Somatosensory System Organization in Mammals and Response to Spinal Injury
Summary and Keywords
Spinal cord injury (SCI) affects well over a million people in the United States alone, and its personal and societal costs are huge. This article provides a current overview of the organization of somatosensory and motor pathways, in the context of hand/paw function in nonhuman primate and rodent models of SCI. Despite decades of basic research and clinical trials, therapeutic options remain limited. This is largely due to the fact that (i) spinal cord structure and function is very complex and still poorly understood, (ii) there are many species differences which can make translation from the rodent to primate difficult, and (iii) we are still some way from determining the detailed multilevel pathway responses affecting recovery. There has also been little focus, until recently, on the sensory pathways involved in SCI and recovery, which are so critical to hand function and the recovery process. The potential for recovery in any individual depends on many factors, including the location and size of the injury, the extent of sparing of fiber tracts, and the post-injury inflammatory response. There is also a progression of change over the first weeks and months that must be taken into account when assessing recovery. There are currently no good biomarkers of recovery, and while axon terminal sprouting is frequently used in the experimental setting as an indicator of circuit remodeling and “recovery,” the correlation between sprouting and functional recovery deserves scrutiny.
The purpose of this article is to provide an overview of our understanding of the organization of somatosensory and motor pathways, in the context of what we currently know about their reorganization/recovery following spinal cord injury (SCI). Since it is not possible to address all aspects of this topic in one review, we will mainly focus on deafferentation injuries and the pathways involved in forelimb function in nonhuman primate (NHP) and rodent research. Hand function is critical to independence in people (Lo, Tran, Anderson, Craig, & Middleton, 2016), and these two animal models are most widely used to inform clinical therapeutic approaches. Spinal injury affects an estimated 1.4 million people living in the United States alone, and the personal and societal toll for these individuals is enormous (statistics from the Christopher and Dana Reeve Foundation). While a vast amount of basic research has been carried out and many clinical trials have been conducted in recent decades, the clinical approach to spinal injury remains limited. The following explores some of the reasons why.
First and foremost we still lack a detailed understanding of the sensorimotor pathways involved in SCI. A thorough knowledge of how they are organized, and how they are able to respond and compensate following injury is key to being able to predict recovery potential in the clinic. The pathways mediating hand function are complex, and are both parallel and distributed, and the extent to which they are able to compensate following injury depends in large part on what is spared by the SCI (after primary and secondary damage has taken place). Rodent models of cervical SCI dominate the literature, but the neuroanatomy mediating paw function is markedly different from that mediating hand function in primates, and many seemingly positive findings in the rodent literature are not directly translational. Having said this, rat and rodent models of SCI remain important, not only because they are better suited to many techniques that are not practicable in NHPs, but also because many basic mechanisms of recovery are conserved across mammalian species.
Second, the immune/inflammatory response to SCI is complex (involves several phases), and is known to affect recovery, both at the injury site and well beyond at other levels of the neuraxis. While a detailed molecular and cellular description falls outside the purview of this review, there are excellent accounts of our current knowledge (Plant, Weinrich, & Kaltschmidt, 2018; Tran, Warren, & Silver, 2018; van Niekerk, Tuszynski, Lu, & Dulin, 2016). We will, however, use examples from our own NHP research and others, to underscore the potential impact of the central nervous system (CNS) immune response on neuronal growth following SCI. We will also identify where we currently fall short in our understanding of the processes involved.
Third, there are almost no good biomarkers of recovery. Axon terminal sprouting has been used most broadly in the experimental setting as an indicator of circuit remodeling and “recovery,” but is there a correlation between axonal sprouting and functional recovery? And, what biomarkers are there that might be helpful in the clinical/experimental setting? We will address each of these questions to date.
Sensorimotor Pathways Subserving Voluntary Forelimb and Hand Function: Normal
Monkeys and Humans
Afferent and Efferent Pathways: The Basics (See Figures 1 and 2)
Sensory information about our reachable world and internal environment is received by different populations of sensory receptors in the skin and deeper tissues of our hand and forearm (e.g., mechanoreceptors, proprioceptors, nociceptors, etc.). This information enters the cervical spinal cord gray matter via the axons of primary afferent neurons (with cell bodies in the dorsal root ganglia), and is transmitted to second-order neurons in the dorsal horn and cuneate nucleus of the medulla (Figure 1). Most, but not all (i.e., spinothalamic) cutaneous input (e.g., tactile mechanoreceptor and proprioceptor information) is transmitted by large-diameter myelinated fibers via the cuneate fasciculus component of the dorsal column to the cuneate nucleus in the medulla. Cuneate nucleus neurons receive this input, cross to ascend in the contralateral medial lemniscus, and synapse on thalamic neurons, primarily within the main somatosensory relay nucleus (i.e., the caudal ventral posterolateral nucleus in macaque monkeys—caudal Ventral Posterolateral [VPLc]). Thalamocortical neurons in turn project to the primary somatosensory cortex (Brodmann areas 3a, 3b, 1, and 2 in monkeys and humans), as well as to the caudal primary motor cortex. Simplistically, this sensory information is then processed in successively more caudal “higher-order” posterior parietal cortical regions. It is then sent forward to frontal and prefrontal cortical regions, which together shape and inform the major descending corticospinal (CST) output to spinal motor neurons. The CST (which comprises > 9 subcomponents in macaques and probably many more in humans), synapses either directly (i.e., this occurs only in the motor CST in primates), or indirectly (e.g., via intraspinal neurons and networks) on motor neurons (Galea & Darian-Smith, 1997a). Afferent and efferent projections are somatotopically organized, such that pathways transmitting input from adjacent body regions neighbor each other.
Of course, every level of the pathway is more complex than this description implies, and this is important when considering potential circuitry that might mediate recovery following spinal injury. For example, primary afferent neurons synapsing in the dorsal horn of the cervical cord do not simply join the dorsal column. They also connect with interneurons (e.g., ipsilateral, commissural, or propriospinal; Figure 3), and interneurons synapse directly or indirectly on to motor neurons, and feed into a number of additional ascending pathways requiring ongoing sensory input. Spinocerebellar, spinothalamic, spinoreticular, and spinovestibular pathways all transmit sensory input from peripheral receptors and send somatosensory input to thalamic, cerebellar, and brainstem targets.
Efferent feedback connections also exist throughout the pathway. These likely inform and modulate or gate afferent input, en route to the brainstem, thalamus, and cortex, though we understand relatively little about them, and even less about their role following injury. Major feedback pathways from the cortex can be grouped broadly into corticothalamic (Darian-Smith, Tan, & Edwards, 1999), corticobulbar (e.g., corticoreticular, see Fregosi et al., 2018; corticocuneate, see Bentivoglio & Rustioni, 1986, and Canedo, 1997), and corticospinal projections (Galea & Darian-Smith, 1997a; Yoshida & Isa, 2018), but each of these projections can be further subdivided into functionally separate subcomponents. Corticocuneate projections, for example, originate from at least the primary motor and somatosensory cortical regions (Fisher, Lilak, Garner, & Darian-Smith 2018, unpublished), terminate in the contralateral cuneate nucleus (in different nuclear subdivisions), and are thought to modulate or gate ascending somatosensory inputs from the periphery, though again, the potential “gating” function of this pathway is poorly understood. Still another example is the corticorubral projection, which in primates feeds almost exclusively into the inferior olivary nucleus (Burman, Darian-Smith, & Darian-Smith, 2000), and forms a feedback loop to the cortex via the cerebellar circuitry, and thalamus (Darian-Smith et al., 1996). This pathway plays a role in motor coordination and learning, and, importantly, in the processing of afferent sensory information received via the spinocerebellar and cuneocerebellar projection pathways. The corticoreticular and reticulospinal projections also play important roles in voluntary forelimb movements (Baker & Perez, 2017), and while this complex tract takes a back seat to the corticospinal tract for fine volitional digit movements in primates, it likely plays an important role in pathway compensation and recovery following spinal injuries affecting the forelimb, digits, and hand (Baker, Zaaimi, Fisher, & Edgeley, 2015). The list of pathways continues, making it clear that even a relatively small primary afferent spinal injury (Figure 2) can affect CNS circuitry throughout the neuraxis.
In terms of intraspinal connections, local interneurons (glycinergic or GABAergic), commissural interneurons, and propriospinal (short and long range) circuitry form complex connections within the cord, and provide a conduit between segments, as well as a link bridging the two sides of the spinal cord (Figure 3). Commissural interneurons have been described in rats, cats, and more recently demonstrated in monkeys (Figure 3 insert; Soteropoulos, Bannatyne, & Maxwell, 2019, personal communication), so there appears to be a considerable conservation of this organization across mammalian species.
The question of how much processing and memory storage occurs intraspinally also remains unclear, though accumulating evidence suggests it is likely to be considerable (see Rossignol et al., 2015; Wolpaw, 2010). Rodent and cat studies demonstrate this by showing coordinated stepping of hindlimbs following complete thoracic spinal transection. Though this has not yet been replicated in monkeys, evidence (Capogrosso et al., 2016) suggests a similar intraspinal capability. This demonstrates that spinal circuitry can act somewhat independently of higher-level input, as long as there is weight support, a stereotypic or learned behavior being activated, and appropriate sensory feedback to the cord from the hind-paw pads to trigger a learned (and stored) coordinated movement.
Rodents versus Primates
Rodents, and particularly rats, dominate the spinal injury field, and rodent research has contributed greatly to our current understanding of the processes and mechanisms mediating circuit reorganization and functional recovery. Of course, as mammals that use their forepaws to explore and manipulate their surroundings, rats share a basic anatomy with primates. However, there are also considerable differences between rats and primates (Darian-Smith, 2007, 2009; Filipp et al., 2019; Friedli et al., 2015), that must be considered in any translational interpretation of research findings.
To begin, rats are strictly tetrapedal, and have grasp, but no digit opposition or sophisticated precision grip. Humans and monkeys can trivially mimic rodent paw action (a fact often used to underscore similarities), but rodents cannot reciprocate. The anatomy of the sensorimotor pathways mediating paw grasp reflects this and (not surprisingly) differs in many ways (Figure 1). Primates quite simply have larger, more complex brains, and with a common ancestor between humans and rats existing > 65 mya, there is > 130my of evolutionary history separating our species.
As an example of a major pathway difference, the CST pathway in rats is mainly situated ventral to the dorsal columns, whereas in all primates it descends almost entirely in the dorsolateral spinal gray matter (see Figure 1). In rats, there is only one primary somatosensory cortical area, whereas in primates there are four different primary somatosensory cortical regions, each with a separate body surface representation (Figure 2). Thus, in rats, the CST originates from no more than 4 cortical regions (Nudo & Masterton, 1990), and in macaque monkeys from 9 or more (Darian-Smith et al., 1996; Galea & Darian-Smith, 1994). Each of these subcomponents has a unique termination pattern within the spinal gray matter, and each plays a slightly different role in shaping the final motor output. Sensorimotor control of forepaw function in rats is less dependent on the CST than on other descending pathways (e.g., the rubrospinal, reticulospinal, and raphespinal connections all play a role), whereas the CSTs dominate fine motor control in the hands of primates, and interruption to this pathway greatly affects hand function.
Spinal Injury and Recovery
The complexities of connections that can be affected by a spinal cord injury (SCI), as well as inherent species differences, mean that SCI research findings are often difficult to translate to the clinic. Even so, there are a number of experimental models that are used to study spinal injury and recovery, and each has its share of pros and cons. As a rule of thumb, recovery in the central nervous system (CNS) should be considered opportunistic, in the sense that reorganization can and will occur wherever there is a sufficient sparing of relevant connections, and the necessary underlying trophic support for the establishment and stabilization of new activity-driven synapses.
Phases of Recovery
When deciphering experimental data, the post-lesion timeline must also be kept in mind, since post-injury reorganization occurs in stages. Though the time frame overlaps somewhat (see Table 1 in Oyinbo, 2011) and each phase scales differently in different species, the stages are generally described in terms of: (i) a primary acute phase (minutes to hours), (ii) a secondary sub-acute phase (minutes to weeks or even months post-injury, during which additional damage can occur), (iii) a subchronic phase (days, weeks, and early months), and (iv) an extended chronic phase (years to decades).
The primary injury involves the initial damage (regardless of the cause, this involves neuronal cell death and vascular events including hemorrhage), and for practical reasons relatively little research has looked to mitigate cell loss at this early time point. As a consequence, the secondary acute injury phase, during which additional (and often extensive) cell loss can occur, has long been the focus of therapeutic intervention, and there are currently more than 25 established secondary injury mechanisms recognized (see Oyinbo, 2011 for review). Though the vast majority of people with SCI are in the extended chronic phase, relatively little is known about recovery potential beyond the first post-injury year. There is, however, clinical evidence that significant recovery is possible many years following the initial injury. For example, the late Christopher Reeve showed significant sensory recovery > 5 years post SCI, and recent epidural stimulation studies (Gill et al., 2018) suggest a latent potential for motor recovery in some chronic patients. However, much work remains to be done to understand and address this potential.
Overview of Experimental SCI Models
The most commonly used experimental SCI models include (1) deafferentation injury models (affecting just the sensory pathways), (2) efferent pathway lesions (affecting specific descending motor pathways), (3) hemisection lesions (involving afferent and efferent pathways, and affecting either one entire side of the cord in monkeys, or the dorsal half of the cord in rodents), and (4) contusion injuries (affecting the central gray matter, as well as a mix of efferent and afferent pathways).
In rodents, contusion injuries are perhaps the most common injury model used, since these closely mimic clinical injuries. In rats, like humans, contusions create a cavitation within the cord which forms a glial scar over time. Cavitations do not occur in mice following contusion injuries, which is a major reason that rats are the preferred rodent model for the study of SCI.
Deafferentation Injury Models that Remove Somatosensory Input
Injuries to primary afferent neuron populations relaying sensory information from the hand to the cortex have been studied extensively over the last 30 years. They include investigations into peripheral nerve injury, amputations, dorsal rhizotomies, and more central injuries to the spinal cord and dorsal columns (see Figure 2). These studies have shown that where the primary afferent neuron is injured (i.e., peripherally, or centrally to the dorsal root ganglion, or centrally within the CNS), makes a very large difference to the post-injury changes that take place. Here we briefly summarize some of this work, and take into consideration multilevel and multi-pathway changes.
Peripheral Nerve and Amputation Injuries
Though technically not an injury to the cord, peripheral nerve injuries are frequently implicated in SCI. In contrast to the CNS, peripheral nerves retain the capacity to regenerate, and will do so unless blocked by a physical or chemical barrier. This means that following nerve cut, central map changes may be relatively minor, though the cortical somatotopic map may never completely recover, even after cut nerve ends are sutured to promote nerve regeneration. Many neurons in the reorganized cortex develop abnormal multifocal receptive fields on nonadjacent skin regions, and the map becomes disorganized. This does not occur following nerve crush, where regenerating fibers innervate their original peripheral target.
Blocking nerves from innervating their original target (e.g., nerve ligation) results in far more extensive reorganization, as the system attempts to compensate for the loss. After nerve transection, there is initially a “silent zone” within the corresponding region of deafferentation in the cortical map. Within minutes to hours, this region is surrounded by an expanded representation of the adjacent body parts, and in the ensuing weeks and months, there is a gradual “filling in” or expansion of the representation of the adjacent skin surface until the new organization stabilizes.
Following the amputation of the hand or digits, the pattern of cortical reorganization mimics that seen following the permanent section of both sensory and motor nerves supplying the hand. Following digit amputation, for example, the representation of the adjacent hand expands into the deprived region of the cortical map. This can take weeks to months, and parts of the cortex may remain silent during the post-injury period. When the amputation is sufficiently large, as in the case of a whole-limb amputation, there is some expansion of the face representation (which lies adjacent to hand representation) into the hand cortex. Imaging studies in humans with hand or forearm amputations have similarly reported face responses within the deprived hand cortex from many months to years following the amputation. The long time frame for these more substantial changes suggests that local neuronal axonal sprouting/growth, synapse formation and modification, and perhaps even some transneuronal atrophy may contribute. Of course in studies of nerve section or amputation, cortical reorganization reflects subcortical reorganization. Cutting nerves supplying the hand in the squirrel monkey, for example, results in an altered “hand” map in the cuneate nucleus and the thalamic ventrobasal complex.
Dorsal Rhizotomy: Dorsal Root Cut
When dorsal roots that innervate the digits of the hand are sectioned, only the central cut axons of the primary afferents degenerate, while their cell bodies (located within the dorsal root ganglion) and peripheral axons remain alive.
When a smaller dorsal rhizotomy is used (e.g., in our lab) that involves dorsal rootlets explicitly innervating the first two-to-three digits of one hand (Darian-Smith, Lilak, & Alarcón, 2013), reorganization is also extensive. Immediately following the lesion, a “silent region” is present in the corresponding spinal dermatomal map, and in somatotopic maps in the cuneate nucleus, thalamus, and somatosensory cortex. These silent zones persist for many weeks and correspond to an initially severe behavioral deficit. However, over many weeks, and with continuous use of the affected hand, digit function largely recovers (Darian-Smith & Ciferri, 2005), and this recovery coincides with a corresponding reorganization in the cuneate nucleus, the thalamus, and the cortex. Within dorsal rootlets adjacent to the lesion, cutaneous responses also become evident, where they were not observed previously. Anatomical experiments show that a small population of afferent fibers are spared by the lesion, and these continue to innervate the “deafferented” digits. These enter the cord adjacent to the lesion site, and for many weeks following the lesion they are too weak to drive the somatotopic map at any level of the neuraxis. These spared primary afferent fibers sprout locally within the spinal dorsal horn and cuneate nucleus over the post-lesion period, forming presumptive additional connections with input-deprived postsynaptic target neurons, and help drive reorganization at the spinal and higher levels of the pathway.
Our studies also demonstrated a limit to the reorganization and recovery that can occur. Extensive post-lesion reorganization was able to occur in the spinal dorsal horn (and at higher levels) as long as a small number (< 5%) of uninjured primary afferents remained intact, and continued to transmit information from the periphery. For example, in monkeys with lesions that removed most input from the thumb and index finger, there was a complete reemergence of the cortical map over the first 3–4 post-lesion months. However, in macaques with a slightly larger lesion (of 2.5–3 segments) the hand map never completely returned within the cuneate nucleus or cortex.
A series of experiments conducted in the early 1980s provided unique insight into the long-term (> 10 years) effects of what was a very large dorsal root section (C2–T4). Though these experiments were understandably controversial (i.e., they completely deafferented the forelimb/s), several important findings emerged from them. First, chronic deafferentation of such a large magnitude can and does lead to significant transneuronal degeneration over many years. This was apparent in an extensive reorganization of the body map in the corresponding thalamic caudal Ventral Posterolateral nucleus(VPLc) and somatosensory cortex, which could be partly explained by long-term atrophy of inactivated regions. In the cortex, silent zones were not detected after this period of time, but the face was represented over a larger region and abutted the trunk representation in the absence of any input from the deafferented limb. Significant atrophy was also apparent in the cuneate fasciculus of the spinal dorsal horn, the cuneate nucleus of the brain stem (37–48% shrinkage by volume), and the thalamic VPLc nucleus. As a final note, when these monkeys were sufficiently motivated, they reportedly retained some capacity to move their deafferented limb/s. This was only possible with visual feedback and an intact motor pathway, but it does highlight the extraordinary capacity of the CNS to compensate.
Dorsal Column Lesions
Dorsal column lesions (DCLs) involve the primary afferents at a central level, and are typically less debilitating than a dorsal root lesion (DRL), because they are partial and leave spinothalamic inputs intact. In one series of experiments in owl monkeys (Reed, Liao, Qi, & Kaas, 2016), a high dorsal column lesion (at C3–4) removed all inputs to the dorsal column nuclei from the forearm and lower body. Though the spinothalamic pathway remained intact, it was not able to reactivate the cortical map. When even a small afferent population was spared by the lesion, however, this was enough to reactivate the corresponding cortex. When no afferents were spared within the dorsal columns, preserved inputs from the face grew (over 6 months) into denervated regions of the cuneate nucleus from the adjacent trigeminal nucleus (Liao, Reed, Qi, Sawyer, & Kaas, 2018). Within the cortex, face inputs expanded into the adjacent silent “forelimb” cortex. This work implied that the spinothalamic pathway was not a major player in the cortical reorganization observed, at least following complete dorsal column transection, though the role of spinothalamic tract input in recovery following a partial spinal deafferentation injury remains unclear.
In our own work in macaque monkeys, a unilateral DCL was made that was confined to the lateral cuneate fasciculus at C5. This resulted in localized cortical reorganization in the region of partially deafferented digit representation. In addition, and surprisingly, this central injury also induced significant bilateral terminal sprouting of both the Primary somatosensory cortex (S1) and Primary motor cortex (M1) corticospinal tracts (CSTs) within the cervical spinal cord (Fisher et al., 2018), which is in direct contrast to what we observed following a DRL (see Figure 4). In fact, the DCL induced sprouting from the S1 and M1 CSTs that more closely resembled the CST responses following a combined DRL/DCL (see next section) (Darian-Smith, Lilak, Garner, & Irvine, 2014; Fisher et al., 2018). In short, it was the involvement of the CNS in this injury, in contrast to the DRL alone, that induced the CST sprouting.
In humans, spinal injuries are rarely restricted to the dorsal column, so it is difficult to assess the neural bases for acute or chronic changes following injury. It can be assumed, however, that the CST sprouting observed in macaque monkeys, which is more dramatic than has been observed in rodent studies, closely reflects, or even underrepresents the clinical response.
Combined Dorsal Root and Dorsal Column Lesion
When a dorsal root (associated with digits 1–3), and the dorsal cuneate fasciculus (at C5) are both targeted and cut unilaterally in macaques, the somatosensory CST sprouting within the cord is even more dramatic than that observed following a dorsal column lesion alone (Darian-Smith et al., 2014; Fisher et al., 2018). This is summarized in Figure 4, where three deafferentation models are compared. It might be expected that a combined lesion would produce a response that is the sum of the two components of the injury (i.e., the dorsal root plus the dorsal cuneate fasciculus lesions), but this was not found to be the case (Fisher et al., 2018), and our recent published work indicates that the peripheral and central lesions interact. Behaviorally, preliminary findings in our lab suggest that monkeys with a combined lesion have a deficit and recovery that is indistinguishable from that of monkeys with a DRL alone.
Contusion and Hemisection Lesions
Contusion lesions are widely used in rat models of SCI, as they closely simulate clinical injuries, but they are difficult to replicate, interpret, or translate to the clinic (Baklaushev et al., 2019), and as a consequence they are seldom used in nonhuman primate (NHP) studies looking at cervical injury (Ye et al., 2016—thoracic), where necessarily small sample numbers demand high reproducibility. They have, however, been invaluable in the study of inflammatory responses, molecular changes, and proof-of-principle mechanisms, which can then be targeted in NHPs.
Hemisections in monkeys (that transect one side of the cord), also involve sensory and motor pathways and have been used in NHP models of spinal injury (Galea & Darian-Smith, 1997a, b; Rosenzweig et al., 2010; Nout et al., 2012a,2012b). These studies have shown that within months of a hemisection lesion, which is initially very debilitating and which involves both ascending and descending pathways, animals are capable of a remarkable amount of functional recovery, which again highlights an extraordinary capacity for compensatory changes in the neuronal circuitry. This underscores an evolutionary redundancy in the circuitry, which is particularly evident in the motor pathways (Isa, 2017; Lemon, Landau, Tutssel, & Lawrence, 2012).
Role of the Major Descending Pathways in SCI Recovery
The corticospinal tract (CST) is the major descending pathway mediating manual dexterity in primates, and as such has long been central to the study of recovery following SCI. It is a complex of pathways with multiple origins, and was aptly described nearly four decades ago as “ . . . a group of corticospinal tracts, each of which is involved in a different aspect of the control of sensorimotor processes” (Murray & Coulter, 1981). When the CST is itself cut as a result of a SCI, the cut fibers notoriously fail to regenerate axons within the cord, and attempts to regenerate these lost axons and terminals has been the focus of considerable research.
In contrast, CST axons that are spared following a central injury are able to sprout when their normal circuitry is disrupted. For example, following a deafferentation injury, where the CST as a whole is entirely spared, we (Fisher et al., 2018) have shown that different subcomponents of the CST are capable of significant spontaneous terminal sprouting within the spinal cord (Figure 4). This is not regeneration but an extension of intact axon terminals.
While the CST is not as prominent in rodent forepaw innervation and function, different CST subcomponents in rats respond to deafferentation injuries in a strikingly similar way to that observed in NHPs. A recent study that looked at the M1 CST following a DRL reported motor CST sprouting within the spinal cord following a cervical DRL (Jiang, Zaaimi, & Martin, 2016). In our own investigations in a rat model of a DRL/DCL (McCann et al., 2019, unpublished), we also observed S1 and Rostral frontal area (RFA) motor CST terminal sprouting within the cervical cord in the post-lesion months.
The reticulospinal tract also contributes to hand and forelimb function, albeit to a lesser extent than the CST, and this pathway has been implicated in post-injury reorganization in primates and rodents (Baker et al., 2015). The details of the role played by this tract, however, are not well understood, in part because the anatomy of this pathway within the brain stem is complex and the circuitry difficult to delineate (Zaaimi, Soteropoulos, Fisher, Riddle, & Baker, 2018).
Though not strictly a descending pathway, the C3–C4 propriospinal network (Isa, 2017; Kinoshita et al., 2012) has also been implicated in circuit reorganization following CST lesions, is accessed directly from the motor cortex, and is known to contribute to the recovery of fine-digit function (Tohyama et al., 2017).
Role of Peripheral Receptors in SCI Recovery
Surprisingly little is currently known about what happens to peripheral receptor populations following central spinal cord lesions. In monkeys, peripheral nerve transection and resuture causes Meissner’s corpuscles (MCs) to atrophy, but they then recover with axonal reinnervation by 9 months post-lesion (Dellon, 1976). Recent work in our lab (Figure 5; Crowley, Lilak, Ahloy-Dallaire, & Darian-Smith, 2019) shows that following a DCL or combined DRL/DCL, the density of Meissner’s corpuscles in the finger pads of deafferented digits falls by as much as 30% over the first 4–5 months post-lesion. Since these lesions permanently remove > 95% of the original innervation to the targeted digits, they are fundamentally different from a peripheral nerve lesion that involves axonal regeneration. Importantly, our work shows that by 1 year post-injury, the density of MCs returns to normal levels in the dermis of the affected skin, and suggests that these MCs are reinnervated by the terminal sprouting of the greatly reduced primary afferent population at the periphery. Further studies are now required to understand the peripheral mechanisms that enable these changes to occur.
Importance of the Inflammatory Response in the Recovery Process Following SCI
Recent data from several sources, including work from our lab, suggests that inflammation within CNS following SCI enhances axonal outgrowth and the recovery process (Fisher et al., 2018; Torres-Espin et al., 2018).
The importance of the inflammatory response within the CNS is demonstrated in our own work, where corticospinal tract sprouting was directly compared in three different deafferentation injuries; a dorsal rhizotomy (DRL), a dorsal column lesion (DCL), and a lesion combining both of these (DRL/DCL) (see Figure 4 and Fisher et al., 2018). All three lesions removed input from the hand, and the DRLs explicitly targeted the thumb, index, and middle fingers used in precision grip. Behaviorally, the greatest deficit in hand function was incurred following a restricted DRL (either on its own, or as part of a combined DRL/DCL). This is because DRLs permanently block all afferent information entering the cord via the cut dorsal roots, and DCLs always leave spinothalamic input intact, so are partial. On its own, the peripheral DRL leads to a significant reduction of input to the cord from the affected somatosensory CST (Darian-Smith et al., 2013). However, as soon as a central injury is involved (i.e., DCL), the S1 and M1 CSTs sprout extensively into the cord (Darian-Smith et al., 2014; Fisher et al., 2018). The most parsimonious explanation for the different outcomes is that the central inflammatory response created an environment that was permissive for the growth of axon terminals with disrupted/lost synaptic connections. Many studies have reported changes to this environment during the post-lesion acute and semi-acute phases, but this exciting field is still relatively nascent, as additional complexities and subpopulations of reactive cell types are discovered (e.g., A1 and A2 astrocyte populations; Liddelow & Barres, 2017).
Biomarkers of Spinal Recovery
Experimentally, there are two main approaches taken to quantify recovery, and attempts are often made to correlate these. A great many behavioral assays have been developed over the years (Darian-Smith, 2007; Fouad, Hurd, & Magnuson, 2013; Geissler, Schmidt, & Schallert, 2013) which are used to directly define physical deficits and to monitor functional recovery/compensation post-lesion. Their utility depends entirely on how closely they test the specific loss of function incurred by the lesion under study. Axonal sprouting is also used as an indirect anatomical correlate to behavior, and has its own set of limitations (see next section).
The Axonal Sprouting Conundrum
Many studies, including the deafferentation work described in this manuscript, have used axon terminal sprouting, particularly from the motor corticospinal tract (CST), to demonstrate circuit remodeling following spinal cord injury. But what exactly do these studies tell us about recovery, and circuit remodeling?
In our own research in macaque monkeys, where lesions only involved the dorsal roots (DRL) or dorsal column (DCL) (or a combination of the two), the CST pathway was entirely spared. In this scenario, as long as a central DCL was involved, both the motor (Area 4) and somatosensory cortex (3b/1) CSTs sprouted spontaneously and terminated bilaterally well beyond their normal range within the cord (Darian-Smith et al., 2014; Fisher et al., 2018). In contrast, when monkeys received a dorsal root lesion alone (Darian-Smith et al., 2014), the primary somatosensory cortex (S1) CST retracted to 60% of its original terminal territory. So, at least some of the axonal sprouting observed in our monkeys following a central lesion may, in effect, be an artifact of the inflammatory response activated by the central injury. Preliminary findings suggest that the behavioral deficit and recovery in hand/digit use is indistinguishable following a DRL and a DRL/DCL, despite the starkly different CST responses produced by each in the cord. If the behavioral similarities are confirmed in additional animals, this would mean that much of the sprouting observed may have little impact on the functional recovery observed.
By comparison, when the CST itself is transected by the spinal injury, as might be observed following a contusion or hemisection injury, it has been shown that cut axons do not spontaneously regrow, even though spared CST axons are able to sprout at their terminal endings, both above and below the site of injury. Much work has been directed toward inducing CST terminal growth (e.g., with the introduction of stem cells, neurotrophic factors, CSPGs, various permissive cytokines, etc.), with some impressive results (Anderson et al., 2018; Lu et al., 2014; Plant et al., 2018; Rosenzweig et al., 2010). However, the idea that axonal growth directly enables functional recovery is not a straightforward rule, and it continues to be of considerable interest in spinal injury research.
There are clearly a number of caveats, then, that must be considered when interpreting the functional relevance of axon terminal sprouting in spinal cord injury (SCI) studies that seek to correlate sprouting with functional recovery. Do the newly formed axons form functional synapses? How do the terminal distributions change over time—is there an initial overabundance of terminal growth and synapse formation which is later pruned in a sequence normally associated with developmental maturation of a pathway? How is the neuronal circuitry affected distant to the lesion site (i.e., in pathways directly or even indirectly affected by the injury)?
Clinical Biomarkers of SCI
The current gold standard for classification of SCI involves scoring using the International Standards for Neurological Classification of SCI (ISNCSCI) examination (grading patients from A to E on the American Spinal Injury Association [ASIA] scale), combined with magnetic resonance imaging (MRI). The idea is to stratify patients according to the severity of injury, which can then inform prognosis to some degree. However, ISNCSCI requires patient compliance and is not always possible to use early post-injury where patients are unconscious or have compounding injuries such as head trauma. Moreover, there is significant heterogeneity within each ASIA classification and hugely variable levels of recovery from injuries which are considered to have similar physiology, so there is no clear indicator which reflects anatomical or functional improvement over time.
There is an obvious need for a quantifiable biomarker which can not only classify injury severity accurately, but which will then guide treatment and ultimately provide a meaningful prognosis for patients (Badhiwala, Wilson, Kwon, Casha, & Fehlings, 2017). Researchers are currently targeting potential candidates in blood or cerebrospinal fluid (CSF). A number of studies have identified CSF markers which are altered following injury such as inflammatory cytokines (e.g., interleukins IL-6 and IL-8), and glial fibrillary acidic protein (GFAP). Levels of both correlate with motor-score improvement over time, and can help identify patients who are unlikely to show spontaneous recovery of at least one ASIA class. Importantly, however, although cytokines are known to fluctuate greatly around the time of the injury, none of these markers have yet been followed long term (beyond a week) and in most cases they have already normalized by this point anyway. In addition, an invasive spinal tap is required to drain CSF which is often not practical immediately post-injury.
Other groups are looking at blood-serum biomarkers for SCI, which would be more easily accessible than CSF. This work reports similar correlations between inflammatory markers and injury severity to those isolated from CSF (Ahadi et al., 2015). Kuhle and colleagues (2015) have also demonstrated that levels of neurofilament light chain are correlated with injury severity and outcome. Such studies show clear potential, but more needs to be done to understand the relationship between molecular factors and the functional state of the spinal cord after injury. Specifically, patients need to be followed long term to track recovery so that the real-world prognostic potential of different markers can be determined. It is also clear that most outcome measures continue to rely solely on ASIA classification, which does not adequately scale subtle changes in motor/sensory ability.
Overall, there are currently no sensitive and specific biomarkers of SCI which are routinely used in the clinic. Research efforts are ongoing, but need to be directed toward customizing treatment, identifying candidates for clinical trials, and predicting recovery.
We are a long way from being able to accurately predict an individual’s capacity to recover functionally after a spinal injury. We are, however, getting to a point where we can, in theory, scan an individual’s injury over time and grossly determine how much sparing of specific fiber tracts there is. It is clear from many studies that where there is sparing of input/output pathways (even < 5% of the original), there is the potential for considerable circuitry reorganization and functional recovery in the form of compensation. Which tracts (and how much of them) are affected does matter. Understanding the anatomy of the injury is important, and understanding that changes can and do occur (to a greater or lesser extent) throughout the neuraxis, from periphery to cortex, wherever the normal flow of input is blocked or disrupted, is also clearly important. Therapies that target, for example, neurotrophins at the site of spinal injury, may be sufficient to alter circuitry within the spinal cord, brainstem, and cortical levels, but as yet we do not know whether recovery might be helped by targeting multiple levels of the pathway simultaneously. We are still trying to work out what changes can occur, and where, in response to different injuries.
Ahadi, R., Khodagholi, F., Daneshi, A., Vafaei, A., Mafi, A. A., & Jorjani, M. (2015). Diagnostic value of serum levels of GFAP, pNF-H, and NSE Compared with clinical findings in severity assessment of human traumatic spinal cord injury. Spine, 40(14), E823–830.Find this resource:
Anderson, M. A., O’Shea, T. M., Burda, J. E., Ao, Y., Barlatey, S. L., Bernstein, A. M., . . . Sofroniew, M. V. (2018). Required growth facilitators propel axon regeneration across complete spinal cord injury. Nature, 561(7723), 396–400.Find this resource:
Badhiwala, J. H., Wilson, J. R., Kwon, B. K., Casha, S., & Fehlings, M. G. (2017). A review of clinical trials in spinal cord injury including biomarkers. Journal of Neurotrauma, 35, 1906–1917.Find this resource:
Baker, S. N., & Perez, M. A. (2017). Reticulospinal contributions to gross hand function after human spinal cord injury. Journal of Neuroscience, 37, 9778–9784.Find this resource:
Baker, S. N., Zaaimi, B., Fisher, K. M., & Edgeley, S. A. (2015). Pathways mediating functional recovery. Progress in Brain Research, 218, 389–412.Find this resource:
Baklaushev, V. P., Durov, O. V., Kim, S. V., Gulaev, E. V., Gubskiy, I. L., Konoplyannikov, M. A., . . . Ahlfors, J. E. (2019). Development of a motor and somatosensory evoked potentials-guided spinal cord Injury model in non-human primates. Journal of Neuroscience, Methods. 311, 200–214.Find this resource:
Bentivoglio, M., & Rustioni, A. (1986). Corticospinal neurons with branching axons to the dorsal column nuclei in the monkey. Journal of Comparative Neurology, 253(2), 260–276.Find this resource:
Burman, K., Darian-Smith, C., & Darian-Smith, I. (2000). Geometry of rubrospinal, ruboolivary, and local circuit neurons in the macaque monkey red nucleus. Journal of Comparative Neurology, 423(2), 197–219.Find this resource:
Canedo, A. (1997). Primary motor cortex influences on the descending and ascending systems. Progress in Neurobiology, 51(3), 287–335.Find this resource:
Capogrosso, M., Milekovic, T., Borton, D., Wagner, F., Moraud, E. M., Mignardot, J. B., . . . Courtine, G. (2016). A brain-spine interface alleviating gait deficits after spinal cord injury in primates. Nature, 539, 284–288.Find this resource:
Crowley, M., Lilak, A., Ahloy-Dallaire, J., & Darian-Smith, C. (2019). Spinal cord injury transiently alters Meissner’s corpuscle density in the digit pads of macaque monkeys. Journal of Comparative Neurology, 527, 1901–1912.Find this resource:
Darian-Smith, C. (2007). Monkey models of recovery of voluntary hand movement after spinal cord and dorsal root injury. ILAR Journal, 48(4), 396–410.Find this resource:
Darian-Smith, C. (2009). Synaptic plasticity, neurogenesis, and functional recovery after spinal cord injury. Neuroscientist, 15(2), 149–165.Find this resource:
Darian-Smith, C. (2017). Somatosensory plasticity. Reference Module in Neuroscience and Biobehavioral Psychology, 2017, 1–13.Find this resource:
Darian-Smith, C., & Ciferri, M. M. (2005). Loss and recovery of voluntary hand movements in the macaque following a cervical dorsal rhizotomy. Journal of Comparative Neurology, 491(1), 27–45.Find this resource:
Darian-Smith, I., Galea, M. P., Darian-Smith, C., Sugitani, M., Tan, A., & Burman, K. (1996). The anatomy of manual dexterity. The new connectivity of the primate sensorimotor thalamus and cerebral cortex. Advances in Anatomy, Embryology and Cell Biology, 133,1–140.Find this resource:
Darian-Smith, C., Lilak, A., & Alarcón, C. (2013). Corticospinal sprouting occurs selectively following dorsal rhizotomy in the macaque monkey. Journal of Comparative Neurology, 521(10), 2359–2372.Find this resource:
Darian-Smith, C., Lilak, A., Garner, J., & Irvine, K. A. (2014). Corticospinal sprouting differs according to spinal injury location and cortical origin in macaque monkeys. Journal of Neuroscience, 34(37), 12267–12279.Find this resource:
Darian-Smith, C., Tan, A., & Edwards, S. (1999). Comparing thalamocortical and corticothalamic microstructure and spatial reciprocity in the macaque ventral posterolateral nucleus (VPLc) and medial pulvinar. Journal of Comparative Neurology, 410(2), 211–234.Find this resource:
Dellon, A. L. (1976). Reinnervation of denervated Meissner corpuscles: a sequential histologic study in the monkey following fascicular nerve repair. Journal of Hand Surgery, 1, 98–109.Find this resource:
Filipp, M. E., Travis, B. J., Henry, S. S., Idzikowski, E. C., Magnuson, S. A., Loh, M. Y., Hellenbrand, D. J., & Hanna, A. S. (2019). Differences in neuroplasticity after spinal cord injury in varying animal models and humans. Neural Regeneration Research, 14 (1), 7–19.Find this resource:
Fisher, K. M., Lilak, A., Garner, J., & Darian-Smith, C. (2018). Extensive somatosensory and motor corticospinal sprouting occurs following a central dorsal column lesion in monkeys. Journal of Comparative Neurology (2018 Oct 15), 526(15), 2373–2387.Find this resource:
Fouad, K., Hurd, C., & Magnuson, D. S. K. (2013). Functional testing in animal models of spinal cord injury: Not as straight forward as one would think. Frontiers in Integrative Neuroscience, 7, article 85.Find this resource:
Fregosi, M., Contestabile, A., Badoud, S., Borgognon, S., Cottet, J., Brunet, J. F., . . . Rouiller, E. M. (2018). Changes of motor corticobulbar projections following different lesion types affecting the central nervous system in adult macaque monkeys. European Journal of Neuroscience, 48(4), 2050–2070.Find this resource:
Friedli, L., Rosenzweig, E. S., Barraud, Q., Schubert, M., Dominici, N., Awai, L., . . . Courtine, G. (2015). Pronounced species divergence in corticospinal tract reorganization and functional recovery after lateralized spinal cord injury favors primates. Science Translational Medicine, 7(302), 302ra134.Find this resource:
Galea, M. P., & Darian-Smith, I. (1994). Multiple corticospinal neuron populations in the macaque monkey are specified by their unique cortical origins, spinal terminations, and connections. Journal of Comparative Neurology, 4, 166–194.Find this resource:
Galea, M. P., & Darian-Smith, I. (1997a). Corticospinal projection patterns following unilateral section of the cervical spinal cord in the newborn and juvenile macaque monkey. Journal of Comparative Neurology, 381(3), 282–306.Find this resource:
Galea, M. P., & Darian-Smith, I. (1997b). Manual dexterity and corticospinal connectivity following unilateral section of the cervical spinal cord in the macaque monkey. Journal of Comparative Neurology, 381(3), 307–319.Find this resource:
Geissler, S. A., Schmidt, C. E., & Schallert, T. (2013). Rodent models and behavioral outcomes of cervical spinal cord injury. J Spine. July 27; (Suppl. 4), 001.Find this resource:
Gill, M. L., Grahn, P. J., Calvert, J. S., Linde, M. B., Lavrov, I. A., Strommen, J. A., & Zhao, K. D. (2018). Neuromodulation of lumbosacral spinal networks enables independent stepping after complete paraplegia. Nature Medicine, 24,1677–1682.Find this resource:
Isa, T. (2017). The brain is needed to cure spinal cord injury. Trends in Neurosciences, 40(10), 625–636.Find this resource:
Jiang, Y. Q., Zaaimi, B., & Martin, J. H. (2016). Competition with primary sensory afferents drives remodeling of corticospinal axons in mature spinal motor circuits. Journal of Neuroscience, 36(1), 193–203.Find this resource:
Kato, S., Kobayashi, K., & Kobayashi, K. (2014). Improved transduction efficiency of a lentiviral vector for neuron-specific retrograde gene transfer by optimizing the junction of fusion envelope glycoprotein. Journal of Neuroscience, Methods, 227, 151–158.Find this resource:
Kinoshita, M., Matsui, R., Kato, S., Hasegawa, T., Kasahara, H., Isa, K., . . . Isa, T. (2012). Genetic dissection of the circuit for hand dexterity in primates. Nature, 487(7406), 235–238.Find this resource:
Kuhle, J., Gaiottino, J., Leppert, D., Petzold, A., Bestwick, J. P., Malaspina, A., . . . Casha, S. (2015). Serum neurofilament light chain is a biomarker of human spinal cord injury severity and outcome. Journal of Neurology, Neurosurgery and Psychiatry, 86(3), 273–279.Find this resource:
Lemon, R. N., Landau, W., Tutssel, D., & Lawrence, D. G. (2012). Lawrence and Kuypers (1968a, b) revisited: Copies of the original filmed material from their classic papers in Brain. Brain, 135(7), 2290–2295.Find this resource:
Liddelow, S. A., & Barres, B. A. (2017). Reactive astrocytes: Production, function, and therapeutic potential. Immunity, 46(6), 957–967.Find this resource:
Liao, C. C., Reed, J. L., Qi, H. X., Sawyer, E. K., & Kaas, J. H. (2018). Second-order spinal cord pathway contributes to cortical responses after long recoveries from dorsal column injury in squirrel monkeys. Proceedings of the National Academy of Sciences of the United States of America, 115(16), 4258–4263.Find this resource:
Lo, C., Tran, Y., Anderson, K., Craig, A., & Middleton, J. (2016). Functional priorities in persons with spinal cord injury: using discrete choice experiments to determine preferences. Journal of Neurotrauma, 33(21), 1958–1968.Find this resource:
Lu, P., Woodruff, G., Wang, Y., Graham, L., Hunt, M., Wu, D., . . . Tuszynski, M. H. (2014). Long-distance axonal growth from human induced pluripotent stem cells after spinal cord injury. Neuron, 83(4), 789–796.Find this resource:
Maxwell, D. J., & Soteropoulos, D. S. (2019). The mammalian spinal commissural system; properties and functions. Journal of Neurophysiology.Find this resource:
McCann, M. M., Fisher, K. M., Ahloy-Dallaire, J., & Darian-Smith, C. (2019). Somatosensory corticospinal tract axons sprout within the cervical cord following a dorsal root/dorsal column spinal injury in the rat. Journal of Comparative Neurology.Find this resource:
Murray, E. A., & Coulter, J. D. (1981). Organization of corticospinal neurons in the monkey. Journal of Comparative Neurology, 195(2), 339–365.Find this resource:
Nout, Y. S., Ferguson, A. R., Strand, S. C., Moseanko, R., Hawbecker, S., Zdunowski, S., . . . Bresnahan, J. C. (2012a). Methods for functional assessment after C7 spinal cord hemisection in the rhesus monkey. Neurorehabilitation and Neural Repair, 26(6), 556–569.Find this resource:
Nout, Y. S., Rosenzweig, E. S., Brock, J. H., Strand, S. C., Moseanko, R., Hawbecker, S., . . . Tuszynski, M. H. (2012b). Animal models of neurologic disorders: a nonhuman primate model of spinal cord injury. Neurotherapeutics, 9(2), 380–392Find this resource:
Nudo, R. J., & Masterton, R. B. (1990). Descending pathways to the spinal cord, III: Sites of origin of the corticospinal tract. Journal of Comparative Neurology, 296(4), 559–583.Find this resource:
Oyinbo, C. A. (2011). Secondary injury mechanisms in traumatic spinal cord injury: a nugget of this multiple cascade. Acta Neurobiologiae Experimentalis, 71, 281–299.Find this resource:
Plant, G. W., Weinrich, J. A., & Kaltschmidt, J. A. (2018). Sensory and descending motor circuitry during development and injury. Current Opinions in Neurobiology, 53,156–161.Find this resource:
Reed, J. L., Liao, C. C., Qi, H. X., & Kaas, J. H. (2016). Plasticity and recovery after dorsal column spinal cord injury in nonhuman primates. Journal of Experimental Neuroscience, 10(Suppl. 1), 11–21.Find this resource:
Rosenzweig, E. S., Courtine, G., Jindrich, D. L., Brock, J. H., Ferguson, A. R., Strand, S. C., . . .Tuszynski, M. H. (2010). Extensive spontaneous plasticity of corticospinal projections after primate spinal cord injury. Nature Neuroscience, 13(12), 1505–1510.Find this resource:
Rossignol, S., Martinez, M., Escalona, M., Kundu, A., Delivet-Mongrain, H., Alluin, O., & Gossard, J. P. (2015). The “beneficial” effects of locomotor training after various types of spinal lesions in cats and rats. Progress in Brain Research, 218, 173–198.Find this resource:
Tohyama, T., Kinoshita, M., Kobayashi, K., Isa, K., Watanabe, D., Kobayashi, K., Liu, M., & Isa, T. (2017). Contribution of propriospinal neurons to recovery of hand dexterity after corticospinal tract lesions in monkeys. Proceedings of the National Academy of Sciences of the United States of America, 114(3), 604–609.Find this resource:
Torres-Espin, A., Beaudry, E., Fenrich, K., Fouad, K. (2018). Rehabilitative training in animals models of spinal cord injury. Journal of Neurotrauma, 35, 1970–1985.Find this resource:
Tran, A. P., Warren, P. M., & Silver, J. (2018). The biology of regeneration failure and success after spinal cord injury. Physiological Reviews, 98(2), 881–917.Find this resource:
van Niekerk, E. A., Tuszynski, M. H., Lu, P., & Dulin, J. N. (2016). Molecular and cellular mechanisms of axonal regeneration after spinal cord injury. Molecular & Cellular Proteomics, 15(2), 394–408.Find this resource:
Wolpaw, J. R. (2010). What can the spinal cord teach us about learning and memory? Neuroscientist, 16(5), 532–549.Find this resource:
Ye, J., Ma, M., Xie, Z., Wang, P., Tang, Y., Huang, L., . . . Zeng, Y. (2016). Evaluation of the neural function of nonhuman primates with spinal cord injury using an evoked potential-based scoring system. Scientific Reports, 6, 33243.Find this resource:
Yoshida, Y., & Isa, T. (2018). Neural and genetic basis of dexterous hand movements. Current Opinions in Neurobiology, 52, 25–32.Find this resource:
Zaaimi, B., Soteropoulos, D. S., Fisher, K. M., Riddle, C. N., & Baker, S. N. (2018). Classification of neurons in the primate reticular formation and changes after recovery from pyramidal tract lesion. Journal of Neuroscience, 38(27), 6190–6206.Find this resource: