Cat Motoneurons Research Articles

Triceps surae motoneuron morphology in the rat: A quantitative light microscopic study

The rat is now the model of choice for many studies of motor function. However, little quantitative information on the structure of rat motoneurons is available. In conjunction with efforts to define the physiologic and anatomic substrates of operantly conditioned plasticity in the spinal cord, 13 physiologically identified triceps surae motoneurons in the rat lumbar spinal cord were labeled intracellularly with horseradish peroxidase and completely reconstructed and measured with a computer-based neuron-tracing system. Somata were all located in the ventral horn of lumbar segments 4-5, had an average diameter of 35 microns, and had 6-12 dendrites. Dendrites ramified throughout the ventral horn and also penetrated the white matter. Their spread was greater in the rostrocaudal and dorsoventral directions (1.53 +/- 0.24 mm and 1.35 +/- 0.23 mm, respectively) than in the mediolateral direction (0.85 +/- 0.14 mm). Regardless of soma location, dendritic fields usually extended throughout the ipsilateral coronal cross-section of the ventral horn. As a result, the ventral or lateral extent of the field was correlated strongly with the soma's distance from the ventral or lateral border, respectively, of the ventral horn. Furthermore, although soma locations in the coronal plane varied widely, the centers of the dendritic fields tended to cluster near the center of the ventral horn. Dendrites constituted 96.2-98.4% (mean +/- SD = 97.3 +/- 0.7%) of the total neuronal surface area. Each of the 104 dendrites studied had an average of 13 branch points and 27 segments. First-order segment diameters ranged from 1.4 to 11.7 microns (mean +/- SD = 5.3 +/- 2.1 microns). Total dendritic length, surface area, volume, number of dendritic segments, and maximum segment order were correlated strongly with diameter of the first-order segment. Proceeding distally between branch points, the mean decrease in dendritic diameter (i.e., tapering) +/- the standard deviation was 22 +/- 8% of the proximal diameter. The average ratio +/- the standard deviation of the sum of the average diameters of each daughter segment raised to the 1.5 power to the average diameter of the parent segment raised to the 1.5 power (i.e., Rall's ratio; Rall, 1959) was 0.87 +/- 0.08. In comparison with cat alpha-motoneurons, rat motoneurons had smaller soma diameters, fewer dendrites, smaller total surface areas, and shorter total dendritic lengths. However, the number of terminations per dendrite was similar in the two species, so that rat motoneurons had more terminations per unit dendritic length.(ABSTRACT TRUNCATED AT 400 WORDS)

Connections between utricular nerve and neck flexor motoneurons of decerebrate cats.

We studied the circuitry between the utricular (UT) nerve and ventral neck motoneurons innervating the longus capitis (LC), a neck flexor muscle, in decerebrate cats. We recorded intracellularly from 63 LC (ipsilateral 37, contralateral 26) motoneurons in C1 and C2 segments. UT nerve stimulation evoked disynaptic, excitatory postsynaptic potentials in all ipsilateral LC motoneurons, and inhibitory postsynaptic potentials that were at least trisynaptic in almost all contralateral LC motoneurons. UT effects on neck motoneurons innervating muscles involved in flexion and lateral turning are similar to the connections between the UT nerve and neck extensor motoneurons. These neuron circuits may play a role in fixing the head and the neck to the body during horizontal linear acceleration.

Discharge properties of medullary reticulospinal neurons during postural changes induced by intrapontine injections of carbachol, atropine and serotonin, and their functional linkages to hindlimb motoneurons in cats.

The present study was aimed at elucidating the pontomedullary and spinal cord mechanisms of postural atonia induced by microinjection of carbachol and restored by microinjections of serotonin or atropine sulfate into the nucleus reticularis pontis oralis (NRPo). Medullary reticulospinal neurons (n = 132) antidromically activated by stimulating the L1 spinal cord segment were recorded extracellularly. Seventy-eight of them were orthodromically activated with mono- or disynaptic latencies by stimulating the NRPo area at the site where carbachol injections effectively induced postural atonia. Most of these reticulospinal neurons (71 of 78) were located in the nucleus reticularis gigantocellularis (NRGc). Following carbachol injection into the NRPo, discharge rates of the NRGc reticulospinal neurons (29 of 34) increased, while the activity of soleus muscles decreased bilaterally. Serotonin or atropine injections into the same NRPo area resulted in a decrease in the discharge rates of the reticulospinal neurons with a concomitant increase in the levels of hindlimb muscle tone. Membrane potentials of hindlimb extensor and flexor alpha motoneurons (MNs) were hyperpolarized and depolarized by carbachol and serotonin or atropine injections, respectively. In all pairs of reticulospinal neurons and MNs (n = 11), there was a high correlation between the increase in the discharge rates and the degree of membrane hyperpolarization of the MNs. Spike-triggered averaging during carbachol-induced atonia revealed that inhibitory postsynaptic potentials (IPSPs) were evoked in 15 MNs by the discharges of nine reticulospinal neurons. Four of them evoked IPSPs in more than one MN. The mean segmental delay and the mean time to the peak of IPSPs were 1.6 ms and 2.0 ms, respectively. Axonal trajectories of reticulospinal neurons (n = 6), which evoked IPSPs in MNs, were investigated in the lumbosacral segments (L1-S1) by antidromic threshold mapping. The stem axons descended through the ventral (n = 2) and ventrolateral (n = 4) funiculi in the lumbar segments. All axons projected their collaterals to the intermediate region (laminae V, VI) and ventromedial part (laminae VII, VIII) of the gray matter. All these results suggest that the reticulospinal pathway originating from the NRGc is involved in postural atonia induced by pontine microinjection of carbachol, and that the pathway is inactivated during the postural restoration induced by subsequent injections of serotonin or atropine. It is further suggested that the pontine inhibitory effect is mediated via segmental inhibitory interneurons projecting to MNs.

Succular effects on neck motoneurons in the decerebrate cats

Succular effects on neck motoneurons in the decerebrate cats

Crossed and uncrossed segmental synaptic effects on rectus abdominis motoneurons from cutaneous and cutaneous-muscle nerve in spinal cats and alpha-chloralose-anesthetized cats.

Segmental synaptic responses evoked by stimulation of the dorsal cutaneous and cutaneous-muscle (CD) nerve or the ventral cutaneous and cutaneous-muscle (CV) nerve were investigated with intracellular recording from rectus abdominis (RA) motoneurons in 49 spinal cats and 28 alpha-chloralose-anesthetized cats. No monosynaptic and few disynaptic connections existed in the circuit from the ipsilateral or contralateral CD or CV nerves to RA motoneurons in spinal cats. Polysynaptic EPSP responses equal to or over trisynaptic connections were dominant. In alpha-chloralose-anesthetized cats, RA motoneurons exhibited significantly more polysynaptic EPSP responses at lower stimulus intensities (1.5T and 2T) of ipsilateral CV nerve stimulation than at those of contralateral CV nerve stimulation. The result that there was no monosynaptic or disynaptic PSP in RA motoneurons produced by low-intensity stimulation of the ipsilateral CD or CV nerve containing the cutaneous-muscle nerves indicated that Ia afferents from the cutaneous-muscle did not connect mono- or di-synaptically to the RA motoneuron. RA motoneurons exhibited more polysynaptic IPSP responses in alpha-chloralose-anesthetized cats than in spinal cats. Segmental pathways from the CD or CV nerves to RA motoneurons might receive inhibitory descending inputs from the upper spinal cord or the brain, in contrast to those from the same nerves to obliquus externus abdominis motoneurons previously reported.

Calcitonin Gene-Related Peptide Increases Following Axotomy of Trochlear Motoneurons

Neuropeptides have long been considered to act as neurotransmitters or neuromodulators, but they may also contribute to a variety of regulatory and trophic neuronal functions. In the present study, we determined the effects of axotomy on the levels of the neuropeptide calcitonin gene-related peptide (CGRP) in trochlear motoneurons (TMNs) of adult cats. The number of neurons with detectable CGRP immunoreactivity, and the intensity of their CGRP immunoreactivity, increased dramatically 1 week postaxotomy, gradually returned toward normal levels, but remained significantly higher than normal 12 weeks postaxotomy—a time when axonal regeneration and functional reinnervation of the superior oblique muscle should be complete. Our observation that CGRP levels in TMNs increase after axotomy suggests a role for this peptide in the response of motoneurons to injury and in regeneration. In addition, since many axotomized TMNs die, we suggest that the maintenance of high CGRP levels even after regeneration is complete may reflect an increased load on those TMNs that survive.

Regrowth of motor axons following spinal cord lesions: Distribution of laminin and collagen in the CNS scar tissue

In previous studies we have demonstrated that spinal motoneurons in the adult cat can regenerate CNS-type axons through CNS scar tissue into denervated ventral roots. This scar tissue, which appears to support and sustain the growth of injured CNS axons, has been shown to have a persistent defect in the blood-brain barrier (BBB). In the present study, the binding of antibodies to nerve growth factor receptor (NGFr), laminin, collagen and a microtubule associated protein (MAP5) was assessed with indirect immunohis-tochemical methods 4 days–20 weeks after a lesion in the ventral funiculus of the spinal cord. An increase in content of collagen-, laminin- and NGFr-like immunoreactivity was observed in the scar tissue during the first 3 weeks. Although type I collagen dominated in superficial areas of the scar, type IV collagen and laminin-like immunoreactivity was observed in expanded perivascular spaces all over the lesion zone. Type IV collagen- and laminin-immunoreactive structures sometimes appeared to form strands which interconnected the ventral horn and the ventral root. Regenerating axons, as revealed by staining with MAP5 or NGFr antibodies, were observed in close association to these paths. It has been suggested that a breakdown of the BBB may play a vital role in certain types of CNS regeneration by increasing the access of blood-borne trophic factors to the lesion area. The demonstration of extracellular matrix proteins like laminin provides further evidence for the notion that the observed regenerative growth takes place in an environment that is markedly different from the normal CNS.

After-potentials and control of repetitive firing in human motoneurones

Characteristics of motoneurone after-potentials in man were derived from the recovery curve of motoneurone excitability after a single discharge evoked by threshold stimulation of Ia afferents or by gentle voluntary muscle contraction. The motoneurone excitability was estimated by the firing index of a single motor unit whose potentials were recorded by needle electrodes. The soleus (a slow muscle) and the flexor carpi ulnaris (a fast muscle) were investigated. The duration of motoneurone after-hyperpolarization of the soleus evaluated by this method ranged between 145 and 255 msec; for the flexor carpi ulnaris it was 55–150 msec. In some motoneurones of the fast muscle, an early short-lasting recovery of excitability (within 5–20 msec after a discharge) was revealed. It was accounted for by delayed depolarization of the motoneurone. The relationship between after-potentials and the characteristics of repetitive firing of motoneurones activated by weak voluntary muscle contraction was analysed. It was observed that the motoneurones with early excitability recovery were capable of firing double discharges with a 5–15 msec interspike interval. It was found also that the minimal firing rate of motoneurones (up to 3.1–5.2 imp/sec in the soleus and 3.8–9.0 imp/sec in the flexor carpi ulnaris) was not correlated with the after-hyperpolarization duration. This differs from the results obtained for cat's motoneurones under intracellular stimulation. The findings suggest that after-hyperpolarization is not the only leading mechanism controlling the low firing rate of motoneurones under conditions of their natural activity in man.

Passive electrophysiological properties of aged and axotomized cat spinal cord motoneurons: The effect of cell size and electrode shunt

Intracellular recordings were obtained from intact and axotomized lumbar motoneurons of aged cats. The sub-threshold electrical properties of these cells were measured, including input resistance, resting membrane potential, and the first two equalizing time constants as well as their associated amplitude constants. These data were used in a semi-infinite cable model of the motoneuron to estimate the size of the shunt resistance (Rshunt) which is created when the electrode penetrates the cell membrane. The average Rshunt for intact aged cells was 5.35 ± 1.01 MΩ, while that for the axotomized aged cells was 8.93 ± 1.20 MΩ. The statistically significant difference in mean shunt magnitude did not affect the measurements of membrane time constant because this constant is independent of the shunt in this model of the motoneuron. However, the determination of cell input resistance, which is not independent of the shunt, was shown to underestimate the real cell input resistance by 23–29%. We therefore conclude that the shunt resistance is an important factor which should be taken into account when measuring input resistance.

Connections between utricular nerve and dorsal neck motoneurons of the decerebrate cat.

1. We studied connections between the utricular (UT) nerve and dorsal neck motoneurons in decerebrate cats. Electrodes were fixed in place on the UT nerve under visual observation; the other branches of the vestibular nerve were transected. 2. The N1 field potential evoked by UT nerve stimulation was recorded in the vestibular nuclei at the start of each experiment. The potential typically grew until it reached a plateau. Stimulus spread (if any) to the central ends of other nerve branches was revealed by an additional increase in N1 amplitude after the plateau was reached. 3. We recorded intracellularly from 55 motoneurons in C1-C3. Some were identified as having axons in the dorsal rami, which innervate dorsal neck muscles. Others projected in nerves that were not available for stimulation. 4. UT nerve stimulation evoked synaptic potentials in essentially all motoneurons studied. The predominant pattern consisted of disynaptic excitatory postsynaptic potentials in ipsilateral motoneurons and inhibitory postsynaptic potentials that were at least trisynaptic in contralateral motoneurons. 5. The results demonstrate the presence of short-latency connections between the utricular nerve and dorsal neck motoneurons. The functional role of this pathway remains to be investigated.

Serotonergic excitatory drive to hypoglossal motoneurons in the decerebrate cat

In decerebrate, paralyzed, vagotomized and artificially ventilated cats, serotonin (5-HT) and its analogues, microinjected into the hypoglossal (XII) motor nucleus, altered the activity of the genioglossal branch of XII nerve. 5-HT, carboxamidotryptamine malcate (5-CT) and DOI (1–5 mM) increased the activity by over 200%. Methysergide reversed this increase. Methysergide, mianserin, or ketanserin (100–250 nl, 1 mM) reduced the spontaneous hypoglossal activity by 20–50%. Buspirone, 8-OH-DPAT and (−)-propranolol were without effect. Thus, 5-HT provides a substantial tonic excitatory drive to XII motoneurons. The 5-HT receptors involved are likely to be type 1C or 2, but uncertainty regarding the affinity profiles of the drugs used in in vivo conditions in the cat precludes a definite identification.

A light and electron microscopic study of intracellularly HRP-labeled lumbar motoneurons after intramedullary axotomy in the adult cat.

In contrast to many other neurons in the central nervous system, spinal motoneurons in adult cats have been shown to regenerate their axons after an axotomy accomplished within the CNS compartment. This regenerative capacity may be the result of extrinsic influences, or intrinsic properties of the motoneurons themselves, or interactions between extrinsic and intrinsic factors. As part of the effort to establish circumstances of importance for this central regeneration, a detailed analysis of the morphology of lumbar motoneurons was performed 3-11 weeks following a ventral funiculus axotomy. Fourteen large neurons considered to be intramedullarly axotomized alpha motoneurons were labeled intracellularly with horseradish peroxidase. Twelve out of the fourteen analyzed neurons had an axonlike regenerating process. These twelve neurons could, in turn, be separated into two groups, based on the proximity of the axonal lesion and the proximal morphology of the regenerating process. Thus, after a comparatively proximal axotomy, new axons were produced, originating either from the cell soma or from a distal dendritic branch. After a more distal axotomy, but still intramedullarly, it seemed as if the proximal part of the original axon always persisted and subsequently regenerated. Analysis of the relation between the cell soma diameter and the diameter and number of its stem dendrites revealed that dendrites become thinner and also decrease in number after an intramedullary axotomy. In this way, it may be calculated that the total dendritic surface area of lesioned motoneurons will decrease by approximately half. In four neurons, most dendrites had an abnormal appearance in the light microscope with increasing diameter of distal branches. Ultrastructural analysis revealed that such dendrites were surrounded by myelin sheaths. Small filopodia in close relation to axon terminals were found to emerge from the cell membrane of the lesioned motoneurons. Their function may be to establish contact with presynaptic elements and then retract them to the cell membrane. We interpret the morphological changes of the motoneurons as signs of a large capacity for axonal regeneration, even after axotomy in the central nervous system.

Multi-phase expiratory inhibition of phrenic motoneurons in the decerebrate cat.

We studied the membrane potential changes of 11 phrenic motoneurons (PMs), located in the C5 portion of the cervical spinal cord, in decerebrate, paralyzed and artificially ventilated cats. Intracellular chloride iontophoresis into PMs could reverse the waves of inhibitory postsynaptic potentials (IPSPs) causing membrane hyperpolarization during expiration. In all PMs, we revealed, in addition to the previously described late expiratory augmenting wave of IPSPs, an early expiratory decrementing wave of IPSPs. Moreover, in three PMs, recorded in two cats exhibiting slow respiratory rhythms, an additional wave of IPSPs was interposed between the early and late ones. The possible origins and roles of these three different waves of inhibition are discussed.

Active electrophysiological properties of spinal motoneurons in aged cats following axotomy

The present study was designed to examine the effects of the aging process on the response of motoneurons to axotomy. Accordingly, in aged cats using intracellular recording techniques, the electrophysiological properties of axotomized lumbar spinal cord motoneurons were compared with those of control (nonaxotomized) motoneurons. In motoneurons that were subjected to axotomy, there was a reduction in axonal conduction velocity compared to that exhibited by control motoneurons. In addition, there were a number of changes in the configuration of the action potential following axotomy. The amplitude of the spike and the overshoot increased as did the slope of the soma-dendritic spike, whereas the delay between the initial segment and the soma-dendritic spikes decreased. The duration of the action potential's afterhyperpolarization increased; its amplitude remained unaffected although the calculated afterhyperpolarization current decreased. Following the spike, most of the axotomized motoneurons exhibited hyperpolarization undershoots and delayed depolarizations. Axotomized motoneurons exhibited a small decrease in the membrane potential and a reduction in the rheobasic current compared to control cells. The changes in the frequency distribution of axonal conduction velocity, afterhyperpolarization duration, afterhyperpolarization current and rheobase measurements suggest that aged motoneurons dedifferentiated following axotomy. These results indicate that axotomy, in aged motoneurons, results in the disruption of a variety of electrophysiological parameters and that the specific patterns of the responses that occur in axotomized motoneurons of adult cats also emerge in axotomized motoneurons of aged animals.

Passive electrical properties of motoneurons in aged cats following axotomy

The objective of this study was to determine whether the aging process influences the changes in the electrophysiological properties of motoneurons that occur as a consequence of axotomy. Accordingly, using intracellular recording and stimulating techniques, the basic electrical properties of control (unaxotomized) and axotomized spinal cord motoneurons of aged cats were determined. Compared with control motoneurons, axotomized motoneurons exhibited increases in input resistance ( R in), membrane time constant (τ b) and the equalizing time constant (τ c). While the electrotonic length ( L) remained unchanged, axotomy induced a decrease in the total cell capacitance ( C cell. The post-axotomy reduction of C cell indicates that the motoneuron surface area was reduced and the increased membrane time constant indicates that there was an increase in membrane resistivity ( R m). The post-axotomy conservation of L accompanied by an increase in R m suggests that aged axotomized motoneurons undergo geometrical changes. Furthermore, calculations based on cable theory suggest that the diameter of the equivalent cylinder ( d) decreased following axotomy, whereas the equivalent cylinder length ( l) remained unaffected. It is concluded that axotomy produces significant alterations in the soma-dendritic portion of aged spinal motoneurons, as indicated by the changes found in their passive electrophysiological properties, and that the pattern of the response that occurs in axotomized motoneurons of adult cats is also present in axotomized motoneurons of aged animals.

Dendritic origin of the A-spike (IS-spike) of motoneurons in adult cats and infant pigs

Dendritic origin of the A-spike (IS-spike) of motoneurons in adult cats and infant pigs

Monosynaptic excitation of motoneurons innervating forelimb muscles following stimulation of the red nucleus in cats

We examined the rubrospinal projection to forelimb motoneurons (Mns) in cats, in comparison with the corticospinal projection. Under pentobarbital anesthesia, intracellular recordings were made from forelimb Mns following stimulation of the red nucleus (RN) and the cerebral peduncle (CP). Single pulse stimulation of the RN produced excitatory postsynaptic potentials (EPSPs) in the majority of forelimb Mns ( 58 80 , 72%). Segmental latencies of RN-EPSPs were shorter than 1.2 ms in 14 out of 22 Mns in which RN-EPSPs were detected in the C 8–T 1 segments, and in 3 of 36 Mns in the C 6–C 7 segments. The results suggest monosynaptic rubro-motoneuronal connections in a substantial portion of forelimb Mns particularly in the C 8–T 1 segments. There was no evidence suggesting monosynaptic connections between cerebral cortex and forelimb Mns.

Behavior of cat abducens motoneurons following the injection of toxic ricin into the lateral rectus muscle

The aim of this investigation was to study the behavior of identified abducens motoneurons in the chronic cat following a single injection of toxic ricin into the lateral rectus muscle. Lateral rectus electromyographic potentials induced by VIth nerve stimulation disappeared, and abducens antidromic field potentials decreased by 90% 3 days following ricin injection. Several abnormalities and a significant decrease in eye position and velocity sensitivities were observed in motoneuron activity up to 8–10 days following ricin injection. Contrary to a previous report for axotomized abducens motoneurons, no functional sign of recovery was observed. Histological analysis showed a survival of 10–15% of the abducens motoneuron population 10 days following ricin injection. From this time on, recorded motoneurons behaved like controls, but showed a specific retraction signal suggesting an exclusive projection onto the retractor bulbi muscle. Although intermingled in the nucleus with motoneurons, no recorded abducens internuclear interneuron was affected by the ricin during one month following the injection.

Axotomy-like changes in cat motoneuron electrical properties elicited by botulinum toxin depend on the complete elimination of neuromuscular transmission

The electrical properties of cat medial gastrocnemius (MG) spinal motoneurons were studied 14-21 d following injection of type A botulinum toxin (BTX) into the MG muscle. Treated MG muscles were atrophic, displayed pronounced fibrillation activity, and were markedly but not completely paralyzed. MG motoneuron electrical properties from animals with the highest MG muscle-twitch forces (greater than 20 gm) appeared normal, while motoneuron properties from animals with the lowest MG muscle-twitch forces (less than 10 gm) exhibited axotomy-like changes, though these changes were less pronounced than after axotomy itself. No changes in the axonal conduction velocity were observed, however. Motoneuron connectivity with MG muscle fibers was determined following intracellular stimulation of MG motoneurons by averaging EMG signals from 3 or 4 pairs of recording electrodes inserted into the BTX-treated MG muscles. Normal electrical properties were observed among motoneurons in which detectable EMG activity linked to the intracellular stimulation pulse was observed. The level of this connectivity, however, indicated that a relatively small number of muscle fibers were activated by individual motoneuron action potentials. Axotomy-like changes of electrical properties were observed in MG motoneurons that could not be associated with detectable EMG activity in the BTX-treated MG muscle following repeated trials of intracellular stimulation. These results indicate that the existence of effective neuromuscular transmission at a small number of motor terminals is sufficient to prevent the appearance of axotomy-like changes in motoneuron electrical properties, and that the absence of such transmission at all motor terminals is associated with the appearance of axotomy-like changes. The results suggest that the effects of axotomy itself on motoneuron properties may be based upon the loss or elimination of a potent interaction between muscle and motoneurons normally mediated by neuromuscular transmission.

Alterations in motoneuron properties induced by acute dorsal spinal hemisection in the decerebrate cat

Using intracellular recording techniques, we studied the response characteristics of two separate populations of triceps surae motoneurons in unanesthetized decerebrate cats, recorded before and after low thoracic hemisection of the spinal cord. In each preparation, we studied the response properties of one group of motoneurons and the protocol was then repeated for a separate group, immediately following the dorsal hemisection. In each group, we examined both the minimum firing rates of motoneurons during intracellular current injection and a range of cellular properties, including input resistance, rheobase current and afterhyperpolarization time course and magnitude. Although earlier studies from this laboratory have shown substantial reductions in minimum firing rate in reflexively active motoneurons in the hemisected decerebrated preparation, the response of motoneurons to intracellular current injection in the current preparation proved to be quite different. Minimum firing rates were either normal or even somewhat higher in the post-lesion group, while the time course of the afterhyperpolarization was shortened. Moreover, these effects were not evenly distributed across the motoneuron pool. The rate effect was most evident in motoneurons with higher conduction velocity, while the afterhyperpolarization effect occurred predominantly in motoneurons with lower conduction velocity. Neither of these effects could be accounted for by lesion-induced changes in other cellular properties. We conclude that tonically active neurons with descending axons traversing dorsolateral white matter may influence both the discharge characteristics and membrane properties of spinal motoneurons in novel ways, presumably by modifying voltage or calcium activated motoneuronal conductances. The previously described reactions in the firing rate of motoneurons after such lesions appear to be mediated by different means, perhaps by alterations in synaptic input from segmental interneurons.