The Corticospinal System and Amyotrophic Lateral Sclerosis: IFCN handbook chapter

(cid:1) This article reviews the organization and function of the human corticospinal tract and its cortico-motoneuronal connections. (cid:1) Accumulating evidence suggests the motor symptoms of ALS reﬂect pathology selectively targeting cortico-motoneuronal connections. (cid:1) Dysfunction and degeneration of cortico-motoneuronal connections impacts not only muscle strength but also motor skill; effects on the latter may be the distinctive feature of ALS.


Introduction
Pathology of the corticospinal system has long been associated with amyotrophic lateral sclerosis (ALS)/motoneurone disease (MND) (Mott, 1895;Smith, 1960;see Eisen, 2021).This chapter will review some of the current knowledge about the anatomy and function of the corticospinal system, with special reference to its key features in humans and other primates.Many of these features are important for our understanding of both the symptoms and disease progression that typify ALS.In parallel with the corticospinal projection from cortex to spinal cord, there is an important corticobulbar projection to brainstem motor centres, which is critical for actions such as speech, chewing and swallowing.Corticospinal and corticobulbar projections are implicated in limb onset and bulbar onset forms of ALS.
The corticospinal system is very well developed in humans compared with other species, with over one million fibres in each side of the pyramidal tract.It is the largest descending pathway in the human motor system, with a widespread cortical origin, a huge range of axon diameters and a highly distributed set of synaptic terminations within all regions of the spinal grey matter.Indeed, it is important to state at the outset that the corticospinal system is multifunctional, and is concerned not only with movement, but also has somatosensory, autonomic and trophic functions (Lemon 2008).During the preparation and execution of movement it does not function in isolation, but in concert with other brainstem and spinal motor systems.

Origins of the corticospinal tract projection (CSP)
Several different frontal cortical areas contribute to the corticospinal tract (CST), including primary motor (M1) and premotor cortex (ventral (PMv) and dorsal divisions (PMd), supplementary motor area and cingulate motor areas.Retrograde labelling of these neurons in the monkey has been used to determine the relative contributions to the CSP from these different areas.The largest comes from M1 (50%) with further large contributions from cingulate (21%) and supplementary motor areas (19%) (Dum and Strick, 1991).Projections from the parietal lobe include a large CSP arises from the primary somatosensory cortex (S1), with smaller projections from areas 5 and 7.All corticobulbar and corticospinal projections arise from layer V pyramidal neurons.Although the presence of an axon reaching the spinal cord is the defining feature of a corticospinal neuron, we should not forget that each of these neurons gives off many collaterals which establish synaptic contacts with a widespread 'connectome' at cortical and subcortical levels.A recent study suggested that in the primate, the main constituents of the connectome are other motor-related structures such as the red nucleus, pontine nuclei and reticular formation (Sinopoulou et al 2022).
Because of its direct action on spinal motor mechanisms, the CSP represents one of most influential of all the descending motor pathways (Lemon 2008).On the other hand, the number of fibres making up the CSP is small compared with those belonging to all the other corticofugal systems with which they are intermingled in the subcortical white matter, internal capsule and cerebral peduncle.Estimates based on the comparison of the crosssectional area of the cerebral peduncle (which would include all the descending cortical projections to midbrain, brainstem and cord) with that of the medullary pyramidal tract, suggest that the CSP makes up less than 10% of all corticofugal projections that arise from the cortex (Glickstein 2014 p418).This must be accounted for when using tractography of the CSP through the peduncle and internal capsule.Many corticofugal fibres project to the pontine nuclei and reticular formation.

Wide range of axon diameters within the corticospinal tract
In most mammals, the CST is dominated by small corticospinal fibres, ranging in diameter from 0.5 lm or even smaller, up to around 3 lm (Kuypers 1981;Leenen et al 1985;Perge et al 2012).Large-bodied primates, including humans, possess thicker fibres that are up to around 12 lm diameter in macaques and somewhat larger in humans (Fig. 4D; Lassek 1942).The result is a remarkably wide spectrum of axon diameters in these primates, around one-hundred fold in the macaque (Firmin et al 2014).Fine axons seem to arise from all cortical motor areas, while the largest fibres arise from area 4/M1 (Innocenti et al 2019), with a bias towards faster fibres from the caudal subdivision of M1 ('new' M1 or area 4p in humans) (Witham et al 2016).
The faster corticospinal fibres (those with conduction velocities of > 60 m/s) are in the minority, but they have a functional importance out of all proportion to their numbers.The larger corticospinal cell somas (including those of the Betz cells) tend to dominate pathological reports of human motor cortex, while studies using non-invasive cortical stimulation tend to focus on the short-latency motor evoked potentials (MEPs; Day et al 1989) mediated by large diameter, fast-conducting axons.High velocities reduce conduction delays, which in a relatively large primate maybe important during transitions from movement to posture in skilled grasp (Venkadesan and Valero-Cuevas 2009).
Very little is known about the function of the numerous corticospinal neurons with small diameter, slowly-conducting axons.They are largely missing from neurophysiological recording studies of single neurons in motor cortex (Kraskov et al 2019), most likely because their action potentials are small and difficult to record with conventional techniques.Newer methods have revealed that it is possible to identify them (Kraskov et al 2020), and therefore possible to begin to study their function, but so far the search has been unrewarding (Ibanez et al 2021).
It is known that the larger CS fibres are particularly vulnerable to disease and to trauma, for example, during spinal cord injury (Quencer et al 1992;Blight 1991).Neuropathological studies have revealed striking evidence for the selective impact of ALS on the larger diameter CS axons (see below).In some autosomal recessive forms of the disease, there is evidence for preservation of a slowlyconducting CM system (Weber et al 2000a).

Corticospinal projections and terminations
The projections from each different neocortical area terminate in a rather specific manner within the spinal grey matter (Maier et al 2002;Morecraft et al 2013Morecraft et al ,2019Morecraft et al ,2023)).The diverse nature of the extensive cortical territory giving rise to the total CSP, com-bined with differences in fibre spectrum and in the pattern of spinal termination, all suggest that the corticospinal system mediates multiple functions (Lemon 2008).For example, there is a heavy CSP from the S1 somatosensory hand representation (area 3b and 1) to the dorsal horn of the spinal cord (Ralston and Ralston 1985), which completely avoids the ventral horn.This is in striking contrast to the characteristic pattern of projections from the M1 hand area, which avoid the dorsal horn, but terminate heavily in Rexed lamina VII and amongst the motor nuclei of Rexed lamina IX (Maier et al 2002;Morecraft et al 2013).
In order to gauge the functions of the different projections, it is important to quantify their connections within the spinal cord.Morecraft and colleagues have achieved this by using stereological techniques to count the number of labelled boutons that belong to Fig. 1.Summary of corticospinal projection from the caudal region of M1 hand/arm cortex in the macaque.In A, illustrating the overall projection, the relative intensity of projections to spinal cord laminae is indicated by line thickness and arrow head size.B illustrates in 3D format the relative strength of the contralateral projection to the anatomical subsectors of laminae VII and IX at each segmental level.The relative height of each column corresponds to the relative number of estimated boutons for each subsector (vm, ventro-medial; vl, ventro-lateral, dm, dorso-medial, dl, dorso-lateral).The mean number of labelled boutons obtained from the three stereology cases is depicted for laminae VII and IX at each cervical level.Note the gradual decrease in the lamina VII projection from rostral to caudal levels and the gradual increase in the lamina IX projection.The projection to lamina VII was consistently most dense in the dorsomedial, dorsolateral, and ventrolateral subsectors at levels C7-T1.The lamina IX projection was preferentially distributed to the dorsal quadrants at C5-C7 but was more widely dispersed among the quadrants at C8 and T1.(with permission from Morecraft et al 2013).

R. Lemon
Clinical Neurophysiology 160 (2024) 56-67 a given CSP within defined sectors/laminae of the spinal grey matter.In an extensive and systematic series of experiments; they labelled boutons using anterograde transport of neuroanatomical tracers injected into physiologically characterised cortical areas (or subdivision) in the macaque monkey (Morecraft et al 2013(Morecraft et al , 2019(Morecraft et al , 2023)).
Their research showed that greatest number of projections to the C5-T1 spinal segments from the caudal hand/arm subdivision of M1 ('new' M1) terminated within lamina VII of the intermediate zone (around 59% of all labelled boutons; Morecraft et al 2013; see Fig. 1).The termination was heaviest in the dorsolateral division of lamina VII; this is the origin of short propriospinal projections which influence primarily motoneurons of distal upper extremity (hand, digit) muscles (Kuypers 1981(Kuypers , 1982)).
Projections terminating in lamina IX represented the second most heavy projection from 'new' M1 (around 18%; Morecraft et al 2013 Fig. 1), confirming that the CM projection is a significant component of the M1 CSP, influencing motoneurons innervating flexors acting on the shoulder and elbow rostrally (C5-C7), along with flexors, extensors, abductors and adductors acting on the digits, hand and wrist caudally (C8-T1).The actual proportion of boutons that are on the dendrites of motoneurons may have been underestimated, since many of the boutons within lamina VII could be located on distal dendrites of alpha motoneurons which spread out throughout the ventral horn (see Sinopoulou et al 2022).
When examining the density of boutons from 'new' M1 arm/ hand area in laminae VII and IX of the different cervical spinal segments, Morecraft et al (2013) noted a gradual decrease in the lamina VII projection from rostral to caudal levels and a gradual increase in the lamina IX projection, peaking at C8 and T1.These latter segments harbour the motoneurons innervating the intrinsic hand muscles.This is consistent with a predominant CM input controlling digit movements (Fig. 1).
Projections from 'old' M1 give rise to only a small proportion of the total CM output from primary motor cortex to the cervical cord (around 5%;Rathelot andStrick 2006,2009;Morecraft et al 2023).In addition, 'old M1 0 projects heavily to brainstem centres giving rise to other descending motor pathways, including the reticu-lospinal tract (Kuypers 1958(Kuypers , 1981;;Darling et al 2018).Other frontal motor areas (such as PMv, PMd) do not give rise to a significant number of CM boutons (Morecraft et al 2019).Their M1 projections are more focused on the ventrolateral sectors of lamina VII, indicating a greater involvement, through onward propriospinal and local segmental projections, in the control of axial and proximal arm muscles.Morecraft et al. (2013Morecraft et al. ( ,2019Morecraft et al. ( ,2023) ) found clear differences in the degree to which CSPs arising from distinct cortical areas terminate in the contralateral spinal grey matter.Almost all the arm/hand 'new' M1 projections (98%) terminate contralaterally.A lower proportion of contralateral boutons was found for projections from 'old M1 0 (M1r) (90%) and from dorsal premotor cortex (79%).Projections from leg M1 give rise to markedly bilateral projections (Lacroix et al 2004).

Somatotopy in the corticospinal tract
Corticofugal fibres originating from the leg, arm, and face regions of M1 remain separate from each other as they descend from the cortex in the corona radiata and internal capsule.However, by the time they reach the cerebral peduncle, there is substantial overlap and intermingling of these fibres (Morecraft et al 2007).But what about the more caudal sectors of the corticospinal tract?Foerster (1936) had first postulated somatotopy of fibres within the pyramidal and lateral corticospinal tract (LCST), with fibres destined for control of leg muscles lying laterally, and those for the arm lying medially.Diagrams showing this 'laminated' organisation' of the LCST still appear in several standard textbooks of anatomy and neurosurgery (see Lemon and Morecraft 2023).The lamination idea is potentially of clinical importance because it has been used to try to explain some features of Central Cord Syndrome and other forms of incomplete spinal injury (iSCI), in which there is pronounced impairment of upper limb and particularly hand function, but lower limb function is relatively unimpaired (Schneider et al 1954).This was explained by suggesting that a central cord lesion in the cervical cord would spread to  Morecraft et al (2021).In no case was there any evidence of somatotopy in the pyramidal decussation or within the contralateral LCST as suggested by Foerster (1936).This same striking pattern of fibre intermingling throughout the cranio-vertebral junction (CVJ) and cervical spinal cord was found following tracer injections in the arm/trunk regions of the premotor cortex and supplementary motor cortex (from Lemon and Morecraft 2023, with permission).
involve fibres in the medial sector of the LCST, while the more laterally located lower limb fibres were spared.
On the basis of new neuroanatomical investigations in the macaque monkey, Morecraft et al (2021) and Lemon and Morecraft (2023) provided unequivocal evidence against any lamination or somatotopy in the LCST: there was almost complete intermingling of leg and arm fibres throughout the spinal course of the CST.They suggested that the much greater deficits in upper vs lower limb function seen after iSCI was because the CST was much more critical for upper limb (especially hand and digit movements) than for lower limb movement.A large amount of evidence consistent with this suggestion was reviewed by Lemon and Morecraft (2023).Importantly, separate MRI-based evidence suggests that iSCI often results in diffuse rather than focal injury to the CST (Huber et al 2018).Therefore it can be anticipated that, because of its greater importance for upper vs lower limb function, and because upper limb fibres are distributed throughout the CST, then such a diffuse injury would be expected to give rise to greater upper than lower limb motor deficits.

Cortico-Motoneuronal projections in monkeys and humans
In the macaque monkey, evidence for CM projections and connections has come from a variety of neuroanatomical tracing techniques and different neurophysiological recording and stimulation techniques (see Morecraft et al 2023).Most neurons with CM pro-jections to upper limb muscles are found in two main cortical areas: 'new' M1 and area 3a (Rathelot andStrick 2006,2009).'New' M1 is restricted to the rostral bank of the central sulcus in the macaque; in humans it is buried deep within the sulcus (Geyer et al 1996;Glasser et al 2016).Some idea of the size of the cortical 'colony' devoted to a single target muscle can be gained from the retrograde transneuronal labelling technique used by Rathelot andStrick (2006,2009) in the macaque.They found that after tracer injection of a single hand or digit muscle, between 248-428 CM neurons were labelled within M1, mostly in 'new' M1.
Intracortical stimulation of 'new' M1 can evoke fast, shortlatency monosynaptic responses in macaque forelimb and hand motoneurons; no such responses were found from 'old' M1 (Witham et al 2016).However, long-latency monosynaptic effects were evoked from both divisions of M1.Other more complex, oligosynaptic effects were also commonly evoked from both divisions.
In the monkey, CM projections arise not only from large pyramidal neurons, as exemplified by the Betz cells, but also from layer V neurons comprising a wide range of soma sizes (Rathelot and Strick 2006).We do not yet have firm data on the exact relationship between soma size and axon diameter, but can assume that CM effects on spinal motoneurons will be mediated by CST axons with a range of conduction velocities.There are certainly clear CM effects from neurons with more slowly-conducting neurons (Porter and Lemon 1993).

R. Lemon
Clinical Neurophysiology 160 (2024) 56-67 In humans, evidence for a CM projection comes from postmortem material showing degenerating terminals of corticospinal fibres amongst motoneuron pools (Kuypers 1958(Kuypers , 1981;;Schoen 1964;Iwatsubo et al 1990).This is consistent with the monosynaptic nature of the earliest component of MEPs generated by noninvasive, transcranial electric (TES) or magnetic (TMS) stimulation of the motor cortex (e.g.Day et al 1989; see Fig. 3, A,D).Some studies have used single motor unit recording to verify that these MEPs include discharge at brief delays that confirm the monosynaptic discharge of the motor unit after cortical stimulation (de Noordhout et al 1999; Palmer and Ashby 1992).TES and TMS of M1 have indicated that many upper and lower limb muscles receive CM projections in humans, with effects on upper limb muscles generally larger than those for the lower limb (e.g.Brouwer and Ashby 1990).It is of course possible that, in humans, the cortical origin of this CM projection may be more extensive than in non-human primates.
In addition, it is likely that CM connections derived from fastconducting axons from M1 mediate the strong beta corticomuscular coherence between EEG/MEG signals, recorded over  29) and clinical progressive muscular atrophy (PMA) a form of motor neurone disease characterised by LMN signs with little UMN involvement.The histograms revealed that percentage of axons measured that were larger than 1 lm was lower in the ALS and PMA groups than in controls.The reverse was true for axons thinner than 1 lm.Note the very small percentage of larger fibres even in the control group.(E) There were also significant differences in the densities of axons that were more than 1 lm in diameter between all pairs of clinical groups: p = 0.001 (*) between the ALS and PMA groups, p = 0.001 (*) between the PMA and controls and p < 0.001 (**) between the ALS and controls.All patients diagnosed with clinical ALS exhibited lower values than controls.In contrast, the results of the clinical PMA group varied widely.(with permission from Riku et al 2014).
motor cortex, and EMG recorded from hand and digit muscles.This is also reflected in EMG-EMG intermuscular coherence (Fig. 3 C, F; Fisher et al 2012;Ibanez et al 2021;Baker et al 2003).
Importantly, both TMS and coherence studies have revealed that there are pronounced differences in the extent of CM influence between muscle groups acting at different upper and lower limb joints.For example, effects on wrist extensors are stronger than those on wrist flexors, while foot dorsiflexors receive stronger effects than plantar flexors (de Noordhout et al 1999; Palmer and Ashby 1992;Brouwer and Ashby 1992;Bawa et al 2002).In ALS patients, these differences may have important clinical consequences (see below).
A final line of evidence comes from studies showing that when the corticospinal tract is damaged there is a clear deficit in the capacity to execute skilled tasks that involve relatively independent finger movements (Lawrence and Kuypers 1968;Lang and Schieber 2003;Eisen and Lemon 2021).
Box 1 Box 1 Primate-specific Properties of the Corticospinal System Compared with rodents, dexterous primates (and particularly macaques), show a number of species-specific features: CM connections: these are particularly well developed for projections to motoneurons supplying the distal muscles of the hand, controlling digit movement, but are also present to a lesser degree in projections to motoneurons of more proximal arm muscles (Lemon 2008).CM effects have also been demonstrated in foot and toe muscle motoneurons (and tail in some primate species).CM effects are generally largest in humans and include corticobulbar targets such as motoneurons supplying intrinsic laryngeal muscles.Fast-conducting corticospinal axons: these may conduct at velocities up to 80-90 m/s (Firmin et al 2014).Fast corticospinal axons are also found in carnivores (Takahashi 1965).High velocity reduces central conduction delays, and thicker axons may be better able to distribute discharges faithfully across collaterals made by a corticospinal neuron (Perge et al 2012).The soma and proximal dendrites of pyramidal neurons in macaque motor cortex express Kv3.1b (fast K+) channels allowing rapid repolarization of the neuron after it has discharged (Soares et al 2017).The fastest conducting corticospinal neurons exhibit 'thin spikes' with trough-topeak duration as brief as 160 lS (Vigneswaran et al 2011;Lemon et al 2021).These last two properties allow for brief bursts of high-frequency discharges in the corticospinal system, particularly at movement onset.

The CM neuron and the UMN/LMN nomenclature
From a clinical perspective, it has always been useful to distinguish the CM cell (and other corticofugal projections) as ''Upper Motor Neurons (UMNs)" from their target alpha motoneurons (Lower Motor Neurons; LMNs).However, when it comes to either the architecture of the CM projection or the complex functional relationship between ''UMN" and ''LMN", there are problems with this nomenclature (see Lemon 2021).
First, a strict one-to-one relationship between a CM cell and its target motoneuron does not exist.CM projections are characterized both by convergence (i.e.many CM cells converge onto the each of the motoneurons innervating a particular muscle) and divergence (i.e.individual CM cells usually project to many motoneurons, including motoneurons belonging to a number of different muscles) (Lemon 2008).Many different convergent excitatory inputs must be active in order to bring the LMN to discharge, and each individual excitatory input generates a relatively small unitary EPSP (in the range of a few hundred lV, which compares with several mV between the motoneuron's resting potential and the threshold for discharge of an action potential).Therefore, discharge of the LMN and the resulting muscle contraction and movement requires synchronous input of many convergent excitatory inputs.This would probably include many of its 'colony' of CM neurons, but also other supraspinal, propriospinal, and segmental inputs.
Second, there is no fixed pattern of recruitment between the UMN and the LMN as voluntary force level increases.Many CM cells already show steady low-frequency activity at rest/very low force levels, and show rapid increases with even small force increments (Cheney and Fetz 1980;Maier et al 1993).Although most CM cells are 'muscle-like' in their pattern of activity, it is not a simple relationship (Griffin et al 2008(Griffin et al , 2015)).Activity in the CM neuron and its target muscle can be readily dissociated by changing task conditions (Fetz andFinocchio 1971,1975;Schieber 2011).
Third, another complication is that a significant proportion of CM cells actually reduce their firing rate with increased force (Maier et al 1993: Glover andBaker 2022), which may be related to cortical mechanisms which disengage particular muscle synergies during skilled manipulation (Maier et al 1993; see also Kraskov et al 2014).It may be that the corticospinal system is more concerned with fine force regulation at low force levels, and that other descending systems (e.g.reticulospinal projections) are responsible for control at higher force levels (Glover and Baker 2022).

Contribution of the CM system to skilled movement
The CM projection is a late evolutionary development which is only present in dexterous primates, such as capuchins, macaques, great apes and humans, but completely lacking in adult rodents, carnivores and many other primates, such as marmosets (Lemon 2008;Gu et al 2017).Why should the direct projection from cortex to a-motoneuron have appeared so late in evolution?One of the earliest ideas was that the CM system could act selectively on the motor apparatus of the upper limb to allow relatively independent digit movements (Phillips 1971;Kuypers 1981;Porter and Lemon 1993).These movements are critical for the execution of all skilled hand tasks, including key human motor characteristics, such as gesture and tool use.While tool use is by no means exclusive to advanced primates and humans with demonstrable CM connections, primates do exhibit a particularly wide range of tool making and tool using behaviours.CM cells have been shown to be strongly active during tool use by trained macaques (Quallo et al 2012).
Some insights into the CM contribution can be gained by looking at the 'muscle field' of a given CM cell: all the muscles receiving CM inputs from that cell (Fetz and Cheney 1980).This muscle field often includes a group of muscles which act together as 'functional synergists', i.e. they act together during complex behaviours such as reach and grasp, or during precision grip (Buys et al 1986;McKiernan et al 1998).This type of muscle field organization may allow the CM system to contribute particular patterns of muscular coordination that are required for the biomechanical stability involved in force control at the most distal extremities.An obvious example is the precision grip between the tips of the index finger and thumb.Consistent with this idea, CM cells have been shown to show strong task-specificity (Muir and Lemon 1983;Quallo et al 2012).
As pointed out above, CM cells generally show a 'muscle-like' pattern of activity i.e., one resembling the timing and pattern of EMG in their target muscles, rather than being 'extrinsic-like' i.e., activity related to the direction of movement produced, independent of arm posture (Griffin et al 2015).However, it is rare for a CM cell to exhibit a pattern of activity similar to that of its target muscle when the muscle is employed as a simple agonist: i.e. a strict one-to-one relationship between socalled 'upper motoneuron' and its target LMN.Instead, individual CM cells can represent agonist, synergist, fixator, and antagonist functions of their target muscles.Indeed, CM outputs organised along these lines show how this system operates to provide a great deal of functional flexibility in the manner of muscle recruitment, a flexibility that cannot be afforded by the relatively fixed synergies represented in spinal motor mechanisms (Rathelot and Strick 2009).A CM cell may exert important inhibitory actions on muscle groups antagonist to their muscle field; this inhibition is probably mediated by spinal inhibitory interneurons.

ALS and the corticospinal system
The involvement of the CST in ALS, recognized in the earliest descriptions of the disease (e.g.Mott, 1895; see Eisen, 2021), is increasingly being understood using behavioural, physiological, and pathological approaches.The accumulating evidence confirms the original suggestion that the CM component of the corticospinal system is implicated in ALS (Eisen et al 1992(Eisen et al , 2017;;Henderson and Eisen 2020;Eisen 2021).

ALS pathology targets the CM system
The evidence suggests that the cell body of corticofugal neurons (including CM cells) in the motor cortex, and their long descending axons, may be involved in the early stages of ALS.The work of Braak et al (2013) has shown that inclusion bodies, characterised by the presence of the phosphorylated form of transactive response DNA binding protein (TDP-43), are found in large pyramidal neurons within the motor cortex of patients with ALS.pTDP-43 pathology is associated not only with dysfunction of the neuronal soma, but also of the axon, with disruption of axonal transport (Coleman 2022).
pTDP-43 pathology then spreads onward via synaptic contacts made by the projections of affected pyramidal neurons to subcortical targets, including alpha motoneurons in brainstem and spinal motor centres, as well as to the reticular formation, pre-cerebellar nuclei and others.Braak et al (2013) suggested that the prion-like spread of pTDP-43 pathology might explain the increasing severity of ALS motor symptoms with time.For example, because of the numerous CM projections to alpha motoneurons of hand and digit muscles, these muscles might be increasingly affected as the disease progressed.The involvement of the CM synapse in the spread of ALS has long been suspected because motoneurons without direct CM input (e.g.Onuf's nucleus, oculomotor nuclei) are generally spared in the early stages of ALS (Braak et al 2013;Eisen et al 2017).
It is interesting that tractography studies of the corticospinal system in patients with so-called 'pure LMN disease' (progressive muscular atrophy) and flail-arm variant of ALS also showed significant CST pathology (Ince et al 2003;Riku et al 2014;Rosenbohm et al 2016Rosenbohm et al , 2022)).But since the load of pTDP-43 varies considerably across different forms of MND/ALS (Nolan et al 2020), this pathology may not predominate in all forms of the disease.
Despite the accumulating evidence for a role in axon dysfunction in ALS, there have been relatively few attempts to quantify the different changes that can be seen in the degenerating axons.
For example, what proportion of axons have fully degenerated and are there differences in the loss of large vs small diameter axons?A selective loss of larger fibres from the region of the lateral corticospinal tract has been reported by several studies which examined the distribution of axon diameters in post-mortem spinal cord tissue from ALS patients.For example, Sobue et al (1987) found a large reduction in the number of large (>7 lm) CST fibres in 6 patients, with much smaller loss of smaller (<7 lm) fibres.Their samples were from the T7 cord segment.Oyanagi et al (1995) made a similar study of the mid-cervical cord from 13 ALS patients and again found a striking loss of large fibres (Fig. 4A-C).
More recently, Riku et al (2014) studied the mid-cervical level in tissue from 29 patients diagnosed with ALS; again, there was a reduction in the proportion of large fibres.Fig. 4D shows that the proportion of all fibres having a diameter > 1 lm was 18.5% of fibres in the ALS cases vs 32.3% in controls.These authors also reported a significant reduction in the density of larger fibres within the LCST compared with controls, reflecting an overall loss of axons from the tract (Fig. 4E).

Corticobulbar vs corticospinal projections in ALS
Frontal motor areas give rise to corticobulbar projections which innervate motor cell groups in the brainstem, including the trigeminal, facial and hypoglossal nuclei, and the motoneurons innervating the larynx and pharynx.These projections are intermingled with corticospinal projections within the internal capsule and cerebral peduncle (Morecraft et al 2017 see Fig. 2); they descend further through the pons and medulla to reach their bulbar targets; some fibres initially travel in the pyramidal tract, leaving the tract and coursing in a dorsal direction along the full length of the medullary pyramidal tract, while others project through the pons and medulla (Kuypers 1958;Morecraft et al 2014).
In parallel with the presence of direct CM connections to spinal motoneurons, there are direct corticobulbar projections to motoneurons located in several brainstem motor nuclei.These appear to be either unique to humans, or at least much better developed in humans than in non-human primates.For example, motoneurons in the n.ambiguus, which innervate the laryngeal muscles, receive direct projections from the cortex, but these are lacking in monkeys (Kuypers 1958;Iwatsubo et al 1990;Espadaler et al 2012;Simonyan 2014).
Projections belonging to either the corticobulbar or corticospinal projection are so heavily intermingled during their descending course (Fig. 2), it is likely that during development they use molecular rather than spatial cues to establish connections with their different targets (Lemon and Morecraft 2023).Differences in the molecular identity of CM projections to brainstem vs spinal motoneurons might then be related to differential vulnerability to diseases such as ALS, with patients then presenting with, respectively, bulbar or limb symptoms.9.3.Changes to the CM system in ALS revealed by TMS Loss of the faster conducting corticospinal axons finally results in a prolongation of the central conduction time, and a less synchronised MEP (Fig. 3A).TMS studies have shown that M1 corticospinal neurons have reduced threshold to non-invasive stimulation early in the disease, reflecting hyperexcitability probably resulting from glutamate toxicity and/or reduced GABA inhibition (Vucic et al 2013).Short-interval intracortical inhibition (SICI) may be reduced or absent.At this early stage, the amplitude of MEPs may be increased.At later stages thresholds rise and the cortex may become 'inexcitable' i.e. no further motor effects can be evoked (Fig. 3 D).

Split phenotypes in ALS and the CM system
Particularly striking insights were made by the detailed study of the 'split-hand syndrome' in ALS (Weber et al., 2000b).It was revealed that the well-known weakness and loss of motor units in hand muscles of ALS patients was more pronounced in thenar (thumb) than in hypothenar (little finger) muscles (Kuwabara et al 1999;Menon et al 2014).Although several different factors are known to contribute to lower (spinal) motoneuron dysfunction, these particular changes were entirely consistent with the loss of the larger CM influence over motoneurons of thenar vs hypothenar muscles, which has been clearly demonstrated in the non-human primate (Clough et al 1968).
For the intrinsic thumb motoneurons, it may be that the CM input represents such a large component of the finely-structured excitatory drive that when this input is compromised by ALS, the result is both a weakness and a poverty of movement (Eisen and Lemon 2021).It is generally accepted that the earliest component of the MEP recorded from a muscle in response to TMS delivered over the motor cortex represents the action of fast, monosynaptic CM inputs to the motoneurons.Weber et al. (2000b) used TMS to show that the MEPs in thenar muscles of ALS patients with splithand syndrome were indeed smaller and more abnormal compared with those in hypothenar muscles.
Interestingly, although the peripheral neuromuscular properties of both these muscle groups showed pathological signs in ALS, these LMN properties were rather similar in thenar and hypothenar muscles, so Menon et al. (2014) suggested that the 'split-hand syndrome' reflects the particular vulnerability of CM projections to the hand, over and above a more general LMN disorder.
Further evidence for the particular vulnerability to ALS of muscle groups with strong CM input has been accumulating.This includes studies of 'split elbow', 'split foot' and 'split ankle' (Henderson & Eisen 2020;Khalaf et al 2019;Ludolph et al 2020).In all three syndromes, the more profound weakness was found in the muscle groups which, in healthy controls, are known to receive relatively strong CM projections.Ludolph et al (2020) included both retrospective and prospective studies of large cohorts of ALS patients whose muscle strength was assessed in different pairs of muscles, using the Medical Research Council scale.When, for each patient, the strength in different pairs of muscle was compared, the study demonstrated statistically greater weakness of the muscle group known to receive stronger CM connections in healthy controls (e.g.wrist extensors vs wrist flexors).This result is again consistent with the loss of CM influence during ALS.

Measures of cortico-muscular coherence reflecting CM system dysfunction in ALS
Since the larger fibres are thought to mediate cortico-muscular beta coherence or the consequent intermuscular coherence (Ibanez et al 2021), it would be expected that both measures would be reduced in ALS.Fisher et al (2012) studied a group of 8 patients diagnosed with primary lateral sclerosis, a variant of ALS with clear signs of 'UMN' involvement, but preservation of LMN function.They found a marked reduction in beta coherence computed between EMG recorded from pairs of either upper limb or lower limb muscles.They also illustrated further reduction in coherence with disease progression (Fig. 3).A similar result was reported by Issa et al (2017).
Proudfoot et al ( 2018) investigated cortico-muscular coherence, between magneto-encephalographic (MEG) activity recorded over motor cortex and EMG recorded from forearm muscles during performance of an auxotonic grip task.They found a significant reduc-tion in beta-band coherence in 17 ALS patients when compared with healthy controls (Fig. 5).They demonstrated that this reduction was not due to there being less cortical beta MEG activity or task-related EMG activity: what was missing was the coherence between these two rhythmic signals, normally provided by fast conducting CST and CM projections.
Although both TMS and coherence studies provide evidence of deficits in the connectivity of the CS/CM systems, this is also true of a number of other motor system disorders that mimic ALS, and is not therefore a unique marker for ALS.For example, Proudfoot et al ( 2018) found reduced cortico-muscular coherence not only in ALS patients but in a group of patients with neuropathy disease mimicking ALS.This finding, and others like it, does not necessarily detract from the importance of understanding corticospinal dysfunction in ALS.This is especially true since TMS and coherence analysis cannot provide insights into the functions transmitted by CS/CM connections, i.e. the functional activity transmitted by these connections.As I stressed at the beginning, the corticospinal system is multifunctional, so that there may well be specific functional activity transmitted by the CS/CM systems which are unique to ALS, and not found in mimics.For example, as Andrew Eisen has highlighted, the progressive deterioration and loss of skilled manipulatory tasks involving fractionation of digit movements is a distinctive and unique feature of ALS (see below; Eisen & Lemon, 2021).A further example might be the long term trophic support of alpha motoneurons that receive CM projections.

R. Lemon
Clinical Neurophysiology 160 (2024) 56-67 9.6.ALS and the performance of skilled movement Many investigators have emphasized ALS as a motoneuron disease, and the major diagnostic tests for confirming ALS/MND involve tests of muscle function and strength.Another view is that ALS is a brain disease with a major impact is on the generation and control of complex movements (Eisen & Lemon, 2021).Recognition of the evolutionary significance of the CM system for human motor capacities such as vocalization and skilled manipulation, together with the vulnerability of this system to ALS (Eisen et al 2014), should result in more attention being directed to the control of skilled movement, in addition to measures of muscle strength.A given muscle can be used in many different ways, and one of the key features of the CM system is the task-specific recruitment of particular groups of muscles (Muir and Lemon 1983;Griffin et al 2015).As pointed out above, there is a task-specific flexibility between activity in a CM neuron and its target motoneurons.This is lost when the CM projection is dysfunctional, and it is predictable that there will be deficits in skill as well as in muscle strength.Interestingly, Proudfoot et al (2018) found no correlation between measures of cortico-muscular coherence and hand strength in ALS patients, whereas a number of investigations have shown a relationship between CMC and task performance and skill in healthy volunteers (Kristeva et al 2007;Perez et al 2006).
The big question is whether standard tests requiring skilled control of the hand and digits can be used diagnostically to identify the effects of ALS on the CM system (Hayden et al 2022).For example, Czell et al (2019) found that the time to complete the standard Nine Hole Peg Test was significantly longer in 20 ALS patients than in age-matched controls, and this measure increased with disease progression.This test was superior to TMS-based measures and was significantly correlated with handwriting scores assessed through the revised ALS functional rating scale questionnaire.Further tests of skill and performance could be carried out to see if they could serve as sensitive biomarkers for disease progression in ALS.
Box 2 Box 2 Animal models Further understanding and development of treatments for ALS depends both on clinical research and on the use of appropriate animal models.ALS is a uniquely human disease and no single animal model can capture all of its features.A macaque model of ALS allows investigation of fast-conducting CST and a well-developed CM system, key elements involved in ALS pathology (Braak et al 2013).Macaque studies (Li et al 2022;Izpisua Belmonte et al 2015) could provide additional insights over and above advances made in rat and mouse studies, which do not share these important features.Indeed, because in the rodent, corticospinal projections from sensorimotor cortex mostly avoid the ventral horn and have limited direct effects on motor control, the layer V pyramidal neurons giving rise to these projections do not qualify as 'upper motor neurons' as widely understood, and are not therefore a good model for study of the effects of ALS on this class of neuron.Small numbers of carefully designed studies in purpose-bred macaques will continue to be needed to progress understanding and treatment of ALS.

Conclusions
There is growing neuropathological, neurophysiological and behavioural evidence that the CM system is critically involved in explaining the debilitating effects of ALS.The recent evolutionary status of this system appears to be consistent with its unique aetiology and progression in humans.Understanding the different consequences of CM dysfunction and degeneration may help to shed light on the symptoms of ALS, including its effects not only on muscle strength but on the performance of skilled actions so characteristic of the human motor repertoire.

Conflict of Interest Statement
I have no financial or conflict of interest to declare.

Fig. 2 .
Fig. 2. Schematic diagram of the intermingling of fibres within the corticospinal tract.Summary diagram illustrating the heavily intermingled pattern of fibres from the M1 arm/hand, M1 shoulder and M1 leg regions in the pyramidal tract in the lower medulla (A), rostral pyramidal decussation (PD) (B), caudal pyramidal decussation (C) and throughout all cervical levels of the contralateral lateral corticospinal tract (cLCST) as shown at the C5 level (D).This summary is based on the data reported byMorecraft et al (2021).In no case was there any evidence of somatotopy in the pyramidal decussation or within the contralateral LCST as suggested byFoerster (1936).This same striking pattern of fibre intermingling throughout the cranio-vertebral junction (CVJ) and cervical spinal cord was found following tracer injections in the arm/trunk regions of the premotor cortex and supplementary motor cortex (from Lemon and Morecraft 2023, with permission).

Fig. 3 .
Fig. 3. A-C Results obtained from an ALS patient on her first laboratory assessment.(A) Motor-evoked potentials (MEPs)are shown from 3 muscles in Patient AB (grey) and an age-matched control subject (black).(B) Raw EMG records show modulation with a precision grip task; arrows indicate trial onset.(C) Intermuscular coherence spectra.Grey boxes indicate the frequency window of interest, and significance levels are represented by grey (Patient AB) and black (control) dashed lines.Note the much lower level of beta coherence compared with control.(D-F) As in A-C, data from the patient's second investigation, 22 months later.Note complete absence of beta coherence in this visit despite clear task-related EMG activity.EDC = extensor digitorum communis; FDI = first dorsal interosseous; FDS = flexor digitorum superficialis; IMCoh = intermuscular coherence.(with permission from Fisher et al 2012).

Fig. 4 .
Fig. 4. Light microscopic photographs of the post-mortem human lateral corticospinal tract in the upper cervical cord.In controls (A) large diameter fibres are clearly present throughout the image (the two large axons in the centre of the image would have diameters around 10-12 lm).(B) In this ALS patient there was a selective loss of large myelinated fibers, whereas in ALS patient (C) there was a more general loss of fibres, including larger ones.Toluidine blue preparation.(with permission from Oyanagi et al (1995).(D) quantified distribution of LCST fibre diameters calculated for post-mortem sections of mid-cervical cord from three groups: age-matched controls (n = 13) clinical ALS (29) and clinical progressive muscular atrophy (PMA) a form of motor neurone disease characterised by LMN signs with little UMN involvement.The histograms revealed

Fig. 5 .
Fig. 5. (A) Average cortico-muscular coherence (CMC) between the forearm EMG and MEG recorded from contralateral motor cortex, averaged across both sides, during auxotonic force production by healthy controls (HC) vs patients diagnosed with ALS.The physiological beta peak of CMC is significantly diminished in patients with ALS.(C) Corticomuscular communication appraised via temporal correlation of fluctuation in band-limited power (amplitude envelope, calculated using the Hilbert transformation) as opposed to phase-based coherence measurement.This independent measure confirmed significantly disrupted communication of beta signals in ALS patients relative to HCs.Shaded area = standard error of mean.(with permission, from Proudfoot et al 2018).