Research Article

Asynchronous therapy restores motor control by rewiring of the rat corticospinal tract after stroke

See allHide authors and affiliations

Science  13 Jun 2014:
Vol. 344, Issue 6189, pp. 1250-1255
DOI: 10.1126/science.1253050

Improving stroke recovery by timing treatment

Patients recovering from strokes often fight a long uphill battle, with mixed results. Studying the effect of physical training on regeneration from damaged nerves in a model of stroke in rats, Wahl et al. show that timing matters. First, the researchers gave the rats a stroke, which damaged their ability to reach for food pellets with their forelimbs. The researchers then gave them physical training and treated them with an antibody to encourage neural regeneration. The rats improved more when the researchers waited until after the antibody treatment to start the training. Damaged circuits, it seems, need a little time to regrow before being called into action.

Science, this issue p. 1250

Abstract

The brain exhibits limited capacity for spontaneous restoration of lost motor functions after stroke. Rehabilitation is the prevailing clinical approach to augment functional recovery, but the scientific basis is poorly understood. Here, we show nearly full recovery of skilled forelimb functions in rats with large strokes when a growth-promoting immunotherapy against a neurite growth–inhibitory protein was applied to boost the sprouting of new fibers, before stabilizing the newly formed circuits by intensive training. In contrast, early high-intensity training during the growth phase destroyed the effect and led to aberrant fiber patterns. Pharmacogenetic experiments identified a subset of corticospinal fibers originating in the intact half of the forebrain, side-switching in the spinal cord to newly innervate the impaired limb and restore skilled motor function.

Stroke is a major cause of severe disability in the elderly population, and recovery after large strokes is limited (1, 2). Current strategies to improve long-term outcome in humans include mostly rehabilitative training and, in experimental models, electrical stimulation and pharmacological interventions (3). However, all of these treatment options have had only limited success thus far (4, 5). Here, we show that rehabilitative training, if preceded by a nerve growth–promoting antibody therapy, almost completely restored skilled forelimb functions after cortical strokes in adult rats. Sequential application of the treatments was essential: When growth promotion by blockade of the neurite growth–inhibitory protein Nogo-A was simultaneously applied with intensive forced-use training of the forelimb during the first 2 weeks after the stroke, functional outcome was poorer compared with training, immunotherapy alone, or no treatment at all (6). Anatomically, Nogo-A neutralization promoted growth of corticospinal fibers from the intact forebrain motor cortex across the midline of the cervical spinal cord. In rats with simultaneous antibody treatment and training, fiber branching was abundant with anatomically aberrant terminations. In contrast, in animals trained for forelimb function subsequent to antibody treatment, axonal fibers originally terminating in the intact spinal hemicord crossed the midline and innervated the ventral motor regions of the spinal hemicord that had lost its input from the motor cortex. To prove the functional relevance of these newly grown, “side-switched” descending corticospinal tract (CST) fibers, we selectively and temporarily blocked these fibers by two different pharmacogenetic techniques, both of which suppressed the restored forelimb function. Our results demonstrate that a sequential strategy of first promoting fiber growth to enhance the low endogenous plastic potential of the brain and spinal cord, followed by rehabilitative training–induced selection and stabilization of functionally meaningful connections can lead to much higher levels of functional restoration after the formation of large brain lesions than currently obtained in conventional rehabilitation medicine.

Success of rehabilitation depends on timing

We compared four different therapy and rehabilitation schedules for promoting functional recovery of fine motor skills of forelimbs in a thrombotic stroke model in rats. Using the well- established technique of photothrombosis (6), we induced blood vessel blockade by multiple microthrombi, which destroyed >90% of the sensory-motor cortex of adult rats. Rats were then treated with intrathecal anti-Nogo-A or control antibody for 2 weeks (6, 7). In addition, we trained rats intensely in skilled forelimb reaching (100 reaches per day), either simultaneously with antibody application (parallel groups) or during the 2 weeks after antibody treatment (sequential groups) (Fig. 1A). To avoid a training effect of testing itself, we did not reassess the sequential groups (anti-Nogo-A/sequential and control sequential) during the first 2 weeks after lesion formation. When the growth-enhancing anti-Nogo-A treatment was followed by the rehabilitative training (anti-Nogo-A/sequential group), animals improved their performance from day 16 poststroke onwards, and their skilled reaching abilities almost completely recovered [reaching 86.3 ± 2.0% of prestroke level; significantly better than all other groups; P < 0.001, two-way repeated measures analysis of variance (ANOVA) with post hoc Bonferroni] (Fig. 1, A and B). This group also performed best in two skilled forelimb use tasks tested at the end of the experiment [Montoya staircase grasping: success rate 34.1 ± 5.1% (Fig. 1C); horizontal ladder crossing: success rate 65.8 ± 3.7% (Fig. 1D)]. As animals had not been exposed to these tasks before, these results indicate a generalization of the recovery of forelimb function in the sequential training group that is transferable to nontrained motor skills.

Fig. 1 Timing matters when a growth-promoting anti-Nogo-A immunotherapy is combined with training.

(A) Success rates in the single-pellet grasping task at baseline (intact, trained), 2 days after a large, unilateral photothrombotic stroke to the sensorimotor cortex of the preferred paw, and during training and retesting sessions until 4 weeks post-insult. The anti-Nogo-A/sequential group showed significant improvement compared with all other groups, whereas the performance of the anti-Nogo-A/parallel group was significantly worse (anti-Nogo-A/parallel, n = 16; control/parallel, n = 8; anti-Nogo-A/sequential, n = 16; and control/sequential, n = 9). (B) Recovery rates expressed as success rates of the last testing session normalized to baseline performance (100%). n.s., not significant. (C and D) Animals in the anti-Nogo-A/sequential group also performed significantly better in grasping tasks, such as the Montoya staircase test (C) or the horizontal ladder crossing task (D), introduced after the completion of the rehabilitation schedules for three consecutive trials. In (A) to (D), data are presented as means ± SEM. Statistical evaluation was carried out with two-way ANOVA repeated measure, followed by Bonferroni post hoc test. Asterisks indicate significances: *P < 0.05, **P < 0.01, ***P < 0.001. (E) Representative picture of an ex vivo magnetic resonance image stack for three-dimensional stroke reconstruction. (F) There was no correlation between stroke volume and end-point success rate in the single-pellet grasping task among all rehabilitation groups (r = 0.04, Spearman correlation).

In contrast to these results, rats receiving intensive forelimb training concurrently to the anti-Nogo-A antibody treatment (anti-Nogo-A/parallel group) performed worse than all other groups in the single-pellet grasping task (success rate 10.0 ± 5.2%) (Fig. 1, A and B). In the tasks evaluated at the end of the experiment, these animals showed either no significant improvement (Fig. 1C) or a tendency to decline over the course of trials (P = 0.1, two-way repeated measures ANOVA with post hoc Bonferroni) (Fig. 1D). The control antibody-treated groups reached low levels of recovery (35 to 40% success rate in pellet grasping) (Fig. 1, A to D), with an early training effect visible in the group trained during the first 2 poststroke weeks (control/parallel) (Fig. 1A). Final success scores did not correlate with stroke volume, as determined ex vivo from either whole-brain magnetic resonance images (r = 0.04, Spearman correlation) (Fig. 1F) or histological Nissl stains (fig. S1), suggesting no specific neuroprotective effect by any of the four therapeutic schedules. We conclude that applying a nerve fiber growth-promoting cellular therapy and rehabilitative physical training in sequence delivers greater functional recovery than when the same protocols are applied concurrently.

Rehabilitative schedules induce distinct neuronal fiber patterns

We investigated the neuroanatomical correlates of functional recovery by labeling the intact, contralesional CST, which normally innervates the spinal cord half opposite to the one that has lost its cortical input, with only few fibers crossing the spinal cord midline. Each of the four experimental groups presented a distinct rehabilitation-induced pattern of fiber sprouting in the cervical spinal cord (Fig. 2C). We counted labeled fibers originating in the cortex of the intact side opposite to the stroke and crossing the midline of the spinal cord (Fig. 2A). We also quantified their elongation and branching within the gray matter of cervical spinal cord, that is, the cord region containing the motor control circuits of the forelimb and paw (Fig. 2, A and B). The greatest number of midline-crossing fibers was seen in the anti-Nogo-A/sequential treatment group, which also had the best functional outcome. In contrast, the anti-Nogo-A/parallel group, showed extensive branching of the midline crossing corticospinal fibers (Fig. 2B) (P < 0.05, two-way repeated measures ANOVA with post hoc Bonferroni).

Fig. 2 Corticospinal tract sprouting depends on timing of rehabilitative training correlating with functional recovery.

The four rehabilitation schedules (anti-Nogo-A/parallel, n = 10; control/parallel, n = 8; anti-Nogo-A/sequential, n = 10; and control/sequential, n = 8) differently influenced the sprouting of corticospinal fibers from the intact side of the spinal cord across the spinal cord midline (M). (A) Low- and high-magnification micrographs of biotinylated dextran amine (BDA)–labeled corticospinal fibers in intact spinal hemicord (left) growing into the stroke-denervated hemicord (right; inset) at spinal cord level C4. D1 to D4: Lines for intersection counts with corticospinal fibers. Scale bar, 200 μm. (B) Fibers crossing the midline (M) and branching in the gray matter at distances D1 to D4 were counted and normalized to the number of BDA-positive labeled fibers in the main tract. (C) Micrographs showing different sprouting patterns of corticospinal fibers from the ipsilateral cortex in the denervated cervical spinal cord (C4) in lamina 7 in the different treatment groups. Scale bars, 200 μm. (D) Combining anti-Nogo-A immunotherapy with simultaneous training (anti-Nogo-A/parallel) results in a significantly higher density of ipsilateral CST fibers in the stroke-denervated cervical spinal cord than does anti-Nogo-A/sequential treatment. (E) The most significant difference in fiber density between anti-Nogo-A/parallel and anti-Nogo-A/sequential animals was detected in lamina 6/7 and lamina 9 of the denervated cervical hemicord. Lamina 7 was also significant for increased fiber branching (F) (branching index = branches per fiber per BDA-positive fibers in the intact CST) and bouton numbers (G) in the anti-Nogo-A/parallel group. (H and I) Significantly more fibers cross the gray matter–white matter boundaries in the dorsolateral (labeled with “A”), the ventrolateral (label “B”), and the ventro-medial funiculus [label “C,” scheme shown in (H)] in the anti-Nogo-A/parallel group. Data are presented as means ± SEM. Statistical evaluation was carried out with two-way ANOVA repeated measure, followed by Bonferroni post hoc (B) and Student’s t test (two-tailed, unpaired) (D to G and I). Asterisks indicate significances: *P < 0.05, **P < 0.01, ***P < 0.001.

A quantitative analysis of the distribution and density of ipsilaterally projecting corticospinal fibers using pattern-recognition algorithms to analyze both single corticospinal fibers and related fiber-growth parameters (see supplementary materials and methods) confirmed overshooting fiber growth and aberrant termination patterns in the anti-Nogo-A/parallel group. In the anti-Nogo-A/sequential group, midline-crossing, sprouting corticospinal fibers displayed a radial organization with few branches and a preference for the premotor and motor spinal cord (laminae 6 to 9) (Fig. 2, D and G). In contrast, fibers in the anti-Nogo-A/parallel group appeared less organized with more than double the number of branches and a different laminar distribution including the dorsal, predominantly sensory laminae 1 to 5. We also assessed the connectivity of the ipsilaterally projecting corticospinal fibers by quantifying the density of axonal boutons recognized morphologically in the premotor interneuron lamina 7: We detected a significantly higher bouton density in the anti-Nogo-A/parallel group compared with the anti-Nogo-A/sequential group (Fig. 2H) (P < 0.05, Student’s t test, two-tailed, unpaired). The anti-Nogo-A/parallel group showed a greater tendency of axons to grow beyond the gray matter–white matter boundary, as well as a highly aberrant growth pattern (Fig. 2, H and I). In the medio-ventral funiculus, such fibers are probably intermixed with sprouts of the small, uncrossed ipsilateral CST.

Nerve cells from the intact forebrain cortex are responsible for recovery

Our results suggest that the recovery of rat forelimb function after stroke in the anti-Nogo-A/sequential group originates from extensive and precise reinnervation of the stroke-denervated spinal hemicord by midline-crossing fibers from the intact motor cortex and CST. We tested this hypothesis in the animals of the anti-Nogo-A/sequential group, all of which showed excellent functional recovery, by using two different experimental approaches for inducible, selective, and reversible inactivation of the ipsilaterally projecting corticospinal fibers on the long and short term, respectively. For long-term blockade, we used a virus to deliver a doxycyclin-inducible tetanus toxin to temporarily inactivate the synaptic release mechanism (8). We injected the highly efficient retrograde gene transfer lentivector HiRet carrying enhanced tetanus neurotoxin light chain (eTeNT) with an enhanced green fluorescent protein (EGFP) downstream of a tetracycline-responsive element (TRE) into the stroke-denervated side of the cervical spinal cord at level C5-C6, and we injected the adeno-associated serotype 2.2 (AAV2) vector carrying the reverse tetracycline transactivator (rtTAV16, Tet-on) into the contralesional, intact premotor and motor cortex (n = 6 animals) (Fig. 3, A and B). Only cortical neurons with axons projecting to the stroke-denervated spinal cord would contain both transgenes and activate tetanus toxin in response to doxycycline. We applied the same procedure in four control animals, except these animals received injections of the HiRet lentivector coding for EGFP only.

Fig. 3 Long-term reversible blockade of midline-crossing corticospinal fibers abolishes the functional recovery after stroke.

(A) Experimental schedule. A­ll animals had anti-Nogo-A immunotherapy followed by intensive reaching training, leading to almost full functional restoration of skilled forelimb functions. (B) Schematic diagram of vector injections into the contralesional motor cortex (AAV2.2-CMV-rtTAV-16) and stroke-denervated cervical spinal cord (HiRet-TRE-EGFP-TeNT or HiRet-TRE-EGFP as control). (C) Induction of tetanus toxin by doxycycline leads to strong impairment of reach and grasping movements within 7 days. The effect was fully reversible within 7 to 10 days of dox removal and could be reinduced by reapplication of dox. TeNT animals, n = 6; control animals, n = 4. (D) Examples of GFP-positive cells (i) in the rostral sensorimotor cortex, 3.7 mm anterior to bregma in comparison to Nissl-positive cells (ii) [scale bars for (i) and (ii), 100 μm], and zooming in to GFP-positive cells in layer five (iii) in relation to Nissl-positive cells (iv) [scale bars for (iii) and (iv), 50 μm]. Roman numerals I to V represent five out of six histological layers in the motor cortex. (E) Quantification of GFP-positive cells in percentage of Nissl-positive cells in the layer five motor cortex. Data are presented as means ± SEM. Statistical evaluation was carried out with two-way ANOVA repeated measure, followed by Bonferroni post hoc test. Asterisks indicate significances: *P < 0.05, **P < 0.01, ***P < 0.001.

After recovery from the surgery and the reassessment of regained grasping skills, doxycycline was orally administered for 2 weeks (Fig. 3, A and C). In the experimental group grasping performance declined within a few days, reaching a very low level from day 7 after doxycycline initiation (P < 0.05, statistical comparison TeNT versus control group, two-way repeated measures ANOVA with post hoc Bonferroni) (Fig. 3C).When the drug administration was ceased, the lost fuction was regained within 2 weeks. All animals again showed a loss of skilled food-pellet grasping movements when oral doxycycline intake was restarted for a second time over the course of another 3 weeks (Fig. 3C). No deterioration of the poststroke recovered skilled grasping was observed in control animals (EGFP instead of eTeNT) under the same dosage of doxycycline and within the same time frame (98.1 ± 0.9% of pre-doxycycline grasping performance). As the retrogradely transported virus was EGFP-tagged, corticospinal neurons projecting to the ipsilateral cervical spinal cord segments C5 and C6 could be quantified. These neurons were concentrated in layer five of a specific region of the rostral, premotor, and primary (M1) forelimb motor cortex (3.7 ± 0.2 mm anterior to bregma). In the center of the labeled region, 41.4 ± 5.6% of Nissl-positive cells of layer five contained the transgene (Fig. 3, D and E). These results demonstrate that midline-crossing corticospinal fibers from the intact hemisphere opposite to the stroke lesion anatomically and functionally switch sides, which is crucial for the recovery of skilled forelimb movements. These fibers maintain their new function, even if they are functionally blocked for weeks. Evidently, their role cannot be compensated by other cortical or subcortical motoneuronal pathways in the injured system.

Temporally blocking rewired corticospinal fibers results in decline of regained function

For short-term reversible inactivation of the midline-crossing corticospinal fibers, we used a pharmacogenetic approach involving virus-mediated gene transfer to express engineered Gi/o-coupled DREADD receptors (“designer receptor exclusively activated by designer drug”). These receptors are only activated by the otherwise pharmacologically inert synthetic ligand clozapine-N-oxide (CNO), resulting in increased intracellular Gi/o-mediated signaling, which leads to membrane hyperpolarization and silencing of the infected neurons (911). We used only rats that had undergone the treatment schedule of growth promotion by anti-Nogo-A antibodies, followed by another 2 weeks of rehabilitative training. The virus injections started at the end of the training period (fig. S2A). To selectively manipulate the corticospinal fibers projecting from the intact, contralesional motor cortex to the denervated cervical spinal cord, we first injected an AAV2.9 vector carrying the Cre sequence into segments C5 and C6 of the stroke-denervated cervical spinal cord, followed by injections of the Cre-dependent AAV2.1 vector carrying the Gi/o-coupled DREADD (hM4Di) receptor into the contralesional pre- and sensorimotor cortex (n = 6) (fig. S2B). The DREADD receptor hM4Di was tagged with mCherry, which allowed neuroanatomical confirmation that only double-infected cells expressed the hM4Di receptor (fig. S2C): We found mCherry-positive cells concentrated in layer five of the same region of the contralesional premotor and motor cortex (3.7 ± 0.2 mm anterior to bregma), as in the tetanus toxin experiment. A portion of Nissl-positive layer five cells in this region (31.6 ± 2.4%) were positive for mCherry (Fig. 4, A and B), with very limited numbers of mCherry-positive cells outside of this area. In control animals that received injection of the Cre-dependent AAV2.1-hM4Di vector in the contralesional cortex without AAV-Cre virus injection in the spinal cord, only 4.3 ± 0.5% of Nissl-positive cells in layer five were positive for mCherry, indicating low background noise (n = 4) (Fig. 4A).

Fig. 4 Short-term reversible blockade of ipsilateral corticospinal fibers abolishes recovered grasping function and forelimb EMGs.

(A and B) Quantification of mCherry-positive cells in percentage of Nissl-positive cells in layer five of contralesional premotor (M2) and motor cortex (M1), 3.7 mm anterior to bregma in the hM4Di/+Cre group (n = 6) and the hM4Di/–Cre control group (n = 4) (A), and the illustration of its location using Neurolucida software reconstruction (B). (C) Activation of the DREADD receptor in the hM4Di/+Cre group induced a rapid and massive but fully reversible impairment of grasping performance 10 to 40 min after CNO application. Reaching success rates were unchanged in the control hM4Di/–Cre CNO-treated group over the same time frame. (D) CNO application disturbed fine motor function of closing the paw around the pellet in the hM4Di/+Cre group (P = 0.02, Kolmogorov-Smirnov test): The figure shows the spatial probability densities for the location of hand closure relative to the sugar pellet during grasping at baseline and after CNO application. “x” represents the position of the sugar pellet relative to the forelimb position during grasping. Scale bar, 10 mm. (E) CNO application leads to a decrease of EMG responses in the hM4Di/+Cre group (n = 5) after ICMS of the contralesional motor cortex compared with ICMS stimulation at baseline and 30 min after CNO application in control animals (n = 3). Heat maps of the cortical stimulation grid are shown (60 stimulation points, 80 μA, +3 to –3 mm anterior-posterior and 1 to 3.5 mm medio-lateral relative to bregma). Each stimulation point is color coded with the mean value of EMG response for a muscle group (wrist, elbow, shoulder) at this stimulation point (in millivolts) normalized to the mean of all stimulation points at the baseline. Data are presented as means ± SEM. Statistical evaluation was carried out with two-way ANOVA repeated measure, followed by Bonferroni post hoc test. Asterisks indicate significances: ***P < 0.001.

Three weeks after virus injection, both hM4Di/+Cre and control hM4Di/–Cre animals performed at >80% success rate in single-pellet grasping (Fig. 4C). Animals were then injected intraperitoneally with the channel-activating drug CNO. Control animals showed no change in reaching and grasping abilities over the 50 min of observation time. However, animals of the hM4Di/+Cre group lost their grasping abilities over 10 to 30 min, with performance declining to 38.9 ± 6.0% success rate, which is significantly lower compared with the control group (Fig. 4C) (P < 0.001, two-way repeated measures ANOVA with post hoc Bonferroni). The defective movements were characterized by a specific failure to target the paw to the pellet and to close the paw around the pellet (Fig. 4D), but with little modifications in overall grasping trajectories (fig. S3). The abatement of distal motor functions was confirmed by performing discriminative classification based on a nonparametric representation (12) of paw posture and its change over time (P = 0.02, grasps at baseline versus after 30 min CNO, Kolmogorov-Smirnov test). This mainly distal impairment may also be due to the fact that only the cervical segments C5 and C6 were injected with the Cre virus. These functional defects were fully reversible, with performance returning to preinjection levels ~40 to 50 min after the CNO injection (Fig. 4, C and D).

Pharmacogenetic inhibition of regained EMG activity

We confirmed the CNO-specific blockade of neuronal firing of ipsilaterally projecting, partially midline-crossing corticospinal fibers of the intact, contralesional motor cortex by electrophysiology using intracortical microstimulation (ICMS) at the end of the behavioral testing: We used a 5-by-12–point stimulation grid (positioned at +3 to –3 mm anterio-posterior and 1 to 3.5 mm medio-lateral relative to bregma) (fig. S4A) and electromyogram (EMG) recordings of wrist, elbow, and shoulder muscles of the impaired paw as readouts (fig. S4B). In all animals, each cortical position within the exploration grid was stimulated twice, first as baseline stimulation, then again 30 min after CNO injection. In all hM4Di/–Cre control animals, the EMG responses at baseline and 30 min after CNO injection were not significantly different for the 60 stimulation points [for wrist, 95% confidence interval between baseline and 30 min CNO was for animal one: 0 (0 to 0.001), animal two: 0 (0 to 0), animal three: 0 (0 to 0.001); one-sample Wilcoxon signed rank test] (Fig. 4E and fig. S4C). In contrast, in four out of five hM4Di/+Cre animals, the ICMS-evoked EMG responses significantly decreased 30 min after CNO injection compared with baseline [for wrist, 95% confidence interval was for animal one: –0.002 (–0.002 to –0.001), animal two: –0.004 (–0.005 to –0.003), animal three: –0.002 (–0.002 to –0.002), animal four: 0 (0 to 0.001), animal five: –0.012 (–0.016 to –0.009); one-sample Wilcoxon signed rank test] (Fig. 4E and fig. S4D). Calculating the mean EMG response for every stimulation point in hM4Di/+Cre animals versus hM4Di/–Cre controls revealed a significant decline of EMG responses in wrist and elbow muscles in the hM4Di/+Cre group 30 min after CNO application compared with controls (P < 0.05, Mann-Whitney test) (Fig. 4E). No significant abatement of EMG responses occurred in shoulder muscles of hM4Di/+Cre animals (P = 0.4, Mann-Whitney test) (Fig. 4E). CNO application resulted in the largest difference in EMG responses of wrist recordings when premotor and rostral forelimb areas of the contralesional motor cortex were stimulated in the hM4Di/+Cre group compared with controls (fig. S2D). These areas also expressed the highest concentration of mCherry-expressing cells (Fig. 4A).

Discussion

Our study shows that in a rat model of large forebrain cortex strokes, timing of rehabilitative training relative to timing of a nerve fiber growth-promoting therapy affects the recovery of lost motor function and the pattern of fiber sprouting. When rats received their first anti-Nogo-A immunotherapy followed by 2 weeks of specific, intense rehabilitative training, forelimb function was almost fully restored (Fig. 1, A and B), indicating a far more extensive recovery rate relative to stroke size than previously obtained by training (1315) or growth-promoting therapy alone (6, 16). Not only did these animals outperform the other rehabilitation groups in the single-pellet grasping task, but they were also able to better transfer their regained skills to novel forelimb tasks (15). The behavioral recovery was associated with crossing of CST fibers from the lesion-spared, intact motor cortex to the stroke-denervated side of the spinal cord. This observation is supported by various stroke and spinal cord injury models (1519) that relate midline-crossing corticospinal fibers to functional outcome. Our analysis shows that not only the quantity of newly out-sprouting corticospinal fibers is relevant but also their termination pattern: Intensive training, when applied too early, induced hyperinnervation and aberrant growth even beyond the gray matter–white matter boundary and into dorsal sensory laminae. Such widespread sprouting may result in wrong circuit connectivity involving cervical interneurons, V2a propriospinal neurons, and motoneurons, thus impairing grasping function, for example, by coactivation of agonistic and antagonistic muscles (20, 21).

Our data suggest the presence of critical time windows during which the brain is most responsive to the application of growth-promoting agents and to training-dependent plasticity. A correct, timed sequence of interventions is required to maximize the effectiveness of rehabilitative therapies after stroke. In a first step, suppression of the action of the endogenous growth-inhibitory factor Nogo-A by immunotherapy may diminish constraints on lesion-induced structural plasticity through mechanisms such as neurotrophic factors, modified electrical properties of motoneurons, alteration in neuronal energy balance (22), and recruitment of new circuits leading to hyperexcitability and prolonged responses to external stimuli (1). In analogy to development, many of these newly formed connections may be weak and imprecise. Training in a second step may then help to shape the spared and new circuits by selection and stabilization of functional connections and pruning of the nonfunctional ones. This second step might involve Hebbian learning rules, in the sense that Hebbian plasticity redistributes synaptic strength to favor functionally relevant pathways that are coincidently active (1). The high degree of recovery of important, cortically controlled motor functions in rats with large ischemic strokes, as demonstrated here, points to a possible avenue to explore growth-inhibitor blockade in combination with rehabilitative training as a treatment strategy for humans with motor cortex stroke. Antibodies against Nogo-A are currently used in clinical trials in humans for amyotrophic lateral sclerosis, multiple sclerosis, and spinal cord injury (www.clinicaltrials.gov). Careful consideration of rehabilitation onset times—particularly with regard to windows of sprouting and circuit plasticity, but also vulnerability of the injured brain—and tailored training adapted to the type and extent of stroke and the patient’s history will be essential for future approaches in the clinic (2, 23).

Supplementary Materials

www.sciencemag.org/content/344/6189/1250/suppl/DC1

Materials and Methods

Fig. S1 to S4

References (2434)

References and Notes

  1. Acknowledgments: We thank T. Isa from the National Institute for Physiological Sciences, Okazaki, Japan, as well as K. Kobayashi, Fukushima Medical University, and D. Watanabe, Kyoto University, for kindly providing plasmids for viral vector production. We thank H. Kasper, C. Bleul, and N. Lindau for technical advice and fruitful discussions, as well as B. Seifert for statistical assistance. This work is supported by the European Union grants FP7 Collaborative Projects ARISE (201024) and PLASTICISE (223524), the advanced European Research Council grant NOGORISE (to M.E.S.), the Swiss National Science Foundation grants Nr. 31-138676 and 3100A0_12252711 (to M.E.S.) and Nr. 31003A_149858 (to F.H.), the Christopher and Dana Reeve Foundation (to M.E.S.), and the Dr. Wilhelm Hurka Foundation (F.H. and W.O.). The work by J.C.R and B.O. was supported by the German Excellence Initiative, Deutsche Forschungsgemeinschaft project number 49/2. A.S.W. and M.E.S. designed the study; A.S.W. and M.G. carried out experiments; A.S.W., W.O., B.O., and F.H. performed data analysis; J.C.R., H.Z., and B.O. developed computer and machine learning algorithm tools for data analysis; J.L.C. and K.K. provided virus vectors; A.S. performed magnetic resonance imaging; O.W. carried out histological analysis; and A.S.W, B.O., F.H., and M.E.S. prepared figures and wrote the manuscript. The University of Zurich holds joint patents with Novartis Pharma for antibodies against Nogo-A and their use in neurological diseases. Otherwise, we have no patents pending or financial conflicts to disclose. Materials and methods are available as supplementary materials on Science Online. Plasmids for lentiviral and AAV vectors for the TeNT experiment were obtained under material transfer agreements (MTAs) with Kyoto University, Japan (pLenti-TRE-EGFP-eTeNT-PEST-WPRE, pLenti-TRE-EGFP-WPRE, pRSV-Rev, pMDLg/pRRE, pAAV2-RC, pAAV2-CMV-rtTAV16), and Fukushima Medical University, Japan (pCAGGS-FuG-B2), respectively. Viral vectors for the DREADD experiment were obtained under a MTA with the University of Pennsylvania, Philadelphia, PA (AAV2.9-CamKIIα.-Cre vector). The anti-Nogo-A antibody (11C7) was a gift from Novartis Pharma. The supplementary materials contain additional data.
View Abstract

Navigate This Article