

Materials and MethodsTissuesThe investigation was conducted on human postmortem ma-terial of seven patients who died 6 hours, 4 months, 8 days,11 days, 13 days, and 14 days, and 1 year after severe trau-matic spinal cord injury. All patients demonstrated anatom-ically complete injuries at either cervical or thoracic levelswith clinical paraplegia or tetraplegia. Postmortem tissues oftwo patients without any neurological disorder have beenused as controls. The analyses conducted on all postmortemsamples were approved by the local ethics committee. Spinalcords were fixed using 10% buffered formaldehyde. Fromeach case, several blocks of spinal cord were processed forembedding in paraffin wax.

In Situ HybridizationThe c-jun RNA probe was generated from human tonsil tis-sue. After RNA isolation and cDNA synthesis, polymerasechain reaction amplification of c-jun cDNA was performedusing the following primers: 5�-cccgaaacttgtgcgcgc-3� and 5�-cccctcctgctcatctgt-3� (accession no. J04111). A 400–base pairfragment of c-jun cDNA and a 700–base pair fragment ofGAP-43 cDNA fragment (GAP-43 plasmid was a gift fromL. Shrama) then were subcloned into a transcription vector

From 1Aachen Spinal Cord Research Center, Department of Neu-rology, University Hospital Aachen, Aachen, Germany; 2Depart-ment of Neuropathology, Royal Hospital, Perth, Australia; 3Depart-ment of Neuropathology, Frenchay Hospital, Bristol, UnitedKingdom; 4Department of Neurosurgery, Sart Tilman Hospital;5Department of Neurology and Neuropathology; and 6Center forCellular and Molecular Neuroscience, University of Liege, Liege,Belgium.

Received Jan 8, 2002, and in revised form Apr 7, 2003. Acceptedfor publication Jun 20, 2003.

Address correspondence to Dr Schmitt, Aachen Spinal Cord Re-search Center, Department of Neurology, University HospitalAachen, Pauwelsstrasse 30, 52057 Aachen, Germany.E-mail: neurodoc@gmx.de

534 © 2003 American Neurological AssociationPublished by Wiley-Liss, Inc., through Wiley Subscription Services

(p-zero1-1; Invitrogen, La Jolla, CA). For nonradioactive insitu hybridization, RNA probes were generated by in vitrotranscription, using the Boehringer digoxigenin RNA label-ing kit. Transverse paraffin sections (5–8�m thick) were de-waxed and rehydrated in 0.1M phosphate-buffered salinebuffer, followed by incubation proteinase K (10�g/ml) for20 minutes at 37°C. For in situ hybridization, the Boehr-inger protocol for nonradioactive in situ hybridization usingdigoxigenin-labeled RNA probes was used as described pre-viously.7

ImmunohistochemistrySections were incubated with monoclonal antibodies recog-nizing the 200kDa phosphorylated neurofilament epitope(dilution 1:200; Sigma, St. Louis, MO), GAP-43 (dilution1:200; Chemicon, Temecula, CA) and activated caspase-3(dilution 1:20; Biovision, Mountain View, CA). Appropriateantibodies for the detection of c-jun protein in paraffin-embedded human tissues were not available. For visualiza-tion, sections were incubated in avidin-biotin-peroxidasecomplex (Vectastain, Vector Laboratories, Burlingame, CA),followed by revelation of reaction product using diamino-benzidine (DAB, Sigma) containing H2O2 (0.1%). For neg-ative controls, the primary antibody was omitted.

Double immunohistochemistry for GAP-43 andneurofilamen was performed according to the method ofLevey.8 In brief, after incubation with the GAP-43 anti-body and visualization with DAB (brown reaction prod-uct), slides were processed for 200kDa NF immunohisto-chemistry. Benzidine-dihydrochloride was used for revelationof the reaction product (punctate dark green reaction prod-uct).

ResultsAt 6 hours after injury, no induction of c-jun orGAP-43 mRNAs or proteins could be detected (Ta-ble). At 8 to 14 days after injury, a drastic upregulationof both c-jun and GAP-43 mRNA could be detected inlesioned Clarke’s nucleus neurons (Fig 1A, C, E). Theexpression of c-jun appeared to be more intense thanthat of GAP-43 (see Table). No induction of c-jun orGAP-43 mRNA could be detected within unaffectedneurons rostral to the lesion. The staining intensity ofClarke’s nucleus neurons rostral to the lesion did notdiffer from control tissue. The induction of c-jun andGAP-43 mRNAs was more prominent close to the le-sion site (see Fig 1 A, E). By 4 months and 1 year after

injury, c-jun and GAP-43 expression has returned tocontrol levels (see Fig 1B, D, E).

The pattern of lesion-induced changes for GAP-43mRNA was closely mirrored by changes in protein ex-pression. No alteration of GAP-43 could be observedat 6 hours, 4 months, and 1 year after injury. How-ever, double immunohistochemistry showed enhancedstaining for GAP-43 protein within axotomized Clar-ke’s nucleus neurons at 8 to 14 days after injury. Thisenhanced staining colocalized in Clarke’s nucleus neu-rons that also demonstrated an accumulation of neuro-filament protein (Fig 2A–C). Immunohistochemistryof sections taken approximately five segmental levelscaudal to the lesion site showed fewer neurons withenhanced NF or GAP-43 staining than could be de-tected in sections taken closer to the lesion site (see Fig2C). The accumulation of NF and enhanced GAP-43staining could not be detected in control cases, nor wasit seen in any Clarke’s nucleus neuron rostral to thelesion, nor in Clarke’s nucleus neurons caudal to thelesion at early (ie, 6 hours after injury) or late (ie, 1month and 1 year after injury) survival times (see Fig2D, Table). A semiquantitative assessment of thelesion-induced changes of mRNA and protein levelswithin Clarke’s nucleus neurons caudal to the lesionwas made independently by three investigators, the re-sults of which are summarized in the Table.

Furthermore, immunohistochemistry to detect theactivated form of caspase-3, a sensitive indicator of ap-optosis, did not show any staining within Clarke’s nu-cleus neurons caudal to the lesion at any time pointinvestigated (see Fig 2E). Sections of a malignant braintumor (medulloblastoma) were used as controls (seeFig 2F).

DiscussionAlthough much experimental data are already availableconcerning the induction of important RAGs, this in-vestigation represents, to the best of our knowledge,the first demonstration of RAG expression by injuredhuman CNS neurons. In comparison with the experi-mental data, there are only relatively few reports of ret-rograde neuronal reactions in human pathologies.10,11

This study showed that both GAP-43 and c-jun

Table. Semiquantitative Assessment of the Investigated mRNA and Protein Expression within Clarke’s Nucleus Neurons Caudal toSpinal Cord Lesions after Different Survival Times

Controls 6 hr 8 days 11 days 13 days 14 days 4 mo 1 yr

GAP-43 mRNA � � �� �� �� �� � �GAP-43 protein � � �� �� �� �� � �C-jun mRNA � � ��� ��� ��� ��� � �NF protein � � �� �� �� �� � �

(�) � negative; (�) � slight; (��) � moderate; (���) � strong staining. Note that controls not only refer to normal, nonlesioned cases,but also to Clarke’s nucleus neurons rostral to the lesion of all survival times.

Schmitt et al: Neuronal Changes After Axotomy 535

Fig 1. (A–F) Nonradioactive in situ hybridization detection of mRNA for c-jun and GAP-43 within the Clarke’s nucleus. (A)Neurons of a patient who died 14 days after trauma demonstrated drastic upregulation of c-jun mRNA in segments caudal andclose to the lesion site. (B) Clarke’s nucleus neurons in a long survival time case (1 year) show a lower level of mRNA expressionthan that seen at 14 days after injury. This basal level of staining is similar to control (nonlesioned) cases or to that seen in Clar-ke’s nucleus neurons rostral to the lesion. (C) Clarke’s nucleus neurons in a case 14 days after injury demonstrate upregulation ofGAP-43 mRNA in segments caudal and close to the lesion. (D) Clarke’s nucleus neurons of a long survival time case (1 year) showa basal expression that is comparable with that of control cases or Clarke’s nucleus neurons rostral to the lesion. (E) Higher magni-fication of neurons in panel C clearly demonstrates enhanced GAP-43 mRNA expression. (F) Higher magnification of neurons inpanel D. Scale bars � 100�m in A–D and 50�m in E and F.

536 Annals of Neurology Vol 54 No 4 October 2003

Fig 2. (A–C) Double immunohistochemistry for NF (dark green punctate reaction product) and GAP-43 (diffuse brown reactionproduct) immunoreactivity within the area of Clarke’s nucleus caudal and close to the lesion site. (A) All Clarke’s nucleus neurons(arrows) in sections taken close to the lesion site demonstrate increased NF and GAP-43 staining at 14 days after severe spinal cordtrauma. (B) Higher magnification of panel A. (C) In caudal segments more remote from the lesion site (more than five segmentallevels), fewer neurons demonstrate the enhanced NF and GAP-43 staining. Both labeled (arrow) and unlabeled (arrowhead) Clar-ke’s nucleus neurons can be identified in the same section. (D) Unlabeled Clarke’s nucleus neurons (arrows) in a long survival timecase (1 year). (E) Clarke’s nucleus neurons caudal to the spinal cord injury are negative for activated caspase-3 (arrows) staining at11 days after injury and at all other postinjury survival times. (F) Positive control for activated caspase-3 staining shows apoptoticcells (arrows) within a section of a medulloblastoma. Scale bars � 100�m in A and D and 50�m in B, C, E, F. Thionin coun-terstain was used for F.

Schmitt et al: Neuronal Changes After Axotomy 537

mRNA and GAP-43 protein are transiently upregu-lated within axotomized, nonregenerating Clarke’s nu-cleus neurons after severe spinal cord injuries in hu-mans. The lack of suitable antibodies for the detectionof c-jun protein in wax-embedded human material pre-vented the complimentary investigation of c-jun pro-tein. Our results confirm experimental data that evenlesioned CNS neurons are capable of upregulatingRAGs, thus displaying several characteristics that aresimilar to effectively regenerating PNS neurons.6,12–14

This supports the notion of a phase of attempted CNSaxon regeneration that fails over time.

Because Clarke’s nucleus neurons are known to un-dergo degenerative changes after axotomy, we per-formed immunohistochemistry for the activated formof caspase-3, which is known to be a sensitive indicatorof apoptosis.9 However, the lack of expression ofcaspase-3 within lesioned Clarke’s nucleus neuronsshowed no evidence for apoptosis, supporting datafrom Sanner and colleagues, who described glutamate-induced excitotoxicity to be the mechanism of neuro-nal degeneration in axotomized Clarke’s nucleus neu-rons.15

Our observation of elevated GAP-43 protein expres-sion within axotomized Clarke’s nucleus neurons is aninteresting finding because the protein usually cannotbe detected within the cell bodies of nonlesioned neu-rons. This is because, after synthesis, the protein is rap-idly transported from the soma to the axons.4 Thepresent postmorten human data, and also that obtainedin experimental animals,16,17 suggest that the inductionof the GAP-43 mRNA leads to enhanced protein pro-duction and its accumulation within the neuronal cellbodies. Furthermore, this accumulation might be trig-gered by disturbed axonal transport caused by axonaltransection.

These data demonstrated an induction of GAP-43and c-jun mRNA and protein that was more promi-nent in sections located close to the lesion site. Thiseffect might reflect the extent of axonal damage thattook place in these lesions; that is, some axons mayhave been spared after the injury. However, this is un-likely because the lesions were described as being ana-tomically complete. The accumulation of NF in thecell bodies of axotomized neurons already has been de-scribed in both humans and experimental animals.10,18

Indeed, our results demonstrated an enhanced stainingfor NF within all Clarke’s nucleus neurons caudal andclose to the lesion site at relatively early survival times,but never in Clarke’s nucleus neurons rostral to thelesion. A similar observation already has been re-ported by Martin and colleagues who described anincreased accumulation of NF in human Clarke’s nu-cleus neurons caudal to a surgically induced tractot-omy at early survival times.10 An alternative reasonfor the more prominent elevation of RAGs by neu-

rons located closer to the lesion site than by thoselocated more distally could be explained by the so-called “near-far” principle. Experimental observationshave indicated that both GAP-43 and c-jun are in-duced in rubrospinal neurons after proximal, but notafter distal axotomy.12,14

Taken together, these data obtained from post-morten human tissues confirm the clinical significanceof numerous experimental observations, indicating thatcertain populations of intrinsic CNS neurons are ableto transiently upregulate important RAGs. This eventprobably reflects an initial regenerative attempt thatfails over time. Furthermore, we did not find any in-dication of apoptosis occurring within affected Clar-ke’s nucleus neurons. This study contributes to a bet-ter understanding of the molecular programs involvedin the process of axonal regeneration in pathologicalhuman tissues. Such comparative investigations arenecessary for the logical transfer of intervention strat-egies from the laboratory bench to the clinical situa-tion.

This study was supported by grants from the Bundesministerium furBildung und Forschung (01K09804, A.B.S.) and the Deutsche For-schungsgemeinschaft (Schm3-1, 2, A.B.S.). This work forms part ofthe doctoral thesis of L.P. (D82, Diss RWTH Aachen).

References1. Schwab ME, Bartholdy D. Degeneration and regeneration of

axons in the lesioned spinal cord. Physiol Rev 1996;76:319–370.

2. Fawcett JW, Asher RA. The glial scar and central nervous sys-tem barrier. Brain Res Bull 1999;49:377–391.

3. Schwab ME. Repairing the injured spinal cord. Science 2002;295:1029-–1031.

4. Skene JHP. Growth-associated proteins and the curious dichot-omies of nerve regeneration. Cell 1984;37:697–700.

5. Herdegen T, Skene P, Bahr M. The c-Jun transcription factor-bipotential mediator of neuronal death, survival and regenera-tion. Trends Neurosci 1997;20:227–231.

6. Schmitt AB, Breuer S, Voell M, et al. GAP-43 (B-50) andC-Jun are up-regulated in axotomized neurons of Clarke’s nu-cleus after spinal cord injury in the adult rat. Neurobiol Dis1999;6:122–130.

7. Brook GA, Schmitt AB, Nacimiento W, et al. distribution ofB-50 (GAP-43) mRNA and protein in the normal adult humanspinal cord. Acta Neuropathol 1998;95:378–386.

8. Levey AI, Bolam JP, Rye DB, et al. A light and electron mi-croscopic procedure for sequential double antigen localizationusing diaminobenzidine and benzidine dihydrochloride. J His-tochem Cytochem 1986;34:1449–1457.

9. Duan WR, Garner DS, Williams SD, et al. Comparison of im-munohistochemistry for activated caspase-3 and cleaved cyto-keratin 18 with the TUNEL method for quantification of ap-optosis in histological sections of PC-3 subcutaneousxenografts. J Pathol 2003;199:221–228.

10. Martin JE, Mather KS, Swash M, et al. Spinal cord trauma inman: studies of phosphorylated neurofilament and ubiquitin ex-pression. Brain 1990;82:802–812.

538 Annals of Neurology Vol 54 No 4 October 2003

11. Kawamura Y, Dyck PJ. Permanent axotomy by amputation re-sults in loss of motor neurons in man. J Neuropathol Exp Neu-rol 1981;40:658–666.

12. Jenkins R, Tetzlaff W, Hunt SP. Differential expression of im-mediate early genes in rubrospinal neurons following axotomyin rat. Eur J Neurosci 1993;5:203–209.

13. Tetzlaff W, Alexander SW, Miller FD, et al. Response of facialand rubrospinal neurons to axotomy: changes in mRNA expres-sion for cytoskeletal proteins and GAP-43. J Neurosci 1991;11:2528–2544.

14. Tetzlaff W, Kobayashi NR, Giehl KMG, et al. Response ofrubrospinal and corticospinal neurons to injury and neurotro-phins. Prog Brain Res 1994;103:271–286.

15. Sanner CA, Cunningham TJ, Goldberger ME. NMDA recep-tor blockade rescues Clarke’s and red nucleus neurons after spi-nal hemisection. J Neurosci 1994;14:6472–6480.

16. Curtis R, Green D, Lindsay RM, et al. Up-regulation ofGAP-43 and growth of axons in rat spinal cord after compres-sion injury. J Neurocytol 1993;22:51–64.

17. Brook GA, Plate D, Franzen R, et al. Spontaneous longitudinallyorientated axonal regeneration is associated with the Schwann cellframework within the lesion site following spinal cord compres-sion injury of the rat. J Neurosci Res 1998;53:51–65.

18. Goldstein ME, Cooper HS, Bruce J, et al. Phosphorylation ofneurofilament proteins and chromatolysis following transectionof rat sciatic nerve. J Neurosci 1987;7:1586–1594.

A Diagnostic Cycle Test forMcArdle’s DiseaseJohn Vissing, MD, PhD,1 and Ronald G. Haller, MD2

We investigated whether the second wind phenomenon(ie, a decrease in heart rate and perceived exertion duringexercise) is pathognomonic for McArdle’s disease.Twenty-four patients with McArdle’s disease, 17 healthysubjects, and 25 patients with other inborn errors ofmuscle metabolism cycled a constant workload for 15minutes. In McArdle’s disease patients, heart rate consis-tently decreased by 35 � 3 beats per minute from the 7th

to the 15th minute of exercise, whereas heart rate in-creased progressively with exercise in all 42 control sub-jects. The findings indicate that cycling at a moderate,constant workload provides a specific, sensitive, and sim-ple diagnostic test for McArdle’s disease.

Ann Neurol 2003;54:539–542

In McArdle’s disease, muscle glycogen breakdown istypically blocked completely because of absence ofmyophosphorylase activity. The presence of a meta-bolic myopathy, particularly the most common defectof muscle carbohydrate metabolism, McArdle’s disease,should always be considered in patients with com-plaints of exercise intolerance, exercise-induced musclepain, and cramps, but a simple screening procedurethat can unequivocally confirm or exclude such a diag-nosis is not available.

Muscle glycogen is the most important fuel forworking muscle early during exercise and at high workintensities.1 McArdle’s disease patients therefore expe-rience severe exercise intolerance, and intense exerciseprovokes cramps, muscle injury, and myoglobinuria.2,3

Diagnosis can be suggested by a forearm exercise test,but final diagnosis rests on either molecular genetic in-vestigations or histochemical or biochemical demon-stration of absent myophosphorylase activity in muscle.

From the 1Department of Neurology and Copenhagen Muscle Re-search Center, National University Hospital, Rigshospitalet, Copen-hagen, Denmark; and 2Neuromuscular Center, Institute for Exerciseand Environmental Medicine, Presbyterian Hospital and Depart-ment of Neurology, VA Medical Center and UT SouthwesternMedical Center, Dallas, TX.

Received Mar 17, 2003, and in revised form Jun 18. Accepted forpublication Jun 24.

Address correspondence to Dr Vissing, Neuromuscular Clinic, De-partment of Neurology, 2082 National University Hospital, Rig-shospitalet Blegdamsvej 9, DK-2100 Copenhagen, Denmark.E-mail: vissing@rh.dk

© 2003 American Neurological Association 539Published by Wiley-Liss, Inc., through Wiley Subscription Services