Introduction

The outflow of the motor cortex to the spinal cord—the corticospinal tract (CST)—drives voluntary motor function predominantly through contralateral projections. Fibers from the primary motor cortex descend through the posterior limb of the internal capsule into the cerebral peduncles forming the pyramidal tract (PT), decussate at the level of the caudal medulla, and descend primarily in the dorsolateral funiculi of the spinal cord (Fig. 1a). Anatomical studies mapping CST connectivity in mammals provided evidence of ipsilateral CST (iCST) projections that descend in the lateral or ventral funiculi of the spinal cord of cats [1, 2] and monkeys [3, 4] (Fig. 1b–d), or that are components of re-crossed contralateral fibers [4] (Fig. 1e). iCST projections are sometimes referred to as “latent” projections with the implicit hypothesis that these projections do not contribute to motor function in the presence of intact contralateral projections, but their role in cortical motor control may arise after a lesion to the crossed CST projections [5]. In fact, unilateral lesions of the CST are associated with significant motor recovery which may suggest the existence or emergence of iCST projections in mammalian models [1,2,3,4] and human patients [6, 7]. However, this concept is challenged by evidence implicating subcortical brainstem descending pathways in the restoration of function after CST lesion especially in the event of bilateral pyramidotomy in monkeys [8, 9]. A potential role of iCST in recovery from injury would provide a rationale for targeting these fibers to modulate the recovery process.

Fig. 1
figure 1

Probable trajectories for contralateral ipsilateral corticospinal projections into spinal interneurons. a Most of CST fibers cross at the caudal medulla and descend through dorsolateral column. b A smaller component of crossed CST fibers descend in the ventral column. c Ipsilateral CST fibers uncrossed at the level of caudal medulla and descend through ipsilateral anterior and dorsolateral columns. d Some crossed CST projections may re-cross the midline and descend ipsilaterally different spinal levels. e Ipsilateral axonal innervation from crossed CST fibers through axonal collaterals that re-cross the midline and innervate the opposite hemicord

Full size image

Using a variety of neuroanatomical and neurophysiological techniques, studies in rodents, cats, monkeys, and humans have significantly advanced our understanding of the neuroanatomical trajectory of the CST. However, no recent studies have reviewed and evaluated both the preclinical and clinical evidence for the presence and functional relevance of iCST projections. The aim of this manuscript is to review the neuroanatomical trajectory and functional relevance of iCST fibers in mammalian species, including human, and how these findings are implicated in neurorehabilitation after brain injury, mainly in the context of cerebral palsy and stroke. It is necessary to emphasize that discussing the role of the ipsilateral hemisphere in recovery from stroke is beyond the scope of this review and has been extensively reviewed [10, 11].

Major Theories on iCST

A common concept in neurodevelopment is that spinal neurons are innervated by bilateral CST projections that get refined during motor development creating a dominant contralateral system [12, 13]. A remnant of ipsilateral projections persists during adulthood and may still contribute to motor functions [3, 14,15,16,17]. In children with congenital hemiplegia, the leading hypothesis is that the developmental machinery is adaptively altered to favor bilateral innervation of spinal neurons from the intact hemisphere and a functional contribution of iCST fibers [18]. Similarly, after brain injury in adults, it is possible that the loss of contralateral CST fibers results in the strengthening of existing iCST fibers or emergence of new iCST collaterals that may contribute to recovery [19,20,21]. However, this possibility is still controversial with the counter-argument suggesting minimal or no role of iCST fibers in normal or compensatory motor activity after brain injury [7, 22,23,24,25,26,27,28].

iCST in non-primates

Preclinical studies provide a unique opportunity to use invasive approaches such as lesion studies and fiber tracing, to determine whether an iCST is evolutionary conserved, and to closely investigate the adaptation of CST projections after injury. Early studies by Armand et al. reported on subtotal transection in the cat spinal cord that spared the lateral or ventral funiculus from one side followed by HRP injection at cortical or spinal levels to anterogradely and retrogradely map motor cortical projections [1, 29]. Anterograde labeling of sections at cervical and lumbar enlargements demonstrated that regions within area 4 (M1) either projected contralaterally through the lateral funiculus or bilaterally through both lateral and ventral funiculi [1]. Around 92% of fibers in the dorsolateral funiculus were derived from contralateral area 4 while the remaining fibers were originating ipsilaterally [29]. Subsequent evidence using WGA-conjugated HRP confirmed that the sensorimotor cortex of cats projects to the bilateral ventral and lateral funiculi [2, 30]. A similar pattern of dominant contralateral CST and a minor iCST was also described in guinea pigs and rats [31]. However, conclusive evidence from these studies is challenged by several limitations. First, HRP may bleed into un-intended spinal segments after injection resulting in false positive findings. Additionally, the use of WGA-HRP, a trans-synaptic tracer, does not eliminate the possibility that the origin of ipsilateral labeling is subcortical nuclei rather than an iCST or that ipsilateral projections are “re-crossed” branching collaterals of initially crossed fibers [2, 9]. Another major concern in animal studies is that the separation of sensory and cortical motor components is difficult to establish [32].

In a thorough investigation of the iCST in adult rats, biotinylated dextran amine (BDA) injection into the right sensorimotor cortex bilaterally labeled terminals in the thalamic, mesencephalic, and pontine nuclei [33]. At the spinal level, BDA was predominantly present in the contralateral dorsolateral funiculus with a smaller ipsilateral fraction in the dorsolateral funiculus, and, occasionally, in the ventral funiculus. No bifurcating axons were detected on sagittal sections of the medulla, suggesting that uncrossed components cannot be explained by bifurcation of crossed CST axons [33]. The majority of iCST fibers terminated in Rexed lamina III-VI, covering predominantly interneurons and sensory nuclei [34]; however, this pattern is similar to that of contralateral CST in rodents that do not terminate directly at motor neurons (lamina IX).

Neurophysiological techniques including intra-cortical micro-stimulation and transcranial magnetic stimulation (TMS) were used to probe the physiological relevance of the iCST; however, they fail to establish evidence of an anatomically distinct iCST [35, 36] (Table 1).

Table 1 Summary of neuroanatomical and neurophysiological approaches used in CST tracking studies
Full size table

iCST in non-human Primates

Non-human primates provide a valuable model to test whether an iCST is conserved in the mammalian hierarchy. Early observations by Hoff et al. showed that unilateral lesions to the PT resulted in extensive bouton degeneration in the contralateral spinal gray matter; however, minor degeneration was observed within the dorsal and ventral horns of ipsilateral gray suggesting the presence of ipsilateral pyramidal projections [37].

Subsequent neuroanatomical and neurophysiological studies characterized those ipsilateral projections regarding quantity, trajectory, termination, and functional relevance. Fiber tracing of spinal neurons in rhesus monkeys mapped cortical projections from area 4 and indicated the presence of iCST fibers that terminated predominantly at spinal laminae VIII [4, 15, 38].

Studies in rhesus monkeys tracing fibers descending from M1 showed that around 85–98% of CST fibers descended contralaterally in the dorsolateral funiculus, <1% descended contralaterally in the ventromedial funiculus, 2–15% descended ipsilaterally in the dorsolateral funiculus, and 2% descended ipsilaterally in the ventromedial funiculus (Table 2, Fig. 2) [3, 14, 39,40,41]. Interestingly, evidence for the crossing of ipsilateral fibers at spinal levels was also observed in multiple studies [3, 39, 40]; therefore, the presence of ipsilateral fibers suggested by BDA only indicates that these fibers descend ipsilaterally and there is a high likelihood that these fibers will re-cross and end up innervating contralateral neurons. When the termination patterns of these ipsilateral fibers were assessed, they were found to cover spinal laminae V-IX, with the highest density (~80%) of innervation in ipsilateral lamina VIII, and very sparse termination in ipsilateral lamina IX (Table 2, Fig. 2) [3, 14, 39,40,41]. The consistent finding that the majority of iCST fibers terminate at lamina VIII challenges the functional significance of iCST fibers since lamina VIII mainly harbors commissural interneurons that project through the midline to the contralateral cord, thus largely contributing to contralateral movement control [46]. In addition, few fibers labeled interneurons that project to motor neurons controlling proximal muscles (lamina VII), and only sparse labeling was found to potentially label motor neurons (lamina IX).

Table 2 Major neuroanatomical and neurophysiological studies describing ipsilateral CST in monkeys
Full size table
Fig. 2
figure 2

Termination patterns of CST fibers descending from the left M1 cortex in monkeys mapped using anterograde tracing. The majority of descending fibers descend contralaterally through the dorsolateral funiculus. Around 2–16% of fibers descend ipsilaterally and terminate predominantly at lamina VII and VIII of the spinal gray

Full size image

Anterograde tracing of WGA-HRP from the supplementary motor cortex (SMA) in rhesus monkeys revealed that 23% of descending fibers were ipsilateral, descended through the dorsolateral funiculus, and terminated in similar patterns to ipsilateral M1 projections (mainly laminae VII-VIII) [14]. Findings in rhesus monkeys were replicated in marmoset monkeys where 11% of M1 CST fibers descended ipsilaterally through the dorsolateral funiculus and terminated in laminae VII-VIII [42]. These studies suggest that ipsilateral M1 may partake in the control of proximal and potentially distal muscles. A major limitation of quantitative studies with labeled tracers is the inability to label all fibers at the injection sites, which may result in variability of the relative proportions of different CST components.

The functional relevance of iCST fibers was investigated using in vivo micro-stimulation of the M1 cortex, SMA, and pyramidal neurons. Stimulation of M1 or SMA cortices while recording from both distal and proximal forearm muscles revealed that the vast majority of M1 and SMA outputs were contralateral [43,44,45]. However, a significant ipsilateral response was predominantly recorded in proximal or truncal muscles upon stimulation of SMA [44, 45], and, occasionally, M1 [43]. In vivo cortical stimulation studies performed in rhesus [24] and marmoset monkeys [42] showed that ipsilateral M1 projections are not monosynaptic. Although the lack of monosynaptic connection is not unanticipated given that the majority of contralateral fibers also do not monosynaptically connect to motor neurons, the presence of polysynaptic connections does not rule out the involvement of the contralateral hemisphere or contralateral brainstem nuclei upon stimulation of ipsilateral cortical regions. In fact, stimulation of the PT in rhesus monkeys elicited clear excitatory post-synaptic potentials in contralateral forearm spinal motor neurons, but failed to induce any ipsilateral responses [24]. Similarly, EMG recordings in forearm muscles demonstrated only contralateral MEPs in response to PT stimulation prior to the decussation. Stimulation of the forearm sites of M1 in awake monkeys resulted in EMG responses and behavioral movements in contralateral forearm muscles only [24]. These results have emphasized that the physiological role of motor cortical stimulation in eliciting responses in ipsilateral muscles is predominantly mediated by polysynaptic connections that do not necessarily involve ipsilateral PT.

Evidence of Reorganization of CST after Injury in Animal Models

Following unilateral lesion to the motor cortex or the PT proximal to decussation in rodents, sprouting axons from the crossed contralateral fibers re-innervate denervated neurons in a compensatory mechanism [47, 48]. This compensatory mechanism is not due to originally uncrossed CST at the pyramidal level that remained unchanged after injury, but due to re-crossing of contralateral fibers at the spinal level [47]. Supporting these findings, unilateral lesions proximal to the CST decussation failed to induce deficit at the ipsilateral limb, and iCST fibers failed to rescue limb function after contralateral pyramidal lesions [49]. Simultaneously, electrical stimulation of affected hemisphere after stroke in rodent model demonstrated improved recovery of skilled functions through promoting axonal sprouting at the subcortical (red nucleus) and spinal levels [9, 50]. However, controversial findings still suggest a role of iCST in recovery from injury. For instance, damage to the motor cortex in rodents can cause deficits in skilled movements ipsilaterally that was associated with increased synaptic plasticity in the undamaged hemisphere [51]. In addition to the sprouting of crossed CST fibers from undamaged neurons in ipsilateral hemisphere, axonal sprouting from the intact motor systems to cover denervated neurons after injury has been described in rodents [52]. Recently, Wahl et al. demonstrated that originally crossed CST fibers from the undamaged hemisphere re-cross at the spinal level to cover denervated neurons initially innervated by the damaged hemisphere [53]. This study used a unique pharmacogenomics approach to silence CST fibers terminating ipsilaterally and demonstrated a significant reduction in EMG response in the ipsilateral limbs and reappearance of functional deficits originally ameliorated by rehabilitation-induced plasticity [53]. Although these findings support a role of the undamaged hemisphere in plasticity after stroke that includes axonal sprouting to cover the ipsilateral neuronal territory, it does not support the presence of latent iCST that gets engaged after stroke.

Studies in primates have supported the contribution of axonal sprouting to recovery after stroke and showed that inhibition of neurite outgrowth inhibitor (NoGo) enhanced compensatory sprouting and improvement motor recovery [54, 55]. Compensatory recovery mechanisms are most prominent in brainstem circuitry. For example, after bilateral pyramidectomy, monkeys were still able to carry out proximal and distal muscle movement [8], pointing to the role of non-pyramidal projections in the recovery of motor performance after an injury to PT in primates. These early findings were further supported by recent work from Zaaimi et al. showing that after a unilateral lesion to the PT, PT stimulation did not result in activity in ipsilateral motor neurons indicating a negligible remodeling in the iCST system 6 months after a unilateral PT lesion [56]. The study demonstrated a limited role for iCST fibers in recovery after unilateral PT lesions and a significant compensatory role of reticulospinal systems in recovery [56].

Evidence of iCST in Healthy Human Subjects

Similar to observations in monkeys, Nathan et al. studied post-mortem spinal cord sections from subjects with supra-spinal lesions and examined the location of degenerating axons in the spinal cord [57]. In one subject with large right middle cerebral artery infarct, degenerating axons were observed both in the left CST and in the right anterior CST, but barely observed in the left anterior CST hinting that an anterior iCST might be present in humans [57]. However, due to limitations of the specificity of this approach, solid evidence supporting the presence of iCST that do not decussate at the medulla in humans was still lacking.

Beyond early neuroanatomical studies, studies investigating the presence of an iCST in healthy human subjects consistently used transcranial magnetic stimulation (TMS) or intra-cortical micro-stimulation (ICMS) to suggest the presence of these fibers. It was suggested that an iCST is more likely to control proximal and truncal muscles rather than the distal muscles since the latter tend to be more severely affected by stroke involving the motor systems and tend to recover last [58]. However, both proximal and distal muscles have been investigated using cortical stimulation and recording of MEP to assess the presence, relative delay, and relative amplitude of ipsilateral muscle responses compared to the contralateral responses (Table 3). Findings from these studies have shown that after TMS stimulation of the dominant hemisphere, bilateral MEPs were detected predominantly in proximal and truncal muscles and were delayed (2–9 ms) and lower in amplitude (10–80%) when compared to contralateral responses [23, 59,60,61,62,63,64, 66]. Similar findings were also reported after ICMS of motor cortex during functional mapping in epileptic patients [67] or intra-operative monitoring during spinal surgeries [68]. In one study, Muller et al. studied the development of the CST innervation and demonstrated that children ≤9 years of age have bilateral MEPs in the FDI, biceps brachii, and brachioradialis muscles after unilateral motor cortical stimulation; however, adults showed only contralateral MEPs in the same muscles [6].

Table 3 Summary of TMS studies investigating bilateral CST projections in humans
Full size table

Although this approach was most commonly adopted in human studies of healthy subjects, several significant challenges limit the utility of this method to establish the presence of an anatomically independent iCST. Studies using TMS to detect ipsilateral MEPs in muscles are limited by the inability to control focal stimulation of the motor cortex unilaterally without influencing the contralateral hemisphere or subcortical structures. Although the use of small coils has provided better resolution, this limitation is still of major concern, especially in studies that reported very short or absent latency between ipsilateral and contralateral responses [59, 60]. In contrast, a delay that reaches up to 9 ms in ipsilateral MEPs indicates a polysynaptic innervation and may involve either interhemispheric facilitation of the contralateral hemisphere or recruitment of bilateral brainstem circuitry rather than a direct projection of uncrossed ipsilateral fibers. This concern is particularly relevant after the finding that TMS stimulation of M1 in monkeys was able to activate the brainstem directly and independent of M1, a phenomenon that is likely to occur in humans [69]. Thus, TMS studies provide little insight into the neuroanatomical substrate of ipsilateral MEPs detected in muscles. An interesting subject in Ziemann’s study showed complete agenesis of the corpus callosum while still exhibiting bilateral MEPs in his FDI after unilateral motor cortical stimulation [64]. This finding indicates that these bilateral MEPs are independent of inter-hemispheric connections despite that they may still relate to direct stimulation of or bilateral cortical projections to brainstem nuclei. It is also noteworthy that in studies reporting bilateral MEPs after cortical stimulation, a subset of subjects showed only contralateral responses, pointing to the presence of individual variation in the laterality of cortical output [23, 59,60,61,62,63,64, 66].

Re-organization of Motor Cortical Projections in Hemiplegic Human Conditions

Although the contribution of iCST to motor control is controversial, disordered organization of CST connectivity, including abnormal decussation or bilateral projections, has been described in several pathological conditions such as essential mirror movement syndrome, Klippel-Feil disease, and progressive scoliosis [22]. In this section, we review the evidence on whether iCST projections play a functional role after brain injury.

One key determinant of motor cortical re-organization after brain injury is the stage of maturation of CST projections during development [70]. In contrast to adult stroke, children with cerebral palsy (CP) show substantial evidence of bilateral cortical innervation from the undamaged cortex to limb muscles [71]. Stimulation of intact motor cortex in children with CP resulted in bilateral contraction of hand muscles with high synchrony on cross-correlograms that sometimes manifested as mirror movements, suggesting that the two muscles may have received the same presynaptic inputs [72, 73]. Based on studies in cats and monkeys [8, 74], this could potentially be explained by the maintenance of dense bilateral CST projections that are otherwise retracted during development, or by bilateral cortical projections into brainstem nuclei [9, 12, 56, 75].

Studies on the neurodevelopment of the corticospinal in neonates have shown that motor cortical stimulation is associated with fast, short latency ipsilateral responses that occur at a similar threshold to contralateral responses [13]. Ipsilateral responses become smaller and more delayed after 18 months of age; however, short latency ipsilateral responses are only preserved in cerebral palsy (CP) patients who had ischemic insults at various gestational stages [76], but not in adults recovering from brain injury [12, 13, 77]. This points to a developmental shift in CST innervation from bilateral to contralateral innervation during early neonatal developmental window that may be inhibited in neonates with CP, but not in adult subjects with ischemic stroke [13, 77]. Thus, it is hypothesized that ipsilateral innervation after CP is likely due to persistent ipsilaterally descending fibers rather than sprouting of contralateral CST. This hypothesis is supported by findings in animals showing preferential withdrawal of iCST fibers during development of the CST [13, 78]. In addition, neuroanatomical data from post-mortem analysis of children showed significantly faster growth of contralateral compared to iCST [12, 79], and a larger number of iCST fibers with larger axons at the pyramidal level of subjects with CP but not those with unilateral lesions that occur during childhood or adulthood [12, 79]. This data indicates that in CP patients, ipsilaterally projecting fibers are not likely due to spinal collaterals of contralateral fibers, but rather from ipsilateral cortico-fugal projections originating from the ipsilateral motor and supplementary motor cortices. The functional relevance of ipsilaterally projecting motor fibers in CP patients was further investigated in subjects with intractable epilepsy who undergo hemispherectomy (HS) to disconnect the damaged hemisphere (extensively reviewed in [80]). Cortical stimulation of the undamaged hemisphere in CP patients continues to elicit bilateral responses in forearm muscles post-HS similar to pre-surgical responses indicating that these responses are independent of the damaged hemisphere [81, 82]. In a large trial of children undergoing HS, Kupper et al. demonstrated that grasping ability in the ipsilesional hand is only preserved in patients with prenatal or perinatal unilateral brain injury (or CP) and is associated with asymmetric structural connectivity of CST projections. This data suggests the reinforcement of developmentally preserved iCST fibers in these patients [83].

Although cortical re-organization involving abnormal iCST projections is infrequently seen in adult stroke, a margin of neuroplastic changes in both hemispheres may still occur and contribute to recovery [7, 13]. In fact, using functional magnetic resonance imaging (fMRI) during task-related movements of the paretic hand in chronic stroke patients, several studies have reported increased activation in the ipsilateral sensorimotor cortex, premotor cortex, SMA, and occipito-parietal cortex compared to the healthy controls [7, 27, 84]. This functional pattern was also associated with microstructural changes bilaterally in the CST of stroke patients, revealed by structural brain imaging (DTI) [85]. Longitudinal studies in patients recovering from stroke have confirmed a shift in task-related cortical activation from ipsilateral (contralesional) to contralateral (ipsilesional) activation, suggesting the importance of ipsilesional M1 activity in mediating recovery processes [7, 25,26,27,28]. Interestingly, bilateral cortical activity was also negatively correlated with the integrity of the affected CST, a finding that was not always associated with the recovery mechanisms [27, 28, 86]. Temporal analysis using electroencephalography (EEG) has also shown patterns of increased activity ipsilateral to the lesion in stroke patients during voluntary movement [5]. However, the timing of ipsilateral activation on EEG occurred after the onset of movement, discounting its role in the activity of the paretic limb [87].

Cortical functional re-organization and connectivity networks involved in stroke recovery do not appear to be uniform. Indeed, a heterogeneous pattern of cortical reorganization seen in fMRI studies reflected the heterogeneity of ischemic stroke populations [28]. It was further confirmed by positron emission tomography studies in patients with internal capsule infarct where the lesion location and the involvement of striatal structures dictated different patterns of re-organization [88]. The heterogeneity of cortical re-organization seen on brain imaging was also consistent with incongruous findings in neurophysiological experiments using TMS or transcranial direct current stimulation of motor cortices of stroke patients. TMS stimulation over the ipsilateral dorsal premotor cortex, M1, and the superior parietal lobe in patients with recovered motor performance after stroke resulted in significant interference with performance that was not observed in healthy controls [84]. However, TMS stimulation elicited ipsilateral MEPs in proximal muscles of stroke patients but more prevalent in stroke patients with poor recovery suggesting the emergence of the contralesional motor drive [89].

Collectively, studies in patients with CP and stroke have supported the role of ipsilesional primary and secondary motor cortices in rapid and better recovery after stroke. Furthermore, despite that contralateral CST projections are the major determinant for motor recovery [90], the functional contribution of iCST projections is still debatable.

Conclusions and Future Directions

Conclusive anatomical evidence that an iCST is present in adult humans is absent; however, animal studies demonstrate that iCST fibers are conserved from rodents to non-human primates supporting the existence of iCST in humans. Neurophysiological studies in humans fail to characterize an independent iCST in healthy adults. The functional relevance of iCST in healthy and stroke adult patients is still controversial, but evidence on the important contribution of iCST projections was demonstrated in pediatric patients with congenital hemiplegia. This supports the hypothesis that CST development exhibits a critical period for lateralization of projections, and that only early cortical insults may promote compensatory neurodevelopmental changes that will protect ipsilateral projections. This raises critical questions whether we can reset the developmental clock of the CST and tune up the damaged tract after brain injury in adults and poses challenges to studies that attempt to stimulate iCST fibers in stroke patients to enhance recovery. In addition, clinical studies investigating the role of undamaged motor cortical activity in stroke recovery should carefully interpret the neuroanatomical substrate of this role taking into account the possibility of involvement of inter-hemispheric commissural connections and bilateral connections to brainstem nuclei in addition to potential iCST fibers.

Moving forward, it is important that work in non-human primates replicates the approach used by Wahl et al. to use optogenetics or pharmacogenomics to specifically inhibit iCST projections in the undamaged cortex and identify the role of these fibers in sub-acute and chronic recovery after stroke. Specifically, the use of viral vectors to encode inhibitory optogenetic channels in cortical neurons with direct ipsilateral projections (not through brainstem nuclei) followed by optogenetic stimulation to specifically inhibit motor or supplementary motor neurons will allow a definite assessment of the presence and role of cortically projecting ipsilateral corticospinal connections both in normal function and after stroke. Additionally, the use of novel neuropathological approaches like CLARITY on post-mortem tissue from stroke patients as well as healthy adults allows high-resolution tracking of projecting fibers to identify whether an anatomically distinct iCST exists at pyramidal levels and to clarify the identity of neurons at the termination of these fibers. Recent proof-of-concept studies in non-human primates demonstrated feasibility of clearing spinal cord tissue with subsequent fluorescent imaging and 3D reconstruction [91]. Finally, more robust evidence on the iCST is anticipated to arise from the advancement of imaging method, like DTI, and non-invasive fiber tracking techniques, and neurophysiological techniques including virtual lesions and plasticity protocols.