ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2016

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Medicine 1272

Deciphering the LocomotorNetwork

The Role of Spinal Cord Interneurons

SHARN PERRY

ISSN 1651-6206ISBN 978-91-554-9741-5urn:nbn:se:uu:diva-305601

Dissertation presented at Uppsala University to be publicly examined in Auditorium Minus,Museum Gustavianum, Akademigatan 3, Uppsala, Friday, 16 December 2016 at 10:15 for thedegree of Doctor of Philosophy (Faculty of Medicine). The examination will be conductedin English. Faculty examiner: Assistant Professor Kimberly Dougherty (Drexel University).

AbstractPerry, S. 2016. Deciphering the Locomotor Network. The Role of Spinal Cord Interneurons.Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Medicine1272. 89 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-554-9741-5.

In the spinal cord, an intricate neural network generates and coordinates the patterning oflimb movements during locomotion. This network, known as the locomotor central patterngenerator (CPG), comprises of various cell populations that together orchestrate the outputof motor neurons. Identification of CPG neurons through their specific gene expression is avaluable tool that can provide considerable insight to the character, intrinsic properties and roleof a population, which represents a step toward understanding locomotor circuit function andcorrelating neural activity to behaviour. We selectively targeted two inhibitory CPG populationsto investigate their molecular characteristics, circuitry and functional role; Renshaw cells (RCs)marked by their specific expression of the cholinergic nicotinic receptor α2 (Chrna2) and asubset of the dI6 population derived by their selective expression of the Doublesex and mab-3related transcription factor 3 (Dmrt3).

We found that RCs have hyperpolarisation-activated cation (Ih) and small calcium-activatedpotassium (ISK) modulatory currents that differentially regulate their excitation and firingproperties, which influence the instantaneous feedback to motor neurons through the recurrentinhibition circuit. Due to previous difficulties isolating RCs from the surrounding locomotorcircuits, their functional role remains poorly defined. For the first time, we selectively silencedRC inhibition and found that all aspects of motor behaviour, including coordination and gaitwere normal. The deletion of RC signalling instead altered the electrical and synaptic propertiesof the recurrent inhibitory circuit, suggesting that developmental plasticity compensates for theloss of RC inhibition.

We reveal Dmrt3 neurons comprise a population of glycinergic inhibitory, spike-frequencyadapting commissural interneurons active during locomotion. Conditional silencing of theDmrt3 population resulted in considerable gait abnormalities in the neonatal and adult mouse.This manifested as an uncoordinated CPG output in vitro, impaired limb coordination in pupsand increased fore- and hindlimb synchrony in adults that was exacerbated at faster locomotorspeeds. Dmrt3 mediated inhibition subsequently impacts locomotion and suggests the Dmrt3population contribute to coordinating speed dependent left-right limb alternation. This thesisprovides cellular, circuit and behavioural insights into the Renshaw cell and Dmrt3 populationsand enhances our knowledge regarding their probable function within the locomotor CPG.

Keywords: Locomotion, central pattern generator, mouse, gait, recurrent inhibition, Renshawcell, Dmrt3, Chrna2

Sharn Perry, Department of Neuroscience, Box 593, Uppsala University, SE-75124 Uppsala,Sweden.

© Sharn Perry 2016

ISSN 1651-6206ISBN 978-91-554-9741-5urn:nbn:se:uu:diva-305601 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-305601)

We have so much time and so little to do.

Strike that. Reverse it.

Willy Wonka

For Balaseca and Gala

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals:

I. Perry S., Gezelius H., Larhammar M., Hilscher M.M., Lamotte d’Incamps B., Leao K.E. and Kullander K. Firing properties of Renshaw cells defined by Chrna2, are modulated by hyperpolarizing and small conductance ion currents, Ih and ISK. 2015. Eur J Neuro-science. Apr; 41(7): 889-900.

II. Enjin A.*, Perry S.*, Nagaraja C., Hilscher M.M., Larhammar M., Gezelius H., Eriksson A., Leao K.E. and Kullander, K. Developmental disruption of recurrent inhibitory feedback results in compensatory adaptation in the Renshaw cell - motor neuron circuit. 2016. The Journal of Neuroscience. In Revision.

III. Perry S.*, Larhammar M.*, Nagaraja C., Hilscher M.M., Tafreshiha A., Potter E., Edwards S.J., Caixeta F.V. and Kullander K. Dmrt3 derived spinal cord neurons regulate locomotor coordination across different speeds. Manuscript.

* Authors contributed equally to this work

Reprints were made with permission from the respective publishers

Other Publications

I. Aresh B., Freitag F.B., Perry S., Blümel E., Lau J., Franck M.C.M and Lagerström M.C. Spinal cord interneurons expressing the gas-trin releasing peptide receptor convey itch through VGLUT2-mediated signalling. 2016. Pain. Submitted.

II. Schnerwitzki D., Perry S., Ivanova A., Caixeta F.V., Cramer P., Günther S., Weber K., Tafreshiha A., Becker L., Schmidt M., Kullander K. and Englert C. The Wilms tumor suppressor Wt1 af-fects spinal circuits responsible for locomotion in mice. Manuscript.

III. Schizas N., Perry S., Andersson B., Wählby C., Kullander K. and Hailer, N.P. Differential neuroprotective effects of Interleukin-1 re-ceptor antagonist on spinal cord neurons after excitotoxic injury. Manuscript.

Contents

Introduction ........................................................................................ 11

Background ......................................................................................... 13 Movement .................................................................................................. 13

Neural Communication ......................................................................... 14 Neurotransmitters, Transporters and Receptors ........................................... 15

Genetic Targeting of Neural Populations .............................................. 17 The Cre-LoxP System .................................................................................. 17

Cortical Control of Movement .................................................................. 18 The Corticospinal Tract ................................................................................ 19

Spinal Control of Movement ..................................................................... 20 The Motor Unit ...................................................................................... 21 Proprioception ....................................................................................... 22

Muscle Spindles ........................................................................................... 23 Sensory Reflexes ................................................................................... 24

Locomotion ............................................................................................... 24 Central Pattern Generators .................................................................... 26

CPG Models ................................................................................................. 29 Spinal Cord Neural Development ......................................................... 29

Developing Spinal Networks ........................................................................ 31 Spinal Cord Neural Populations in Locomotion ................................... 32

Motor Neurons ............................................................................................. 32 Locomotor Related Interneurons .................................................................. 33

CPG Coordination ................................................................................. 34 Left-Right Alternation .................................................................................. 34 Flexor-Extensor Alternation ......................................................................... 36

V1 Interneurons ....................................................................................... 36 Speed Modulation ......................................................................................... 38

dI6 Neurons and Dmrt3 ............................................................................. 39 Recurrent Inhibition .................................................................................. 40

Renshaw Cells ....................................................................................... 42 Clinical Applications ................................................................................. 44

Aims .................................................................................................... 47

Methodological Considerations .......................................................... 48 Animal Models .......................................................................................... 48 Electrophysiology ...................................................................................... 49

Fictive Locomotion ............................................................................... 49 Patch-Clamp Electrophysiology ............................................................ 51

Retrograde Tracing .................................................................................... 53 Behaviour Assays ...................................................................................... 54

Results and Discussion ....................................................................... 56 Paper I ....................................................................................................... 56 Paper II ...................................................................................................... 59 Paper III ..................................................................................................... 63

Summary and Future Perspectives ..................................................... 68

Acknowledgements ............................................................................ 71

References .......................................................................................... 73

Abbreviations

5-HT 5-Hydroxytryptamine or Serotonin ALS Amyotrophic lateral sclerosis AHP Afterhyperpolarisation potential BAC Bacterial artificial chromosome BMPs Bone morphogenetic proteins Chrna2 Cholinergic nicotinic receptor alpha2 ChAT Choline acetyltransferase CIN Commissural interneuron CNS Central nervous system CPG Central pattern generator Cre Cyclic recombinase DA Dopamine DAPI 4’,6-diamidino-2-phenylindole Dmrt3 Doublesex and mab-3 related transcription factor 3 DNA Deoxyribonucleic acid DREADD Designer receptors exclusively activated by designer drugs DRG Dorsal root ganglion FDA Fluorescein-dextran-amine GABA Gamma(γ)-aminobutyric acid HCN Hyperpolarisation-activated cyclic nucleotide-gated channels IaIN Ia inhibitory interneuron Ih Hyperpolarisation-activated cation current ISK Small conductance calcium-activated potassium current MN Motor neurons MN-RC Motor neuron – Renshaw cell synapse mRNA Messenger RNA (Ribonucleic acid) nAChR Nicotinic acetylcholine receptor NMDA N-methyl-D-aspartate PIC Persistent inward currents RFP Red fluorescent protein RC Renshaw cells RCα2 Renshaw cells expressing Chrna2 Shh Sonic hedgehog VAChT Vesicular acetylcholine transporter VGLUT Vesicular glutamate transporter VIAAT Vesicular inhibitory amino acid transporter ZFN Zinc finger nuclease

11

Introduction

… So come on, come on Do the loco-motion with me.

Kylie Minogue, 1988

If you’re not so familiar with the global hit ‘The Loco-motion’, it is an ex-tremely catchy song that persuades the listener, despite their perceived level of dance ability, that they can do the loco-motion easier than doing their ABCs. Now, despite one’s given enthusiasm or reluctance to get up and move, the premise of these lyrics, and the ease and control with which the moves are executed relies on our ability to generate and coordinate an exten-sive range of movements. Regardless of the type of motor action, be it high spirited 80’s dance moves, or perhaps just reaching to turn the music off… every move we make, every slight adjustment in force, direction or speed, is the result of a highly complex and refined series of movement commands continually being delivered to our muscles. Generating movements is an extremely robust and flexible process as we are not only continually aware of our body’s position, but our nervous system is endowed with internal mechanisms that allow for rapid movement adjustments in response to envi-ronmental changes or to the internal state of our muscles.

Movements, no matter how simple or complex, are generated, controlled and refined by networks of neurons in the brain and spinal cord. In the spinal cord alone, a specialised neural network exists that autonomously produces the rhythm and pattern of locomotion that underlies movements including walking, running and swimming. This locomotor network is comprised of populations of neurons that together form a central pattern generator (CPG) to orchestrate the patterned, rhythmical activation of motor neurons ensuring correct and coordinated motor output. Within the CPG, each interneuron population has a critical role in directing and organising motor output.

Motor neurons, as the final common motor pathway, execute the motor commands generated by brain and spinal networks and relay these com-mands to muscles to generate movement. If any part of the motor network breaks down, the information conveyed to the muscles is disrupted and movements become uncoordinated, uncontrolled or in the worst case, para-lysed. Signalling and communication errors within brain and spinal motor control centres can manifest as clinical disorders that include Parkinson’s disease, amyotrophic lateral sclerosis (ALS) and full or partial spinal cord

12

injuries. Understanding the circuitry and communication between locomotor CPG populations, especially the role of each population during movement, is crucial to extend our basic understanding of the locomotor network and iden-tify future treatments and therapeutic targets for motor disorders. This thesis, Deciphering the Locomotor Network, focuses on the spinal con-trol of movement and how individual neural populations underlie, generate, control and tailor specific movement patterns. In particular, this work reports on two interneuron populations investigated in the mouse spinal cord; Dmrt3 neurons and Renshaw cells that interestingly represent the discovery of one of the most recent (Dmrt3) and one of the first (Renshaw cell) spinal popula-tions. The juxtaposition of these populations highlights the significant tech-nical and genetic advancements that have propelled locomotor research in recent years, which now allow the selective and specific manipulation of defined populations to uncover their functional role.

13

Background

Movement During development, babies progressively learn to replace rudimentary step-ping and reaching actions with more complex movements like running, skip-ping and catching. Over time, and with insurmountable practice, they even-tually learn and develop an extensive repertoire of basic and skilled move-ments that they can execute with precise control, many times over, with little conscious thought. These actions encompass everything from the reflexive rapid, physical withdrawal to pain to complex movement sequences like those occurring during tennis that depend highly on physical aptitude, coor-dination, timing and sensory input.

Involuntary movements or reflexes, are generated through often deceiv-ingly simple, local spinal cord reflex pathways that lead to rapid, stereotypi-cal and reproducible motor responses to stimuli. Voluntary movements, of-ten significantly more complex then reflexes, are consciously controlled demanding a higher level of regulation and coordination, and therefore re-quire pathways linking the brain to the periphery.

To execute skilled movements quickly, the nervous system stores motor information as a series of pre-programmed muscle commands known as motor programs, that are ready to be selected at will. Information relating to movement patterns, the external environment and previous motor experienc-es are coded into the motor programs, which are then selected, initiated, modified and executed by motor command centres in the brain and spinal cord (Morris et al., 1994).

The highly organised series of descending motor commands are relayed from the brain to skeletal muscles via spinal motor neurons that form the final motor command pathway. These spinal motor neurons integrate brain, peripheral and spinal inputs to coordinate the timely contraction and relaxa-tion of skeletal muscles, ultimately producing a desired movement.

In order to understand how neural networks produce coordinated move-ment, it is first necessary to consider how neurons communicate to reliably transmit, modify and propagate movement signals throughout local circuits and peripheral and cortical pathways.

14

Neural Communication Neurons form vast electrical networks that have the capability to rapidly transmit signals throughout the central nervous system (CNS). Two neural features facilitate this transmission; a specialised anatomy that primes the neuron to receive and send electrical signals and the formation of chemical and electrical synapses on surrounding neurons to relay signals throughout the network.

Incoming information is received through neural dendrites, extensive ar-borisations that project from the cell body, and transduced into an electrical impulse or action potential (Hodgkin and Huxley, 1952) sent along the neu-ral axon to the axon terminal. At the axon terminal, the nerve impulse travels across a synapse (see Neurotransmitters, Transporters and Receptors) and is relayed to downstream cells. Neurons communicate by generating trains of action potentials at certain frequencies in response to the strength, duration and type of input they receive. Thus, neurons encode information in the tim-ing and firing rate or frequency of their action potentials. The equivalent activation of individual neurons within a population can elicit vastly differ-ent action potential firing patterns, which attests that neurons have different intrinsic coding properties (Dougherty and Kiehn, 2010a; Prescott et al., 2008; Zhong et al., 2010; Zhu et al., 2012). As such it is possible to identify and discriminate subsets of neurons within a population based on their action potential firing frequency and firing pattern. One particular feature of neural firing, common to neurons involved in rhythmic circuits, is spike-frequency adaptation. Spike-frequency adaptation is the phenomenon in which a neu-ron has a reduction in firing frequency after an initial increase, recruiting an internal means of modifying its own frequency. Spike-frequency adaptation has been reported in spinal locomotor populations (Borowska et al., 2013; Brownstone et al., 2011; Dougherty and Kiehn, 2010a; Miles et al., 2005; Zhang et al., 2008) as such we investigated whether the spinal cord popula-tions investigated in Papers I and III had spike adaptation as a firing fea-ture.

Neurons have inherent differences within their electrical properties that underlie and permit diverse firing patterns. Neural membranes are laden with ion channels, membrane pores that allow the flow of ions (current) to change cellular membrane potentials in stereotypical ways. Ion channels are funda-mental to the electrical signalling capabilities of neurons and underlie both the generation of the action potential and the transduction of chemical sig-nals into electrical messages (see Neurotransmitters, Transporters and Re-ceptors). The expression of different types, compositions and variants of ion channels, activated and inactivated by different stimuli, individualise firing frequencies and tailor the neuron’s response to inputs. Pharmacologically blocking ion channels modulates both channel activity and conductance, making it possible to isolate and examine individual currents. Papers I and III, investigated the following ion channels and explored their singular cur-

15

rent (I) contributions to membrane excitability, action potential generation and firing frequency; the hyperpolarisation-activated cyclic nucleotide-gated channels (HCN, Ih, Pape, 1996) and the small conductance calcium-activated potassium channels (SK, ISK, Deardorff et al., 2013).

Neurotransmitters, Transporters and Receptors Neurons employ two mechanisms to communicate and transmit neural sig-nals to downstream neurons in robust and dependable ways; electrical and chemical synapses. Adjacent neurons can be physically connected through specialised connections known as gap junctions. As an electrical thorough-fare, gap junctions permit the direct transfer of electrical signals between two neurons in extreme proximity. Chemical synapses, considerably more complex than electrical synapses, permit neural communication via a process called synaptic transmission, which involves the transmission of specialised molecules called neurotransmitters between pre- and postsynaptic neurons (Figure 1). Chemical synapses are the predominant form of signal transmis-sion within the CNS, although some spinal cord neural populations com-municate via both chemical and electrical synapses (Song et al., 2016; Zhong et al., 2010).

Neurotransmitters, synthesised in the cell body, axon or axon terminal, are packaged and maintained in synaptic vesicles by transporters lodged in the vesicular membrane. In the spinal cord, glutamate packaged into vesicles by the vesicular glutamate transporters (VGLUTs; Shigeri et al., 2004), me-diates excitatory neurotransmission. Inhibitory neurotransmission occurs through the independent or co-release of γ-aminobutyric acid (GABA) and glycine, packaged into vesicles by a common vesicular inhibitory amino acid transporter (VIAAT, Aubrey, 2016) that supports the co-transmission of both molecules across a synapse (Figure 1). Modulatory neurotransmitters, such as acetylcholine packaged by vesicular acetylcholine transporters (VAChT, Lawal and Krantz, 2013), can either be excitatory or inhibitory depending on their mechanism of action on the postsynaptic membrane. Within locomotor circuits, acetylcholine serves vital functions in shaping the motor command and in generating movement, as motor neurons signal to both local spinal circuits and muscles via cholinergic transmission.

Neurons can synthesise, store and co-release multiple neurotransmitters in combination with neuromodulatory molecules such as neuropeptides, which act to enhance, inhibit or further modulate neurotransmission at a synapse broadening neural functionality and increasing the complexity of neural communication (El Mestikawy et al., 2011; Kupfermann, 1991). Transport-ers serve critical functions in neural communication where neural signalling is severely affected if transporters are blocked, abolished or malfunction. The central function of VIAAT in spinal cord locomotor populations has been exploited in Papers II and III.

16

Figure 1. Synaptic transmission. GABA (orange) and glycine (blue) in the presyn-aptic terminal are packaged into synaptic vesicles via the vesicular inhibitory amino acid transporter (VIAAT, grey). When the vesicle fuses with the membrane GABA and glycine are released into the cleft to diffuse and bind to their respective iono-tropic receptors on the postsynaptic membrane. As inhibitory neurotransmitters, the GABA/glycine-receptor interaction opens chloride (Cl-) channels to conduct a chlo-ride current (green) that hyperpolarises the postsynaptic cell.

Depolarisation, due to the axonal propagation of action potentials to the axon terminal, causes vesicles fuse with the presynaptic membrane and release their contents. Once released, neurotransmitters diffuse across the synaptic cleft and bind to ionotropic or metabotropic receptors on the postsynaptic membrane that differ in their mechanism of transduction and time course. The neurotransmitter –receptor interaction opens ion channels that ultimately depolarise (excitatory, linked to a cation channel) or hyperpolarise (inhibito-ry, linked to a chloride channel) the postsynaptic membrane, which make the postsynaptic neurons more or less susceptible respectively, to firing action potentials (Figure 1). Glutamate and glycine bind exclusively to ionotropic receptors, whilst GABA and acetylcholine bind to both ionotropic and metabotropic receptors (reviewed in Purves et al., 2012).

Spinal locomotor interneuron populations are largely glutamatergic or GABA-/glycinergic, but their communication within the locomotor network can be triggered and influenced by modulatory neurotransmitters including N-methyl-D-aspartate (NMDA), dopamine (DA) and serotonin (5-HT). To impact, alter or elucidate the actions of neurotransmitters at the postsynaptic membrane, receptors can be pharmacologically manipulated. We have used agonists of NMDA, dopamine and serotonin receptors (Papers II and III) and antagonists of GABAA receptors and nicotinic acetylcholine receptors (NAChRs, Paper II), to investigate the functions of spinal neural popula-tions within the locomotor circuit. Additionally, we have used the specific and restricted expression of the cholinergic nicotinic receptor subunit α2

Cl- >

>GABAA Receptor

Glycine Receptor

Cl-

VIAAT

Glycine

GABA

Presynaptic Cell

Postsynaptic Cell

Cl-

Cl-

17

(Chrna2) as a genetic marker to selectively identify, target and manipulate Chrna2 expressing interneurons (Paper I and II).

Genetic Targeting of Neural Populations To understand the generation of movement, it is beneficial to identify and study each neural population within the locomotor network to decipher how population based activity generates behavioural output. In the spinal cord this has been challenging, due to the difficulty identifying individual neural populations from surrounding networks. Significant progress in identifying spinal locomotor interneuron populations has come from the generation of a number of genetic, molecular and developmental tools. The identification and use of tailored genetic markers specifically expressed by neural popula-tions, permits the creation of targeted, unique transgenic models where nor-mal gene function has been disrupted through the introduction of a foreign gene (Jaenisch, 1988). Subsequent genomic manipulations of the targeted populations include gene knockouts, where the gene is removed from the genome, or conditional knockouts where gene regulation is temporally or spatially controlled such as cell silencing or cell ablation (Sauer and Henderson, 1988).

The Cre-LoxP System The development of the Cre-LoxP system has revolutionised genetic manip-ulations and the study of specific neural populations. This two-part system conditionally targets genes in a region specific manner and, in this thesis, facilitates the manipulation of the mouse genome to elucidate the functional role of neural populations within locomotor networks. The Cre-LoxP system is fundamentally based on cellular specific expression and activity of cyclic recombinase (Cre), which causes genetic deletions, insertions, translocations and inversions at specific LoxP sites inserted in cellular DNA (Sauer and Henderson, 1988). Cre mice, most often generated by the random insertion of the Cre transgene into the genome (transgenesis), are mated with floxed mice, where the gene of interest is flanked by LoxP sites, to produce condi-tional knockout mice. Cre is inserted under the control of a known, tissue specific, inducible genetic reporter activated during development, whilst identical LoxP sites are placed around the gene to be removed. Once activat-ed, Cre binds to and causes recombination of the LoxP sites, excising the gene of interest and subsequently pasting the genome back together without the excised DNA sequence (Figure 2, Wang, 2009).

18

Figure 2. Generation of transgenic mice using the Cre-LoxP system. Cre is in-serted into the genome under the control of a cell specific promoter (A). The target gene to be manipulated is flanked by the insertion of LoxP sites (arrows, B). When Cre mice are mated with LoxP mice, Cre excises the target gene in Cre specific populations (C).

The Cre-LoxP system is an extremely useful and powerful tool to efficiently and effectively investigate the functional consequences of removing or ma-nipulating specific genes within a population. The ability to cross Cre lines with various reporter lines, such as the tdTomatoRosa26 reporter strain (Madisen et al., 2010), allows Cre specific populations to be visualised. Cre activity excises a LoxP flanked STOP signal, which turns on the expression of a reporter gene where, in the case of the tdTomatoRosa26 mice, red fluo-rescent protein (RFP) is expressed (Figure 2). Moreover, the Cre-LoxP sys-tem is readily used to delete or modify cell signalling components rendering the population silent, or to entirely ablate the Cre expressing population dur-ing development.

This thesis used the Cre-LoxP system to visualise spinal interneuron pop-ulations (Papers I - III) and to study the behavioural and circuitry conse-quences when VIAAT-mediated signalling capabilities of these populations were removed (Papers II and III).

Cortical Control of Movement There are five distinct but highly integrative and interactive brain systems governing movement; the cerebral cortex, basal ganglia, cerebellum, brain-stem and spinal cord (Figure 3). Communication between the brain and spi-nal motor neurons is vital to initiate, direct, plan and adjust complex move-ments. These systems, each with specialised functions, communicate through descending pathways and local neural circuits to regulate and control muscle coordination necessary for organised movement.

Movement occurs within a dynamic environment, requiring internal and external cues and stimuli to be continually incorporated into the ongoing movement pattern. Within the cerebral cortex, the cortical motor systems that select and execute appropriate motor programs integrate sensory feed-back into the occurring motor plan by contacting diverse spinal interneurons,

>

Cre>

Cell specific promoter

Target gene

LoxP LoxP

Target gene

Cre

Target gene excised in Cre specific tissue

+ Target gene

A B C

19

creating an interconnected sensorimotor system that minimises motor dis-crepancies due to environmental changes (Harrison and Murphy, 2014; Moreno-López et al., 2016).

The cortex works in conjunction with other brain areas where cortical mo-tor output is modulated through continuous, sensory, basal ganglia and cere-bellar feedback mechanisms for the effortless, coordinated, well timed exe-cution of movements. The basal ganglia, a group of deep brain nuclei, facili-tate the selection and initiation of appropriate motor actions, whilst simulta-neously supressing unwanted actions. The cerebellum at the base of the brain (Figure 3), ensures that planned and initiated motor actions occur in a coor-dinated, smooth fashion, functioning to calibrate fine movements and cor-rect, adjust and attenuate motor deviations.

All descending motor commands pass through the brainstem, a small, yet vital structure, physically connecting the brain and spinal cord (Figure 3). Brainstem circuits facilitate signal transmission throughout the CNS and work in cooperation with voluntary motor cortex commands to regulate mo-tor output. The brainstem is the origin of motor pathways that use sensory information from the muscle and visual environment to maintain balance, posture, position and gaze, all necessary for supporting skilled motor behav-iour. Whilst voluntary movement is under direct cortical control, autonomic movements including breathing and the initiation and halting of locomotion are under brainstem control (Bouvier et al., 2015; Hägglund et al., 2010).

Descending pathways established between the brain, efferent motor neu-rons and local spinal interneurons work together with cortical and spinal circuits as a functional network to produce movement. Motor signals are transmitted through tracks of nerve fibres traversing the spinal cord white matter, linking the spinal cord with the brain. Descending tracts are charac-terised and organised by neuroanatomical, neuropharmacological and neu-roinformatic features and each have a specific spinal cord target and func-tional role (Lemon, 2008). Together these systems, known as the corticospi-nal, propriospinal, rubrospinal and reticulospinal tracts, facilitate the transfer of descending motor commands to spinal cord targets, maintaining con-trolled and integrated motor function. Although general principles underlie the neural organisation of these tracts, species-specific differences in de-scending pathways reflect specialised functionality and sensorimotor behav-iours (Lemon, 2008; Purves et al., 2012)

The Corticospinal Tract The corticospinal tract, originating in the pyramidal cells of the motor cor-tex, is the major descending pathway carrying motor information from the cortex to multiple dorsal and ventral spinal cord targets. The axons of these ‘upper’ motor neurons descend through lower brain structures including the thalamus, midbrain and medulla before dividing to form the lateral (contrala-teral) and ventral (ipsilateral) corticospinal tracts. The lateral tract forms a direct corticospinal pathway, where fibres terminate extensively within the

20

dorsal and ventral horns to establish direct or indirect via premotor interneu-rons, cortical connections with motor neurons. Ventral corticospinal fibres terminate ipsi- or bilaterally in the ventral horn. Corticospinal neurons there-fore have a dual function, activating and inhibiting motor and sensory neu-rons, to facilitate sensorimotor integration. The diverse spinal targets of the corticospinal fibres, infer that the corticospinal tract modulates sensory func-tions in addition to the direct control of motor circuits (Lemon, 2008; More-no-López et al., 2016; Purves et al., 2012).

The loss of functional corticospinal connections, often through lesions or damage to the spinal cord, results in significant movement disturbances, abnormalities and paralysis. Current investigations into the neural conse-quences of corticospinal tract lesions and motor disorders aim to understand and explore both the fate of the motor circuitry and the plasticity within local and corticospinal pathways that often leads to partly restored motor function (see Clinical Perspectives; Blanco et al., 2007; Han et al., 2015; Nakagawa et al., 2015; Siegel et al., 2015).

Spinal Control of Movement As the central messenger passageway connecting the brain with the periph-ery, the spinal cord runs within the vertebral canal from the base of the skull (rostral) to the first lumbar vertebra (caudal). Along this rostro-caudal axis the spinal cord is segmented into cervical, thoracic, lumbar and sacral seg-ments (Figure 3), from which bundles of dorsal and ventral nerves emerge forming the dorsal and ventral roots. Within the spinal tissue, the grey matter homing populations of spinal interneurons, is divided into the dorsal and ventral horns that are further classified into laminae based on the size, shape and density of neurons (Figure 4, Rexed, 1952, 1954). These laminae are of importance when categorising neural subtypes, peripheral inputs and signal projections involved in the transmission and integration of sensory pathways and the processing and execution of locomotor output.

Figure 3. Schematic of the mouse central nervous system. The spinal cord is segmented along the rostro-caudal axis where bundles of nerves (roots) emerge from each segment.

Brain

Brainstem Spinal Cord

Cervical Thoracic Lumbar Sacral

Cerebellum

Rostral Caudal

21

Dorsal neural populations within laminae I-VI represent the first stage of central sensory signal processing, acting as a relay station to transmit senso-ry signals. Primary afferent fibres constitute the main source of sensory input and transmit sensory information from the periphery to the spinal cord dorsal horn through the dorsal root, where sensory fibres relaying muscle and joint information terminate in lamina VI, VII and IX (Brown, 1982). Dorsal inter-neuron circuits process and modulate sensory information from primary afferents to be incorporated into motor commands or transmitted to the brain for the integration and perception of sensory stimuli. The proper functioning of these neural circuits is crucial as it is widely accepted that the symptoms of chronic pain, hyperalgesia and allodynia, are in part, attributed to the dys-function of dorsal horn neurons (Sivilotti and Woolf, 1994; Yaksh, 1989; Zeilhofer et al., 2012).

Where the dorsal horn represents the first stage of sensory processing, the ventral horn (laminae VII-IX) initiates, modulates and shapes the final out-put for movement. Ventral horn interneuron populations receive, respond and integrate a myriad of descending, local and peripheral inputs for a cohe-sive motor signal to muscles. The most prominent feature of the ventral horn is the large clusters, or nuclei, of motoneurons located throughout lamina IX, whose axons leave the ventral horn through the ventral root to innervate skeletal muscle (Figure 4).

The Motor Unit Motor units are the basis of the neural control of movement and underlie the functional relationship and dependence between muscle fibres and spinal motor neurons. The motor unit, a term first described by Charles Sherrington (Sherrington, 1906, 1925) defines the innervation of a group of muscle fibres by the axon of a motor neuron. As movement is initiated, executed and re-vised, motor units are sequentially selected, recruited and de-recruited into contractions as needed based on the outgoing motor command.

Motor units adhere to a strict organisation that function to regulate muscle contractions. During development, motor neurons that innervate the same muscle fibres are collectively grouped along the spinal cord into distinct longitudinal columns of motor pools. Muscle fibres are strictly innervated by one motor neuron that simultaneously innervates multiple (up to 1000) mus-cle fibres distributed across the muscle to evenly spread contractile force. Moreover, muscle fibres and motor neurons collectively show a diversity of size, which is reflected in the motor unit where distinct motor units are de-fined by the size of the motor neuron, the size and type of the innervated muscle fibre and the number of innervated fibres (Burke, 1981; Eccles and Sherrington, 1930; Miles, 1994; Purves et al., 2012).

Motor units are recruited (Sherrington, 1906, 1925) into movement through the activation of motor neurons in a highly reproducible, sequential

22

pattern that is coupled with the preferential recruitment of slow muscle fibres before fast muscle fibres. The stereotypical, size ordered activation of motor units allows contractile force to be regulated; smaller, prolonged contrac-tions activate primarily small motor units but larger motor units will be se-quentially recruited as voluntary contractions become more powerful (re-view see Burke, 1981; Miles, 1994; Purves et al., 2012).

The electrical properties of motor neurons vary with motor neuron size and influence when motor units will be recruited into contractions. This re-cruitment is believed to be largely dependent on input resistance which de-termines the amount of excitation a motor neuron needs to be recruited; smaller neurons have high input resistances requiring less excitation to be activated, as such small motor units are recruited earlier than larger motor units (review see Miles, 1994; Purves et al., 2012). Recent evidence in zebrafish offers a novel modular organisation of motor neuron circuits link-ing excitatory premotor neurons to corresponding groups of motor units (Ampatzis et al., 2014). These premotor neuron–motor unit microcircuits, segregated into slow, intermediate and fast motor unit pools, are sequentially recruited into movement based on their electrical properties that are largely unrelated to input resistance (Ampatzis et al., 2014). This distinct innerva-tion forms discrete modules that are activated during specific movements. A modular motor unit organisation whilst suggested, has not been well sup-ported in the mouse spinal cord, however, recent findings suggest a modular organisation of mouse locomotor circuits where gait could be dependent on the activation of different circuitry modules (Bellardita and Kiehn, 2015).

Proprioception We are constantly aware, even if we are blindfolded or upside down, of where and how our bodies are positioned in space. This enables us to per-form detailed, precise and skilful voluntary movements in many circum-stances and environments as we have the extraordinary capacity to shape, modify and reselect our movements in response to continual environmental changes and physical obstacles. Primarily this is due to a critical, specialised component of our somatosensory system called proprioception that enables us with an inherent awareness of our body position. Proprioception is an intrinsic sense of self-responsiveness that is set apart from classical senses such as sight or hearing, as it reports the internal state of the body, rather than conveying information from the external world.

Proprioception, first described by Sherrington (1906) as the sense of body position, is the sensory information derived from the skin, joints and muscles that communicates the relative movement of our body parts in space. Spe-cialised sensory organs embedded in skeletal muscle known as muscle spin-dles, continuously relay fast, accurate and detailed information about the instantaneous state of the muscle where information pertaining to muscle length, tension and position is transmitted to the spinal cord via sensory

23

neurons (Windhorst, 2007). These proprioceptive signals are processed through spinal cord networks for rapid movement adjustments and are re-layed to the brain, via the cerebellum, for central processing and to regulate the timing of muscle contractions.

Figure 4. The spinal cord lamination and stretch reflex. A schematic of a transected spinal cord illustrating the grey matter lamina divisions (I-X), dorsal and ventral roots and dorsal root ganglion (DRG). The proprioceptive stretch reflex is depicted showing the innervation of muscle spindles by group 1a afferents and their reciprocal innervation of alpha motor neurons (MN) and 1a inhibitory interneurons (IaIN, blue) in the ventral horn; + excitation, - inhibition.

Muscle Spindles Intrafusal muscle fibres are the sensory encoding, rather than force produc-ing, component of skeletal muscle and are the foundation of the muscle spin-dle, a specialised organ central to relaying proprioceptive input (Hunt, 1990). The muscle spindle comprises of intrafusal fibres innervated by sensory af-ferent neurons that have their receptor endings coiled around (primary, Group Ia afferents) or embedded within (secondary, Group II afferents) the fibres (Figure 4). This unique innervation relays specific and instantaneous feedback about different overlapping modalities of muscle stretch and length where Ia afferents relay the velocity of muscle stretch, whilst Group II affer-ents relay the degree of muscle stretch (Windhorst, 2007). Stretch related mechanical force, i.e. stretching the muscle and muscle fibre, activates the mechanosensitive Ia and group II sensory receptors. This local activation triggers a feedback signal, transmitted via afferent fibres to the spinal cord, for incorporation into the ongoing motor plan. Muscle spindles have both sensory and motor elements, since intrafusal fibres are innervated by special-ised (gamma) motor neurons that regulate the sensitivity of muscle fibres to stretch, ascertaining a continuous relay of proprioceptive feedback to the spinal cord.

IIIIIIIV

V

VII

VIIIIX

XVI

Dorsal

Ventral

++

-

1a Sensory Afferent

Muscle Spindle

MN

DRG

IaIN

Antagonist Muscle

Agonist Muscle

24

Proprioceptive input is not required for the initiation of directed movements but as a central component of movement processing, acts as continuous, necessary feedback to guide, manage and time skilful voluntary actions. Without sufficient proprioceptive inputs, movements become erratic and jerky (Lajoie et al., 1996).

Sensory Reflexes Reflexes are involuntary, instantaneous and highly stereotypical movement reactions to sensory stimuli that are generated purely within the spinal cord (Sherrington, 1906, 1925). Reflexes can be monosynaptic, where sensory neurons directly target effector neurons or polysynaptic, where sensory neu-rons activate interneurons that subsequently innervate the effector neurons. Often, reflexes are not absolutely one or the other and frequently involve both monosynaptic and polysynaptic pathways.

Many motor reflexes rely on proprioceptive feedback from muscle spin-dles to initiate the reflex reaction. When a muscle is stretched, activated muscle spindles transmit proprioceptive feedback to the spinal cord via Ia sensory afferents. These Ia afferents are monosynaptically connected to mo-tor neurons innervating agonist musculature where direct proprioceptive input activates the motor neurons to contract the muscle. Ia afferents also target Ia inhibitory interneurons (IaINs) positioned upstream of motor neu-rons, which innervate and inhibit antagonistic motor neurons that relax op-posing muscles. The simultaneous activation (contraction) and inhibition (relaxation) of muscles pairs around the same joint is called reciprocal inhi-bition (Figure 4).

Ia afferents follow a strict innervation structure strongly innervating either motor neurons that activate agonist musculature or Ia interneurons that sub-sequently inhibit antagonistic motor pools. Peripheral signals help to main-tain the strength of these afferent synapses ensuring functional reflex circuits are maintained throughout development (Mentis et al., 2010).

The stretch reflex response and subsequent interneuron populations in-volved in the monosynaptic and polysynaptic stretch reflex pathways are investigated in Papers II and III.

Locomotion In its most basic definition, locomotion is the act of self-motion or propul-sion and in bipeds and quadrupeds, is comprised of rhythmical movements of limb flexion and extension. When a limb is in contact with the ground it extends and propels the animal forwards. The limb is then subsequently lift-ed from the ground by an act of flexion and reset ready to repeat the cycle (Brown, 1911). This innate, rhythmical, stepping behaviour is a fixed motor

25

pattern present from birth, as neonates exhibit rhythmical step-like reflexes prior to weight bearing age (Robinson and Goldberger, 1986).

Locomotion results from the interplay between the biomechanical con-straints of the muscles and the neural circuits underlying muscular control. Locomotor movements are characterised by repetitive oscillatory bursting of alternate flexor-extensor and left-right motor neurons that activate limb mus-cles resulting in a step cycle. Locomotion is both robust and highly flexible, as locomotor sequences must adapt to changes in internal and external envi-ronments adjusting speed, force and coordination. The locomotor machinery is therefore a hard-wired circuit generating repetitive step cycles that are constrained by physical limitations such as stride length, but can be modu-lated by extraneous sensory input.

A step cycle can be broken down into stance; designating the period the limb is in contact with the ground and swing; the period the foot remains airborne for advancement. Changes to stance and swing phases cause varia-tions to an animal’s gait, the locomotor patterning of the limbs, which in-creases or decreases the velocity of locomotion. Thus, at different locomotor speeds, animals use different gaits (Bellardita and Kiehn, 2015; Lemieux et al., 2016).

Most quadrupeds including horses, cats and rats have a substantial reper-toire of locomotor gaits, however, there has been surprisingly little evidence detailing the capabilities of mice to move at high speeds (gallop or bound) like their other quadruped counterparts. The full locomotor gait repertoire of the mouse during overground and treadmill locomotion has recently been established (Bellardita and Kiehn, 2015; Lemieux et al., 2016) and is far more extensive than first thought. Mice have three distinct gaits that emerge at different speeds; walking, trot and bound (Bellardita and Kiehn, 2015) where mice showed a preference to trot (fore- and hindlimbs move in diago-nal synchrony) at walking speeds and bound (fore- and hindlimbs in syn-chrony) at higher speeds (Lemieux et al., 2016). Between these two predom-inant gaits mice shifted between transitional gaits including gallop and half-bound in a predictable sequential manner, where gait switches occurred ab-ruptly or with minimal transitional steps (Bellardita and Kiehn, 2015; Lemieux et al., 2016). These biological observations have recently been supported by a computational model that recreated the shift in locomotor gaits of the mouse, from left-right alternation (walk and trot) at lower speeds, to left-right synchrony (gallop and bound) at higher speeds (Danner et al., 2016).

Humans (bipeds) predominantly shift between five natural gaits that range from walking at lower speeds to sprinting at higher speeds. Quadrupeds, in particular horses and cats, are able to perform three natural gaits; walk, trot and canter/gallop; however some species, in particular Icelandic horses, are capable of performing additional gaits. These additional gaits are integrated as part of the natural movement and are not to be misconstrued as abnormal gaits. Gait abnormalities, in both humans and quadrupeds are often a conse-

26

quence of underlying genetic or physiological conditions (Andersson et al., 2012; Bellardita and Kiehn, 2015; Guertin, 2012; Kullander et al., 2003; Lerer et al., 2005). Observations of gait can therefore provide critical infor-mation about possible musculoskeletal, neurological and genetic conditions underlying locomotion and assist in deciphering the neural circuitry generat-ing locomotor patterns. The spinal cord neural circuitry supporting locomo-tion is investigated in Papers II and III, where the concept of speed-dependent gait changes and the subsequent neural circuitry underlying these changes is considered in Paper III.

Central Pattern Generators Locomotion was initially suggested to be the result of a subset of rhythmi-cally alternating flexor-extensor reflexes. This was shortly overruled as cats under the effect of deep anaesthesia, at which all peripheral reflexes are abolished, exhibited spontaneous hindlimb stepping movements (Brown, 1911). Reflexive stepping remained present when peripheral input was sev-ered, suggesting locomotion does not derive from purely reflexive or sensory circuits. In the de-afferented, decerebrate, mesencephalic cat devoid of de-scending, peripheral and proprioceptive spinal inputs, locomotion was ob-served to be normal (Brown, 1911; Grillner, 1975). Thus, it was proposed that the neural components generating the locomotor rhythm must be con-tained within the spinal cord and peripheral and descending inputs modulate, select and shape locomotive outputs rather than generating the inherent rhythm (Grillner, 1975). These intrinsic, neural networks produce a rhythmi-cal, patterned output in absence of sensory feedback.

Coordinated, stereotypical patterns are generated by networks of neurons known as central pattern generators (CPG). These networks generate and sustain the internal rhythmic drive behind oscillatory behaviours including respiration and locomotion and serve two basic functions; rhythm generation and pattern formation. The spinal cord locomotor CPG is a tightly integrated, highly complex circuit of local interneurons that modulate and transmit lo-comotor signals, and motor neurons that execute the final CPG command. As a network, the locomotor CPG repetitively activates the correct motor pools with spatial and temporal precision without the need for cortical inputs or peripheral feedback. Whilst interneurons are integral elements of the CPG, it is speculated as to whether motor neurons are actively included within the CPG or passive recipients of CPG commands generated by up-stream interneuron circuits. Recently, motor neurons were identified as inte-gral components of the zebrafish locomotor network, as they selectively control the firing and recruitment of an upstream excitatory interneuron pop-ulation via electrical gap junctions (Song et al., 2016).

Localising the anatomical position and distribution of locomotor CPG neurons began in cats (Grillner and Zangger, 1979) and continued in the isolated rodent spinal cord preparation (Bertrand and Cazalets, 2002; Bonnot

27

et al., 2002; Cazalets et al., 1995; Kjaerulff and Kiehn, 1996). Specific tran-sectioning or lesioning of the isolated spinal cord along both the rostro-caudal and dorso-ventral axis elucidated the origin of the rhythmic, patterned CPG drive. These experiments confirmed the core CPG components are distributed through the ventral laminae (Cina and Hochman, 2000; Tresch and Kiehn, 1999) where the neurons necessary for generating locomotor rhythmicity were localised and distributed within the rostral lumbar spinal cord (Bertrand and Cazalets, 2002; Cazalets et al., 1995; Grillner and Zangger, 1979), whilst circuits coordinating left-right alternation were dis-tributed rostro-caudally along lumbar segments (Kjaerulff and Kiehn, 1996).

CPG neural circuits in each half of the spinal cord are proposed to be driving and controlling the three underlying features of locomotion; the rhythm, the coordination of flexor and extensor muscles and the coordination of left and right limbs (Kiehn, 2006; Figure 5). Each limb is suggested to be controlled by a separate CPG that simultaneously modulates left-right and flexor-extensor motor output, as cats can walk on split-belt treadmills run-ning at different speeds (Frigon et al., 2013). The complexity of the locomo-tor CPG network renders it challenging to distinguish between neurons gen-erating the rhythmic drive and neurons passively driven by the CPG. It is widely accepted that locomotor rhythm generation is initiated and driven primarily by spinal excitatory interneurons that set the baseline locomotor frequency (Hägglund et al., 2010, 2013; reviewed in Kiehn et al., 2008, 2010). The critical role of spinal excitatory neurons in locomotor rhythm generation was reinforced when the targeted activation of Vglut2+ glutama-tergic neurons was sufficient to generate a normal locomotor-like rhythm, whilst the selective inhibition of glutamateric interneurons halted the ongo-ing rhythm (Hägglund et al., 2010, 2013). A specific subset of ventral excita-tory interneurons has recently been identified and implicated as part of the rhythm-generating core (Dougherty et al., 2013). In direct contrast, locomo-tor-like activity was generated in isolated spinal cords deficient in glutama-tergic signalling (Vglut2 null), suggesting that whilst excitatory neurons may be involved in generating the rhythm, another neurotransmitter system, namely the remaining inhibitory connections, must produce the coordinated, rhythmic locomotor output in their absence (Gezelius et al., 2006; Talpalar et al., 2011; Wallén-Mackenzie et al., 2006). This phenomenon seemed re-stricted to drug induced locomotion as electrical stimulation of Vglut2 null isolated spinal cords failed to activate a rhythm, more in keeping with the abovementioned results (Hägglund et al., 2010; Talpalar et al., 2011). The direct stimulation of spinal inhibitory interneurons cannot elicit a locomotor rhythm, arguing against a purely inhibitory rhythmic drive in a normal spinal cord (Hägglund et al., 2013).

28

Figure 5. The locomotor central pattern generator (CPG) in the spinal cord ventral horn. A schematic spinal cord depicting the CPG where neurons form a network to generate the key features of locomotion; rhythm generation (green) and pattern formation (grey network). The CPG output is sent to ipsilateral and contrala-teral motor neurons (black) to coordinate limb movements. Within the pattern for-mation networks (right), interneuron populations including excitatory interneurons (eIN, orange), commissural neurons (CIN, blue) and inhibitory ipsilateral interneu-rons (iIN), which includes Renshaw cells (RC), orchestrate, shape and direct motor neuron output to coordinate left-right and flexor-extensor alternation.

The cross-inhibitory actions between CPG networks located on different rostro-caudal spinal cord levels are important for flexor-extensor coordina-tion. These reciprocal networks are ipsilateral, as flexor-extensor alternation persists in hemisected spinal cords (Kjaerulff and Kiehn, 1997) and inhibito-ry, as flexor-extensor alternation was lost when inhibition was blocked (Kiehn, 2006). The cross-inhibitory actions between CPG networks on either side of the spinal cord control left-right alternation as left-right alternation was lost when the spinal cord midline was hemisected (Cowley and Schmidt, 1995; Kjaerulff and Kiehn, 1996). Multiple commissural interneuron popula-tions, both excitatory and inhibitory have been identified and are involved in coordinating left-right alternation (for review of the locomotor CPG see Kiehn, 2006; Kiehn et al., 2008).

The locomotor CPG has been extensively characterised in the lumbar spi-nal cord that controls hindlimb patterning. Even though direct intraspinal interactions exist between cervical and lumbar segments, the established connections between CPGs controlling fore- and hindlimb networks are poorly understood. Recent evidence, however, proposes the control of fore-and hindlimb coordination in the cat spinal cord is bidirectional, asymmetric and flexible (Thibaudier et al., 2013). The CPG is not an isolated network and requires the balanced interplay between neural circuits to form a func-tioning CPG that produces coordinated output. The neurons comprising the

MNMN

RhythmeIN

CINiIN RC

Sensory Input

CPG

29

locomotor CPG, in particular those coordinating left-right and flexor-extensor alternation, will be readdressed in ‘Locomotor Related Interneu-rons’.

CPG Models Over the years CPG models have emerged in an attempt to understand and explain how fore- and hindlimb locomotor circuits, accounting for rhythm generation and pattern formation, interact to produce coordinated output. Sherrington initially proposed the idea of a neural ‘half centre’ controlling locomotion, which was subsequently elaborated upon by Thomas Graham Brown (1911, 1914) and Anders Lundberg (1981), to become a ‘half centre model’. Here, the locomotor pattern was generated by flexor and extensor neurons, reciprocally organised to mutually inhibit each other, ensuring only one ‘half-centre’ could be activated at a time. Following the ‘half-centre’ model, Miller and Scott, (1977) proposed that inhibitory interneurons could initiate and sustain coordinated alternation between flexor-extensor motor pools, but their model failed to account for the origin of the locomotor rhythm, although recent evidence does lend support to the model (Talpalar et al., 2011). The ‘unit burst generator’ model proposed by Grillner, (1981) and recently supported by Hägglund et al. (2013) accounts for the complexity of locomotor patterns, and proposes separate ‘models’ or burst generators that control defined subsets of motor neurons. Each side of the spinal cord con-tains networks that can generate rhythmic motor activity to control different motor pools independently (Hägglund et al., 2013), and is supported by ob-servations that hemisected spinal cords can produce rhythmic activity (CPG models reviewed in Guertin, 2009).

Despite the number of proposed locomotor models, there is no clear con-sensus on the organisation of the locomotor CPG at a network level. CPG activity is likely far more complex than these models can predict, and is modulated by descending and peripheral inputs that existing models have limited means of incorporating. Models are continually being reworked and improved as our understanding of the circuitry and specific neural functions increases. Thus, emerging models that incorporate computational modelling with the results of functional rhythmogenic and patterning studies offer a more comprehensive view of the CPG network (Danner et al., 2016; Rybak et al., 2006a, 2006b, 2015).

Spinal Cord Neural Development Within an organism, developing cells are arranged and organised into de-fined bodily structures through a process known as patterning, which en-compasses the complex cellular and molecular mechanisms that program cell fate. In the developing embryo, cells within the ectoderm eventually become the nervous system. Neural populations start their differentiation from a spe-cialised ectoderm structure, the neural tube, where cells migrate away from

30

the neural tube to form the peripheral and autonomic nervous systems. The remaining neural tube becomes the CNS, where spinal cord populations are exposed to signals, particularly the specific expression of genes and chemi-cals that influence, and ultimately decide their individual cellular fate (Wolpert et al., 1998).

Chemical signals secreted by the developing embryo function both broad-ly to pattern anatomical features such as cellular symmetry and dorso-ventral and anterior-posterior distinctions, and finely to define precise subsets of interneurons. These morphogens and the subsequent gene expression they control, follow a specific spatial and temporal expression pattern where cel-lular differentiation and tissue specification depend on the specific sensitivi-ty to and interplay between the signals (Gómez-Skarmeta et al., 2003). Be-tween embryonic day (E) 9.5-18.5, two morphological gradients are estab-lished that determine the neural identity of developing spinal cord neurons and influence their end placement along the spinal cord dorso-ventral axis (Lu et al., 2015). Bone morphogenetic proteins (BMPs), secreted by the dor-sal epidermis/roof plate (Lee and Jessell, 1999), and sonic hedgehog (Shh) secreted ventrally from the notochord and floor plate (Echelard et al., 1993; Placzek, 1995), regulate the dorso-ventral patterning of the neural tube (Fig-ure 6). The opposing activity of BMPs and Shh causes a repression of the signals, restricting the expression of subsequent patterning proteins to de-fined spatial regions along the dorso-ventral axis (Lee et al., 2000; Liem et al., 2000; Placzek et al., 1991; Wilson and Maden, 2005; Yamada et al., 1991).

Figure 6. Spinal cord neural development. Bone morphogenetic proteins (BMPs) and Sonic hedgehog (Shh) secreted from the roof and floor plate (RP, FP) respec-tively, create a dorso-ventral gradient that specifies genetically distinct progenitor domains that differentiate into specific cell populations including the locomotor populations (dI6, V0, V1, V2, V3 and MN), shown. Adapted from Goulding, 2009.

pd1

pd2

pd3

pd4

pd5

pd6

p0

p1

p2pMN

p3

dI6

V0

V1

V2

V3

MN

FP

RP

Shh

BMPs

Sensory

Motor

Progenitor Domains Cell Type

Dorsal

Ventral

31

These dorso-ventral signals regulate the expression of transcription factors that form an integrative transcriptional code from which genetically distinct spinal neural populations are derived at defined dorso-ventral regions in the developing spinal cord. The result is eleven established dorso-ventral pro-genitor domains that specify six dorsal interneuron (dI) domains (dI1 – dI6), one motor neuron domain and four ventral (V) cellular domains (V0 – V3) characterised by the combinatorial expression of different transcription fac-tors and distinct genetic backgrounds (Alaynick et al., 2011; Gross et al., 2002; Lu et al., 2015, Figure 6). Each domain contains molecularly distinct neural subtypes that differ in their end targets, roles and neurochemical phe-notype, but share fundamental properties including origins from the same progenitor cell, migration, final lamina location, medio-lateral positioning and axonal projection (Alvarez et al., 2013; Lu et al., 2015). Interneurons within the ventral domains (V0-V3) and the most ventral dorsal domain (dI6) are thought likely to be associated with the CPG and are further differ-entiated into distinct populations.

Developing Spinal Networks During development, neurons and neural precursors exhibit spontaneous activity that initially starts on a cell-to-cell basis but as synaptic connections form and mature, emerges as correlated spontaneous activity across the forming network. This spontaneous activity shapes circuit characteristics including network connectivity, function, synaptic strength and excitability (review see Blankenship and Feller, 2010). In the spinal cord spontaneous network activity is a general property of developing motor networks and is mediated by cholinergic, glycinergic and GABAergic pathways (Chub and O’Donovan, 1998; Hanson and Landmesser, 2003; O’Donovan, 1999). It is inconclusive as to which neural populations trigger the spontaneous activity, but motor neurons have been postulated to play a prominent role as they exhibit episodes of larger rhythmic depolarisation coupled with periods of inactivity that have been observed from early embryonic days right up until CPG circuits are functional. Once spontaneous activity has been initiated, the spinal interneuron network propagates and sets the rhythmicity of the activity through reciprocal excitatory connections (review see Blankenship and Feller, 2010).

Spontaneous network activity is most prominent in developing circuits when the actions of GABA and glycine are excitatory rather than inhibitory (Ben-Ari, 2002; Ben-Ari et al., 2007; Gonzalez-Islas et al., 2010; Gulledge and Stuart, 2003; Marty and Llano, 2005; Vinay and Clarac, 1999). Due to an increased intracellular chloride concentration in immature neurons, GABA and glycine initially depolarise neurons, however, the excitatory actions of GABA have recently been challenged and suggested to be an arte-fact of intracellular recording procedures (Bregestovski and Bernard, 2012; Zilberter, 2015). As neurons mature there is a developmental shift to inhibi-

32

tory GABA-/glycinergic signalling that is modulated by the ongoing neu-ronal activity (Ben-Ari, 2002).

The dynamics of neural circuits arise through a balanced interplay be-tween synaptic inputs, network connections and the electrical properties of neurons within that network (Prinz et al., 2004). Developing networks are incredibly robust and have phenomenal capacities to recover and re-establish ‘normal’ network activity after being deprived of vital inputs or when critical components of the network are lost (review see Blankenship and Feller, 2010). When network activity is perturbed, neurons adjust their cellular ex-citability, rhythmogenic capabilities, strength of excitatory and inhibitory synapses and ion channel protein levels and conductances to maintain origi-nal network activity levels within the defined physiological range (Banks et al., 2005; De Zeeuw et al., 2003; Fogarty et al., 2016; Harley et al., 2015; Leao et al., 2004; MacLean et al., 2003; Sapir et al., 2004; Swensen and Bean, 2005; Wilhelm et al., 2009). In the spinal cord, compensatory network adjustments have recently been attributed to GABAergic signalling (Wenner, 2014; Wilhelm and Wenner, 2008; Wilhelm et al., 2009). Of particular inter-est, the genetic elimination of gephyrin on motor neurons caused dramatic changes to motor neuron spontaneous inputs, where even though the net synaptic input was conserved, there was a shift in the proportions of inhibito-ry and excitatory synapses (Fogarty et al., 2016). The capacity of the devel-oping nervous system to maintain normal physiological output in the ab-sence of specific network inputs is investigated in Paper II.

Spinal Cord Neural Populations in Locomotion In the locomotor CPG, descending signals and incoming sensory information are acted upon by interneurons that converge, mediate and modulate diverse afferent and local circuit inputs. Motor neurons are generated, selected and targeted to execute the commands of the CPG and whilst seemingly simple in function, represent a diverse, yet highly organised population that together orchestrate a common motor goal.

Motor Neurons Spinal motor neurons are the body’s effector cells; the command neurons relaying the final motor output signal to muscles to produce movement. Mo-tor neurons derive from the premotor neuron progenitor domain and migrate ventrally towards their final position and project their axons shaping the ventral root to innervate skeletal muscle (Lu et al., 2015). Right before the motor neuron axon leaves the spinal cord, an axon collateral projects back into the ventral horn to target local interneurons (see Renshaw cells). In the periphery, the motor neuron axon and muscle fibre form a specialised syn-apse known as the neuromuscular junction (Ribchester, 2009). Motor neu-rons communicate both peripherally and centrally via cholinergic transmis-

33

sion to activate their postsynaptic targets (Fletcher and Forrester, 1975), however, their central transmission is more complex than initially thought (see Renshaw cells). Motor neurons are reliant on inhibitory inputs during late embryonic development for their survival, circuit activity and innerva-tion of musculature (Banks et al., 2005; Fogarty et al., 2013, 2016).

Motor neurons adhere to a strict rostro-caudal and medio-lateral topo-graphical organisation both within the ventral horn and segmentally along the spinal cord. Alpha (α), Beta (β) and Gamma (γ) motor neurons innervate different muscle fibres where motor neurons of different types are common-ly intermingled within the same motor pool (review see Burke, 1981), and are classified based on their anatomical size, morphological and molecular profiles, conduction velocities, electrical and firing properties and the specif-ic expression of transcription factors (Enjin et al., 2010, 2012; Kanning et al., 2010; Zhu et al., 2012). In this thesis, the term ‘motor neuron’ refers to alpha motor neurons.

Motor neurons are primed with their large cell bodies and extensive den-dritic trees, to receive information from three main sources; proprioceptive input from the muscle spindle, descending input from cortical and brain re-gions and local input from the CPG. The CPG drive causes the similar rhythmical activation of motor neurons within motor columns that is inde-pendent of motor neuron subtype or anatomical position. During locomotor activity, motor neurons at a given lumbar segment burst in phase with one another showing comparable activity profiles (Hinckley et al., 2015). Paper III investigates the population activity of motor neurons during locomotor-like activity when an inhibitory CPG population is silenced.

Locomotor Related Interneurons The term ‘locomotor related interneurons’ encapsulates the diverse CPG interneuron populations (dI6, V0, V1, V2 and V3) that are integrated in rhythm generation and pattern formation locomotor networks. Perhaps the coarsest classification of CPG interneurons is by their action as either an excitatory or inhibitory population (Figure 5). Inhibitory interneurons derive from dI6 and V1 populations and have major roles in networks that refine, sculpt and time motor neuron activity to generate patterned output (review see Goulding, 2009; Goulding et al., 2014; Kiehn, 2006; Sapir et al., 2004). Excitatory interneurons are largely, but not exclusively, involved in CPG rhythm generation and include the V3 population (Hägglund et al., 2010, 2013; Hinckley et al., 2005; Kiehn et al., 2008; Wilson et al., 2005; Zhang et al., 2008). V0 and V2 interneurons comprise of both inhibitory and excitato-ry populations where ventral V0 (V0V) and V2a neurons are excitatory, whilst dorsal V0 (V0D) and V2b neurons are inhibitory (Al-Mosawie et al., 2007; Lanuza et al., 2004).

CPG interneurons can be divided by their axonal projections, independent of their neurochemical phenotype, into ipsilateral or commissural interneu-ron classes. The axons of ipsilateral interneurons target downstream neurons

34

on the same side as their cell bodies and include neurons in the V1 and V2 populations. Commissural interneurons (CINs) define a class of cells whose axons cross the midline via a central commissure and target neurons contra-lateral to their own cell bodies and in the spinal cord are the dI6, V0, and V3 populations (Gross et al., 2002; Moran-Rivard et al., 2001; Zhang et al., 2008). The axons of both ipsilateral and commissural interneurons can pro-ject locally within one spinal cord segment (intrasegmental), or over multiple segments (intersegmental), where intersegmental interneurons can further be subdivided into ascending, descending and bifurcating interneurons (Butt et al., 2002; Kiehn, 2006; Stokke et al., 2002). Regardless of their neurochemi-cal phenotype, the common end target for both commissural and ipsilateral interneurons is to directly or indirectly modulate the activity of motor neu-rons (Butt and Kiehn, 2003; review see Kiehn, 2006; Figure 5).

CPG Coordination The connectivity and axonal projections of interneurons within locomotor CPG circuits, in addition to their neurochemical phenotype, gives inferences to the population’s potential role in locomotion. Generation of the locomotor rhythm is directly linked to the activity of excitatory interneurons, among which are a subset of the V2a interneurons defined by the expression of Shox2 (see Central Pattern Generator; Dougherty et al., 2013; Hägglund et al., 2013). Excitatory and inhibitory, commissural and ipsilateral interneu-rons are part of complex rhythmic networks that play a critical role in pattern formation, controlling and coordinating the alternation between left-right and flexor-extensor motor pools. Genetic approaches, selectively targeting CPG interneuron populations based on their specific transcription factor expres-sion, have been successful in uncovering the potential function of multiple locomotor interneuron populations within the CPG.

Left-Right Alternation Commissural inhibitory interneurons were initially thought to directly con-trol left-right alternation, where a dual-inhibitory commissural system impli-cated both inhibitory and excitatory commissural interneurons acting in al-ternate pathways (Kiehn et al., 2010; Quinlan and Kiehn, 2007; Figure 7). In this system, inhibitory commissural pathways directly innervate contralateral motor neurons, whilst excitatory commissural interneurons indirectly inhibit contralateral motor neurons through the activation of ipsilateral inhibitory populations (Quinlan and Kiehn, 2007). The V0 population underpins this dual-inhibitory system as the inhibitory dorsal (V0D) and excitatory ventral (V0V) subsets support both the direct and indirect inhibition of contralateral motor neurons (Figure 7). Ablation of the entire V0 population in neonatal mice caused disturbed locomotor patterns where mice displayed bilateral synchrony rather than normal left-right alternation (Lanuza et al., 2004).

35

This finding was supported in adult animals where mice with selective abla-tion of V0 neurons lost the ability to perform alternating gaits (Bellardita and Kiehn, 2015). Moreover, the selective deletion of either the V0D population or the V0V population was sufficient to disrupt left-right alternation and promote synchrony (Talpalar et al., 2013) as mice had selective and defined difficulties performing and maintaining alternation across certain gaits (Bellardita and Kiehn, 2015).

Excitatory, ipsilateral subsets of the V2a interneurons were also found to be integral in coordinating left-right alternation as mice with the genetic deletion of the V2a population developed a synchronous gait (Crone et al., 2008, 2009). V2a interneurons project to commissural excitatory interneu-rons including the V0V population and thus indirectly mediate the contrala-teral inhibition of motor neurons (Dougherty and Kiehn, 2010b). The im-portance of excitatory commissural interneurons in left-right alternation was realised when mice with increased excitatory commissural projections due to faulty axonal guidance, displayed a hopping gait in place of left-right alter-nation (Kullander et al., 2003). Subsequent studies reported that blocking the synaptic drive from V3 commissural interneurons lead to a variable locomo-tor output with weaker left-right alternation (Zhang et al., 2008), whilst the spared V3 commissural projections in netrin mutants generated synchronised hindlimb activity (Rabe et al., 2009). The latter findings directly implicate V3 neurons in left-right alternation, where these direct excitatory, commissu-ral projections are suggested to comprise a direct excitatory pathway to con-tralateral motor neurons that supports synchrony (Quinlan and Kiehn, 2007, Figure 7). Together the V0, V2a and V3 populations contribute to left-right alternation via direct (V0D and V3), and indirect (V0V driven by V2a) path-ways (review see Kiehn et al., 2010).

36

Figure 7. The proposed dual-inhibitory system underlying left-right alternation. Left-right alternation is suggested to be coordinated by excitatory and inhibitory commissural interneurons (blue, CIN) that directly or indirectly inhibit or excite contralateral motor neurons (MN). The V0 population supports the dual pathway; inhibitory V0D neurons directly inhibit motor neurons whilst excitatory V0v neurons activate ipsilateral inhibitory neurons (yellow) including Renshaw cells (RC), Ia interneurons (IaIN) and currently unidentified neurons (?IN) that subsequently in-hibit motor neurons. The V3 population comprise a direct excitatory pathway that supports left-right synchrony. Inhibitory synapses represented by closed circles; excitatory synapses by lines. Dotted line represents spinal cord midline. Adapted from Quinlan and Kiehn, 2007.

Flexor-Extensor Alternation Flexor-extensor alternation is suggested to be generated by the reciprocal connections between groups of ipsilateral inhibitory interneurons (Kiehn, 2006), however, the functional organisation of flexor-extensor networks is currently not well defined. V1 and V2b interneurons are thought to be in-volved in flexion-extension coordination because flexor-extensor alternation around a limb joint depends on the regulated activity of ipsilaterally project-ing inhibitory networks (Zhang et al., 2014). Moreover, the selective abla-tion of V1 or V2b interneurons in adult mice affirm and extend the findings by Zhang et al. (2014) revealing the V1 and V2b populations independently control the step cycle through different mechanisms. V1 interneurons short-en the bursting of flexor motor neurons to promote limb extension, whilst V2b interneurons limit the firing of extensor motor neurons and promote flexion (Britz et al., 2015). Previously, V1 interneurons have been proposed to coordinate flexor-extensor motor neuron patterning (McCrea et al., 1980), however, mice that develop without the V1 interneuron population have normal flexor-extensor motor neuron activity (Gosgnach et al., 2006; review see Goulding, 2009).

V1 Interneurons V1 interneurons derived from the p1 progenitor domain, that transiently express engrailed 1 early during development, comprise a heterogeneous

< MN

RC

IaIN

?IN

CIN

CIN

CIN

DualInhibitoryPathway

ExcitatoryPathway

V0v

V0D

V3

37

population that account for one third of all ventral inhibitory populations (Alvarez et al., 2005, 2013; Benito-Gonzalez and Alvarez, 2012; Sapir et al., 2004). After exiting the cell cycle between E9.5-12, V1 interneurons follow a ventrolateral migration pathway, project to ipsilateral motor neurons and up- or downregulate different transcription factors and proteins that start to shape cellular characteristics and neurochemical phenotype (Alvarez et al., 2005; Benito-Gonzalez and Alvarez, 2012). Although flexor-extensor coor-dination is not the primary role of V1 neurons, the population was revealed to shape motor neuron output and determine the speed of locomotion (Gosgnach et al., 2006; Zhang et al., 2014).

V1 neurons are further divided into subpopulations based on their transi-ent expression of transcription factors, exit time from the cell cycle, dorso-ventral migration, expression of specific calcium buffering proteins, connec-tivity within locomotor circuits and electrophysiological properties (Alvarez et al., 2005, 2013; Lu et al., 2015; Sapir et al., 2004). Recently V1 interneu-rons were fragmentally subdivided into approximately 50 transcriptionally defined subtypes by their restricted, transient expression of 19 transcription factors. This genetic diversity broadly defined four V1 subpopulations or clades that were anatomically, physiologically and functionally discrete (Bikoff et al., 2016), documenting extensive diversity within V1 interneu-rons. Despite this diversity only two V1 populations have been well charac-terised; Ia inhibitory interneurons (IaIN) and Renshaw cells (RCs), which account for approximately 25% of the V1 population (Alvarez et al., 2005) leaving the vast majority of V1 interneurons without a defined population or functional role.

Whilst IaIN and RCs have similar functional features, they are both rhythmically active and primarily responsible for the inhibition of motor neurons during locomotion (Eccles and Lundberg, 1958; Eccles et al., 1954; Geertsen et al., 2011; Pratt and Jordan, 1987; Talpalar et al., 2011) and share properties common to V1 interneurons, they represent characteristically dis-tinct subtypes (Alvarez et al., 2005; Benito-Gonzalez and Alvarez, 2012; Sapir et al., 2004). The specification of Renshaw cells and IaINs from the V1 lineage is dependent on the temporal control of their neurogenesis where Renshaw cells are generated early during development (E9.5–10.5), whilst IaINs are born later at E11–12.5 (Benito-Gonzalez and Alvarez, 2012). This differential generation influences neural differentiation and migration, which ultimately affects the integration of each neuron into functionally distinct inhibitory circuits. Renshaw cells form a unique feedback circuit with ho-monymous and synergistic motor neurons, known as recurrent inhibition (see Recurrent Inhibition; Eccles et al., 1954; Sapir et al., 2004), whilst IaINs act as a connection between sensory input and motor output, reciprocally inhibiting motor neurons innervating antagonistic muscles (see Sensory Re-flexes; Figure 4). Renshaw cells and IaINs receive excitatory inputs arising from proprioceptive afferents, but whilst these proprioceptive inputs are

38

‘deselected’ from Renshaw cells during development, they are strengthened on IaINs supporting their reciprocal function (Mentis et al., 2006; Siembab et al., 2010). Recently, IaINs have been proposed to originate from other non-V1 derived progenitor domains (Wang et al., 2008; Zhang et al., 2014), whilst Renshaw cells develop exclusively from V1 interneurons (Alvarez et al., 2005; Sapir et al., 2004). Furthermore, the IaINs are the only known inhibitory cell population to receive inputs from and thus be inhibited by, Renshaw cells (Jankowska, 1992), which further increases the complexity of CPG inhibitory circuits.

Speed Modulation In quadrupeds and bipeds, changes in locomotor velocity are accompanied by functional changes to gait, where shifts in both stance and swing phases alter locomotor step patterns and permit an increase or decrease in speed. Neurologically, modulation of locomotor speed is mediated by changes in the excitatory drive within locomotor circuits, where an increase in excitato-ry drive increases locomotor speed (Danner et al., 2016). In vivo, this excita-tion descends from cortical and brainstem areas and is manifested through the CPG, but in vitro, this excitation is derived from external stimulation or pharmacological activation of spinal locomotor circuits (see Fictive Loco-motion). The transition from slower alternating gaits such as walking, to faster synchronous gaits such as bound, has been defined by the balance of activity in commissural interneuron pathways (Danner et al., 2016), which local CPG circuits must potentiate and interpret into a cohesive locomotor output.

The CPG interneurons implicated in speed-dependent modulation of lo-comotion are varied and diverse, not strictly belong to a single population and appear to exert differential effects at particular speeds (Bellardita and Kiehn, 2015; Danner et al., 2016). In both mice and zebrafish, the V2a popu-lation are readily involved in speed-dependent networks. In zebrafish, modu-lating the activity of these neurons accelerated and decelerated the speed of locomotion (Ampatzis et al., 2014), whilst V2a interneurons in mice are implicated in coordinating locomotion at moderate to high locomotor speeds (Crone et al., 2009). Inhibitory populations are also involved in driving the speed of locomotion. Motor neuron activity decreased when the V1 inhibito-ry interneurons were genetically ablated resulting in a longer step cycle and slower locomotor frequency (Gosgnach et al., 2006). Moreover, the V0 pop-ulation coordinates left-right alternation in a speed-dependent manner where the V0D and V0V populations coordinate locomotion at slower and faster speeds respectively, and are permissive for a mouse’s ability to shift between gaits (Bellardita and Kiehn, 2015; Talpalar et al., 2013). Mice lacking the V0V neurons were unable to trot, whilst bound was the default gait in mice lacking both V0V and V0D populations (Bellardita and Kiehn, 2015). Paper III further investigates locomotor interneurons contributing to loco-

39

motor speed modulation and focuses on a possible speed-dependent role of a subpopulation of dI6 neurons that express the Doublesex and mab-3 related transcription factor 3 (Dmrt3).

dI6 Neurons and Dmrt3 Neural populations from adjacent progenitor domains in the absence of spe-cific transcription factors can acquire a molecular character similar to that of neighbouring populations (Vallstedt and Kullander, 2013). The dI6 and V0 populations share common properties as they arise from progenitor domains with similar characteristics (Gross et al., 2002; Lanuza et al., 2004). In par-ticular, the V0D subpopulation demonstrate a molecular character similar to the dI6 population when the transcription factor differentiating the V0 popu-lation is lost (Lanuza et al., 2004). Whilst the potential roles and functions of the V0 population have been extensively researched, knowledge surrounding the dI6 population is limited and remains one of the least characterised spi-nal cord interneuron populations.

dI6 neurons originate from dorsal pd6 progenitor cells that leave the cell cycle at approximately E11.5 and migrate ventro-medially to settle in ventral horn laminae VII/VIII (Alaynick et al., 2011; Gross et al., 2002). The dI6 neurons are an inhibitory population of ipsi- and commissural interneurons that innervate contralateral motor neurons, and are proposed to be involved within circuits coordinating left-right alternation (Andersson et al., 2012; Dyck et al., 2012; Goulding, 2009; Rabe et al., 2009). Dorsal dI6 neurons could be further divided into subpopulations based on their intrinsic and functional properties, where a subset of neurons were observed to be rhyth-mically active during locomotion (Dyck et al., 2012). Within this subset, a proportion of neurons were suggested to comprise part of the rhythm-generating network, whilst the remaining neurons comprise part of the pat-tern-forming network (Dyck et al., 2012).

Genetic markers to identify and subdivide the dI6 population have been sorely lacking. Recently, the gene Doublesex and mab-3 related transcription factor 3 (Dmrt3) was found to mark a subset of the dI6 neurons and has ma-jor implications in the gait and locomotion of both horses and mice (Andersson et al., 2012). ‘Gaitedness’, the ability of horses to perform two alternate gaits in addition to the three natural gaits, is readily selected for in many breeds (Albertsdóttir et al., 2007; Jäderkvist et al., 2014). A nonsense mutation introducing a premature STOP codon in the gene DMRT3 was found to be permissive for a horses’ ability to perform alternate gaits (Andersson et al., 2012; Kristjansson et al., 2014) where the DMRT3 muta-tion was found at a high frequency in all gaited horse breeds worldwide (Promerová et al., 2014). Locomotor studies in Dmrt3-/- knockout mice, con-firmed Dmrt3 is also central to locomotor coordination in the mouse. Dele-tion of Dmrt3 affects dI6 neuron development and subsequently Dmrt3-/-

40

mice had a disturbed CPG output that resulted in pups displaying severe difficulties coordinating their hindlimbs and were unable to maintain left-right alternation (Andersson et al., 2012). In the adult Dmrt3-/- mouse, motor coordination and balance were largely unaffected, but mice had significant gait abnormalities showing an increased stride length with subsequent diffi-culties running at high speeds (Andersson et al., 2012). Whilst the Dmrt3 gene has significant consequences for locomotion in mice, the characteristics and functional role of the population of spinal cord neurons derived from Dmrt3 is unknown. Thus, the Dmrt3 derived population is explored in Paper III.

Recurrent Inhibition The locomotor CPG is a functioning neural network that generates a single patterned output in the absence of external inputs. Within this network addi-tional circuits function to tightly regulate motor neuron output that work largely independently of the unique CPG rhythm generation and pattern formation network. These circuits include the reciprocal inhibition of motor neurons by IaINs and recurrent inhibition mediated by Renshaw cells.

Recurrent inhibition is an additional feedback circuit that instantaneously reads the motor signal sent to the muscle for rapid movement output adjust-ments (Eccles et al., 1961; Windhorst, 1996). The recurrent inhibitory circuit was one of the first functional circuits described in the nervous system, and the interneurons governing the inhibition the Renshaw cells (Renshaw, 1946), have been extensively described in many organisms and share similar features and characteristics (Hultborn et al., 1979; Obeidat et al., 2014; Szczupak, 2014; Wenner and O’Donovan, 1999). In the rodent and cat recur-rent inhibitory circuit, motor neurons and Renshaw cells act as both pre- and postsynaptic targets. Renshaw cells are monosynaptically connected to mo-tor neurons via motor neuron axon collaterals and in turn, project to inner-vate and inhibit homonymous and heteronymous motor neurons (Figure 8). The recurrent inhibitory network provides additional control to functioning motor circuits, as Renshaw cells simultaneously receive, process and re-spond to the motor neuron signal being delivered to the muscle. The selec-tive activation of the recurrent inhibitory pathway is effective in supressing motor neuron firing as just one action potential emitted from the Renshaw cell is sufficient to disrupt the motor neuron signal (Bhumbra et al., 2014). Conversely, single motor neuron action potentials reliably activate and drive Renshaw cell firing (Moore et al., 2015), where Renshaw cells reportedly reliably fire in response to motor neuron inputs upwards of 50 Hz (Ross et al., 1975). The generated recurrent inhibition can alter the firing rate and temporal timing of motor neuron spikes, act to variably change the gain mo-

41

tor neurons or strongly suppress or stop motor neuron activity (Hultborn et al., 2004; Obeidat et al., 2014; Pratt and Jordan, 1987).

Figure 8. The recurrent inhibitory circuit. Renshaw cells (RC) innervated by a motor neuron (MN) axon collateral, project their axons back to inhibit the motor neuron.

The connectivity, synaptic inputs and function of the recurrent inhibitory circuit are shaped by the migration and positioning of Renshaw cells during development (Alvarez et al., 2013; Benito-Gonzalez and Alvarez, 2012; Bikoff et al., 2016). The motor neuron-Renshaw cell relationship is facilitat-ed by the early generation of Renshaw cells, coupled with their unique cir-cumferential migration to follow and cross the path of motor neurons as they migrate towards the ventral root (Alvarez et al., 2013; Benito-Gonzalez and Alvarez, 2012). The inability of motor neuron axons to leave the ventral root and their need to find innervation targets reinforces the recurrent circuitry where, at E13, motor neuron axons project to and establish a close connec-tion with Renshaw cells. Renshaw cell innervation of motor neurons begins later at E15 and increases in frequency up until E17 when motor neuron synapses are cemented and a functional circuit is established (Alvarez et al., 2013).

Although in principle the recurrent inhibitory circuit appears straightfor-ward, the wider connectivity and inhibitory influence is more diverse. Any single Renshaw cell is innervated by motor neurons innervating homony-mous or heteronymous muscles (Hamm, 1990), where in turn, Renshaw cells generate recurrent inhibition on targeted motor neurons, IaINs and other Renshaw cells (Jankowska, 1992; Windhorst, 1996). Moreover, recurrent inhibition is not uniform along the cord or across motor pools as inhibitory strength decreases as Renshaw cells target motor neurons at increasing dis-tances along the cord (Eccles et al., 1961; Hamm et al., 1987), and as they innervate motor neurons that control more distal muscles (McCurdy and Hamm, 1992). Innervation of Renshaw cells by larger motor neurons pro-vides more powerful recurrent inhibition than innervation by smaller motor neurons (Burke, 1981) although this has been challenged and instead motor neurons innervating fast-fatigable muscles generate stronger recurrent inhibi-

MNRC

42

tion than those innervating slow muscles (Windhorst, 1996). The capacity of the recurrent inhibition circuit to maintain functional output depends on the extent of convergence of excitatory and inhibitory inputs within the circuit (Moore et al., 2015).

Renshaw Cells The Renshaw cells (RC), named after their discoverer Birdsey Renshaw, represent only 2-3% of all ventral interneurons (Alaynick et al., 2011; Alvarez et al., 2005; Fitzsimons et al. 2006). Whilst the molecular, morpho-logical, synaptic and electrical profiles of the Renshaw cells have been com-prehensively documented since their discovery, their definitive functional role remains elusive. This thesis highlights the main physiological and char-acteristic attributes of the Renshaw population.

Renshaw cells are the inhibitory interneuron governing recurrent inhibi-tion to motor neurons, where a reduction of motor neuron firing observed in response to ventral root stimulation, is a hallmark property of this network (Eccles et al., 1954; Renshaw, 1941; Renshaw, 1946). Renshaw cells have a multipolar or fusiform morphology with small cell bodies (Fyffe, 1990) and axons that project ipsilaterally, bifurcating at the ventral furniculus, extend-ing rostro-caudally into lamina IX to target motor neuron proximal dendrites (Mentis et al., 2006; Sapir et al., 2004). Contralateral Renshaw cell projec-tions, although suggested (Ryall et al., 1971), have not been confirmed.

Renshaw cells can be identified by their anatomical location within the ‘Renshaw cell area’ in lamina IX (Thomas and Wilson, 1965) and their char-acteristic immunohistochemical profile (review see Alvarez et al., 2013). However, the only unequivocal confirmation of Renshaw cells in vitro is the generation of an evoked monosynaptic response from ventral root stimula-tion. To date confident immunohistochemical identification of Renshaw cells relies on the unique expression of large gephyrin clusters that develop be-tween P10 and P14, and the expression of Calbindin (Alvarez et al., 1997; Carr et al., 1998; Geiman et al., 2000; Gonzalez-Forero et al., 2005; Sapir et al., 2004). Renshaw cell gephyrin clustering is discriminant from other in-hibitory interneurons and motor neurons (Alvarez et al., 1997), whilst cal-bindin, under the control of transcription factors, is upregulated soon after Renshaw cells begin differentiation and retained in mature Renshaw cells but down regulated in other early V1 populations (Alvarez et al., 2013; Benito-Gonzalez and Alvarez, 2012; Stam et al., 2012). However, both gephyrin and calbindin are abundant in the neonatal spinal cord and thus remain unreliable markers for Renshaw cells (Alvarez et al., 1997; Carr et al., 1998; Geiman et al., 2000; Gonzalez-Forero et al., 2005; Sapir et al., 2004).

During the first two postnatal weeks, the Renshaw cell-motor neuron cir-cuit undergoes considerable maturation that coincides with weight bearing

43

locomotion, reflex development and improved motor coordination. During this time, postnatal modifications increase and adjust synaptic strength, where there is a considerable shift in the primary innervation of Renshaw cells (Mentis et al., 2006). With a predicted RC:MN ratio of 1:5 (Fitzsimons et al., 2006), Renshaw cells receive considerable excitatory inputs from mul-tiple motor neuron axon collaterals on their proximal dendrites that have a large probability of neurotransmitter release (Nishimaru et al., 2006; Siem-bab et al., 2010; Alvarez et al., 1999; Mentis et al., 2006; Moore et al., 2015; Bhumbra et al., 2014). In fact, it is estimated that there are approximately 75 motor neurons innervating any one Renshaw cell, whilst 40 Renshaw cells innervate any one motor neuron (Alvarez et al., 1999; Moore et al., 2015). Motor neuron synapses on neonatal Renshaw cells are established early and continue to expand their innervation domain during postnatal maturation (Mentis et al., 2006). Coupled with the innervation from motor neurons, Renshaw cells are innervated at E18 by VGLUT1 positive proprioceptive afferents on their distal dendrites that whilst initially expand, subsequently decline in density after P20 (Mentis et al., 2006). Maintaining the innerva-tion balance is critical for normal circuit and synaptic development as alter-ing the strength of proprioceptive and motor neuron inputs considerably impacts the synaptic density on Renshaw cells (Siembab et al., 2016).

Adding to the complexity, Renshaw cells receive considerable mon-osynaptic inhibitory inputs on their most proximal dendrites from other spi-nal inhibitory interneurons and inputs from excitatory interneurons and some descending fibres (D’Acunzo et al., 2014; Kiehn, 2006; Mentis et al., 2006; Ryall, 1970, 1981; Alvarez et al., 1997; Nishimaru et al., 2006; Nishimaru et al., 2010; Windhorst 1996). Coupled with their proximal dendritic innerva-tion of motor neurons (Bhumbra et al., 2014; Eccles et al., 1954; Renshaw, 1946), Renshaw cells innervate and inhibit other Renshaw cells, IaINs and Ib interneurons (Hultborn et al., 1971; Lamotte d’Incamps and Ascher, 2008). Thus the local recurrent inhibition circuit, whilst fundamentally simple in principle, has many confounding factors that increase the complexity of the circuit, making functional isolation of the Renshaw cell from its surrounding networks difficult.

The MN-RC synapse was the first example of a fast, cholinergic synapse in the CNS (Eccles et al., 1954). Cholinergic transmission at the MN-RC synapse has since been confirmed, as pharmacological blockade of the nico-tinic acetylcholine receptors (nAChR α2 and α4) in isolated spinal cords, silenced Renshaw cell firing (Noga et al., 1987). Of note, whilst nAChR α4 is expressed in other spinal interneurons, nAChR α2 has an expression pat-tern restricted to Renshaw cells and has been hinted as a potential Renshaw cell marker (Dourado and Sargent, 2002; Enjin et al., 2010; Ishii et al., 2005). Currently glutamate and/or aspartate are suggested to be co-released with acetylcholine at central motor neuron synapses (O’Donovan et al., 2010; Richards et al., 2014) as antagonists of nAChR failed to completely inhibit evoked motor neuron activation of Renshaw cells (Lamotte

44

d’Incamps and Ascher, 2008; Mentis et al., 2006; Nishimaru et al., 2005). Vesicular glutamate transporters (VGLUTs) are not co-localised with ve-sicular acetylcholine transporters in the terminal, suggesting that although glutamate may contribute, aspartate, capable of activating NMDA receptors independently of VGLUT mechanisms, might be a better candidate (Richards et al., 2014). Although controversial, synaptic transmission at the MN-RC synapse is thus suggested to be cholinergic, aspartaergic and glu-tamateric (Lamotte d’Incamps and Ascher, 2008; Mentis et al., 2006; Ni-shimaru et al., 2005; Richards et al., 2014). Renshaw cell inhibitory neuro-transmission is both glycinergic and GABAergic distinguishing Renshaw cell inhibition from that of other spinal interneurons (Schneider and Fyffe, 1992). Renshaw cell GABAergic transmission prolongs the inhibitory signal on motor neurons, as GABAergic currents have a slower time course than glycinergic currents (Cullheim and Kellerth, 1981).

Renshaw cells are known to be rhythmically active during locomotion (McCrea et al., 1980; Nishimaru et al., 2006), but not integral for generating the rhythm (Pratt and Jordan, 1987), thus the precise functional role of the population in locomotion remains unclear. Renshaw cells fire rhythmical bursts in-phase with the motor neuron pool they are in innervated by and thus potential roles are to shorten the burst duration of motor neurons, con-tribute to rhythm generation and coordinate flexor-extensor alternation (McCrea et al., 1980; Pratt and Jordan, 1987; Nishimaru et al., 2006). Fur-ther theories suggest that Renshaw cells may change the gain of motor pools (Hultborn et al., 2004), influence the de-correlation of motor neuron firing (Maltenfort et al., 1998) or reduce motor neuron firing during locomotion (Noga et al., 1987). Convincing evidence supporting a functionally distinct role for Renshaw cells in locomotion is lacking despite the comprehensive array of neurochemical, immunohistochemical, pharmacological, anatomical and electrophysiological data contributing to the Renshaw cell profile. Iden-tifying a genetic marker to selectively isolate the Renshaw cells from sur-rounding circuits was a focus of Paper I, and has been supported by func-tional studies investigating the role of the Renshaw cell in Paper II.

Clinical Applications Spinal cord injuries and motor neuron diseases severely impair the quality of life and significantly reduce the life expectancy of patients living with these afflictions. Whilst the prevalence of motor related conditions continues to rise due to environmental, genetic and lifestyle factors, most motor diseases have an unknown cause with uncertain pathophysiology and progression, making it difficult to identify therapeutic targets and possible treatment in-terventions.

The underlying general pathophysiology of motor diseases is a disruption to normal motor circuitry signals and subsequent communication with mus-

45

culature. Alterations in, and the possible loss of neuronal inhibitory mecha-nisms often lead to deficits within cortical and spinal motor circuits, which manifest as severe functional disturbances when executing motor tasks. Dys-tonia, a common form of behavioural disturbance encompasses irregular hyperkinetic, tremor-like movements that originates as defective neural net-works in multiple brain regions leading to reduced inhibition in the spinal cord, brainstem and cortex producing dysfunctional movements (review see Albanese et al., 2013; Hallett, 2011; Phukan et al., 2011). There are many underlying genetic causes for dystonia (review see Phukan et al., 2011; Liu et al., 2015), but little clarity surrounding the exact neural populations con-tributing to the phenotype.

Amyotrophic lateral sclerosis (ALS) is the progressive degeneration of cortical and spinal motor neurons that ultimately render patients paralysed. Although there are underlying genetic mutations associated with ALS (Chen et al., 2013; Kenna et al., 2016), in most cases there is an incomplete under-standing behind the mechanism of selective motor neuron death. Excitotoxi-city, presenting as intrinsic motor neuron hyperexcitability has been pro-posed as a likely mechanism of motor neuron degeneration possibly stem-ming from increased glutamate-mediated transmission, the loss of inhibitory neurotransmission or alterations to inhibitory circuits (review see Martin and Chang, 2012; Ramírez-Jarquín et al. 2013). The recurrent inhibition circuit is an attractive candidate for the source of the hyperexcitability as Renshaw cells are reported to have the earliest neuronal damage leading to either di-minished or lost recurrent inhibition, supporting findings in ALS patients (Fornai et al., 2008; Pasquali et al., 2009; Raynor and Shefner, 1994). In work using experimental animal models of ALS, although there is a univer-sal agreement that glycinergic inhibition is affected during ALS progression (Chang and Martin, 2009, 2011; Hossaini et al., 2011; Martin and Chang, 2012; Wootz et al., 2013), it is not agreed as to how this loss of inhibition manifests. A loss of Renshaw cell innervation on motor neurons, followed by a complete loss of the Renshaw cell population could underlie the hyper-excitability (Chang and Martin, 2009, 2011), however, significant recurrent inhibitory circuit and synaptic changes occur prior to complete motor neuron degradation that are not a direct consequence of Renshaw cell loss (Hossaini et al., 2011; Wootz et al., 2013). Rather Renshaw cell loss is initiated by motor neuron degeneration, catalysed by a retraction of motor neuron synap-ses on Renshaw cells, that disconnects the recurrent inhibition circuit (Wootz et al., 2013) This suggests that both Renshaw cells and their subsequent cho-linergic innervation present as attractive targets for continued ALS research.

Patients with full or partial spinal cord injuries have previously had little hope of ever improving or regaining control over their movements. Now, armed with an increased understanding of the locomotor network in particu-lar how rehabilitation based activities promote the re-wiring of brain and spinal cord circuits following disease or trauma, new therapeutic treatments show considerable promise to induce permissive locomotion from paralysed

46

states. These approaches are founded and governed on the ability of the lo-comotor CPG to generate rhythmical output in absence of descending inputs. In humans, similar to cats or rodents, lumbar spinal cord stimulation induces patterned, rhythmical activity in muscles and initiates step-like lower limb movements resembling those underlying volitional locomotion (Courtine et al., 2009; Danner et al., 2015; Dimitrijevic et al., 1998; Ladenbauer et al., 2010; Minassian et al., 2007). The stereotypical motor output generated by these networks, combined with chemical and electrical initiation of move-ment in isolated preparations, provides a solid physiological framework in which to build clinical and rehabilitation based therapies (Musienko et al., 2009).

In 2012, Van den Brand and colleagues designed an electrochemical neu-roprothesis for paralysed rodents that when combined with a robotic support interface, locomotor training and local stimulation of spinal neural networks, encouraged the brain to reuse and re-wire locomotor circuits to promote stepping. After a period of active locomotor training, rats with experimental-ly induced spinal cord injuries were capable of initiating and sustaining full weight bearing locomotion during electrochemical enabled motor states, even navigating obstacles and making task specific adjustments (van den Brand et al., 2012). This approach capitalised on the plasticity of the nervous system to remodel to construct novel and highly functional supraspinal pathways from intact, dormant spinal circuits to regain cortical control of locomotor networks under defined conditions. The success in generating bipedal locomotion from paralysed quadrupeds strongly suggests this thera-peutic outlet could be beneficial for sufferers of a rare human gait disorder, Uner Tan Syndrom, characterised by quadrupedal locomotion and for in-complete spinal cord injury patients (Guertin, 2012).

Several enhancers of inhibitory signalling combined with therapeutic ap-proaches have been used to treat both ALS and dystonia disorders with little success (see review Hallett, 2011; Ramírez-Jarquín et al. 2013). Investigat-ing and understanding the intricacies of normal motor networks remains of critical importance to identify genes and neurons involved in disease pro-gression that could present novel treatment targets. Moreover, combining active rehabilitation and training with electrochemical stimulation of func-tionally isolated, neural networks offers a promising therapeutic modality for limb paralysis and genetic CPG abnormalities, and reiterates the need for a functional and detailed understanding of the spinal cord locomotor network.

47

Aims

The overall aim of this work was to decipher the spinal cord locomotor net-work and identify the functional roles and contributions of individual neural populations to the locomotor output. Specific aims are detailed below:

Paper I To confirm that Renshaw cells can be genetically labelled by their expres-sion of Chrna2 and to identify potential modulatory currents and their con-tribution to the electrophysiological profile and firing patterns of Renshaw cells.

Paper II To investigate the functional, electrical and anatomical importance of VIAAT-mediated Renshaw cell signalling during recurrent inhibitory circuit development and the overall impact and role of recurrent inhibition within the locomotor CPG.

Paper III To characterise the morphological, synaptic and electrophysiological proper-ties of the commissural interneuron population derived by Dmrt3 and inves-tigate their functional role and contribution to locomotor output and subse-quent motor behaviour.

48

Methodological Considerations

This thesis presents results obtained through a diverse array of methods, techniques and experimental procedures that are described in detail in the materials and methods section of Papers I – III. The following section is therefore limited to outlining the methodological considerations of the main methods either performed by myself, or those underlying the foundation of the papers (animal models and treadmill locomotion).

Animal Models Animal models are invaluable in the scientific investigation of normative and patho-physiological mechanisms and behaviours, and offer the potential for preclinical to clinical research translations. The creation of the C57BL mouse line by Clarence Little (Crow, 2002), in combination with advancing transgenic technologies and the full sequencing of the mouse genome (Mouse Genome Sequencing Consortium et al., 2002) has made the mouse a powerhouse in biological research. The short gestational period and life cy-cle, large litter size, considerable genetic similarities to humans and the pos-sibility to manipulate the mouse genome continue to make mice an attractive research tool. Within recent years, mouse models have been purposefully designed and tailored to investigate specific developmental, physiological and functional aspects of the spinal cord locomotor CPG network.

This thesis details the generation of Cre-LoxP transgenic mouse models (see Genetic Targeting of Neural Populations), to investigate the characteris-tics and function of the Renshaw cell and Dmrt3 populations during locomo-tion and motor behaviour. The animal models and husbandry procedures are described in brief below:

All animal procedures were approved by the local Swedish ethical commit-tee (permit C248/11, C135/14). Heterozygous Chrna2Cre and Dmrt3Cre posi-tive or wild-type (WT) control animals of either sex maintained on a C57BL6 background were used for analysis. Cre mouse lines were generated in-house using Bacterial Artificial Chromosome (BAC) and Zinc Finger Nuclease (ZFN) technologies. Chrna2Cre and Dmrt3Cre knock in mice were crossed with the reporter line tdTomato (Gt(ROSA)26Sortm14(CAG-tdTomato)Hze; Allen Brain Institute; (Madisen et al., 2010) to visualise neural populations

49

and were used in combination with Viaatlx/lx mice (Vgatlx/lx) (Tong et al., 2008) to conditionally silence neurons. Additionally, Chrna2Cre mice were used in combination with the reporter line TaumGFP-nlslacZ (Hippenmeyer et al., 2005) and B6.Cg-Tg mice (Thy1-Brainbow1.0)HLich/J (Jackson lab).

Electrophysiology Electrophysiology is the measurement of membrane electrical currents and voltage changes as a consequence of ionic conductance. While electrophysi-ological recordings encompass everything from electrical measurements from the heart to the electrical potentials of muscles, this thesis focuses on single neuron recordings from spinal cord populations and extracellular compound action potentials recorded from the ventral roots.

Fictive Locomotion Fictive locomotion is an in vitro protocol in which induced ‘fictive’ locomo-tor-like activity is recorded in an isolated spinal cord devoid of descending or sensory inputs. Locomotor signals recorded as left and right (l/r) and flex-or and extensor (f/e) motor bursts through the ventral roots are used to ex-trapolate and reconstruct the locomotor pattern. Individual locomotor param-eters, including burst cycle length, temporal cycle variations and the overall coordination of l/r and f/e burst patterns can be identified from the recorded activity. As an animal steps with one limb at a time, locomotor bursts should alternate in parallel between l/r motor pools. As the animal uses opposing muscles around the joint to propel the limb forward, f/e motor pools burst in alternation (Figure 9).

In an intact nervous system the ‘drive’ for locomotion initiation descends from the brain to activate CPG locomotor networks. In a fictive locomotion preparation, this excitatory drive is provided either electrically through stim-ulation of brainstem neurons (Hägglund et al., 2010, 2013; Talpalar and Kiehn, 2010) or chemically through application of a combination of drugs artificially mimicking descending input from higher structures (Harris-Warrick, 2011; Miles and Sillar, 2011). In combination with an excitatory amino acid (most often N-methyl-D-aspartate, NMDA) that provides the main excitation (Kudo and Yamada, 1987), the neuromodulatory amines, dopamine (DA), serotonin (5-HT) and noradrenaline (NA) found extensively throughout the brain, are an effective means of inducing locomotion (Beliez et al., 2014).

The ‘optimal’ locomotor-like output in vitro generates alternating motor bursts with a stable rhythmicity and cycle pattern (Figure 9), but tempera-ture, the ionic composition of surrounding buffers and the combination and concentration of locomotor drugs, influence the rhythm. Several drug com-binations and concentrations induce the ‘optimal’ locomotor-like pattern (Dougherty and Kiehn, 2010a; Hägglund et al., 2010; Hinckley et al., 2005; Nishimaru et al., 2006; Zhong et al., 2010), however, the working concentra-

50

tions and active combinations of drugs must be considered as each amine exerts independent actions on locomotor activity (Beliez et al., 2014). In fact the selective activation of ionotropic glutamate receptors, differentially af-fect locomotor speed (Talpalar and Kiehn, 2010). Amines in combination with NMDA produce motor bursts with longer cycle duration, a stable rhythm and a stable f/e relationship, ideal for locomotor coordination studies (Beliez et al., 2014; Talpalar and Kiehn, 2010).

Figure 9. Fictive locomotion preparation and patterned motor output. A sche-matic of an isolated spinal cord with recording electrodes place on left (l) and right (r) lumbar (L)2 and 5 ventral roots (A). Application of locomotor drugs induces locomotor-like activity with alternation of motor bursts between l/r (rL2 Vs. lL2) and f/e roots (lL2 Vs. lL5) (B). Cycle period, burst and interburst are identified on rectified trace.

The experiments presented in this thesis investigate the contributions of the Renshaw cell and Dmrt3 populations to locomotor coordination with a focus on locomotor frequency, burst duration and the coherency between l/r and f/e motor outputs. Fictive locomotion was repeatedly induced under similar circumstances as deviations to the locomotor activity due to mismatches or variations in drug concentrations or variable experimental temperatures could falsely be perceived as phenotypic outcomes, potentially skewing the data. As such all experiments were completed as described in brief below:

The neonatal (<P3) spinal cord was freely dissected in ice cold, K-gluconate cutting solution (Dugué et al., 2005; Lamotte d’Incamps and Ascher, 2008) before being transferred to equilibrated (95% O2 and 5% CO2) artificial cer-ebrospinal fluid (aCSF) for 30 minutes prior to recordings and subsequently held at room temperature (22-24°C) throughout experimental procedures. Suction electrodes were placed on left and right lumbar (L)2 and L5 ventral roots. The following drugs were added to the perfusing aCSF and bath ap-plied to spinal cords to induce locomotor-like activity:

51

Paper II: 5 µM NMDA + 10 µM 5-HT induced baseline locomotion whilst 3, 6 and 10 µM NMDA + 10 µM 5-HT + 50 µM dopamine varied locomotor speed. Mecamylamine (50 µM), a nicotinic receptor antagonist was occa-sionally added to the perfusate. Paper III: 5 µM NMDA + 10 µM 5-HT + 50 µM dopamine induced base-line locomotion whilst 7.5 µM NMDA was substituted for experiments ad-dressing increased speed.

Recorded ventral root signals were amplified and band-passed filtered be-fore being digitized and recorded using Axoscope 10.2 (Axon Instruments Inc.) or WinEDR for analysis. Data was rectified, occasionally downsampled and low pass filtered. Analyses including coherence plots, preferential phase alignment, and burst parameters over at least 20 sequential activity cycles were performed using SpinalCore (version 1.1, Mor and Lev-Tov, 2007), Neurodata (Zhang et al., 2008) or an in-house designed Matlab program.

A considerable advantage of fictive locomotion is its adaptability and readiness to be combined with other techniques. In Paper III, fictive loco-motion was combined with 2-photon imaging to record the population activi-ty of the Dmrt3Cre neurons during locomotor-like activity and whole-cell patch-clamp techniques to record individual neural activity during locomo-tor-like patterns.

Patch-Clamp Electrophysiology The ionic current and resulting voltage change across a membrane can be directly measured through intracellular recording techniques. In the 1940’s, Kenneth Cole established the voltage-clamp technique where the effect of the membrane potential on membrane currents was recorded by holding or ‘clamping’ the neuron at predetermined voltages. This technique was suc-cessfully used by Hodgkin and Huxley (1952), in their pioneering experi-ments in the squid axon, where they revealed the ionic mechanisms underly-ing the action potential. Erwin Neher and Bert Sakmann in the late 1970’s to early 1980’s, further refined the voltage clamp method by adding a giga-seal thereby establishing the widely used patch-clamp technique (Neher and Sakmann, 1992; Sakmann and Neher, 1984) where the current through volt-age gated, ligand gated and mechanically gated ion channels could be rec-orded. The patch-clamp method works by electrically isolating a patch of membrane, through which ionic conductance can be measured, by creating a giga-seal between the membrane and a glass micropipette. Once the seal has been established one of four patch clamp configurations can be obtained, the most common of which is the whole-cell configuration (Figure 10). Whole-cell patch-clamp experiments record the electrical potentials and ionic cur-rents from the entire cell through two recording modes; current-clamp that records the membrane potential to current inputs and voltage-clamp that records membrane currents at any given membrane voltage.

52

Figure 10. Patch-clamp recording configurations. Adapted from Zhao et al., 2008

In Papers I–III, whole-cell patch-clamp recordings were made on Chrna2Cre, Dmrt3Cre and motor neurons in spinal cord slices. In Paper III, Dmrt3Cre neurons were patched in a dorsal-horn-removed spinal cord, often in combination with fictive locomotion. The electrophysiological protocol and tissue preparation procedures were fundamentally similar and are de-scribed in brief below: Neonates (P0-P10) were anesthetized by hypothermia (<P4) or oral inhala-tion of isoflurane (>P4), decapitated and spinal cord freely dissected in oxy-genated (95% O2 and 5% CO2), ice cold, K-gluconate cutting solution.

Dorsal-horn-removed experiments: The dorsal horn of the lumbar spinal cord was carefully removed from segments L2 – L5 leaving the ventral horn and roots intact. The spinal cord was incubated in oxygenated aCSF for 30 minutes before being transferred to the recording chamber, where a free ven-tral root was attached to a suction electrode. Fictive locomotion was induced as described above. After stable locomotor-like activity had been induced (>15 minutes) Dmrt3 neurons were patched (described below) either contra-laterally or ipsilaterally to the recorded ventral root.

Spinal cord slice electrophysiology: The spinal cord was angled (35°) on an agar block, secured with gelatin and placed in a pre-cooled slicing cham-ber filled with ice cold cutting solution. Transverse, lumbar, spinal cord sec-tions (350µm) were made using a vibrating blade microtome, incubated for 45 minutes at 35°C in equilibrated aCSF and subsequently held at room temperature (22-25°C) throughout experimental procedures. Neurons were visualised, by their stereotypical morphology or RFP expression. Ventral roots were occasionally mounted into glass suction electrodes for antidromic response to ventral root stimulation. Images were routinely taken of patched RFP expressing neurons often in combination with ventral root recording electrodes, to confirm lumbar segment, dorso-ventral and/or medio-lateral placement. Patch pipettes (5-9MΩ) contained a K+ based internal solution,

Cell-attached

Whole-cell

Inside-out

Outside-out

<

<<

53

with an osmolarity between 280-300mOsm/L. The liquid junction potential was 14.4mV.

Whole-cell patch-clamp recordings were filtered and digitized using WinWCP software (Dr. J. Dempster, University of Strathclyde, Glasgow, UK) or Axograph data acquisition and analysis software (Sydney, Australia). Additional analysis was performed using Matlab and Prism. For current clamp recordings depolarising and hyperpolarising current protocols were applied. For voltage clamp recordings, neurons were voltage clamped at -60mV and access resistance was compensated by >66% prior to the applica-tion of voltage steps. Drugs added to the perfusate were ZD7288 20µM (Pa-per I and III), Apamin 100 nM, BaCl2 50 µM, CsCl 3, 9 and 12 mM (Paper I) Picrotoxin 10 µM, and Tetrodotoxin 1µM (Paper II).

Retrograde Tracing Retrograde tracing is a powerful method to identify the axonal and dendritic projections and pathways of neurons within neural networks. Retrograde tracing works by injecting one or two fluorescent retrograde tracers into a target site or sites, where cells that project to the targeted region are identi-fied by the presence of tracer in their cell body. The flexibility and ease of combining tracing techniques with immunohistochemical and physiological recordings makes retrograde tracing advantageous in many studies, but given specific tracer characteristics, selecting the appropriate tracer should be con-sidered. Although the percentage of traced neurons is invariably quite high it is not absolute and often underestimates the prevalence of projections, as such procedural methods exist to improve retrograde labelling (review see Schofield, 2008).

In this thesis, retrograde dextran tracing was used to identify commissural ascending, descending and bifurcating axonal projection patterns of spinal Dmrt3Cre;tdT neurons. Dextran tracers are advantageous as they are readily viewed on fluorescent and confocal microscopes and are highly sensitive, often extensively labelling neural dendrites (review see Schofield, 2008). Any severed axon will pick up dextran tracers. As such, to avoid tracing ipsilateral neurons it was imperative the spinal cord midline was kept intact throughout dissection and commissural tracing procedures and superfluous tracer was immediately removed. The retrograde tracing procedures are de-scribed in brief below:

Neonatal spinal cords (P0/P1) were dissected (see Electrophysiology) and transferred to a chamber filled with oxygenated aCSF at room temperature (22 -24°C). Two horizontal cuts were made in the spinal cord; one at L1 and a second between L3 and L4. Fluorescein-dextran-amine (FDA, 3000 MW) and Alexa fluor 647 (10,000 MW) were applied to L1 and L3/4 cuts respec-

54

tively and spinal cords were left to incubate in the dark overnight. Traced cords were fixed for one week before transverse sectioning (60 µm).

Some dextran tracers commercially available as conjugates with biotin (biocytin) are injected into neurons during electrophysiological recordings linking electrical characteristics to morphology. In Papers I and III, Chrna2Cre;tdT and Dmrt3Cre;tdT patched neurons were readily filled with biocytin (2mg/ml) during recording protocols and their morphology was reconstructed post hoc using the avidin-biotin reaction: spinal cord slices were fixed overnight and processed with Streptavadin-Alexa-Fluor® 488 (1:1000) together with 4’,6-diamidino-2-phenylindole (DAPI), before being washed and mounted in Mowiol. Images for both retrograde tracers and bio-cytin-streptavidin were acquired on a fluorescence or confocal microscope and processed in Velocity software (Improvision, Coventry, UK), ImageJ and Adobe Photoshop.

Behaviour Assays Genetic targeting is a common way to identify genes important in cellular and physiological processes that underlie behaviour. As genetic targeting becomes more specific the need for a well-developed and characterised set of behavioural assays arises that account for a plethora of behavioural phe-notypes. In the mouse, behavioural paradigms have been adopted from the rat and are continually adapted and improved by contributions from psy-chology, pharmacology and neuroscience. Although observable behaviours, both normal and abnormal, have underlying cellular, molecular or physiolog-ical causes, it can often be difficult to determine cause and effect relation-ships due to extraneous variables impacting animal behaviour or behavioural interpretations. Thus it is challenging, yet imperative, to choose the correct behavioural paradigms to observe and test, robust and reliable mouse behav-iour (review see Bućan and Abel, 2002).

When examining mouse behaviour the limitations and physiological chal-lenges presented by the test and the animal must be considered. This is par-ticularly relevant to motor tests as motor coordination and strength develop with age, stressing the importance of examining these parameters in compa-rable age groups. Although treadmill tests are an excellent means of as-sessing adult locomotion, they are relatively ineffective in determining the walking capacities of pups as gravity overrides their muscular strength. Swimming and suspended ‘air stepping’ assays instead allow the unburdened execution and analysis of neonatal locomotor behaviour, which facilitate the study of locomotion throughout development.

This thesis uses various motor behavioural assays, including treadmill lo-comotion, grip strength, beam walking, air stepping and swim tests to eluci-date the role of Chrna2Cre and Dmrt3Cre populations in motor output. The

55

procedures for treadmill locomotion and swim tests are described in brief below:

Adult animals were placed on the treadmill band moving at different speeds (9, 15, 20, 25 and 30 cm/s) and their steps during a 20 second period were recorded. Gait parameters were automatically analysed using Treadscan software (CleverSys Inc) or calculated from 10 manually tracked stride cy-cles. Treadmill locomotion at faster speeds (40, 45 and 50 cm/s) was per-formed by animals (P26-P28) that were trained between P14-P26 using a training schedule devised by Lemieux et al. (2016) with slight modifications. The proportion of synchronous fore- and hindlimb steps were quantified for each treadmill speed.

Swim patterns were assessed in adults and pups (P2, P4, P8 and P16). An-imals were allowed to swim freely in a chamber proportionate to animal size for a maximum time of 30 seconds. Mice were videotaped from the side and from below to analyse postural development, swim style and movement patterns. Limb patterns (alternating, uncoordinated or immobile) were as-sessed from hindlimb movements during 20 full stroke cycles.

56

Results and Discussion

Paper I Deciphering a functional role for the Renshaw cell in locomotion has, over the past decades, been challenging. Their function remains poorly defined due, at least in part, to their central position within intricately assembled networks of CPG interneurons and motor neurons. Thus, the use for a specif-ic Renshaw cell marker that can functionally isolate the Renshaw cell popu-lation is needed.

A microarray screen of the mouse spinal cord, identified the cholinergic nicotinic receptor alpha 2 (Chrna2) could be a potential Renshaw cell mark-er due to the high expression of Chrna2 within the most ventral horn (Enjin et al., 2010). To selectively target Renshaw cells, we generated a transgenic mouse line where Cre, driven by the Chrna2 promoter, was expressed in Chrna2 positive cells. Chrna2Cre positive neurons were visualised by cross-ing the Chrna2Cre mouse with various reporter lines. Chrna2 was specifically expressed in a subpopulation of ventral horn neurons with an anatomical location and pattern of expression reminiscent of Renshaw cells. Further validation using the known Renshaw cell immunohistochemical markers calbindin and gephyrin, and the generation of an evoked response through ventral root stimulation, confirmed Chrna2 expressing neurons in the most ventral horn are Renshaw cells. CHRNA2 as a membrane bound receptor, is expressed early (E11) in cells during embryogenesis and maintained within the cellular membrane throughout cell life (Hogg et al., 2003). The early and continued expression of cholinergic receptors in the membrane combined with the high overlap between Chrna2-mRNA and calbindin makes Chrna2 an attractive, reliable and persistent marker for Renshaw cells, since Cre will continually be expressed as long as Chrna2-mRNA is transcribed.

The identification of a Renshaw cell specific marker permitted, for the first time, a thorough investigation of Renshaw cell electrophysiological properties without the need for post hoc confirmation (Alvarez and Fyffe, 2007). As such, we patched neonatal (P0-P10) Renshaw cells from Chrna2-Cre/R26Tom mice (RCα2). We initially confirmed the presence of reported characteristic Renshaw cell firing properties in RCα2 (Eccles et al., 1954; Lamotte d’Incamps and Ascher, 2008). Furthermore, we established the fun-damental electrophysiological profile including both passive (input re-sistance and resting membrane potential) and active (action potential charac-teristics) properties for the RCα2 population, which had been absent in previ-

57

ous reports. Interestingly, upon strong hyperpolarisation of the RCα2 mem-brane, a large inward rectification or depolarising ‘sag’ was observed in neu-rons and often, post-inhibitory rebound potentials were generated at the ter-mination of hyperpolarising steps. These features, stereotypical of the hy-perpolarisation-activated cation current (Ih) were abolished when Ih was blocked by the Ih antagonist ZD7288 (Marshall et al., 1993; Wu et al., 2012). These findings prompted the investigation of the specific ionic contributions to RCα2 membrane properties, in particular Ih and the small conductance, calcium-activated potassium current (ISK).

RCα2 have confirmed expression of HCN4 channels, (Hughes et al., 2013), which we found to conduct a functional, prominent Ih that was signif-icantly attenuated when blocked by ZD7288. The activation time constant for RCα2 Ih was similar to previous studies investigating the kinetics of HCN4 channels (Ishii et al., 1999; Seifert et al., 1999) supporting the expres-sion of HCN4 in Renshaw cells. HCN2 channels are also abundantly ex-pressed in the ventral horn (Santoro et al., 2000) and it is possible that Ih in RCα2 could be conducted through both HCN2 and HCN4 channels, although the co-expression of HCN2 and HCN4 in RCα2 is yet to be determined. Ih has diverse effects on neuronal membranes and amongst other roles, is implicat-ed in establishing the resting membrane potential, participates in rhythmic firing and modulates neuronal excitability (Pape, 1996). Given the strikingly negative voltage at which Ih is activated in postnatal RCα2, a role of Ih in setting the RCα2 resting membrane potential was ruled out. The kinetic nature of Ih, promotes an inherent cellular rhythmicity shifting the membrane poten-tial from depolarisation to hyperpolarisation (Pape, 1996), and as Renshaw cells are rhythmically active during locomotion (McCrea et al., 1980; Pratt and Jordan, 1987), it was likely that Ih contributed to Renshaw cell firing patterns. Although we did not observe burst firing in RCα2 as previously reported (Nishimaru et al., 2006) blocking Ih decreased the firing frequency of RCα2. Since Renshaw cells match motor neuron firing, often firing at a frequency in excess of 50 Hz (Moore et al., 2015) and HCN4 channels are suggested to be linked to high action potential discharge (Hughes et al., 2013), we suggest that Ih in RCα2 contributes to high frequency firing in Ren-shaw cells to tightly regulate motor neuron output.

Mature Renshaw cells have distinctive, large gephyrin clusters on their soma and proximal dendrites that co-localise with both glycine and GABAA receptors (Alvarez et al., 1997; Geiman et al., 2000, 2002) suggesting that Renshaw cells receive potent inhibition from other neurons. In support of this, the half-activation potential for Ih in RCα2 was strongly negative, pro-posing that Renshaw cells have a mechanism in place to deal with the in-coming powerful inhibition. HCN4 channels, more abundant in the spinal cord than first thought, have the slowest activation and inactivation kinetics (Santoro and Baram, 2003; Santoro et al., 2000), which can lead to the gen-eration of rebound action potentials at the termination of hyperpolarising events. We postulate that the activation of RCα2 Ih at strongly negative poten-

58

tials, by the simultaneous convergence of multiple inhibitory inputs, gener-ates rebound action potentials turning strong inhibition into rebound excita-tion. Thus, inhibition of the Renshaw cell through Ih activation, could result in motor neuron inhibition, or at the very least a disturbance in motor neuron firing, as one Renshaw cell action potential is enough to disrupt motor neu-ron activity (Bhumbra et al., 2014). This scenario is supported for Renshaw cells at rest, however, in an active recurrent inhibitory network with a push/pull of excitatory and inhibitory inputs, inhibition of the Renshaw cell would primarily result in disinhibition of the motor neuron, increasing motor neuron firing. We suggest Ih acts to modulate RCα2 neuronal excitability by regulating firing frequencies and potentially turning strong inhibition into rebound excitation. We reveal a potential dual action of Renshaw cell activi-ty, which can both inhibit and activate neurons, adding to the complexity of Renshaw cell function during locomotion. This potential dual action be-comes even more complicated when considering the vast circuitry estab-lished between Renshaw cells, motor neurons and other locomotor and sen-sory neurons.

RCα2 firing can also be modulated by the small calcium-activated potassi-um current (ISK). RCα2 firing frequency significantly increased when Apamin was applied to cells. Apamin, a specific ISK antagonist, blocked the action potential afterhyperpolarisation (AHP) significantly reducing interspike in-tervals and increasing RCα2 spike rate. The AHP modulates neuronal firing patterns, where blocking ISK can shift firing from tonic to bursting (Edgerton and Reinhart, 2003). Thus in RCα2, ISK limits the firing frequency and poten-tially prevents Renshaw cells from burst firing under certain stimuli (McCrea et al., 1980). ISK channels are activated by intracellular calcium levels, thus the stronger the activation of Renshaw cells e.g. during pro-longed depolarisation, the more ISK will act to regulate firing. Under normal circumstances, ISK has the potential to change the pattern of Renshaw cell firing and appears to limit Renshaw cell activity in response to increased activity and could possibly act as protection from over-excitation.

Together, we reveal two previously unidentified RCα2currents that inde-pendently act to modulate activity through different activation mechanisms; one by strong inhibition and the other by profound excitation. The interplay of these currents with other ionic conductances and the possible implications this has on both Renshaw cell and motor neuron excitability remains to be investigated.

Although Chrna2 is a specific Renshaw cell marker, Chrna2 marks addi-tional neural populations found within recurrent inhibitory circuits in the hippocampus and cortex with similar specificity (Leao et al., 2012). In the hippocampus, the expression of Chrna2 defines the Oriens/Lacunosum-moleculare (OLM) interneuron population (Leao et al., 2012; Mikulovic et al., 2015), which contribute to oscillatory rhythms and are involved in anxie-ty behaviours (Mikulovic et al., submitted) and memory formation (França et al. submitted). In the cortex, Chrna2 defines the Martinotti cell population

59

(Hilscher et al., submitted) that contributes to correlating, synchronising and coordinating the output of cortical pyramidal cells (Berger et al., 2010; Silberberg and Markram, 2007). In the ventral spinal cord, hippocampus and cortex, Chrna2+ inhibitory interneurons have molecular, synaptic (circuitry) and electrophysiological similarities and are accountable for supressing, modifying or gating the output of a primary neuron. The similarities between the Renshaw, OLM and Martinotti cells and their signalling circuits are be-ing considered in an upcoming review (Hilscher, Mikulovic, Perry and Kullander, in preparation). Additionally, within the spinal cord Chrna2 is expressed in dorsal interneuron populations and marks a population of cur-rently unidentified, midline ventral horn neurons.

The Chrna2-Cre/R26Tom mouse has made it possible to visualise and rec-ord from Chrna2 expressing populations in whole spinal cord and brain preparations. Thus we can extrapolate our findings from single cells to fit population based analyses. This has been particularly advantageous in inves-tigating the electrical and functional properties of the RCα2 population within an isolated spinal cord using 2-Photon imaging population-based recording techniques (Nagaraja et al., submitted). Moreover, we have used Chrna2-Cre/R26Tom mice in CLARITY techniques (Chung and Deisseroth, 2013) for high resolution imaging, visualising the density, distribution and projections of Chrna2 populations within whole CNS structures.

Even though Renshaw cell research has persisted since their discovery and significant contributions have been made to the Renshaw cell profile, the role of the Renshaw cell in motor control remains elusive. The Chrna2Cre mouse is the first example of a Renshaw cell genetic marker and broadens the possibilities to decipher the functional role of Renshaw cells in locomo-tion and to investigate other Chrna2 expressing populations throughout the CNS. The focus of Paper II is to address the functional role of Renshaw cells in locomotion using Chrna2Cre mice

Paper II This work extended the characterisation of RCα2 and used the Chrna2Cre mouse to specifically remove VIAAT-mediated inhibitory signalling from the Renshaw cell population (Chrna2Cre;Viaatlx/lx). For the first time, we spe-cifically silenced Renshaw cells within locomotor circuits and investigated their functional role during locomotion. Previous studies have only targeted the V1 population that includes both Renshaw cells and IaINs, and although locomotor disturbances were observed when V1 interneurons were ablated, primarily to the speed of locomotion and difficulties generating limb exten-sion, they could not be directly attributed to the role of the Renshaw cell (Britz et al., 2015; Gosgnach et al., 2006).

Surprisingly, we found no observable phenotypic behaviours in adult Chrna2Cre;Viaatlxlx mice, which contradicts previous results (Britz et al.,

60

2015; Gosgnach et al., 2006). Removing Renshaw cell mediated inhibition during development did not impact motor behaviour as all tested aspects of motor functioning including gait, coordination, strength, and balance in Chrna2Cre;Viaatlx/lx mice were similar to control animals. We cannot exclude the possibility that silencing Renshaw cell function influenced aspects of motor behaviour not observable by the behavioural assays in this study, alt-hough given the extent of testing, this seems unlikely.

With no clear behavioural phenotype in adults, locomotor patterns were induced and examined in neonates. In keeping with adult results, the fre-quency and rhythm of the induced locomotor-like activity were not affected by the removal of Renshaw cell inhibition in Chrna2Cre;Viaatlxlx mice, nor were hindlimb coordination or alternation patterns altered suggesting that CPG circuits in Chrna2Cre;Viaatlxlx mice function normally from birth. These results do not support previous theories implicating Renshaw cells in the regulation of flexor-extensor motor neuron activity (McCrea et al., 1980; Nishimaru et al., 2006) or support the finding that mice with ablated V1 neurons had flexor-extensor related difficulties with gait (Britz et al., 2015). Our findings do however, support the previous observation that flexor-extensor alternation remains unaffected when the V1 population is genetical-ly ablated or silenced (Gosgnach et al., 2006; Zhang et al., 2014).

In an attempt to provoke the locomotor circuit Renshaw cell activation was blocked by the application of mecamylamine, a nicotinic receptor an-tagonist (Sapir et al., 2004), that should alter the locomotor frequency in control cords (Nishimaru et al., 2006), but not affect the frequency of Chrna2Cre;Viaatlxlx cords. Interestingly, mecamylamine reduced the locomo-tor frequency of Chrna2Cre;Viaatlxlx mice by a similar magnitude to control mice, suggesting the actions of mecamylamine are independent of Renshaw cell function and possibly act on other CPG neurons. There is little literature surrounding the actions of mecamylamine on spinal cord neurons other than Renshaw cells, but potential targets could include other populations express-ing Chrna2 (Chrna2-Cre/R26Tom midline population) or other nicotinic ace-tylcholine receptor subtypes (e.g. Chrna5). As a counter approach to explore a possible Renshaw cell functional role, fictive locomotor speed was in-creased. Locomotor-like activity similarly increased in both control and Chrna2Cre;Viaatlxlx cords, suggesting that Renshaw cells are not central to the regulation or coordination of locomotor patterns across speeds contrary to previous reports for the V1 population (Gosgnach et al., 2006). Together these results infer that, perhaps, Renshaw cells are entirely dispensable for locomotion as removing their inhibition does not impact functional locomo-tor parameters.

In light of these findings, the efficiency of the VIAAT deletion in Chrna2Cre;Viaatlxlx mice is questioned, although there is considerable evi-dence arguing for the successful elimination of VIAAT in other neurons (Leao et al., 2012; Rahman et al., 2015; Wojcik et al., 2006). Viaat mRNA was absent from all calbindin+ ventral horn neurons in Chrna2Cre;Viaatlxlx

61

mice, and present in all calbindin+ ventral horn neurons in control mice. Fur-thermore, Chrna2Cre;Viaatlxlx motor neurons received a lower frequency of miniature inhibitory post synaptic currents than control mice, which physio-logically supports a VIAAT deletion in Renshaw cells. Although unlikely, we cannot entirely exclude the possibility that inhibitory neurotransmitters may be released via mechanisms independent of VIAAT-mediated release. To equivocally confirm functional VIAAT deletion in Chrna2Cre;Viaatlxlx mice, antidromic stimulation of recurrent inhibitory circuits, should not de-press the motor neuron response to afferent stimulation (Eccles et al., 1954). These experiments are currently underway.

With an absence of behavioural phenotype, we investigated whether there were synaptic or electrophysiological alterations to the recurrent inhibition circuit. Previous alterations in the recurrent inhibitory circuit were reported in Engrailed 1 mutant mice, where although Renshaw cell inhibitory input to motor neurons decreased by almost half, the total number of inhibitory syn-apses remained normal, suggesting increased innervation by other inhibitory neurons (Sapir et al., 2004). VIAAT-mediated Renshaw cell signalling influ-enced the development of inhibitory and sensory synapses as Chrna2Cre;Viaatlxlx motor neurons had significantly increased calbindin+ Renshaw cell contacts, coupled with an increased number of proprioceptive glutamatergic sensory VGLUT1+ contacts (Figure 11). Interestingly, the altered synaptic innervation of Chrna2Cre;Viaatlxlx motor neurons did not translate to altered synaptic function, as Chrna2Cre;Viaatlxlx motor neurons received a lower frequency of spontaneous inhibitory currents and main-tained normal stretch reflex responses. The normal stretch reflex despite increased sensory innervation suggests changes to the electrical properties of the circuit.

Removing VIAAT-mediated Renshaw cell signalling impacted the devel-opment of motor neuron and Renshaw cell electrical properties. Chrna2Cre;Viaatlxlx motor neurons and Renshaw cells had significantly al-tered excitability parameters and exhibited changes in their afterhyperpolari-sation potentials, that slowed their firing frequencies. Chrna2Cre;Viaatlxlx motor neurons required more current to fire an action potential, showing a hypoexcitable phenotype, whilst Renshaw cells required less excitation showing a hyperexcitable phenotype (Figure 11). Interestingly, when com-paring the relative changes between Chrna2Cre;Viaatlxlx motor neurons and Renshaw cells, in many properties, the percentage change in Chrna2Cre;Viaatlxlx motor neurons was matched by an opposite change in Chrna2Cre;Viaatlxlx Renshaw cells. Thus, there were significant electrophysi-ological changes to motor neurons and Renshaw cells as a consequence of VIAAT deletion, and in the majority of altered parameters, the changes were reciprocal.

62

Figure 11. Summary of the synaptic and electrophysiological compensatory changes in the recurrent inhibition circuit in Chrna2Cre;Viaatlxlx mice. Chrna2Cre;Viaatlxlx motor neurons (black, right) had increased calbindin+ (Calb+, red) and afferent (VGLUT1+, green) contacts compared to control motor neurons (left). These synaptic changes were accompanied by altered electrical properties where motor neurons became hypoexcitable whilst Renshaw cells (red) became hyperex-citable. ChAT+ (choline acetyltransferase) and VAChT+ (vesicular acetylcholine transporter, grey) contacts remained constant.

The absence of behavioural deficiencies, combined with the altered synaptic and electrophysiological properties of Chrna2Cre;Viaatlxlx mice, suggest that the physiological role of Renshaw cells is compensated for during network development. The expression of Cre removes VIAAT from Renshaw cells early during embryogenesis, during a time in which GABA and glycine are presumably excitatory (Ben-Ari, 2002; Ben-Ari et al., 2007). As such, the corresponding effects and potential changes to the recurrent inhibition circuit are likely due to a reduction in excitation rather than a reduction in inhibi-tion. Compensation within developing networks is a consequence of early neural and chemical alterations to re-establish normal physiological func-tions to keep them within defined levels (Banks et al., 2005; Fogarty et al., 2016). In a similar scenario to Chrna2Cre;Viaatlxlx motor neurons, Fogarty et al. (2016) observed that motor neurons with a specific deletion of gephyrin received decreased inhibition, which was not only coupled with a two-fold increase in excitation, but both inhibitory and excitatory currents had altered functional properties. These results are similar to our observations, where the remaining inhibitory currents to Chrna2Cre;Viaatlxlx motor neurons had al-tered rise times.

We saw several, independent pieces of evidence that suggest there is pre-sumable, but considerable compensation within the recurrent inhibition cir-cuit in Chrna2Cre;Viaatlxlx mice, rendering behavioural output normal (Figure 11). Chrna2Cre;Viaatlxlx motor neurons received reduced inhibition, detected as a lower frequency of inhibitory currents, which the Renshaw cells alt-

Motor NeuronRenshaw Cell

Calb+

ChAT+ChAT+

VGLUT1+VAChT+

Control Chrna2Cre;Viaatlx/lx

Calb+

VGLUT1+

ChAT+ChAT+

>

Excitability Excitability

VAChT+

>

63

hough silenced, could possibly attempt to counteract by increasing the num-ber of Renshaw cell contacts to Chrna2Cre;Viaatlxlx motor neurons. The de-creased excitability of neonatal Chrna2Cre;Viaatlxlx motor neurons was poten-tially balanced by an increase in the number of excitatory proprioceptive contacts projecting to motor neurons, possibly attempting to provide the additional excitation the motor neurons need to activate. In turn, the de-creased excitability in motor neurons was matched by an increase in the ex-citability of Chrna2Cre;Viaatlxlx Renshaw cells. This complementary compen-sation between motor neurons and Renshaw cells, could underlie the normal locomotor output in Chrna2Cre;Viaatlxlx mice. Our results highlight that the spinal cord recurrent inhibitory circuit is endowed with several mechanisms to maintain a purposeful output, and promote the further analysis of Ren-shaw cell function using acute silencing methods.

There has been much speculation about the role of the Renshaw cell in motor control, and our results neither fully support nor entirely negate exist-ing evidence. In fact, if anything, our results argue that Renshaw cells could be entirely dispensable for locomotor control as the developmental silencing of the Renshaw cell population still renders entirely functional behavioural output, most likely through complementary compensation mechanisms.

Paper III Dmrt3 knock out animals (Dmrt3-/-) illustrate the need for a functional Dmrt3 gene in the development and generation of coordinated locomotor patterns controlling limb movements (Andersson et al., 2012). In the mouse spinal cord, Dmrt3 defines a subset of inhibitory dI6 interneurons located around the central canal in laminae VII and VIII that are primed to contrib-ute to locomotion. Whilst previous studies have evaluated the properties of putative subsets of dorsal dI6 neurons (Dyck et al., 2012) or the Dmrt3 gene (Andersson et al., 2012), this study selectively targets the Dmrt3 derived population to determine the role of Dmrt3 neurons in the locomotor CPG.

Genetically targeted Dmrt3 neurons identified by RFP expression (Dmrt3Cre;tdT) were confirmed as the Dmrt3 population, due to their high expression of Dmrt3 mRNA. Some Dmrt3 neurons also expressed WT1, indicating diversity in their neural background. The dI6 and neighbouring V0 population can be further subdivided based on their unique intrinsic, functional and anatomical properties (Dyck et al., 2012; Lanuza et al., 2004). Whilst the majority of Dmrt3 cells were located in lamina VII, the neuronal distribution spanned across both laminae VII and VIII, where more ventral Dmrt3 cells were morphologically larger than dorsal Dmrt3 cells. Unlike the V0 and V3 populations (Borowska et al., 2013; Lanuza et al., 2004), dorsal and ventral Dmrt3 cells do not represent functionally distinct subgroups, as membrane and action potential properties were similar between cells. The dI6 population reportedly projects both contralaterally and ipsilaterally

64

(Andersson et al., 2012; Lanuza et al., 2004; Rabe et al., 2009). We con-firmed Dmrt3 neurons are inhibitory, commissural interneurons whose axons bifurcate and extend intersegmental ascending and descending projections to target motor neurons and other inhibitory interneurons.

As a population Dmrt3 neurons had fundamentally similar action poten-tial properties, but showed diversity in the occurrence of spontaneous firing, frequency adaptation and active membrane currents. Considerable spike-frequency adaptation in response to depolarisation was seen in Dmrt3 neu-rons, which became more prominent at higher current inputs. Almost all (90%) neurons exhibited spike adaptation at 150pA whilst only 40% of Dmrt3 neurons adapted their firing frequency at 50pA, revealing a small proportion of the Dmrt3 population fired regularly across all current inputs. Adaptation is found in rhythmically firing neurons and although not a unique feature of the Dmrt3 population, adaptation could contribute to their physio-logical and functional role. We suggest that spike-frequency adaptation in Dmrt3 neurons could potentially act to maintain a pre-set level of inhibition onto motor neurons in response to increasing excitatory inputs. Thus, with a higher locomotor drive, Dmrt3 neurons permit CPG excitation to override inhibition, which could promote synchronous ‘bounding’ gaits at higher locomotor speeds (Lemieux et al., 2016).

Many rhythmically active CPG interneurons have electrophysiological properties, including adaptation and hyperpolarisation-activated cation cur-rents (Ih) that support their inherent rhythmical firing (Borowska et al., 2013; Butt et al., 2002; Dougherty and Kiehn, 2010a; Kiehn et al., 2000; Perry et al., 2015; Wilson et al., 2005; Zhong et al., 2010). Ih currents, present in 76% of Dmrt3 neurons, generated stereotypical depolarising ‘sags’ and repolar-ised membranes from hyperpolarised potentials. Upon termination of hy-perpolarisation, half the Dmrt3 neurons responded with rebound action po-tentials. Ih was not present in all Dmrt3 neurons suggesting possible differ-ences in ion channel expression and current between cells, which propose the existence of subgroups. The expression of HCN channels, or a detailed characterisation of Ih functionality has not been investigated in the Dmrt3 population and would be of potential interest to classify neurons into poten-tial subsets particularly since Ih exerts diverse effects on neural membranes (Pape, 1996). In any case, Dmrt3 neurons have adaptive firing and Ih, two features that support inherent rhythmical activity.

Previous studies of inhibitory CPG interneurons reported that neuronal firing patterns were heterogeneous during locomotor activity (Dyck et al., 2012; Wu et al., 2011). Our findings support these results as only 63% of Dmrt3 neurons had rhythmic activity that was either in or out of phase with the locomotor pattern recorded through the ventral root. The remaining Dmrt3 neurons showed no discernable rhythm and fired continuously throughout locomotor activity in accordance with previous reports (Dyck et al., 2012; Wu et al., 2011). As a population Dmrt3 firing activity was not correlated to the locomotor rhythm, even though some neurons were strongly

65

coupled to motor bursts. These findings corroborated the multiple Dmrt3 rhythmicities seen in 2-photon imaging experiments. The variable firing patterns during locomotor activity could likely arise from variations in the firing and membrane properties of Dmrt3 neurons especially since subsets of the dI6 population had persistent inward currents (PIC) that supported their intrinsic rhythmogenic properties (Dyck et al., 2012). The specific current contributions to individual firing patterns were not investigated in any neu-rons, as such it is not possible to conclusively link neural firing to defined cellular electrical properties. It is plausible, however, that Dmrt3 neurons exhibiting both spike adaptation and active Ih would more likely be generat-ing a rhythmic output.

To ascertain the locomotor role of the Dmrt3 neurons themselves, Dmrt3 neurons were genetically stripped of their ability to inhibit downstream pop-ulations (Dmrt3Cre;ViaatKO mice). In an isolated preparation, Dmrt3Cre;ViaatKO cords had markedly slower and considerably disturbed locomotor patterns that transitioned between periods of left-right and flexor-extensor alternation and synchrony and showed variable burst parameters similar to previous findings (Crone et al., 2008; Lanuza et al., 2004; Zhang et al., 2008). Increasing the locomotor speed further augmented left-right locomotor pattern abnormalities, suggesting Dmrt3 neurons could be in-volved in speed dependent coordination of locomotion.

Defined populations of CPG interneurons are implicated in segmental left-right alternation of locomotion. Selective elimination or preservation of these neurons including the V2a, V3, and V0 populations in mice, resulted in abnormal synchronous hindlimb ‘hopping’ gaits (Bellardita and Kiehn, 2015; Crone et al., 2009; Kullander et al., 2003; Rabe et al., 2009; Talpalar et al., 2013), some of which had strong speed-dependent modulation of lo-comotion (Bellardita and Kiehn, 2015; Crone et al., 2009; Talpalar et al., 2013). Dmrt3Cre;ViaatKO mice had impaired hindlimb coordination and dis-played an abnormal gait, taking considerably more synchronous hindlimb steps, during motor tasks. During increased treadmill locomotion, Dmrt3Cre;ViaatKO mice sequentially increased their number of synchronous hindlimb steps until they switched to using a predominantly bounding gait, falling into both fore- and hindlimb synchrony at speeds above 35cm/s (Fig-ure 12). This shift to fore- and hindlimb synchrony is seen in control mice running at speeds above 75cm/s (Lemieux et al., 2016), suggesting Dmrt3 neurons contribute to maintaining left-right coordination more prominently at higher speeds. During high speed running, Dmrt3Cre;ViaatKO mice had shorter stance (extensor), but longer swing (flexor) phases during a single locomotor stride, similar to Dmrt3-/- null mice (Andersson et al., 2012), sup-porting the role of Dmrt3 neurons in flexor-extensor coordination. Similar to Dmrt3Cre;ViaatKO mice, DMRT3 was important for fore- and hindlimb coor-dination in horses where mutations in DMRT3 is permissive for a horse’s ability to remain in trot at high speeds rather than shifting to gallop (Andersson et al., 2012). This observation is in keeping with recent results in

66

the mouse, where the selective ablation of V0v populations removed the ability of mice to shift from walking into trot (Bellardita and Kiehn, 2015).

Figure 12. Dmrt3Cre;ViaatKO mice display fore- and hindlimb synchrony at high-er locomotor speeds. Still images taken during treadmill locomotion at 50cm/s showing the footfall pattern of control (left) and Dmrt3Cre;ViaatKO (right) mice.

In quadrupeds, a balance between excitation and inhibition is necessary to produce alternation between left and right motor pools (walking, trot) until such a point, that excitation naturally overrides inhibition and there is a shift to synchronous limb movements to support a faster gait (bound, Bellardita and Kiehn, 2015). More specifically, all gaits depended on the balance be-tween commissural pathways supporting left-right alternation and those sup-porting synchrony (Danner et al., 2016; Rybak et al., 2015). Thus in the mouse the switch from alternation to synchrony appears to be influenced by the level of inhibition originating from commissural populations (Paper III, Bellardita and Kiehn, 2015; Danner et al., 2016; Talpalar et al., 2013) and upstream excitatory ipsilateral populations (Crone et al., 2009) that indirectly facilitate inhibition. The firing adaptation of the Dmrt3 population could act to gate excitatory inputs, such that in collaboration with other interneurons, the Dmrt3 population balance the excitation within the CPG. Previous find-ings suggest the majority of rhythmic neurons within dorsal dI6 subsets con-tribute to locomotor rhythm generation, whilst the minority contribute to pattern formation (Dyck et al., 2012). The behavioural findings of Paper III, combined with previous suggestions positioning the dI6 neurons within CPG models (Danner et al., 2016; Dougherty and Kiehn, 2010b), imply the Dmrt3 neurons are positioned within both the left-right/flexor-extensor pattern-formation and speed dependent circuitry, contributing minimally, if at all, to rhythm generation.

There have been many CPG models proposing the circuitry between ex-citatory rhythm generating inputs and the ventral interneuron populations modulating motor neuron activity. Dmrt3 neurons, as a newly described population, are not exclusively included within existing models (Dougherty and Kiehn, 2010b; Kiehn et al., 2010; Rybak et al., 2015). Thus, we consider how the Dmrt3 population might fit within the CPG circuitry and reflect on the possible connections of Dmrt3 neurons with other locomotor neurons.

Control Dmrt3Cre;ViaatKO

67

Dmrt3 neurons are activated, inhibited and modulated by synaptic inputs from largely unknown origins. Dmrt3 neurons do receive excitatory afferent input, however, the origin of other excitatory inputs mainly derived from VGLUT2+ terminals can only be speculated upon. These excitatory inputs may arise from the rhythm-generating core, possibly from Shox2 interneu-rons (Dougherty et al., 2013); upstream V2a interneurons that drive commis-sural interneurons (review see Dougherty and Kiehn, 2010b), or other cur-rently unidentified local excitatory inputs. Monoaminergic and cholinergic modulatory input, is presumably arising from descending projections origi-nating in monoaminergic nuclei in the brain or local cholinergic partition cells. Unlike the Renshaw cells, cholinergic input to the Dmrt3 population is not arising from motor neuron collaterals (Eccles et al., 1954). Local inhibi-tory interneurons or inhibitory cortical Dmrt3 neurons could provide the inhibitory input to the population. Dmrt3 neurons innervate contralateral motor neurons and project to Renshaw cells, however, Dmrt3-RC connec-tions are not confirmed. Other interneuron populations innervated by Dmrt3 neurons have not been investigated neither has the possibility that the Dmrt3 population inhibits itself, which could be a source of inhibitory inputs to Dmrt3 neurons.

The presumable dual commissural and ipsilateral Dmrt3 projections chal-lenge the position of Dmrt3 neurons in CPG models. We confirmed contrala-teral Dmrt3 projections, however, given the flexor-extensor locomotor dis-turbances in Dmrt3Cre;ViaatKO mice, ipsilateral Dmrt3 projections likely exist. Whether a single Dmrt3 neuron projects both ipsi- and contralaterally, or whether there exist separate ipsilateral and contralateral populations, re-mains to be investigated. This influences our interpretation of Dmrt3 circuit-ry and presents multiple scenarios on how Dmrt3 neurons contribute to lo-comotion. Currently, no other locomotor interneuron populations are report-ed to have dual left-right and flexor-extensor roles, however, the V3 popula-tion which act to balance locomotor output, have a small proportion of ipsilateral projections (Zhang et al., 2008). Nonetheless, the left-right/flexor-extensor function of the Dmrt3 population combined with both the heteroge-neous firing patterns during locomotor activity and current contributions, suggests heterogeneity within the Dmrt3 neurons that could possibly repre-sent subpopulations. As such, deciphering the Dmrt3 circuitry is of consider-able interest to the functioning CPG.

The findings from Paper III provide a morphological, cellular and behav-ioural profile of the Dmrt3 neurons, however, their heterogeneous morphol-ogy, axonal projections, rhythmic firing and definitive role in locomotion, complicates our understanding of the population. Nevertheless, taken togeth-er our results suggest a functional significance for Dmrt3 neurons within CPG circuits.

68

Summary and Future Perspectives

Neuroscientists are continually tailoring, reworking, expanding and refining Sherrington and Brown’s initial circuitry ideals to form a current representa-tion of the locomotor CPG. Spinal locomotor networks are now considered to be comprised of neural populations derived from specific developmental gene expression, with their own explicit characteristics, function and role within locomotor circuits. The work in this thesis contributes to the contin-ued and increased understanding of established neural populations such as the Renshaw cells (Paper I and II), and to the identification, classification and possible functional roles of a previously undescribed population marked by Dmrt3 (Paper III).

Many questions surrounding the locomotor CPG remain open ended and none more so than those surrounding the function and role of Renshaw cells. We found that Renshaw cells can be exclusively marked by Chrna2 (Paper I) allowing them to be discriminated from other ventral horn interneurons, but their apparent role in locomotor control still remains unclear (Paper II). Instead, we found Renshaw cell mediated inhibition has profound effects on the components, but not the functional outcome of the recurrent inhibition circuit. Our results do not support a role for the Renshaw cells in locomotion contrary to recent views surrounding the V1 population, but rather indicate the importance of the conserved function of the recurrent inhibitory circuit through compensatory mechanisms. Future studies addressing the role of Renshaw cells should combine specific Renshaw cell targeting with optoge-netic and DREADD (designer receptors exclusively activated by designer drugs) technologies to selectively and temporally manipulate neural signal-ling. Specific temporal manipulations, especially during late postnatal or adult stages, would circumvent possible developmental compensation by the nervous system, potentially revealing a more defined functional role for the Renshaw cell population. These acute manipulation techniques in combina-tion with existing electrophysiological and emerging in vivo methods, should address whether the acute activation or inhibition of the Renshaw cell popu-lation will alter the ongoing locomotor pattern or motor behaviour.

Specific questions raised from Papers I and II are currently being incor-porated in future projects within the laboratory and aim to address whether Renshaw cells constitute multiple subpopulations based on ion channel and gene expression, and whether additional Renshaw cell specific markers can be identified. Moreover, whilst not directly related to the Renshaw cell func-

69

tional role but still of considerable interest to the field, the specific actions of mecamylamine on Renshaw cell and other spinal populations should be more thoroughly investigated.

The gene Dmrt3 has recently gained interest due to its central role in the gait patterning of horses and mice, but until now, the role of the population derived from Dmrt3 expression remained obscure. We found the inhibitory Dmrt3 population had a profound impact on locomotor coordination in mice, which became more prominent at faster locomotor speeds. In particular, mice lacking Dmrt3 mediated inhibition had severe difficulties coordinated left-right alternation at high speeds, falling into a synchronous gait (Paper III). Additionally we found possible Dmrt3 subpopulations that fire at indi-vidual and heterogeneous frequencies during locomotor activity. Future stud-ies should aim to functionally, electrophysiologically and morphologically classify the neural characteristics of these presumed subpopulations and follow the activity patterns of the neurons across different locomotor speeds. Part of this work, following single Dmrt3 neural firing across different speeds is already underway and preliminary results suggest that although the firing frequency of Dmrt3 neurons changes with speed, the neural firing pattern and relationship to the locomotor rhythm is unaltered. Paper III did not address whether the inherent firing and electrical properties of Dmrt3 or motor neurons were altered in Dmrt3Cre;ViaatKO mice. This could be of po-tential interest to better understand the physiological effects on the locomo-tor circuitry in the absence of Dmrt3 inhibition. Optogenetic manipulation to artificially regulate the firing frequency of Dmrt3 neurons, combined with existing electrophysiological methods would be an elegant way to investi-gate whether Dmrt3 firing drives CPG patterning. Many unanswered ques-tions remain surrounding the connectivity and synaptic inputs to the Dmrt3 population, which currently hinder efforts to confidently place Dmrt3 neu-rons within CPG circuitry models. Future experiments should therefore con-firm presumed ipsilateral Dmrt3 projections and aim to elucidate neurons up- and downstream of the Dmrt3 population. Moreover, novel in vitro ex-periments should be considered and devised to confirm whether Dmrt3 neu-rons directly coordinate flexor-extensor alternation.

Although we did not observe any direct compensatory effects resulting from removing Dmrt3-mediated inhibition in CPG circuits, and saw similar locomotor behavioural phenotypes both in isolated preparations and in vivo, we cannot completely exclude the possibility that compensatory mechanisms contribute to the phenotype. Therefore, similar to the suggested Renshaw cell experiments, temporal induction mechanisms including optogenetics and DREADD techniques, should be used in a Dmrt3 cell specific manner to modulate cellular functions and control neural firing to reiterate and extend the current findings of Paper III. Other strategies include using DTR-DTA (Diptheria toxin) viruses to conditionally ablate Chrna2 and Dmrt3 popula-tions in vivo and observe the consequences to behaviour. Dmrt3 and Chrna2

70

are expressed extensively throughout the brain (Paper III, Leao et al., 2012; Mikulovic et al., 2015), making system wide neural ablation through Cre-LoxP dependent DREAAD or optogenetic mice less preferable. Localised lumbar spinal cord injections of Cre specific DREAAD or optogenetic virus-es would avoid potential expression in cortical populations, restricting phe-notypic outcomes to spinal interneurons. These approaches, not without their own technical limitations and challenges, are currently being implemented within the laboratory and will be used in conjunction with a battery of motor tests to elucidate and confirm the functional role of both Renshaw cells and Dmrt3 neurons.

Although we have made considerable advances in our understanding of CPG populations there is still significant work to be done to comprehensive-ly elucidate the functional roles and circuitry of CPG interneuron popula-tions. The continual genetic identification of tightly defined CPG interneu-ron populations and subpopulations drives the re-investigation of molecular, anatomical, physiological and functional properties of more restricted groups of neurons. Our experiments have a distinct advantage in the progression of research surrounding these populations since our novel mice lines selectively target the Renshaw cell (Chrna2Cre) and Dmrt3 (Dmrt3Cre) populations. In general, future studies should aim to exploit the genetic specificity offered by these mouse models and target these populations with novel methods to determine their functional role within the locomotor circuitry and how their inherent activity correlates to behaviour.

71

Acknowledgements

And now, the end is near and so I face the final curtain.

My friend, I’ll say it clear, I’ll state my case, of which I’m certain.

Frank Sinatra

There are so many people that have contributed to the academic, scientific and personal experiences I have had during the course of my PhD. I thank you all for making this a truly memorable, enriching and enjoyable time. Thank you for your input, your contributions, your time and your patience. Thank you for your dedication to research, for your continued effort and persistence and of course last, but certainly not least, your friendship. As I walk away from the last five years, I take with me a tremendous amount of memories, knowledge and shared accomplishments.

First and foremost, I would like to express my sincere gratitude to my super-visor Klas Kullander. Thank you for offering me the opportunity to develop and grow into the person and researcher that stands before you today. Thank you for your support, guidance and encouragement throughout the years.

Kia Leão, it has been an absolute pleasure working with you. Thank you for always taking time to share your knowledge, skills and opinions; they have made, and will continue to make me a better researcher.

Throughout my PhD I have been asked time and time again how I came to be in Sweden. Now that I am almost at the end of my time as a PhD student, I often reflect about the choices I made and the opportunities I had that al-lowed me the fortune of moving to Sweden. Karin Nordström, I don’t think I can accurately express just how much you have influenced and impacted both my academic and personal journey. Just to say thank you, seems terri-bly inadequate for the support, belief, encouragement and mentorship you have given me over the last nine years.

72

Malin Lagerström, Henrik Bojie and Richardson Leão, thank you all for being a continual support, for fostering my knowledge and skills and for bringing your unique and positive perspectives to my academic endeavours. To all my co-authors and colleagues, thank you for your continued hard work, co-operation, collaboration and enthusiasm. The work in this thesis would not have been possible without you all. A special mention to Fabio Caixeta, Markus Hilscher, Martin Larhammar, Anders Enjin and Danny Schnerwitzki; I have learnt so much from each of you – thank you for al-ways having the time to teach, explain and accommodate my endless ques-tions and queries; your input is truly valued. Sanja – thank you for your sci-entific prowess and friendship. To the past and present members of the Developmental Genetics Unit, you are all a huge part of why this journey has been so enjoyable. It has been a true adventure to work beside and with each of you. To friends and col-leagues within the Department of Neuroscience and the animal facility – thank you for your help and advice. Jenny, I am immensely grateful for your assistance with the mice over the years.

To my friends and family, both in Australia and Sweden, I couldn’t have done this without you and I am so incredibly lucky to have each of you in my life. Tim– Thank you for everything you have done for me, I truly appre-ciate it. Kasia, Rebecca and Shiuli – I couldn’t have imagined this journey without you; thank you for everything, you are family. Marcus – Thank you for your endless patience; my life was made for the better when I met you. Mum, Dad, Caitlin, Nan and Pop – I cannot find the words to express how incredibly proud I am that you are my family, nor can I adequately convey what your unwavering love, support and belief means to me. Thank you for both showing and instilling in me where dedication, hard work and persever-ance can take you and above all, for always encouraging and understanding my decisions.

73

References

Alaynick, W.A., Jessell, T.M., and Pfaff, S.L. (2011). SnapShot: spinal cord devel-opment. Cell 146, 178–178.e1.

Albanese, A., Bhatia, K., Bressman, S.B., Delong, M.R., Fahn, S., Fung, V.S.C., Hallett, M., Jankovic, J., Jinnah, H.A., Klein, C., et al. (2013). Phenomenology and classification of dystonia: a consensus update. Mov. Disord. Off. J. Mov. Disord. Soc. 28, 863–873.

Albertsdóttir, E., Eriksson, S., Näsholm, A., Strandberg, E., and Árnason, T. (2007). Genetic analysis of competition data on Icelandic horses. Livest. Sci. 110, 242–250.

Al-Mosawie, A., Wilson, J.M., and Brownstone, R.M. (2007). Heterogeneity of V2-derived interneurons in the adult mouse spinal cord. Eur. J. Neurosci. 26, 3003–3015.

Alvarez, F.J., and Fyffe, R.E.W. (2007). The continuing case for the Renshaw cell. J. Physiol. 584, 31–45.

Alvarez, F.J., Dewey, D.E., Harrington, D.A., and Fyffe, R.E. (1997). Cell-type specific organization of glycine receptor clusters in the mammalian spinal cord. J. Comp. Neurol. 379, 150–170.

Alvarez, F.J., Dewey, D.E., McMillin, P., and Fyffe, R.E.W. (1999). Distribution of cholinergic contacts on Renshaw cells in the rat spinal cord: a light microscopic study. J. Physiol. 515, 787–797.

Alvarez, F.J., Jonas, P.C., Sapir, T., Hartley, R., Berrocal, M.C., Geiman, E.J., Todd, A.J., and Goulding, M. (2005). Postnatal phenotype and localization of spinal cord V1 derived interneurons. J. Comp. Neurol. 493, 177–192.

Alvarez, F.J., Benito-Gonzalez, A., and Siembab, V.C. (2013). Principles of inter-neuron development learned from Renshaw cells and the motoneuron recurrent inhibitory circuit. Ann. N. Y. Acad. Sci. 1279, 22–31.

Ampatzis, K., Song, J., Ausborn, J., and El Manira, A. (2014). Separate Microcircuit Modules of Distinct V2a Interneurons and Motoneurons Control the Speed of Locomotion. Neuron 83, 934–943.

Andersson, L.S., Larhammar, M., Memic, F., Wootz, H., Schwochow, D., Rubin, C.-J., Patra, K., Arnason, T., Wellbring, L., Hjälm, G., et al. (2012). Mutations in DMRT3 affect locomotion in horses and spinal circuit function in mice. Na-ture 488, 642–646.

Aubrey, K.R. (2016). Presynaptic control of inhibitory neurotransmitter content in VIAAT containing synaptic vesicles. Neurochem. Int.

Banks, G.B., Kanjhan, R., Wiese, S., Kneussel, M., Wong, L.M., O’Sullivan, G., Sendtner, M., Bellingham, M.C., Betz, H., and Noakes, P.G. (2005). Glyciner-gic and GABAergic synaptic activity differentially regulate motoneuron surviv-al and skeletal muscle innervation. J. Neurosci. Off. J. Soc. Neurosci. 25, 1249–1259.

74

Beliez, L., Barrière, G., Bertrand, S.S., and Cazalets, J.-R. (2014). Multiple mono-aminergic modulation of posturo-locomotor network activity in the newborn rat spinal cord. Front. Neural Circuits 8, 99.

Bellardita, C., and Kiehn, O. (2015). Phenotypic characterization of speed-associated gait changes in mice reveals modular organization of locomotor net-works. Curr. Biol. CB 25, 1426–1436.

Ben-Ari, Y. (2002). Excitatory actions of gaba during development: the nature of the nurture. Nat. Rev. Neurosci. 3, 728–739.

Ben-Ari, Y., Gaiarsa, J.-L., Tyzio, R., and Khazipov, R. (2007). GABA: a pioneer transmitter that excites immature neurons and generates primitive oscillations. Physiol. Rev. 87, 1215–1284.

Benito-Gonzalez, A., and Alvarez, F.J. (2012). Renshaw cells and Ia inhibitory in-terneurons are generated at different times from p1 progenitors and differentiate shortly after exiting the cell cycle. J. Neurosci. Off. J. Soc. Neurosci. 32, 1156–1170.

Berger, T.K., Silberberg, G., Perin, R., and Markram, H. (2010). Brief bursts self-inhibit and correlate the pyramidal network. PLoS Biol. 8.

Bertrand, S., and Cazalets, J.-R. (2002). The respective contribution of lumbar seg-ments to the generation of locomotion in the isolated spinal cord of newborn rat. Eur. J. Neurosci. 16, 1741–1750.

Bhumbra, G.S., Bannatyne, B.A., Watanabe, M., Todd, A.J., Maxwell, D.J., and Beato, M. (2014). The recurrent case for the Renshaw cell. J. Neurosci. Off. J. Soc. Neurosci. 34, 12919–12932.

Bikoff, J.B., Gabitto, M.I., Rivard, A.F., Drobac, E., Machado, T.A., Miri, A., Bren-ner-Morton, S., Famojure, E., Diaz, C., Alvarez, F.J., et al. (2016). Spinal Inhib-itory Interneuron Diversity Delineates Variant Motor Microcircuits. Cell 165, 207–219.

Blanco, J.E., Anderson, K.D., and Steward, O. (2007). Recovery of forepaw grip-ping ability and reorganization of cortical motor control following cervical spi-nal cord injuries in mice. Exp. Neurol. 203, 333–348.

Blankenship, A.G., and Feller, M.B. (2010). Mechanisms underlying spontaneous patterned activity in developing neural circuits. Nat. Rev. Neurosci. 11, 18–29.

Bonnot, A., Whelan, P.J., Mentis, G.Z., and O’Donovan, M.J. (2002). Locomotor-like activity generated by the neonatal mouse spinal cord. Brain Res. Brain Res. Rev. 40, 141–151.

Borowska, J., Jones, C.T., Zhang, H., Blacklaws, J., Goulding, M., and Zhang, Y. (2013). Functional subpopulations of V3 interneurons in the mature mouse spi-nal cord. J. Neurosci. Off. J. Soc. Neurosci. 33, 18553–18565.

Bouvier, J., Caggiano, V., Leiras, R., Caldeira, V., Bellardita, C., Balueva, K., Fuchs, A., and Kiehn, O. (2015). Descending Command Neurons in the Brain-stem that Halt Locomotion. Cell 163, 1191–1203.

van den Brand, R., Heutschi, J., Barraud, Q., DiGiovanna, J., Bartholdi, K., Huerli-mann, M., Friedli, L., Vollenweider, I., Moraud, E.M., Duis, S., et al. (2012). Restoring voluntary control of locomotion after paralyzing spinal cord injury. Science 336, 1182–1185.

Bregestovski, P., and Bernard, C. (2012). Excitatory GABA: How a Correct Obser-vation May Turn Out to be an Experimental Artifact. Front. Pharmacol. 3, 65.

75

Britz, O., Zhang, J., Grossmann, K.S., Dyck, J., Kim, J.C., Dymecki, S., Gosgnach, S., and Goulding, M. (2015). A genetically defined asymmetry underlies the in-hibitory control of flexor-extensor locomotor movements. eLife 4.

Brown, A.G. (1982). Review Article the Dorsal Horn of the Spinal Cord. Exp. Phys-iol. 67, 193–212.

Brown, T.G. (1911). The Intrinsic Factors in the Act of Progression in the Mammal. Proc. R. Soc. Lond. Ser. B Contain. Pap. Biol. Character 84, 308–319.

Brown, T.G. (1914). On the nature of the fundamental activity of the nervous cen-tres; together with an analysis of the conditioning of rhythmic activity in pro-gression, and a theory of the evolution of function in the nervous system. J. Physiol. 48, 18–46.

Brownstone, R.M., Krawitz, S., and Jordan, L.M. (2011). Reversal of the late phase of spike frequency adaptation in cat spinal motoneurons during fictive locomo-tion. J. Neurophysiol. 105, 1045–1050.

Bućan, M., and Abel, T. (2002). The mouse: genetics meets behaviour. Nat. Rev. Genet. 3, 114–123.

Burke, R.E. (1981). Motor units: anatomy, physiology, and functional organization. In Handbook of Physiology, V. B. Brooks (ed.). Section 1: The Nervous Sys-tem. (Bethseda, MD: American Physiological Society).

Butt, S.J.B., and Kiehn, O. (2003). Functional identification of interneurons respon-sible for left-right coordination of hindlimbs in mammals. Neuron 38, 953–963.

Butt, S.J.B., Lebret, J.M., and Kiehn, O. (2002). Organization of left-right coordina-tion in the mammalian locomotor network. Brain Res. Brain Res. Rev. 40, 107–117.

Carr, P.A., Alvarez, F.J., Leman, E.A., and Fyffe, R.E. (1998). Calbindin D28k expression in immunohistochemically identified Renshaw cells. Neuroreport 9, 2657–2661.

Cazalets, J.R., Borde, M., and Clarac, F. (1995). Localization and organization of the central pattern generator for hindlimb locomotion in newborn rat. J. Neuro-sci. Off. J. Soc. Neurosci. 15, 4943–4951.

Chang, Q., and Martin, L.J. (2009). Glycinergic innervation of motoneurons is defi-cient in amyotrophic lateral sclerosis mice: a quantitative confocal analysis. Am. J. Pathol. 174, 574–585.

Chang, Q., and Martin, L.J. (2011). Glycine receptor channels in spinal motoneurons are abnormal in a transgenic mouse model of amyotrophic lateral sclerosis. J. Neurosci. Off. J. Soc. Neurosci. 31, 2815–2827.

Chen, S., Sayana, P., Zhang, X., and Le, W. (2013). Genetics of amyotrophic lateral sclerosis: an update. Mol. Neurodegener. 8, 28.

Chub, N., and O’Donovan, M.J. (1998). Blockade and recovery of spontaneous rhythmic activity after application of neurotransmitter antagonists to spinal net-works of the chick embryo. J. Neurosci. Off. J. Soc. Neurosci. 18, 294–306.

Chung, K., and Deisseroth, K. (2013). CLARITY for mapping the nervous system. Nat. Methods 10, 508–513.

Cina, C., and Hochman, S. (2000). Diffuse distribution of sulforhodamine-labeled neurons during serotonin-evoked locomotion in the neonatal rat thoracolumbar spinal cord. J. Comp. Neurol. 423, 590–602.

76

Courtine, G., Gerasimenko, Y., van den Brand, R., Yew, A., Musienko, P., Zhong, H., Song, B., Ao, Y., Ichiyama, R.M., Lavrov, I., et al. (2009). Transformation of nonfunctional spinal circuits into functional states after the loss of brain in-put. Nat. Neurosci. 12, 1333–1342.

Cowley, K.C., and Schmidt, B.J. (1995). Effects of inhibitory amino acid antago-nists on reciprocal inhibitory interactions during rhythmic motor activity in the in vitro neonatal rat spinal cord. J. Neurophysiol. 74, 1109–1117.

Crone, S.A., Quinlan, K.A., Zagoraiou, L., Droho, S., Restrepo, C.E., Lundfald, L., Endo, T., Setlak, J., Jessell, T.M., Kiehn, O., et al. (2008). Genetic ablation of V2a ipsilateral interneurons disrupts left-right locomotor coordination in mam-malian spinal cord. Neuron 60, 70–83.

Crone, S.A., Zhong, G., Harris-Warrick, R., and Sharma, K. (2009). In Mice Lack-ing V2a Interneurons, Gait Depends on Speed of Locomotion. J. Neurosci. 29, 7098–7109.

Crow, J.F. (2002). C. C. Little, cancer and inbred mice. Genetics 161, 1357–1361. Cullheim, S., and Kellerth, J.O. (1981). Two kinds of recurrent inhibition of cat

spinal alpha-motoneurones as differentiated pharmacologically. J. Physiol. 312, 209–224.

D’Acunzo, P., Badaloni, A., Ferro, M., Zimarino, V., Ripamonti, M., Malgaroli, A., and Consalez, G.G. (2014). A conditional transgenic reporter of presynaptic terminals reveals novel features of the mouse corticospinal tract. Front. Neuro-anat. 7, 50.

Danner, S.M., Hofstoetter, U.S., Freundl, B., Binder, H., Mayr, W., Rattay, F., and Minassian, K. (2015). Human spinal locomotor control is based on flexibly or-ganized burst generators. Brain 138, 577–588.

Danner, S.M., Wilshin, S.D., Shevtsova, N.A., and Rybak, I.A. (2016). Central con-trol of interlimb coordination and speed-dependent gait expression in quadru-peds. J. Physiol.

De Zeeuw, C.I., Chorev, E., Devor, A., Manor, Y., Van Der Giessen, R.S., De Jeu, M.T., Hoogenraad, C.C., Bijman, J., Ruigrok, T.J.H., French, P., et al. (2003). Deformation of network connectivity in the inferior olive of connexin 36-deficient mice is compensated by morphological and electrophysiological changes at the single neuron level. J. Neurosci. Off. J. Soc. Neurosci. 23, 4700–4711.

Deardorff, A.S., Romer, S.H., Deng, Z., Bullinger, K.L., Nardelli, P., Cope, T.C., and Fyffe, R.E.W. (2013). Expression of postsynaptic Ca2+-activated K+ (SK) channels at C-bouton synapses in mammalian lumbar -motoneurons. J. Physiol. 591, 875–897.

Dimitrijevic, M.R., Gerasimenko, Y., and Pinter, M.M. (1998). Evidence for a spinal central pattern generator in humans. Ann. N. Y. Acad. Sci. 860, 360–376.

Dougherty, K.J., and Kiehn, O. (2010a). Firing and cellular properties of V2a inter-neurons in the rodent spinal cord. J. Neurosci. Off. J. Soc. Neurosci. 30, 24–37.

Dougherty, K.J., and Kiehn, O. (2010b). Functional organization of V2a-related locomotor circuits in the rodent spinal cord. Ann. N. Y. Acad. Sci. 1198, 85–93.

Dougherty, K.J., Zagoraiou, L., Satoh, D., Rozani, I., Doobar, S., Arber, S., Jessell, T.M., and Kiehn, O. (2013). Locomotor rhythm generation linked to the output of spinal shox2 excitatory interneurons. Neuron 80, 920–933.

77

Dourado, M., and Sargent, P.B. (2002). Properties of nicotinic receptors underlying Renshaw cell excitation by alpha-motor neurons in neonatal rat spinal cord. J. Neurophysiol. 87, 3117–3125.

Dugué, G.P., Dumoulin, A., Triller, A., and Dieudonné, S. (2005). Target-Dependent Use of Coreleased Inhibitory Transmitters at Central Synapses. J. Neurosci. 25, 6490–6498.

Dyck, J., Lanuza, G.M., and Gosgnach, S. (2012). Functional characterization of dI6 interneurons in the neonatal mouse spinal cord. J. Neurophysiol. 107, 3256–3266.

Eccles, J.C., and Sherrington, C.S. (1930). Numbers and Contraction-Values of Individual Motor-Units Examined in some Muscles of the Limb. Proc. R. Soc. Lond. Ser. B Contain. Pap. Biol. Character 106, 326–357.

Eccles, R.M., and Lundberg, A. (1958). Integrative pattern of Ia synaptic actions on motoneurones of hip and knee muscles. J. Physiol. 144, 271–298.

Eccles, J.C., Fatt, P., and Koketsu, K. (1954). Cholinergic and inhibitory synapses in a pathway from motor-axon collaterals to motoneurones. J. Physiol. 126, 524–562.

Eccles, J.C., Eccles, R.M., Iggo, A., and Ito, M. (1961). Distribution of recurrent inhibition among motoneurones. J. Physiol. 159, 479–499.

Echelard, Y., Epstein, D.J., St-Jacques, B., Shen, L., Mohler, J., McMahon, J.A., and McMahon, A.P. (1993). Sonic hedgehog, a member of a family of putative sig-naling molecules, is implicated in the regulation of CNS polarity. Cell 75, 1417–1430.

Edgerton, J.R., and Reinhart, P.H. (2003). Distinct contributions of small and large conductance Ca2+-activated K+ channels to rat Purkinje neuron function. J. Physiol. 548, 53–69.

El Mestikawy, S., Wallén-Mackenzie, A., Fortin, G.M., Descarries, L., and Trudeau, L.-E. (2011). From glutamate co-release to vesicular synergy: vesicular gluta-mate transporters. Nat. Rev. Neurosci. 12, 204–216.

Enjin, A., Rabe, N., Nakanishi, S.T., Vallstedt, A., Gezelius, H., Memic, F., Lind, M., Hjalt, T., Tourtellotte, W.G., Bruder, C., et al. (2010). Identification of nov-el spinal cholinergic genetic subtypes disclose Chodl and Pitx2 as markers for fast motor neurons and partition cells. J. Comp. Neurol. 518, 2284–2304.

Enjin, A., Leao, K.E., Mikulovic, S., Le Merre, P., Tourtellotte, W.G., and Kulland-er, K. (2012). Sensorimotor function is modulated by the serotonin receptor 1d, a novel marker for gamma motor neurons. Mol. Cell. Neurosci. 49, 322–332.

Fitzsimons, E., Van Zundert, B., Constaine-Paton, M., Brown Jr, R.H., and Alvarez, F.J. (2006) Alterations in the Renshaw cell recurrent inhibitory circuit in the G93A SOD1 mouse model ALS. Abstr Soc Neurosci 678. 679.

Fletcher, P., and Forrester, T. (1975). The effect of curare on the release of acetyl-choline from mammalian motor nerve terminals and an estimate of quantum content. J. Physiol. 251, 131–144.

Fogarty, M.J., Smallcombe, K.L., Yanagawa, Y., Obata, K., Bellingham, M.C., and Noakes, P.G. (2013). Genetic deficiency of GABA differentially regulates res-piratory and non-respiratory motor neuron development. PloS One 8, e56257.

Fogarty, M.J., Kanjhan, R., Bellingham, M.C., and Noakes, P.G. (2016). Glyciner-gic Neurotransmission: A Potent Regulator of Embryonic Motor Neuron Den-dritic Morphology and Synaptic Plasticity. J. Neurosci. Off. J. Soc. Neurosci. 36, 80–87.

78

Fornai, F., Siciliano, G., Manca, M.L., Murri, L., Paparelli, A., and Ruggieri, S. (2008). Lithium in ALS: from the bench to the bedside. Amyotroph. Lateral Scler. Off. Publ. World Fed. Neurol. Res. Group Mot. Neuron Dis. 9, 123–124.

França, A.S.C., Siwani, S., Reis, A., Mikulovic, S., Hilscher, M.M., Edwards, S.J., Leão, R.N., Tort, A.B.L., and Kullander, K. OLMα2 cells control encoding of recognition and avoidance memory.

Frigon, A., Hurteau, M.-F., Thibaudier, Y., Leblond, H., Telonio, A., and D’Angelo, G. (2013). Split-Belt Walking Alters the Relationship between Locomotor Phas-es and Cycle Duration across Speeds in Intact and Chronic Spinalized Adult Cats. J. Neurosci. 33, 8559–8566.

Fyffe, R.E. (1990). Evidence for separate morphological classes of Renshaw cells in the cat’s spinal cord. Brain Res. 536, 301–304.

Geertsen, S.S., Stecina, K., Meehan, C.F., Nielsen, J.B., and Hultborn, H. (2011). Reciprocal Ia inhibition contributes to motoneuronal hyperpolarisation during the inactive phase of locomotion and scratching in the cat. J. Physiol. 589, 119–134.

Geiman, E.J., Knox, M.C., and Alvarez, F.J. (2000). Postnatal maturation of gephyrin/glycine receptor clusters on developing Renshaw cells. J. Comp. Neu-rol. 426, 130–142.

Geiman, E.J., Zheng, W., Fritschy, J.-M., and Alvarez, F.J. (2002). Glycine and GABA(A) receptor subunits on Renshaw cells: relationship with presynaptic neurotransmitters and postsynaptic gephyrin clusters. J. Comp. Neurol. 444, 275–289.

Gezelius, H., Wallén-Mackenzie, A., Enjin, A., Lagerström, M., and Kullander, K. (2006). Role of glutamate in locomotor rhythm generating neuronal circuitry. J. Physiol. Paris 100, 297–303.

Gómez-Skarmeta, J.L., Campuzano, S., and Modolell, J. (2003). Half a century of neural prepatterning: the story of a few bristles and many genes. Nat. Rev. Neu-rosci. 4, 587–598.

Gonzalez-Forero, D., Pastor, A.M., Geiman, E.J., Benítez-Temiño, B., and Alvarez, F.J. (2005). Regulation of gephyrin cluster size and inhibitory synaptic currents on Renshaw cells by motor axon excitatory inputs. J. Neurosci. Off. J. Soc. Neu-rosci. 25, 417–429.

Gonzalez-Islas, C., Chub, N., Garcia-Bereguiain, M.A., and Wenner, P. (2010). GABAergic synaptic scaling in embryonic motoneurons is mediated by a shift in the chloride reversal potential. J. Neurosci. Off. J. Soc. Neurosci. 30, 13016–13020.

Gosgnach, S., Lanuza, G.M., Butt, S.J.B., Saueressig, H., Zhang, Y., Velasquez, T., Riethmacher, D., Callaway, E.M., Kiehn, O., and Goulding, M. (2006). V1 spi-nal neurons regulate the speed of vertebrate locomotor outputs. Nature 440, 215–219.

Goulding, M. (2009). Circuits controlling vertebrate locomotion: moving in a new direction. Nat. Rev. Neurosci. 10, 507–518.

Goulding, M., Bourane, S., Garcia-Campmany, L., Dalet, A., and Koch, S. (2014). Inhibition downunder: an update from the spinal cord. Curr. Opin. Neurobiol. 26, 161–166.

Grillner, S. (1975). Locomotion in vertebrates: central mechanisms and reflex inter-action. Physiol. Rev. 55, 247–304.

79

Grillner, S. (1981). Control of locomotion in bipeds, tetrapods, and fish. In: Hand-book of physiology. the nervous system; motor control (Brookhart JM, Mountcastle VB., eds) (Bethseda, MD: American Physiological Society).

Grillner, S., and Zangger, P. (1979). On the central generation of locomotion in the low spinal cat. Exp. Brain Res. 34, 241–261.

Gross, M.K., Dottori, M., and Goulding, M. (2002). Lbx1 specifies somatosensory association interneurons in the dorsal spinal cord. Neuron 34, 535–549.

Guertin, P.A. (2009). The mammalian central pattern generator for locomotion. Brain Res. Rev. 62, 45–56.

Guertin, P.A. (2012). Central pattern generator for locomotion: anatomical, physio-logical, and pathophysiological considerations. Front. Neurol. 3, 183.

Gulledge, A.T., and Stuart, G.J. (2003). Excitatory Actions of GABA in the Cortex. Neuron 37, 299–309.

Hägglund, M., Borgius, L., Dougherty, K.J., and Kiehn, O. (2010). Activation of groups of excitatory neurons in the mammalian spinal cord or hindbrain evokes locomotion. Nat. Neurosci. 13, 246–252.

Hägglund, M., Dougherty, K.J., Borgius, L., Itohara, S., Iwasato, T., and Kiehn, O. (2013). Optogenetic dissection reveals multiple rhythmogenic modules underly-ing locomotion. Proc. Natl. Acad. Sci. U. S. A. 110, 11589–11594.

Hallett, M. (2011). Neurophysiology of dystonia: The role of inhibition. Neurobiol. Dis. 42, 177–184.

Hamm, T.M. (1990). Recurrent inhibition to and from motoneurons innervating the flexor digitorum and flexor hallucis longus muscles of the cat. J. Neurophysiol. 63, 395–403.

Hamm, T.M., Sasaki, S., Stuart, D.G., Windhorst, U., and Yuan, C.S. (1987). Distri-bution of single-axon recurrent inhibitory post-synaptic potentials in a single spinal motor nucleus in the cat. J. Physiol. 388, 653–664.

Han, Q., Cao, C., Ding, Y., So, K.-F., Wu, W., Qu, Y., and Zhou, L. (2015). Plastici-ty of motor network and function in the absence of corticospinal projection. Exp. Neurol. 267, 194–208.

Hanson, M.G., and Landmesser, L.T. (2003). Characterization of the circuits that generate spontaneous episodes of activity in the early embryonic mouse spinal cord. J. Neurosci. Off. J. Soc. Neurosci. 23, 587–600.

Harley, C.M., Reilly, M.G., Stewart, C., Schlegel, C., Morley, E., Puhl, J.G., Nagel, C., Crisp, K.M., and Mesce, K.A. (2015). Compensatory plasticity restores lo-comotion after chronic removal of descending projections. J. Neurophysiol. 113, 3610–3622.

Harrison, T.C., and Murphy, T.H. (2014). Motor maps and the cortical control of movement. Curr. Opin. Neurobiol. 24, 88–94.

Harris-Warrick, R.M. (2011). Neuromodulation and flexibility in Central Pattern Generator networks. Curr. Opin. Neurobiol. 21, 685–692.

Hilscher, M.M., Leão, R.N., Edwards, S.J., Leão, K.E., and Kullander, K. Chrna2-Martinotti Cells Synchronize layer 5 type A Pyramidal Cells via Rebound Exci-tation.

Hinckley, C.A., Hartley, R., Wu, L., Todd, A., and Ziskind-Conhaim, L. (2005). Locomotor-like rhythms in a genetically distinct cluster of interneurons in the mammalian spinal cord. J. Neurophysiol. 93, 1439–1449.

80

Hinckley, C.A., Alaynick, W.A., Gallarda, B.W., Hayashi, M., Hilde, K.L., Driscoll, S.P., Dekker, J.D., Tucker, H.O., Sharpee, T.O., and Pfaff, S.L. (2015). Spinal Locomotor Circuits Develop Using Hierarchical Rules Based on Motorneuron Position and Identity. Neuron 87, 1008–1021.

Hippenmeyer, S., Vrieseling, E., Sigrist, M., Portmann, T., Laengle, C., Ladle, D.R., and Arber, S. (2005). A developmental switch in the response of DRG neurons to ETS transcription factor signaling. PLoS Biol. 3, e159.

Hodgkin, A.L., and Huxley, A.F. (1952). A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. 117, 500–544.

Hogg, R.C., Raggenbass, M., and Bertrand, D. (2003). Nicotinic acetylcholine re-ceptors: from structure to brain function. Rev. Physiol. Biochem. Pharmacol. 147, 1–46.

Hossaini, M., Cardona Cano, S., van Dis, V., Haasdijk, E.D., Hoogenraad, C.C., Holstege, J.C., and Jaarsma, D. (2011). Spinal inhibitory interneuron pathology follows motor neuron degeneration independent of glial mutant superoxide dis-mutase 1 expression in SOD1-ALS mice. J. Neuropathol. Exp. Neurol. 70, 662–677.

Hughes, D.I., Boyle, K.A., Kinnon, C.M., Bilsland, C., Quayle, J.A., Callister, R.J., and Graham, B.A. (2013). HCN4 subunit expression in fast-spiking interneurons of the rat spinal cord and hippocampus. Neuroscience 237, 7–18.

Hultborn, H., Jankowska, E., and Lindström, S. (1971). Recurrent inhibition from motor axon collaterals of transmission in the Ia inhibitory pathway to motoneu-rones. J. Physiol. 215, 591–612.

Hultborn, H., Lindström, S., and Wigström, H. (1979). On the function of recurrent inhibition in the spinal cord. Exp. Brain Res. 37, 399–403.

Hultborn, H., Brownstone, R.B., Toth, T.I., and Gossard, J.-P. (2004). Key mecha-nisms for setting the input-output gain across the motoneuron pool. Prog. Brain Res. 143, 77–95.

Hunt, C.C. (1990). Mammalian muscle spindle: peripheral mechanisms. Physiol. Rev. 70, 643–663.

Ishii, K., Wong, J.K., and Sumikawa, K. (2005). Comparison of alpha2 nicotinic acetylcholine receptor subunit mRNA expression in the central nervous system of rats and mice. J. Comp. Neurol. 493, 241–260.

Ishii, T.M., Takano, M., Xie, L.-H., Noma, A., and Ohmori, H. (1999). Molecular Characterization of the Hyperpolarization-activated Cation Channel in Rabbit Heart Sinoatrial Node. J. Biol. Chem. 274, 12835–12839.

Jäderkvist, K., Kangas, N., Andersson, L.S., and Lindgren, G. (2014). Gaitedness is associated with the DMRT3 “Gait keeper” mutation in Morgan and American Curly horses. Anim. Genet. 45, 908–909.

Jaenisch, R. (1988). Transgenic animals. Science 240, 1468–1474. Jankowska, E. (1992). Interneuronal relay in spinal pathways from proprioceptors.

Prog. Neurobiol. 38, 335–378. Kanning, K.C., Kaplan, A., and Henderson, C.E. (2010). Motor neuron diversity in

development and disease. Annu. Rev. Neurosci. 33, 409–440. Kenna, K.P., van Doormaal, P.T.C., Dekker, A.M., Ticozzi, N., Kenna, B.J., Dieks-

tra, F.P., van Rheenen, W., van Eijk, K.R., Jones, A.R., Keagle, P., et al. (2016). NEK1 variants confer susceptibility to amyotrophic lateral sclerosis. Nat. Genet. advance online publication.

81

Kiehn, O. (2006). Locomotor circuits in the mammalian spinal cord. Annu. Rev. Neurosci. 29, 279–306.

Kiehn, O., Kjaerulff, O., Tresch, M.C., and Harris-Warrick, R.M. (2000). Contribu-tions of intrinsic motor neuron properties to the production of rhythmic motor output in the mammalian spinal cord. Brain Res. Bull. 53, 649–659.

Kiehn, O., Quinlan, K.A., Restrepo, C.E., Lundfald, L., Borgius, L., Talpalar, A.E., and Endo, T. (2008). Excitatory components of the mammalian locomotor CPG. Brain Res. Rev. 57, 56–63.

Kiehn, O., Dougherty, K.J., Hägglund, M., Borgius, L., Talpalar, A., and Restrepo, C.E. (2010). Probing spinal circuits controlling walking in mammals. Biochem. Biophys. Res. Commun. 396, 11–18.

Kjaerulff, O., and Kiehn, O. (1996). Distribution of networks generating and coordi-nating locomotor activity in the neonatal rat spinal cord in vitro: a lesion study. J. Neurosci. Off. J. Soc. Neurosci. 16, 5777–5794.

Kjaerulff, O., and Kiehn, O. (1997). Crossed rhythmic synaptic input to motoneu-rons during selective activation of the contralateral spinal locomotor network. J. Neurosci. Off. J. Soc. Neurosci. 17, 9433–9447.

Kristjansson, T., Bjornsdottir, S., Sigurdsson, A., Andersson, L. s., Lindgren, G., Helyar, S. j., Klonowski, A. m., and Arnason, T. (2014). The effect of the “Gait keeper” mutation in the DMRT3 gene on gaiting ability in Icelandic horses. J. Anim. Breed. Genet. 131, 415–425.

Kudo, N., and Yamada, T. (1987). N-methyl-D,L-aspartate-induced locomotor activ-ity in a spinal cord-hindlimb muscles preparation of the newborn rat studied in vitro. Neurosci. Lett. 75, 43–48.

Kullander, K., Butt, S.J.B., Lebret, J.M., Lundfald, L., Restrepo, C.E., Rydström, A., Klein, R., and Kiehn, O. (2003). Role of EphA4 and EphrinB3 in local neuronal circuits that control walking. Science 299, 1889–1892.

Kupfermann, I. (1991). Functional studies of cotransmission. Physiol. Rev. 71, 683–732.

Ladenbauer, J., Minassian, K., Hofstoetter, U.S., Dimitrijevic, M.R., and Rattay, F. (2010). Stimulation of the human lumbar spinal cord with implanted and surface electrodes: a computer simulation study. IEEE Trans. Neural Syst. Rehabil. Eng. Publ. IEEE Eng. Med. Biol. Soc. 18, 637–645.

Lajoie, Y., Teasdale, N., Cole, J.D., Burnett, M., Bard, C., Fleury, M., Forget, R., Paillard, J., and Lamarre, Y. (1996). Gait of a deafferented subject without large myelinated sensory fibers below the neck. Neurology 47, 109–115.

Lamotte d’Incamps, B., and Ascher, P. (2008). Four excitatory postsynaptic iono-tropic receptors coactivated at the motoneuron-Renshaw cell synapse. J. Neuro-sci. Off. J. Soc. Neurosci. 28, 14121–14131.

Lanuza, G.M., Gosgnach, S., Pierani, A., Jessell, T.M., and Goulding, M. (2004). Genetic Identification of Spinal Interneurons that Coordinate Left-Right Loco-motor Activity Necessary for Walking Movements. Neuron 42, 375–386.

Lawal, H.O., and Krantz, D.E. (2013). SLC18: Vesicular neurotransmitter transport-ers for monoamines and acetylcholine. Mol. Aspects Med. 34, 360–372.

Leao, R.N., Oleskevich, S., Sun, H., Bautista, M., Fyffe, R.E.W., and Walmsley, B. (2004). Differences in Glycinergic mIPSCs in the Auditory Brain Stem of Nor-mal and Congenitally Deaf Neonatal Mice. J. Neurophysiol. 91, 1006–1012.

Leao, R.N., Mikulovic, S., Leao, K.E., Munguba, H., Gezelius, H., Enjin, A., Patra, K., Eriksson, A., Loew, L.M., Tort, A.B.L., et al. (2012). OLM interneurons dif-

82

ferentially modulate CA3 and entorhinal inputs to hippocampal CA1 neurons. Nat. Neurosci. 15, 1524–1530.

Lee, K.J., and Jessell, T.M. (1999). The specification of dorsal cell fates in the ver-tebrate central nervous system. Annu. Rev. Neurosci. 22, 261–294.

Lee, K.J., Dietrich, P., and Jessell, T.M. (2000). Genetic ablation reveals that the roof plate is essential for dorsal interneuron specification. Nature 403, 734–740.

Lemieux, M., Josset, N., Roussel, M., Couraud, S., and Bretzner, F. (2016). Speed-Dependent Modulation of the Locomotor Behavior in Adult Mice Reveals At-tractor and Transitional Gaits. Syst. Biol. 42.

Lemon, R.N. (2008). Descending Pathways in Motor Control. Annu. Rev. Neurosci. 31, 195–218.

Lerer, I., Sagi, M., Meiner, V., Cohen, T., Zlotogora, J., and Abeliovich, D. (2005). Deletion of the ANKRD15 gene at 9p24.3 causes parent-of-origin-dependent inheritance of familial cerebral palsy. Hum. Mol. Genet. 14, 3911–3920.

Liem, K.F., Jessell, T.M., and Briscoe, J. (2000). Regulation of the neural patterning activity of sonic hedgehog by secreted BMP inhibitors expressed by notochord and somites. Dev. Camb. Engl. 127, 4855–4866.

Liu, Y.B., Tewari, A., Salameh, J., Arystarkhova, E., Hampton, T.G., Brashear, A., Ozelius, L.J., Khodakhah, K., and Sweadner, K.J. (2015). A dystonia-like movement disorder with brain and spinal neuronal defects is caused by mutation of the mouse laminin β1 subunit, Lamb1. eLife 4.

Lu, D.C., Niu, T., and Alaynick, W.A. (2015). Molecular and cellular development of spinal cord locomotor circuitry. Front. Mol. Neurosci. 8, 25.

Lundberg, A. (1981). Half-centres revisited In: Regulatory functions of the CNS. Motion and organization principles (Szentagothai J, Palkovits M, Hamori J., eds) (Budapest, Hungary: Pergamon Akadem Kiado).

MacLean, J.N., Zhang, Y., Johnson, B.R., and Harris-Warrick, R.M. (2003). Activi-ty-Independent Homeostasis in Rhythmically Active Neurons. Neuron 37, 109–120.

Madisen, L., Zwingman, T.A., Sunkin, S.M., Oh, S.W., Zariwala, H.A., Gu, H., Ng, L.L., Palmiter, R.D., Hawrylycz, M.J., Jones, A.R., et al. (2010). A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat. Neurosci. 13, 133–140.

Maltenfort, M.G., Heckman, C.J., and Rymer, W.Z. (1998). Decorrelating actions of Renshaw interneurons on the firing of spinal motoneurons within a motor nu-cleus: a simulation study. J. Neurophysiol. 80, 309–323.

Marshall, P.W., Rouse, W., Briggs, I., Hargreaves, R.B., Mills, S.D., and McLough-lin, B.J. (1993). ICI D7288, a novel sinoatrial node modulator. J. Cardiovasc. Pharmacol. 21, 902–906.

Martin, L.J., and Chang, Q. (2012). Inhibitory synaptic regulation of motoneurons: a new target of disease mechanisms in amyotrophic lateral sclerosis. Mol. Neuro-biol. 45, 30–42.

Marty, A., and Llano, I. (2005). Excitatory effects of GABA in established brain networks. Trends Neurosci. 28, 284–289.

McCrea, D.A., Pratt, C.A., and Jordan, L.M. (1980). Renshaw cell activity and re-current effects on motoneurons during fictive locomotion. J. Neurophysiol. 44, 475–488.

83

McCurdy, M.L., and Hamm, T.M. (1992). Recurrent collaterals of motoneurons projecting to distal muscles in the cat hindlimb. J. Neurophysiol. 67, 1359–1366.

Mentis, G.Z., Siembab, V.C., Zerda, R., O’Donovan, M.J., and Alvarez, F.J. (2006). Primary Afferent Synapses on Developing and Adult Renshaw Cells. J. Neuro-sci. 26, 13297–13310.

Mentis, G.Z., Alvarez, F.J., Shneider, N.A., Siembab, V.C., and O’Donovan, M.J. (2010). Mechanisms regulating the specificity and strength of muscle afferent inputs in the spinal cord. Ann. N. Y. Acad. Sci. 1198, 220–230.

Mikulovic, S., Restrepo, C.E., Hilscher, M.M., Kullander, K., and Leão, R.N. (2015). Novel markers for OLM interneurons in the hippocampus. Front. Cell. Neurosci. 9, 201.

Mikulovic, S., Restrepo, C.E., Bauer, P., Pupe, S., Siwani, S., Tort, A.B.L., Kullander, K., and Leao, R.N. Ventral hippocampus OLMα2 cells control type 2 theta oscillations and response to predator odor.

Miles, T.S. (1994). The control of human motor units. Clin. Exp. Pharmacol. Phys-iol. 21, 511–520.

Miles, G.B., and Sillar, K.T. (2011). Neuromodulation of vertebrate locomotor con-trol networks. Physiol. Bethesda Md 26, 393–411.

Miles, G.B., Dai, Y., and Brownstone, R.M. (2005). Mechanisms underlying the early phase of spike frequency adaptation in mouse spinal motoneurones. J. Physiol. 566, 519–532.

Miller, S., and Scott, P.D. (1977). A model of the spinal locomotor generator in the cat [proceedings]. J. Physiol. 269, 20P–22P.

Minassian, K., Persy, I., Rattay, F., Pinter, M.M., Kern, H., and Dimitrijevic, M.R. (2007). Human lumbar cord circuitries can be activated by extrinsic tonic input to generate locomotor-like activity. Hum. Mov. Sci. 26, 275–295.

Moore, N.J., Bhumbra, G.S., Foster, J.D., and Beato, M. (2015). Synaptic Connec-tivity between Renshaw Cells and Motoneurons in the Recurrent Inhibitory Cir-cuit of the Spinal Cord. J. Neurosci. 35, 13673–13686.

Mor, Y., and Lev-Tov, A. (2007). Analysis of rhythmic patterns produced by spinal neural networks. J. Neurophysiol. 98, 2807–2817.

Moran-Rivard, L., Kagawa, T., Saueressig, H., Gross, M.K., Burrill, J., and Gould-ing, M. (2001). Evx1 is a postmitotic determinant of v0 interneuron identity in the spinal cord. Neuron 29, 385–399.

Moreno-López, Y., Olivares-Moreno, R., Cordero-Erausquin, M., and Rojas-Piloni, G. (2016). Sensorimotor Integration by Corticospinal System. Front. Neuroanat. 10, 24.

Morris, M.E., Summers, J.J., Matyas, T.A., and Iansek, R. (1994). Current status of the motor program. Phys. Ther. 74, 738-748-752.

Mouse Genome Sequencing Consortium, Waterston, R.H., Lindblad-Toh, K., Bir-ney, E., Rogers, J., Abril, J.F., Agarwal, P., Agarwala, R., Ainscough, R., Alex-andersson, M., et al. (2002). Initial sequencing and comparative analysis of the mouse genome. Nature 420, 520–562.

Musienko, P., Brand, R. van den, Maerzendorfer, O., Larmagnac, A., and Courtine*, G. (2009). Combinatory Electrical and Pharmacological Neuroprosthetic Inter-faces to Regain Motor Function After Spinal Cord Injury. IEEE Trans. Biomed. Eng. 56, 2707–2711.

84

Nagaraja, C., Helmchen, F., Peuckert, C., and Kullander, K. Multisegmental sensory inputs modulate Renshaw cell activity during circuit maturation.

Nakagawa, H., Ninomiya, T., Yamashita, T., and Takada, M. (2015). Reorganization of corticospinal tract fibers after spinal cord injury in adult macaques. Sci. Rep. 5, 11986.

Neher, E., and Sakmann, B. (1992). The patch clamp technique. Sci. Am. 266, 44–51.

Nishimaru, H., Restrepo, C.E., Ryge, J., Yanagawa, Y., and Kiehn, O. (2005). Mammalian motor neurons corelease glutamate and acetylcholine at central synapses. Proc. Natl. Acad. Sci. U. S. A. 102, 5245–5249.

Nishimaru, H., Restrepo, C.E., and Kiehn, O. (2006). Activity of Renshaw cells during locomotor-like rhythmic activity in the isolated spinal cord of neonatal mice. J. Neurosci. Off. J. Soc. Neurosci. 26, 5320–5328.

Nishimaru, H., Koganezawa, T., Kakizaki, M., Ebihara, T., and Yanagawa, Y. (2010). Inhibitory synaptic modulation of renshaw cell activity in the lumbar spinal cord of neonatal mice. J. Neurophysiol. 103, 3437–3447.

Noga, B.R., Shefchyk, S.J., Jamal, J., and Jordan, L.M. (1987). The role of Renshaw cells in locomotion: antagonism of their excitation from motor axon collaterals with intravenous mecamylamine. Exp. Brain Res. 66, 99–105.

Obeidat, A.Z., Nardelli, P., Powers, R.K., and Cope, T.C. (2014). Modulation of motoneuron firing by recurrent inhibition in the adult rat in vivo. J. Neurophys-iol. 112, 2302–2315.

O’Donovan, M.J. (1999). The origin of spontaneous activity in developing networks of the vertebrate nervous system. Curr. Opin. Neurobiol. 9, 94–104.

O’Donovan, M.J., Bonnot, A., Mentis, G.Z., Chub, N., Pujala, A., and Alvarez, F.J. (2010). Mechanisms of excitation of spinal networks by stimulation of the ven-tral roots. Ann. N. Y. Acad. Sci. 1198, 63–71.

Pape, H.C. (1996). Queer current and pacemaker: the hyperpolarization-activated cation current in neurons. Annu. Rev. Physiol. 58, 299–327.

Pasquali, L., Longone, P., Isidoro, C., Ruggieri, S., Paparelli, A., and Fornai, F. (2009). Autophagy, lithium, and amyotrophic lateral sclerosis. Muscle Nerve 40, 173–194.

Perry, S., Gezelius, H., Larhammar, M., Hilscher, M.M., Lamotte d’Incamps, B., Leao, K.E., and Kullander, K. (2015). Firing properties of Renshaw cells de-fined by Chrna2 are modulated by hyperpolarizing and small conductance ion currents Ih and ISK. Eur. J. Neurosci. 41(7), 889-900.

Phukan, J., Albanese, A., Gasser, T., and Warner, T. (2011). Primary dystonia and dystonia-plus syndromes: clinical characteristics, diagnosis, and pathogenesis. Lancet Neurol. 10, 1074–1085.

Placzek, M. (1995). The role of the notochord and floor plate in inductive interac-tions. Curr. Opin. Genet. Dev. 5, 499–506.

Placzek, M., Yamada, T., Tessier-Lavigne, M., Jessell, T., and Dodd, J. (1991). Control of dorsoventral pattern in vertebrate neural development: induction and polarizing properties of the floor plate. Dev. Camb. Engl. Suppl. Suppl 2, 105–122.

Pratt, C.A., and Jordan, L.M. (1987). Ia inhibitory interneurons and Renshaw cells as contributors to the spinal mechanisms of fictive locomotion. J. Neurophysiol. 57, 56–71.

85

Prescott, S.A., De Koninck, Y., and Sejnowski, T.J. (2008). Biophysical basis for three distinct dynamical mechanisms of action potential initiation. PLoS Com-put. Biol. 4, e1000198.

Prinz, A.A., Bucher, D., and Marder, E. (2004). Similar network activity from dis-parate circuit parameters. Nat. Neurosci. 7, 1345–1352.

Promerová, M., Andersson, L.S., Juras, R., Penedo, M.C.T., Reissmann, M., Tozaki, T., Bellone, R., Dunner, S., Hořín, P., Imsland, F., et al. (2014). Worldwide fre-quency distribution of the “Gait keeper” mutation in the DMRT3 gene. Anim. Genet. 45, 274–282.

Purves, D., Augustine, G.J., Fitzpatrick, D., Hall, W.C., LaMantia, A.-S., and White, L.E. (2012). Neuroscience Fifth Edition (Sunderland, MA: Sinauer Associates).

Quinlan, K.A., and Kiehn, O. (2007). Segmental, synaptic actions of commissural interneurons in the mouse spinal cord. J. Neurosci. Off. J. Soc. Neurosci. 27, 6521–6530.

Rabe, N., Gezelius, H., Vallstedt, A., Memic, F., and Kullander, K. (2009). Netrin-1-dependent spinal interneuron subtypes are required for the formation of left-right alternating locomotor circuitry. J. Neurosci. Off. J. Soc. Neurosci. 29, 15642–15649.

Rahman, J., Besser, S., Schnell, C., Eulenburg, V., Hirrlinger, J., Wojcik, S.M., and Hülsmann, S. (2015). Genetic ablation of VIAAT in glycinergic neurons causes a severe respiratory phenotype and perinatal death. Brain Struct. Funct. 220, 2835–2849.

Ramírez-Jarquín, U.N., Lazo-Gómez, R., Tovar-y-Romo, L.B., and Tapia, R. (2013) Spinal inhibitory circuits and their role in motor neuron degeneration. Neuro-pharmacology.

Raynor, E.M., and Shefner, J.M. (1994). Recurrent inhibition is decreased in pa-tients with amyotrophic lateral sclerosis. Neurology 44, 2148–2153.

Renshaw, B. (1941). Influence of Discharge of Motoneurons Upon Excitation of Neighboring Motoneurons. J. Neurophysiol. 4, 167–183.

Renshaw, B. (1946). Observations on interaction of nerve impulses in the gray mat-ter and on the nature of central inhibition. Am. J. Physiol. 146, 443–448.

Rexed, B. (1952). The cytoarchitectonic organization of the spinal cord in the cat. J. Comp. Neurol. 96, 415–495.

Rexed, B. (1954). A cytoarchitectonic atlas of the spinal coed in the cat. J. Comp. Neurol. 100, 297–379.

Ribchester, R.R. (2009). Mammalian neuromuscular junctions: modern tools to monitor synaptic form and function. Curr. Opin. Pharmacol. 9, 297–305.

Richards, D.S., Griffith, R.W., Romer, S.H., and Alvarez, F.J. (2014). Motor axon synapses on renshaw cells contain higher levels of aspartate than glutamate. PloS One 9, e97240.

Robinson, G.A., and Goldberger, M.E. (1986). The development and recovery of motor function in spinal cats. I. The infant lesion effect. Exp. Brain Res. 62, 373–386.

Ross, H.G., Cleveland, S., and Haase, J. (1975). Contribution of single motoneurons to renshaw cell activity. Neurosci. Lett. 1, 105–108.

Ryall, R.W. (1970). Renshaw cell mediated inhibition of Renshaw cells: patterns of excitation and inhibition from impulses in motor axon collaterals. J. Neurophys-iol. 33, 257–270.

86

Ryall, R.W. (1981). Patterns of recurrent excitation and mutual inhibition of cat Renshaw cells. J. Physiol. 316, 439–452.

Ryall, R.W., Piercey, M.F., and Polosa, C. (1971). Intersegmental and intrasegmen-tal distribution of mutual inhibition of Renshaw cells. J. Neurophysiol. 34, 700–707.

Rybak, I.A., Shevtsova, N.A., Lafreniere-Roula, M., and McCrea, D.A. (2006a). Modelling spinal circuitry involved in locomotor pattern generation: insights from deletions during fictive locomotion. J. Physiol. 577, 617–639.

Rybak, I.A., Stecina, K., Shevtsova, N.A., and McCrea, D.A. (2006b). Modelling spinal circuitry involved in locomotor pattern generation: insights from the ef-fects of afferent stimulation. J. Physiol. 577, 641–658.

Rybak, I.A., Dougherty, K.J., and Shevtsova, N.A. (2015). Organization of the Mammalian Locomotor CPG: Review of Computational Model and Circuit Ar-chitectures Based on Genetically Identified Spinal Interneurons(1,2,3). eNeuro 2.

Sakmann, B., and Neher, E. (1984). Patch clamp techniques for studying ionic chan-nels in excitable membranes. Annu. Rev. Physiol. 46, 455–472.

Santoro, B., and Baram, T.Z. (2003). The multiple personalities of h-channels. Trends Neurosci. 26, 550–554.

Santoro, B., Chen, S., Luthi, A., Pavlidis, P., Shumyatsky, G.P., Tibbs, G.R., and Siegelbaum, S.A. (2000). Molecular and functional heterogeneity of hyperpolar-ization-activated pacemaker channels in the mouse CNS. J. Neurosci. Off. J. Soc. Neurosci. 20, 5264–5275.

Sapir, T., Geiman, E.J., Wang, Z., Velasquez, T., Mitsui, S., Yoshihara, Y., Frank, E., Alvarez, F.J., and Goulding, M. (2004). Pax6 and engrailed 1 regulate two distinct aspects of renshaw cell development. J. Neurosci. Off. J. Soc. Neurosci. 24, 1255–1264.

Sauer, B., and Henderson, N. (1988). Site-specific DNA recombination in mamma-lian cells by the Cre recombinase of bacteriophage P1. Proc. Natl. Acad. Sci. U. S. A. 85, 5166–5170.

Schneider, S.P., and Fyffe, R.E. (1992). Involvement of GABA and glycine in recur-rent inhibition of spinal motoneurons. J. Neurophysiol. 68, 397–406.

Schofield, B.R. (2008). Retrograde axonal tracing with fluorescent markers. Curr. Protoc. Neurosci. Editor. Board Jacqueline N Crawley Al Chapter 1, Unit 1.17.

Seifert, R., Scholten, A., Gauss, R., Mincheva, A., Lichter, P., and Kaupp, U.B. (1999). Molecular characterization of a slowly gating human hyperpolarization-activated channel predominantly expressed in thalamus, heart, and testis. Proc. Natl. Acad. Sci. U. S. A. 96, 9391–9396.

Sherrington, C.S. (1906). The Integrative Action of the Nervous System (New Ha-ven: Yale Univ. Press).

Sherrington, C.S. (1925). Remarks on some Aspects of Reflex Inhibition. Proc. R. Soc. Lond. Ser. B Contain. Pap. Biol. Character 97, 519–545.

Shigeri, Y., Seal, R.P., and Shimamoto, K. (2004). Molecular pharmacology of glutamate transporters, EAATs and VGLUTs. Brain Res. Brain Res. Rev. 45, 250–265.

Siegel, C.S., Fink, K.L., Strittmatter, S.M., and Cafferty, W.B.J. (2015). Plasticity of intact rubral projections mediates spontaneous recovery of function after corti-cospinal tract injury. J. Neurosci. Off. J. Soc. Neurosci. 35, 1443–1457.

87

Siembab, V.C., Smith, C.A., Zagoraiou, L., Berrocal, M.C., Mentis, G.Z., and Alva-rez, F.J. (2010). Target selection of proprioceptive and motor axon synapses on neonatal V1-derived Ia inhibitory interneurons and Renshaw cells. J. Comp. Neurol. 518, 4675–4701.

Siembab, V.C., Gomez-Perez, L., Rotterman, T.M., Shneider, N.A., and Alvarez, F.J. (2016). Role of primary afferents in the developmental regulation of motor axon synapse numbers on Renshaw cells. J. Comp. Neurol. n/a-n/a.

Silberberg, G., and Markram, H. (2007). Disynaptic inhibition between neocortical pyramidal cells mediated by Martinotti cells. Neuron 53, 735–746.

Sivilotti, L., and Woolf, C.J. (1994). The contribution of GABAA and glycine re-ceptors to central sensitization: disinhibition and touch-evoked allodynia in the spinal cord. J. Neurophysiol. 72, 169–179.

Song, J., Ampatzis, K., Björnfors, E.R., and El Manira, A. (2016). Motor neurons control locomotor circuit function retrogradely via gap junctions. Nature 529, 399–402.

Stam, F.J., Hendricks, T.J., Zhang, J., Geiman, E.J., Francius, C., Labosky, P.A., Clotman, F., and Goulding, M. (2012). Renshaw cell interneuron specialization is controlled by a temporally restricted transcription factor program. Dev. Camb. Engl. 139, 179–190.

Stokke, M.F., Nissen, U.V., Glover, J.C., and Kiehn, O. (2002). Projection patterns of commissural interneurons in the lumbar spinal cord of the neonatal rat. J. Comp. Neurol. 446, 349–359.

Swensen, A.M., and Bean, B.P. (2005). Robustness of burst firing in dissociated purkinje neurons with acute or long-term reductions in sodium conductance. J. Neurosci. Off. J. Soc. Neurosci. 25, 3509–3520.

Szczupak, L. (2014). Recurrent inhibition in motor systems, a comparative analysis. J. Physiol.-Paris 108, 148–154.

Talpalar, A.E., and Kiehn, O. (2010). Glutamatergic Mechanisms for Speed Control and Network Operation in the Rodent Locomotor CPG. Front. Neural Circuits 4.

Talpalar, A.E., Endo, T., Löw, P., Borgius, L., Hägglund, M., Dougherty, K.J., Ryge, J., Hnasko, T.S., and Kiehn, O. (2011). Identification of minimal neuronal networks involved in flexor-extensor alternation in the mammalian spinal cord. Neuron 71, 1071–1084.

Talpalar, A.E., Bouvier, J., Borgius, L., Fortin, G., Pierani, A., and Kiehn, O. (2013). Dual-mode operation of neuronal networks involved in left-right alter-nation. Nature 500, 85–88.

Thibaudier, Y., Hurteau, M.-F., Telonio, A., and Frigon, A. (2013). Coordination between the fore- and hindlimbs is bidirectional, asymmetrically organized, and flexible during quadrupedal locomotion in the intact adult cat. Neuroscience 240, 13–26.

Thomas, R.C., and Wilson, V.J. (1965). Precise localization of Renshaw cells with a new marking technique. Nature 206, 211–213.

Tong, Q., Ye, C.-P., Jones, J.E., Elmquist, J.K., and Lowell, B.B. (2008). Synaptic release of GABA by AgRP neurons is required for normal regulation of energy balance. Nat. Neurosci. 11, 998–1000.

Tresch, M.C., and Kiehn, O. (1999). Coding of locomotor phase in populations of neurons in rostral and caudal segments of the neonatal rat lumbar spinal cord. J. Neurophysiol. 82, 3563–3574.

88

Vallstedt, A., and Kullander, K. (2013). Dorsally derived spinal interneurons in locomotor circuits. Ann. N. Y. Acad. Sci. 1279, 32–42.

Vinay, L., and Clarac, F. (1999). Antidromic discharges of dorsal root afferents and inhibition of the lumbar monosynaptic reflex in the neonatal rat. Neuroscience 90, 165–176.

Wallén-Mackenzie, A., Gezelius, H., Thoby-Brisson, M., Nygård, A., Enjin, A., Fujiyama, F., Fortin, G., and Kullander, K. (2006). Vesicular glutamate trans-porter 2 is required for central respiratory rhythm generation but not for locomo-tor central pattern generation. J. Neurosci. Off. J. Soc. Neurosci. 26, 12294–12307.

Wang, X. (2009). Cre transgenic mouse lines. Methods Mol. Biol. Clifton NJ 561, 265–273.

Wang, Z., Li, L., Goulding, M., and Frank, E. (2008). Early postnatal development of reciprocal Ia inhibition in the murine spinal cord. J. Neurophysiol. 100, 185–196.

Wenner, P. (2014). Homeostatic synaptic plasticity in developing spinal networks driven by excitatory GABAergic currents. Neuropharmacology 78, 55–62.

Wenner, P., and O’Donovan, M.J. (1999). Identification of an Interneuronal Popula-tion that Mediates Recurrent Inhibition of Motoneurons in the Developing Chick Spinal Cord. J. Neurosci. 19, 7557–7567.

Wilhelm, J.C., and Wenner, P. (2008). GABAA transmission is a critical step in the process of triggering homeostatic increases in quantal amplitude. Proc. Natl. Acad. Sci. U. S. A. 105, 11412–11417.

Wilhelm, J.C., Rich, M.M., and Wenner, P. (2009). Compensatory changes in cellu-lar excitability, not synaptic scaling, contribute to homeostatic recovery of em-bryonic network activity. Proc. Natl. Acad. Sci. U. S. A. 106, 6760–6765.

Wilson, L., and Maden, M. (2005). The mechanisms of dorsoventral patterning in the vertebrate neural tube. Dev. Biol. 282, 1–13.

Wilson, J.M., Hartley, R., Maxwell, D.J., Todd, A.J., Lieberam, I., Kaltschmidt, J.A., Yoshida, Y., Jessell, T.M., and Brownstone, R.M. (2005). Conditional rhythmicity of ventral spinal interneurons defined by expression of the Hb9 homeodomain protein. J. Neurosci. Off. J. Soc. Neurosci. 25, 5710–5719.

Windhorst, U. (1996). On the role of recurrent inhibitory feedback in motor control. Prog. Neurobiol. 49, 517–587.

Windhorst, U. (2007). Muscle proprioceptive feedback and spinal networks. Brain Res. Bull. 73, 155–202.

Wojcik, S.M., Katsurabayashi, S., Guillemin, I., Friauf, E., Rosenmund, C., Brose, N., and Rhee, J.-S. (2006). A shared vesicular carrier allows synaptic corelease of GABA and glycine. Neuron 50, 575–587.

Wolpert, L., Beddington, R., Brockes, J., Jessell, T., Lawrence, P., and Meyerowitz, E. (1998). Principles of Development (Oxford: Oxford University Press).

Wootz, H., Fitzsimons-Kantamneni, E., Larhammar, M., Rotterman, T.M., Enjin, A., Patra, K., André, E., Van Zundert, B., Kullander, K., and Alvarez, F.J. (2013). Alterations in the motor neuron-renshaw cell circuit in the Sod1(G93A) mouse model. J. Comp. Neurol. 521, 1449–1469.

Wu, L., Sonner, P.M., Titus, D.J., Wiesner, E.P., Alvarez, F.J., and Ziskind-Conhaim, L. (2011). Properties of a Distinct Subpopulation of GABAergic Commissural Interneurons That Are Part of the Locomotor Circuitry in the Ne-onatal Spinal Cord. J. Neurosci. 31, 4821–4833.

89

Wu, X., Liao, L., Liu, X., Luo, F., Yang, T., and Li, C. (2012). Is ZD7288 a selec-tive blocker of hyperpolarization-activated cyclic nucleotide-gated channel cur-rents? Channels Austin Tex 6, 438–442.

Yaksh, T.L. (1989). Behavioral and autonomic correlates of the tactile evoked allo-dynia produced by spinal glycine inhibition: effects of modulatory receptor sys-tems and excitatory amino acid antagonists. Pain 37, 111–123.

Yamada, T., Placzek, M., Tanaka, H., Dodd, J., and Jessell, T.M. (1991). Control of cell pattern in the developing nervous system: polarizing activity of the floor plate and notochord. Cell 64, 635–647.

Zeilhofer, H.U., Wildner, H., and Yévenes, G.E. (2012). Fast synaptic inhibition in spinal sensory processing and pain control. Physiol. Rev. 92, 193–235.

Zhang, J., Lanuza, G.M., Britz, O., Wang, Z., Siembab, V.C., Zhang, Y., Velasquez, T., Alvarez, F.J., Frank, E., and Goulding, M. (2014). V1 and V2b Interneurons Secure the Alternating Flexor-Extensor Motor Activity Mice Require for Limbed Locomotion. Neuron 82, 138–150.

Zhang, Y., Narayan, S., Geiman, E., Lanuza, G.M., Velasquez, T., Shanks, B., Akay, T., Dyck, J., Pearson, K., Gosgnach, S., et al. (2008). V3 Spinal Neurons Establish a Robust and Balanced Locomotor Rhythm during Walking. Neuron 60, 84–96.

Zhao, Y., Inayat, S., Dikin, D.A., Singer, J.H., Ruoff, R.S., and Troy, J.B. (2008). Patch clamp technique: Review of the current state of the art and potential con-tributions from nanoengineering. Proc. Inst. Mech. Eng. Part N J. Nanoeng. Nanosyst. 222, 1–11.

Zhong, G., Droho, S., Crone, S.A., Dietz, S., Kwan, A.C., Webb, W.W., Sharma, K., and Harris-Warrick, R.M. (2010). Electrophysiological Characterization of V2a Interneurons and Their Locomotor-Related Activity in the Neonatal Mouse Spi-nal Cord. J. Neurosci. 30, 170–182.

Zhu, J., Feng, F., Ni, K., Zheng, Y., and Zhang, J. (2012). Four subtypes of motor neurons exhibiting mutually exclusive firing patterns in the spinal ventral horn. Neurosci. Lett. 525, 163–167.

Zilberter, Y. (2015). Commentary: GABA Depolarizes Immature Neurons and In-hibits Network Activity in the Neonatal Neocortex In vivo. Front. Pharmacol. 6.

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Medicine 1272

Editor: The Dean of the Faculty of Medicine

A doctoral dissertation from the Faculty of Medicine, UppsalaUniversity, is usually a summary of a number of papers. A fewcopies of the complete dissertation are kept at major Swedishresearch libraries, while the summary alone is distributedinternationally through the series Digital ComprehensiveSummaries of Uppsala Dissertations from the Faculty ofMedicine. (Prior to January, 2005, the series was publishedunder the title “Comprehensive Summaries of UppsalaDissertations from the Faculty of Medicine”.)

Distribution: publications.uu.seurn:nbn:se:uu:diva-305601

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2016