Command neuron for a half-center oscillator

Our latest paper just came out in the Journal of Neuroscience.

Command or obey? Homologous neurons differ in hierarchical position for the generation of homologous behaviors
Akira Sakurai and Paul S. Katz 
J Neurosci 17 June 2019, 3229-18
DOI: https://doi.org/10.1523/JNEUROSCI.3229-18.2019

Click here for the reprint

Nudibranchs have homologous neurons that can be identified across species. Cross-species comparisons of motor system organization provide fundamental insights into their function and origin. This paper shows that an identified cerebral ganglion neuron serves as a command neuron for the swimming behavior in a nudibranch species. The same neuron serves as a member of a central pattern generator (CPG) in another species. We described the synaptic and neuromodulatory mechanisms by which the command neuron initiates and accelerates rhythmic motor patterns.


Two sea slug species show homologous swimming behaviors
In motor systems, higher-level components issue commands that are carried out by lower-level circuits. In this paper, we describe the physiological actions of an identified neuron, which turned out to be a "command" neuron for the swimming behavior of a giant sea slug, Dendronotus iris. We determined which functional components of the swim CPG are modulated by the command inputs to initiate, maintain, and terminate the rhythmic activity of a central pattern generator circuit.

Among the swimming nudibranchs, two species Melibe leonina and Dendronotus iris show the homologous swimming behavior by flexing their bodies from left to right (Sakurai et al., 2011).
 



Homologous behaviors are produced by homologous neurons
Phylogenetic analysis indicates that the most recent common ancestor of these species likely swam in this manner, making the swimming behaviors homologous (Goodheart et al., 2015; Sakurai and Katz, 2017).

The brains of Melibe leonina (left) 
and Dendronotus iris (right)

Homologous behaviors are produced by distinct neural circuit designs
The neural circuits underlying their behaviors have been studied extensively in both species. All neurons in the swim CPGs have been identified, and their synaptic connections have been determined with careful pairwise electrophysiological recordings (Sakurai et al., 2014; Sakurai and Katz 2016). The two swim CPGs employ different network architectures for producing similar rhythmic motor patterns.

The swim CPGs of Melibe (left) and Dendronotus (right)
have distinct synaptic organizations 

Si1 as a command neuron in Dendronotus
In this study we found that Si1 neurons in Dendronotus iris serves as a neuromodulatory "command" neuron for the swim CPG.

We further revealed how such command actions were mediated by performing dynamic clamp experiments and electrophysiological manipulations.

Providing artificial synaptic boost of the Si3-to-Si2 synapse and tonic synaptic excitation of Si3 mimicked the command actions of Si1 neurons.


Synaptic and neuromodulatory actions underlie the command input
It turned out that the organization of the Dendronotus swim CPG closely resembles the model that was originally proposed for a half-center oscillator with excitatory drive arising from a command neuron (Friesen, 1994).

A classical model of the half-center oscillator (left) 
and the Dendronotus swim CPG (right)

The command neuron Si1 provides not only the overall excitatory drive but also the neuromodulation of synaptic potentiation within each half of the oscillator. Our results also suggest that the functional position of neurons in a motor hierarchy can shift from one level (CPG) to another (a command neuron) over evolutionary time.

• Friesen WO (1994) Reciprocal inhibition: a mechanism underlying oscillatory animal movements. Neuroscience and biobehavioral reviews 18:547-553.
• Goodheart JA, Bazinet AL, Collins AG, Cummings MP (2015) Relationships within Cladobranchia (Gastropoda: Nudibranchia) based on RNA-Seq data: an initial investigation. R Soc Open Sci 2:150196. 
• Sakurai A, Katz PS (2016) The central pattern generator underlying swimming in Dendronotus iris: a simple half-center network oscillator with a twist. J Neurophysiol 116:1728-1742.
• Sakurai A, Katz PS (2017) Artificial Synaptic Rewiring Demonstrates that Distinct Neural Circuit Configurations Underlie Homologous Behaviors. Curr Biol 27:1721-1734 e1723.
• Sakurai A, Newcomb JM, Lillvis JL, Katz PS (2011) Different roles for homologous interneurons in species exhibiting similar rhythmic behaviors. Curr Biol 21:1036-1043.
• Sakurai A, Gunaratne CA, Katz PS (2014) Two interconnected kernels of reciprocally inhibitory interneurons underlie alternating left-right swim motor pattern generation in the mollusc Melibe leonina. J Neurophysiol 112:1317-1328. 

Journal club: The cortex as a central pattern generator

NATURE REVIEWS NEUROSCIENCE 6(6) 2005

by Rafael Yuste, Jason N. MacLean, Jeffrey Smith and Anders Lansner

A vast number of neurons in the brain cortex together generates synchronized oscillatory activity. In this review paper, the authors discuss how the neocortex neural network shares basic designs with the central pattern generator circuits in the spinal cord and brain stem.

For rhythmogenesis in a neural network, it seems like the general rule that self-boosting excitation within a functional unit is more important than reciprocal inhibition.  In the neocortical circuit, a population of neurons forms an excitatory kernel by being interconnected through recurrent excitation. Such excitatory neuronal population is capable of producing synchronized oscillatory activity through their active membrane properties and short-term synaptic plasticity. This circuit architecture is similar to the CPGs in the spinal cord and brain stem of vertebrates. The inhibitory connections are necessary for the regulation of the rhythm and spatiotemporal pattern of the network output.

The extent of spatial distribution is one of the major differences between the cortical network and the CPG in the brain stem/spinal cord.  In the cortex, the recurrent excitatory network is widely spread out over the neocortex, whereas in those for locomotion and breathing have bilaterally-grouped functional units. The other difference is high plasticity in the cortical network in contrast to the hard-wired CPGs in the spinal cord and brain stem. The connectivity in the cortex develops through activity-dependent Hebbian plasticity. We can call the cortex network as a learning CPG that is based on Hebbian assemblies and is specialized for learning and storing or retrieving memories.  Because of these properties, the neocortex network is highly fluidic.

Journal Club: Fast Silencing Reveals a Lost Role for Reciprocal Inhibition in Locomotion

By Peter R. Moult, Glen A. Cottrell, and Wen-Chang Li
Neuron Volume 77, Issue 1, 9 January 2013, Pages 129-140

Reciprocal inhibition is considered as a fundamental building block in neural circuits for rhythmic motor pattern generation. Despite commonality in many rhythmic systems, however, the reciprocal inhibition is often found to be not a necessary component for rhythmogenesis. This is because a network is often capable of exhibiting rhythmic activity even when the reciprocal inhibition was removed mechanically or pharmacologically. The excitatory synaptic components within each half of the network are more crucial.  In this study, Wen-Chang Li's group challenged this idea by using optogenetics to hyperpolarize the entire population of neurons in one side of the spinal cord in a tadpole.

When the entire half of the spinal cord was suppressed by flashing yellow light, the other half also stopped bursting. This indicates that, in this reciprocally inhibitory network, the excitation in one side is necessary for the excitation of the other. Firing in one side would cause a bombardment of IPSPs, which induce post-inhibitory rebound. The authors also get a similar result by giving a strong hyperpolarizing current into one of the reciprocally inhibitory neurons. They concluded that the reciprocal inhibition is necessary in the tadpole swim circuit; it induces rebound excitation in the contralateral neurons.

In this paper, it is noteworthy that the author also mentioned about the "functional homeostasis" of the neural circuit. They said the rhythmic activity recovers after half an hour by itself without the reciprocal inhibition. The mechanism of the recovery still unknown.

Melibe and Dendronotus Dynamic Clamp paper

My latest paper will appear in Current Biology next weekend.

Artificial Synaptic Rewiring Demonstrates that Distinct Neural Circuit Configurations Underlie Homologous Behaviors
http://www.cell.com/current-biology/fulltext/S0960-9822(17)30552-3
by Akira Sakurai and Paul S. Katz

Behaviors can be homologous just like any other trait can be. This study directly compared neural circuit mechanisms underlying homologous behaviors in two closely-related species.

   This work originates from two questions. First, we wanted to grasp a clue to figure out how species-specific behaviors have evolved. Mollusks are good models to study this because of their wide variety of speciation and the simplicity of the nervous system. In other animal models, functional elements of a neural circuit often consist of a population of neurons having the same function. This makes it difficult to manipulate because there are so many. In contrast, mollusks have one large neuron playing a key role in generating motor output for behavior. Their behaviors are also simple and reliable. The neuronal activity can be precisely manipulated so that one can easily relate one neuron to one behavior. By looking into molluscan species, we hoped we might be able to witness how a species-specific behavior has evolved.
   Secondly, we have been wondering what would happen if we swap a neural circuit of one species with that of the other species. There is a technique called "dynamic clamping", by which one can modify the strength of synapses or membrane conductances by injecting electrical current into neurons. The amount of the injected current is calculated in real-time by a computer based on the membrane potential that is being recorded. With this technique, we hoped we could reveal a crucial element that provides a neural circuit function.

   In this paper, we compared two neural circuits underlying swimming behaviors of two nudibranchs, Melibe leonina and Dendronotus iris. These sea slugs swim by flexing their body from left to right. These behaviors are likely to be homologous because both species belong to a clade that consists only of families that contain species that swim in the same way.
   We found that their swimming behaviors are produced by distinct neural circuit mechanisms. In Melibe, one of the neurons called Si3 makes an inhibitory synapse to fine-tune the rhythm made by other neurons.  In Dendoronotus, Si3 provides excitatory drive to other neurons to induce their rhythmic activities. When the Si3 synapses were blocked by curare, the swim rhythm slowed down in Melibe; whereas in Dendronotus, curare abolished the motor pattern. Replacing these synapses by artificial computer-generated synapses using "dynamic clamping" immediately restored the motor pattern. Using the dynamic clamp, we also rewired the Dendronotus circuit to the Melibe circuit. Then the Dendronotus neurons started to burst like the Melibe neurons in curare.

   From the results, we discussed that the neural mechanisms underlying homologous behaviors appear to have diverged but are still interchangeable in the other species. This has important significance for making inferences into neural mechanisms based on behavior. It also provides a real life instantiation for a prediction based on models, that there are multiple circuit architectures that can produce the same pattern of activity.

Journal club: Mechanisms of oscillation in dynamic clamp constructed two-cell half-center circuits

A. A. Sharp, F. K. Skinner, E. Marder
Journal of Neurophysiology Published 1 August 1996 Vol. 76 no. 2, 867-883

Dynamic clamping is merely current injection triggered by presynaptic voltage. The amount of current is determined mainly by postsynaptic voltage. There is nothing magical.

In this paper, the authors studied how systematic alterations in intrinsic and synaptic parameters affected the network behavior by using their newly-developed "dynamic clamping" on a pair of the gastric mill motor neurons, GMs. They isolated GMs by blocking synaptic transmission with picrotoxin and created reciprocally inhibitory two-cell circuits by the dynamic clamp.

The author demonstrated in this system that there was no bursting without the hyperpolarization-activated inward current (I_H). In the presence of additional I_H, a variety of circuit dynamics, including stable half-center oscillatory activity, was produced. The increase in synaptic conductance increased the burst period, whereas the increase in IH conductance reduced it. Discussion went on about synaptic threshold, saying that changes in the synaptic threshold might play a large role in turning on and off bursting activity. However,  these discussion are not so informative to others because nonspiking synapses are rarely seen other than crustaceans.

The authors often stated that they "depolarized" or "hyperpolarized" the thoreshold for synapse. This is wrong. The threshold does not "polarize." Membrane potential does. They should have stated that the threshold was changed to more depolarized or hyperpolarized levels.

Journal club: Recent Clione papers

Arshavsky published a series of studies on the Clione swim CPG. Those papers are "must read" for researchers who work on CPG. I pay the highest respect to them.They were published in late 80's and now can be categorized as classical papers.
   More recently, there came a series of papers performing more detailed(?) analyses on this system. The following are my impressions with regard to those recent Clione papers.

2004
"Cellular mechanisms underlying swim acceleration in the pteropod mollusk Clione limacina"
by Pirtle and Satterlie
Integr. Comp. Biol. (2004) 44 (1): 37-46. doi: 10.1093/icb/44.1.37

Clione can swim faster when attacking a prey. It had been suggested that 5-HT mediates the acceleration of the swim motor pattern. In this study, the authors examined what membrane current components were modified by 5-HT.
   It seems to me that the experiments were not well designed. The way they compared the effects of drugs seems problematic (e.g., Fig. 7D). Figures were not well arranged. A control swim activity was not presented in Fig. 2 but then shown in Fig. 3. The authors called ion channel blockers as "antagonists." Antagonists are for receptors. No description of what neurons they used. The introduction was great, though.


2006
"The contribution of the pleural type 12 interneuron to swim acceleration in Clione limacina"
by Pirtle and Satterlie
Invert Neurosci (2006) 6:161–168. DOI 10.1007/s10158-006-0029-8

In this paper, the authors examined the role of type-12 neurons in the swim acceleration.
   Type-12 neuron is recruited into the type7/8 half-center network when the CPG gets into the fast swim mode. Arshavsky et al. (1985) suggested that this neuron plays an important role in the swim acceleration by rewiring the circuit. It turned out that activation of type-12 had only a transient effect on the swim cycle frequency. They suggested that type 12 has rather minor roles in maintaining the fast swim mode. Its function may be bilateral coordination or to increase the stability of the network. Despite negative results, this paper is more interesting than the other two.


2010
"A hyperpolarization-activated inward current alters swim frequency of the pteropod mollusk Clione limacina"
Pirtle, Willingham, and Satterlie
Comparative Biochemistry and Physiology, Part A 157 (2010) 319–327. DOI:10.1016/j.cbpa.2010.07.025

I don't see much difference between this and the previous 2004 paper. The only change was that they now show ZD7288 data instead of CsCl. In Fig. 5E. They stated ZD7288 blocked the 5-HT effect. I cannot agree with that. If compared with Fig. 4D data, I see that 5-HT still increases the swim cycle frequency even in the presence of ZD7288. Oh and again they were still calling ZD7288 as an "antagonist" for Ih current.

A simple half-center network oscillator with a twist

My latest paper came out from J Neurophys:

The central pattern generator underlying swimming in Dendronotus iris: A simple half-center network oscillator with a twist  

by Akira Sakurai and Paul S. Katz

DOI: 10.1152/jn.00150.2016

This paper describes the central oscillatory circuit underlying rhythmic swimming of a nudibranch sea slug, Dendronotus iris.

Dendronotus iris swims by rhythmically flexing its body to left and right.

The Dendronotus brain is a cluster of lobes or "ganglia." The neurons that produce the rhythmic motor output for swimming have their cell bodies in the pedal ganglia. They all project the axons toward the other side of the brain to synapse with their contralateral counterparts.  

The circuit is a typical "half-center oscillator" that consists of only two bilateral pairs of neurons. The paired neurons each inhibit their contralateral counterparts.  The circuit has a “twisted” organization; that is, a neuron in one pair is excitatory-coupled contralaterally to a neuron in the other pair.

The Dendronotus swim CPG is a half-center oscillator.  The left illustration shows actual synaptic connections. The left Si3 (L-Si3) neuron forms an excitatory synapse and electrical connection onto the right Si2 (R-Si2), forming a twisted configuration. A modified version is shown on the right. Coupled Si2 and Si3 neurons form a functional unit that works as a half-center to produce rhythmic bursting.

The half-center oscillator is the simplest design of a network oscillator, in which two neuronal elements with no endogenous rhythmicity form reciprocally inhibititory synapses. The half-center theory was first proposed by T. Graham Brown in 1911 after the historical experiment with a walking cat with transected spinal cord.

In addition to the reciprocal inhibition, each functional unit of the half-center oscillator generally contains the excitatory neurons that provide rhythmic excitatory drive onto the mutually-inhibitory neurons (eg., Clione, tadpole, lamprey, zebrafish, and mice). Because of high complexity with so many neurons involved, the role of the excitatory neurons have not been clearly understood.  Here, we found that the Dendronotus swim circuit consists of only 4 neurons. By using "Dynamic Clamping" technique, we manipulated the strength of the excitatory synapse and found that they play crucial roles for the circuit to function as the half-center oscillator.  To our knowledge, this is probably the simplest half-center oscillator described to date. Because of such simplicity, this circuit is also highly manipulable, and hence may provide a good system to study the fundamental properties of a network oscillator. 

Variability, compensation and homeostasis in neuron and network function

Eve Marder and Jean-Marc Goaillard

Hebbian learning can be appropriately balanced by stability mechanisms that allow neurons and synaptic connections to be maintained in appropriate operating ranges (by Turrigiano and Nelson, various mechanisms including synaptic scaling and changes in individual ionic currents).

omeostatic tuning rules that maintain a constant activity pattern could, in principle, operate to tune conductances so that an individual neuron remains within a given region of parameter space, although its values for one or more conductances may be substantially
altered.

Variability in channel densities
How can we reconcile the apparent sensitivity of many neurons to rapid pharmacological treatments with new data indicating that individual neurons within a class can differ by as much as two- to fourfold in the densities of many of their currents?
Computational models show that a number of different compensating combinations of conductances can result in similar activity patterns38,51.

In contrast to pharmacological manipulations, slow mechanisms that function during development and over days and weeks can result in a set of compensating conductances that give rise to a target activity pattern.

Figure 2 | Neurons with similar intrinsic properties have different ratios of conductances.

Figure 3 | Comparison of short-term pharmacological manipulations and long-term genetic deletions.

Slow developmental and homeostatic mechanisms can ‘find’ multiple solutions of correlated
and compensating values of membrane conductances consistent with a given activity pattern, even while rapid pharmacological treatments that vary the value of one current at a time result in altered activity57.

Recovery of Locomotion After Spinal Cord Injury: Some Facts and Mechanisms

Serge Rossignol and Alain Frigon

The model provide...The model has...This model produces...How this model can be approached...What can we learn from it.
This study focuses on...by first describing...Specifically we here show...We propose that...
First figure explains the system we study...How the system functions...We first describe the effects of a procedure...We then discuss the effects of...to establish new interactions for the generation of hindlimb locomotion.

LOCOMOTOR RECOVERY AFTER COMPLETE SPINAL TRANSECTION

Cats with a complete SCI (i.e., spinalization) at the last thoracic segment (T13) gradually recover hindlimb locomotion on a treadmill following a few weeks of locomotor training.

The Inescapable Central Spinal Pattern Generator

In acutely spinalized and paralyzed cats, fictive locomotion can be recorded with pharmacological stimulation (L-DOPA) in the complete absence of overt movement (Grillner & Zangger 1979).
In chronic spinal cats, fictive locomotion can occur spontaneously without drugs, indicating that functional changes have occurred within the spinal locomotor circuitry enabling the spontaneous expression of this endogenous pattern (Pearson & Rossignol 1991).

Functional organization of locomotorgenerating circuits.
The mammalian locomotor CPG is thought to be composed of interconnected modules that coordinate activity around specific joints (Grillner 1981). A multilayered spinal locomotor CPG, in which rhythm-generation and pattern formation are functionally separated, has been proposed to account for some experimental findings (reviewed in McCrea & Rybak 2008).

Spinal localization of locomotor-generating circuits.
For instance, although rhythmogenic properties within the lumbosacral spinal cord are somewhat distributed over several segments, the L3-L4 segments in cats (Marcoux & Rossignol 2000, Langlet et al. 2005, Delivet-Mongrain et al. 2008) and L1-L2 segments in rodents (Cazalets et al. 1995, Kiehn 2006) are critical for rhythm generation. This segmental
heterogeneity has important implications for the recovery of walking after SCI.

A balance between excitation and inhibition.
Function within the spinal locomotor network is governed by excitatory and inhibitory connections. During locomotion, motoneurons receive rhythmic alternating pushpull patterns of glutamatergic excitation and glycinergic inhibition during the active and inactive phases, respectively (Shefchyk & Jordan 1985, Cazalets et al. 1996, Grillner 2003). Excitatory connections are sufficient to drive rhythmic bursting because blocking inhibitory transmission, through GABAA (i.e., bicuculline) and glycine (i.e., strychnine) receptor antagonists, does not abolish oscillatory activity (Kjaerulff & Kiehn 1997, Grillner & Jessell 2009). However, inhibition is necessary to produce appropriate flexion/extension (Cowley & Schmidt 1995) and left-right (Cowley & Schmidt 1995, Kremer & Lev-Tov 1997, Hinckley et al. 2005) alternations (Grillner &Jessell 2009).

Intrinsic properties of central pattern generator neurons.
voltage-dependent persistent inward currents (PICs) that amplify excitatory synaptic inputs and sustain neuronal firing are thought to facilitate rhythmogenesis by timing and shaping locomotor output (Brownstone et al. 1994, Kiehn et al. 1996, Tazerart et al. 2008).

Cellular changes in spinal locomotorgenerating circuits.
Immediately after SCI, the excitability of spinal interneurons and motoneurons is depressed because of the loss of excitatory neuromodulatory inputs from brainstem-derived pathways. The return of neuronal excitability is required for functional recovery and can be mediated by several factors.
Some 5-HT receptors became constitutively active following SCI in adult rats, indicating that intracellular signaling occurred without normal ligand binding. Some receptors can also become supersensitive to remaining endogenous sources of neurotransmitters.
Changes in inhibitory circuits could also play a part in modifying neuronal excitability following SCI; increased levels of inhibitory neurotransmitters (i.e., more inhibition) could depress neuronal excitability and impair specific spinal circuits.
The switch from inhibition to facilitation in adult rats was partly attributed to downregulation of potassiumchloride cotransporter 2 (KCC2) in lumbar neurons (Boulenguez et al. 2010, Boulenguez & Vinay 2009). KCC2 expression progressively decreased within the ventral horn following complete or incomplete SCI, and increased levels of intracellular Cl− diminished the efficacy of synaptic inhibition.

The Key Role Played by Sensory Inputs

Sensory inputs play a key role in the regulation of normal locomotion, which can be altered after SCI. After complete SCI, intrinsic changes at the cellular level of the CPG promote the return of hindlimb locomotion through interactions with peripheral sensory inputs.

Spinal reflexes during locomotion.
Reflex responses are　state- and phase-dependent, indicating that sensory processing is regulated by context or, in other words, the current configuration of the spinal circuitry.
Changes at the cellular level following SCI will directly impact the regulation of peripheral sensory inputs and their interaction with the spinal locomotor CPG.

Changes in spinal reflexes after spinal lesion.
If a small portion of sensory feedback is reduced by lesioning specific peripheral nerves before a complete SCI in adult cats, the recovery of hindlimb locomotion is severely impaired.

LOCOMOTOR RECOVERY AFTER PARTIAL SPINAL LESIONS

After incomplete SCI, spared pathways originating from supraspinal and propriospinal structures can play an active role in the recovery process, and also in restoring some voluntary
control. However, intrinsic spinal circuits and peripheral afferents still remain to initiate and
organize hindlimb locomotion.

Accessing the Locomotor Circuitry by Descending Inputs

Ventral and ventrolateral lesions (reticulospinal and vestibulospinal pathways).

Dorsal/dorsolateral lesions (corticospinal and rubrospinal pathways).

Other pathways.Propriospinal pathways. Propriospinal pathways appear to be of considerable importance for volitional aspects of locomotor recovery.Noradrenergic and serotonergic pathways. The loss of neurotransmitters will in turn
have important consequences on themembran properties of target neurons.

Multiple pathways severed by contusions or hemisections.Contusions. Locomotor recovery did not depend on the sparing of corticospinal or long propriospinal pathways (Basso et al. 1996), indicating a role for short intraspinal circuits.
Hemisections.

Mechanisms of Locomotor Recovery After Partial Spinal Cord Injury

Intrinsic spinal mechanisms and afferent mechanisms are still critical in locomotor recovery after an incomplete SCI. In turn, new interactions can modify spared structures throughout the CNS, not just the spinal cord.
Functional recovery is often thought to result from a combination of regeneration, sprouting, or other ill-defined plastic changes in descending pathways (Cafferty et al. 2008).

Compensation by sensory afferents.
Sprouting of sensory afferents on the lesioned side is prominent and could partly account for the functional recovery of various motor patterns (Goldberger 1977, Helgren & Goldberger 1993).
Sensory feedback is of crucial importance in the recovery process.

Compensation by descending pathways.
New circuits could result from new anatomical connections (new circuits) or from enhanced connectivity(enhancing existing circuits).

Regeneration and sprouting. There is a lack of hard evidence that regenerated lesioned axons induce significant functional improvements because of the small number of regenerating axons.
A critical question is whether regenerated fibers are even functional.
Experiments using staggered spinal hemisections show that the regeneration of long descending pathways is not necessary (Kato 1989).
Locomotor recovery depends more on intrinsic spinal mechanisms and contributions from sensory afferents.

New/old circuits. There is no doubt that the propriospinal systems (long and short) can reach the CPG.
The recovery of hindlimb locomotion after a complete SCI in cats, rats, and mice absolutely requires a spinal circuitry capable of generating the basic locomotor pattern independently of descending commands (Grillner 1981; Rossignol 1996, 2006; Rossignol et al. 2006).

Compensation by the Intrinsic Spinal Circuitry. Recent work on the escape swim of a mollusc (Tritonia diomedea) showed near immediate changes within the functional connectivity of the swim CPG following a lesion within the intrinsic circuitry that compensated for the loss of long projections and reinstated function in the absence of regeneration (Sakurai&Katz 2009). A considerable portion of the recovery could be mediated within spinal circuits rather than by a functional takeover by descending pathways.
We propose that the recovery of function by descending or afferent inputs after SCI essentially depends on how the circuitry has adapted to the total or partial absence of descending inputs.

IMPLICATIONS FOR HUMANS WITH SPINAL CORD INJURY

Organization of spinal circuitry for rodent locomotion

Neuronal activity in the isolated mouse spinal cord during spontaneous deletions in fictive locomotion: Insights into locomotor CPG organization

by Guisheng Zhong, Natalia A Shevtsova, Ilya A Rybak, Ronald M Harris-Warrick

Journal of Physiology (2012)

Deletions are spontaneous errors in the rhythmic locomotor pattern when a set of synergist motoneurons (for example, flexor motoneurons on one side) loses rhythmic firing or falls silent during a time period when they are normally active. 

In the non-resetting deletions, the phase of the rhythm after the deletion did not change. The resetting deletions show rhythm resetting which was recognized by a shift in the phase of the motor bursts after the deletion. This study follows Rybak-McCrea model of the locomotor CPG, which has two functional levels: a half-center rhythm-generator and pattern formation networks. 

Deletions occur simultaneously in motor activity across more than one spinal segment. Spontaneous non-resetting deletions on one side are independent of the other side. The CPG can be functional within an isolated hemisegment. The reduction in locomotor frequency after simulated hemisection results mainly from the elimination of excitatory input to the rhythm-generating ipsilateral RG-F population from the contralateral RG-E population. The resumption of activity an integer number of cyces later does not require input from other parts of the spinal cord. Each hemicord contains an independent rhythmogenic network that can function in the absence of the other hemicord, although the left and right networks are normally coupled via commissural interneurons.

Among V2a interneurons, there are deletion-sensitive types and insensitive type (type I and II V2a interneurons). Type I V2a does not respond to non-locomotory firing in iL2, whereas type II V2a does. Both neurons were depolarized by synaptic drive. The type I V2a interneurons are involved in rhythm generation and/or coordination between left and right networks via the CINs. In contrast, the type II V2a interneurons do not belong to rhythm generator networks, but can be components of the pattern gormation network and/or last-order interneurons that directly project to motoneurons. Commissural interneurons (CINs) send their axons to the opposite side of the cord and coordinate left-right alternation. CINs were not affected by motoneuronal deletion. There was asymmetry in deletion: During all of the flexor deletions, the ipsilateral extensor root showed sustained activity with no interruptions at the times of the missing flexor bursts. In contrast, during all extensor deletions, the ipsilateral flexor root continued unperturbed bursting.

Their computational model combines the Rybak-McCrea concept of the two-level locomotor CPG (Rybak et al., 2006a,b; McCrea and Rybak, 2007,2008) with the Duysens-Pearson concept of an asymmetric rhythm generator with a dominant flexor half-centre (Pearson and Duysens, 1976).

The recovery paper

My latest paper will appear in the Journal of Neuroscience.
October 21, 2009 | Volume 29 | Number 42 |

They will introduce my paper in "This Week in The Journal"
Development/Plasticity/Repair - "Nerve Transection Induces Circuit Reorganization in Tritonia"

The article title is:
"Functional Recovery after Lesion of a Central Pattern Generator"
by Akira Sakurai and Paul S. Katz

In this paper, we found that severing a set of connections between some CPG neurons impaired motor pattern production but that the system spontaneously recovered over the course of a few hours to a day. Furthermore, we observed corresponding changes in synaptic strength that can account for the functional recovery.

Re-expression of Locomotor Function After Partial Spinal Cord Injury

S. Rossignol, G. Barrière, O. Alluin and A. Frigon

The CPG is defined as a spinal network of neurons capable of generating a rhythmic pattern consisting of alternating activity between flexor and extensor motoneurons on the same side with reciprocal activation of homologous motoneurons in the other limb of
the same girdle. In general, during walking or trotting, this network ensures that flexor motoneurons on one side are active with contralateral extensors and vice versa for extensor motoneurons.

...In such a preparation, rhythmic activity, evoked by injecting the noradrenaline precursor l-dihydroxyphenylalanine (L-DOPA), is recorded from peripheral muscle nerves and is termed “fictive locomotion.”

...Indeed, electrical stimulation of a circumscribed brain stem region called the mesencephalic locomotor region (MLR), ... Various gait patterns, such as walk, trot, and gallop, can be
evoked with increasing stimulation intensity.

...Other descending pathways release specific neurotransmitters, which are synthesized by cells in well defined brain stem nuclei (e.g., noradrenaline in the locus coeruleus and serotonin in the raphe and parapyramidal nuclei). These neurotransmitters exert powerful effects on the spinal circuitry and can change characteristics of the locomotor pattern.

...it is important to know that, after a complete spinal transection, most quadruped mammals will
recover some degree of locomotor function in the limbs below the lesion. Cats, rats, and mice can re-express hindlimb locomotion provided the spinal cord below the complete lesion is properly
stimulated, either pharmacologically or through locomotor training.

...locomotion is controlled at multiple levels of the central nervous system, and a subtle and intricate balance is established between these levels of control. This then leads to the question of how an optimal equilibrium is re-established when this exquisite balance is perturbed following lesions of the spinal cord.

...Instead of taking over lost spinal functions, remnant descending pathways or regenerating pathways could direct the reorganization of the spinal circuitry so that it can function optimally and with a greater level of independence so that, after the complete section, the full pattern of hindlimb locomotion can be expressed by an already autonomous CPG.

The main conclusion of this brief review of multiple types of lesions is that there are several ways through which the CNS and peripheral afferent inputs can access the spinal locomotor circuitry. This apparent redundancy points to the fact that the rhythm is generated at the spinal level and that various degrees of control levels can modulate this spinal circuitry through multiple pathways. The removal of certain pathways produces specific locomotor deficits, but the
spinal circuitry and other intact pathways are still able to optimize remnant locomotor functions.

Prominent Role of the Spinal Central Pattern Generator in the Recovery of Locomotion after Partial Spinal Cord Injuries

Grégory Barrie`re, Hugues Leblond, Janyne Provencher, and Serge Rossignol

The general accepted model of locomotor control is tripartite.
1) CPG
2) sensory feedback
3) descending pathways

After partial spinal cord injury (SCI), this optimal balance is perturbed because communication between the brain and the spinal CPG is altered (Barbeau and Rossignol, 1994).

Q: Are there lastic changes within descending pathways?
Or, the spinal CPG retains its function and that changes in descending commands aim at maintaining an optimal control of the spinal locomotor network?

The role of the CPG in the recovery of locomotion after incomplete spinal cord lesion is mostly unknown.

RESULT
The first step consisted of an incomplete section of the spinal cord at the thoracic level T10 or T11.
The second step was a complete transection of the spinal cord at T13 or L1.

1) The recovery of quadrupedal locomotion after partial lesions is mostly the result of an intrinsic reorganization of the spinal locomotor network below the lesion.

2) Locomotor training is a major factor in facilitating the recovery process because cats intensively trained after the partial lesion expressed a very high locomotor
performance bilaterally within hours of the complete spinalization, whereas in untrained animals only a unilateral locomotion was observed in the limb ipsilateral to the partial lesion.

3) Plastic reorganization of the spinal CPG may still occur after the complete spinal section because bilateral locomotor performance improved with training over time in all cats after complete spinalization.

Altogether, this work highlights the importance of promoting spinal neuroplasticity in rehabilitation strategies in SCI patients, especially to maintain the spinal circuitry in an optimal condition to generate locomotion.