Rehabilitative Training Promotes Rapid Motor Recovery but Delayed Motor Map Reorganization in a Rat Cortical Ischemic Infarct Model

Subjects and Group Assignments

A total of 25 adult, male, Long-Evans hooded rats (Harlan; 4-5 month old, 300-400 g) was used in accordance with National Institutes of Health regulations, and approved by the Institutional Animal Care and Use Committee of the University of Kansas Medical Center. Rats were singly housed in a Plexiglas cage with ad libitum food and water on a 12 hour:12 hour light–dark cycle, with ambient temperature maintained at 68 to 71°F. Animals were randomly assigned to 1 of 4 post-lesion groups that varied by survival time (18 or 38 days) and postinjury behavioral experience (rehabilitative training, referred to as “rehab” group, or no rehabilitative training, referred to as “no-rehab” group). Thus, there were 4 groups: (a) rehab/short-term survival (n = 6), (b) rehab/long-term survival (n = 7), (c) no-rehab/short-term survival (n = 6), and (d) no-rehab/long-term survival (n = 6). Through post-lesion day (PLD) 18, rats in the 2 rehab groups received the same preoperative and postoperative experiences; rats in the 2 no-rehab groups also received the same experiences. Therefore, these groups were combined as rehab (n = 13) and no-rehab (n = 12) groups for analyses through PLD 18.

Pre-Infarct Behavioral Training and Motor Assessment

Postinfarct Rehabilitative Training

Rehabilitative training was initiated in the rehab groups on PLD 8. On PLD 8 to 12, a tray-reaching task was used. This task requires less precise reaching and grasping than the single-pellet retrieval task since rats retrieve pellets for 20 minutes from a tray filled with pellets. ¹³ On PLD 13 to 17, the single-pellet retrieval task was implemented (60 training trials per day). Rats assigned to the no-rehab groups had a similar number of food pellets available from the floor of the reaching box. Randomly selected videos of motor performance and movement kinematics were independently scored by separate examiners blind to the experimental condition as a reliability check.

Cortical Infarct Procedure

Anesthesia was induced with 3% isoflurane gas followed by ketamine (100 mg/kg, intraperitoneal) and xylazine (5 mg/kg, intramuscular). Additional doses of ketamine (20 mg/kg intramuscular) were used as needed. Six 0.7-mm diameter holes were drilled over the CFA contralateral to the dominant forelimb at anteroposterior +1.5, +0.5, and −0.5 mm and mediolateral +2.5 and +3.5 mm from bregma. ¹⁴ To induce cortical ischemia, 0.33 µL of endothelin-1 (ET-1; Bachem Laboratories, 0.3 mg/mL) was injected into each hole at a depth of 1.5 mm from the cortical surface, through a micropipette (160 µm o.d.) attached to a Hamilton syringe using a microsyringe injector (UltraMicro Pump III, World Precision Instruments). Appropriate postoperative care was provided under veterinary supervision.

Postinfarct Neurophysiological Assessment

Standard intracortical microstimulation (ICMS) techniques were used to derive forelimb movement maps in the cortex rostral to the ischemic lesion. ⁸ Rats in short-term and long-term survival groups underwent an ICMS mapping procedure on PLD 18 and 38, respectively. Rats were anesthetized with ketamine/xylazine and secured in a stereotaxic frame. Anesthesia was maintained throughout the procedure with bolus injections of ketamine (20 mg/kg, intramuscular) as needed to minimize spontaneous movements and toe pinch reflex. A craniectomy was performed over the frontal cortex. A digital image of the surface vasculature was obtained and imported into a graphics program to guide the placement of the microelectrode on a 250 µm grid pattern. A glass microelectrode (tapered to 15-20 µm o.d. with beveled tip; impedance = 500-700 kΩ) was filled with 3.5 M NaCl and connected to a constant-current stimulator (Model BSI-2, BAK Electronics) through a platinum wire. The electrode was lowered to 1700 µm (approximately Layer V) using a hydraulic microdrive (Model 650, David Kopf Instruments). The ICMS stimulus consisted of 13 cathodal pulses (200 µs each) delivered at 350 Hz. Movements and their threshold current levels (maximum = 80 µA) were recorded for each stimulation site. ICMS-evoked movements were defined via visual observation by an observer blind to the electrode placement. A second observer, blind to the experimental condition, verified the evoked movement. Distal forelimb movements were defined as visually observable movements of the wrist (extension, flexion, supination, pronation) and/or digits (extension, flexion). Proximal forelimb movements were defined as visually observable movements of the elbow (extension, flexion) and/or shoulder (extension, flexion). Movement representation maps were reconstructed from movement topography and areal extents measured with imaging software (NIH IMAGE vl.61). Rostral, caudal, medial, and lateral extents of forelimb representations were recorded based on the boundaries of reconstructed maps (rostral, caudal relative to bregma; medial, lateral relative to midline).

Histology

Immediately following the ICMS mapping procedure, rats were euthanized by an overdose of Buthanasia and perfused transcardially with normal saline followed by 4% paraformaldehyde in 0.1 M phosphate-buffered saline. Brains were postfixed in 20% glycerol, sectioned coronally (30 µm), and stained with cresyl violet. Lesion volume estimation was obtained by the difference of the cortical volume of the injured hemisphere subtracted from that of the intact hemisphere, ⁸ using the Cavalieri method in StereoInvestigator (Microbrightfield, Inc).

Statistical Analysis

Statistical analyses were performed using JMP v10.0 (SAS Institute, Inc, Cary, NC). Two-way ANOVAs were used to examine the effects of Group, Time, and Group × Time interactions on lesion volume, motor performance, kinematic scores, topography of forelimb motor maps, and ICMS current thresholds. Post hoc comparisons were performed with Tukey tests when appropriate (α = .05).

Histological Results

The lesion extended through all cortical layers leaving the underlying white matter intact in all infarcted rats (Figure 1). No cortical damage was evident at the level of RFA in any of the cases. While the main effect of Group was not statistically significant (F_1,21 = 3.9206; P = .0603), mean lesion volume in the rehab group was 36% larger than in the no-rehab group.

OPEN IN VIEWER

Figure 1.

Histological results.

Effects of Rehabilitative Training on Motor Behavior

Behavioral results are described separately for 3 post-lesion phases (Figure 2): (a) Lesion phase (PLD −1 to PLD 7); (b) Rehabilitative training phase (single pellet retrieval training; PLD 12 to PLD 17); and (c) Follow-up phase (PLD 22 to 37).

OPEN IN VIEWER

Figure 2.

Motor performance results.

Movement Kinematics on Single-Pellet Retrieval Task

Lesion phase

There was a significant effect of Time (lesion effect) for supinate I (F_1,10 = 21.922, P = .0009), supinate II (F_1,10 = 44.705, P < .0001), and release (F_1,10 = 26.406, P = .0004) (Figure 3). The effect of Time for grasp approached significance (F_1,10 = 4.755, P = .054). There was no effect of Time for pronate, advance, or digit extend. Thus, the remaining analyses were limited to supinate I, supinate II, release, and grasp.

OPEN IN VIEWER

Figure 3.

Kinematic results.

Rehabilitative training phase

The effects of Time on supinate I approached significance (F_1,10 = 4.54, P = .059). Significant Group differences were found for supinate I (F_1,10 = 6.47, P = .029), supinate II (F_1,10 = 23.17, P = .0007), and release (F_1,10 = 30.39, P = .0003), reflecting superior kinematic endpoints in the rehab group.

Follow-up phase

Group differences favoring the rehab group persisted for supinate I (F_3,30 = 15.5, P = .0005), supinate II (F_3,30 = 18.40, P = .0002), and release (F_3,30 = 25.42, P < .0001). There was an effect of Time for release (F_3,30 = 3.62, P = .024). Post hoc analysis comparing PLD22 to PLD37 revealed that both rehab and no-rehab groups showed significant improvement in release.

Effects of Rehabilitative Training on Motor Output Maps

Location of Forelimb Representations

ICMS maps were successfully obtained in 21 of 25 rats. Two rats died prior to the mapping procedure (long-term rehab group = 1, long-term no-rehab group = 1). In 2 additional rats (long-term rehab group = 1, long-term no-rehab group = 1), no evoked movements were observed from ICMS stimulation at the maximum current level (80 µA). Such outcomes are not uncommon in ICMS experiments and are typically attributable to improper anesthetic depth that cannot be corrected during the course of the procedure. ⁸ Subsequent analyses focused on the 21 rats with successful maps.

On both PLD 18 and PLD 38, the spared forelimb motor representations were largely contained within the typical RFA location (Figure 4). ⁸ In both rehab and no-rehab groups, neck, jaw, orofacial, and vibrissae movements were evoked on the medial, rostral, and lateral borders of the forelimb representation (as well as the caudal border in rats with more rostrally located forelimb maps). Because only the forelimb representation was mapped in its entirety, we did not examine these additional representations systematically.

OPEN IN VIEWER

Figure 4.

Forelimb movement representations rostral to infarct.

On PLD 18, in 2 rats in each group, movements could not be evoked rostral to the forelimb representation using the maximum current (80 µA). In 3 rats in each group, the forelimb representation extended further caudally to include the peri-infarct area immediately rostral to the infarct, where neck and orofacial representations typically are found.^11-13 As this location shift was relatively small (<500 µm), we cannot rule out the roles of tissue cavitation (in the infarcted area), edema, or inflammation in such shifts. Thus, since it could not be determined that these forelimb representations were entirely within RFA, we conservatively refer to forelimb representations rostral to the lesion.

On PLD 38, in 3 rats in the rehab group, forelimb movements were evoked caudal to the typical RFA territory. There was a significant effect of Time in the rostral extent of forelimb movement maps (F_3,20 = 5.7161, P = .0287), but no effect of Group nor Group × Time interaction. The map extended more rostrally in long-term versus short-term survival groups. There were no significant differences in medial or lateral extent of the maps.

Size of Forelimb Representations

On PLD 18, the total forelimb area rostral to the lesion in the rehab and no-rehab groups was 0.34 ± 0.06 mm² (mean ± SEM) and 0.54 ± 0.05 mm², respectively. This contrasts with the much larger RFA area in historical controls (0.95 ± 0.7 mm²).^8,15 On PLD 38, the total RFA area was 1.08 ± 0.11 mm (14% larger than historical controls) and 0.60 ± 0.09 mm (37% smaller than historical controls) in the rehab and no-rehab groups, respectively.

There was a significant effect of Time in the total forelimb area rostral to the lesion (F_3,20 = 23.4417, P = .0002) and a significant Group × Time interaction (F_3,20 = 19.5069, P = .0004). Post hoc tests indicate that the forelimb area of the rehab/PLD 38 group was larger than any of the other 3 groups (P < .05, Tukey’s HSD; Figure 5A).

OPEN IN VIEWER

Figure 5.

Motor map areas and thresholds.

Further analysis to differentiate distal versus proximal representations revealed that most of the change in the total forelimb area was due to changes in the distal area. While the effect of Group approached significance (F_3,20 = 3.9535, P = .0631), there was a significant effect of Time (F_3,20 = 16.6762, P = .0008) and a significant Group × Time interaction (F_3,20 = 18.4697, P = .0005). Post hoc tests showed that the distal forelimb area was larger in the rehab/38 day group than any of the other 3 groups (P < .05). There were no significant effects for proximal forelimb area.

There were no significant linear correlations between map size and final performance or kinematic scores. The strongest relationship was between the distal forelimb area and release scores (PLD 38 rats; F = 2.5018, P = .1524).

Effects of Rehabilitative Training on Motor Performance and Movement Quality

The performance deficits and training-induced performance gains observed in this study are similar to those reported previously in rats after motor cortex lesions, though the severity of initial deficits and the time course of improvements are largely dependent on lesion size and location.^16-19 Altered movement kinematic patterns during the single-pellet retrieval task also have been demonstrated frequently in rats after focal cortical lesions in motor cortex, especially in supination/pronation, aim, grasp, and release.^11,16,17,20 Some impairments, such as grasp kinematics, are much more severe in rats with subcortical infarcts. ²¹ Since the rehabilitative training was focused on the retrieval task, requiring skilled use of the digits, it is not surprising that rehabilitative training did not affect performance on the foot-fault task. This suggests that training effects are task-specific and do not generalize to tasks that do not require skilled digit use.

It is typically assumed that motor performance gains are due to a combination of compensation (use of alternative kinematic patterns) and true recovery (return of baseline kinematic patterns).^16,17,22-27 One of the key translational questions regarding poststroke rehabilitation is the following: Can rehabilitative interventions result in more extensive recovery of normal kinematic patterns, that is, movement quality? In the present study, rehabilitative training resulted in a sustained improvement in kinematics, in addition to greater and more rapid gains in functional performance. In a previous study after motor cortex lesions in rats, neither motor performance nor normalization of kinematic patterns was aided by poststroke practice. ¹⁶ Also, a recent study of chronic human stroke survivors undergoing a constraint-induced movement therapy intervention demonstrated improvement in functional outcomes, but no improvement in kinematic outcomes. ²⁸ However, other studies have demonstrated that behavioral experience can alter not only the trajectory of motor recovery but can at least partially reduce compensatory movement patterns.^17,18

It can be argued that performance gains by rats in the no-rehab group in the present study were largely compensatory. But the present results suggest that rehabilitative training provides benefits beyond simply improving compensatory skills and may promote true recovery. The results parallel those in a human stroke population in which motor performance improved over time, but those subjects with the greatest recovery of normal kinematic patterns had the most extensive functional improvement. ²⁹

Temporary Disruption of Spared Motor Maps After Motor Cortex Lesions

The reduced size of forelimb representations on PLD 18 suggests that, at least for a few weeks after the infarct, the functional integrity of cortex rostral to the lesion (including RFA) is disrupted, perhaps by a diaschisis-like effect. Focal cortical infarcts produce hypometabolism throughout a large region of ipsilesional cortex, closely corresponding to areas with known corticocortical connections with the infarct core. ³⁰ Thus, the functional integrity of neurons connected to the infarct is likely to be compromised. Since the CFA and RFA have dense reciprocal interconnections, neurons involved with motor control of the forelimb are most likely to be disrupted, allowing a competitive advantage for neurons more involved with motor control of more proximal (neck, orofacial, vibrissae) musculature. Similar hypotheses have been proposed to explain map changes after focal cortical impact injuries. ⁸ Such injuries result in widespread sublethal effects in the cortex ipsilateral to the damage. ³¹ It has long been known that the overlap of face/neck and forelimb representations in motor cortex provides a substrate for rapid map plasticity. Forelimb motor sites can convert to face/neck motor sites and vice versa within hours.^32,33 If sublethal effects differentially compromise forelimb related neurons in RFA due to the dense reciprocal connections with CFA, then more proximal movement fields are likely to emerge. The present results suggest that this bias toward face/neck representations persists at least for a few weeks after the injury.

Temporal Relationship Between Cortical Plasticity and Behavioral Recovery

While forelimb motor maps were substantially smaller on PLD 18, motor performance had already plateaued. Thus, behavioral improvement preceded map expansion. Similar results were found in the supplementary motor area of monkeys undergoing spontaneous recovery after infarcts that damaged primary motor cortex and nearby premotor areas. ⁶ It was suggested that undetected changes in behavior, such as improvement in kinematic patterns, may have occurred during later stages and that late motor map reorganization may reflect improved movement quality.

Part of the rationale for conducting the present study was to determine whether late motor map changes reflect late improvements in kinematic patterns. However, as with motor performance measures, kinematic endpoints had already improved by the end of the rehabilitative training period. As there continued to be improvements in release during the follow-up period, it is still possible that some of the late map expansion was due to refinement in movement quality. It is possible that the smaller PLD 38 forelimb maps in the no-rehab group reflect the continued use of maladaptive compensatory movement strategies, sometimes called “learned-bad use.” ³⁴ Such compensation may have dampened both motor performance gains and forelimb map expansion.

The lack of an effect of rehabilitative training on motor maps on PLD 18 contrasts with the large number of neuroimaging studies in stroke survivors demonstrating rapid structural and functional changes after rehabilitative interventions. ³⁵ However, human studies correlating the effects of interventions on neuroimaging endpoints are typically done in a more chronic state. The present results suggest that as clinical trials are conducted at earlier time points after stroke, neuroimaging results may not be entirely predictable from results in chronic populations. Changes in functional maps may only be observable at later time points.

There is considerable evidence that cortical injury initiates a plethora of presumably regenerative neurophysiologic and neuroanatomic events in the peri-infarct and remote tissue that last for at least a few weeks.^27,36 A time-dependent expression of both neuronal growth-promoting and growth-inhibiting genes occurs within the first week after stroke. Altered gene expression profiles are found in the peri-infarct tissue ³⁷ and in neurons in the RFA that project to the infarcted zone. ¹⁴ Other early changes include synaptogenesis, axonal sprouting, dendritic arborization, and dendritic spine remodeling.^38-44 Peri-infarct neuronal excitability changes occur rapidly and are mediated by extrasynaptic GABA_A (γ-aminobutyric acid type A) receptors.^45-47 Pharmacologic blockade of tonic GABAergic transmission during this early stage can rapidly improve recovery in mice after stroke. ⁴⁷ Also, decreased excitation in the peri-infarct cortex is mediated by altered NMDA (N-methyl-d-aspartate) and AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) receptors. Modulation of AMPA receptors 5 days after stroke enhances functional recovery mediated by release of BDNF (brain-derived neurotrophic factor). ⁴⁸ Thus, while motor maps (and sensory maps^36,49) are disrupted and disorganized at these early time points, repair processes are already initiated and are amenable to therapeutic interventions.

In conclusion, during the first few weeks after injury, 2 competing processes are at play: (a) Diaschisis results in sublethal, presumably reversible functional disruption of local and connected neurons; less affected neurons in these same regions gain a competitive advantage, resulting in altered map topography. (b) Regenerative processes are set into motion and are modifiable via behavioral or pharmacological interventions. Thus, rehabilitative training during the first few weeks postinjury results in rapid behavioral gains, but such gains typically are not expressed in motor output maps due to the ongoing diaschisis. However, cortical reorganization continues long after behavioral recovery has stabilized. It is likely that rehabilitative training guides the eventual neuroanatomical and neurophysiological changes that will persist in chronic stages, much like motor training in healthy rats induces synaptogenesis that is reflected later. ⁵⁰ The implication of these results for clinical stroke rehabilitation is that the process of postinjury neural plasticity can be guided in a powerful way by the type and quality of early postinjury motor experience.