CO-INDUCTION OF NITRIC OXIDE SYNTHASE, BCL-2 AND

GROWTH-ASSOCIATED PROTEIN-43 IN SPINAL MOTONEURONS DURING AXON

REGENERATION IN THE LIZARD TAIL

L. CRISTINO,*² A. PICA,* F. DELLA CORTE* and M. BENTIVOGLIO²³

*Department of Evolutionary and Comparative Biology, University of Naples ªFederico IIº, Naples, Italy

²Department of Morphological and Biomedical Sciences, Section of Anatomy and Histology, Medical Faculty, University of Verona,Strada Le Grazie 8, 37134 Verona, Italy

AbstractÐIn lizards, tail loss transects spinal nerves and the cut axons elongate in the regrowing tail, providing a natural paradigmof robust regenerative response of injured spinal motoneurons. We previously ascertained that these events involve nitric oxidesynthase induction in the axotomized motoneurons, suggesting a correlation of this enzyme with regeneration-associated geneexpression. Here we investigated, in lizards, whether the cell death repressor Bcl-2 protein and growth-associated protein-43 (GAP-43) were also induced in motoneurons that innervate the regenerated tail in the ®rst month post-caudotomy. Single and multipleimmunocytochemical techniques, and quantitative image analysis, were performed. Nitric oxide synthase, GAP-43 or Bcl-2immunoreactivity was very low or absent in spinal motoneurons of control lizards with intact tail. Nitric oxide synthase andGAP-43 were induced during the ®rst month post-caudotomy in more than 75% of motoneurons which innnervate the regenerate.Bcl-2 was induced in approximately 95% of these motoneurons at ®ve and 15 days, and in about 35% at one month. The intensity ofBcl-2 and GAP-43 immunostaining peaked at ®ve days, and nitric oxide synthase at 15 days; immunoreactivity to these proteinswas still signi®cantly high at one month. Immuno¯uorescence revealed co-localization of nitric oxide synthase, GAP-43 and Bcl-2in the vast majority of motoneurons at ®ve and 15 days post-caudotomy.

These ®ndings demonstrate that co-induction of nitric oxide synthase, Bcl-2 and GAP-43 may be part of the molecular repertoireof injured motoneurons committed to survival and axon regeneration, and strongly favor a role of nitric oxide synthase inmotoneuron plasticity. q 2000 IBRO. Published by Elsevier Science Ltd. All rights reserved.

Key words: axotomy, free radicals, neuroplasticity, neurotoxicity, gecko.

Lizards detach their tail for defense, and tail loss is promptlyfollowed by regeneration of the lost tail portion. The intact tailcontains the caudal segments of the spinal cord and dorsalroots, which provide a segmental peripheral innervation. Dueto the oblique course of spinal nerves along three segments,tail loss transects three pairs of spinal nerves deriving fromthe segments rostral to the amputation.29,33 The regrowing tailis innervated by these axotomized sensory ganglion cells andspinal motoneurons12,13 and their target, when the regenera-tion is complete, can reach an extent 20 times greater than inthe original tail. Thus, the lizard tail loss and regenerationprovides a unique biological model of target ablation andaxotomy, followed by remarkable target expansion andaxonal elongation.

In the ®rst days after tail loss, changes typical of a retro-grade reaction to injury are evident in the axotomized spinalmotoneurons of the lizard.12,15,32 Neurons in which retrogradechanges are induced by axonal damage can undergo degen-eration or recovery, depending upon several parameters,which include the lesion severity and the age of the animal,16

but the crucial factors determining neuronal fate and nerverepair are still undetermined. In the lizard, the retrograde cellchanges consequent to tail loss were found to recover, with no

evidence of motoneuron loss or apoptotic phenomena.12

Therefore, this paradigm is well suited for the study ofregeneration-associated molecular events.

We have demonstrated in a previous study that the neuronalisoform of nitric oxide synthase (nNOS), the syntheticenzyme of the free radical nitric oxide (NO), is induced inthe lizard spinal motoneurons that innervate the regenerate.12

The enzyme expression was found to be temporally regulatedand persisted for at least some months after complete tailregeneration.12 These ®ndings indicated that nitric oxidesynthase (NOS) induction in damaged spinal motoneurons,which has been repeatedly described in rodents after ventralroot avulsion,8,24,35 is a feature highly conserved in evolution.At variance with the wealth of data indicating that damage-induced NOS plays a neurotoxic role in injured moto-neurons,11,20,26,31,35 these ®ndings favored instead an involve-ment of the enzyme in plastic phenomena.

In view of the potential role played by NOS inductionand oxidative stress in motoneuron disease and degenera-tion,34 an understanding of the molecular correlates of NOSexpression deserves special attention. We thus examinedwhether, in the lizard motoneurons that innervate the regen-erate, this enzyme was induced concomitantly with proteinsinvolved in neuronal survival and plasticity, such as Bcl-2and the growth-associated protein-43 (GAP-43), and whetherthese proteins were co-expressed in the same motoneurons.Several data have indicated that the Bcl-2 protein plays amajor role as a cell death repressor.2,25 However, GAP-43, aneuronal protein highly conserved in evolution, is selectivelyexpressed during peripheral axon regeneration in the adultnervous system.27

Motoneuron plasticity and NOS 451

451

Neuroscience Vol. 101, No. 2, pp. 451±458, 2000q 2000 IBRO. Published by Elsevier Science Ltd

Printed in Great Britain. All rights reserved0306-4522/00 $20.00+0.00PII: S0306-4522(00)00393-6

Pergamon

www.elsevier.com/locate/neuroscience

³To whom correspondence should be addressed. Tel./fax: 139-045-8027-158; fax: 139-045-8027-163.

E-mail address: marina@anatomy.univr.it (M. Bentivoglio).Abbreviations: ALS, amyotrophic lateral sclerosis; ANOVA, analysis of

variance; DRG, dorsal root ganglion; GAP-43, growth-associated proteinof 43,000 mol. wt; IR, immunoreactive; NGS, normal goat serum; NHS,normal horse serum; nNOS, neuronal nitric oxide synthase; NO, nitricoxide; NOS, nitric oxide synthase; PBS, phosphate-buffered saline.

EXPERIMENTAL PROCEDURES

Animals and histology

The study was performed on 18 male adult Tokay geckos (Gekkogecko; body weight 75±100 g, length 22±24 cm), purchased from anauthorized breeder (Schneider, Varese, Italy). The experiments wereperformed following protocols that had received approval by theItalian Ministry of Health, and all efforts were made to minimize thenumber of animals used and to avoid any suffering. The animals, noneof which had lost the tail prior to the experiments, were maintained at aconstant temperature of 268C, under a 12-h/12-h light±dark cycle, andwere nourished with mealworms and water ad libitum. The animalswere randomly divided into four groups, as indicated in Table 1. Thetail was left intact in four cases, that were used as controls. In the other14 animals, amputation was achieved by gently pulling the tail througha nylon slipknot at the level of the sixth vertebral tail segment, thusmimicking the conditions of autotomy in the natural environment ofthe animals. After ®ve (n� 5), 15 (n� 5) and 30 days (n� 4), theanimals were anesthetized with ethyl ether and transcardially perfusedwith saline (6.9% NaCl) followed by 4% paraformaldehyde in 0.1 Mphosphate buffer (pH 7.4). The three segments of the tail spinal cord(i.e. caudal levels c3±c5) rostral to the level of amputation, and thesame levels in the intact animals, were dissected out and post®xed for1 h.

The tissue sampled from the control animals and from 10 caudoto-mized animals (three that had survived ®ve and 15 days, respectively,and the four geckos that had survived one month) was paraf®nembedded. Sections were then cut at a thickness of 10±13 mm inthree series, each consisting of pairs of serial sections: one sectionwas processed for immunohistochemistry with either anti-nNOS orGAP-43 or Bcl-2 antibodies, and the adjacent matched sections wereNissl stained.

The tissue sampled from the other four caudotomized animals (twocases at ®ve days and two at 15 days after tail amputation) was soakedfor cryoprotection in 30% sucrose in phosphate buffer and then cut on acryostat into 12-mm-thick frozen sections that were collected in threeseries: the ®rst section was processed for NOS/GAP-43 doubleimmuno¯uorescence, the second for Bcl-2/GAP-43 double immuno-¯uorescence, and the third was Nissl stained. After processing forimmuno¯uorescence, the sections were air dried and coverslippedwith glycerol. All the other sections were dehydrated through the alco-hol series and coverslipped with Entellan.

Immunohistochemical procedures

Single antigen immunohistochemistry. For single NOS immunocyto-chemistry, the sections were reacted for 15 min in 3% H2O2 to inacti-vate endogenous peroxidase activity and incubated for 60 min at roomtemperature in 5% normal goat serum (NGS; Dako, Glostrup,Denmark) in 0.1 M phosphate-buffered saline (PBS; pH 7.4), contain-ing 0.1% Triton X-100 (Sigma, St Louis, MO, USA). The sectionswere then incubated overnight, in a humid chamber, with rabbit poly-clonal antibodies that recognize nNOS (Boehringer Mannheim,Mannheim, Germany), diluted 1:100 in NGS. After several rinses,the sections were incubated for 2 h in biotinylated goat anti-rabbitimmunoglobulins G (Vector Laboratories, Burlingame, CA, USA),diluted 1:50 in NGS, followed by incubation for 1 h at room tempera-ture in avidin±biotin±peroxidase solution (ABC Kit; Vectastain,Vector) in PBS, and then for 10 min in 0.05% 3,3 0-diaminobenzidine

and 0.01% H2O2 in 0.01 M Tris±HCl-buffered saline (pH 7.6). As anegative control, some sections derived from caudotomized animalswere processed with the same protocol omitting the primary antibody;no immunostaining was detected in these sections.

The same procedure was followed for Bcl-2 immunostaining usingrabbit polyclonal anti-Bcl-2 antibodies (Santa Cruz Biotechnology,Santa Cruz, CA, USA; diluted 1:100), and for GAP-43 immunostain-ing, using mouse monoclonal anti-GAP-43 antibodies (BoehringerMannheim; diluted 1:50), normal horse serum (NHS; Dako) and bio-tinylated horse anti-mouse immunoglobulins G (Vector).

Multiple labeling with immuno¯uorescence. For triple labeling ofthe same motoneuronal cell populations, we exploited a strategy inwhich double immuno¯uorescence (NOS/GAP-43 and GAP-43/Bcl-2, respectively) was performed in consecutive cryostat-cut thinsections, using the same primary antibodies described above.

The sections were ®rst incubated for 1 h in 5% NHS in PBS contain-ing 0.1% Triton X-100 and then overnight in the monoclonal anti-GAP-43 antibodies diluted 1:50 in NHS. GAP-43 immuno¯uorescencewas visualized by incubation for 3 h in rhodamine-conjugated horseanti-mouse immunoglobulins G (Santa Cruz Biotechnology; diluted1:25). After rinsing, alternate sections were incubated overnight witheither polyclonal anti-nNOS or anti-Bcl-2 antibodies diluted 1:50 inNGS. Immuno¯uorescence to the antibody used in the second step ofthe procedure was visualized by incubation for 3 h in ¯uoresceinisothiocyanate-conjugated goat anti-rabbit immunoglobulins G(Santa Cruz Biotechnology; diluted 1:25).

Data analysis

The sections processed for immuno¯uorescence were studied usingan epi¯uorescence microscope equipped with the appropriate ®lters;all the other material was investigated at the microscope under bright-®eld illumination. Quantitative analyses were performed blind withrespect to the animal group assignment.

The relative proportion of NOS-immunoreactive (IR), GAP-43-IRor Bcl-2-IR motoneurons at each time-point was evaluated in the seriesof sections processed for single immunocytochemistry with respect tothe matched Nissl-stained sections. Cell counts were performed withthe same criteria adopted previously.12 In brief, the immunostainedmotoneurons were counted on both sides of the spinal cord in 30sections per group (10 sections per animal) and expressed as a percent-age over the total number of motoneurons with their nucleoli in thefocal plane in the adjacent Nissl-stained sections.

Quantitative evaluation of the intensity of the immunostaining ofeach marker in the caudal spinal motoneurons at the different timeswas performed with image analysis in the sections processed for singleimmunocytochemistry. A sample of 60 immunopositive motoneuronsper marker (NOS, GAP-43 or Bcl-2), with nuclei (unstained or lightlystained) in the focal plane, were randomly selected in the caudalsegments c3±c5 from three geckos per group. Images were acquired,under constant light illumination and at the same magni®cation, usingthe digital camera JVC KY-F58 connected to the microscope and theimage analysis software Image Pro Plusw 4.0 for Windows (MediaCybernetics, Silver Springs, MD, USA). In each section, the zerovalue of optical density was assigned to the background, i.e. a portionof the spinal ventral horn devoid of stained cell bodies and neuropil.For statistical analysis of densitometric measures, the differences of theaverage optical density values obtained in each group were evaluated

L. Cristino et al.452

Table 1. Experimental parameters

Immunohistochemistry Multiple labeling

Group(number of animals)

Survival aftercaudotomy

(days)

Nissl staining nNOS GAP-43 Bcl-2 nNOS/GAP-43 Bcl-2/GAP-43

1. Control (n� 4) (no caudotomy) n� 4 n� 4 n� 4 n� 4 Ðoriginal tail2. Amputated tail (n� 5) 5 n� 5 n� 3 n� 3 n� 3 n� 23. Regenerating tail (n� 5) 15 n� 5 n� 3 n� 3 n� 3 n� 2regenerate length: 3±4 mm4. Regenerating tail (n� 4) 30 n� 4 n� 4 n� 4 n� 4 Ðregenerate length: 6±7 mm

using one-way analysis of variance (ANOVA), followed by theBonferroni post hoc test, with P , 0.05 set as the level of con®dence.Unless otherwise stated, data are presented as mean^ S.D.

Quantitative evaluation of the co-expression of Bcl-2, GAP-43 andNOS was performed in spinal motoneurons of the segments c3±c5 at®ve and 15 days post-caudotomy identi®ed in the sections processedfor double immuno¯uorescence, and derived from ®ve pairs of adja-cent sections per animal in each group. To avoid double counting, eachpair of sections was at least 100 mm apart.

RESULTS

The cell changes observed in Cresyl Violet-stained sectionsin the spinal ventral horn at the examined time-points duringthe ®rst month after caudotomy are consistent with thosedescribed previously,12 and included perikaryal swellingand nuclear eccentricity (Fig. 1B±D). Marked variation ofthe immunohistochemical phenotype of motoneurons, witha different temporal regulation of each marker, was foundin the pattern of immunopositivity of the analysed proteins(Fig. 1E±R).

Features of nitric oxide synthase, growth-associated protein-43 or Bcl-2 immunoreactivity

In control animals with intact tails, basal expression of theexamined proteins was either absent or very low at the thresh-old of immunohistochemical revelation, and a few verylightly NOS- or Bcl-2-immunostained perikaryal pro®leswere seen above background (Fig. 1E, I, O).

At ®ve days after caudotomy, NOS-IR motoneurons weredistinctly evident in the ventral horn (Fig. 1F), and NOSimmunostaining persisted up to 30 days after caudotomy(Fig. 1G, H). Immunopositive neurons and neuropil immuno-reactivity were also seen in the other portions of the spinalgray matter, and especially in the dorsal horn (Fig. 1G).

Selective and intense induction of GAP-43 immuno-reactivity was detected in motoneurons at ®ve days post-caudotomy (Fig. 1L). The induction was clearly evident inmotoneurons also at 15 days, when no other immunostainedcell bodies were seen, but marked GAP-43 immunopositivitywas detected throughout the neuropil of the spinal graymatter, as well as in ®bers (Fig. 1M). GAP-43 immuno-reactivity was markedly down-regulated one month aftercaudotomy, when, however, the immunopositivity was stillevident throughout the spinal gray matter and in lightlyimmunopositive motoneuron pro®les (Fig. 1N).

Bcl-2 immunoreactivity exhibited a time-course similar tothat described above for GAP-43, but it was intensely andwidely expressed in spinal neurons also outside the ventralhorn. Thus, at ®ve days, Bcl-2 immunostaining was veryintense in motoneurons, in other neuronal cell bodies distrib-uted throughout the gray matter and in the neuropil of thedorsal horn (Fig. 1P). Bcl-2 immunoreactivity was thendown-regulated, but was clearly evident in the ventral hornat 15 days (Fig. 1Q). At one month post-axotomy, Bcl-2immunostaining was still above basal levels, and was detectedin motoneuron pro®les and in the neuropil, especially in thedorsal horn (Fig. 1R).

The quantitative analysis of the immunostained cell popu-lations (Fig. 2A) con®rmed that, in control animals with intacttails, a few motoneurons exhibited GAP-43 (mean^S.D.:3.73^ 0.6%) or Bcl-2 (2.11^ 0.7%) immunoreactivityabove background; the proportion of motoneurons exhibitingbasal NOS immunostaining was slightly higher (6.8^ 0.8%).

The relative proportion of immunopositive motoneuronsincreased dramatically at ®ve days after tail transection(Fig. 2A), when the vast majority of motoneurons wereNOS-IR (84.2^ 1.9%) or GAP-43-IR (77.8^ 1.6%), andBcl-2 was induced in approximately 95% (94.3^ 2.5%) ofmotoneurons. Such percentages were not substantially modi-®ed at 15 days. The proportion of NOS-IR (83.3^ 2.2%) orGAP-43-IR (77.3^ 1.9%) motoneurons was still relativelystable at one month, when the number of motoneurons exhi-biting Bcl-2 immunoreactivity had markedly decreased, andabout one-third (34.2^ 1.7%) of these cells were Bcl-2immunopositive.

The densitometric analysis of the immunostaining intensitycon®rmed the qualitative observation of a marked inductionof the three proteins in motoneurons after tail transection(Fig. 2B). This set of data indicated that Bcl-2 and GAP-43immunoreactivities were highest early after caudotomy (i.e. at®ve days), whereas NOS immunopositivity peaked at 15 days.The immunostaining decreased at one month, when, however,it was still higher than in the animals with intact tails.

The statistical evaluation revealed a highly signi®canteffect of time in the variation of NOS immunostainingintensity at the different times (one-way ANOVA, F3 �84.975, P , 0.001), as well as of Bcl-2 (F3� 95.271,P , 0.001) and GAP-43 (F3� 155.651, P , 0.001) immuno-reactivities. At the post hoc evaluation, highly signi®cantdifferences were found between the intensity of immuno-positivity of each marker at the different time-points, aswell as with respect to controls (Bonferroni test, P , 0.005for all multiple comparisons).

Multiple immuno¯uorescent labeling

In the animals in which double immuno¯uorescence (Bcl-2or NOS combined with GAP-43) was performed in consecu-tive sections, strong labeling of either marker could be iden-ti®ed in single-, double- and triple-labeled motoneurons (Fig.3). The same cells were easily recognized in the pairs ofadjacent sections (Fig. 3), and the features of immunolabelingwere similar to those observed with single immunocyto-chemical procedures.

All the motoneurons identi®ed in the sections processed forimmuno¯uorescence were Bcl-2 positive; NOS, GAP-43, orboth, were consistently co-localized with Bcl-2 immunoposi-tivity. However, single Bcl-2-labeled cells were also seen,and accounted for less than 10% of the immuno¯uorescentmotoneurons at ®ve days; this proportion increased to about16% at 15 days (Table 2). The results of the quantitativeevaluation of the co-localization of NOS and/or GAP-43with Bcl-2 in the same cells are summarized in Table 2. Itis noteworthy that NOS, GAP-43 and Bcl-2 were co-localizedin the vast majority of motoneurons at both ®ve and 15 days(Table 2, Fig. 3). Approximately 20% of the Bcl-2-labeledmotoneurons also exhibited NOS immuno¯uorescence, butwere GAP-43 immunonegative (Table 2, Fig. 3D±F).

Thus, altogether the NOS-positive motoneurons (NOS1/GAP-432 and NOS1/GAP-431 in Table 2) representedmore than 90% of the Bcl-2-IR motoneurons (90.1^ 5.3%)at ®ve days post-caudotomy, and more than 80% (83.3 ^4.6%) at 15 days. Finally, it is worth noting that GAP-43was always co-localized not only with Bcl-2, but also withNOS, and the GAP-43-positive motoneuron subset represented

Motoneuron plasticity and NOS 453

L. Cristino et al.454

Fig. 1. The caudal c3 spinal cord segment of geckos with intact tail (A, E, I, O), and at ®ve (B, F, L, P), 15 (C, G, M, Q) and 30 days (D, H, N, R) after tailamputation. Structural post-caudotomy changes are shown in the Nissl-stained sections (B±D): note the marked hypertrophy and nuclear eccentricity inmotoneurons with respect to those of the corresponding segment of the intact tail (A). Note that NOS (E), GAP-43 (I) and Bcl-2 (O) immunoreactivities arebarely detectable in the intact spinal cord, and are markedly induced in motoneurons after caudotomy (F±H, L±N, P±R), as well as in other spinal neurons and/or neuropil (G, H, M, P±R). This plate and Fig. 3 were generated, without alterations, from digital images. Scale bar in R� 62 mm (also applies to A±Q).

80% and 75% of the NOS-positive motoneurons at ®ve and15 days post-caudotomy, respectively.

DISCUSSION

The present data demonstrate that NOS is co-expressedwith Bcl-2 and GAP-43 in the spinal motoneurons thatprovide de novo innervation of the regenerating lizard tail.To our knowledge, this is the ®rst report of the co-localizationof the three proteins in the same neurons, and it is of specialinterest that this occurred in injured motoneurons.

Induction of the proteins, and in particular of NOS andBcl-2, was also observed in the spinal gray matter outside theventral horn. It should be noted in this respect that, in thelizard, tail amputation also results in marked variation ofgene expression in dorsal root ganglion (DRG) cells and inthe dorsal horn.13 In addition, tail loss causes transection ofthe caudal segments of the spinal cord that extend originallythrough the intact lizard tail, and thus results in injury ofcentral spinal pathways.15 Variation of gene expression inthe spinal segments rostral to the amputation level may

therefore re¯ect a response to damage, as described forNOS induced by spinal cord injury in mammals.8,35

Bcl-2 induction in injured motoneurons

The bcl-2 gene and its product, an intracellular membrane-associated protein, act as negative regulators of cell death.2,5

In particular, Bcl-2 has been repeatedly associated with moto-neuron protection: Bcl-2 overexpression in transgenic micewas found to rescue immature facial motoneurons fromaxotomy-induced degeneration,14 and changes in Bcl-2expression have been associated with the regulation ofaxotomy-induced death of mature motoneurons in the rat.3

Taken together with the present ®ndings, the data indicatethat, in our study, Bcl-2 was speci®cally induced in moto-neurons in response to a potentially lethal injury.

In the rat, Bcl-2 was found to be induced followingsciatic nerve transection in both DRG cells and lumbar moto-neurons, but at much higher levels in the former cell popula-tion.18 In the present study, DRGs were not investigated, but itis worth emphasizing that Bcl-2 induction in motoneuronswas very high in our paradigm, in which axons were trans-ected at a short distance from the parent cell bodies. Theintense Bcl-2 immunoreactivity we observed in motoneuronsat early stages after injury agrees not only with the previousevidence that these cells do not die,12 but also with theirmarked regenerative response, since Bcl-2 was found topromote axonal growth in addition to its anti-apoptoticactivity.9

Motoneuron plasticity and NOS 455

Fig. 2. Graphical representation of the quantitative analysis. (A) Histogram of the proportion of nNOS-, GAP-43- or Bcl-2-IR motoneurons in geckos withoriginal tail and at different time-points after caudotomy. Values are expressed as mean percentage^ S.D. (B) Histogram of the immunostaining intensity in

the nNOS-, GAP-43- or Bcl-2-IR motoneurons in geckos at the same time-points. Values are expressed as mean optical density units^S.D.

Table 2. Proportion of motoneurons labeled by immuno¯uorescence

5 days* 15 days

Bcl-21/NOS2/GAP-432 9.1^ 1.9 16.6^ 2.2Bcl-21/NOS1/GAP-432 18.2^ 2.4 20.8^ 2.3Bcl-21/NOS1/GAP-431 72.7^ 3.8 62.5^ 3.6

Numerical data are expressed as mean percentage^S.D.*Days post-caudotomy.

Growth-associated protein-43 and motoneuron plasticity

The neuronal GAP-43 phosphoprotein (also known asB50 or by other designations) has been closely asso-ciated with the growth and regeneration of axonal pro-cesses in vivo and in vitro, and is speci®cally concentratedat the tips of growing axons.5,27 GAP-43 is expressed athigh levels during development in the peripheral andcentral nervous systems, and declines sharply after theestablishment of mature synapses. In adulthood, GAP-43expression is selectively maintained in populations of centralneurons presumed to undergo synaptic remodeling,22 and isinduced during peripheral nerve regeneration after axonalinjury.27

The present ®nding of a very low basal expression ofGAP-43 immunoreactivity in lizard motoneurons corre-lates with the data observed in adult rats, in which GAP-43 mRNA expression is undetectable or very weak inmature somatic motoneurons under normal conditions.22,23

GAP-43 was found to be rapidly up-regulated in ratmotoneurons following axotomy.21,28,30 Thus, the present®ndings con®rm that not only GAP-43 expression but alsoits injury-induced changes are highly conserved in evolu-tion, and that the elongation of axons in the regeneratinglizard tail provides an effective paradigm of motoneuronplasticity.

Co-localization of neuronal nitric oxide synthase, Bcl-2 andgrowth-associated protein-43

In our study, the co-expression of NOS with cell deathrepressor and regeneration-associated proteins was found tocharacterize motoneurons at the early stages of axonregrowth. In agreement with previous data,12 high NOSexpression was found to persist at 30 days, and the present®ndings indicate a co-localization of NOS with Bcl-2 andGAP-43 throughout this period of time.

After sciatic nerve damage, co-induction of NOS and GAP-43 was demonstrated in rat DRG cells,6,19 and in peripheralaxons and growth cones.19 These ®ndings were taken as anindication that NOS up-regulation plays a protective andregenerative role in DRG cells after peripheral injury, atvariance with the wealth of evidence pointing to a toxic effectof NOS induction in motoneurons.8,24,33,35 The present datashowed that NOS can also be co-expressed with GAP-43 ininjured motoneurons. In addition, the co-localization withBcl-2 provided further evidence that these events are part ofthe shift in protein synthesis correlated with motoneuronprotection. In support of this assumption, the number ofBcl-2-positive motoneurons and Bcl-2 immunoreactivitywere highest early after axotomy in the paradigm weinvestigated.

In rat DRGs, the proportion of GAP-43-positive neurons,

L. Cristino et al.456

Fig. 3. Triple ¯uorescent labeling of motoneurons obtained with double immuno¯uorescence of nNOS/GAP-43 in one section, and of GAP-43/Bcl-2 in theadjacent section of spinal cord (caudal c3 level) of geckos at ®ve (A±C) and 15 days (D±F) after caudotomy. The same ®elds were observed under twodifferent wavelengths to elicit red (rhodamine) GAP-43 immuno¯uorescence and green (¯uorescein) nNOS or Bcl-2 immuno¯uorescence, respectively. Theasterisks mark the ventral horn areas enlarged in the insets. Note, in A±C, that all motoneurons are triple labeled; in D±F, one motoneuron, indicated byarrowheads in the insets, is nNOS/Bcl-2 double labeled (white arrowheads), but GAP-43 immunonegative (open arrowhead), whereas the other motoneurons

are triple labeled. Scale bar in A� 64 mm (also applies to B±F), 36 mm (for insets of A±C), 32 mm (for insets of D±F).

and of GAP-43/NOS double-labeled neurons, was found toincrease progressively in the ®rst month after axotomy.19 Inthe present study, the proportion of GAP-43-IR cells in whichthis protein was consistently co-localized with NOS was rela-tively high and stable throughout the ®rst month after axot-omy, indicating an early and marked regenerative response ofthe majority of motoneurons.

Implications for motoneuron pathology

Reactive oxygen species may be generated by NOS, whichis not constitutively expressed in somatic motoneurons, butcan be induced in these cells by noxious stimuli.8,24,35 Atvariance with the present ®ndings, reports based on variouslines of evidence have indicated that damage-induced nNOScan play a role in motoneuron death.11,20,26,31,35 The productionof free radicals has also been implicated in the selectivedegeneration that characterizes, in humans, the familial andsporadic forms of amyotrophic lateral sclerosis (ALS).34

Mutations of the gene encoding Cu±Zn superoxide dismutasehave been discovered in patients affected by familial ALS,and mice carrying superoxide dismutase-mutated geneprovide experimental models of the disease,7 pointing to thepotential toxic role played by oxidative stress. In addition,NOS is induced in degenerating motoneurons during ALS,

and the increase in nitrotyrosine suggested the production ofperoxynitrite in motoneurons of ALS victims.1,4,10 NOS isalso expressed in degenerating spinal motoneurons in wobblermouse motoneuron disease, the progression of which wasfound to be ameliorated by NOS inhibitors, that have beenproposed as potentially therapeutic agents.20

The free radical NO is a pleiotropic molecule and, depend-ing on its redox states, can exert neurotoxic and neuroprotec-tive effects in the central nervous system.23 The present®ndings do not argue against a potentially harmful action ofhigh doses of NO in certain conditions, but draw attention tothe fact that the effects of NO production in motoneurons canalso represent a bene®cial response, counteracting celldamage and promoting regenerative events. In support ofthis assumption, recent data have implicated NO productionin axon viability and in the milieu of regenerating nervestumps.17,36 Taken together with the present ®ndings, thesedata suggest that caution should be exercised in designingtherapeutic strategies for motoneuron disease based on NOSinhibiting activity.

AcknowledgementsÐThis work was supported by grants of the ItalianMinistry of University and Scienti®c Research (MURST Co®n 1998),the National Research Council (CNR) and the Fund for Scienti®cResearch of the University of Naples ªFederico IIº.

REFERENCES

1. Abe K., Pan L.-H., Watanabe M., Konno H., Kato T. and Itoyama Y. (1997) Upregulation of protein-tyrosine nitration in the anterior horn cells ofamyotrophic lateral sclerosis. Neurol. Res. 19, 124±128.

2. Adams J. M. and Cory S. (1988) The Bcl-2 protein family: arbiters of cell survival. Science 281, 1322±1325.3. Baba N., Takehiko K., Itoh M. and Mizuno A. (1999) Reciprocal changes in the expression of Bcl-2 and Bax in hypoglossal nucleus after axotomy in

adult rats: possible involvement in the induction of neuronal cell death. Brain Res. 827, 122±129.4. Beal M. F., Ferrante R. J., Browne S. E., Matthews R. T., Kowall N. W. and Brown R. H. Jr (1997) Increased 3-nitrotyrosine in both sporadic and familial

amyotrophic lateral sclerosis. Ann. Neurol. 42, 646±654.5. Benowitz L. I. and Routtenberg A. (1997) GAP-43: an intrinsic determinant of neuronal development and plasticity. Trends Neurosci. 20, 84±91.6. Bergman E., Carlsson K., Liljeborg A., Manders E., HoÈkfelt T. and Ulfhake B. (1999) Neuropeptides, nitric oxide synthase and GAP-43 in B4-binding

and RT97 immunoreactive primary sensory neurons: normal distribution pattern and changes after peripheral nerve transection and aging. Brain Res. 832,63±83.

7. Brown R. H. Jr (1995) Amyotrophic lateral sclerosis: recent insights from genetics and transgenic mice. Cell 80, 687±692.8. Callsen-Cencic P., Hoheisel U., Kaske A., Mense S. and Tenschert S. (1999) The controversy about spinal neuronal nitric oxide synthase: under which

conditions is it up- or downregulated? Cell Tiss. Res. 295, 183±194.9. Chen D.-F., Schneider G. E., Martinou J.-C. and Tonegawa S. (1997) Bcl-2 promotes regeneration of severed axons in mammalian CNS. Nature 385,

434±439.10. Chou S. M., Wang H. S. and Komai K. (1996) Colocalization of NOS and SOD1 in neuro®lament accumulation within motor neurons of amyotrophic

lateral sclerosis: an immunohistochemical study. J. chem. Neuroanat. 10, 249±258.11. Clowry G. J. (1993) Axotomy induces NADPH diaphorase activity in neonatal but not adult motoneurones. NeuroReport 5, 361±364.12. Cristino L., Pica A., Della Corte F. and Bentivoglio M. (2000) Plastic changes and nitric oxide synthase induction in neurons that innervate the

regenerated tail of the lizard Gekko gecko: I. Response of spinal motoneurons to tail amputation and regeneration. J. comp. Neurol. 417, 60±72.13. Cristino L., Pica A., Della Corte F. and Bentivoglio M. (2000) Plastic changes and nitric oxide synthase induction in neurons which innervate the

regenerated tail of the lizard Gekko gecko: II. The response of dorsal root ganglion cells to tail amputation and regeneration. Brain Res. 871, 83±93.14. Dubois-Dauphin M., Frankowski H., Tsujimoto Y., Huarte J. and Martinou J. -C. (1994) Neonatal motoneurons overexpressing the bcl-2 protooncogene

in transgenic mice are protected from axotomy-induced cell death. Proc. natn. Acad. Sci. USA 91, 3309±3313.15. Duffy M. T., Liebich D. R., Garner L. K., Hawrych A., Simpson S. B. Jr and Davis B. M. (1992) Axonal sprouting and frank regeneration in the lizard tail

spinal cord: correlation between changes in synaptic circuitry and axonal growth. J. comp. Neurol. 316, 363±374.16. Fawcett J. W. and Keynes R. J. (1990) Peripheral nerve regeneration. A. Rev. Neurosci. 13, 43±60.17. Garthwaite G., Goodwin D. A. and Garthwaite J. (1999) Nitric oxide stimulates cGMP formation in rat optic nerve axons, providing a speci®c marker of

axon viability. Eur. J. Neurosci. 11, 4367±4372.18. Gillardon F., Klimaschewski L., Wickert H., Krajewski S., Reed J. C. and Zimmermann M. (1996) 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. 739, 244±250.19. GonzaÂlez-HernaÂndez T. and Rustioni A. (1999) Nitric oxide synthase and growth-associated protein are coexpressed in primary sensory neurons after

peripheral injury. J. comp. Neurol. 404, 64±74.20. Ikeda K., Iwasaki Y. and Kinoshita M. (1998) Neuronal nitric oxide synthase inhibitor, 7-nitroindazole, delays motor dysfunction and spinal motoneuron

degeneration in the wobbler mouse. J. neurol. Sci. 160, 9±15.21. Kobayashi N., Kiyama H. and Tohyama M. (1994) GAP-43 (B50/F1) gene regulation by axonal injury of the hypoglossal nerve in the adult rat. Molec.

Brain Res. 21, 9±18.22. Kruger L., Bendotti C., Rivolta R. and Samanin R. (1993) Distribution of GAP-43 mRNA in the adult rat brain. J. comp. Neurol. 333, 417±434.23. Lipton S. A., Singel D. J. and Stamler J. S. (1994) Nitric oxide in the central nervous system. Prog. Brain Res. 103, 359±364.24. Mariotti R. (1998) Induction of nitric oxide synthase activity and motoneuronal cell damage. Eur. J. Anat. 2, 55±61.25. Merry D. E. and Korsmeyer S. J. (1997) Bcl-2 gene family in the nervous system. A. Rev. Neurosci. 20, 245±267.

Motoneuron plasticity and NOS 457

26. Novikov L., Novikova L. and Kellerth J. -O. (1997) Brain-derived neurotrophic factor promotes axonal regeneration and long-term survival of adult ratspinal motoneurons in vivo. Neuroscience 79, 765±774.

27. Oestreicher A. B., De Graan P. N. E., Gispen W. H., Verhaagen J. and Schrama L. H. (1997) B-50, the growth associated protein-43: modulation of cellmorphology and communication in the nervous system. Prog. Neurobiol. 53, 627±686.

28. Palacios G., Mengod G., Sarasa M., Baudier J. and Palacios J. M. (1994) De novo synthesis of GAP-43: in situ hybridization histochemistry and light andelectron microscopy immunocytochemical studies in regenerating motor neurons of cranial nerve nuclei in the rat brain. Molec. Brain Res. 24, 107±117.

29. Pannese E. (1963) Investigation on the ultrastructural changes of the spinal ganglion neurons in the course of axon regeneration and cell hypetrophy. I.Changes during axon regeneration. Z. Zellforsch. 60, 711±740.

30. Piehl F., Hammarberg H., Tabar G., HoÈkfelt T. and Cullheim S. (1998) Changes in the mRNA expression pattern, with special reference to calcitoningene-related peptide, after axonal injuries in rat motoneurons depends on age and type of injury. Expl Brain Res. 119, 191±204.

31. Ruan R.-S., Leong S.-K. and Yeoh K.-H. (1995) The role of nitric oxide in facial motoneuronal death. Brain Res. 698, 163±168.32. Simpson S. B. and Duffy M. T. (1994) The lizard spinal cord: a model system for the study of spinal cord injury and repair. Prog. Brain Res. 103,

229±241.33. Terni T. (1920) Sulla correlazione tra ampiezza del territorio d'innervazione e grandezza delle cellule gangliari. 2. Ricerche sui gangli spinali che

innervano la coda rigenerata, nei Sauri (Gongylus ocellatus). Arch ital. Anat. Embriol. 17, 507±543.34. Torreilles F., Salman-Tabcheh S., GueÂrin M.-C. and Torreilles J. (1999) Neurodegenerative disorders: the role of peroxynitrite. Brain Res. Rev. 30,

153±163.35. Wu W. (2000) Response of nitric oxide synthase to neuronal injury. In Functional Neuroanatomy of the Nitric Oxide System, Handbook of Chemical

Neuroanatomy (eds Steinbusch H. W. M., De Vente J. and Vincent S. R.), Vol. 17, pp. 315±353. Elsevier, Amsterdam.36. Zochodne D. W., Levy D., Zwiers H., Sun H., Rubin I., Cheng C. and Lauritzen M. (1999) Evidence for nitric oxide and nitric oxide synthase activity in

proximal stumps of transected peripheral nerves. Neuroscience 91, 1515±1527.

(Accepted 23 August 2000)

L. Cristino et al.458