fibroblast growth factor-2 mrna expression in the brainstem and spinal cord of normal and chronic...

Fibroblast Growth Factor-2 mRNAExpression in the Brainstem and SpinalCord of Normal and Chronic SpinallyTransected Urodeles

Marie Moftah,1,2,3 Marc Landry,1,2 Frederic Nagy,1,2 and Jean-Marie Cabelguen1,2*1INSERM U 862, Neurocentre Magendie, Pathophysiology of Spinal Networks, Bordeaux, France2Universite de Bordeaux, INSERM U 862, Neurocentre Magendie, Pathophysiology of Spinal Networks, Bordeaux, France3Zoology Department, Faculty of Science, Alexandria University, Alexandria, Egypt

Descending pathways in the spinal cord of adult urodeleamphibians show a high regenerative ability after bodyspinal cord transection; regenerated axons regrow intothe transected spinal cord, and hindlimb locomotor re-covery occurs spontaneously. Little is currently knownabout the molecular basis of spinal cord regeneration inurodeles, but it is believed that fibroblast growth factor-2(FGF2) may play an important role by inducing prolifera-tion of neural progenitor cells. The aim of our study,using in situ hybridization in adult Pleurodeles waltlii,was twofold: 1) to document FGF2 mRNA expressionpattern along the brainstem-spinal cord of intact sala-manders and 2) to investigate the changes in this patternin animals unable to display hindlimb locomotor move-ments and in animals having fully recovered hindlimblocomotor activity after body spinal cord transection.This design establishes a firm basis for further studieson the role of FGF2 in functional recovery of hindlimblocomotion. Our results revealed a decreasing rostro-caudal gradient in FGF2 mRNA expression along thebrainstem-spinal cord in intact animals. They furtherdemonstrated a long-lasting up-regulation of FGF2mRNA expression in response to spinal transection atthe midtrunk level, both in brainstem and in the spinalcord below the injury. Finally, double immunolabelingshowed that FGF2 was up-regulated in neuroglial, pre-sumably undifferentiated, cells. Therefore, we proposethat FGF2 may be involved in cell proliferation and/or neuronal differentiation after body spinal cord trans-ection in salamander and could thus play an importantrole in functional recovery of locomotion after spinallesion. VVC 2008 Wiley-Liss, Inc.

Key words: FGF2; brainstem; spinal cord; injury;salamander

The adult salamander can recover locomotionabout 3 months following a complete spinal cord trans-ection, even without any treatment (Piatt, 1955; Daviset al., 1990; Chevallier et al., 2004). Recovered locomo-tor activity results from functional regeneration of some

of the lesioned brainstem and spinal fiber tracts acrossthe transection (Davis et al., 1990; Chevallier et al.,2004).

Little is currently known about the molecularevents underlying axon regeneration from brainstem orspinal neurons in salamanders. However, a potential rolefor fibroblast growth factor-2 (FGF2) is suggested by itsability to increase proliferation of neural progenitor cellsduring regeneration of the tail spinal cord (Zhang et al.,2000). Nevertheless, several studies suggest that regener-ation processes underlying spinal cord regeneration inthe amputated tail and in spinal cord transected morerostrally are different (see Chernoff et al., 2003). Thus,the role played by FGF2 in body spinal cord regenera-tion and locomotor recovery remains unclear.

For other systems, reports strongly suggest thatFGF2 is involved in locomotor recovery after spinalcord injury (SCI). Indeed, in adult rats, a significant andrapid up-regulation of FGF2 expression was observedboth above and below the epicenter of a compressioninjury (Follesa et al., 1994; Mocchetti et al., 1996; Zaiet al., 2005). The up-regulation in the distal spinal cordis thought to play a prominent role in motoneuron sur-vival below the injury and in partial recovery of loco-motor function over time (Teng et al., 1998). Further-more, FGF2 administration following a moderate or asevere SCI promotes hindlimb movement recovery inthe adult rat (Rabchevsky et al., 2000). More impor-

Contract grant sponsor: Region Aquitaine; Contract grant number:

20040301208N; Contract grant sponsor: Agence Universitaire de la Fran-

cophonie (AUF; to M.M.); Contract grant sponsor: Fondation Singer-

Polignac (to M.M.).

*Correspondence to: Jean-Marie Cabelguen, INSERM U 862, Neuro-

centre Magendie, Pathophysiology of Spinal Networks, 146 rue Leo Saignat,

33077 Bordeaux Cedex, France. E-mail: [email protected]

Received 20 November 2007; Revised 2 April 2008; Accepted 17 April

2008

Published online 14 July 2008 in Wiley InterScience (www.

interscience.wiley.com). DOI: 10.1002/jnr.21776

Journal of Neuroscience Research 86:3348–3358 (2008)

' 2008 Wiley-Liss, Inc.

tantly, FGF2 mRNA expression in the distal cord wasshown to be up-regulated after a complete spinal cordtransection in infant (P11) rats, whereas this did notoccur in adult rats (Qi et al., 2003). It was suggestedthat this up-regulation provides the neurotrophic supportfor previously reported axonal regeneration and recon-nection of some descending tracts in infant rats (Waka-bayashi et al., 2001). Interestingly, FGF2 was shown topromote neurite outgrowth of bulbospinal neuronsmediating the locomotor drive in young (E8) chicken(Pataky et al., 2000).

This prompted us to study the role of FGF2 in thefunctional regeneration of axotomized bulbospinal neuronsin the spinally transected salamander. A prerequisite was todescribe FGF2 distribution and cellular source, and theirchanges at different times after a chronic transection.

Therefore, the first objective of our study was todocument FGF2 mRNA expression pattern along thebrainstem-spinal cord of intact salamanders using in situhybridization. The second was to investigate possiblechanges in this pattern in animals unable to display hin-dlimb locomotor movements (1–2 weeks postopera-tively), and animals having fully recovered hindlimblocomotor activity (15 weeks postoperatively) after a spi-nal cord transection at the midtrunk level (Davis et al.,1990; Chevallier et al., 2004). We also aimed at identify-ing FGF2 mRNA cellular sources in intact and spinallytransected animals by immunohistochemistry.

Our data provide the first comprehensive view ofFGF expression along the neural axis of a salamandermodel of spinal cord lesion. They revealed a decreasingrostrocaudal gradient in FGF2 mRNA expression alongthe brainstem-spinal cord in intact animals and its long-lasting up-regulation in response to spinal transection atthe midtrunk level, both in brainstem and in spinal cordbelow the injury. Finally, double immunolabelingshowed that FGF2 was up-regulated in neuroglial, pre-sumably undifferentiated, cells. Therefore, we proposethat FGF2 may be involved in cell proliferation and/orneuronal differentiation after body spinal cord transectionin salamander and could thus play an important role infunctional recovery of locomotion after spinal lesion.

MATERIALS AND METHODS

Animals

Experiments were carried out on 25 Urodele amphib-ians (Pleurodeles waltlii) with snout vent length (SVL) rangingfrom 35 to 45 mm. Animals obtained from Blades BiologicalLtd. (Kent, United Kingdom) were kept in aquaria at 198C incirculating filtered tap water and fed twice per week with Chi-ronomes larvae. Water was completely renewed twice per week.Surgical procedures and handling and housing of the animalswere in accordance with protocols approved by the INSERMEthics Committee and conformed to NIH guidelines.

Spinalization

Surgery was performed under aseptic conditions withanimals under general anesthesia. Anesthesia was induced by

immersion in a 0.1% aqueous solution of tricaine methanesul-fonate (MS-222; Sigma, Saint Quentin Fallavier, France). Spi-nal cord was exposed from the dorsal midline at the 0.70SVL level by removing a small dorsal part of the twelfth ver-tebra. The dura was cut open, and the spinal cord was com-pletely transected between spinal segments 12 and 13 with apair of microdissecting scissors. The completeness of transec-tion was confirmed by retraction (about 1 mm) and visualinspection of the two spinal cord stumps. Once transectionwas verified, the wound was sutured, and spinally transectedanimals were kept in individual empty tanks, the bottom ofwhich was covered with fresh wet paper. When wound heal-ing was complete (8–10 days postoperatively), spinally trans-ected animals were transferred back to their aquaria torecover. The bladder was pressed manually two or three timeseach week for the first month. Animals did not seem to be indistress, as judged from the absence of mucus secretion oraggressive behavior. In sham-operated animals, spinal cordswere exposed by laminectomy but were not transected, andthereafter the animals were treated exactly like spinally trans-ected ones.

In Situ Hybridization

Normal (n 5 5), sham-operated (n 5 10, 5 for eachtime point), and spinally transected (n 5 10, 5 for eachtime point) animals were anesthetized as previouslydescribed. The dorsal half of animals was cut off, discardingthe whole viscera and tail. Skin and muscles were removedto expose the vertebral column in a cold oxygenated Uro-dele Ringer’s solution (in mM: NaCl, 130; KCl, 2.1;CaCl2, 2.6; MgCl2, 0.2; HEPES, 4; D-glucose, 5;NaHCO3, 1; pH 7.4; at 48C) as previously tested (Cheval-lier et al., 2004). Brainstem and spinal cord were exposedfrom the dorsal midline by craniectomy and laminectomy,respectively. The dura was cut open, and brainstem andbody spinal cord were completely taken out with a pair ofmicrodissecting scissors. Spinal cord was then divided intothree segments, namely, SC1, SC2, and SC3, correspond-ing, respectively, to two prelesional and one sublesionalportion of trunk spinal cord. Brainstem (BS) and the threespinal cord regions were immediately but separately frozenin embedding medium (Tissue-Tek, Sakura, Cergy-Pon-toise, France) without fixation by immersion in –508C iso-pentane (Merck, Darmstadt, Germany).

Oligonucleotide probes. Two oligonucleotideprobes (Eurogentech, Seraing, Belgium) were used in thisstudy for in situ detection of FGF2 mRNA. Oligonucleotideswere chosen in regions presenting few homologies withsequences of related mRNAs, and they were checked againstthe GenBank database. Fifty-mer oligonucleotides weredesigned as follows from published gene sequences (Zhanget al., 2000): 50GTTGATCCGCAGAAAGAAGCCCCCGTTCTTGCAGTACAGCCTCTTGGGTC30 and 50TTCAGCGCCATAAGCCTGCCGTCATCCTTCATAGCGAGATAGCGGTTTGC30.

Oligonucleotides were labelled as previously described(Landry et al., 2000) at the 30-end using terminal deoxynu-cleotidyl transferase (TdT; Amersham, Amersham, United

FGF2 Expression After Spinal Transection 3349

Journal of Neuroscience Research

Kingdom) in cobalt-containing buffer with 35S-dATP (Amer-sham) to a specific activity of 1–4 3 109 cpm/lg and purifiedby ethanol precipitation.

In situ hybridization procedure. Spinal cord andbrainstem sections (265 section/region/animal/experiment)were cut at 14 lm thickness using a cryostat (Microm, Hei-delberg, Germany) and thaw-mounted onto Super-FrostGold slides (CML, Nemours, France). Sections were thenprocessed as described earlier (Landry et al., 2000). In brief,tissue sections were air dried and incubated for 16 hr at 428Cwith 0.5 ng of each of the radioactively labelled probes,which were diluted in a hybridization solution containing50% deionized formamide (Sigma, Bordeaux, France), 43standard saline citrate (SSC; 13 SSC 5 0.15 M NaCl, 0.015M NaCitrate), 13 Denhardt’s solution [0.02% bovine serumalbumin, 0.02% Ficoll (Sigma), 0.02% polyvinylpyrrolidone,0.02 M NaPO4, pH 7.0], 1% N-lauroylsarcosine, 10% dex-tran sulfate (Sigma), 500 mg/l denatured salmon testis DNA(Sigma), and 200 mM dithiothreitol (Sigma). After hybridiza-tion, sections were rinsed in 13 SSC, and four times for 15min each at 558C, followed by 30 min at room temperature(RT). Radioactively labelled sections were then dipped intoIlford K5 nuclear emulsion (Ilford, Mobberly, Cheshire,United Kingdom) diluted 1:1 with distilled water, exposedfor 6 weeks, developed in Kodak D19 for 3.5 min, and fixedin Kodak 3000 for 10 min. Before mounting in glycerol, sec-tions were counterstained with 0.25% cresyl violet acetate(pH 4; Sigma).

Immunohistochemistry

Animals (n 5 5/experiment) were anesthetized as pre-viously described and perfusion fixed via the ascending aortawith 10 ml Urodele Ringer’s solution, followed by 10 ml ofan ice-cold fixative containing 4% paraformaldehyde and0.2% picric acid in 0.1 M phosphate buffer (PB; pH 7.4;Pease, 1962; Zamboni and De Martino, 1967). Brainstemand spinal cord were dissected out as previously described,immersed in the same fixative for 1 hr, and rinsed for atleast 24 hr in 0.1 M PB (pH 7.4) containing 15% sucroseand 0.02% sodium azide (Sigma). They were then frozenand sectioned (see above). Sections were air dried, rinsedwith Tris-buffered saline (TBS; 0.1 M Tris-HCl, 0.15 MNaCl, pH 7.5), and triply labelled with rabbit anti-FGF2(Chemicon, Temecula, CA), mouse anti-NeuN (Abcam,Paris, France), and rabbit anti-GFAP (Dako SA, Trappes,France) as follows to avoid cross-reactivity between same-species-raised antibodies.

Tissue sections were preincubated with 0.01 M phos-phate-buffered saline (PBS) for 20 min, and then endogenousperoxidases were blocked with H2O2 (1%) in PBS for 10min at RT. A 10-min rinse was applied using TBS contain-ing 0.05% Tween 20 (TNT), followed by a 30-min incuba-tion with 1% (w/v) BSA (Sigma) in TNT (TNT/BSA). Sec-tions were incubated overnight at 48C with rabbit anti-FGF2primary antibody (1:1,000 in TNT/BSA). After rinsing inTNT, sections were incubated with anti-rabbit horseradishperoxidase (HRP)-conjugated secondary antibody (Dako,Carpinteria, CA) for 1 hr at RT. Peroxidase activity was

then developed with fluorescein-conjugated tyramide(Perkin-Elmer, Boston, MA) according to the manufacturer’sinstructions, as previously described (Landry et al., 2004).Cross-reaction with the subsequent detection was avoided byusing a blocking reagent (Dako; Landry et al., 2004). Afteradditional rinsing, sections were incubated overnight at 48Cwith mouse anti-NeuN (1:100 in TNT/BSA) and rabbitanti-GFAP (1:500 in TNT/BSA) and rinsed in TNT. GFAPwas detected by incubating sections with Alexa 647-conju-gated donkey anti-rabbit secondary antibody (Invitrogen,Cergy Pontoise, France; 1:500 in TNB) for 2 hr at RT. Im-munohistochemical detection of NeuN was visualized withbiotinylated anti-mouse antibody (Vector, Burlingame, CA;1:100 in TNB) and Alexa 568-conjugated streptavidin (Invi-trogen; 1:500 in TNB) for 2 hr at RT. Sections were thenrinsed for 3 3 10 min in TNT and finally mounted on cov-erslips and viewed.

Imaging

Slides were examined and brightfield light microscopymicrographs were taken using a Zeiss Axiophot 2 microscope(Zeiss, Jena, Germany). Triple immunostainings were analyzedwith a Leica DMR PCS SP2 AOBS confocal microscope(Leica, Heidelberg, Germany) using a 340 or a 363 oil-immersion lens. In all cases, scans were carried out sequen-tially with the 488 nm, 568 nm, and 647 nm lines of the laser.Only single optical sections were used for illustrations. Digitalimages were optimized for image resolution (300 dpi final re-solution), brightness, and contrast in Adobe Photoshop 6.0(Adobe System, San Jose, CA).

Data Analyses

Comparisons between groups (5 animals each) weremade on sections treated together on the same slides underidentical conditions. Nonadjacent sections from SC2-3 and BSwere randomly chosen after the in situ hybridization proce-dure. The number of labelled cells and the labelling intensitywere quantified as previously described (Landry et al., 2000).Cellular profiles containing three times more grains than meanbackground grain densities were considered labelled. Manualoutlines were made for cell profiles with clearly distinguishablenuclei. They were delineated only based on staining, notshape, size, or other measurable quantities. The number of sil-ver grains per cresyl violet-counterstained cell was semiauto-matically counted in MetaMorph Offline 6.1 software (Uni-versal Imaging Corporation). Data were expressed as graindensity per square micrometer 6 SEM after calibrating thephotomicrographs. In each experiment, transmitted light pho-tomicrographs were taken for at least 500 fields (700 lm2

each).Data were imported into a spreadsheet program (Sigma

Plot software; Jandel Scientific) that calculated and graphedthe density of FGF2 mRNAs expression and the number oflabelled cells. Data were compared by using one-wayANOVA tests and processed with the use of standard statisticalanalyses (Sigma Stat software). Differences were considered tobe significant at P < 0.05.

3350 Moftah et al.


RESULTS

The distribution of FGF2 mRNA and the in situhybridization labelling intensity are presented in intact(Fig. 1a–c,g–i) and in spinally transected (Fig. 1d–f,j–l)animals. The in situ hybridization signal has been quan-

tified and allows us to compare the expression of FGF2transcript among the different segments of the neuralaxis in intact animals (Fig. 2) but also to analyzechanges after spinal transection both in brainstem(Fig. 3) and in spinal cord (Figs. 4, 5). Finally, Figure 6

Fig. 1. Localization of FGF2 mRNA in cross-sections of brainstemand spinal cord in intact and spinally transected animals. a–l: Local-ization of FGF2 mRNA in brainstem (BS; a–f) and posterior spinalcord (SC3; g–l) using in situ hybridization in intact animals (a–c,g–i),and at 15 weeks after spinal cord transection (d–f,j–l). b,e,h,k Aremagnifications of boxed areas in a,d,g,j, respectively. c,f,i,l Are exam-ples of the Metamorph-assisted procedure to quantify in situ hybrid-

ization. All cells are numbered, but only labelled cells are delineated(arrow); nonlabelled cells (arrowhead) are not taken into account forgrain counting. Counted silver grains are threshold and highlightedin color. Labelling intensity in individual cells is more intense afterspinal transection as shown in f vs. c and l vs. i. V, ventricle; CC,central canal; Cont, control; spinal, spinally transected. Scale bars5 100 lm in a,d,g,j; 20 lm in b,c,e,f,h,i,k,l.



provides information on the cell types that expressFGF2 after lesion.

FGF2 Expression in Brainstem and SpinalCord in Normal Animals

Expression of FGF2 mRNA was assessed in thebrainstem and in three different (anterior, middle, andposterior) portions of body spinal cord of the juvenilePleurodeles waltlii by using radioactive in situ hybridiza-tion. Brainstem labelled-cells (‘‘FGF21 cells’’) werelocated mainly laterally, flanking the fourth ventricle(Fig. 1). Cells in the middle part of the brainstem, liningthe fourth ventricle, showed poor FGF2 mRNA. In par-allel, in the spinal cord, FGF2 mRNA was not observedin the ependymal cells lining the central canal, but it waspresent in ventrolaterally located cells (Fig. 1g). In bothstructures, silver grains were seen overlying large profilesof spinal cell bodies (Fig. 1b,h). Although this type ofcells was proposed to be a motor neuron on the basis ofthe ventrolateral location and morphology (Zhang et al.,2000), the exact nature remains to be established in ourconditions. The density of silver grain appeared moder-ate in brainstem (Fig. 1c) and spinal cord (Fig. 1i).

To provide more detailed information on thenumber of labelled cell profiles and on the cellular inten-sity of the labelling, FGF2 mRNA was then quantifiedand compared between the brainstem and the differentportions of the spinal cord. FGF2 mRNA grain densityreflects the cellular labelling intensity. As shown in Fig-ure 2 (left), it was gradually declining along the neuralaxis. Indeed, the highest FGF2 mRNA grain density wasin brainstem (0.0752 6 0.0027 grains/lm2) and wasapproximately twofold that of the posterior part of bodyspinal cord (SC3: 0.0371 6 0.0034 grains/lm2). FGF2mRNA grain densities in the anterior (SC1: 0.0626 60.0057 grains/lm2) and middle (SC2: 0.0532 6 0.0032grains/lm2) portions of the spinal cord were betweenthose of the brainstem and SC3. Note that there was nosignificant (P 5 0.220) difference between FGF2

mRNA grain densities in SC1 and SC2 portions of thebody spinal cord.

To assess whether FGF2 mRNA distribution patternalong the neural axis reflected differences in the numberof FGF21 cells, we counted those. In the brainstem,approximately 70% (n 5 208 6 17 cells) of the total pop-ulation of cells (n 5 288 6 24 cells) was FGF21 cells,whereas, in the posterior spinal cord, FGF21 cells consti-tuted about 17% (n 5 30 6 2 cells) of the total populationof cells (n 5 186 6 5 cells; Fig. 2, right). The number ofFGF21 cells in the anterior and middle portions of thespinal cord did not differ significantly (P 5 0.389) but wassignificantly lower than that of FGF21 cells in brainstem(P 5 0.008) and higher than the number of FGF21 cellsin posterior spinal cord (P 5 0.011).

Altogether, our results evidence a decreasing gradi-ent in FGF2 mRNA along the neural axis of the sala-mander. They also reveal that the FGF2 mRNA patternresulted from a greater number of FGF21 cells and ahigher level of FGF2 mRNA in individual FGF21 cellsin brainstem than in body spinal cord.

FGF2 Expression in Brainstem and SpinalCord in Spinally Transected Animals

FGF2 mRNA density was assessed in spinally trans-ected animals, which were unable to display hindlimblocomotor movements (1–2 weeks postoperatively) andin animals that have fully recovered hindlimb locomotormovements (15 weeks postoperatively). This timing hasbeen chosen on the basis of our previous electrophysio-logical study on locomotor recovery in spinally trans-ected salamanders (Chevallier et al., 2004). It is worthnoting that the locomotor recovery 15 weeks after cordtransection is associated with regeneration of descendingpathways through the transection and reinnervation ofthe spinal cord below the lesion (Piatt, 1955; Daviset al., 1990; Chevallier et al., 2004).

The distribution of FGF2 mRNA was first assessedin tissue sections of brainstem (15 weeks postoperatively,

Fig. 2. FGF2 mRNA expression along the brainstem and spinal cordin intact animals. Left panel: FGF2 mRNA grain density (numberof grains/lm2). Right panel: Number of FGF21 cells. Note thatboth the grain density and the number of FGF21 cells are highest inthe brainstem (BS) and lowest in the caudal spinal cord (SC3). Note

also that these two parameters were not significantly different in theanterior (SC1) and the midlle (SC2) portions of the spinal cord. Sym-bols above each bar indicate the statistical significance vs. the previ-ous bar (*0.01 < P � 0.05, **0.001 < P � 0.01, ***P � 0.001).

3352 Moftah et al.


Fig. 1d–f) and spinal cord (15 weeks postoperatively,Fig. 1j–l). Positive and negative cell profiles were found,the former being more heavily labelled than in intactanimals, both in brainstem (Fig. 1e,f) and in spinal cord(Fig. 1k,l).

In brainstems of sham-operated animals, FGF2mRNA grain density and the number of FGF21 cellsmeasured 1–2 weeks after surgery were, respectively,slightly higher than (P 5 0.013) and close to (P 50.064) control (Fig. 3A,B, white bars). By contrast,FGF2 mRNA grain density in the brainstem of spinallytransected animals was, at that time, significantly (P <0.001) more elevated than in the control condition. Onthe other hand, the number of FGF21 cells was notice-ably (P 5 0.008) reduced (Fig. 3A,B, white bars). Thisresult suggested an up-regulation of the expression ofFGF2 mRNA in individual brainstem FGF21 cells, 1–2weeks after spinalization.

Compared with the control condition, FGF2 mRNAgrain density was significantly (P < 0.001) increased 15weeks after surgery, both in sham-operated and in spinallytransected animals (Fig. 3A, gray bars). Note, however, thatFGF2 mRNA grain densities at 1–2 weeks and 15 weekspostoperatively in spinally transected animals were not sig-nificantly (P 5 0.165) different. Furthermore, the numberof FGF21 cells was still not significantly (P 5 0.690)altered in sham-operated animals, although it was signifi-cantly increased (P 5 0.046) to be close to control in spi-nally transected ones (Fig. 3B, gray bars).

Altogether these results suggested that, 1–2 weeksafter spinalization, FGF2 mRNA expression in brainstemcells was up-regulated and remained so at 15 weeks afterspinal cord transection. The long-lasting postinjury

increase in FGF2 mRNA grain density not only results,at least partially, from the cord transection per se butalso reflects the effect of laminectomy as suggested bythe increase seen in sham-operated animals. By contrast,because the number of FGF21 cells was, at any postsur-gery time, close to control in sham-operated animals,the observed changes in the number of FGF21 cells inspinally transected animals were very likely a result ofthe cord injury per se. Hence, the important transientreduction of the number of FGF21 cells after spinaliza-tion could reflect a death of some of the descendingbrainstem neurons projecting below the level of spinalcord transection.

As for body spinal cord, we have focused ourinvestigation on SC2 and SC3, which are the spinalcord stumps flanking the transection site. FGF2 mRNAgrain density in SC2 remained close to the control level1–2 weeks after surgery, both in sham-operated and inspinally transected animals (Fig. 4A, white bars). Insham-operated animals, FGF2 mRNA grain densitymeasured 15 weeks after surgery (0.0670 6 0.0498grains/lm2) was not significantly (P 5 0.996) differentfrom control (Fig. 4A). By contrast, grain density meas-ured in spinally transected animals (0.0770 6 0.0596grains/lm2) was slightly, but significantly (P 5 0.024),larger than in control conditions (Fig. 4A). Figure 4Bfurther shows that, 1–2 weeks after surgery, the numberof FGF21 cells significantly (P 5 0.004) increased insham-operated animals, whereas it dramatically (P <0.001) decreased in spinally transected animals. Note,however, that 15 weeks after surgery, the number ofFGF21 cells was not significantly (P > 0.1) differentfrom control in either preparation. These results sug-

Fig. 3. FGF2 mRNA expression in the brainstem of spinally trans-ected animals. A: FGF2 mRNA grain density. The black bar repre-sents FGF2 mRNA grain density measured in intact animals. Insham-operated animals, the level of grain density was close to control1–2 weeks postoperatively (sham, white bar), although a highly sig-nificant increase appeared 15 weeks postoperatively (sham, gray bar).Spinally transected animals had almost doubled grain density levels 1–2 weeks postoperatively (spinal, white bar), and this increase was stillpresent 15 weeks postoperatively (spinal, gray bar). B: Number of

FGF21 cells. The black bar represents the number of FGF21 cellsmeasured in intact animals. No significant change was observed insham-operated animals (sham, white and gray bars). By contrast, inspinally transected animals, the FGF21 cells number was greatlydecreased 1–2 weeks postoperatively (spinal, white bar), whereas itwas increased 15 weeks postoperatively (spinal, gray bar). Symbolsabove each bar indicate the statistical significance vs. the intact ani-mal’s level (*0.01 < P � 0.05, **0.001 < P � 0.01, ***P � 0.001).



gested that postinjury changes in FGF2 mRNA distribu-tion pattern in SC2 were basically similar to, albeit lessmarked than, what we have observed in brainstem.

Sham-operated and spinally transected animalsshowed a large increase in FGF2 mRNA grain densityin SC3 1–2 weeks after surgery (Fig. 5A, white bars).The increase was preserved, although at a lower level,15 weeks after surgery (Fig. 5A, gray bars). Interestingly,FGF2 mRNA grain density was always significantly

(P < 0.001) higher in spinally transected than in sham-operated animals. This suggested that the spinal transec-tion was by itself responsible for at least a part of theincrease in FGF2 mRNA grain density observed in spi-nally transected animals.

The number of FGF21 cells in sham-operated ani-mals, measured 1–2 weeks or 15 weeks after surgery,was significantly (P < 0.001) higher than in the controlcondition, although it was not significantly (P 5 0.257)

Fig. 4. FGF2 mRNA expression in the spinal cord stump (SC2) inspinally transected animals. A: FGF2 mRNA grain density. Samerepresentation and same conventions as in Figure 3A. Both at 1–2weeks (white bars) and at 15 weeks (gray bars) postoperatively, thelevel of grain density in sham-operated (sham) or spinally transected(spinal) animals was not significantly different from that in intact ani-mals. B: Number of FGF21 cells. Same representation and sameconventions as in Figure 3B. In sham-operated animals, the number

of FGF21 cells was significantly increased 1–2 weeks postoperatively(sham, white bar), but it was close to control 15 weeks postopera-tively (sham, gray bar). In spinally transected animals, the number ofFGF21 cells was strongly decreased 1–2 weeks postoperatively (spi-nal, white bar) but was close to control 15 weeks postoperatively(spinal, gray bar). Symbols above each bar indicate the statistical sig-nificance vs. the intact animal’s level (*0.01 < P � 0.05, **0.001 <P � 0.01, ***P � 0.001).

Fig. 5. FGF2 mRNA expression in the spinal cord stump (SC3) inspinally transected animals. A: FGF2 mRNA grain density. Samerepresentation and same conventions as in Figure 3A. Sham-operatedanimals displayed a significant increase in grain density 1–2 weeksand 15 weeks postoperatively (sham, white and gray bars). Note,however, the slight decrease 15 weeks postoperatively. Spinally trans-ected animals had almost 2.5-fold the grain density levels 1–2 weekspostoperatively (spinal, white bar). This decreased to 1.25-fold byweek 15 postoperatively (spinal, gray bar). B: Number of FGF21

cells. Same representation and same conventions as in Figure 3B. Insham-operated animals, the number of FGF21 cells was significantlyincreased, both 1–2 weeks and 15 weeks postoperatively (sham,white and gray bars). Note, however, the slight decrease 15 weekspostoperatively in spinally transected animals; either 1–2 weeks or 15weeks postoperatively, the number of FGF21 cells was close to con-trol. Symbols above each bar indicate the statistical significance withthe intact animal’s level (**0.001 < P � 0.01, ***P � 0.001).

3354 Moftah et al.


different from control in spinally transected animals (Fig.5B). This suggested that the increase in FGF21 cellnumbers, which should be induced by laminectomy, wasprevented by spinalization.

Altogether these results suggested that FGF2mRNA expression was increased in individual cellswithin SC3. They further suggested that spinalizationlimited the number of FGF21 cells in spinal cord belowthe injury.

Nature of FGF21 Cells

To examine whether the above-mentioned increasein mRNA after spinal cord injury leads to increasingprotein synthesis, we performed immunohistochemicalanalyses on the same sets of experiments. Only fewFGF21 cells were visualized by immunohistochemistry

in brainstem (Fig. 6A–D) and posterior spinal cord (Fig.6I–L) of intact animals. Codetection of FGF2 with neu-ronal and glial markers indicated that FGF2 was presentin both cell types in the brainstem (Fig. 6D) but essen-tially in neuronal cells in the spinal cord (Fig. 6L), whichconfirms the probability of these FGF21 cells beingmotoneurons. At 15 weeks after spinal lesion, the immu-nolabeling of FGF2 was markedly increased, both inbrainstem (Fig. 6E–H) and in sublesional spinal cord(Fig. 6M–P). After lesion, FGF2 was seen in neuronalcells in brainstem, flanking the fourth ventricle, the samecells in which we saw the increase in mRNA in FGF21

cells. However, a majority of FGF2-immunopositivecells expressed both markers for neuron and glia inbrainstem and spinal cord. Virtually all FGF2-immuno-positive cells were neuroglial in the spinal cord (Fig.6P). FGF2-containing cells lay mainly in the periphery

Fig. 6. Immunohistochemical analysis of FGF2 distribution in thebrainstem and the distal (sublesional) spinal stump (SC3) of spinallytransected animals. FGF2-immunoreactive (-IR) cells are shown ingreen (A,E,I,M), NeuN-IR cells in red (B,F,J,N), GFAP-IR cellsin blue (C,G,K,O). D,H,L,P: Overlays. Brainstem: A–D: Sham-operated animals; E–H: spinally transected animals 15 weeks postop-eratively. In sham-operated animals (A), FGF2 was weakly expressed.It was seen in a few glial cells (arrowheads in A,C,D) and neurons(arrows in A,B,D). FGF2 expression increased after 15 weeks of spi-

nal cord transection. It was obvious mainly in neurons (single arrowsin E,F,H) but also in some neuroglial cells lining the fourth ventricle(double arrows in E–H). Distal spinal stump: I–L: Sham-operatedanimals; M–P: spinally transected animals 15 weeks postoperatively.In sham-operated animals, FGF2 was vaguely seen in neurons (singlearrows in I,J,L). It increased at 15 weeks postoperatively, to becomeprominent in ependymal cells (double arrows in M–P). Ependymalcells, which were neuroglial cells, expressed both NeuN and GFAP(double arrows in N,O). Scale bar 5 40 lm.



of the fourth ventricle in brainstem and around the ep-endymal canal in spinal cord (Fig. 6E,M, respectively).

DISCUSSION

Intact Animals

Our study using in situ hybridization indicated adecreasing rostrocaudal gradient of FGF2 mRNAexpression along the brainstem and body spinal cord inintact salamanders. We further provide evidence that spi-nal FGF21 cells were located ventrolaterally and thatspinal FGF2-immunopositive cells were essentially neu-rons. Zhang and colleagues (2000) previously showedthat FGF2 mRNA expression and FGF2 immunoreac-tivity were localized in motoneurons of the intact Uro-dele tail cord. Moreover, Grothe and Wewetzer (1996)detected FGF2 immunoreactivity in motoneurons of therat body spinal cord. Altogether, these data suggest that,in the intact Urodele body cord, FGF2 is expressedmainly in motoneurons. Notwithstanding, this observa-tion requires further confirmation.

Our result also revealed that brainstem FGF21 cellslined the fourth ventricle and that they could be neuro-nal or glial cells. Interestingly, this brainstem region con-tains the body of spinally projecting neurons whoseaxons can regrow after axotomy, when performed at themidtrunk level (Davis et al., 1990; Chevallier et al.,2004). Our previous electrophysiological study has dem-onstrated, as well, that regrowth of these presumablyreticulospinal axons to the level of the spinal cord belowthe injury is responsible for the recovery of locomotionin the spinal salamander (Chevallier et al., 2004). More-over, our anterior in vitro studies showed that exoge-nous application of FGF2 induced brainstem neuriteoutgrowth (Moftah, 2007). It is worth noting that FGF2mRNA and protein are expressed in some brainstemneurons in the adult rat (Gonzalez et al., 1995).

The wide distribution of FGF21 cells in the brain-stem and body spinal cord of the salamander is in agree-ment with the idea that FGF2 serves multiple functions,depending both on the type of FGF receptors (FGFRs)stimulated and on their cellular localization (Grothe andWewetzer, 1996). Mapping of FGFRs along the brain-stem and body spinal cord is still lacking, so it seems tobe premature to discuss these functions further. How-ever, it is worth noting that FGFR1 and FGFR4 showdistinct patterns of expression in the tail spinal cord ofPleurodeles waltlii, and this has been related to distinctroles for these receptors during regeneration of the tail(Zhang et al., 2002).

Other growth factors such as BDNF could play animportant role as well in the process of spinal cordregeneration. However, in this study, we focused onFGF2 because it is known to support proliferation andmultipotentiality of progenitor cells, whereas BDNF cre-ates a neurogenic environment allowing for neuronaldifferentiation of adult progenitor cells (Chen et al.,2007). In a context of regeneration, it seems that FGF2might be acting at an earlier stage than BDNF in order

to allow injured spinal cord to regain its original struc-ture and eventually to recover its function after lesion.

Spinally Transected Animals

Our data first revealed short- and long term up-regulation of FGF2 mRNA expression in brainstem inresponse to a midtrunk spinal cord transection. Thepostinjury up-regulation of FGF2 mRNA expression inbrainstem could facilitate the high regenerative capacityof reticulospinal neurons and locomotor recovery in spi-nally transected salamanders (Davis et al., 1990; Cheval-lier et al., 2004). It could also facilitate the regenerationand functional reconnection of damaged spinal pathwaysthat ascend throughout the brainstem to the diencepha-lon (Munoz et al., 1997).

Our data further indicated that the distal spinalcord stump, the target for the regenerated descendingfibers, also responded by short- and long term up-regu-lation of FGF2 mRNA expression. This up-regulationcould facilitate the reestablishment of the functionalinnervation of spinal locomotor networks below theinjury by regenerated brainstem and intraspinal descend-ing neurons. In line with this assumption, the expressionof FGF2 mRNA and protein was also shown to be up-regulated in the distal cord after a compression injury ora complete spinal cord transection in infant rats, but notin adult rats (Follesa et al., 1994; Qi et al., 2003). Thisdifferential expression of FGF2 has been related to thehigher regenerative capacity of some descending tractsand functional recovery in spinally transected infant rats(Wakabayashi et al., 2001).

Furthermore, the up-regulation of FGF2 mRNAaround the transection site observed here could facilitatethe regeneration of damaged axons not only from de-scending but also from ascending intraspinal neurons.Regeneration of damaged axons from intraspinal neuronsis thought to play an important role in the reestablish-ment of the intersegmental locomotor coordinationacross a spinal cord transection in the salamander (Che-vallier et al., 2004) and in the lamprey (Cohen et al.,1986, 1988; McClellan, 1990). Interestingly, the densityof FGF2-immunoreactive cells following a contusive spi-nal cord injury in adult rats is the greatest a few milli-meters rostral to the lesion epicenter (Zai et al., 2005).Further investigations are necessary to test whether asimilar postinjury pattern of FGF2-immunoreactive cellsis present in the salamander.

In the present study, we observed that FGF2mRNA expression in the distal spinal cord firstincreased, thereafter gradually decreasing to remain sig-nificantly higher than control 15 weeks after injury. Asimilar temporal pattern of expression, albeit over 3weeks, was reported in the spinal cord of regeneratingtails following amputation (Zhang et al., 2000). Thetemporal course of FGF2 mRNA expression observedhere could be related to the period of progressive recov-ery of hindlimb locomotion in spinally transected ani-mals (Chevallier et al., 2004). However, FGF2 mRNA

3356 Moftah et al.


expression level remained elevated after the recovery ofhindlimb locomotor movements was complete. Thissuggests that FGF2 continued to play a role, whichcould be a remodelling of locomotor networks and de-scending controls in order to improve further the recov-ered hindlimb locomotor movements. Still adult axolotsrequire about 1 year for a complete replacement of de-scending pathways in the regenerated lumbar spinal cord(Clarke et al., 1988).

A previous study in Pleurodeles waltlii has shownthat the FGF2 mRNA level is up-regulated in regenerat-ing spinal cord following tail amputation (Zhang et al.,2000). Therefore, we can conclude that FGF2 up-regu-lation can be triggered in response to spinal cord injury,no matter where the lesion takes place.

The spatiotemporal distribution of FGF2 protein af-ter spinal transection was not examined in the presentstudy. Nonetheless, data from the regenerating spinal cordafter tail amputation in Pleurodeles waltlii suggest that theFGF2 protein pattern of expression follows that of theFGF2 mRNA expression (Zhang et al., 2000, 2002).

Our immunohistochemical data revealed thatFGF2-immunopositive cells in the distal spinal cordstump of long-term spinally transected animals wereessentially neuroglial cells located around the ependymalcanal. This finding is in line with the view that FGF2may play an important role in the proliferation of neuralprogenitor cells, which have the potential to regeneratespinal cord (Zhang et al., 2000). The role of FGF2 inbrainstem remains to be elucidated, however. Interest-ingly, FGF2 promotes both survival and neurite out-growth of bulbospinal neurons in vitro in chicken(Pataky et al., 2000).

FGFRs, which mediated the proliferation of neuralprogenitors in brainstem and body spinal cord, are stillunknown. It has been previously suggested that FGFR1activation is associated primarily with FGF2-inducedproliferation of neural progenitors in the regeneratingspinal cord following tail amputation (Zhang et al.,2002). That FGFR1 is also involved in body spinal cordregeneration is likely, but this remains to be established.

In conclusion, we have provided evidence that acomplete spinal cord transection at a midtrunk level inadult salamanders results in a pronounced FGF2 mRNAup-regulation for a long period in both brainstem andspinal cord. The up-regulation span corresponds to theinterval during which regeneration of damaged axonsand recovery of hindlimb locomotion take place afterspinal lesion. Therefore, further studies identifying theFGFRs involved, their signalling systems, and their cel-lular distribution will provide new therapeuticapproaches for human spinal cord impairment.

REFERENCES

Chen K, Henry RA, Hughes SM, Connor B. 2007. Creating a neuro-

genic environment: the role of BDNF and FGF2. Mol Cell Neurosci

36:108–120.

Chernoff EA, Stocum DL, Nye HL, Cameron JA. 2003. Urodele spinal

cord regeneration and related processes. Dev Dyn 226:295–

307.

Chevallier S, Landry M, Nagy F, Cabelguen JM. 2004. Recovery of

bimodal locomotion in the spinally transected salamander, Pleurodeles

waltlii. Eur J Neurosci 20:1995–2007.

Clarke JD, Alexander R, Holder N. 1988. Regeneration of descending

axons in the spinal cord of the axolotl. Neurosci Lett 89:1–6.

Cohen AH, Mackler SA, Selzer ME. 1986. Functional regeneration fol-

lowing spinal transection demonstrated in the isolated spinal cord of the

larval sea lamprey. Proc Natl Acad Sci U S A 83:2763–2766.

Cohen AH, Mackler SA, Selzer ME. 1988. Behavioral recovery follow-

ing spinal transection: functional regeneration in the lamprey CNS.

Trends Neurosci 11:227–231.

Davis BM, Ayers JL, Koran L, Carlson J, Anderson MC, Simpson SB Jr.

1990. Time course of salamander spinal cord regeneration and recovery

of swimming: HRP retrograde pathway tracing and kinematic analysis.

Exp Neurol 108:198–213.

Follesa P, Wrathall JR, Mocchetti I. 1994. Increased basic fibroblast

growth factor mRNA following contusive spinal cord injury. Brain Res

Mol Brain Res 22:1–8.

Gonzalez AM, Berry M, Maher PA, Logan A, Baird A. 1995. A compre-

hensive analysis of the distribution of FGF-2 and FGFR1 in the rat

brain. Brain Res 701:201–226.

Grothe C, Wewetzer K. 1996. Fibroblast growth factor and its implica-

tions for developing and regenerating neurons. Int J Dev Biol 40:403–

410.

Landry M, Holmberg K, Zhang X, Hokfelt T. 2000. Effect of axotomy

on expression of NPY, galanin, and NPY Y1 and Y2 receptors in dor-

sal root ganglia and the superior cervical ganglion studied with double-

labelling in situ hybridization and immunohistochemistry. Exp Neurol

162:361–384.

Landry M, Bouali-Benazzouz R, El Mestikawy S, Ravassard P, Nagy F.

2004. Expression of vesicular glutamate transporters in rat lumbar spinal

cord, with a note on dorsal root ganglia. J Comp Neurol 468:380–

394.

McClellan AD. 1990. Locomotor recovery in spinally transected lamprey:

role of functional regeneration of descending axons from brainstem

locomotor command neurons. Neuroscience 37:781–798.

Mocchetti I, Rabin SJ, Colangelo AM, Whittemore SR, Wrathall JR.

1996. Increased basic fibroblast growth factor expression following con-

tusive spinal cord injury. Exp Neurol 141:154–164.

Moftah MZ. 2007. Differential effects of FGF-2 on cell cultures from ju-

venile brainstem and body spinal cord of the Urodele Pleurodeles waltlii.

Egypt Zool 48:21–44.

Munoz A, Munoz M, Gonzalez A, ten Donkelaar HJ. 1997. Spinal

ascending pathways in amphibians: cells of origin and main targets. J

Comp Neurol 378:205–228.

Pataky DM, Borisoff JF, Fernandes KJL, Tetzlaff W, Steeves JD. 2000.

Fibroblast growth factor treatment produces differential effects on sur-

vival and neurite outgrowth from identified bulbospianl neurons in

vitro. Exp Neurol 163:357–372.

Pease PC. 1962. Buffered formaldehyde as a killing agent and primary

fixative for electron microscopy. Anat Rec 142:342.

Piatt J. 1955. Regeneration of the spinal cord in the salamander. J Exp

Zool 29:177–208.

Qi ML, Wakabayashi Y, Haro H, Shinomiya K. 2003. Changes in FGF-

2 expression in the distal spinal cord stump after complete cord transec-

tion: a comparison between infant and adult rats. Spine 28:1934–

1940.

Rabchevsky AG, Fugaccia I, Turner AF, Blades DA, Mattson MP, Scheff

SW. 2000. Basic fibroblast growth factor (bFGF) enhances functional

recovery following severe spinal cord injury to the rat. Exp Neurol

164:280–291.



Teng YD, Mocchetti I, Wrathall JR. 1998. Basic and acidic fibroblast

growth factors protect spinal motor neurones in vivo after experimental

spinal cord injury. Eur J Neurosci 10:798–802.

Wakabayashi Y, Komori H, Kawa-Uchi T, Mochida K, Takahashi M,

Qi M, Otake K, Shinomiya K. 2001. Functional recovery and regener-

ation of descending tracts in rats after spinal cord transection in infancy.

Spine 26:1215–1222.

Zai LJ, Yoo S, Wrathall JR. 2005. Increased growth factor expression and

cell proliferation after contusive spinal cord injury. Brain Res 1052:147–155.

Zamboni I, De Martino C. 1967. Buffered picric acid formalde-

hyde. A new rapid fixative for electron microscopy. J Cell Biol

35:148A.

Zhang F, Clarke JD, Ferretti P. 2000. FGF-2 up-regulation and prolifera-

tion of neural progenitors in the regenerating amphibian spinal cord in

vivo. Dev Biol 225:381–391.

Zhang F, Clarke J, Santos-Ruiz L, Ferretti P. 2002. Differential regula-

tion of fibroblast growth factor receptors in the regenerating amphibian

spinal cord in vivo. Neuroscience 114:837–848.

3358 Moftah et al.


fibroblast growth factor-2 mrna expression in the brainstem and spinal cord of normal and chronic...

Documents