Keywords

1 Introduction

Motoneurons of different classes and subtypes (fast/slow, alpha/gamma) are grouped together into motor pools each of which innervates a single muscle in vertebrates (Kanning et al. 2010). An adult motoneuron innervates a group of muscular fibers, both forming a motor unit. Most muscles are composed of different motor units with different ratios between slow, fast-resistant and fast fatigable depending on the muscle (Burke et al. 1971; Burke 1981). In neonatal animals, a single muscle fiber is innervated by several motoneurons (polyinnervation) up to the third postnatal week and motor units achieve a complete maturation depending on the muscle a month after birth in rodents (Navarrete and Vrbova 1983, 1993; Balice-Gordon and Thompson 1988). Gap junctions are present during the two first postnatal weeks in rodents facilitating synchronization of motor discharge firing between motoneurons. At adulthood, an orderly recruitment from slow to fast motor units is a general rule for most movements (Henneman 1957; Henneman et al. 1965). Motoneuron discharge firing is modulated and permanently adjusted during the muscle contraction by proprioceptive afferents from muscle spindles (Matthews 1982; Rossi-Durand 2006, for review).

During the neonatal period, the motoneuron is the main target for numerous abnormalities caused by neurological diseases. Early abnormalities have been described in motoneurons from animal models of neurodegenerative diseases during embryonic and postnatal periods in disorders such as spinal muscular atrophy (Mentis et al. 2011; Vrbova 2007), Charcot–Marie–Tooth diseases (Irobi et al. 2010) and amyotrophic lateral sclerosis (Amendola and Durand 2008; Amendola et al. 2004; Bories et al. 2007; Durand et al. 2006; Raoul et al. 2002; Van Zundert et al. 2012; Vinsant et al. 2013). Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease mainly affecting motor neurons and is one of the most complex and dramatic diseases with no available curative therapy (Leigh and Swash 1996; Kanning et al. 2010). It is known that large motoneurons are affected first in humans. In transgenic mouse model of ALS, the motoneuron subtypes are not equally vulnerable (Pun et al. 2006; Hadzipasic et al. 2014). A number of sporadic and familial ALS patients have mutations in the gene coding the superoxide dismutase 1 (SOD1) (Rosen et al. 1993; Boillee et al. 2006; Robberecht and Philips 2013). Most mechanisms leading to the disease have been described in the SOD1 mice, (Gurney et al. 1994; Dal Canto and Gurney 1997; Bruijn et al. 1997, 1998, 2004). Early changes in excitability were described in the motoneurons of SOD1 mice including both increased and decreased excitability (Pieri et al. 2003; Kuo et al. 2004, 2005; Bories et al. 2007; Van Zundert et al. 2008; Pambo-Pambo et al. 2009; Martin et al. 2013; Elbasiouny et al. 2010). Perturbed intracellular trafficking in SOD1 mice was indicated by the reduced transport of selective cargoes described months before disease onset (Williamson and Cleveland 1999), the alterations in fast axonal transport (Kieran et al. 2005) and neurofilament accumulation (Bruijn et al. 2004). A direct link between retrograde axonal transport and mutant SOD1 was made with dynein interaction (Zhang et al. 2007).

We described morphological and electrical pathological signs in motoneurons during the postnatal period of SOD1 mice (Amendola et al. 2004, 2007; Durand et al. 2006; Bories et al. 2007; Amendola and Durand 2008; Filipchuk and Durand 2012; Filipchuk et al. 2008, 2021). The time course of the disease may differ depending on the mutations and the number of copies of mutated proteins expressed in the mutant animals. In our experiments, we have used low expressor lines of SOD1 mice (G85R and G93A-low) that become paralyzed several months after birth (6 to 8 months) (Dal Canto and Gurney 1997; Bruijn et al. 1997, 1998; Amendola et al. 2004; Pambo-Pambo et al. 2009) in contrast to high expressor strains (SOD1 G93A-high) that are paralyzed at 4 months (Gurney et al. 1994; Vinsant et al. 2013). In the high expressor SOD1-G93A strain, neuromuscular junctions start to disconnect between 14 and 30 days after birth depending on the muscles (Vinsant et al. 2013). The advantage of the former models (low expressor lines) is the longer presymptomatic time allowing us to decorticate the earlier pathological mechanisms (Amendola et al. 2004; Bories et al. 2007; Pambo-Pambo et al. 2009). The homozygous SOD1G85R strain of mice was first chosen for several reasons. First, all the SOD1G85R homozygous pups are viable and carries the mutation causing the disease; thus no genetic screening was needed to study sensorimotor reflexes using Fox battery tests (Fox 1965) and spinal networks using electrophysiological recordings in the whole brainstem spinal cord preparation (Amendola et al. 2004). Second, the SOD1G85R strain was found to be a good model because the disease-onset did not depend on the enzyme activity and astrocytes were involved in the pathology (Bruijn et al. 1997, 1998). Homozygous and heterozygous SOD1G85R mice develop clinical symptoms at 6 and 12 months of age, respectively (Bruijn et al. 1997, 1998; Amendola et al. 2004, 2007; Amendola 2008). We studied the electrical and morphological properties of motoneurons in both low expressor models (G85R and G93A-low) with comparable time courses (Amendola et al. 2007; Pambo-Pambo et al. 2009; Filipchuk et al. 2021). We also studied the electrical properties of lumbar motoneurons from the high expressor line of SOD1 G93A mice (G93A-high) in the whole brainstem-spinal cord preparation to compare with most studies performed in slice or in cultured neurons from this transgenic strain (Pieri et al. 2003; Kuo et al. 2004, 2005; Quinlan et al. 2011; Leroy et al. 2014).

2 Altered Sensorimotor Development in SOD1 Mice

In earlier work, we addressed the question of whether the SOD1 mutation has an impact on spinal motor networks in postnatal mutant mice (Amendola et al. 2004). We detected subtle functional alterations in postnatal SOD1G85R mice during sensorimotor development. Six sensorimotor responses appearing from P1 to P6 were measured daily in all male and female mice (Fox 1965; Amendola et al. 2004; Amendola 2008). When the appropriate responses were observed, the pup was given a score of 1 for the corresponding test. Tests were as follows. (i) Righting: the pup was placed on it back; it righted itself within10 s. (ii) Cliff-drop aversion: the pup was placed on the edge of a cliff (a table), the forepaws and the head over the edge; it turned and crawled away from the cliff. (iii) Forepaw and (iv) hind paw grasping: when the inside of one paw was gently stroked with an object, the paw flexed to grasp the object. (v) Forelimb and (vi) Vibrissae placing responses were also tested between P3 and P8 (Amendola et al. 2004). Righting and hind paw grasping responses were significantly delayed for SOD1G85R pups up to P7 (Amendola et al. 2004). Our behavioral testing from WT and SOD1G85R showed a transient motor deficit with a motor impairment of only the hind limbs. Behavioral deficits were also found in postnatal SOD1G93A mice (Van Zundert et al. 2008).

Electrophysiological studies using extracellular recordings from the ventral roots in entire brainstem-spinal cord preparations in vitro during perfusion of NMDA and 5HT, showed a lack of alternating rhythmic activities in SOD1 mice, in contrast to WT preparations recorded in the same chamber. This lack of rhythmic responses suggested early hypoexcitability of lumbar networks probably due to changes in endogenous membrane properties and/or alterations in lumbar network connectivity.

A delayed maturation of supraspinal descending pathways has been suggested (Amendola et al. 2004; Amendola 2008). Recent results show a delay in the development of 5HT immunoreactive descending fibers in SOD1G93A mice (Martin et al. 2020). Further, we recently brought some arguments in support of a lower density of boutons in motoneuronal pools of SOD1 G85R mice (Filipchuk et al. 2021). In transgenic SOD1G93A mice, it was shown that NMA and 5HT could generate fictive locomotor activity just after birth at P0-P3 (Milan et al. 2014). This was not in contrast with our results since they were obtained in younger animals and in a different strain. Further, in previous experiments we could get alternating rhythmic activities in SOD1G85R in in vitro preparations when dopamine was added to the cocktail NMDA, 5HT (Amendola 2008). Gross deficits in locomotion were found in SOD1 juvenile mice (Wooley et al. 2005; Van Zundert et al. 2008; Mead et al. 2011; Vinsant et al. 2013).

3 Electrical Properties of Developing Motoneurons in Postnatal Mice

Electrical properties of developing spinal motoneurons in neonatal mice have been investigated in several studies (Mynlieff and Beam 1992; Bories 2004; Bories et al. 2007; Pambo-Pambo et al. 2009; Pambo-Pambo 2010; Nakanishi and Whelan 2010; Quinlan et al. 2011; Durand et al. 2015).

3.1 Early Changes in Electrical Properties of Lumbar Motoneurons from WT Mice

Here we focus on the different subgroups of motoneurons defined by their pattern of discharge during the postnatal period (Pambo-Pambo et al. 2009; Durand et al. 2015). Using intracellular current pulses, we found three subgroups of postnatal motoneurons according to their discharge firing pattern (Fig. 1).

Fig. 1
3 illustrations of lumbar motoneurons firing patterns with 3 subgroups. A, a negative 52 m V of Transient with 0.4 n A, 0.6 n A, and 1 n A. B, a negative 65 m V of sustained with 0.9 n A, 1 n A, and 1.2 n A. C, a negative 76 m V of delayed onset with 2.2 n A, 2.3 n A, and 2.9 n A.

Three subgroups of lumbar motoneurons defined according to their discharge firing pattern in postnatal mice. (a) The transient firing is present in 32% of motoneurons at postnatal days P3-P5 (Durand et al. 2015). (b) The sustained firing is the most frequent after P7-P8. (c) The delayed onset firing is observed in 20–30% of cases when recorded in the whole brainstem-spinal cord preparation. Intracellular recordings of three lumbar motoneurons are illustrated with voltage deflections in response to current pulse injections of increasing intensities indicated above each black bar (nA). Membrane potentials are shown below first upper traces for each motoneuron. Same calibration bars for all recordings in the lower right panel

Full size image

The delayed-onset firing motoneurons have the highest rheobase and, consequently, are the least excitable motoneurons. We suggested they might represent the future fast motoneurons in adult animals (Durand et al. 2015). A transient firing pattern with a single or a short train of spike (Fig. 1a) was the most preponderant between P0 and P3 as already described in neonatal rat and these motoneurons are considered as the least mature (Vinay et al. 2000a, b, 2002; Mentis et al. 2007). The sustained firing pattern (Fig. 1b) was the most frequent in older animals and was expressed by 100% of the motoneurons at P9 when recorded in an entire brainstem-spinal cord preparation (Durand et al. 2015). The distribution of the different patterns depends on the postnatal age of the mouse (Durand et al. 2015). The delayed-onset firing pattern (Fig. 1c) was found in 20–30% of lumbar motoneurons in mice aged between P3 and P10. However, delayed-onset firing pattern (also called late spiking) was recorded in 65–70% motoneurons in spinal cord slices (Pambo-Pambo et al. 2009; Leroy et al. 2014). We suggested that the absence of neuromodulatory controls by supraspinal pathways in slice preparations was responsible for this different ratio (Durand et al. 2015). Several electrophysiological properties indicate that the three different patterns of discharge firing induced by intracellular current pulses injection in motoneurons are reflecting different states of maturation during this postnatal period (Durand et al. 2015). We also described five types of neonatal motoneurons using current ramp injection named transient, linear, linear with sustained firing, clockwise and counter-clockwise hysteresis (Amendola et al. 2007; Durand et al. 2015). Except for the transient groups, we found no correlation between the subgroups defined by their discharge pattern induced by current pulses and ramps (Pambo-Pambo et al. 2009; Durand et al. 2015).

3.2 Early Alterations in Electrical Properties of Lumbar Motoneurons from SOD1 Mice

We found differences in electrical properties of WT and SOD1G85R motoneurons as soon as the beginning of the second postnatal week (Bories 2004; Durand et al. 2006; Bories et al. 2007; Pambo-Pambo et al. 2009; Elbasiouny et al. 2010). We measured a lower gain indicated by a lower slope of the frequency (F)– intensity (I) relationship in SOD1G85R motoneurons, a lower input resistance and a higher rheobase indicating hypoexcitability of lumbar motoneurons (Bories et al. 2007). The slope of the frequency-intensity curve (F-I) measured during the steady state of the discharge firing induced by intracellular injection of current pulses is the gain of a neuron. It has been measured in lumbar motoneurons (n = 32) and neighbor neurons (n = 20) in WT and SOD1 mice (Fig. 2). Such changes in excitability were not present in P3-P6 motoneurons (but only in P7-P10) nor in spinal neurons, suggesting specific alterations of the motoneurons at that particular postnatal period (Bories et al. 2007; Amendola 2008).

Fig. 2
3 illustrations. A, three recordings of 0.65 n A, 0.75 n A, 1.1 n A. B, a graph depicts two scatter plots of W T and S O D 1, G 85 R, in an increasing trend. C, a histogram depicts highest value of W T in motoneurons, and highest value of S O D 1 in spinal neurons.

Lower gain in SOD1 motoneurons. The frequency-intensity curve (F-I) measured in motoneurons and spinal neurons during the steady state of discharge firing induced by intracellular injection of current pulses with increasing intensities. In (a) the upper trace shows the extracellular recording from the L5 ventral root of the orthodromic spike evoked in the intracellularly recorded motoneuron (lower trace) confirming the identification of the motoneuron. In the lower traces, the voltage deflections of a WT motoneuron intracellularly recorded at P7 in response to injected pulses at three different current intensities (Em = −65 mV) illustrate the increased discharge firing frequency of action potentials. (b) Steady-state F-I curves of two motoneurons distinguished by different colors (WT in blue and SOD1-G85R in red). (c) Bar plots of the slopes of F-I relationships of the motoneurons (n = 32) and spinal neurons (n = 20). The slope of the F-I relationship of the SOD1 motoneurons was less than that of the WT motoneurons (P = 0.02; permutation exact test with general scores) whereas no difference was observed between the spinal neurons. (Adapted from Bories et al. 2007)

Full size image

Recently we confirmed hypoexcitability of lumbar motoneurons from SOD1G93A low mice (Filipchuk et al. 2021). Further, alterations in the electrical properties including a lower gain and higher voltage threshold (signs of hypoexcitability) were specific to the delayed-onset firing SOD1 postnatal motoneurons. We concluded that the delayed-onset firing motoneurons are the first functionally affected motoneurons in the second postnatal week in the mouse model of ALS (Filipchuk et al. 2021).

4 Morphological Properties of Developing Motoneurons in Postnatal Mice

Because the spatial pattern of the dendritic arborizations influences neuronal functions (Korogod and Tyc-Dumont 2009), it is important to determine how dendrites acquire their characteristic size and morphology during development, when neuronal branching patterns expand in space by increasing the number and length of segments. In cats and rodents dendritic maturation is completed at different times, it takes the first 6 postnatal weeks in the kitten (Krishnan 1980; Ulfhake et al. 1988; Ulfhake and Cullheim 1988) and the first 3 weeks after birth in the rat (Altman and Sudarshan 1975; Navarrete and Vrbova 1983, 1993). The complexity of dendritic arborizations depend on many intrinsic and extrinsic factors (Libersat and Duch 2004; Parrish et al. 2007).

4.1 Early Changes in Morphological Properties of Lumbar Motoneurons in WT Mice

In contrast to cat and rat motoneurons in which the branching structure (topology) of motoneuron dendrites exhibits some simplification during postnatal development (Cameron et al. 1991; Kalb 1994; Nunez-Abades et al. 1994; Núñez-Abades and Cameron 1995), it seems that motoneuronal complexity in the mouse (number of branches per neuron) does not change during the two first postnatal weeks (Li et al. 2005; Filipchuk and Durand 2012). Well branched dendritic trees of spinal motoneurons were found as early as P3 (average 75 ± 20 branching points per neuron) in postnatal mouse (Filipchuk 2011; Filipchuk and Durand 2012). In embryonic lumbar motoneurons, dendritic arborizations are already well developed in WT mice (Martin et al. 2013) although the total dendritic length and the number of branching points at E17.5 are much lower than that measured at P3-P4 (Martin et al. 2013; Filipchuk and Durand 2012). The number of branching points per motoneuron ranges between 34 and 100 and remains stable on average during the first two postnatal weeks in WT mice (Amendola and Durand 2008; Filipchuk and Durand 2012). The number of dendrites per motoneuron, number of terminals per motoneuron and the other parameters related to tree topology remain relatively constant (Amendola and Durand 2008; Li et al. 2005; Filipchuk and Durand 2012; Filipchuk et al. 2021). Lumbar motoneurons develop in mice by elongation of their terminal segments between P3 and P8 (Filipchuk 2011; Filipchuk and Durand 2012).

The complete morphologies of single WT motoneurons have been described in detail in Amendola and Durand (2008), Filipchuk (2011) and Filipchuk and Durand, (2012).

4.2 Early Changes in Morphological Properties of Lumbar Motoneurons in SOD1 Mice

We compared the detailed morphology of lumbar motoneurons from WT and SOD1 mice at the age P3-P4 and correlated it with the geometry of WT and SOD1 motoneurons at P8-P9. By measuring the length of intermediate and terminal dendritic segments, we determined how dendrites grow and we detected the first morphological anomalies in postnatal SOD1 motoneurons (Filipchuk and Durand 2012). Before the formation of supernumerary branches, SOD1 motoneurons exhibit longer terminal segments compared to WT motoneurons at P3-P4 (Fig. 3).

Fig. 3
2 illustrations of elongated distal dendrites in S O D 1 motoneurons. A, titled as W T P 4, M n 3 with two images. B, titled as S O D 1 P 4, M n 13 with two images.

Elongated distal dendrites in SOD1 motoneurons. Early morphological abnormalities in SOD1 motoneurons at P3-P4 concern the terminal dendrites. Dendrograms illustrate high percentage of short terminals (in red) in the WT motoneurons. Intermediate segments are shown in black; short terminal segments (<35 pm) are highlighted in red and longer ones (>35 pm) in blue. (Adapted from Filipchuk and Durand 2012)

Full size image

Excessive dendritic overbranching was first measured in SOD1G85R lumbar motoneurons at postnatal days P8-P9 (Allene 2006; Amendola et al. 2007; Amendola and Durand 2008; Filipchuk and Durand 2012) and then confirmed in the SOD1-G93A-low expressor line (Filipchuk et al. 2010, 2021). Abnormal proliferation of dendritic arborizations of SOD1 lumbar motoneurons was reported at the beginning of the second postnatal week in SOD1G85R mice, concomitant with changes in input resistance, rheobase current and membrane capacitance measured in motoneurons recorded using the brainstem/spinal cord preparation (Bories et al. 2007).

The number of branches was twofold higher than that in the WT motoneurons at postnatal days P8 to P9. However, the rostrocaudal extension was found to be similar between WT and SOD1 motoneurons as well as the transverse extension. In the examples of Fig. 4, digitized full reconstructions showing the complete dendritic trees of two WT and two SOD1 motoneurons are illustrated in the transverse view. The motoneurons with the lowest number of branching points (upper two) and with the highest number of branching points (lower two) were compared. In both cases, the SOD1 motoneurons have a higher number of branching points (see Amendola and Durand 2008. Filipchuk et al. 2021).

Fig. 4
An illustration depicts dendritic trees of two W T and two S O D 1 motoneurons. Four parts presented are: dorsal, ventral, lateral, medial.

Examples of full 3D reconstructions showing the dendritic trees of two WT and two SOD1 motoneurons (P8-P9) in transverse view. SOD1 motoneurons have a higher number of branching points in all cases. Dendritic overbranching in SOD1 motoneurons has been found originally in SOD1-G85R at postnatal days P8-P9. (Adapted from Amendola and Durand 2008)

Full size image

The shape of dendritic arborizations depends on many intrinsic and extrinsic factors during development (Libersat and Duch 2004). The location of cell bodies in the ventral horn of the spinal cord also influences the orientation of dendrites. The dendrites projected mainly in 3 directions (dorsal, dorsolateral and medial) when the soma was in close proximity to the ventral horn boundary (Fig. 5a, c, e). Dendrites extended also ventrally when the soma was located centrally (Fig. 5b, d, f).

Fig. 5
6 illustrations depict intracellularly stained lumbar motoneurons through various digital images.

Examples of intracellularly stained lumbar motoneurons from SOD1 mice at P8-P9: (a, b) The shape of the dendritic arborizations depends on the location of cell bodies in the ventral horn of the spinal cord. Digital images of lumbar motoneurons in sections (75 pm) containing soma and proximal dendrites. Note the high number of primary dendrites and branching points, the abnormal and opposite directions taken by some proximal dendrites direction in other branches. (c, d) Digital images of the sections n−1 or n+1 adjacent to the section containing the soma and showing the dendritic extension of SOD1 G93A motoneurons in both cases

Full size image

In many instances, multiple staining was observed following a single intracellular injection of neurobiotin (see also Amendola and Durand 2008). This could be due to gap junctions precluding the full reconstructions of motoneurons (Amendola et al. 2007; Amendola and Durand 2008; Filipchuk et al. 2021). In the example illustrated in Fig. 6, a second stained soma is visible in the background of the image.

Fig. 6
4 images of dendrites. First one depicts a 3 D, digitized image. Rest of the three are, of a closeup view, of the part of a dendrite.

Digitized full 3D reconstruction of a P8 lumbar SOD1 G93A-low motoneuron (from Filipchuk 2011). Upper right: Example of multiple staining following a single intracellular staining during the postnatal period; a second somatic region is visible in the background. Lower images: Small spines are detected in distal dendrites. Notice their very small size (<0.7 pm). lower right: an example of trifurcation in dendrite of a SOD1 motoneuron. Rare spines are present on thicker dendrite

Full size image

During the course of these experiments, we could observe on rare occasions trifurcations in dendrites of SOD1 motoneurons (Fig. 6d). At higher magnification only a few and very small spines (<0.5 μm) are detected 100 μm from the soma (Fig. 6c) and in distal dendrites, C). This is in contrast with golgi-cox staining where many big spines are observed in motoneuron dendrites at that postnatal age (Kanjhan et al. 2016).

5 Discussion

Several decades ago, Swash and Schwartz (1995) wrote about ALS “The onset of the disease is difficult to delineate since the clinical manifestations develop gradually; retrospective analysis may suggest that the disease began several years before presentation with muscular weakness or wasting), perhaps with a long preclinical phase during which the disease, although active is not symptomatic”. This long pre-symptomatic period has now been confirmed in the transgenic SOD1 mice, models of ALS. We recently suggested that structural alterations perturbing intracellular trafficking might provoke an early dysfunction such as hypoexcitability of motoneurons (Durand et al. 2021). We will discuss here the possible causes of pathological alterations of dendritic trees in SOD1 motoneurons, the possible link between toxic misfolded SOD1 protein (Saxena et al. 2013) and the specific hypoexcitabilty in the delayed onset firing motoneurons.

5.1 Possible Mechanisms Involved in Pathological Alterations of Dendritic Tree

In SOD1G85R mutant mice, we showed that dendrites of SOD1 motoneurons had significantly longer terminal segments than the WT motoneurons at P3-P4 (Filipchuk and Durand 2012). This excessive elongation was then followed by pathological ramification of most individual dendrites between P4 and P8. During this short time, WT motoneurons grew only by elongation of their terminal segments, the number of branches being stable. In both WT and SOD1G85R motoneurons, the average length of intermediate segments remained invariant. An algorithm of SOD1G85R dendritic growth between P4-P8 was built, based on the detailed morphometric analysis, to outline the dynamics of abnormal dendritic development (Filipchuk and Durand 2012). The dendritic elongation could be a sign of acceleration of dendritic growth at P3-P4 which is consistent with reported results demonstrating accelerated maturation of electrical properties in SOD1 motoneurons (Quinlan et al. 2011; Van Zundert et al. 2008). On the other hand, excessively elongated terminals segments might indicate some extrinsic mechanisms influencing dendritic morphology. Immunofluorescent labeling of synaptic terminals revealed significantly lower densities of synaptophysin in SOD1G85R motoneuronal pools at postnatal day P3 compared with matched WT pools (Filipchuk et al. 2010, 2021). This indicated that SOD1G85R motoneurons may have lower synaptic input than WT motoneurons at that early age P3-P4. It’s known that dendrites extend into rich synaptogenic regions in normal conditions (Vaughn et al. 1988; Vaughn 1989) and that dendritic density is proportional to the number of synapses which supports the idea that formation of a synaptic contact could stabilize particular filopodia (Vaughn et al. 1988). In the absence of synaptic activity, dendrites elongate until they encounter active presynaptic elements.

In cultured Purkinje cells, inhibition of electrical activity by tetrodotoxin (TTX) decreasing intracellular calcium level stops dendritic branching and provokes dendritic elongation (Schilling et al. 1991). It was hypothesized that for branching to occur, not only are synaptic contacts necessary but they must be functional, both in terms of neurotransmission and the generation of postsynaptic activity. It might be suggested that at P3-P4, dendrites elongate “looking for functional presynaptic partners” and likely between P4 and P8 they manage to establish contacts which trigger branching. A study on adult Drosophila motoneurons demonstrated that decreased excitability leads to increased dendritic branch elongation, whereas increased excitability causes branching (Duch et al. 2008). In chicken motoneurons, it was shown that blocking of Ca+ permeable AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) receptors at different embryonic ages has different effect on dendritic morphology (elongation or branching) (Ni and Martin-Caraballo 2010).

Dendrite elongation and branching may involve the same mechanisms based on the microtubule-associated proteins (MAP2) phosphorylation (Hely et al. 2001). Perturbation of microtubule dynamism and impairment of axonal transport occur in ALS (Fanara et al. 2007; Farah et al. 2003; Kieran et al. 2005; Williamson and Cleveland 1999). Alterations in the integrity of microtubules have been observed very early in SOD1 mice with decreased levels of microtubule-associated proteins (Farah et al. 2003) as in human ALS patients (Kikuchi et al. 1999). Dendritic branching depends on microtubule assembly and disassembly under the control of the microtubule-associated proteins MAP2 phosphorylation (Audesirk et al. 1997; Friedrich and Aszodi 1991; Hely et al. 2001). MAP regulates many factors of microtubule dynamics, including polymerization (depolymerization), bundling, spacing, and interaction with actin filaments (Maccioni and Cambiazo 1995). MAP2 was found predominantly in dendrites. The phosphorylation state of MAP2 modulates its interaction with microtubules (Audesirk et al. 1997). Increased phosphorylation of MAP2 protein stimulates neurite branching. Dephosphorylated MAP2 favors elongation by promoting microtubule polymerization. It may play a key role in switching neurite development between elongation and branching. MAP2 phosphorylation depends on the intracellular calcium concentration (Maccioni and Cambiazo 1995).

5.2 Synaptic Activity and NMDA

The timing of afferent innervation and synapse formation coincides with the period of maximum growth and dendritic remodeling (Cline and Haas 2008; Curfs et al. 1993). In the spinal cord the first postnatal weeks are an important period for the establishment of the connections of descending pathways (Vinay et al. 2000b). In the activity-dependent regulation of dendrite development, NMDA-type glutamate receptors play a central role (Kalb 1994; Lee et al. 2005). Their activation during development promotes motoneurons dendrite growth. Incorporation of NR2B dramatically increases the number of secondary and tertiary dendritic branches of ventral spinal cord neurons (Sepulveda et al. 2010). The reduction of dendritic branching is related to the reduction of NR2A subunit in the cortex of SOD1 mice (Spalloni et al. 2011). We have already some preliminary indications that NR2A subunit expression is changed in the developing SOD1 mouse lumbar spinal cord (unpublished data).

5.3 Maturation of Motoneurons

In the newborn kitten, motoneurons innervating the short plantar muscles of the foot show less mature characteristics than those innervating triceps surae muscles (Ulfhake and Cullheim 1988) suggesting that dendritic maturation may be completed at different times in different motoneuron pools. It may correspond to different maturation of electrical properties of functionally distinct motoneurons (Russier et al. 2003; Cotel et al. 2009; Vinay et al. 2000a). Thus changes in excitability and distribution of firing patterns depend on the motoneuron populations under study during the maturation period. Postnatal maturation of dendrites of extensor motoneurons within cervical in rat spinal cord is delayed compared to flexor motoneurons (Curfs et al. 1993) which can be related to the fact that flexor muscles are innervated at an earlier stage than extensor ones. The development of repetitive firing of ankle extensor motoneurons is also delayed compared to the flexor ones (Vinay et al. 2000a). Sustained repetitive and transient firings were observed for neonatal rat brain stem and spinal motoneurons (Gao and Ziskind-Conhaim 1998; Viana et al. 1995; Vinay et al. 2000a, 2002). Transient features are presumably related to different maturation states of underlying slow Na+ and Ca2+currents in lumbar motoneurons.

Signs of denervation appear very early in certain muscles of the SOD1 G93A high expressor (Vinsant et al. 2013; Hadzipasic et al. 2014). We hypothesized that the early electrical alterations in motoneurons are pathological signs in SOD1 mice as they might be at the origin of the peripheral disconnection at the neuromuscular junction (Durand et al. 2021). Hypoexcitability was found in the delayed-onset firing motoneurons in postnatal SOD1 mice (Filipchuk et al. 2021). It was also described in adult transgenic SOD1 mice (Martínez-Silva et al. 2018), in iPSC derived from human patients (Sances et al. 2016) and also in ALS patients (Marchand-Pauvert et al. 2019). A direct consequence of early motoneuronal hypoexcitability may be weakening of motor units.

6 Conclusion

We studied in detail the earliest modifications in dendrites, their morphological evolution during a crucial period of maturation and attempted to establish a structure-functional correlation in the context of synaptic input integration in SOD1G85R and SOD1G93A-low mouse models of ALS. We have shown that excessive elongation of terminal segments represents the earliest signs of abnormal dendritic development in SOD1 motoneurons and is followed by intensive overbranching, starting between P4 and P8. The possible scenario of dendritic growth during this period has been elaborated (Filipchuk and Durand 2012). We have found that dendritic overbranching is a common feature in both SOD1G85R and SOD1G93A model mice at the beginning of the second postnatal week (Filipchuk et al. 2021).

We suggested that the sustained firing motoneurons are more resistant to pathological mechanisms involving the toxic accumulation of mutated proteins (Saxena et al. 2013).

Our data support the fact that pathological changes begin significantly earlier than overt clinical symptoms and suggest that treatment should start much earlier to be effective.

It seems that the denervation of muscles precipitates the clinical symptoms, but denervation starts very much earlier that the obvious motor deficits. Our data support the central origin of the denervation following motoneuronal hypoexcitability in ALS.

The transgenic SOD1 animals compensate for a long period during which clinical symptoms are not detected even when a number of motoneurons have been lost. Such compensatory mechanisms could explain a long asymptomatic period in ALS patients (Leigh and Swash 1996).