Effects of lesion of the primary motor cortex (M1) were investigated in a large number of studies [23–27]. Among them, the ability of finger dexterity was investigated in detail in a most recent study by Murata et al. [28]. In this study, following mapping of the somatotopical organization of the M1 with intracortical microstimulation (ICMS) technique, the hand region was destroyed by injection of ibotenic acid. Following the lesion, digit movements were severely impaired, but with daily rehabilitative training, the ability of precision grip recovered in 3–4 month period. Interestingly, the recovery course took two steps; initially, the monkey tended to make “alternate grip” with which a small spherical food pellet was grasped between the tip of the index finger and the point around the metacarpophalangeal joint or interphalangeal joint of the thumb. However, gradually the monkey changed the strategy to grasp between the pads of the index finger and tip (distal to the interphalangeal joint) of the thumb (“precision grip”) and the alternative grip disappeared. But if the monkey was not sufficiently trained, they continued to use the alternative grip. In some of these monkeys, the brain activity during the recovery course was observed by using the positron-emission tomography (PET) and it was found that the activity of the ventral premotor cortex (PMv) on the lesioned side and the M1 area surrounding the lesion was enhanced during the recovery course [29]. Possible change in the somatotopic organization of the M1 and PM after the M1 lesion was studied by Liu and Rouiller by using ICMS [23]; however, any visible motor response was induced from the lesioned M1. On the other hand, inactivation of the PM on the lesioned side with muscimol suppressed the recovered dexterous movements, which suggested contribution of the PM to the recovery [29]. Based on the above results, a question may arise on how the PM contributes to the recovery. One possible explanation is the PM influences the motor function via the M1 as shown in the normal monkey [30] in the “hierarchical order” with the M1. Another possibility is that the PM directly influences the motor control via its direct projection to the subcortical centers and the spinal cord [31, 32] in the “parallel order” with the M1. At this moment, we have no clear answer to this question.

“Pyramidotomy” at the level of the brainstem pyramid has been tested in a large number of studies (e.g., [33–35]). In these studies, dexterous hand movements were permanently impaired. In Lawrence and Kuypers study [35], the authors hypothesized that the recovered movements (less dexterous) must be controlled by the descending pathways mediated by the brainstem nuclei such as the red nucleus and pontomedullary reticular formation. Additional lesion of the lateral brainstem pathway (namely the cortico-rubrospinal pathway) and ventromedial brainstem pathway (cortico-reticulospinal pathway) revealed that interruption of the former resulted in the impairment of hand movements, while interruption of the latter pathway resulted in a severe impairment of movements in the proximal extremity. These findings suggested that the recovery of hand movements after the pyramidotomy was chiefly compensated by the cortico-rubrospinal pathway. In humans, Bucy et al. [36] reported the results of transection of the cerebral peduncle in patients with dyskinesias and showed recovery of limb movements. Bucy discussed that such recovery was due mainly to the subcortical pathways.

On the other hand, more recently, Baker and colleagues investigated the role of the cortico-reticulospinal pathway for recovery from lesion of the brainstem pyramid. They showed that the amplitude of the ecitatory synaptic potentials evoked by electrical stimulation of the reticulospinal tract in the motoneurons of hand muscles was facilitated [37], which suggested plasticity enhancing the transmission through the reticulospinal tract. Based on these findings, it was proposed that not only the cortico-rubrospinal tract but also the cortico-reticulospinal tract is responsible for the recovery of hand movements after stroke [38].

In lower mammals, the corticospinal tract (CST) does not make direct connections with motoneurons. The shortest pathway to forelimb motoneurons in a cat, for example, is disynaptic, with a substantial portion of the disynaptic excitation being mediated by propriospinal neurons (PNs) located in the mid-cervical segments, C3–C4 (C3–C4 PNs) [39–41]. The existence of such PNs in primates has been doubted, primarily because disynaptic excitation of motoneurons was rarely observed in either intracellular recordings from motoneurons in anesthetized monkeys [42, 43] or in EMG recordings from hand muscles in awake monkeys [44] following stimulation of the CST or the motor cortex. However, it has been shown that disynaptic corticospinal excitation, which was mediated by C3–C4 PNs, became evident after glycinergic inhibition was reduced by intravenous injection of strychnine [45] and the existence of C3–C4 PNs, which receive monosynaptic pyramidal excitation and transmit the excitation to forelimb motoneurons, has been demonstrated in anesthetized monkeys [46]. The reason why the disynaptic excitation was not observed without strychnine was suggested to be that feedforward inhibition from the CST to the C3–C4 PNs was so potent that stimulation of the brainstem pyramid caused substantial disynaptic inhibition. It prevented the PNs from firing to the pyramidal stimulation; the monosynaptic excitation might be substantially curtailed by the following disynaptic inhibition [46]. The C3–C4 PNs have also been shown to exist in humans, and similarly, potential feedforward inhibition from the CST has been demonstrated in this system [47]. Thus, the existence of the C3–C4 PNs has been shown in electrophysiological experiments in anesthetized monkeys, but only after application of strychnine. As the next step, the functional significance of the PNs was needed to be clarified in behavioral studies. To study the role of the C3–C4 PNs in the control of fractionated digit movements, we investigated the effect of lateral corticospinal tract (l-CST) lesion at the C5 and C2 segments in macaque monkeys. First, selective lesion of the dorsolateral funiculus which covers the location of the l-CST was made at C5 segments. Immediately after the lesion, the monkeys showed deficit in precision grip. However, fractionated digit movements could still be observed immediately after the lesion (<2 weeks) in all the animals. In these animals, precision grip recovered in a few weeks to a few months (Fig. 7.2). Until now, we have tested the similar lesion in more than 20 monkeys and all of them showed recovery of precision grip to the success ratio higher than 90 %. In contrast, in the animals with similar lesion at the C2 level, the monkeys showed no trace of fractionated digit movements immediately after the lesion. Comparison of the hand movements immediately after the l-CST lesion at C5 and C2 suggests that the C3–C4 PNs can mediate the command for the precision grip in intact animals [48]. In these animals, longer post-lesional observation revealed limited recovery of fractionated digit movements, similar to the lesion brainstem pyramid reported by Lawrence and Kuypers [35] (unpublished observation by the same authors as Alstermark et al. 2011). All these results suggested that the PNs are involved in the control of fractionated digit movements and also their long-term recovery after the l-CST lesion at C5.

Several groups investigated the effects of hemisection at the lower cervical level (C5–C7) [50–54]. These studies showed that the monkeys exhibited severe flaccid hemiparesis during the first month, and predominant recovery of the affected limb could be observed later. However, the deficit in hand dexterity remained. Recently, Rosenzweig et al. [53] showed extensive sprouting of midline-crossing axons from the corticospinal fibers descending in the intact side of the spinal cord, which may underlie the recovery of hand movements.

There is no systematic report on the effect of damage to the internal capsule in the nonhuman primate model, presumably because of difficulty in making an appropriate lesion of such a large fiber tract located subcortically. In humans, the cerebral infarction frequently occurs around the internal capsule. Fries et al. [55] studied patients with capsular or striatocapsular stroke and found that lesion of the anterior (plus caudate/putamen) or posterior limb of the internal capsule led to an initially severe motor impairment; however, it is mostly followed by excellent recovery in hand function. In contrast, lesions of the posterior limb in combination with the lateral thalamus compromised the motor outcome. They subsequently studied the trajectories of the corticofugal axons of the M1, PM, and SMA with an anterograde tracer in macaque monkeys. It was found that axons from each cortical area were differentially localized in the internal capsule; those from the M1 passed through the middle third of the posterior limb, those from the PM passed through the capsular genu, and those from the SMA passed through the anterior limb. Thus, if the lesion covers the whole capsule, the impairment should be severe; however, if a part of the capsule spared the lesion, considerable recovery can be expected. These results support the “parallel order” rather than “hierarchical order” operation of cortical motor areas, which provides the capability of substituting each other in the process of recovery.

As described above, the hand dexterity showed marked recovery after the CST lesion at the C4–C5 segments. We used this model to study the plasticity during the recovery of hand dexterity. Actually, the plasticity was found not only at the spinal cord level but also in higher-order structures. We will summarize the results below.