Studies of the axon initial segment of motor neurons
Tor Halle
Termin 6
Abstract
The axon initial segment has characteristics that make it well suited to be a key regulator of the excitability. In a knock out of the transporter protein VIAAT required for functional Renshaw cell synapses, Enjin and co-‐-‐-‐workers found that the alpha motor neuron excitability was reduced. We therefore wanted to investigate if this reduced excitability is due to alterations in the axon initial segment length, distance to soma or ion channel density.
Background
Motor neurons
The process by which you generate a movement is complex; it involves a wide variety of cells. The cells that
ultimately trigger the movement are the motor neurons, the link between CNS and the muscles. Thereby the definition; a motor neuron is a neuron with the cell body in the central
nervous system that projects its axon to skeletal muscles (1). Motor neurons (MNs) were first described in the late nineteenth century (4). When first described the motor neurons were dived in two subtypes, alpha and gamma, based on which muscle fibres
they innervate (3, 5). Alpha motor
Figure 1 Sketch showing the motor pool principle (Kanning et al 2010) (3)
neurons (α-‐-‐-‐MNs) innervate extrafusal muscle fibres (commonly referred as muscle fibres) that generate force and thereby a movement, this gives α-‐-‐-‐MNs a skeletomotor function. Gamma motor neurons (γ-‐-‐-‐MNs) innervate intrafusal muscle fibres that modulate muscle spindles, the sensory organ of muscle stretch, thus giving γ-‐-‐-‐MNs a fusimotor function (3, 6). In the first half of the twentieth century another motor neuron subtype was discovered. This subtype
Termin 6
innervated both extrafusal and intrafusal muscle fibres, and thereby have the combined function of α-‐-‐-‐MNs and γ-‐-‐-‐MNs (skeletofusimotor function), this subtype was called beta motor neurons (β-‐-‐-‐MNs) (3, 6). All three of the motor neuron subtypes are located along the entire spinal cord, and in separate nuclei in the brain stem, organized in columns that is divided into motor pools based on the muscle the MN innervate (7). Henceforth, this report will focus on α-‐-‐-‐MNs.
The alpha motor neurons, the archetypal motor neuron, can further be divided into three groups (fast fatigable, fatigue resistant and slow) based upon the electrophysiology of the motor neuron (3). α-‐-‐-‐MNs have the largest cell body among the motor neurons (almost twice the size as the other MNs). The cell body of α-‐-‐-‐MNs is also more irregular and has a more branched dendritic tree than the other motor neurons (3, 6). All these differences make it easier to identify α-‐-‐-‐MNs under a microscope. As
mentioned before, in
response to efferent input α-‐-‐-‐MNs stimulate extrafusal skeletal muscle fibres, and thereby generate a muscle contraction. It was long believed that the α-‐-‐-‐MNs
axon only innervated
Figure 2 Motor neuron -‐-‐-‐ Renshaw cell circuitry
muscle fibres, but in the mid-‐-‐-‐twentieth century intraspinal collaterals were discovered on the α-‐-‐-‐MN axon (8). These collaterals were almost exclusively found on α-‐-‐-‐MN and they all innervated the same interneuron type. This
interneuron was named after its discoverer, Birdsey Renshaw, to Renshaw cell (3, 8, 9).
Renshaw cells
Renshaw cells (RS cells) are located in the ventral horn, close to the α-‐-‐-‐MNs, and are derived from engrailed-‐-‐-‐1-‐-‐-‐expressing V1-‐-‐-‐embryonic interneurons (8). The discovery of the RS cell was based on the observation of reduced excitation by the Ia afferents of motor neurons upon antidromic stimulation of the ventral root (8, 10). They subsequently showed that this effect was caused by cholinergic
Termin 6
activation of RS cells by the α-‐-‐-‐MNs. α-‐-‐-‐MNs activate RS cells with the
neurotransmitter acetylcholine (Ach) that interacts with ACh-‐-‐-‐receptors on the surface of the RS cell. The RS cells responds with a high frequency signalling with the inhibitory neurotransmitters glycine and GABA to the motor neurons, and consequently inhibit the MNs. Thus, RS cells are activated each time the α-‐-‐-‐MNs releases ACh to generate a movement, and the RS cells in turn inhibit the α-‐-‐-‐MNs (8-‐-‐-‐10). This describes phenomena called recurrent inhibition, a phenomena that occurs in various parts of the CNS. The recurrent inhibition is more intense over motor pools involved with more stereotyped motor output, i.e. proximal muscles (8). One RS cell receives input from several motor neurons and conversely the RS cells output is distributed to several motor neurons. The characteristics of the inhibitory synapses develop after the first week postnatal and are firmly established after three weeks postnatal (8, 11).
Even though the activity of RS cells is well established, little is known about the function of RS cells. It has been suggested that RS cells have an important role during the development and maturation of spinal circuits, but no reliable data have been obtained (8). RS cell dysfunction has been predicted to cause
alterations in force generation/maintenance, rigidity, spasticity or tremor. These hypotheses have been difficult to test due to lack of experimental tools to
selectively knockout RS cells (8). RS cells are characterized by high expression of calbindin Dk28, a calcium-‐-‐-‐buffering protein. Even though RS cells are not the only spinal cord cells expressing calbindin Dk28, there are very few other neurons in the ventral horn expressing calbindin Dk28 (8). Studies of the ACh-‐-‐-‐
receptors, mediating the α-‐-‐-‐MN cholinergic signals, expressed on RS cell surface revealed that there were two different subtypes expressed, α4 and α2. Whereas the α4 subunit also was expressed by other spinal interneurons it seemed that the α2 subunit was restricted to RS cells (Henrik Gezelius, unpublished data).
Axon Initial Segment Axon potential initiation is the culmination of the synaptic stimulation of neurons. Whether the axon potential origin was
Figure 3 Ion channel expressed in the AIS (Bender &
Trussell 2012) (2)
Termin 6
somatic or axonal was a controversy during the first part of the twentieth century. It was not until 1950 before they received experimental data that yielded a solution to the issue, and these data favored an axonal origin (2). The expermintel data sugsted that axon potential intiating begins somewhere in the region between the axon hillock, the part of the soma that is connected to the mainbody of the axon, and the first segment of myelination. This region was termed the axon initial segment (AIS) (2, 12). Features that influence the spiking capability of the AIS are distance from the soma, segment length and ion channel density & compostion(2, 13-‐-‐-‐15). The AIS region on the axon often starts tens of micrometer away from the soma (14). Although there are a significant variation, both across and within different cell types, of the precise position of the AIS within the axon. This variation may give rise to differences in neuronal excitability (14). Deprivation of presynaptic activity can directly regulate the length and position (both distal and proximal) of the AIS. This change of the AIS length and position can in turn influence the excitability of the neuron, for example cells were the AIS lies more distal in the axon have a higher current threshold for action potential (thus a decreased excitability) (13-‐-‐-‐15). The fact that the myelin sheath starts immediately after the AIS may affect the AISs
capability to increase either its length or distance from the soma (12). In addition to AIS length and location can AIS ion channel density, composition and
gradients influence the neuron excitability (2, 12, 15). The ion channels in the AIS that have greatest influence on spiking properties of the AIS are sodium (Nav) and potassium (Kv) channels, in addition to these ion channels there is also calcium channel present at the AIS (2, 12). The Nav channel density in the AIS is very high, about 50-‐-‐-‐fold higher than density at the soma and proximal dendrites, which lowers the threshold for axon potential initiation, thus increasing the neuron excitability(16, 17). Three subtypes of Nav channels are present at the AIS, Nav1.1, Nav1.2 and Nav1.6. These Nav channels (especially Nav1.6) are anchored and clustered to the AIS by the cytoskeletal protein Ankyrin G (Ank G).
This anchoring protein is specific for the AIS and can therefore be used to specific stain the AIS through immunohistochemistry (14, 17, 18). It has been shown that Ank G-‐-‐-‐ deficient Purkinje cells can’t concentrate Nav channels at the AIS; these cells show reduced action potential firing responses to excitatory stimulation (18). The Kv channels are important for the modulation of action
Termin 6
potential repolarisation, firing rate and pattern (17). Two Kv channel subtypes, Kv7.2/KCNQ2 & Kv7.3/KCNQ3, are thought to be well suited for controlling the excitability of the axon. These two Kv channels are anchored to the AIS by the cytoskeletal protein Ank G, the same protein that anchors Nav channels to the AIS. In addition to these two Kv channels two other Kv subtypes, Kv1.1 and Kv1.2, are expressed at the AIS (2, 12). AIS can be heterogeneous within a single neuronal population, both in terms of ion channel composition, density and localisation, and thus confer different excitability properties (12, 17). This AIS plasticity is much more significant in excitatory neurons, for example motor neurons (13). These AIS characteristics generally apply for the alpha motor neuron, but there are some specific features of the α-‐-‐-‐MN AIS that are described by Duflocq and his colleagues (12). The AIS of α-‐-‐-‐MN can be either soma or dendritic derived, where the dendritic derived AIS logically have a longer
distance to the soma, but in most cases the AIS is soma derived. The length of the AIS in α-‐-‐-‐MN differs between 25 and 35 µm, which can affect the spiking
properties. In spinal alpha motor neurons the major axonal ion channel is the sodium channel subtype Nav1.1. The same also applies in the AIS, where the Nav1.6 subtype also is expressed. However, there is no expression of the sodium channel subtype Nav1.2 in the α-‐-‐-‐MN AIS. When an immunohistochemical
staining of Nav1.1 and Nav1.6 was made it was found that the intensity of the Nav1.1 staining decreased when the intensity of the Nav1.6 staining increased, accordingly there is no expression of Nav1.1 and Nav1.6 simultaniosly. The potasium channel subtyoe KCNQ2/Kv7.2 is expressed uniformly throughout the Ank G positive parts of the AIS in α-‐-‐-‐MN. In addition to the Kv7.2 subtype there is also an expression of Kv1.1, Kv1.2 and Kvβ2 at the α-‐-‐-‐MN AIS.
Knockout of VIAAT in Chrna2cre mice
In the beginning of the twenty-‐-‐-‐first century Anders Enjin and colleagues (1) did a successful conditional knockout of the vesicular inhibitory amino acid
transporter (VIAAT), that is the protein responsible for filling synaptic vesicles with the inhibitory neurotransmitters glycine and GABA . Thereby, removing VIAAT from synapses between Renshaw cells and motor neurons will block RS cells from releasing glycine and GABA at the synapse. To make this conditional knockout specific to the RS population, the Cre/lox system was taken advantage of. A Cre-‐-‐-‐expressing cassette was inserted at the initiating ATG codon of the first
Termin 6
coding exon of the Chrna2 gene, this gene express the α2 subunit which is selectively expressed in cholinergic receptors on RS cell surface, that in the ventral horn of the spinal cord can be used as a specific marker for RS cells (19).
To construct the viaat-‐-‐-‐lox mice line exon 2 of the Slc32a1 gene (expressing Viaat) was flanked with loxP sites. Thereby, by crossing the Chrna2-‐-‐-‐cre mice with Viaat-‐-‐-‐
lox animals a conditional knockout of VIAAT in Chrna2::Cre expressing cells (RS cells) was obtained (20). The cre expressing mice was further crossed with a commercial line carrying the tdTomato protein (constructed using a CAG-‐-‐-‐lox-‐-‐-‐
stop-‐-‐-‐lox-‐-‐-‐tdTomato-‐-‐-‐WPRE-‐-‐-‐pA plasmid inserted into the Gt(Rosa26sor locus)).
Thereby Chrna2-‐-‐-‐cre/tdTomato mice expressed the fluorescent protein tdTomato when the stop codon was removed by Cre, thus in Chrna2::Cre expressing cells (RS cells). This expression of fluorescent protein enabled the identification of RS cell under a fluorescent microscope without having to perform any
immunohistochemistry. As a small fraction of α-‐-‐-‐MNs also express the Chrna2 gene, these mice could also be used readily identify large MNs expressing tdTomato. These rare labellings of MNs could therefore be taken advantage of, when wanting to investigate single, well separated MNs.
The conditional knockout of VIAAT has not been shown to result in alterations in motor coordination, nor any alterations in the cholinergic input to the RS cells.
But there were an increased number of synapses from RS cells to the α-‐-‐-‐MNs, however there were no alterations in the monosynaptic response in the motor neurons. Enjin and colleagues also found that the action potential threshold was more depolarized and the action potential amplitude lower in α-‐-‐-‐MNs in knockout mice. In addition to that the α-‐-‐-‐MNs in the knockout mice had an impaired
capacity to increase spike frequency in response to increased input. This suggests α-‐-‐-‐MNs to be hypoexcitable when developing without RS cell input.
This hypoexcitability is the subject for my study. My hypothesis is that the hypoexcitability may be caused by changes in α-‐-‐-‐MN AIS location, ion channel density or composition.
Termin 6
Methods
Mice
In this study two mice line were used, Viaatlx/lx (a conditional knockout) and Chrna2::Cre (control). For the conditional knockout mice a Cre-‐-‐-‐expressing cassette has been inserted at the initiating ATG codon of the first coding exon of the Chrna2 gene (expressed in the spinal cord in Renshaw cells and motor neurons(8, 19, 21, 22)) and exon 2 of the Slc32a1 gene (expressing Viaat) was flanked with loxP sites. This leads to a knockout of Viaat in Chrna2::Cre
expressing cells(Renshaw cells). In the control mice has only a Cre-‐-‐-‐expressing cassette been inserted, the cassettes position is the same as in the conditional knockout mice (the initiating ATG codon of the first coding exon of the Chrna2 gene) (19). In both mice lines a CAG-‐-‐-‐lox-‐-‐-‐stop-‐-‐-‐lox-‐-‐-‐tdTomato-‐-‐-‐WPRE-‐-‐-‐pA plasmid has been inserted into the Gt(ROSA)26sor locus. To generate the conditional knockout mice the two different mice lines (Chrna2::Cre and Viaatlx/lx) had been mated with each other. The offspring expresses Cre resulting in a knockout of exon 2 of the Slc32a1 and thereby a conditional knockout of Viaat. The expression of Cre also results in a removal of the Stop cassette in the CAG-‐-‐-‐lox-‐-‐-‐
stop-‐-‐-‐lox-‐-‐-‐tdTomato-‐-‐-‐WPRE-‐-‐-‐pA plasmid, and thereby expression of tdTomato. The control mice had been generated by mating Chrna2::Cre mice with each other, thereby creating offspring that express Viaat and Tomato. Cre positive mice are identified by PCR genotyping. Due to the fact that Chrna2 also is expressed in nasal cells of the mouse (21) tdTomato will also be expressed there. This tdTomato expression can be visualized by illuminating the mouse’s nose with laser, and it is with this method tdTomato mice were identified. Other lab
members have done the tdTomato identification previously. The same mice lines were initially used by Anders Enjin in his thesis Neural Control of Movement;
Motor Neuron Subtypes, Proprioception and Recurrent Inhibition, not published yet (1). The VIAATlx/lx and Gt(ROSA)26sor::tdTomato mice lines are described in Tong et al 2008 (20) and Madisen et al 2010 (23).
Dissection
Mice were anaesthetized with isoflurane and decapitated and the spinal cord was dissected out in cold 1x PBS. The spinal cord was placed in 4 %
paraformaldehyde and fixed for 2 hours. Spinal cords samples were
Termin 6
cryoprotected by immersion in 30 % sucrose and PBS over night. Finally the samples were frozen slowly in OCT medium. The frozen samples were stored at -‐-‐-‐
20 degrees.
Cryostat
Spinal cord samples were cut into 20 µm or 35 um slices on a cryostat at a temperature of -‐-‐-‐19°C. The slices were collected on glass slides, 3-‐-‐-‐4 sections on each slide. For each spinal cord sample 30-‐-‐-‐35 glass slides were collected, each was marked with date, genotype, thickness, postnatal stage, and area of spinal cord, angle and slide number. Collected slides were kept dark and dried for an hour and then placed in a cover box. The slides were stored at -‐-‐-‐20 degrees.
Immunohistochemistry
Antibody Binding site Visualized structure
Calbindin D-‐-‐-‐28K: mouse monoclonal
Calcium-‐-‐-‐binding proteins In Spinal cord ventral horn: Renshaw cells
Ankyrin G: Rabbit polyclonal
Anchoring protein (Ankyrin G) for sodium and potassium channels in the AIS
Axon Initial Segment
PanNav: Rabbit Polyclonal
Voltage-‐-‐-‐gated sodium channels
Voltage-‐-‐-‐gated sodium channels
DAPI Adenine and Tyrosine
rich regions in the DNA
Cell Nucleus
Following primary antibody solutions were used:
Solution 1
• Anti Calbindin-‐-‐-‐32 monoclonal mouse (1:1000)
• Anti PanNav polyclonal rabbit (1:500) Solution 2
• Anti Calbindin-‐-‐-‐32 monoclonal mouse (1:1000)
• Anti AnkG polyclonal rabbit (1:500)
Alexa 488 and Alexa 647 conjugated secondary antibodies were used to detect mouse and rabbit antibodies. Dapi was also added as a marker for DNA (cell
Termin 6
nucleus). The same secondary antibody solution was used for both primary antibody solutions.
During the immunohistochemistry the slides were kept in a humidifying chamber. The spinal cord sections on each slide were circled with a hydrophobic barrier marker. Each slide was first incubated in antibody blocking solution for 1 hour (10 ml 0,1M PB, 1 ml 4 % goat serum and 30 µl Triton-‐-‐-‐X-‐-‐-‐100), to minimize the background fluorescence. The blocking solution was removed and the primary antibody solution was applied. The slides were incubated in room temperature overnight (occasionally in a fridge over the weekend). After incubation the slides were washed in PBS for 3x10 min. Then the secondary antibody solution was applied, and the slides were incubated for 1 hour at room temperature. When the incubation was finished the slides were washed as earlier in PBS. Finally the slides were dried and sealed with cover slips and Mowiol. The stained slides were stored in a cover box in a freezer.
Confocal microscopy
Images were obtained with a confocal laser-‐-‐-‐scanning microscope, using the software LSM5. Overview images were taken with a 20x objective and zoomed images were taken with a 63x objective. The overview images were taken with only the genetically expressed fluorescent protein tdTomato visible. Two zoomed images were taken of each motor neuron of interest. One image with only the tdTomato protein visible and one image with tdTomato protein, Calbindin and Ankyrin G/PanNav stainings visible, both images were taken with a Z-‐-‐-‐stack mode. Distance from the soma and the length of the Axon Initial Segment (AIS) where measured in the software LSM Images, that is used to view the images. To quantify the density of Nav channels the intensity of the immunofluorescence intensity of the PanNav staining where measured with the same software.
Termin 6
Result
The aim with this study was to measure the AIS length, distance from soma and sodium channel density in VIAATlx/lx (conditional knockout) and control mice (Chrna2::Cre). First overview pictures were taken with using a 10x objective, these pictures were taken with a regular fluorescence microscope (see figure 4).
The overview pictures were used to identify MNs in the spinal cord samples.
This was done using the tomato staining. Red cells, with irregular form, twice the size of other tomato labelled cells located in the ventral part of the spinal cord were identified as MNs (se figure 4 & 5).
Termin 6
To identify MN axons pictures were taken with a 20x objective in a confocal microscope (see figure 5). Axons can be identified by that they are much thinner than dendrites. In order to distinguish between motor neuron axons and
Renshaw cell axons terminating on the MN soma, the calbindin staining was used. Thereby, thin tomato positive neuritis can be distinguished as either axons (only positive for tomato) or if calbindin positive, labelling RS cells, it shows the tomato stain (RS cell origin) however overlayed with calbindin.
Termin 6
Zoomed pictures on selected motor neurons were taken to make measurements on (see figure 6).
Due to technical difficulties no images with clearly visible MN axons was
obtained that could be used for subsequent measurements.
Discussion
This conditional knockout with the synapses from the Renshaw cells to the alpha motor neurons deleted is used in several other studies in the Kullander lab. The fluorescent protein tdTomato is one of the strongest fluorescent proteins available today, and the selective expression in only a few α-‐-‐-‐MNs in each spinal cord section has intrigued members of this lab. However, the calbindin staining, was shown to not work well (personal communication with PhD student S.
Perry). The stainings of Ankyrin G and sodium ion channels could not be evaluated due to the fact that no motor neuron axons were found in the pictures.
Despite the fact that fluorescent protein tdTomato labels a few α-‐-‐-‐MNs it may be better to use another marker that label more α-‐-‐-‐MNs. Duflocq 2011 (12) used an anti-‐-‐-‐Peripherin antibody in his study. This antibody specifically labels somatic MNs in the ventral horns of the spinal cord. Thus, this antibody can be used as a
Termin 6
complement to Tomato in identifying α-‐-‐-‐MNs. Another factor that may have affected the occurrence of α-‐-‐-‐MN axons is the thickness of the obtained sections.
All sections used in study were 20 µm thick, which reduces the chance to find a motor neuron with a visible axon. This is due to the fact that α-‐-‐-‐MNs have diameter of about 50 µm (5), consequently in 20 µm section only a small part of the motor neuron be included. Thicker sections, about 30 – 40 µm, may increase the chance of finding a α-‐-‐-‐MN with a visible axon, and still allowing for good penetrance of antibodies. All mice used in my study were between 5 and 15 days old, this might be a cause to lack of result. Alvarerz and Fyffe 2007 (8) write in their article that “the characteristics of inhibitory synapses develop after the firs week postnatal and are firmly established by P20”, accordingly the hypothetical changes in the AIS might not occur until the postnatal day 20 when the inhibitory synapses are firmly developed. However, all the results showing hypoexcitability in MNs of Chrna2-‐-‐-‐cre/viaat-‐-‐-‐lox was obtained from mice of age P0-‐-‐-‐P6 (K. Leao, unpublished data). In a study by Pan et al 2006 (18) they noticed that Purkinje cells in mice lacking Ankyrin G expression failed to concentrate sodium ion channels at the AIS, these cells showed reduced action potential firing responses to excitatory stimulation. If the conditional knockout of the Renshaw cells synapses somehow caused an alteration in the Ankyrin G expression in the α-‐-‐-‐MNs, it may give rise to the changes in α-‐-‐-‐MN found by Enjin et al (1). The lack of Ankyrin G expression would also have made the staining for this protein useless.
Even though I did not get any result proving my hypothesis, there are many facts about AIS plasticity in favour of the hypotheses. Long-‐-‐-‐term changes in electrical activity, such as chronic depolarization or chronic stimulation, are known to cause significant changes in the AIS location (14, 15). Consequently, neurons can respond to both increases and decreases of the activity with changes in AIS position. It has been shown that AISs with a more distal position have higher current thresholds for action potential initiation, and thus a lower excitability (14). Consequently, the AIS in the motor neurons in the conditional knockout mice may have a more distal position. Alterations in the sodium channel density and composition at the AIS have been shown to cause reduced action potential firing in response to excitatory stimulation (18). These alterations can be caused
by chronic stimulation or chronic depolarization (15), and thus might be a cause to the reduced motor neuron excitability in the conditional knockout mice.
Bender and Trussell 2012 (2) suggested that the potassium ion channel subtypes Kv7.2 & 7.3 (KCNQ2 and KCNQ3), clustered to the AIS through the anchoring protein Ankyrin G, are well suited for controlling the excitability of the axon. If this two potassium ion channels subtypes in fact controls the excitability of the axon, might alterations in the density or composition of these two subtypes be the cause of the reduced motor neuron excitability in the conditional knockout mice.
It may be one or several of these factors that causes the reduced excitability in the conditional knockout mice. However in order to answer this question, more studies are needed.
References
1. A. Enjin et al., Development of spinal motor circuits in the absence of VIAAT-‐-‐-‐mediated Renshaw cell signaling. (2011-‐-‐-‐03-‐-‐-‐28T17:44:34+0200, Not published yet).
2. K. J. Bender, L. O. Trussell, The Physiology of the Axon Initial Segment.
Annu Rev Neurosci, (Mar 20, 2012).
3. K. C. Kanning, A. Kaplan, C. E. Henderson, Motor neuron diversity in development and disease. Annu Rev Neurosci 33, 409 (2010).
4. M. Manuel, D. Zytnicki, Alpha, beta and gamma motoneurons: functional diversity in the motor system's final pathway. J Integr Neurosci 10, 243 (Sep, 2011).
5. R. N. Bryan, D. L. Trevino, W. D. Willis, Evidence for a common location of alpha and gamma motoneurons. Brain Res 38, 193 (Mar 10, 1972).
6. D. R. Westbury, A comparison of the structures of alpha and gamma-‐-‐-‐
spinal motoneurones of the cat. J Physiol 325, 79 (Apr, 1982).
7. S. McHanwell, T. J. Biscoe, The localization of motoneurons supplying the hindlimb muscles of the mouse. Philos Trans R Soc Lond B Biol Sci 293, 477 (Aug 12, 1981).
8. F. J. Alvarez, R. E. Fyffe, The continuing case for the Renshaw cell. J Physiol 584, 31 (Oct 1, 2007).
9. J. C. Eccles, P. Fatt, K. Koketsu, Cholinergic and inhibitory synapses in a pathway from motor-‐-‐-‐axon collaterals to motoneurones. J Physiol 126, 524 (Dec 10, 1954).
10. B. Lamotte d'Incamps, P. Ascher, Four excitatory postsynaptic ionotropic receptors coactivated at the motoneuron-‐-‐-‐Renshaw cell synapse. J Neurosci 28, 14121 (Dec 24, 2008).
11. U. Windhorst, On the role of recurrent inhibitory feedback in motor control. Prog Neurobiol 49, 517 (Aug, 1996).
12. F. C. Amandine Duflocq, Marco Giovannini, François Couraud, Marc Davenne, Characterization of the axon initial segment (AIS) of motor neurons and identification of a para-‐-‐-‐AIS and a juxtapara-‐-‐-‐AIS, organized by protein 4.1B. BMC Biology 9, 66 (2011, 2012).
13. M. S. Grubb, J. Burrone, Activity-‐-‐-‐dependent relocation of the axon initial segment fine-‐-‐-‐tunes neuronal excitability. Nature 465, 1070 (Jun 24, 2010).
14. M. S. Grubb, J. Burrone, Building and maintaining the axon initial segment.
Curr Opin Neurobiol 20, 481 (Aug, 2010).
15. M. N. Rasband, The axon initial segment and the maintenance of neuronal polarity. Nat Rev Neurosci 11, 552 (Aug, 2010).
16. M. H. Kole et al., Action potential generation requires a high sodium channel density in the axon initial segment. Nat Neurosci 11, 178 (Feb, 2008).
17. A. Lorincz, Z. Nusser, Cell-‐-‐-‐type-‐-‐-‐dependent molecular composition of the axon initial segment. J Neurosci 28, 14329 (Dec 31, 2008).
18. Z. Pan et al., A common ankyrin-‐-‐-‐G-‐-‐-‐based mechanism retains KCNQ and NaV channels at electrically active domains of the axon. J Neurosci 26, 2599 (Mar 8, 2006).
19. H. Gezelius, dissertation; Studies of Spinal Motor Control Networks in Genetically Modified Mouse Models, Uppsala universitet (2009).
20. Q. Tong, C. P. Ye, J. E. Jones, J. K. Elmquist, B. B. Lowell, Synaptic release of GABA by AgRP neurons is required for normal regulation of energy balance. Nat Neurosci 11, 998 (Sep, 2008).
21. K. Ishii, J. K. Wong, K. Sumikawa, Comparison of alpha2 nicotinic acetylcholine receptor subunit mRNA expression in the central nervous system of rats and mice. J Comp Neurol 493, 241 (Dec 12, 2005).
22. M. Dourado, P. B. Sargent, Properties of nicotinic receptors underlying Renshaw cell excitation by alpha-‐-‐-‐motor neurons in neonatal rat spinal cord. J Neurophysiol 87, 3117 (Jun, 2002).
23. L. Madisen et al., A robust and high-‐-‐-‐throughput Cre reporting and characterization system for the whole mouse brain. Nat Neurosci 13, 133 (Jan, 2010).