Rewiring of hindlimb corticospinal neurons after spinal cord injury

Transcription

1 Rewiring of hindlim corticospinl neurons fter spinl cord injury Arko Ghosh 1,4, Florent Hiss 2, Esther Sydekum 3, Regul Schneider 1, Mirim Gullo 1, Mtthis T Wyss 2, Thoms Mueggler 3, Christof Bltes 3, Mrkus Rudin 2,3, Bruno Weer 2 & Mrtin E Schw 1 Little is known out the functionl role of xotomized corticl neurons tht survive spinl cord injury. Lrge thorcic spinl cord injuries in dult rts result in impirments of hindlim function. Using retrogrde trcers, we found tht xotomized corticospinl xons from the hindlim sensorimotor cortex sprouted in the cervicl spinl cord. Mpping of these neurons reveled the emergence of new forelim corticospinl projection from the rostrl prt of the former hindlim cortex. Voltge-sensitive dye (VSD) imging nd lood-oxygen-level dependent functionl mgnetic resonnce imging (BOLD fmri) reveled stle expnsion of the forelim sensory mp, covering in prticulr the former hindlim cortex contining the rewired neurons. Therefore, xotomised hindlim corticospinl neurons cn e incorported into the sensorimotor circuits of the unffected forelim. Lrge spinl cord injuries result in the disruption of descending nd scending sensorimotor circuits, ut most corticospinl neurons survive the injury 1 3. In intct nimls, the neurons of the corticospinl trct (CST) re essentil for the initition nd control of skilled movements. Although it hs een shown tht these neurons survive fter injury, their functionl role remins unknown. Spinl cord injuries cn induce sprouting of descending motor pthwys t severl levels of the dult CNS 4,5, nd corticospinl neurons hve een shown to sprout in the spinl cord of dult rodents nd primtes 6 1. Wht functions do these neurons exert nd wht inputs do these re-wired corticl neurons receive? Electrophysiologicl studies hve shown tht the somtosensory cortex cn e physiologiclly remodeled fter spinl cord injuries in primtes 11,12. Denervted corticl res cn e rectivted y expnding representtions of unffected ody prts, proly ccompnied y remodeling of sucorticl nd corticl connectivity. In dult rts, it is uncler whether sensory representtions reorgnize fter spinl cord injury nd to wht extent ntomicl rewiring occurs. A thorcic ilterl dorsl spinl cord hemisection in the dult rt results in hindlim presis. In this injury, the hindlim cortex is deprived of tctile nd proprioceptive inputs, nd corticospinl xons originting from this cortex re xotomized. Interestingly, fter the thorcic injury in rodents, some corticospinl xons originting from the hindlim cortex were seen to sprout into the cervicl spinl cord, fr rostrl of the injury site 7,8. This ntomicl rerrngement ws ccompnied y the occurrence of forelim movements upon electricl microstimultion of the hindlim cortex 8. However, the functionlity of these neurons remined unknown. If the sprouting corticospinl neurons re to e incorported into the forelim circuitry, forelim sensory input must rech these neurons nd intrcorticl connectivity must e modified in line with the chnges in the CST. Whether the denervted hindlim cortex cn e ctivted y sensory input from the forelims remins unresolved in the light of pprently contrdictory findings from previous studies We creted lrge thorcic lesions in dult rts tht resulted in incresed ehviorl dependence of the nimls on the forelim in the sence of functionl hindlims. Using retrogrde trcings from the spinl cord we identified nd mpped xotomised corticospinl neurons in the hindlim cortex tht were rewired to the cervicl spinl cord. Mny sprouting fiers originted from the hindlim field. Susets of these neurons remined rewired to the cervicl spinl cord severl months lter. VSD imging nd BOLD fmri of the sensorimotor cortex reveled tht the forelim representtion ws expnded fter injury. The enlrged forelim representtion included those prts of the hindlim sensorimotor cortex tht gve rise to the rewired corticospinl neurons. By recording ction potentils from xotomized hindlim corticospinl neurons, we found tht they responded fster to forelim sensory input thn when intct. Together, these results show tht xotomised corticospinl neurons rewire rostrl to the injury to ecome forelim neurons. These neurons might e functionlly relevnt to the motor control of the unffected ody prts. RESULTS Forelim use fter thorcic spinl cord injury Lrge thorcic ilterl dorsl hemisections of the rt spinl cord (Fig. 1 nd Supplementry Fig. 1) resulted in severe hindlim presis with little spontneous recovery 16,17. We ssessed lim movements during overground locomotion on runwy (Ctwlk R ) tht mesured 1 Brin Reserch Institute, University of Zurich nd Swiss Federl Institute of Technology Zurich, Zurich, Switzerlnd. 2 Institute of Phrmcology nd Toxicology, University of Zurich, Zurich, Switzerlnd. 3 Institute for Biomedicl Engineering, University of Zurich nd Swiss Federl Institute of Technology Zurich, Zurich, Switzerlnd. 4 Present ddress: Institute of Neuroinformtics, University of Zurich nd Swiss Federl Institute of Technology Zurich, Zurich, Switzerlnd. Correspondence should e ddressed to A.G. (rko@ini.phys.ethz.ch). Received 21 July; ccepted 15 Octoer; pulished online 13 Decemer 29; doi:1.138/nn.2448 nture NEUROSCIENCE VOLUME 13 NUMBER 1 jnury 21 97

2 Figure 1 Behviorl nlysis fter thorcic spinl cord injury. () The injury t the T8 segment of the spinl cord (lck rrowhed) resulted in prlyzed hindlims. () Quntifiction of the numer of fore- nd hindlim steps tken while crossing the horizontl runwy. (c) Pw prints mde y rt efore nd eight weeks fter injury, cptured using the nlysis system (Ctwlk). (d) Men pixel intensities (ritrry unit) of pw plcements mesured using the nlysis system. (e) Schemtic representtion of nimls crossing horizontl ldder used to evlute their skilled movements efore nd fter injury. Blck mrks on the rung denote forelim plcements. Scle r, 6 cm. (f) Quntifiction of the numer of correct plcements on the horizontl ldder. n = 13 rts, error rs represent ± s.e.m., dt ws sujected to t-test (unpired, two-tiled), P <.5, P <.1. c Before injury 8 weeks fter injury Pw plcements (% of totl plcements) d Intct Forelims Hindlims 1 week 4 weeks 8 weeks the numer of pw plcements nd the intensity of the foot prints (indicting the mount of odyweight supported) 18. Intct nimls wlked on the runwy with lternting fore- nd hindlim steps (Fig. 1,c). The thorcic injury (t verter T8) resulted in mrked reduction of hindlim use nd the forelims were used extensively to cross the runwy. The forelims propelled the ody forwrd, drgging the prlyzed hindlims. The forelims performed 9 ± 3.4% (± denotes s.e.m.) of the steps fter injury compred to 52 ±.34% in intct nimls (P = , Fig. 1,c). At oth four nd eight weeks fter injury, forelim use remined high even though the hindlim recovered some stepping ility. This spontneous recovery of the hindlim ws very poor; the hindlims were not used s much s efore injury nd even when they were used to step, the ody weight supported ws much lower (Fig. 1,d). The forelims lso showed greter weight ering fter injury (135 ± 2.87 pixel intensity (ritrry unit 18 ) in intct rts compred with 145 ± weeks fter injury, P =.15). We evluted the nimls s they crossed horizontl ldder s test of precise sensory-guided pw plcements (Fig. 1e,f). This tsk depends strongly on sensory input nd descending motor control, including tht crried y the CST 19. Trined intct nimls crossed the ldder with ese, mking regulr forelim nd hindlim steps (1 ±.% correct forelim plcements nd 95 ± 1.5% correct hindlim plcements). After spinl cord injury, the nimls crossed the ldder with gret difficulty. The hindlims were often drgged long; the forelims gripped the rungs nd enled the rts to cross the ldder. e FL HL FL HL Before injury After injury Men intensity of pw plcements f Correct plcement (percentge of totl ttempts) The drop in the success rte of the hindlim on the horizontl ldder fter injury (2 ±.8% correct plcements 1 week fter injury) persisted till the lst time point of ssessment t 8 weeks (3 ± 2.% correct plcements t 8 weeks fter injury). Therefore, in the skilled tsk of crossing the horizontl ldder, prpretic rts were entirely dependent on their forelims in the sence of functionl recovery of the hindlims. Reorgniztion of xotomized hindlim corticospinl xons In cutely injured nimls, creful retrogrde leling of the corticospinl neurons tht project to the cervicl gry mtter with the dye Intct 1 week 4 weeks 8 weeks Time fter injury Intct 1 week 4 weeks 8 weeks Time fter injury C 6/7 Fst lue T 8 Mini ruy No. of corticospinl neurons (counts per hemisphere) 4,5 3, 1,5 Midline HL Dy 1 week RFA n = 5 rts CFA Axotomized neurons (hindlim) Doule-leled neurons (hindlim nd forelim) Axotomised (dy ) Time of second trcer injection fter injury 3 12 weeks weeks S-ll c Axotomized neuron (hindlim only) Doule-leled neuron (from hindlim nd forelim) d e f Dy M 1 week fter injury 3 weeks fter injury R 12 weeks fter injury Figure 2 Retrogrde leling of hindlim corticospinl neurons projecting to the cervicl spinl cord gry mtter. () A representtive non-flttened corticl mp of retrogrde-leled neurons in n cutely injured niml (horizontl plne of view of the reconstruction); red dots indicte hindlim corticospinl neurons, lck dots indicte neurons projecting to the cervicl spinl cord (forelim corticospinl neurons). The retrogrde trcer Mini Ruy ws plced t the site of injury (red rrowhed, T8) nd second trcer, Fst Blue, ws injected unilterlly into the cervicl spinl cord gry mtter t five injection sites (C6/7, lck spots). RFA: rostrl forelim re, CFA: cudl forelim re, S-II: secondry somtosensory re, HL: hindlim corticospinl representtion. Scle r, 1 mm. () Quntifiction of doule-leled neurons in the corticl hemisphere contrlterl to injection in intct nd injured rts. Y xis denotes cell counts from the hemisphere contrlterl to the injection s shown in (derived from 5-µm-thick rin sections collected with 1-µm gp). In cutely injured rts, very smll proportion of the xotomized hindlim neurons (red r) were doule-leled (lso projecting to the forelim (gry rs)). (c f) Representtive mps of doule-leled (xotomised hindlim corticospinl neurons projecting to the cervicl spinl cord) nd single-leled (xotomized corticospinl neurons without detectle uptke from the cervicl cord) in the sensorimotor cortex t different time points fter injury. R, rostrl. M, medil. n = 5 rts, error rs represent ± s.e.m., dt ws sujected to Mnn-Whitney U test, P < VOLUME 13 NUMBER 1 jnury 21 nture NEUROSCIENCE

3 Figure 3 BOLD-fMRI mpping of forelim ctivtion in the sensory motor cortex efore nd fter injury. (,c,e) Representtive grdientecho EPI scns (horizontl plne of view) with BOLD-fMRI ctivtion mp in rt efore injury (), 1 week fter injury (c) nd 4 weeks fter injury (e) in response to forepw stimultion (1.6 ma). Insert in shows the loction of the hindlim in the sme niml efore injury. In, c nd e the red line depicts the hindlim re. (,d,f) Superimposed ctivtion mps of intct rts () nd rts 1 (d) or 4 (f) weeks fter injury. Blue nd red res show the ctivtion fter forepw nd hindpw stimultion, respectively. After injury, there is no response from the hindpw; the originl re is repeted in gry (d,f). (g,h) Men forelim representtion re (lck outlined columns) nd forelimhindlim overlp re (gry columns) in response to 1.6 ma (g) nd 6 ma (h) forelim stimultion. Sttistics on re quntifictions re s descried in Figure 2. Mps were derived without flttening the corticl surfce. Intct Fst Blue (without leling the neighoring, densely pcked, CST), followed y ppliction of nother trcer (Mini Ruy) to the cut CST xons reveled the seprtion of the forelim nd hindlim corticospinl representtions in the sensorimotor cortex (Fig. 2). Although hindlim corticospinl neurons ppered from confined re of the sensorimotor cortex, forelim neurons ppered from much lrger re. Forelim CST fiers primrily originte from the cudl forelim re (CFA, homologous to primry motor cortex in primtes 2 ) nd from the rostrl forelim re (RFA, homologous to premotor motor cortex nd supplementry motor re in primtes 21 ). In greement with previous studies, we found tht only few of these forelim neurons in the hindlim field were doule-leled with the two retrogrde trcers (27.6 ± 2.2, counts per hemisphere contrlterl to the cervicl injection); tht is, they were neurons with fiers tht projected oth to the cervicl spinl cord nd to the lower thorcic or lumo-scrl cord (Fig. 2,c) 22. When the trcer Fst Blue ws injected in the cervicl gry mtter 1 week, 3 weeks or 12 weeks fter spinl cord injury (Mini Ruy plced t the injury site), there ws significnt increse in the numer of doule-leled cells (Fig. 2). One week fter injury, the numer of doule-leled cells incresed to 393 ± 144 (P =.1). Axotomized fiers originting minly from the rostrl regions of the hindlim sensorimotor cortex projected more frequently to the cervicl spinl cord, forming dense cluster of doule-leled cells (Fig. 2d). Three weeks fter injury, the numer of doule-leled cells remined high (Fig. 2,e), ut t twelve weeks there ws significnt decrese in the numer of doule-leled cells (158 ± 28 compred to 384 ± 9 t 3 weeks, P =.4). These cells were preferentilly locted in the rostrl hindlim field (Fig. 2f). Over the course of 12 weeks, the totl numer of retrogrdely leled xotomized hindlim corticospinl neurons did not lter significntly (4,866 ± 62 in cutely injured nimls, 3,319 ± 27 twelve weeks fter injury, P =.3), ruling out lrge loss of cells y cell deth, in greement with previous studies 1,2. The numer of single-leled forelim corticospinl neurons in the hindlim field lso remined unchnged (487 ± 72 in cutely injured nimls, 42 ± 84 one week fter injury, 384 ± 47 three weeks fter injury, 456 ± 53 twelve weeks fter injury). These 12 t vlue 2 c e 1 week fter injury 4 weeks fter injury Distnce from regm (mm) d Distnce from regm (mm) f Distnce from regm (mm) No. of rts Forepw 8 1 Hindpw Lterl distnce from midline (mm) Lterl distnce from midline (mm) Lterl distnce from midline (mm) Forelim response re Forelim response re in hindlim re 1.6-mA forepw stimultion n = 8 rts results show tht xotomised hindlim corticospinl neurons cn increse their innervtion of the cervicl cord. Representtion of the forelim s shown y BOLD-fMRI The increse in fier connections from xotomized hindlim corticospinl neurons to the cervicl spinl cord led us to sk whether forelim sensory input cn rech the hindlim sensorimotor cortex. In the rt, forelim sensory nd motor cortices re well seprted (s in higher mmmls), ut the hindlim sensory nd motor representtions completely overlp 23. We mpped the ctivted regions fter sucutneous electricl stimultion of the forepw nd hindpw (Fig. 3). We compred the effects of stimultion t 1.6 ma nd 6 ma (3 Hz) to determine whether the detected mp ltertions depend on the stimultion current mplitude. The higher stimultion intensity hs een used in BOLD-fMRI experiments under isoflurne nesthesi nd is considered to e non-pinful 24,25. At the lower current stimultion, the forelim representtion of intct nimls ws confined to n re of 4.1 ±.3 mm 2 in the somtosensory cortex (Fig. 3,,g). We found no pre-existing overlp of forelim nd hindlim somtosensory representtions (Fig. 3,g). At no point fter injury could BOLD ctivtion e elicited fter hindlim stimultion. One week fter injury, the re ctivted y forelim stimultion expnded to 7. ±.85 mm 2 (P =.135, Fig. 3c,d,g). This expnded re included regions mostly medil to the originl forelim representtion (Fig. 3c,d). Prts of the originl hindlim representtion lso responded to forelim stimultion (Fig. 3c,d,g). This response ws consistent cross nimls nd occurred minly in the rostrl prt of the hindlim representtion. Three weeks lter the enlrged forelim representtion, overlpping with the rostrl prt of the originl hindlim field, remined (Fig. 3e,f,g). The non-invsive nture of BOLD-fMRI llowed us to follow up individul nimls, g Are (mm 2 ) h Are (mm 2 ) mA forepw stimultion n = 8 rts Intct 1 week 4 weeks Time fter injury Intct 1 week 4 weeks Time fter injury 12 weeks nture NEUROSCIENCE VOLUME 13 NUMBER 1 jnury 21 99

4 Figure 4 Voltge-sensitive dye imging of the sensory nd motor cortices responding to forelim stimultion efore nd fter injury. () Sptio-temporl dynmics of VSD signls in intct nd injured rts, one week fter injury. The digrm of the rt hed shows the field of view (purple) used to detect responses fter contrlterl forepw stimultion. Blue nd gry outlines show 5% isocentric contours t 15 ms nd 25 ms fter stimultion, respectively. The 15 ms contour corresponds ntomiclly to nd is used s the forelim sensory region of interest. The 25 ms contour corresponds ntomiclly to nd is used s the forelim motor region of interest. (,c) Men time ctivtion curves t different time points fter injury with quntifiction of pek mplitude expressed s column plots. Regions of interest were the forelim sensory cortex in nd the forelim motor cortex in c. Stimultion intensity is mentioned elow the ctivtion curves. Thin lines nd error rs represent ± s.e.m. n = 5 rts, dt were sujected to Mnn-Whitney U test, P <.5, P <.1. Green rrowhed points to the time of stimultion. Scle r, 1.5 mm. 1 ms Intct nd we found tht ll nimls hd enlrged forelim representtions 1 week fter injury (Supplementry Fig. 2). To wht extent is 35 ms the mp expnsion into the rostrl portion of hindlim field opportunistic? Although only frction of the expnsion covered the 4 ms hindlim field, the re of overlp incresed proportionlly with the chnge in forelim mp size (Supplementry Fig. 2). Most of the expnsion occurred medilly into the forelim motor re, suggesting tht there ws n increse in sensorimotor excitility in ddition to n opportunistic encrochment of the hindlim field fter injury. At the higher current stimultion, the re occupied y the forelim representtion in intct nimls ws in the sme rnge s in the lower stimultion (3.8 ±.4 mm 2, Fig. 3g,h). Similrly, forelim nd hindlim representtions were well seprted (Fig. 3h nd Supplementry Fig. 3). One week fter injury, the higher stimultion reveled n expnsion into the rostrl hindlim field tht remined for up to 12 weeks (Fig. 3h nd see Supplementry Fig. 3). In ddition, the forelim mps were mrkedly lrger when stimulted with the 6-mA current stimultion thn the 1.6-mA current t 1 week fter injury (12.4 ± 1.6 mm 2, Fig. 3g,h). Voltge-sensitive dye imging of the sensorimotor cortex Next we ddressed the sptio-temporl dynmics of forelim sensory input in the sensorimotor cortex with temporl resolution of 5 ms using VSD imging. In comprison to BOLD-fMRI experiments, much lower stimultion strengths could e used under the sme level of nesthesi, highlighting the sensitivity of VSD imging: the forepw ws stimulted sucutneously with 1-ms pulse of current (2 or 6 µa). In intct rts, neuronl responses in the forelim sensory cortex were seen t ltency of 15 ms (Fig. 4). In the following 5 1 ms, the ctivtion spred medilly to the forelim motor cortex. One week fter spinl cord injury, the sme stimultion resulted in stronger nd lrger response t the forelim sensory cortex (Fig. 4). 15 ms 2 ms 25 ms 3 ms 1 week fter injury 6-µA forepw stimultion Percentge F/F.7.4 percentge F/F c 1 ms Intct 1 week fter injury 4 weeks fter injury 12 weeks fter injury Forelim sensory cortex n = 5 rts 2 µa 6 µa Forelim motor cortex 2 µa 6 µa At 2 25 ms fter stimultion, oth forelim sensory nd motor cortices were highly ctivted compred to intct nimls. In the following milliseconds the ctivtion encompssed regions cudl to the originl forelim motor cortex, the hindlim sensorimotor cortex s descried lter. To quntify the ctivtions, two regions of interest were identified: the forelim sensory cortex nd the forelim motor cortex. In response to the low (2 µa) nd high current (6 µa) stimultions, oth the sensory nd motor cortices showed incresed ctivtion one week fter injury (Fig. 4 c); for exmple, t 6 µa stimultion, the forelim sensory cortex incresed its ctivtion from.32 ±.5% F/F to.57 ±.3% F/F (P =.9). The incresed ctivtions were sent 4 or 12 weeks fter injury (Fig. 4,c). In intct nimls, the sensory cortex response mplitude of.3% F/F (t 6 µa) resulted in slightly weker ctivtion of the motor cortex reflected y motor:sensory cortex ctivtion rtio of.8 ±.1. One week fter injury, t 2 µa stimultion, the sensory response mplitude ws.3% F/F, nd the susequent ctivtion in the motor cortex ws eqully strong resulting in rtio of 1.1 ±.1. Therefore, the reltive excitility of the motor cortex ws higher fter spinl cord injury (P =.36). Next, we gin investigted the ctivtion of the hindlim sensorimotor cortex y forelim sensory input fter injury. We estimted the hindlim field using two pproches. The first relies on the locliztion Percentge F/F Percentge F/F 1 VOLUME 13 NUMBER 1 jnury 21 nture neuroscience

5 .4 percentge F/F Intct 2 ms R7 R1 Intct Percentge R7 F/F.6 1 week fter injury R1 Percentge F/F 1 week fter injury R1.6 R7 R7 6-µA forepw stimultion R1 Time 8 ms c d e HL FL Intct 1 week fter injury n = 1 trils 5 ms 8-µA hindpw stimulton 5 ms 25 ms 45 ms 65 ms 85 ms 15 ms 125 ms Intct 1 Week fter injury 6-µA forepw stimulton Pek mplitude (percentge F/F ).2 percentge F/F.6.3 Intct 1 week fter injury n = 7 rts R1 R7 2-µA forepw stimultion.6.3 R R7 6-µA forepw stimultion Intct 1 week fter injury 4 weeks fter injury 12 weeks fter injury HL sensory-motor cortex P1 P2 n = 5 rts.3 1 ms Percentge F/F.4 2 µa 6 µa P1 P2 P1 P Percentge F/F Percentge F/F Figure 5 Response of the hindlim sensorimotor cortex to forelim stimultion in intct nd injured rts. () Top left: schemtic view of the rt hed showing the region of dt cquisition in purple. The hindlim field lies cudo-medil to the forelim sensory cortex. Bottom left: seven 3-µm dimeter circulr ROIs (R1 R7 white circles) were plced djcent to the forelim representtion such tht R1 included the cudl edge of the 5% isocentric counter drwn 2 ms fter forepw stimultion. The ROIs were plced t n ngle of 35 from the midline. Right: ctivtion representtions in the ROIs. Note tht the ROIs closest to the forelim re ctivted first nd the ctivtion is much higher in ll ROIs fter injury. Green rrowheds in ll pnels point to the time of stimultion. Blck rrowhed in points to 2 ms fter stimultion. () Quntifiction of ctivtions in the distinct ROIs show strong ctivity in the hindlim field fter spinl cord injury t oth 2 nd 6 µa stimultions. (c) Representtive, superimposed, 5% isocentric contours t the time of pek response mplitude evoked y hindpw stimultion (HL; 8 µa current) in n intct (gry outline) nd injured (red outline, one week fter injury) rt nd t 15 ms fter forepw stimultion (FL; 6 µa current). Below, time ctivtion curves of the enclosed hindlim re responding to 8 µa hindpw stimultion. (d) Representtive sptio-temporl dynmics of neuronl ctivity fter forepw stimultion in the identified hindlim cortex in intct nd injured rts. Grey nd red outlines s in c. (e) Men time ctivtion curves of responses in the hindlim cortex fter forelim stimultion nd column plots of pek response mplitude in the two temporl locks P1 nd P2. Stimultion intensity is indicted elow the plots. n = 5 rts, sttistics re s descried in Figure 4. Scle rs, 1.5 mm. of the forelim representtion nd its known sptil reltionship to the hindlim field, nd the second is sed on the ctivtion of the hindlim field in lesioned nimls y stimultion of the spred spinothlmic trct. In intct nimls the hindlim re is cudo-medil to the forelim sensorimotor field. We therefore ssessed the hindlim re in injured nimls y plcing seven djcent regions of interest (ROIs) 35 from the midline strting with the cudl prt of the forelim field. In intct nimls, forepw stimultion resulted in low-mplitude ctivtion of the hindlim ROIs (Fig. 5,). The ctivtion ppered first in the ROI closest to the forelim re nd lst in the ROI frthest wy. One week fter injury, ll the ROIs showed sequentil ctivtion t much higher mplitude (Fig. 5,). The second pproch relied on ctivtion of the spred spinothlmic trct. Inputs from this trct rech the hindlim corticl res tht respond to tctile sensory input 26. Stimultion t 8 µa resulted in wek nd delyed ctivtion of the hindlim re (Fig. 5c). In intct nimls, the hindlim sensorimotor re received forelim sensory input 25 ms fter stimultion in very restricted rostro-lterl order re only. A second more generl, weker wve of ctivtion occurred t ~1 ms (Fig. 5d). The first ctivtion occurred exclusively t the rostrl rim of the hindlim sensorimotor re in proximity to the CFA. The second wve included the cudl spects of the hindlim field. One week fter injury, the initil ctivtion t 25 ms ws much stronger nd encroched well into the hindlim field (Fig. 5d). The second wve of ctivtion occurred sooner, ppering y 45 ms nd peking t 85 ms in the cudl hindlim field. We quntified the ctivtions in the identified hindlim re fter 2 nd 6 µa forepw stimultions (Fig. 5e). At the low current stimultion, the mplitude of the first wve of ctivtion ws significntly higher one week fter injury. Both wves of ctivtion hd significntly higher mplitudes when the forepw ws stimulted t 6 µa. Four nd twelve weeks fter injury, the rostrl rim nd the cudl hindlim field continued to respond to forelim stimultion (Fig. 5e). These results show tht the hindlim sensorimotor cortex in prpretic rts responds to forelim sensory input. Response of hindlim corticospinl xons to forelim input From oth BOLD-fMRI nd VSD imging, we conclude tht the denervted hindlim sensorimotor re receives forelim sensory input. However, it remins uncler whether this sensory input ctivtes hindlim corticospinl neurons tht originte from corticl lyer V (which is too deep to e ssessed y VSD imging). Sensory input cn induce ction potentils in corticospinl neurons It remins unknown whether forelim sensory input reches the hindlim CST. nture NEUROSCIENCE VOLUME 13 NUMBER 1 jnury 21 11

6 Figure 6 Forelim stimultion evoked ction potentils in the corticospinl trct t thorcic level T7. () The forelim ws stimulted t 6 ma (green rrowhed) for 1 ms nd proe with 16 electrodes inserted t T7 (golden rrowhed) ws used to detect ction potentils. Red rrowhed shows T8, the loction of injury. () Peri-stimulus time histogrm (PSTH, time lock = 1 ms) of n intct rt responding to hindlim nd forelim stimultion (6 ma, 1 ms). (c) Men ctivtion curves sed on PSTH fter forelim stimultion in intct nd injured nimls (n = 6). Shded re depicts ± s.e.m. (d) Quntifiction of the ltency to pek nd pek mplitude from PSTH of intct nd injured rts. n = 6 rts, sttistics re s descried in Figure 4. Green rrowhed points to the time of stimultion. We recorded extrcellulr ction potentils evoked y sucutneous forepw stimultion from the CST t level T7 (multi-unit extrcellulr recordings) in intct nd injured nimls. Most of these xons project to the lumr enlrgement nd re therefore involved in hindlim function. Recording from the hindlim CST in the spinl cord we could specificlly ddress the chnges in this cell type fter injury. First, we stimulted the hindlim of four intct rts to determine tht t 6 ma current stimultion (1 pulse, 1 ms) the CST response ws significnt (Fig. 6,). In intct nimls, forelim stimultion resulted in significnt response in T7 CST xons t ms (ltency to pek mplitude, men 27 ± 2 ms, n = 4) fter stimultion, 6 ms fster thn the response to hindlim stimultion (men 33 ± 1 ms, n = 4). One week fter injury t T8, the hindlim CST responses to the forelim stimultion were mrkedly fster, with ltency to pek t 18 ± 1 ms (n = 6, Fig. 6c,d). In oth groups, the numer of spikes detected vried gretly etween nimls, proly due to lck of distinction etween single nd compound ction potentils nd differences in the positioning of the 177 µm 2 electrode in the densely pcked CST (Fig. 6d). Importntly, there ws no correltion etween the ltency to pek mplitude nd the mplitude itself in the injured group. However, intct nimls showed tendency for longer ltency to pek with increse in mplitude of the response (r 2 =.55, liner fit, Supplementry Fig. 4). The responses in the CST of oth intct nd injured nimls could e eliminted y corticl ltion tht included the hindlim re (Supplementry Fig. 5). Therefore, xotomised hindlim corticospinl neurons cn respond to sensory input from the forelim. This input might contriute to the rewiring nd forelim function of the hindlim neurons fter spinl cord injury. DISCUSSION We hve shown tht, in response to thorcic ilterl dorsl hemisection, which renders the hindlims dysfunctionl, out 1% of the xotomized hindlim corticospinl neurons rewired to the cervicl spinl cord. Over months, out 5% of these rewired neurons were eliminted from the cervicl segments ut neurons in the rostrl hindlim domin, in proximity to the forelim cortex, remined. Using VSD imging nd BOLD-fMRI of the cortex, we showed tht the forelim sensory representtion ws ltered fter injury nd expnded into the hindlim sensorimotor re. The enhnced forelim sensory input into the hindlim sensorimotor cortex ws lso reflected t the level of the hindlim corticospinl neurons responding to forelim sensory input (see Supplementry Fig. 6 for schemtic summry). After thorcic injury, the hindlims showed no spontneous recovery in the skilled wlking tsks. Owing to the significnt steps tken in the lst decde to induce regenertion in the dult CNS, it is now envisioned tht spinl cord injury tretments tht promote regenertion nd structurl plsticity will eventully led to functionl improvements 4. However, our understnding of the chnges induced y the injury itself remins incomplete. Do xotomised c Multi-unit ctivity (pro. in 1-ms in) Recording site t T7 Injury site t T8 Intct niml Intct.4 Forepw stimultion Hindpw stimultion Injured Forepw stimultion Intct One week fter injury 2 Time (ms) Cudl Rostrl n = 6 rts 4 Multi-unit ctivity (pro. in 1-ms in) Ltency to pek (ms) Pek mplitude (pro. in 1-ms in) 2 Time (ms) 15 trils corticospinl cells remin dormnt fter injury? Our dt show tht the cells cn form new connections, distnt from the site of injury. The presented numer of rewired corticospinl neurons (detected s doule-leled neurons) is gross underestimte, when the efficiencies of trcers used re tken into ccount. Retrogrde leling from the site of injury t T8 with the dextrn dye Mini-ruy leled ~4,9 corticospinl neurons. We estimte tht the leling revels only out one-seventh of the corticospinl xons tht pss through the thorcic spinl cord, ccording to xonl counts 3. The other trcer, Fst Blue, lels ~1.5 times more cells thn the dextrn 31. Therefore, we expect tht the numer of rewired corticospinl neurons is t lest seven times higher thn is shown y the retrogrde method. Adult neurons in the CNS cnnot regenerte over long distnces, ut xotomized neurons cn form short sprouts from their dmged xons Previously, nterogrde trcing experiments in the dult rt fter thorcic injury showed tht corticospinl neurons from the hindlim cortex cn sprout into the cervicl spinl cord gry mtter 7,8. Here, retrogrde trcing llowed us to identify xotomized hindlim neurons tht rewired to cervicl segments. This rewiring might involve the formtion of entirely new xonl collterls nd/or the enlrgement of rre pre-existing xonl rors in the cervicl gry mtter fter xotomy; very few collterls of the hindlim xons into the cervicl spinl cord were detected in intct nimls y using nterogrde trcing, confirming the present retrogrde trcing results. Counts nd topogrphic mps of the rewired neurons showed tht some pruning followed within severl weeks fter the initil sprouting of the xotomized hindlim xons in the cervicl segments. In greement with the nterogrde experiments, out hlf of the rewired hindlim neurons remined connected to the cervicl spinl cord 7,8. Neurons tht did not undergo pruning were mostly locted in the rostrl prt of the hindlim field, in proximity to the originl forelim corticospinl re. The forelim corticospinl representtion is thus permnently strengthened y recruitment of former hindlim CST neurons. These rewired xons presumly connect to the spinl neurons in cervicl cord nd the significnce of this is demonstrted y the existence of forelim movements upon intrcorticl stimultion of the hindlim cortex fter thorcic spinl cord injury 8. The persistence of rewired d n = 6 rts Intct 1 week fter injury 4 12 VOLUME 13 NUMBER 1 jnury 21 nture neuroscience

7 neurons in proximity to the forelim cortex could suggest role for forelim inputs in inducing sprouting or determining the fte of the xotomized sprouting corticospinl neurons. After focl forelim corticl stroke, hindlim neurons (in the superficil lyers) ecome responsive to forelim sensory input 35. After spinl cord injury, if the ntomiclly rewired hindlim sensorimotor cortex is to prticipte in meningful forelim movements, it must lso receive forelim sensory input nd e responsive to it. To ddress this, we used VSD imging nd BOLD-fMRI. Studies in prplegic humns filed to gree whether the denervted leg representtion in the sensory cortex ecomes responsive to input from the hnd 36,37. In prplegic rts too, n expnsion of the forelim sensory representtion into neighoring corticl res including the hindlim sensorimotor cortex hs een descried using BOLD-fMRI, ut n electrophysiologicl study filed to oserve such n ctivtion 13,14. In our study, using forepw electricl stimultion, BOLD-fMRI of intct nimls reveled exclusive ctivtion of the forelim sensory cortex. In contrst, VSD imging reveled roust dditionl ctivtion of the forelim motor cortex (medil to forelim sensory cortex). After injury, the BOLD response re expnded medilly to the forelim motor cortex. This shows tht pre-existing forelim sensorimotor circuits chnge significntly fter injury to result in n expnded BOLD ctivtion re. The rewired CST representtion continues to refine over weeks fter injury. Bsed on the BOLD results it would e premture for us to conclude tht refinement of the sensory input does not tke plce. A etter understnding of the reltionship etween ntomicl rewiring nd BOLD mp ltertions would e possile if these mesurements were tken from the sme niml. Importntly, the rostrl prt of the hindlim representtion, in oth BOLD-fMRI nd VSD experiments, responded consistently to forelim sensory stimultion fter injury. Thus, the denervted hindlim sensorimotor cortex receives forelim sensory inputs fter spinl cord lesion. The expnsion of the forelim representtion into the hindlim field nd the ltertion of the forelim sensorimotor circuitry might e relted to the incresed ehviorl relince of the rts on their forelims fter injury. Using extrcellulr recordings, we found tht xotomised hindlim corticospinl neurons responded fster to sensory inputs from the forelim thn when intct. Exctly which spect of the sensory input these cells were responding to remins uncler given the strength of the stimulus used. Both the sensory current threshold for CST ctivtion nd the pthwys involved remin unidentified. Moreover, given tht our recordings rostrl to the injury might include xotomised xons tht do nd tht do not give rise to collterls in the cervicl spinl cord, it remins uncler whether the rewired neurons respond fster to inputs from the forelim. Shorter ltency to corticl responses from injury spred ody prts hs een previously oserved in experiments involving digit denervtion, within hours of injury 38. In greement with previous in vitro study, we find tht xotomized corticospinl neurons not only survive the injury ut lso continue to respond to synptic input 39. Our VSD imging experiments show tht the forelim sensory cortex, the forelim motor cortex nd the hindlim sensorimotor cortex re ll more excitle in response to forelim sensory input fter injury. With two different mplitudes of stimultion, resulting in the sme response mplitude in the sensory cortex efore nd fter injury, we showed tht the forelim motor cortex is more excitle one week fter injury. Chnges in cellulr physiology, synptic strength nd intr- nd su-corticl circuits might ll contriute to the fster response of the xotomised hindlim CST in prpretic rts. The oserved excitility in the forelim sensorimotor cortex within 7 dys of spinl cord injury might in prticulr e due to ltered effectiveness of horizontl connections 4. The suppression of GABA-medited inhiition might unmsk pre-existing horizontl connections, leding to enlrged corticl ctivtion mps 41. With reduced intr-corticl inhiition, synptic strengthening of excittory horizontl inputs to pyrmidl neurons is likely 42. The suggested ltertions to the input of xotomised corticospinl neurons might lso led to chnges in the corticl hrdwre, including xonl rors nd dendrites. Notly, in dult mice fter spinl cord injury, there is n initil loss of dendritic spine density in lyer V pyrmidl neurons in the motor cortex 43. This loss is prtilly reversed within few weeks fter injury. It remins to e determined whether the chnges reported in our study re driven y the ehviorl compenstions y the forelim. It is known tht lerning of new motor skill results in incresed excitility of the motor cortex 44. Therefore, we speculte tht the incresed forelim use results in ltertions in corticl excitility tht then influences ntomicl ltertions of the xotomised corticospinl neurons. This specultion is in line with erlier studies tht show tht corticl ctivity cn influence sprouting of the CST 45. In summry, we oserved significnt ltertions in how the sensorimotor cortex received nd processed inputs from the forelim fter thorcic spinl cord injury. The forelim sensory representtion expnded nd invded prts of the denervted hindlim re. The regions in the hindlim cortex tht received forelim sensory input gve rise to collterls in the cervicl spinl cord from xotomised neurons. Thus, xotomised corticospinl neurons re-wire fter injury nd re incorported into the circuitry of the injury spred ody prt. Methods Methods nd ny ssocited references re ville in the online version of the pper t Note: Supplementry informtion is ville on the Nture Neuroscience wesite. Acknowledgments We thnk J. Scholl, E. Hochreutener, R. Schö nd L. Schnell for technicl ssistnce. We thnk A. Buchli, I. Mier, M. Strkey, V. Pernet, D. Mrgolis nd B. Kmp for helpful discussions. This work ws supported y the Swiss Ntionl Science Foundtion, Grnt , the Ntionl Center of Competence in Reserch Neurl Plsticity nd Repir of the Swiss Ntionl Science Foundtion, the Spinl Cord Consortium of the Christopher Reeve Prlysis Foundtion, nd NeuroNe, Network of Excellence of the Europen Consortium for Reserch in Neurodegenertive Diseses (Sixth Frmework EU Progrm). AUTHOR CONTRIBUTIONS A.G. nd M.E.S. designed the experiments nd prepred the mnuscript. A.G., F.H. nd E.S. crried out nd interpreted the experiments. R.S., M.G., M.T.W., T.M. nd C.B. performed the experiments. M.R. nd B.W. prepred the mnuscript. Pulished online t Reprints nd permissions informtion is ville online t reprintsndpermissions/. 1. Klil, K. & Schneider, G.E. Retrogrde corticl nd xonl chnges following lesions of the pyrmidl trct. Brin Res. 89, (1975). 2. Wnnier, T., Schmidlin, E., Bloch, J. & Rouiller, E.M. A unilterl section of the corticospinl trct t cervicl level in primte does not led to mesurle cell loss in motor cortex. J. Neurotrum 22, (25). 3. Hins, B.C., Blck, J.A. & Wxmn, S.G. Primry corticl motor neurons undergo poptosis fter xotomizing spinl cord injury. J. Comp. Neurol. 462, (23). 4. Mier, I.C. & Schw, M.E. Sprouting, regenertion nd circuit formtion in the injured spinl cord: fctors nd ctivity. Phil. Trns. R. Soc. Lond. B 361, (26). 5. Rineteu, O. & Schw, M.E. Plsticity of motor systems fter incomplete spinl cord injury. Nt. Rev. Neurosci. 2, (21). 6. Aoki, M., Fujito, Y., Stomi, H., Kurosw, Y. & Ks, T. The possile role of collterl sprouting in the functionl restitution of corticospinl connections fter spinl hemisection. Neurosci. Res. 3, (1986). nture NEUROSCIENCE VOLUME 13 NUMBER 1 jnury 21 13

8 7. Breyre, F.M. et l. The injured spinl cord spontneously forms new intrspinl circuit in dult rts. Nt. Neurosci. 7, (24). 8. Foud, K., Pedersen, V., Schw, M.E. & Brosmle, C. Cervicl sprouting of corticospinl fiers fter thorcic spinl cord injury ccompnies shifts in evoked motor responses. Curr. Biol. 11, (21). 9. Weidner, N., Ner, A., Slimi, N. & Tuszynski, M.H. Spontneous corticospinl xonl plsticity nd functionl recovery fter dult centrl nervous system injury. Proc. Ntl. Acd. Sci. USA 98, (21). 1. Ghosh, A. et l. Functionl nd ntomicl reorgniztion of the sensory-motor cortex fter incomplete spinl cord injury in dult rts. J. Neurosci. 29, (29). 11. Ks, J.H. et l. Corticl nd sucorticl plsticity in the rins of humns, primtes, nd rts fter dmge to sensory fferents in the dorsl columns of the spinl cord. Exp. Neurol. 29, (28). 12. Jin, N., Florence, S.L., Qi, H.X. & Ks, J.H. Growth of new rinstem connections in dult monkeys with mssive sensory loss. Proc. Ntl. Acd. Sci. USA 97, (2). 13. Endo, T., Spenger, C., Toming, T., Brene, S. & Olson, L. Corticl sensory mp rerrngement fter spinl cord injury: fmri responses linked to Nogo signlling. Brin 13, (27). 14. Jin, N., Florence, S.L. & Ks, J.H. Limits on plsticity in somtosensory cortex of dult rts: hindlim cortex is not rectivted fter dorsl column section. J. Neurophysiol. 73, (1995). 15. Wll, P.D. & Egger, M.D. Formtion of new connexions in dult rt rins fter prtil defferenttion. Nture 232, (1971). 16. Liescher, T. et l. Nogo-A ntiody improves regenertion nd locomotion of spinl cord-injured rts. Ann. Neurol. 58, (25). 17. Schucht, P., Rineteu, O., Schw, M.E. & Foud, K. Antomicl correltes of locomotor recovery following dorsl nd ventrl lesions of the rt spinl cord. Exp. Neurol. 176, (22). 18. Hmers, F.P., Koopmns, G.C. & Joosten, E.A. CtWlk-ssisted git nlysis in the ssessment of spinl cord injury. J. Neurotrum 23, (26). 19. Metz, G.A. & Whishw, I.Q. Corticl nd sucorticl lesions impir skilled wlking in the ldder rung wlking test: new tsk to evlute fore- nd hindlim stepping, plcing, nd co-ordintion. J. Neurosci. Methods 115, (22). 2. Sievert, C.F. & Nefsey, E.J. A chronic unit study of the sensory properties of neurons in the forelim res of rt sensorimotor cortex. Brin Res. 381, (1986). 21. Nefsey, E.J. & Sievert, C. A second forelim motor re exists in rt frontl cortex. Brin Res. 232, (1982). 22. Akintunde, A. & Buxton, D.F. Differentil sites of origin nd collterliztion of corticospinl neurons in the rt: multiple fluorescent retrogrde trcer study. Brin Res. 575, (1992). 23. Hll, R.D. & Lindholm, E.P. Orgniztion of motor nd somtosensory neocortex in the lino rt. Brin Res. 66, (1974). 24. Liu, Z.M., Schmidt, K.F., Sicrd, K.M. & Duong, T.Q. Imging oxygen consumption in forepw somtosensory stimultion in rts under isoflurne nesthesi. Mgn. Reson. Med. 52, (24). 25. Sicrd, K.M. & Duong, T.Q. Effects of hypoxi, hyperoxi, nd hypercpni on seline nd stimulus-evoked BOLD, CBF, nd CMRO2 in spontneously rething nimls. Neuroimge 25, (25). 26. Guilud, G., Benoist, J.M., Levnte, A., Gutron, M. & Willer, J.C. Primry somtosensory cortex in rts with pin-relted ehviors due to peripherl mononeuropthy fter moderte ligtion of one scitic nerve: neuronl responsivity to somtic stimultion. Exp. Brin Res. 92, (1992). 27. Adrin, E.D. & Moruzzi, G. Impulses in the pyrmidl trct. J. Physiol. (Lond.) 97, (1939). 28. Swett, J.E. & Bourss, C.M. Short ltency ctivtion of pyrmidl trct cells y Group I fferent volleys in the ct. J. Physiol. (Lond.) 189, (1967). 29. McComs, A.J. & Wilson, P. An investigtion of pyrmidl trct cells in the somtosensory cortex of the rt. J. Physiol. (Lond.) 194, (1968). 3. Schreyer, D.J. & Jones, E.G. Axon elimintion in the developing corticospinl trct of the rt. Brin Res. 466, (1988). 31. Choi, D., Li, D. & Rismn, G. Fluorescent retrogrde neuronl trcers tht lel the rt fcil nucleus: comprison of Fst Blue, Fluoro-ruy, Fluoro-emerld, Fluoro-Gold nd DiI. J. Neurosci. Methods 117, (22). 32. Schw, M.E. & Brtholdi, D. Degenertion nd regenertion of xons in the lesioned spinl cord. Physiol. Rev. 76, (1996). 33. Fwcett, J.W. & Geller, H.M. Regenertion in the CNS: optimism mounts. Trends Neurosci. 21, (1998). 34. Foerster, A.P. Spontneous regenertion of cut xons in dult rt rin. J. Comp. Neurol. 21, (1982). 35. Brown, C.E., Aminoltejri, K., Er, H., Winship, I.R. & Murphy, T.H. In vivo voltgesensitive dye imging in dult mice revels tht somtosensory mps lost to stroke re replced over weeks y new structurl nd functionl circuits with prolonged modes of ctivtion within oth the peri-infrct zone nd distnt sites. J. Neurosci. 29, (29). 36. Turner, J.A., Lee, J.S., Schndler, S.L. & Cohen, M.J. An fmri investigtion of hnd representtion in prplegic humns. Neurorehil. Neurl Repir 17, (23). 37. Curt, A., Bruehlmeier, M., Leenders, K.L., Roelcke, U. & Dietz, V. Differentil effect of spinl cord injury nd functionl impirment on humn rin ctivtion. J. Neurotrum 19, (22). 38. Doetsch, G.S., Hrrison, T.A., McDonld, A.C. & Litker, M.S. Short-term plsticity in primry somtosensory cortex of the rt: rpid chnges in mgnitudes nd ltencies of neuronl responses following digit denervtion. Exp. Brin Res. 112, (1996). 39. Tseng, G.F. & Prince, D.A. Structurl nd functionl ltertions in rt corticospinl neurons fter xotomy. J. Neurophysiol. 75, (1996). 4. Huntley, G.W. Correltion etween ptterns of horizontl connectivity nd the extend of short-term representtionl plsticity in rt motor cortex. Cere. Cortex 7, (1997). 41. Jcos, K.M. & Donoghue, J.P. Reshping the corticl motor mp y unmsking ltent intrcorticl connections. Science 251, (1991). 42. Hess, G. & Donoghue, J.P. Long-term potentition of horizontl connections provides mechnism to reorgnize corticl motor mps. J. Neurophysiol. 71, (1994). 43. Kim, B.G., Di, H.N., McAtee, M., Vicini, S. & Bregmn, B.S. Remodeling of synptic structures in the motor cortex following spinl cord injury. Exp. Neurol. 198, (26). 44. Rosenkrnz, K., Willimon, A. & Rothwell, J.C. Motorcorticl excitility nd synptic plsticity is enhnced in professionl musicins. J. Neurosci. 27, (27). 45. Slimi, I., Friel, K.M. & Mrtin, J.H. Pyrmidl trct stimultion restores norml corticospinl trct connections nd visuomotor skill fter erly postntl motor cortex ctivity lockde. J. Neurosci. 28, (28). 14 VOLUME 13 NUMBER 1 jnury 21 nture neuroscience

9 ONLINE METHODS Experimentl sujects. Adult femle Lewis rts (2 22 g, otined from R. Jnvier, Le Genest-St Isle, Frnce) were used. Animls were housed in groups of four in stndrdized cges (Mrcolon type IV) t 12:12 h light:drk cycle with food nd wter d liitum. All nimls were given one week to cclimtize to their housing efore experimenttion. All experimentl procedures performed were in dherence to the guidelines of the Veterinry office of the Cnton of Zurich, Switzerlnd. Spinl cord injury. Spinl cord dorsl ilterl hemisections were performed s descried elsewhere 7. Briefly, dorsl ilterl lminectomy ws performed t thorcic segment 8 (verter) under nesthesi (Hypnorm, 12 µl per 2 g ody weight sucutneous (VetPhrm) nd Dormicum,.75 mg in 15 µl per 2 g odyweight sucutneous (Roche Phrmceuticls)). Using shrp iridectomy scissors, the dorsl hlf of the spinl cord ws trnsected (Supplementry Fig. 1). The retrogrde trcer ws pplied to the injury site s descried elow. After surgery, nimls were dministered nlgesic (2.5 mg per kg of ody weight sucutneous injection of Rimdyl; Pfizer Lortories) once dy for five dys nd ntiiotic (Bytril 1.25 mg per 25 g ody weight intrperitonel; Byer AG) once dy for five dys. Bldders were checked nd emptied three times dy until ldder function hd completely recovered. Spinl cord injuries were histologiclly verified. Indequtely injured or over-lesioned rts were susequently eliminted from nlysis. Behviorl dt cquisition nd nlysis. Animls were evluted on the Ctwlk R (Noldus Informtion Technology) nd on the horizontl ldder. Before injury, nimls were trined on the ehviorl tests for 3 weeks (twice per week in the first 2 weeks, once in the following week). The ctwlk nlysis system hs een descried in detil 18. Briefly, this system consists of glss runwy, with customized lighting, which is video recorded s the nimls trverse the runwy. From these videos the system determines the percentge of usge y ech pw nd the intensity of footprints during mximum pw contct t every step. The horizontl ldder consists of eqully spced (t 6-cm gps) rungs. The ldder ws 1 m long nd elevted t 1 m. Three trils were video recorded over 6-cm stretch. Videos were nlyzed offline. When the pw ws plced such tht the lim did not slip from the rung (weight-supported steps), step ws noted s successful 1. The percentge of correct plcements for ech niml ws determined y verging over three trils. Retrogrde trcing nd leled cell nlysis. After spinl cord injury ws performed s descried ove, 2 µl of retrogrde trcer Mini Ruy ( 3, MW dextrn conjugted to tetr-methyl rhodmine, 1% in.1m PB; Invitrogen) ws plced t the site of injury. Twenty minutes lter, excess dye seeping out of the lesion site ws removed y plcing sterile cotton piece on the dorsl surfce of the injured cord, covering the lesion for 1 15 min. Injections into the cervicl gry mtter were done using modified stereotxic frme under deep nesthesi (Hypnorm nd Dormicum s mentioned ove): the hed ws fixed using er rs nd the cervicl verter ws fixed using clmps plced on either side of the verterl column. A unilterl lmenectomy ws performed t C6 nd C7 nd the dur ws removed. Using 28 G Nnofil syringe ttched to n UltrMicroPump (WPI) mounted on the stereotxic frme, 1% Fst Blue (EMS-Polyloy) suspension in.1 M PB nd 2% DMSO ws injected into the spinl cord. The injection ws monitored with fluorescence inoculrs. Ech niml hd five injections of 12 nl of Fst Blue spced t.5 mm, long the spinl cord. The coordintes for injection were.75 mm lterl from the midline nd 1.2 mm elow the spinl cord surfce. Unilterlity of the injection ws verified post-mortem. Ten dys fter injections, nimls were perfused with prformldehyde (4%, post-fixed for 4 h fter perfusion) nd processed for cryosectioning in 3% sucrose. Series of 5-µm spinl cord sections of the injection site in the cervicl segment confirmed the confinement of the trcer to the cervicl gry mtter nd lck of spred into the densely pcked white mtter. Series of 5-µm rin cross-sections (with 1-µm gps) were used for quntifiction of leled cells (cell counts were multiplied y 3 to incorporte the skipped sections into our estimte of the numer of cells per hemisphere) nd for the reconstruction of forelim nd hindlim corticospinl representtions. Fst Blue, Mini Ruy nd doule-leled neurons were identified on fluorescent microscope configured to Neurolucid (ver 7.; MicroBrightfield). Apprent doule-leled cells with fint lel of either trcer were confirmed using confocl microscopy. BOLD fmri. Rts were scnned efore nd fter spinl cord injury on Bruker Biospec 94/3 horizontl ore smll niml MR system (BGA-12 grdient insert; grdient strength 4 mt m 1, minimum rise time of 8 µs; Bruker Biospin GmH) under isoflurne nesthesi (1.5%). The muscle relxnt Gllmine (3 mg per 2 g odyweight; Sigm-Aldrich) ws used to immoilize the nimls. Antomicl imges were otined using multi-slice rpid cquisition with relxtion enhncement (RARE) spin echo sequence (prmeters: field of view (FOV) = mm 2, mtrix dimension (MD) = , slice thickness (SLTH) = 1 mm, inter-slice distnce (ISD) = 1.25 mm, echo dely (TE) = 6 ms, repetition dely (TR) = 1,259 ms, RARE fctor = 8, numer of cquisitions (NA) = 1, cquisition time T cq = 1.25 min). BOLD-fMRI ws performed using seril grdientecho echo-plnr-imging (GE-EPI) sequence (FOV = mm 2, MD = 64 64, SLTH = 1 mm, temporl resolution = 1 s, 5 repetitions, T cq = 8 min 2 s). Two horizontl imges were cquired to cover ll corticl lyers. Both forepws nd the hindpw were stimulted using needle electrodes plced sucutneously. Stimultion prmeters were: mplitude = ma, pulse durtion =.5 ms nd frequency = 3 Hz. A repetitive lock design hs een used with 5 cycles consisting of n on-period of 4 s nd n off-period of 6 s. Digit flexion threshold ws determined efore the ppliction of Gllmine. Dt were nlyzed using Biomp (ver. 4; Mrtin Rusch, Novrtis) with sttisticl nlysis done using generl liner model. Thresholds for considering ctivtion were P <.1 nd minimum re of 3 voxels. Without further processing, these dt were sptilly normlized using lndmrks identified directly on EPI imges s descried 46 (using custom-written softwre sed on IDL (ITT)). Briefly, mximum intensity projections of the two horizontl sections were clculted nd susequently summed to generte overlpping mps (using custom-written progrm sed on Mtl (The MthWorks), extrction with t-vlue threshold of 4). Opticl imging of the cortex using voltge-sensitive dye. Opticl imging using the VSD RH1691 ws performed s descried, under isoflourne nesthesi (4% for induction, mintined t 1.5% during imging, 3% while not imging) 1,47. Briefly, the sensorimotor cortex ws exposed crefully nd the VSD (1 mg ml 1 in Ringer s solution) ws topiclly dded nd then llowed to diffuse for 1 h. The undiffused dye ws wshed from the surfce using Ringers solution. The cortex ws then covered with 2% gr solution nd with glss coverslip. Animls were stimulted (sucutneous forepw or hindpw electrodes) with single pulse current (2 8 µa mplitude, 1 ms durtion). Stimultion strength s high s used in BOLD-fMRI resulted in sturtion of the VSD signl nd mde the sensorimotor ctivtion difficult to chrcterize. VSD ws excited with 63 nm light from 1-W hlogen lmp. Imges (1 1 pixels, 6 6 mm FOV, 5 ms exposure time) were cquired using high speed CMOS cmer (Micm Ultim, Scimedi). Anlyses were sed on lech-corrected verge of 1 trils. Blech correction (doule exponentil fit), conversion to chnges in VSD fluorescence intensity (% F/F ) nd region-of-interest nlysis were performed using PMOD (PMOD Technolgies Ltd). Multi-unit recordings from the CST. Animls under deep isoflurne nesthesi (ventilted, monitoring rteril pressure) were fixed on the modified stereotxic frme s mentioned ove, with the clmps plced on the thorcic verter. A lminectomy ws performed t T7. A 16-chnnel NeuroNexus proe (16 proes spced t 1 µm, M, NeuroNexus Technologies) coted with DiI ws used to record multi-unit responses nd descended into the dorsl funiculus.15 mm wy from the midline. Prior to recording, isoflurne nesthesi ws grdully replced with Urethne (1,2 mg kg 1 intrperitonel, Sigm- Aldrich) nd the muscle relxnt pncuronium romide injected intrvenous (.1 mg kg 1, Pvulon, Orgnon). Upon forepw stimultion in injured nimls nd fore- or hindpw stimultion in intct nimls t 6 ma (1 ms), ction potentils were initited. The stimultion mplitude ws sed on previous study 29. Responses from the corticospinl trct were otined from electrodes t depth of mm. Before ltion, the cortex ws stimulted using pltinum tungsten glss coted electrode (elow 1 MΩ) plced in lyer V of the hindlim cortex or using silver ll electrode (.5 mm dimter) plced on the doi:1.138/nn.2448 nture NEUROSCIENCE