1. The Mesencephalic Locomotion Region (MLR) Contains ACh Neurons
Large populations of cholinergic neurons are present in the caudal mesencephalon and rostral pons, in regions traditionally associated with the location of the mesencephalic locomotion region (MLR). The initial studies describing the MLR in decerebrate cats
[1], revealed that the cuneiform nucleus (CuN) was part of this locomotor region (for reviews see
[2][3]). The most striking observation at the time was the tight coupling between the MLR stimulation strength and the intensity of the locomotor output that ensued: as the MLR stimulation intensity increased, the generated locomotor speed also increased proportionally, as if locomotion was controlled by a rheostat. The initial discovery of the MLR ignited significant new interest in the supraspinal control of locomotion. Several animal species became research subjects in the field, and the MLR was functionally identified in all of the vertebrate species in which it was investigated (reviewed in
[3][4]). Perhaps the most noticeable observation was the similar location of the MLR in those vertebrate species. Moreover, it was found that the MLR projections were very similar (reviewed in
[5]). In the cat, the comparison between the distribution of the choline acetyltransferase-labeled cells (according to Kimura et al.
[6]) and the regions that efficiently induced locomotion upon their electrical stimulation
[7][8] showed a striking similarity
[9]. It is now generally accepted that the MLR contains a mixture of glutamatergic, GABAergic, and cholinergic neurons distributed over several adjacent brainstem nuclei.
The pedunculopontine nucleus (PPN) was also shown in rats to be part of the MLR thanks to the elegant work of E. Garcia-Rill
[10][11] (for reviews see
[12][13]). The CuN and PPN occupy a large part of the mesencephalon, and, therefore, it is likely that the different parts of these two nuclei could be associated with different aspects of locomotor control. Additionally, it was shown in both lampreys and salamanders, that the laterodorsal tegmental nucleus (LDT) also controlled locomotion efficiently. The LDT is located medially near the caudal pole of the PPN, and it contains a large proportion of cholinergic neurons. In lampreys and in salamanders, the LDT was shown to be the most efficient region from which locomotion could be elicited and controlled
[14][15] (for reviews see
[3][16]).
It appears therefore that the PPN and CuN, and the LDT, at least in basal vertebrates, constitute the main brainstem nuclei responsible for the locomotor effects of the MLR stimulation (e.g.,
[10][14][15][17]). In addition, recent work carried out on subjects with Parkinson’s disease
[18] showed that the mesencephalic deep brain stimulation of the PPN and CuN areas efficiently treated the freezing of gait, suggesting that the PPN and CuN may also be constitutive of the MLR in humans. These nuclei project to a variety of brain areas
[19], and contain different proportions of glutamatergic and cholinergic projection neurons
[20][21][22][23].
2. MLR Implications in Locomotor Control: Targets, Pathways, and Pharmacology
Imaging experiments in humans revealed that both the CuN and the dorsal PPN are active during imaginary fast walking
[24]. Cholinergic neurons of the PPN have also been proposed to be involved in several motor functions including REM sleep, cervical tone, startle responses and locomotion (reviewed in
[25]). In primates including humans, PPN cholinergic neurons play a role in controlling gait and posture
[24]. Similarly, activating the ventral PPN, which mostly contains cholinergic neurons in cats, suppresses tonic muscle tone, subsequently allowing locomotion
[26], whereas the genetic suppression of the vesicular ACh transporter in cholinergic neurons of both the PPN and LDT generates dramatic motor deficits in mice
[27]. In contrast, it was shown that a complete excitotoxic lesion of the PPN alone failed to affect gait significantly in rats
[28].
The MLR sends cholinergic axons to many key forebrain and brainstem structures controlling locomotion
[19]. Therefore, it is likely that the MLR constitutes the major source of ACh modulation for locomotion. The basal ganglia regulate goal-directed behaviors by exerting an inhibitory control on the MLR, as shown by experiments in which a MLR disinhibition allowed for the initiation of locomotion (see
[29]). In turn, MLR neurons project to the basal forebrain structures involved in locomotor regulation such as the basal ganglia
[30][31] and the ventral tegmental area (VTA; e.g.,
[32]). Indeed, in mammals, the PPN and LDT cholinergic neurons projecting to the VTA were proposed to modulate locomotion related to drug- and novelty-seeking behaviors, specifically (reviewed in
[33]) via both mAChRs
[34] and nAChRs
[35]. In rats, the optogenetic activation of PPN cholinergic neuron terminals in the VTA increases locomotion, whereas the activation of cholinergic terminals from the LDT tends to reduce locomotion
[32] (but see also
[36]). These effects presumably result from modulatory effects exerted by the VTA on the striatum.
Stimulation of the PPN in the rat triggers a dopamine release in the basal ganglia through the activation of both ionotropic glutamate receptors and nAChRs, and through the activation of mAChRs, all found on substantia nigra neurons. On the other hand, the M2-type muscarinic autoreceptors located on PPN neurons decrease the nigrostriatal dopamine
[30]. In addition, ACh interneurons are present in the striatum where they exert a counterbalancing effect to dopamine inputs, thus regulating the striatum activity (reviewed in
[37]). Modeling experiments suggested that this interaction is necessary to induce locomotion
[38]. In contrast, the cholinergic innervation originating in the PPN was proposed to allow the basal ganglia to modulate/adapt ongoing locomotion to behavioral constraints (reviewed in
[39]). Nevertheless, whatever the source of ACh (local interneurons within basal ganglia or PPN neurons), interactions between ACh and dopamine in the striatum appear essential for motor control, and motor dysfunctions in Parkinson’s disease have been correlated with the dysregulation of such interactions (e.g.,
[24][40]; for a review, see
[37]). Although an increasing amount of data illustrate the impact of the MLR stimulation on forebrain motor structures, the control exerted by the MLR on locomotion is largely associated with the regulation of downstream motor nuclei of the brainstem (reviewed in
[3][4]).
The cholinergic neurons seem to represent only about one quarter of the PPN neurons projecting to the reticular nuclei in mice
[41]. In contrast, various studies in another rodent species, the rat, showed that most descending PPN neurons
[42][43] and a substantial part of the reticular cell-connecting CuN neurons
[44] are cholinergic. It is noteworthy that rat CuN cholinergic neurons are involved most likely in sensory modulation and cardiovascular regulation
[44], which suggests a secondary role for these cholinergic neurons in locomotor control in this species. Therefore, recent optogenetic studies in rodents have attempted to establish the specific contribution of the glutamatergic and cholinergic neuronal populations in the MLR to the control of locomotion.
Glutamatergic cells in the CuN were shown to play a role in generating the locomotion at different speeds
[45][46][47][48], whereas glutamatergic neurons of the PPN rather control locomotion at lower speeds
[46][47]. In some cases, locomotion was halted by activating PPN glutamatergic neurons
[47][49][50], and the induction or termination of locomotion was shown to depend on the respective activation of the glutamatergic and GABAergic reticulospinal systems of the lower brainstem
[41]. Such a MLR-controlled start/stop system seems remarkably well conserved in vertebrates since comparable results were found in the lamprey
[51]. The same approach has been used to examine the role of the PPN cholinergic neurons in the control of locomotion. As of now, the results are contradictory as increases and decreases in locomotor output were observed
[45][46][47].
All these results strongly suggest that glutamatergic MLR neurons play a predominant role in triggering locomotion through the activation of downstream RSNs compared to cholinergic MLR neurons
[52]. This is supported by the observation in cats that cholinergic antagonists seem to produce only little and temporary effects on spontaneous
[53] or MLR stimulation-induced locomotion
[54]. In striking contrast however, the direct injection of ACh agonists in the reticular formation is sufficient to trigger locomotor bouts in rats, birds, and lampreys
[15][55][56], while ACh antagonists block MLR-induced RSN depolarization and locomotion in rats
[55][57]. The latter results indicate that ACh may also play a significant role in MLR-Induced locomotion. Therefore, activating either the glutamatergic or cholinergic system independently may yield an incomplete picture of the specific role of these two neurotransmitter systems in controlling locomotor behavior. In addition, because slight variations in the experimental approaches (stimulation site, intensity/frequency parameters) may trigger different motor performances (e.g.,
[58]) a degree of caution is required in the interpretation of the behavioral observations resulting from localized brain stimulation.
Recently, the group of Bretzner
[47] has shown in mice that the joint activation of glutamatergic and cholinergic neurons in the PPN modulated CuN-evoked locomotion, converting running into walking, thereby suggesting that both the glutamatergic and cholinergic systems participate in the MLR command. Interestingly, neurotransmitter interactions in the context of the MLR-controlled locomotion has already been reported in the lamprey, and it may be conserved across vertebrates
[59], where forebrain-originating dopamine enhances both the MLR command onto RSNs
[60] and RSNs directly
[61]. Both the glutamatergic and cholinergic MLR inputs were shown electrophysiologically in the lamprey to converge onto RSNs, and each transmitter system can trigger a sustained RSN depolarization associated with locomotion in a semi-intact preparation (reviewed in
[4]). Therefore, analyzing the RSN responses to both the glutamatergic and cholinergic MLR inputs is necessary to eventually understand the exact contribution of each neurotransmitter in the MLR-evoked locomotion in mammals.
3. Cholinergic Effects on Reticulospinal Neurons
Whereas the cholinergic effects of the MLR on other supraspinal structures is still not understood at the cellular level, the lamprey model has allowed researchers to gain a better understanding about the MLR-related ACh effects on downstream RSNs in the context of goal-directed locomotion.
The local application of ACh or nicotine on pontine reticulospinal nuclei (MRRN) consistently produced excitatory postsynaptic potentials (EPSPs) in a dose-dependent fashion in the RSNs. Large doses triggered locomotor activity in either isolated brainstem/spinal cord (fictive locomotion in spinal ventral roots) or in semi-intact preparations (active locomotion)
[15]. In addition, the ACh-evoked EPSPs showed summation properties coherent with the known graded effect of the MLR stimulation. Moreover, during spontaneous swimming in a semi-intact preparation, both nicotinic agonists significantly accelerated the ongoing rhythm
[15]. As the first direct demonstration of ACh inputs on RSNs, this indicates that the ACh inputs from the MLR are likely to play a substantial role in the initiation and the control of locomotion in lampreys, by allowing the pontine RSNs to generate sustained depolarization and firing under cholinergic effects. In lampreys, the pontine RSNs and their activation by ACh are crucial for MLR-induced locomotion, while the bulbar RSNs (PRRN) seem to produce less powerful effects, since only the glutamatergic antagonists prevented the acceleration of the locomotor rhythm by a stronger MLR stimulation
[62]. The latter finding underlines the major contribution of the glutamatergic MLR inputs onto the bulbar neurons of the lateral paragigantocellular nucleus reported in mammals
[52]. These similarities between lampreys and mammals certainly warrant a much needed, redirected attention to the pontine RSNs in mammals.
4. A Parallel Muscarinic Hyperdrive to Boost the Locomotor Output
As indicated above, the MLR projections to RSNs are essential to control locomotion. In lampreys, the MLR projects directly to RSNs via monosynaptic glutamatergic and cholinergic connections, and it also projects to RSNs indirectly via cholinergic connections. The direct cholinergic excitation is nicotinic in nature
[15], while the indirect projection is muscarinic
[63]. A bath application of muscarine elicits sustained and recurring depolarizations in RSNs. Calcium imaging revealed oscillations in calcium levels that occurred synchronously within the entire RSN population that was imaged. These oscillations were abolished by the bath application of TTX indicating that they were not intrinsic but driven by other neurons. Subsequent experiments revealed that such RSN oscillations were driven by a group of cells located at the junction between the pontine and bulbar reticular formation. The driving cells were named “muscarinoceptive”, and anatomical studies revealed that these neurons project directly to the reticular formation. The resulting effect of activating this group of cells is a marked amplification of the activity in the RSNs and an increase in the duration of locomotor output. The muscarinoceptive cells were shown themselves to respond to muscarine with long-lasting bouts of activity, to receive a direct muscarinic excitation from the MLR, and to send a glutamatergic excitation to the RSNs. Blocking the mAChRs on these neurons dramatically reduced the MLR-induced excitation of RSNs and slowed locomotion. These results further explain the previous observations in the lamprey that muscarine elicited a sustained depolarization in RSNs
[64].
The presence of such a boosting mechanism has yet to be demonstrated in other vertebrate species. However, findings from older studies revealed that the activation of muscarinic receptors in the same region of the reticular formation in birds and mammals can elicit bouts of locomotion. Indeed, the locomotor behavior in birds can result from brainstem injections of carbachol, a nonspecific cholinergic agonist, and these effects are blocked by the muscarine receptor antagonist, atropine
[56]. Cholinergic inputs are also believed to activate brainstem neurons in mammals
[55][57], and a group of muscarine-sensitive neurons receiving cholinergic inputs from the PPN was described in rats in the ventromedial medulla close to the pontine border
[65], at a location similar to that of the muscarinoceptive cells in lampreys. The role of these neurons was not described in relation to locomotion in mammals, but the similarities in their properties and location support the idea that they could amplify the reticulospinal descending signals to boost the locomotor output, just as has been observed in lampreys.
The presence of muscarinic effects in birds and mammals, as in lampreys, suggests that the muscarinic amplifying mechanism is conserved in evolution. It further suggests that the supraspinal control of locomotion in vertebrates is not exclusively composed of linear projections down from the forebrain to the spinal cord. Rather, there exists an additional feedforward cholinergic “hyperdrive” component, originating in the MLR and allowing for a supplementary activation of the locomotor system.
5. Muscarinic Control of Brainstem Sensory Inputs
In lampreys
[66][67] as in mammals (e.g.,
[68]) the reticular neurons, and especially the RSNs, integrate information from various sensory modalities in order to generate adaptive motor commands. It is also well established that cat RSNs that are rhythmically active during locomotion, in addition receive sensory inputs from a large portion of the body
[69]. Interestingly, the MLR has been shown to modulate sensory inputs that reach the RSNs in the lamprey
[70]. Indeed, a short duration electrical stimulation of the MLR depresses, for several tens of minutes, the EPSPs evoked in RSNs by trigeminal nerve stimulation. These effects were prevented by perfusing the brainstem with the muscarinic antagonist atropine. Moreover, mAChRs were immunohistochemically identified on the cells that relay trigeminal sensory signals to the RSNs
[71], as well as on the RSNs themselves. Similar muscarinic effects were observed after ejecting ACh or the muscarinic agonist pilocarpine directly onto intracellularly recorded RSNs while electrically stimulating the trigeminal nerve
[72]. This muscarinic modulation was not related to any changes in the RSN membrane properties, but rather depended on a presynaptic control of the synaptic transmission between the trigeminal relay cells and RSNs. In contrast, the perfusion of atropine strongly potentiated the RSN responses to trigeminal nerve stimulation, suggesting the presence of a tonic inhibitory mAChR-mediated regulation of RSNs and/or trigeminal relay cells. These observations in lampreys suggest that predictable sensory inputs would be gated by MLR inputs, allowing for a smooth behavioral output to occur during goal-directed locomotion. Such a filtering has been proposed in numerous species for phasic sensory inputs impinging on the locomotor CPGs in the spinal cord where there is a strong presynaptic modulation of the primary afferent terminal inputs during fictive locomotion (see below and
[73]), but was not envisioned yet in the brainstem. In addition, when the MLR is inactive the muscarinic modulation would be strongly decreased, and this would open the gates for sensory inputs and thus facilitate sensory-evoked locomotion (e.g., escape swimming
[66][67]). Moreover, blocking the muscarinic inputs unmasked persistent RSN membrane potential oscillations after applying NMDA onto the RSNs
[72] that could further support escape swimming.