Regulation of the Motor Output of the Spinal Cord

Home , Mephenesin

REGULATION OF THE MOTOR OUTPUT OF THE SPINAL CORD:

BURST FIRING GENERATION AND SENSORIMOTOR

INTEGRATION

A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy

AMR A. MAHROUS B.S., Cairo University, 2006 M.S., Cairo University, 2012

2018 Wright State University

AMR A. MAHROUS

2018

WRIGHT STATE UNIVERSITY

GRADUATE SCHOOL

April 4, 2018 I HEREBY RECOMMEND THAT THE DISSERTATION PREPARED UNDER MY SUPERVISION BY Amr A. Mahrous ENTITLED Regulation of the motor output of the spinal cord: burst firing generation and sensorimotor integration BE ACCEPTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF Doctor of Philosophy.

Sherif Elbasiouny, Ph.D. Dissertation Director

Mill W. Miller, Ph.D. Director, Biomedical Sciences Ph.D. Program

Barry Milligan, Ph.D. Interim Dean of the Graduate School Committee on Final Examination

Sherif Elbasiouny, Ph.D.

Mark Rich, Ph.D.

Mill W. Miller, Ph.D.

David Ladle, Ph.D.

Lynn Hartzler, Ph.D.

ABSTRACT

Mahrous, Amr A. Ph.D., Biomedical Sciences Ph.D. program, Wright State University, 2018. Regulation of the motor output of the spinal cord: burst firing generation and sensorimotor integration

Spinal motoneurons are the final common path for motor signals. Thus, the firing activity recorded from spinal motoneurons or their axons in the ventral roots represents the motor output of the nervous system. This dissertation investigates two major phenomena that help generate and shape the neuronal motor output.

Specialized groups of spinal neurons called central pattern generators (CPGs) are capable of orchestrating rhythmic bursting activity without involvement of the brain. This activity underlies stereotyped movements such as locomotion. If known, the ionic mechanisms responsible for generating this activity can offer new pharmacological interventions to restore walking in paralyzed patients with intact spinal circuits. In part-I of this dissertation, we examined the role of the small conductance Ca2+-activated potassium (SK) channels in the generation of rhythmic activity. SK channels represent a strong candidate for controlling bursting behaviors because they have a dual regulatory function on both the synaptic inputs and the firing output of spinal neurons. Rhythmic activity was studied in murine spinal cord preparations in vitro using intracellular and extracellular electrophysiological recordings. The data showed that inhibition of SK channels -using multiple approaches- facilitates the initiation of rhythmic motor outputs. Additionally,

iv graded pharmacological modulation of SK channels determined the amplitude of the motor bursts in a dose-dependent manner. Similar manipulations of other types of ion channels did not replicate these effects.

The autonomous activity of the spinal CPGs can continue for long periods of time without extrinsic inputs. However, to meet the demands of the environment, the locomotor activity has to undergo continuous modulation by peripheral sensory inputs and descending supraspinal inputs (sensorimotor integration). These synaptic inputs exhibit use-dependent plasticity at physiological firing frequencies (either facilitation or depression). Using a combination of intracellular and extracellular recordings, part-II of this dissertation investigated how variable sensory and descending inputs of different frequencies, amplitudes, and neuromodulatory states generate a steady motor output at the cellular and system levels. The data indicated that integration of multiple, despite adapting, excitatory inputs help generate a stable motor output by maintaining the synaptic potentials above the firing threshold, which was more readily achievable at higher neuromodulatory states.

TABLE OF CONTENT

INTRODUCTION ...... 1

Locomotion and rhythmic activity...... 2

Central pattern generators (CPGs) ...... 3

Neuronal bursting in the spinal cord...... 4

Origin of alternating and synchronized bursting ...... 5

SK channels: gating and function ...... 7

SK channels in motoneurons ...... 8

SK channels and burst firing ...... 9

Sensorimotor integration in the spinal cord ...... 10

Sensorimotor integration in spinal motoneurons ...... 12

Neuromodulation of networks, neurons, and synapses ...... 13

HYPOTHESES, SPECIFIC AIMS, AND CLINICAL SIGNIFICANCE

...... 15

PART I: Cellular mechanisms underlying rhythmic motor outputs ...... 15

PART II: Sensorimotor network inputs mediating stable motor output ...... 15

Clinical significance of the research questions ...... 16

MATERIALS AND METHODS ...... 18

Animals ...... 18

In vitro preparations ...... 18

Physiological solutions ...... 20

Electrophysiological recordings ...... 20

Rhythmic motor bursting activity ...... 23

Sensory and descending inputs in the sacral cord ...... 24

Drugs and chemicals ...... 27

Statistical analysis...... 27

RESULTS ...... 31

PART-I: CELLULAR MECHANISMS UNDERLYING RHYTHMIC MOTOR

OUTPUTS...... 31

SPECIFIC AIM 1: TO STUDY THE ROLE OF SK CHANNELS IN THE

INITIATION OF MOTOR BURSTS IN THE MAMMALIAN SPINAL CORD. ... 31

Specific aim 1A: The role of SK channels in the initiation of right-left alternating

bursting in the neonatal spinal cord...... 31

Specific aim 1B: The role of SK channels in the initiation of synchronized burst

firing in the adult spinal cord...... 38

SPECIFIC AIM 2: TO INVESTIGATE THE ROLE OF SK CHANNELS IN MOTOR

BURST AMPLITUDE MODULATION...... 52

Specific aim 2A: Regulation of synchronized burst amplitude by graded SK channel

inhibition...... 52

Specific aim 2B: The role of gap junctions in synchronizing motor bursts and/or

setting the burst amplitude...... 58

vii

PART-II: SENSORIMOTOR NETWORK INPUTS MEDIATING STABLE

MOTOR OUTPUT ...... 70

SPECIFIC AIM 3: TO STUDY THE PLASTICITY PROFILE AND INTEGRATION

OF SENSORY AND MOTOR INPUTS TO MOTONEURONS ...... 70

Specific aim 3A: Plasticity patterns of sensory and motor inputs in the isolated sacral

cord...... 71

Specific aim 3B: Spinal sensorimotor integration at the cellular and system levels.

...... 87

DISCUSSION ...... 108

THE ROLE OF SK CHANNELS IN THE INITIATION AND AMPLITUDE

MODULATION OF RHYTHMIC MOTOR ACTIVITY ...... 109

Induction of burst firing in the mature spinal cord tissue in vitro ...... 109

Evidence suggesting SK channel inhibition during bursting...... 110

SK channel inhibition facilitates burst initiation ...... 111

The possible mechanisms underlying burst initiation by SK inhibition ...... 112

SK inhibition and burst amplitude modulation...... 114

Specific SK channel role vs. non-specific excitability modulation ...... 115

Synaptic conductances mediating bursting activity ...... 116

Cholinergic modulation of motoneuron bursting through SK channels ...... 118

The effect of SK modulators on motoneurons vs. interneurons ...... 119

SENSORIMOTOR INTEGRATION IN THE SPINAL CORD ...... 121

viii

Short-term plasticity of sensory and motor inputs ...... 121

Integration of sensory and descending (motor) inputs ...... 124

REFERENCES ...... 127

LIST OF FIGURES

Figure 1: Different forms of rhythmic bursting induced in vitro in the isolated neonatal rat spinal cord preparation...... 6

Figure 2: Major excitatory inputs to motoneurons in the spinal cord...... 11

Figure 3: Electrophysiology experimental set-up for recording spontaneous rhythmic bursting activity...... 22

Figure 4: Experimental setup for recording synaptic responses to sensory and descending stimulation...... 26

Figure 5: Neurotransmitter-induced alternating bursting in the spinal cord of functionally- mature mice...... 33

Figure 6: Effect of methoxamine on alternating activity in the lumbosacral cord...... 35

Figure 7: Locomotor bursts are inhibited by SK activators and restored by SK blockers.37

Figure 8: Synchronized bursts induced in the adult spinal cord...... 40

Figure 9: SK channel inhibition facilitates synchronized bursting initiation...... 41

Figure 10: Effect of reducing extracellular free Ca2+ on burst initiation...... 43

Figure 11: Muscarinic inhibition of SK channels facilitates bursting...... 45

Figure 12: In absence of SK channel inhibitors, electrical stimulation does not evoke bursting in the disinhibited spinal cord...... 47

Figure 13: Blocking Ca2+PIC does not affect burst initiation or amplitude...... 50

Figure 14: SK channel inhibition is specifically necessary for burst initiation...... 51

Figure 15: Graded direct blockade of SK channels modulated the burst amplitude...... 54

Figure 16: Physiological inhibition of SK channels grades the burst amplitude...... 55

Figure 17: Burst amplitude gradation requires availability of SK channels...... 57

Figure 18: Different levels of synchrony during spontaneous synchronized bursting. .... 59

Figure 19: Gap junctions do not contribute to burst amplitude or synchrony...... 61

Figure 20: Synchronized bursts are not exclusively-driven by NMDA receptors...... 63

Figure 21: Both AMPA and NMDA receptors generate the synchronized bursts...... 64

Figure 22: The effect of apamin vs. strychnine/picrotoxin on the low- amplitude/low- frequency inputs to the motor pool...... 67

Figure 23: The effect of apamin vs. strychnine/picrotoxin on the high-amplitude/high- frequency inputs to the motor pool...... 68

Figure 24: synchronized bursting is evoked through polysynaptic pathways...... 69

Figure 25: Motor pool and single cell responses to 25 Hz 5-pulse train of electrical stimulation of the dorsal roots in the sacral cord...... 74

Figure 26: Plasticity pattern of the sensory inputs at the system and cellular levels...... 76

Figure 27: Depression of the sensory response is not caused by stimulation failure...... 77

Figure 28: Response to 25 Hz 5-pulse train of electrical stimulation of the descending axons in the sacral cord ...... 80

Figure 29: Plasticity pattern of the descending inputs at the system and cellular levels. . 82

Figure 30: Descending stimulation does not spread non-specifically in the tissue...... 83

Figure 31: Measurements of timing parameters for the synaptic potentials...... 85

Figure 32: Response to 25 Hz 5-pulse train of simultaneous stimulation of the dorsal roots and descending axons...... 88

Figure 33: Integration of sensory and motor inputs at 1.5xT/25 Hz stimulation...... 91

Figure 34: Integration of sensory and motor inputs at 1.5xT/50 Hz stimulation...... 93

Figure 35: Dependence of EPSP integration on driving force...... 96

Figure 36: Integration of sensory and motor inputs at 10xT/25 Hz stimulation...... 99

Figure 37: Integration of sensory and motor inputs at 10xT/50 Hz stimulation...... 101

Figure 38: The effect of methoxamine on plasticity and integration of sensorimotor inputs at the system level...... 106

Figure 39: Methoxamine increases the amplitude and slows the decay of sensory and motor synaptic potentials...... 107

Figure 40: Possible mechanisms for regulation of motor rhythmic bursting by SK channels in the spinal cord...... 113

xii

LIST OF TABLES

Table 1: Drugs and chemicals used in the current study...... 29

Table 2: Different timing parameters for the synaptic potentials of the sensory and motor inputs...... 86

Table 3: Parameters for computer simulations of two excitatory synapses in a membrane patch...... 95

Table 4: Summary of the type of summation of sensorimotor inputs at different intensities/frequencies combinations...... 102

xiii

LIST OF ABBREVIATIONS

5-HT 5-Hydroxytryptamine (serotonin)

ACSF Artificial cerebrospinal fluid

AHP / mAHP Afterhyperpolarization / medium afterhyperpolarization

AMPA α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid

AMPARs AMPA receptors

BIC Bicuculline

CoAP Compound action potential

CPG Central pattern generator

DA Dopamine

EPSP Excitatory postsynaptic potential

GABA gamma-aminobutyric acid

Gi Gigantocellular reticular nucleus

L Lx / R Lx Left side of the lumbar spinal segment number (x) / Right side of the same segment

LSx / RSx Left side of the sacral spinal segment number (x) / Right side of the same segment

NMDA N-methyl D-aspartate

NMDARs NMDA receptors

PIC Persistent inward current

PnO Pontine reticular nucleus

PTX Picrotoxin

xiv

RM one/two-way Repeated-measures one/two-way analysis of variance ANOVA

S&M Simultaneous stimulation of sensory and motor synaptic inputs

S+M Mathematical (theoretical linear) summation of sensory and motor responses

SEM Standard error of the mean

SK Small conductance calcium-activated potassium channel

STD Short-term depression

STF Short-term facilitation

STR Strychnine xT Times threshold

ACKNOWLEDGEMENT

The work in this dissertation wouldn’t have come to fruition without the help, support, and encouragement I received from several people who, each in their own way, made my time in the lab more productive and enjoyable. First, I would like to thank my great mentor, favorite colleague, and exceptional friend Dr. Sherif Elbasiouny. I owe him a debt of gratitude for investing long hours in nurturing and modeling my scientific mentality! I truly appreciate everything he has done over the past three years for my professional, and personal development. I would also like to thank all the current and previous members of the “NERD” lab for all the thoughtful discussions in the lab meetings, helpful advice, and fun they shared with me. Special thanks to Teresa Garrett who made my life in the lab easier by ensuring I always had all the animals and supplies that I needed.

My sincere thanks to Dr. David Bennett and all the members of his lab at the University of

Alberta, especially Dr. Yaqing Li, for hosting me in the lab during my training and giving me invaluable technical advices on performing the electrophysiological recordings. I am also thankful to Drs. Mark Rich, Mill Miller, David Ladle, and Lynn Hartzler, for serving on my committee and for all their support, comments, feedback and guidance in navigating me through the different stages of the program. I owe them my gratitude for sharing their scientific perspective and expertise with me and sometimes their lab equipment as well!

xvi

I owe special gratitude to the current and former BMS directors, Drs. Mill Miller and

Gerald Alter, and to Karen Luchin for their numerous contributions to my journey at

Wright State.

Special thanks to my classmate, BMS colleague, and dear friend Dr. Mahmoud Alghamri for all the snowy nights we stayed up late talking about science, family, and home. Last but not least, I am very grateful to my mentors and colleagues at the Faculty of Pharmacy,

Cairo University for their contribution to my scientific background, and to Dr. Robert

Putnam for planting the seeds of love for electrophysiology in my heart.

xvii

DEDICATION

This dissertation is dedicated to my family for their unconditional love and unwavering

support.

To my parents, Nadia and Abdelhameed, for teaching me to dream with no limits, and for

devoting their lives to help me and my siblings attain our full potential.

To my lovely wife, Lobna and our children, Ahmad and Kareem for making me cheerful

and keeping me sane.

To my siblings, Shereen, Mahmoud, and Shaimaa for all the beautiful memories we share

and unstoppable encouragement.

Thank you!

xviii

INTRODUCTION

“To move things is all that mankind can do, and for such the sole executant is muscle, whether in whispering a syllable or felling a forest.”

Sir Charles Sherrington (1857-1952).

A huge number of sensory inputs constantly reach the brain and spinal cord encoding various environmental information. These inputs are integrated in complex neuronal networks to generate an appropriate output. Despite the diverse nature of the sensory information, the resulting output of the nervous system is almost always in the form of a movement. The flow of neuronal signals to different types of muscles underlie the regulation of multiple physiological systems such as the circulatory, respiratory, digestive, urinary, and musculoskeletal systems. Therefore, understanding how different forms of movement are initiated in the nervous system and how their amplitude is modulated to meet various needs is crucial for treatment of many neurological disorders. Although, the work presented in this dissertation focuses on the neuronal control of skeletal muscles and limb movement, the basic concepts investigated here can be extended to other forms of neuronal outputs.

Locomotion and rhythmic activity

Locomotion (the ability to move) is a crucial skill for survival that allows the search for food, escape from predators, or migration to areas of more tolerable weather conditions. It also facilitates the interaction with the surrounding environment in many different ways. It takes different forms in different species, such as flying in birds, swimming in fish, and walking in other animals. Even though humans and animals perform these motor activities seemingly mindless, most of these tasks are difficult to replicate in sophisticated robots.

This is because movements -even as simple as walking- are the result of complex integrated neuronal network activities and highly coordinated muscle contractions. This complexity usually becomes appreciated when the motor nervous system is damaged by injury, ischemia or degenerative diseases.

The command for initiation of voluntary movements takes place in supraspinal centers including the cortex (Drew and Marigold 2015) and basal ganglia (Grillner and Robertson

2015). Repetition and practice can drive neural plasticity in these structures which underlie the development of what is commonly known as the “muscle memory”. This neuroplasticity can enable a person to perform quite varied and complex patterns of motor activity such as playing a musical instrument or performing a complicated dance.

Nonetheless, simpler repetitive/stereotyped movements can be generated without much involvement from the brain. The spinal cord, in all vertebrates, contains the basic networks needed to coordinate simple rhythmic activity, such as breathing, walking, swimming, flying, peristalsis, chewing, and scratching (Grillner 1975). For example, locomotive limb

2 movements have been seen in decerebrate1 animal preparations (Steeves, Sholomenko, and

Webster 1987, Meehan et al. 2012, Whelan 1996).

Central pattern generators (CPGs)

It was initially believed that the sensory feedback from the skin and muscles in decerebrate preparations was responsible for generating coordinated movements. In the early twentieth century, however, T. Graham Brown showed flexor-extensor alternating activity in a decerebrate de-afferented2 spinal3 cat (Brown 1911). This indicated the existence of specialized cells in the spinal cord, later called “central pattern generators (CPGs)”, capable of generating patterned activity without neither descending inputs from the brain nor sensory inputs from the periphery. These findings triggered the development of isolated spinal cord preparations for studying rhythmic motor outputs in vitro (Kudo and Yamada

1987, Smith and Feldman 1987).

Some of the rhythm-generating CPGs, e.g. respiratory ones, are continuously active but subject to modulation (Johnson, Smith, and Feldman 1996, Di Pasquale et al. 1992, Arata,

Onimaru, and Homma 1998). Conversely, others, e.g. locomotor CPGs, are normally inactive and require descending supraspinal inputs for activation (Grillner 1996, Kiehn

2016). Once turned on, locomotor CPGs are then able to sustain periodic activity without further inputs. However, they still need sensory information to produce the proper output

(discussed below). Locomotor rhythmic activity induced without phasic sensory input from the periphery, like in case of in vitro preparations or in decerebrate animals when the

1 in which the cerebral cortex is surgically removed 2 The dorsal roots were cut (no sensory afferent feedback) 3 The spinal cord was transected to disconnect the lumbar area from the brainstem 3 neuromuscular junction is blocked, is called “fictive locomotion”, see Perret (1983). When studying locomotion in vitro, the activation and modulation of the CPGs are attained by the addition of drugs and/or neuromodulators.

Intracellular recording of spinal motoneurons during locomotion revealed alternating periods of depolarization and hyperpolarization, known as locomotor drive potentials

(Jordan 1983). During depolarizing phases, motoneurons fire repetitively in response to synaptic excitation (Brownstone et al. 1992), and then stops until the next wave of excitation. This phasic repetitive firing activity in neurons is called bursting.

Neuronal bursting in the spinal cord

Neuronal bursting is a behavior in which the neuron alternates between periods of repetitive firing and periods of inactivity (Izhikevich 1999). Bursts serve a lot of functions in the nervous system including increased synaptic transmission reliability, short and long term synaptic plasticity, selective communication between neurons, and strengthening synapses during development (Lisman 1997, Gonzalez-Islas and Wenner 2006, Izhikevich et al.

2003, Izhikevich 1999). In locomotion, motoneuron bursting is required to produce adequate muscle contraction needed for coordinated limb movements, something that single spikes cannot achieve.

Spinal motoneurons are not spontaneously active; they require excitatory inputs to fire.

However, there are specialized burster interneurons (locomotor CPGs) which can be activated in vitro via application of different combinations of neuromodulators (Kiehn

2006). These interneurons drive the phasic firing of multiple motoneuron pools in a rhythmic spatially-organized manner to produce behavior-relevant bursting activity.

Depending on its pattern and segmental location, this rhythmic activity could underlie various motor behaviors such as locomotion, breathing, and chewing (Guertin 2012, Kiehn

2006).

Different bursting patterns can be induced in isolated spinal cord preparations by administering cocktails of neuromodulators (Figure 1). For instance, the application of serotonin, NMDA, and dopamine, produces right-left alternating bursting (Figure 1A), which underlies alternating motor behaviors such as walking and running (Cazalets, Sqalli-

Houssaini, and Clarac 1992, Whelan, Bonnot, and O'Donovan 2000, Kudo and Yamada

1987). Conversely, the application of synaptic inhibition blockers, such as strychnine and bicuculline (glycine and GABAA receptor blockers, respectively), produces synchronized bursting (Figure 1B), which drives synchronized motor behaviors such as hopping and jumping (Bracci, Ballerini, and Nistri 1996b, O'Donovan et al. 2010).

Origin of alternating and synchronized bursting

In order to locate the anatomical rhythmogenic origin of synchronized bursting, lesions were done to the lumbar spinal cord after the bursting was established. The lesions caused changes in different characteristics of the synchronized bursting such as frequency, amplitude and duration (Bracci, Ballerini, and Nistri 1996a). However, the bursting was preserved in any two segments of the cord. In agreement with this, synchronized activity can be evoked in the isolated sacral cord (Jiang et al. 2009). Moreover, similar electrical activity was recorded in cultured slices of embryonic spinal cord (Streit 1993). These earlier studies indicate that the CPGs for synchronized bursting are part of a repeated excitatory network distributed all over the spinal cord.

Figure 1: Different forms of rhythmic bursting induced in vitro in the isolated neonatal rat spinal cord preparation.

A: Alternating bursting activity recorded in vitro from the ventral roots at the third lumbar segment in a P0-P3 rat preparation. Bursting was induced via the administration of 100 µM serotonin. The right and left sides of the cord fire bursts out of phase. Modified from Cazalets, Sqalli-Houssaini, and Clarac (1992). B: Synchronized bursting activity recorded from the ventral roots at the fifth lumbar segment of a P5-P8 rat preparation in a different study. Bursting was induced via the administration of 1 µM strychnine and 20 µM bicuculline. During this form of bursting, both sides of the cord fire synchronously. Modified from Bracci, Ballerini, and Nistri (1996b). Note the difference in shape, duration, and frequency of the two bursting patterns.

On the other hand, the alternating locomotor CPGs are thought to be located mainly in lumbar and thoracic segments, and several segments of the spinal cord are needed to generate this activity in vitro (Cazalets, Borde, and Clarac 1995, Kiehn 2006, Barbeau and

Rossignol 1987, Kiehn and Kjaerulff 1998). However, some spinal network components may be shared between the alternating and synchronized bursting since both are similarly modulated by some neurotransmitters (Bracci, Ballerini, and Nistri 1996b, Beato and Nistri

1999). Additionally, when commissural inhibitory pathways were genetically inhibited, the neurotransmitters used to induce fictive locomotion produced more synchronous activity

(Kiehn 2011, Lanuza et al. 2004).

In both patterns of bursting, an excitatory input from a specific population of interneurons periodically stimulates repetitive firing in motoneurons. The first part of the current study investigated some ionic mechanisms which regulate this activity, with more focus on the role of small conductance calcium-activated potassium (SK) channels.

SK channels: gating and function

The gating of SK channels is controlled by intracellular Ca2+ levels ([Ca2+]i) (Schumacher et al. 2001). The channel subunits interact constitutively with calmodulin, which causes a conformational change and pore opening when local [Ca2+]i is increased (Adelman,

Maylie, and Sah 2012). Potassium (K+) efflux through the channel hyperpolarizes the membrane and regulates the activation of voltage-gated ion channels (Li and Bennett 2007,

Nanou et al. 2013, Adelman, Maylie, and Sah 2012, Viana, Bayliss, and Berger 1993).

Thus, when coupled with voltage-dependent Ca2+-permeable channels, neuronal SK channels can serve as a negative feedback loop that regulates membrane potential and

7 excitability. SK channels are expressed all over the nervous system where they serve different functions (Adelman, Maylie, and Sah 2012, Faber and Sah 2007).

SK channels in motoneurons

In spinal motoneurons, SK channels are expressed in multiple locations and coupled with various Ca2+ sources to perform several functions (Deardorff et al. 2014). On the motoneuron soma, SK channels form large clusters at the cholinergic C-bouton synapses

(Wilson, Rempel, and Brownstone 2004, Deardorff et al. 2013), and are co-localized with

N-type Ca2+channels (Viana, Bayliss, and Berger 1993, Li and Bennett 2007). The N-type

Ca2+ channels are activated during the action potential spike causing a rise in [Ca2+]i and activation of the neighboring SK channels. This generates a long post-spike hyperpolarization period known as the medium afterhyperpolarization (mAHP) (Ransom,

Barker, and Nelson 1975, Meech 1978, Wikstrom and El Manira 1998). While the action potential spike lasts 1-2 milliseconds, the duration of the motoneuron AHP can exceed 100 milliseconds, which makes the AHP a major determinant of motoneuron firing rate (Eccles,

Eccles, and Lundberg 1958).

On the motoneuron dendrites, however, SK channels are activated by persistent L-type

Ca2+channels and by NMDA receptors; their activation causes reduction in the amplitudes of the persistent inward Ca2+ current and excitatory postsynaptic potentials (EPSPs), respectively (Nanou et al. 2013, Li and Bennett 2007, Ngo-Anh et al. 2005, Hounsgaard and Mintz 1988).

Accordingly, their position on the motoneuron dendrites and soma, where the cell’s inputs and output arrive/emerge, allows SK channels to modulate synaptic inputs and firing

8 output, thereby giving them substantial control over the input-output relationship of motoneurons. Since rhythmic bursting is triggered through synaptic inputs from interneurons to motoneurons and involves repetitive firing in both interneurons and motoneurons (two processes regulated by SK channels), SK channels represent a potential candidate for regulation of bursting activity.

SK channels and burst firing

In the brain, SK channels regulate bursting in different areas such as thalamocortical neurons, hippocampal neurons, and raphe nuclei (Kleiman-Weiner et al. 2009, Lappin et al. 2005, Rouchet et al. 2008). In the spinal cord, it has been shown that SK channels only serve as a mechanism for burst termination (el Manira, Tegner, and Grillner 1994, Hill et al. 1992, Grillner 2003). However, multiple indirect experimental observations suggest that

SK channels may play a bigger role in burst regulation in the spinal cord. For example, some neuromodulators used to evoke bursting such as serotonin, dopamine, cholinergic agonists, and bicuculline also inhibit SK channels (Han et al. 2007, Grunnet, Jespersen, and Perrier 2004, Khawaled et al. 1999, Giessel and Sabatini 2010). In addition, the AHP amplitude is reduced during fictive locomotion both in vivo and in vitro (Brownstone et al.

1992, Schmidt 1994), suggesting active inhibition of SK channels during bursting.

Furthermore, cholinergic interneurons, which activate C-boutons fire in phase with ventral root during bursts (Zagoraiou et al. 2009).

Despite their candidacy, SK channels’ potential role in initiating and modulating burst- firing behaviors has not been directly investigated, therefore, the goal of part-I of this study was to test this possibility. Both alternating and synchronized bursting were induced pharmacologically in isolated spinal cord preparations from neonatal as well as fully adult

9 mice. Electrophysiological recordings were made from single motoneurons and from ventral roots to record bursts at the cellular as well as system levels. Our results showed that SK channel inhibition is required for both initiation and amplitude modulation of bursting in motoneurons. The data presented in Part-I of this dissertation have been published in two separate manuscripts (Mahrous and Elbasiouny 2017a, b).

Sensorimotor integration in the spinal cord

The activity of spinal locomotor CPGs has to be first initiated by descending motor commands from the brain (Grillner 2003, Kiehn 2016). Once started, their spontaneous activity can continue for long periods of time without further input. However, in awake behaving animals, the amplitude and speed of locomotor activity has to be continuously modulated and fine-tuned to meet the changing demands of their surrounding environment.

There are two main sources of regulatory inputs for the locomotor CPG activity (Figure 2).

First, sensory inputs from the periphery (Frigon and Rossignol 2006, Rossignol, Dubuc, and Gossard 2006), such as the proprioceptive inputs that gives the perception of position in space. This allows the performance of skilled motor tasks without visual cues, such as typing on a keyboard while looking on the computer screen. However, even seemingly simple motor activities, like walking, might become extremely difficult without guidance from these sensory inputs (McNeill, Quaeghebeur, and Duncan 2008). Second, descending inputs from supraspinal centers that initiate the movement and regulate different aspects like balance and direction also help shape the locomotor output (Kim et al. 2017).

Figure 2: Major excitatory inputs to motoneurons in the spinal cord.

Spinal motoneurons (red) receive three main inputs: 1) Input from the central pattern generators (black) located inside the spinal cord, 2) Sensory input (green) from the periphery such as cutaneous and muscle afferents, and 3) Descending motor inputs (cyan) from the supraspinal centers such as the motor cortex and basal ganglia.

Consequently, generating the appropriate motor output for a specific task relies on the integration of the sensory and descending inputs in the spinal cord. In most neurological disorders which affect the motor systems, sensorimotor integration is impaired and contributes to the symptoms (Brown et al. 2018, Swash 2018, Lewis and Byblow 2002).

Sensorimotor integration in spinal motoneurons

The sensory neurons projecting from the periphery and the descending fibers from supraspinal structures form direct connections with the lower motoneurons (Eccles 1946,

Brown and Fyffe 1981, Witham et al. 2016, Riddle, Edgley, and Baker 2009, Liang et al.

2014). Therefore, the integration of sensory and motor inputs can be studied in spinal motoneurons. Integration of synaptic potentials in motoneurons is influenced by several factors (Magee 2000, Spruston 2008, Heckman and Lee 1999, Heckman, Kuo, and Johnson

2004). One key element is the pattern of synaptic plasticity in each input. With repeated activation at physiological frequencies, almost all types of synapses undergo multiple short-term and long-term plasticity alterations. Some of these processes result in reduced synaptic strength while others cause synaptic enhancement (Zucker and Regehr 2002,

Regehr 2012). Recent studies showed that sensory and motor signals to motoneurons exhibit variable forms of short-term plasticity (Jiang et al. 2015, Barriere et al. 2008).

However, the motor output of the spinal cord during any specific motor task can be maintained steady for long periods of time. Hence, it is quite unclear how these varying sensory and motor inputs could generate a steady motor output.

Furthermore, the importance of integrated sensorimotor signaling upsurges as the motor task becomes more challenging. For instance, walking on a hanging rope demands a higher

12 level of sensorimotor regulation than taking a walk in the park. The nervous system adapts to either scenario by changing the neuromodulatory state (Lee and Heckman 2000).

Neuromodulation of networks, neurons, and synapses

Neuromodulators such as serotonin, dopamine, acetylcholine, and noradrenaline are known to modulate excitability in the nervous system (Avery and Krichmar 2017, Hounsgaard et al. 1988, Lee and Heckman 1999). Spinal locomotor networks receive neuromodulators mainly from descending brainstem fibers (Hounsgaard et al. 1988, Carlsson, Magnusson, and Rosengren 1963, Jordan et al. 2008). These neuromodulators are known to enhance the excitability in the spinal cord at different levels. At the network level, the neuromodulatory effect on excitability depends on the physiological state and initial level of activity in the network (Sharples and Whelan 2017, Marder, O'Leary, and Shruti 2014).

When the network is at a basal state of activity, e.g. in vitro preparations, the application of neuromodulators (e.g. serotonin, dopamine, and/or noradrenaline) increases network excitability and can activate rhythmic bursting activity (discussed above). At the single neuron level, excitability is determined by its voltage-gated currents and the cell input resistance. Neuromodulators can regulate neuronal excitability through altering many aspects of voltage-gated ionic currents such as voltage-dependence, gating, and inactivation (Kaczmarek and Levitan 1986). At the synapse level, neuromodulation of presynaptic terminals and postsynaptic membranes can affect the patterns of synaptic plasticity (Nadim and Bucher 2014). Presynaptically, Ca2+ influx and vesicular release probability are targets for neuromodulators (Higley and Sabatini 2010, Logsdon et al.

2006). Postsynaptic neuromodulation, on the other hand, is usually achieved through controlling ionic conductances on the dendrites that amplify or suppress synaptic currents

(Lee and Heckman 2000, Miles et al. 2007, Sun, Zhao, and Wolf 2005, Heckman, Kuo, and Johnson 2004). The effect of neuromodulators on synaptic plasticity can be as powerful as converting the dynamics from depression to facilitation (Barriere et al. 2008, Zhao et al.

2011, Bevan and Parker 2004).

In part-II of the current study, we investigated the synaptic plasticity, integration, and neuromodulation of sensory peripheral inputs and motor descending inputs in vitro. The synaptic effects under multiple conditions were recorded from single motoneurons as well as the ventral roots to study sensorimotor integration at the cellular and system levels. Our data explains how a stable motor output is generated from varying sensorimotor inputs.

HYPOTHESES, SPECIFIC AIMS, AND CLINICAL

SIGNIFICANCE

PART I: Cellular mechanisms underlying rhythmic motor outputs

Hypothesis: The small conductance Calcium-activated potassium (SK) channels regulate both the initiation and amplitude modulation of burst firing in the mammalian spinal cord.

SPECIFIC AIM 1: To study the role of SK channels in the initiation of motor bursts in the mammalian spinal cord

A. The role of SK channels in the initiation of right-left alternating bursting in the neonatal

spinal cord.

B. The role of SK channels in the initiation of synchronized burst firing in the adult spinal

cord.

SPECIFIC AIM 2: To investigate the role of SK channels in the modulation of motor burst amplitude

A. Regulation of synchronized burst amplitude by graded SK channel inhibition.

B. The possible role of gap junctions in synchronizing motor bursts and/or setting the burst

amplitude.

PART II: Sensorimotor network inputs mediating stable motor output

Hypothesis: Plastic sensory and motor inputs integrate non-linearly to generate a steady motor output.

SPECIFIC AIM 3: To study the plasticity profile and integration of sensory and motor inputs to motoneurons

A. Plasticity patterns of sensory and motor inputs in the isolated sacral cord.

B. Sensorimotor integration at the cellular versus the system levels.

Clinical significance of the research questions

Part-I of this dissertation provides novel insights into the mechanisms of motor burst initiation in the spinal cord. It is of clinical importance to better understand the ionic basis of operation of the locomotor CPGs, especially how bursting activity is initiated and modulated. The ability to manipulate the spinal locomotor CPGs pharmacologically may be a way to improve the quality of movement/life of patients with movement dysfunction such as spinal cord injury and stroke. The combination of pharmacotherapy, targeted microstimulation of the spinal cord, and peripheral stimulation of sensory afferents might restore the ability to walk in paralyzed patients with intact spinal circuits.

Part-II of this dissertation provides more details about the motor responses generated by stimulation of the sensory afferents and remaining descending axons in a transected spinal cord and explains several physiological phenomena such as the ability to produce stable motor output from varying inputs. These data are significant to electrical stimulation studies which use surface or implanted electrodes to induce locomotion in patients with spinal cord injury. The research question is also of direct relevance to the clinical/neuroengineering efforts to create closed-loop somatosensory limb prosthetics.

These prosthetics optimally will be fully-integrated with the nervous system to supply the amputee with naturalistic sensory percepts by electrically stimulating peripheral nerves in 16 the residual limb. Our data provide detailed insights on the response to this electrical stimulation in both individual motoneurons as well as in the entire motor pool.

Furthermore, we show how these sensory inputs interact at the cellular level with descending supraspinal motor commands to create the final output. This knowledge will facilitate the design of better decoders for the neuronal signals recorded from peripheral nerves or re-innervated muscles and used to move the somatosensory prosthetic.

MATERIALS AND METHODS

Animals

Single motoneuron behavior and ventral root activity were recorded in vitro from spinal cord tissue obtained from mice (B6SJL, Jackson laboratory, Bar Harbor, ME) of different age groups. All surgical and experimental procedures were reviewed and approved by the

Wright State University Animal Care and Use Committee.

In vitro preparations

Neonatal lumbosacral cord preparation

For isolation of the spinal cord from young functionally-mature mice (P10-P15), the animal was deeply anesthetized with intraperitoneal injection of Euthasol (1 mg/Kg).

Supplemental amounts of anesthetic were given as needed until the animal did not respond to toe and tail pinching. The mouse was then decapitated and the skin of the back was opened. The spinal column was quickly cut and transferred to a dissection dish full of modified cerebrospinal fluid (mACSF, see below) at room temperature. The spinal cord was isolated by dorsal laminectomy and longitudinal incision of the dura mater. The cord was transected at the upper lumbar segments (around L1), and all roots were cut close to the side of the cord except for the ventral roots at L5 to S2, which were kept for recording.

The lumbosacral cord was then transferred to a recording chamber and continuously perfused with normal artificial cerebrospinal fluid (nACSF, see below) at a rate of 2.5-3 ml/min. The L5-S2 ventral roots were mounted on bipolar wire electrodes. During the first

30-40 min in the chamber, the ventral root response to dorsal root stimulation steadily increased in amplitude. Therefore, the cord was allowed to recover for at least 1 hour before any recordings were started, to stabilize the activity and wash out any remaining anesthetic in the tissue.

Adult sacrocaudal cord preparation

For surgical isolation of the sacrocaudal cord from adult mice (P30-P120), the procedures described in Jiang and Heckman (2006) were followed. Animals were deeply anesthetized using urethane (0.18 g/100 g, injected intraperitoneally). Supplemental amounts of anesthetic were injected, if needed, until the animal did not respond to foot pinching and the respiration was slow and steady. The animal was then placed in a dissection dish, in sternal recumbency, and supplied with carbogen (95% O2/5% CO2) through a face mask.

The skin of the back was opened and the muscles were dissected from the vertebral column.

The sacrocaudal spinal cord (S1-Ca2) was exposed by performing dorsal laminectomy and longitudinal incision of the dura mater. The cord was continuously perfused with mACSF at a rate of nearly 6-7 ml/min. The mouse was then decapitated, and the spinal cord was immediately transected around the L4 lumbar segment. The cord was carefully lifted up and the spinal roots at the caudal cord were cut at their outlets. The isolated sacrocaudal cord, with attached roots, was then moved to a Petri dish full of mACSF and aerated continuously with carbogen (95% O2/5% CO2). In the dish, the ventral and dorsal S1–S3 spinal roots were separated and cleaned of any remaining dura.

The cord was transferred to a recording chamber and pinned with the ventral side upward to facilitate recording of the motoneurons. The chamber was continuously perfused with

19 nACSF at a rate of 2.5-3 ml/min. The ventral and dorsal roots were mounted on bipolar wire electrodes on both the left and right sides of the chamber and immediately covered with petroleum jelly to prevent drying. The tissue was then washed for about 1 hour before any recordings were started.

Physiological solutions

Normal artificial cerebrospinal fluid (nACSF)

The nACSF was composed of the following (in mM): 128 NaCl, 3 KCl, 1.5 MgSO4, 1

NaH2PO4, 2.5 CaCl2, 22 NaHCO3 and 12 glucose. The osmolarity of the solution was ̴ 295 mOsm, and the pH was 7.35 - 7.4 when aerated with carbogen (a mixture of carbon dioxide and oxygen gas in the following proportions: 95% O2 and 5% CO2).

Modified artificial cerebrospinal fluid (mACSF)

The mACSF was composed of the following (in mM): 118 NaCl, 3 KCl, 1.3 MgSO4, 5

MgCl2, 1.4 NaH2PO4, 1.5 CaCl2, 24 NaHCO3 and 25 glucose, and aerated continuously with carbogen. The osmolarity of the solution was ̴ 310 mOsm, and the pH was 7.35 - 7.4.

Electrophysiological recordings

Root recordings

The ventral roots were mounted on bipolar wire electrodes (Figure 3) and connected to a custom-built six-channel amplifier (Kinetic Software, GA) in differential mode with 1000x gain. The recordings were low-pass filtered at 3 kHz and high-pass filtered at 300 Hz.

When needed, the dorsal roots’ electrodes were connected to an isolated stimulator

(Isoflex, AMPI) for electrical stimulation. Brief electrical stimuli (0.2 ms) at different intensities (1xT to 10xT, T is threshold) and different frequencies (0.066 to 50 Hz) were 20 used to activate interneuronal circuits and/or induce synaptic inputs to motoneurons.

Threshold was defined as the lowest current needed to stimulate the dorsal roots to produce a minimal ventral root response, and ranged from 1.5 to 6 µA.

Motoneuron recordings

Intracellular recordings were made from single motoneurons while the whole sacrocaudal cord tissue was maintained intact in vitro. Sharp intracellular electrodes were pulled using a micropipette puller (P97, Sutter instruments, CA). The electrodes were filled with 3M potassium acetate (with or without 100 mM KCl) and beveled to a resistance of 25-40 MΩ using a beveler (BV-10, Sutter instruments, CA). The electrodes were advanced into the ventral horn using a micro-positioner (2660, Kopf instruments, CA). Single cell recordings were sampled at 20 kHz and low-pass filtered at 3 kHz with an Axoclamp-2A (or

Axoclamp-2B) amplifier (Molecular Devices, CA) running in bridge or discontinuous current clamp (DCC) modes. Motoneurons were identified by antidromic stimulation of the ventral root and were accepted for recording when the resting membrane potential was below -60 mV and the antidromic spike was ≥ 60 mV.

The outputs of the amplifiers were digitized using Power 1401-3 data acquisition interface

(CED, UK) at 10-20 kHz. Data were acquired into a computer controlled by Spike2 software (version 8.06, CED) and stored for offline analysis.

Figure 3: Electrophysiology experimental set-up for recording spontaneous rhythmic bursting activity.

In-vitro whole-tissue spinal cord preparation (ventral side up) in which intracellular recording of motoneurons is combined with extracellular recording of ventral root activity on both the left and right side of the same segment of the spinal cord.

Rhythmic motor bursting activity

To induce spontaneous bursting activity in vitro, neurotransmitters and/or pharmacological agents were added to the recording solution. Once a stable pattern of activity is established, modulators of SK channels and/or other ion channels were added to examine their effect.

Alternating rhythmicity was induced by adding serotonin, dopamine, NMDA, and methoxamine to the nACSF solution. Alternating bursting was defined as right-left alternating temporary increase in ventral root activity at a particular segment.

Synchronous bursting activity was initiated by adding strychnine and bicuculline to the recording solution or alternatively a mixture of strychnine, picrotoxin, and one of the SK channel inhibitors. A synchronized burst in the ventral roots was defined as sustained increase in activity (> 10 times the standard deviation of baseline noise calculated using an automated script in Spike2®) in at least two ventral roots at the same segment (one on each side) and starting less than 50 ms from each other. A synchronized increase in bursting amplitude before the ongoing burst completely ended was considered as a sub-burst (a component of the ongoing burst).

Burst amplitude was measured as the peak amplitude of the rectified trace and is proportional to the number of activated motoneurons (Jiang et al. 2015). Burst duration is the time during which the root activity remained above the preset threshold. Burst cycle was defined as the time period from the start of one burst until the start of the next one. In each experiment, burst parameters were calculated as the average from at least 10 bursts, and then a grand mean was calculated for multiple experiments.

Sensory and descending inputs in the sacral cord

The sensory input (S) was induced by electrical stimulation of the dorsal roots. Dorsal roots were connected through bipolar wire electrodes (Figure 4, electrode “A”) to a stimulator

(Isoflex, AMPI), and stimulated with 0.1 ms pulses at either 1.5 or 10 times threshold

(1.5xT or 10xT). The threshold for the sensory input was defined as the smallest amount of current needed to stimulate the dorsal roots to produce a minimal response (compound action potential, coAP) in the ventral roots, and ranged from 1.5 to 6 µA. In some experiments, another bipolar electrode (Figure 4, electrode “C”) was placed on the dorsal root more proximally to record the root potential.

The descending motor input (M) was induced by electrical stimulation of the remaining descending axons (Figure 4, electrode “D”). A custom-built concentric bipolar electrode

(0.125 mm stainless steel contact diameter insulated by Teflon from a stainless steel tube

0.2 ID/0.35 OD) was placed on the ventral surface close to the midline at the L6 ventral root entrance (Jiang et al. 2015). Brief electrical pulses (0.1 ms) were delivered through the concentric electrode at 1.5xT or 10xT. The threshold for descending inputs was defined as the smallest amount of current passed through the concentric electrode that produced a response in the ventral roots, and ranged from 100 to 200 µA.

The response to stimulation was recorded in the ventral roots using bipolar wire electrodes

(Figure 4, electrode B”), and from motoneurons using sharp glass microelectrodes (see details above). To be able to measure the exact amplitude of the excitatory postsynaptic potentials (EPSPs) during intracellular recordings, we had to inhibit action potential firing.

Therefore, a blocker of the voltage-gated Na+ channel, QX-314 (50-100 mM), was used in the internal electrode solution. The amplitude of the EPSPs was measured as the difference between the peak of the EPSP and the resting membrane potential before the first pulse.

The largest ventral root response to motor stimulation was at S1, and it became smaller as it moved caudally. However, the response to dorsal roots’ stimulation was stable only at the S3-Ca2 ventral roots. Therefore, the data collected from S3 to Ca2 were analyzed for both inputs.

The response to each synaptic input was recorded separately, and then both pathways were stimulated simultaneously to study integration. This paradigm was repeated in presence of methoxamine (10 µM) to simulate a higher neuromodulatory state (Lee and Heckman

1999).

Figure 4: Experimental setup for recording synaptic responses to sensory and descending stimulation.

The spinal cord is placed ventral side up in a perfusion chamber. Ventral and dorsal roots are mounted on bipolar wire electrodes above the solution level, and covered with petroleum jelly. The dorsal roots (A) are connected to a stimulator; while the ventral roots (B) are connected to a multichannel extracellular amplifier for recording. Another electrode is placed on the dorsal root (C) more proximally and connected to the extracellular amplifier to record the dorsal root volley. A concentric electrode (D) is placed on the ventral side to stimulate the descending axons. Intracellular recording is performed using sharp electrodes (E) advanced through the ventral surface.

Drugs and chemicals

Several pharmacological agents in different combinations were used in the present study.

A detailed list of all the pharmacological agents used in the current study is provided in

Table 1. All the components of the physiological solutions were purchased from Thermo

Fisher Scientific® (Waltham, MA), and all the drugs were purchased from Sigma®, except strychnine and nimodipine, which were purchased from ACROS®.

Statistical analysis

The non-parametric Mann-Whitney test was used to assess the effects of methoxamine on the parameters of the 5-HT/DA/NMDA-induced bursts. To test the effect of apamin or

STR/PTX on the normalized amplitude of the root reflex, the non-parametric Wilcoxon matched-pairs signed rank test was used. The same test was also used to study the changes in burst characteristics before and after the application of nimodipine, APV, or carbenoxolone. Repeated-measures (RM) two-way ANOVA with Tukey post-hoc analysis was used to study the changes in root reflex amplitudes elicited by high frequency stimulation before and after administration of apamin or STR/PTX. The same test was also used to investigate the type of synaptic integration (Sensory & Motor vs. Sensory + Motor).

Repeated-measures (RM) one-way ANOVA was used to study the effect of different apamin or oxotremorine concentrations on the burst amplitude. The same test was also used to study the pattern of adaptation of a certain synaptic input with a 5-pulse train of stimuli.

The correlations between normalized burst amplitude and the log values of drug concentrations were analyzed using the Pearson correlation coefficient. Analysis of covariance (ANCOVA) was used to compare the slopes of linear regression of burst amplitude and stimulus intensity at different apamin concentrations. p value < 0.05 was

27 considered statistically significant for all tests. The statistical analyses were performed using Graphpad Prism software (Version 7.01; GraphPad Software, La Jolla, CA) or

Statistica (Version 6.1; StatSoft Inc, Tulsa, OK).

Table 1: Drugs and chemicals used in the current study.

Chemical Pharmacological Effect

Apamin SK channel antagonist

APV (DL-2-Amino-5- NMDA receptor antagonist phosphonopentanoic acid)

Antagonist of GABA receptors and SK Bicuculline methiodide A channels

Carbenoxolone Connexin-36 gap junction antagonist

CyPPA (Cyclohexyl-[2-(3,5- dimethyl-pyrazol-1-yl)-6-methyl- SK channels activator pyrimidin-4-yl]-amine)

DNQX (6,7-Dinitroquinoxaline- AMPA receptor antagonist 2,3(1H,4H)-dione)

Dopamine Agonist of dopamine receptors

Non-specific antagonist of SK channel (also a D-tubocurarine blocker of neuromuscular transmission).

EGTA (Ethylene glycol-bis (2- aminoethylether)-N,N,N′,N′- Calcium chelator tetraacetic acid)

Anesthetic combination (pentobarbital and Euthasol phenytoin)

Mephenesin Blocker of polysynaptic transmission

Methoctramine M2 muscarinic receptor antagonist

Methoxamine α1-adrenergic receptor agonist

Nimodipine L-type Ca2+ channel antagonist

NMDA (N-methyl-D-Aspartate) Agonist of NMDA receptors

NS13001 (4-Chlorophenyl)[2-(3,5- dimethylpyrazol-1-yl)-9-methyl- SK channel activator 9H-purin-6-yl]amine)

Oxotremorine sesquifumarate Agonist of M1 and M2 muscarinic receptors

Picrotoxin GABAA receptor antagonist

QX-314 bromide (N-(2,6- Dimethylphenylcarbamoylmethy)tr Voltage-gated Na+ channel antagonist iethylammonium bromide)

Serotonin Agonist of serotonin receptors

SKA-19 (2-amino-6- SK channel activator trifluoromethylthio-benzothiazole)

Strychnine Glycine receptor antagonist

Telenzepine M1 muscarinic receptor antagonist

Urethane Anesthetic

RESULTS

PART-I: CELLULAR MECHANISMS UNDERLYING RHYTHMIC

MOTOR OUTPUTS.

In this part of the study, we used a pharmacological approach to investigate the role of SK channels in the generation of rhythmic activity in the spinal cord. Drug/neurotransmitter mixtures were administered in vitro to induce stable burst firing in isolated spinal cord preparations. The effect of different concentrations/types of SK channel blockers or activators was then examined.

SPECIFIC AIM 1: TO STUDY THE ROLE OF SK CHANNELS IN THE

INITIATION OF MOTOR BURSTS IN THE MAMMALIAN SPINAL CORD.

Specific aim 1A: The role of SK channels in the initiation of right-left alternating bursting in the neonatal spinal cord.

Alternating bursting in the functionally-mature murine cord

Induction of in-vitro locomotive alternating activity in the spinal cord of fully-mature adult animals has been shown to be challenging [see Jiang, Carlin, and Brownstone (1999) for review]. The oldest age where this pattern was successfully induced in-vitro in a mouse spinal cord is postnatal day 10 (P10), soon after the animals are capable of weight bearing and walking (Jiang, Carlin, and Brownstone 1999). In a pilot study, our attempts to induce stable alternating activity in spinal cords of adult animals (P30-P100) in vitro were unsuccessful (n=9, data not shown). Therefore, we decided to use young, yet functionally-

31 mature, mice (P11-P15). At this age, mice have enough developmental locomotor maturity manifested as being able to walk and weight-bear. The firing activity of the ventral roots was recorded at the lower lumbar (L5-L6) and upper sacral segments (S1-S2). For consistency, the activity of the right and left ventral roots at L6 and S2 roots will be shown in the figures.

In normal artificial cerebrospinal fluid (nACSF), the ventral roots are usually quiet except for random, low-amplitude spiking activity of no consistent pattern. As reported before

(Jiang, Carlin, and Brownstone 1999), the addition of a combination of serotonin (5-HT), dopamine (DA), and N-methyl-D-aspartate (NMDA) produces right-left alternating activity (Figure 5A). However, our recordings at that age showed that locomotor-like activity was mostly restricted to one segment, was usually unstable, and had low amplitude

(0.15±0.08 mV, n=4) and long burst cycles (24.2±6 sec, n=4). In addition, the same combination of neuromodulators sometimes failed to produce alternating activity, especially as mice became older (Figure 5B).

Figure 5: Neurotransmitter-induced alternating bursting in the spinal cord of functionally-mature mice.

A: The activity of the ventral roots at the lower lumbar (L6) and mid-sacral (S2) segments on the left (L) and right (R) sides of a spinal cord obtained from a mouse at P11. The application of serotonin (5-HT), dopamine (DA), and NMDA induces low- amplitude, left-right alternating, rhythmic bursting at the sacral, but not the lumbar, segments. B: Recordings from a P12 mouse showing the effect of the same concentrations of serotonin, dopamine, and NMDA. Only the left side ventral root at L6 is showing low amplitude bursts (arrows).

Induction of stable alternating bursting in the functionally-mature murine cord

It has been shown that methoxamine, an α1-adrenergic receptor agonist, induces alternating activity in the isolated sacral cord of neonatal rats (Gabbay and Lev-Tov 2004). We thus tested whether methoxamine can stabilize the 5-HT/DA/NMDA-induced rhythm in our preparation. The addition of 100 µM methoxamine to the locomotion cocktail drastically changed the overall activity (Figure 6A, n=6). At a relatively large time scale, the combination seemed to produce rhythmic synchronous activity across multiple segments of the cord (Figure 6A, left and middle panels). The cycle time of this synchronous activity was 33.7±18.37 sec (n=6) and the duration of each episode was 7.68±3.05 sec (n=6).

However, on a smaller time scale, this synchronized activity is composed of short bursts that alternate on the left and right sides of the cord (Figure 6A, right panel), with an almost eighty-times-shorter burst cycle (0.31±0.02 sec, n=6) and a burst duration of 0.16±0.02 sec

(n=6). The cycle and duration of the fast alternating activity was consistent across different experiments (compare the right panel in Figure 6A to Figure 6B). Figure 6C shows the changes in burst parameters after methoxamine administration expressed as percentage of control (i.e., values before methoxamine administration).

Figure 6: Effect of methoxamine on alternating activity in the lumbosacral cord.

A: Left: The addition of the α1-adrenergic agonist (methoxamine) to the 5- HT/DA/NMDA cocktail, in the same experiment as Figure 5B, produces a seemingly synchronized high-amplitude activity in all ventral roots. Middle: Each one of these episodes of synchronous activity is a cluster of short bursts. Right: This clustered activity is composed of short alternating bursts between the left and right sides; with the lumbar bursts in phase with sacral bursts on the same side of the cord (notice the different time scales). B: An episode of fast, alternating activity produced by the same chemical cocktail in a spinal cord of a P15 mouse. Note that burst duration and cycle are similar to the experiment in (A). C: The burst parameters after methoxamine administration, plotted as a percentage of the value before adding the drug. The bars represent the mean ± SD. * Statistically significant, p value < 0.05.

Effect of SK modulators on alternating bursting activity

Serotonin, dopamine, and methoxamine have been reported to inhibit SK channels (Han et al. 2007, Grunnet, Jespersen, and Perrier 2004, Wagner, Ronnekleiv, and Kelly 2001). To test whether SK channels regulate fictive locomotor activity in the spinal cord, SK modulators were administered to the recording solution after stable bursting had been established. Figure 7 shows episodic alternating bursting activity in the lumbosacral cord induced by 5-HT, DA, NMDA, and methoxamine. The application of CyPPA, an SK channel activator (Hougaard et al. 2007) at 20 µM reduced the burst amplitude. When the concentration of CyPPA was increased to 40 µM, the bursting was completely inhibited, indicating that SK inhibition is needed to induce bursting (n=3). Similar results were obtained using another SK channel activator SKA-19,(Coleman et al. 2015) (n=3). To further confirm, a specific SK channel inhibitor, apamin, was then added to the recording solution, which restored the rhythmic activity (Figure 7, n=6). The locomotor bursts generated after adding apamin had longer duration (compare the two insets in Figure 7), consistent with the role of SK channels in terminating the burst (el Manira, Tegner, and

Grillner 1994, Grillner 2003, Hill et al. 1992).

Figure 7: Locomotor bursts are inhibited by SK activators and restored by SK blockers.

The application of the SK activator (CyPPA) inhibits the episodic alternating bursting activity in a dose-dependent manner. After the bursting has completely ceased, the addition of the SK blocker, apamin, restores the rhythmic alternating activity. The insets show, on a different time scale, representative alternating bursts from two different episodes of activity.

Specific aim 1B: The role of SK channels in the initiation of synchronized burst firing in the adult spinal cord.

The SK channel expression levels continue to change over the course of motoneuron maturation in mice until the adult stage (unpublished observation from our lab). To investigate if SK channels continue to control bursting activity in the fully-mature spinal cord, we aimed at inducing bursting activity in an adult cord preparation. Unlike, the locomotion-like alternating bursting, the synchronous pattern of bursting (in which the right and left sides at the same segment fire bursts in phase) can be induced in vitro in an adult spinal cord (Jiang et al. 2009). While alternating activity is induced by adding excitatory neurotransmitters, synchrony is evoked by blocking inhibitory synaptic transmission. The two rhythmic patterns may share some neuronal circuitry, but they are thought to originate from different central generators (Beato and Nistri 1999). In this set of experiments, we will record the firing activity of the ventral roots at the sacral segments

(S1-S3) from fully-adult mice (P80-P120). This region of the adult spinal cord can be reliably maintained in vitro for several hours from both rats (Bennett, Li, and Siu 2001) and mice (Jiang and Heckman 2006). For consistency, the activity of the right and left ventral roots at the third sacral segment (R S3 and L S3, respectively) will be shown in the figures.

Facilitation of synchronized burst initiation by SK channel blockers in the adult mouse spinal cord

Similar to the neonatal cord preparation, ventral roots are normally quiet except for random low-amplitude spiking activity (Figure 8A, ACSF phase) with no rhythmic activity.

Synchronized bursts were induced by applying 0.25 µM of strychnine (STR), a glycine

38 receptor blocker, and 1 µM of bicuculline (BIC), a blocker of both GABAA receptors and

SK channels (Khawaled et al. 1999), which triggered high amplitude low-frequency synchronized bursts in all ventral roots. The bursts started and ended simultaneously on both sides with a few milliseconds differences (compare Figure 8A to Figure 6).

Intracellular recordings of single motoneurons confirmed that the spontaneous synchronized root activity is produced by action potentials in motoneurons (Figure 8B, n=13). The cells fired at high frequencies at the beginning of each burst (typically 120-200

Hz), followed by adaptation to a lower frequency range (50-100 Hz), and eventually fired a few spikes at less than 10 Hz before it stops around the end of the burst (Figure 8A).

Some cells stopped firing before the burst ended in the ventral root. However, the membrane potential would typically exhibit a plateau (i.e. remain depolarized) until the burst ends (see Figure 18B)

Given that BIC blocks GABAA receptors and SK channels (Debarbieux, Brunton, and

Charpak 1998, Khawaled et al. 1999), we replaced it with picrotoxin (PTX, 5 µM), a

GABAA receptor blocker that does not block SK channels (Pflieger, Clarac, and Vinay

2002). There was no change in ventral root spontaneous activity in presence of STR and

PTX (Figure 9A). However, when apamin (100 nM, a specific SK channel blocker) was applied to the ACSF solution with STR and PTX, synchronized bursting was initiated

(Figure 9A, n=14). To further confirm that direct SK channel blockade is responsible for evoking burst firing in the disinhibited cord, we used another, yet non-specific, SK channel blocker, d-tubocurarine (d-TC, 100 µM) (Dun, Jiang, and Mo 1986, Ishii, Maylie, and

Adelman 1997), which successfully evoked synchronized bursting when administered with

STR/PTX (Figure 9B, n=4).

Figure 8: Synchronized bursts induced in the adult spinal cord.

A: The application of strychnine and bicuculline induces spontaneous bursts in the spinal cord. The bursts are synchronized on both sides of the cord as shown at the S3 segment at different time scales. B: Recording of a motoneuron in current clamp mode showing spikes and firing rates (lower panel) during spontaneous bursting. Upper panel shows the simultaneous extracellular recording of the segmentally-aligned ventral root.

Figure 9: SK channel inhibition facilitates synchronized bursting initiation.

A: Strychnine and picrotoxin did not induce bursting until apamin was added. B: A different SK channel blocker (d-tubocurarine) also initiated synchronized bursting in the disinhibited spinal cord.

Reduction of extracellular Ca2+ facilitates synchronized bursting

SK channels are gated by Ca2+; where increases in free intracellular Ca2+ causes a conformational change in the channel protein which results in pore formation and K+ flux through the membrane (Adelman, Maylie, and Sah 2012). Depletion of extracellular Ca2+

2+ ([Ca ]e) eventually reduces the level of SK channel activation (Viana, Bayliss, and Berger

1993). We investigated whether using a different recording solution in which 90% of

2+ 2+ [Ca ]e is replaced with barium ions (Ba ) can mimic the effect of apamin.

Ba2+ can pass through Ca2+ channels but it does not bind or activate SK channels (Enomoto et al. 1991). The application of the Ba2+ ACSF containing STR and PTX results in tonic activity, in addition to the rhythmic synchronized bursts (Figure 10A, n=4/5). The tonic activity probably originated from motoneurons themselves due to enhanced excitability by multiple effects of Ba2+ (Standen and Stanfield 1978, Hille 2001, Lu et al. 2010). The synchronized bursts, but not the tonic activity, were eliminated by DNQX, an AMPA receptor (AMPAR) blocker (data not shown), indicating that only the synchronized bursts were synaptically-triggered.

To further confirm, extracellular Ca2+ concentration was lowered via adding the Ca2+ chelator EGTA to the recording solution. In presence of STR/PTX, adding EGTA (1.5 to

2 mM) evoked synchronized bursts (Figure 10B, n=3). When EGTA was increased to 3 mM, spontaneous bursting was inhibited and electrical stimulation of the dorsal roots failed to elicit a response in the ventral roots (data not shown), indicating that synaptic transmission is inhibited at this concentration. These data show that lowering SK channel activation, either by chelating Ca2+ or replacing it with Ba2+, evoked burst firing in the

disinhibited cord.

Figure 10: Effect of reducing extracellular free Ca2+ on burst initiation.

A: Substituting extracellular Ca2+ with Ba2+ evokes tonic activity on which synchronized bursts are superimposed (arrows). B: Reduction of extracellular free Ca2+ through the application of the Ca2+ chelator EGTA similarly evokes bursting.

Cholinergic inhibition of SK channels also facilitates burst firing

To examine whether inhibiting SK channels via a physiological pathway, specifically muscarinic receptors (Miles et al. 2007, Giessel and Sabatini 2010, Adelman, Maylie, and

Sah 2012), would produce synchronized bursting, we used muscarinic agonists. When 5µM oxotremorine sesquifumarate salt (an M1/M2-receptor agonist with preferential activity on

M2) was administered, synchronized bursts emerged in all ventral roots (Figure 11A, n=13). To test that burst firing is mediated through SK channels, we examined the effect of CyPPA, a specific SK channel activator (Hougaard et al. 2007), on oxotremorine- induced bursting. CyPPA (40 µM) inhibited the bursting in full (Figure 11A, n=6). Similar results were obtained using other SK channel activators including SKA-19 and NS13001

(data not shown).

To investigate the type of muscarinic receptor which mediates the effect of oxotremorine, we used specific blockers of M1 and M2 receptors. The effects of oxotremorine were not blocked by the pre-application of methoctramine (20 µM), an M2-receptor antagonist

(Figure 11B, n=5) or telenzepine (10 µM), an M1-receptor antagonist (Figure 11C, n=4).

However, the pre-application of both methoctramine and telenzepine inhibited oxotremorine-induced bursting (Figure 11D, n=3). The direct blockade of SK channels using apamin, in presence of the muscarinic antagonists, successfully induced bursting

(Figure 11D, n=2).

Taken collectively, the data show - via multiple approaches - that SK channels’ inhibition facilitates motor burst initiation in the spinal cord.

Figure 11: Muscarinic inhibition of SK channels facilitates bursting.

A: The muscarinic agonist, oxotremorine, induces bursting in presence of strychnine and picrotoxin; this bursting can be inhibited by the SK channel activator, CYPPA. The pre- application of either the selective M2 receptor antagonist, methoctramine (B), or the selective M1 receptor antagonist, telenzepine (C), does not inhibit oxotremorine-induced bursting. D: The combined M1 and M2 receptors blockade completely inhibits oxotremorine effect; however, the application of the direct SK channel blocker, apamin, induces synchronized bursts in presence of the muscarinic antagonists.

A specific role for SK channels versus non-specifically altering the excitation- inhibition balance.

To test if excitability modulation, not necessarily through SK channels, could also evoke or stop the synchronized bursts, we used two different approaches to modulate the excitability of motoneurons and spinal networks in ways not involving SK channels. First, in presence of STR/PTX only, excitation was increased by electrical stimulation of the dorsal roots to test if this can mimic the effect of SK inhibitors. Second, bursting was induced by STR/PTX/apamin and then excitability was reduced by blocking the Ca2+ persistent inward current (PIC) to test if this can mimic the effect of SK activators.

Increased excitation through electrical stimulation does not mimic SK inhibitors:

In presence of STR/PTX and without any SK modulator, we increased the network excitability by delivering brief electrical stimuli to the dorsal roots at different intensities

(1xT-10xT), and recorded the ventral roots’ response (Figure 12A). With synaptic inhibition blocked, this electrical stimulation activates strong excitatory signals across the network and eventually excitatory inputs to motoneurons. Importantly, this strong graded excitation did not evoke bursting in the ventral roots; only single compound action potentials were recorded in response to each stimulus (Figure 12A, left inset and Figure

12B, arrow heads in the top panel, n=6). However, when apamin (100 nM) was added in presence of STR and PTX, the same electrical stimulation paradigm evoked bursts at all stimulation intensities, when delivered in between spontaneous bursts (Figure 12A, right inset and Figure 12B, circles in the lower panel). These data show that increasing the excitability using electrical stimulation in presence of STR and PTX was not sufficient to evoke bursting until SK channels were inhibited.

Figure 12: In absence of SK channel inhibitors, electrical stimulation does not evoke bursting in the disinhibited spinal cord.

A: Recording of ventral root response to electrical stimulation of the ipsilateral dorsal roots in presence of STR and PTX only, or in presence of STR, PTX, and apamin (100 nM); at 1x, 1.5x, 2x, 5x, and 10xT. Insets show, on a smaller time scale, the ventral root recording at 2xT. Bursting was evoked when apamin was applied (right inset) but not in presence of STR/PTX only (left inset). B: Top: average ventral root response to dorsal root stimulation at each intensity in presence of STR and PTX; only compound action potentials were elicited (arrow heads). Bottom: in presence of STR/PTX + Apamin, the response has 2 components, a compound action potential (arrow heads) followed by a burst (circles).

Decreased excitability through Ca2+ PIC blockers does not mimic SK activators:

We tested the effect of reduced motoneuron and interneuron excitability by blocking the

L-type Ca2+ channels which mediate the persistent inward Ca2+ current (Ca2+ PIC). If SK channels regulate bursting by non-specifically affecting excitability, then nimodipine, a

Ca2+ PIC blocker, would be able to mimic the effect of SK activators and eliminate the bursting. In Figure 13A, synchronized bursting was first evoked by using apamin, STR, and PTX. When nimodipine (20 μM) was then applied, neither the burst amplitude nor the frequency was changed (Figure 13B, p=0.26 and 0.37 respectively, n=7). Notably, this nimodipine concentration has been shown to block the Ca2+ PIC in full (Li and Bennett

2003). Interestingly, the burst duration was decreased (Figure 13B, p=0.015), indicating that the Ca2+ PIC acts to prolong the burst duration but does not contribute to its initiation or amplitude.

A counter argument is that acutely spinalized preparations may have small PICs due to the loss of descending serotoninergic inputs (Hounsgaard et al. 1988, Li and Bennett 2003).

This could alternatively explain the weak effect of nimodipine on the bursting in Figs. 6A-

B. To address that, we used serotonin, in presence of STR and PTX, in order to enhance the Ca2+ PIC as well as induce synchronized bursts. It is important to mention that serotonin has been shown to fully activate the Ca2+ PIC in acutely spinalized preparations (Li et al.

2007) and also to inhibit SK channels (Berger, Bayliss, and Viana 1992, Grunnet,

Jespersen, and Perrier 2004, White and Fung 1989). Intracellular recordings from motoneurons in presence of serotonin, STR, and PTX indicated that PICs are activated.

With ramp current injection, cells stopped firing at a lower current value than the one at which firing started (∆I= 1.2 ± 0.11 nA) (Figure 14A).

Figure 14B shows synchronized bursting induced by serotonin in presence of STR and

PTX. The application of increasing concentrations of nimodipine (15 µM and 30 µM) did not affect the bursting characteristics (n=4). However, after nimodipine had been completely washed out, administration of the SK channel activator CyPPA (at 20 µM and

40 µM) gradually reduced the burst amplitude and eventually eliminated the bursting

(Figure 14B). These data indicate that: 1) nimodipine does not affect the synchronized burst properties in preparations with high PICs, 2) The SK activator, CyPPA, reduces the amplitude and eventually stops the bursting; indicating that SK channel activation has stronger effects on burst firing than Ca2+ PIC inhibition.

Taken collectively, these data demonstrate that SK channels’ inhibition specifically mediate bursting initiation, and cannot be mimicked by altering the excitation-inhibition balance.

Figure 13: Blocking Ca2+PIC does not affect burst initiation or amplitude.

A: Application of the Ca2+ PIC blocker nimodipine does not change the synchronized bursting. B: A summary of the effect of nimodipine on the burst characteristics. Data plotted as the grand mean obtained from multiple experiments, and error bars represent SD.

Figure 14: SK channel inhibition is specifically necessary for burst initiation.

A: Intracellular recording from an S3 motoneuron after the application of serotonin (20 µM) in presence of STR and PTX. In response to ramp current injection (lower panel), the cell fires repetitively and continues to fire at lower current values on the descending ramp (middle panel); indicating that persistent inward currents are activated. The top panel shows the instantaneous firing frequency. B: Synchronized bursts induced by serotonin in presence of STR and PTX are not changed by the application of nimodipine; while CyPPA eliminates the bursting.

SPECIFIC AIM 2: TO INVESTIGATE THE ROLE OF SK CHANNELS IN

MOTOR BURST AMPLITUDE MODULATION.

Physiologically, the motor output has to be graded to generate smooth movements. The amplitude of the ventral root burst is determined by the number of motoneurons firing synchronously at the beginning of the burst; and is proportional to the force generated by the innervated muscle. In the experiments shown in Figure 14, the SK activators seemed to produce a dose-dependent decrease in the burst amplitude. Thus, we further investigated whether the effects of SK inhibitors on synchronized bursting are “all or none” or the burst amplitude could be modulated proportionally to the level of SK channel inhibition.

Specific aim 2A: Regulation of synchronized burst amplitude by graded SK channel inhibition.

Two approaches were used to study the role of SK channels in grading the motor output.

First, SK channel inhibitors were titrated in increasing concentrations and the mean peak amplitude of spontaneous bursts at each concentration was measured and compared.

Second, we tested whether graded electrical stimulation can grade the burst amplitude at fixed SK blocker concentrations.

SK channel direct blockade changes the burst amplitude in a dose-dependent manner

Apamin was titrated in 5 different concentrations (5, 10, 20, 50, and 100 nM) (Figure 15A, n=9) and the burst amplitude at each concentration was measured and compared. The amplitude of the rectified trace was normalized to the maximum burst amplitude (the amplitude at 100 nM apamin) for any particular root (Figure 15B). As apamin concentration increased, the amplitude of generated bursts on all ventral roots were graded

52 linearly, then saturated (Figure 15A-B). Strong correlation was found between the burst amplitude and the logarithm of apamin concentration (Pearson correlation: r2=0.9, p=0.006). RM one-way ANOVA statistical analysis also revealed that the burst amplitude changed with different apamin concentrations (p<0.0001). Post-hoc Tukey’s multiple comparisons tests showed that the burst amplitudes were significantly different at each apamin concentration from the preceding one; until the response saturated at 50 and 100 nM (Figure 15B).

Burst amplitude regulation by physiological inhibition of SK channels

To investigate whether physiological inhibition of the channel through the muscarinic receptors can achieve similar gradation in burst amplitude, we titrated oxotremorine at increasing concentrations (from 1 to 5 μM, Figure 16A). The burst amplitude increased progressively (Figure 16B); and was found to strongly correlate with the logarithm of the drug concentration (Pearson correlation: r2=0.98, p=0.003, n=8). Using RM one-way

ANOVA test, the burst amplitudes were found statistically different at various oxotremorine concentrations (p<0.0001). The post-hoc Tukey’s multiple comparisons tests showed that burst amplitudes were different until the response saturated at 4 and 5 µM

(Figure 16B).

Figure 15: Graded direct blockade of SK channels modulated the burst amplitude.

A: Bursts in the ventral roots increase in amplitude with increasing apamin concentrations in a dose-dependent fashion. B: The burst peak amplitude is correlated with apamin concentration. * indicates significant difference from the preceding data point. The bursts are normalized to the maximum burst amplitude obtained at 100 nM apamin. Each data point on the graphs is the grand mean obtained from multiple experiments; and error bars represent SD.

Figure 16: Physiological inhibition of SK channels grades the burst amplitude.

A: Oxotremorine induces similar graded response, which is inhibited by CYPPA in a dose-dependent manner. B: Similar to apamin, oxotremorine concentration determines the burst amplitude. * indicates significant difference from the preceding data point. The bursts are normalized to the maximum burst amplitude obtained at 5 µM oxotremorine. Each data point on the graphs is the grand mean obtained from multiple experiments; and error bars represent SD.

Another explanation for the burst amplitude gradation observed with apamin and oxotremorine could be the increased excitability of interneurons resulting from these drugs, leading to increased excitatory input to motoneurons, thereby grading the burst amplitude.

That is, burst amplitude modulation is achieved through synaptic input gradation, not necessarily via SK channel inhibition. To investigate this potential mechanism, we performed experiments where the magnitude of the synaptic input was graded using brief electrical stimulation of the dorsal roots at different intensities (from 1xT to 10xT) in presence of high apamin concentration (100 nM) + STR/PTX – in order to ensure that SK channels are fully blocked. The ventral root response at different intensities consisted of two components, an early monosynaptic component (Figure 17A, arrow head) whose amplitude was graded with the stimulus intensity, and a later component that was the burst

(dotted circles in Fig 12A, the inset shows the 1xT response at a different scale). When the normalized burst amplitude was compared at different stimulation intensities, there was no gradation (i.e., the burst amplitude remained 100% at all intensities, Figure 17B, n=6), indicating that graded dorsal root stimulation did not grade the burst amplitude when SK channels are fully blocked. However, when a lower concentration of apamin was used (5 nM, n=6) to evoke bursting without full SK channel blockade, the relationship between the stimulus intensity and the burst amplitude was significantly steeper (p=0.046), indicating that burst amplitude gradation requires the availability of SK channels.

Figure 17: Burst amplitude gradation requires availability of SK channels.

A: Ventral root responses to electrical stimulation of the ipsilateral dorsal roots at 1x, 1.5x, 2x, 5x, and 10xT in presence of STR/PTX and apamin. The response has 2 components, a compound action potential (arrow head in the 1xT response) followed by a burst (circles). The inset shows the 1xT response at a less compressed time scale. B: At high apamin concentration (100 nM), the burst amplitude does not change with the stimulus intensity; however, the relationship is steeper at lower apamin concentrations (5 nM). The bursts are normalized to the maximum burst amplitude obtained at 10xT. Each data point on the graphs is the grand mean obtained from multiple experiments; and error bars represent SD.

Specific aim 2B: The role of gap junctions in synchronizing motor bursts and/or setting the burst amplitude.

A characteristic feature of the bursting behavior in this study is the apparent synchronization of activity at different levels. First, synchrony occurs across multiple ipsilateral segments of the spinal cord (Figure 18A): Even the burst components were synchronized across different segments (Figure 18A, vertical arrows). On a given side of the cord, the bursts usually started at the S3 segment then travelled towards S1 with a delay of about 10 ms between each two adjacent segments (Figure 18A, inset). Second, synchrony occurred between contralateral sides of the cord at the same segment level, in which the bursts spread to the contralateral side of the cord with a delay of about 5 ms

(Figure 18B, upper panel and inset and Figure 19B). Third, synchrony occurred among motoneurons in each motor pool. Our intracellular recordings from motoneurons showed a one-to-one correspondence between the individual motoneuron firing pattern and the discharge activity of the segmentally-aligned ventral root during spontaneous bursting

(Figure 18B and Figure 8B). When spike-triggered averaging was used to average the activity of the aligned ventral root triggered by individual spikes in motoneurons during bursting, a peak was always seen in the average (Figure 18C). In one experiment, two different cells triggered peaks in the averaged activity of the same root. This indicates that, during synchronized bursting, motoneurons in each motor pool fire synchronously and exhibit similar firing rate profiles. One potential mechanism for this synchrony across segments and within the motor pool is the presence of gap junctions between interneurons and/or motoneurons, respectively. If this is true, gap junction expression and activity would be a major regulator of burst amplitude.

Figure 18: Different levels of synchrony during spontaneous synchronized bursting.

A: At different segments on the same side of the cord, the bursts and their subcomponents are time-locked (arrows) with about 10 ms delays (see expanded time scale, inset). B: At the same segment, bursts and subcomponents are synchronized with a shorter delay (about 5 ms, see inset). Lower panel shows intracellular recording from an S2 motoneuron on the right side of the segment with the firing behavior matching the root discharge on both sides of the cord. C: Spike-triggered average of the ventral roots’ activity in B triggered by spikes in the motoneuron in the lower panel. The spikes in the motoneuron trigger a peak in the averaged root activity on the ipsilateral side.

Gap junction coupling does not contribute to the burst amplitude

To test whether gap junctions affect the amplitude or synchrony, we induced bursting using

STR/PTX/Apamin and then applied carbenoxolone (Figure 19A), which binds and blocks

CX36 gap junctions (Rozental, Srinivas, and Spray 2001, Salameh and Dhein 2005). The burst amplitude was not changed after carbenoxolone (Figure 19B, n=8, p= 0.11). This indicates that synchrony within the motor pool is not mediated by gap junctions. Similarly, the contralateral delay did not change after carbenoxolone application (Figure 19B), which suggests a lack of effect on interneurons. The burst cycle was not changed (p=0.55) but the burst duration was shortened (p=0.04). The reduction in burst duration was probably due to non-specific effects of carbenoxolone on presynaptic release probability and Ca2+ influx

(Tovar, Maher, and Westbrook 2009, Vessey et al. 2004).

These results indicate that the burst amplitude as well as the synchrony among the motor pool and across contralateral sides are not mediated by gap junctions between motoneurons or interneurons.

Amplification of excitatory inputs by STR/PTX vs. apamin

Since the application of apamin alone without STR and PTX does not evoke bursting, we examined the effect of blocking SK channels vs. blocking synaptic inhibition on excitatory synaptic transmission. Both SK inhibitors and synaptic inhibition blockers have been shown to enhance glutamatergic transmission in the nervous system (Ngo-Anh et al. 2005,

Faber, Delaney, and Sah 2005, Streit 1993). However the relative effect of these drugs on the excitatory drive to the motor pool has not been investigated.

Figure 19: Gap junctions do not contribute to burst amplitude or synchrony.

A: The gap junction blocker, carbenoxolone, does not change the bursting synchrony. B: A summary of the effects of carbenoxolone on the burst characteristics. Data plotted as the grand mean obtained from multiple experiments; and error bars represent SD.

First, we investigated the synaptic conductances involved in generating these bursts (Figure

20 and Figure 21). Second, we tested how these conductances are affect by either apamin or STR/PTX (Figure 22 and Figure 23).

Excitatory synaptic conductances generating synchronous activity:

Synchronized bursts were evoked in all roots by the application of STR, PTX, and apamin.

Then, we added selective blockers to investigate the involvement of different excitatory receptors. First, we applied APV (100 µM), a selective NMDA receptor (NMDAR) blocker. APV silenced the bursting activity for 2 to 5 minutes which was then spontaneously restored (Figure 20A, n=6) with similar amplitude (p=0.93) and shorter duration (p=0.031). These data suggest that NMDARs do not affect the number of recruited motoneurons during the burst but may prolong the burst duration. The bursting also had lower frequency in presence of APV (Figure 20B, p= 0.03), suggesting an effect on the

CPGs. These data indicate that synchronized bursting is not mediated exclusively by

NMDARs.

In contrast, when DNQX, an AMPAR blocker, spontaneous bursting was eliminated

(Figure 21, n=8). However, electrical stimulation of the dorsal roots (at 10xT) was still capable of eliciting bursts in the ventral roots (Figure 21, left inset), an effect probably mediated by the NMDARs. The subsequent application of NMDA (30 µM) restored the bursting (n=3). When APV was now added, with DNQX maintained in the solution, the bursting was completely eliminated and dorsal root stimulation (at 10x) was unable to evoke bursting anymore (Figure 21, right inset). The data in Figure 20 and Figure 21 indicate that synchronized bursting is driven by both NMDA and AMPA receptors.

Figure 20: Synchronized bursts are not exclusively-driven by NMDA receptors.

A: Blocking NMDA receptors with APV does not eliminate the spontaneous bursting, but decreases its frequency. B: A summary of the effect of APV on the burst characteristics. APV slows down the bursting and shortens the burst duration but does not affect the burst amplitude. Data plotted as the mean ± SD.

Figure 21: Both AMPA and NMDA receptors generate the synchronized bursts.

DNQX, the AMPA receptor blocker, completely stops the bursting; however electrical stimuli to dorsal roots (arrows, left inset) can still trigger bursting. When NMDA is added, spontaneous bursting can be resumed while AMPA receptors are still blocked; furthermore, this effect was inhibited by adding APV. Electrical stimulation of the dorsal roots no longer evokes bursting when both receptors are blocked (right inset).

Differential amplification of glutamatergic drive by SK inhibition versus blocking inhibitory receptors:

We examined the relative effect of blocking SK channels (using 100 nM apamin) vs. blocking synaptic inhibition (using 0.25 µM STR/5 µM PTX) on glutamatergic inputs in the spinal circuits. Dorsal root-evoked ventral root response is known to be driven by glutamatergic synapses, specifically AMPARs (King, Lopez-Garcia, and Cumberbatch

1992). Therefore, we used the amplitude of compound action potentials (coAPs) recorded in the ventral roots as an estimate of the net glutamatergic synaptic drive in the circuit.

Dorsal root stimulation was quantified in terms of threshold (T, the current used to stimulate the dorsal roots to produce a minimal ventral root response).

Two types of dorsal root stimulation paradigms were tested: 1) single stimulation pulses with intensity of 1x threshold (1xT), to assess the drug effect on the monosynaptic responses of the circuit (Figure 22), and 2) five stimulation pulses at 50Hz with intensity of 10xT, to assess the repeated monosynaptic and polysynaptic responses (Figure 23). First, when dorsal roots were stimulated by single 1xT pulses in presence of apamin, the coAP amplitude showed an increase relative to control (Figure 22A, n=11, p<0.01). However,

STR/PTX had no effect on the coAP amplitude at 1xT (Figure 22B, n=11, p=0.38). Second, repeated (5 pulses, at 50Hz) dorsal roots stimulation at 10xT, under control conditions (i.e., no drugs applied) showed a gradual decline in amplitude (Figure 23, black bars), in agreement with short-term synaptic depression (STD) characterized in this preparation

(Jiang et al. 2015).

The application of apamin did not change the coAP amplitudes at any pulse as indicated by repeated-measures (RM) two-way ANOVA and Tukey post-hoc test (Figure 23A, n=7,

65 p=0.16). On the other hand, blocking synaptic inhibition with STR/PTX markedly reduced

STD of coAP amplitudes with repeated stimulation (Figure 23B, n=8, p<0.001). Tukey post-hoc analysis showed that the coAP amplitudes at the 3rd, 4th, and 5th pulses (P3, P4, and P5) were significantly larger in presence of STR/PTX relative to control with an amplification factor ranging between 3x and 4.5x (Figure 23B); indicating that blocking inhibition in the circuit amplifies the high frequency polysynaptic components.

To investigate if the synchronized bursts have a polysynaptic circuit component, we tested the effect of mephenesin on the bursts. Mephenesin is a blocker of NMDA receptors that preferentially interferes with polysynaptic transmission (Lev-Tov and Pinco 1992, Kaada

1950, Jiang et al. 2015). Mephenesin (1-1.5 mM, n=7) eliminated the synchronized bursts, demonstrating a significant polysynaptic component in the bursts (Figure 24).

Together, these data demonstrate that: 1) Inhibition of SK channels amplifies the monosynaptic excitatory drive to the spinal circuit; 2) Blocking inhibitory synapses amplifies the polysynaptic excitatory drive to the spinal circuit during high-frequency repeated activation, which is crucial during bursts (Figure 8B and Figure 18B); and 3) the combined blocking of SK channels and inhibitory synapses, as opposed to individual blocking, amplifies the excitatory drive substantially to trigger burst firing. Therefore, these results demonstrate the relative roles for SK channels and inhibitory synapses in burst initiation and explain why both apamin and STR/PTX are needed to trigger synchronized bursting.

In conclusion, the data in part-I of this dissertation showed for the first time a unique role for SK channels in controlling the initiation and setting the amplitude of spontaneous bursts in the spinal cord generated by interneuronal inputs to the motoneurons.

Figure 22: The effect of apamin vs. strychnine/picrotoxin on the low- amplitude/low-frequency inputs to the motor pool.

Average ventral root response to single pulses of dorsal root stimulation at 1xT before and after bath application of either apamin (A) or STR/PTX (B). Each ventral root response in presence of the drug(s) is normalized to its control response to account for the variability among preparations. Quantification of the effect of apamin or STR/PTX was done from multiple experiments. Data represented as the grand mean ± SD. * indicates a statistically significant difference.

Figure 23: The effect of apamin vs. strychnine/picrotoxin on the high- amplitude/high-frequency inputs to the motor pool.

Ventral root response to 5 pulses of dorsal root stimulation at 50 Hz with intensity of 10xT before and after bath application of apamin (A) or STR/PTX (B). The coAP amplitudes are normalized to the amplitude of the first pulse recorded in nACSF without any drugs. Quantification of the effect of apamin or STR/PTX was done from multiple experiments. Data represented as the grand mean ± SD. * indicates a statistically significant difference.

Figure 24: synchronized bursting is evoked through polysynaptic pathways.

Spontaneous bursting was first induced using strychnine, picrotoxin, and apamin. Mephenesin, a drug that blocks the polysynaptic pathways, was then added to the bath solution where it successfully blocked the synchronized bursts.

PART-II: SENSORIMOTOR NETWORK INPUTS MEDIATING

STABLE MOTOR OUTPUT

Synaptic inputs from multiple peripheral and central sources carrying different information get integrated in the final pathway to movement via spinal motoneurons. Depending on the synaptic strength and firing frequency of the presynaptic neuron, these inputs exhibit different patterns of short-term plasticity (either facilitation or depression) though the motor output of the spinal cord is maintained steady during most motor tasks. This part of the study investigated how varying sensory and descending (motor) inputs to motoneurons in the adult mouse spinal cord could generate a steady motor output.

SPECIFIC AIM 3: TO STUDY THE PLASTICITY PROFILE AND

INTEGRATION OF SENSORY AND MOTOR INPUTS TO MOTONEURONS

The inputs were induced by a short train of electrical stimuli delivered either to the dorsal roots (sensory input), the remaining descending axons (Motor input), or both of them simultaneously (Sensorimotor inputs). Stimulation was delivered at different frequencies, intensities, and neuromodulatory states. The motor output in response to stimulation was recorded from single motoneurons (excitatory postsynaptic potentials, EPSPs) using intracellular sharp electrodes, and from ventral roots (compound action potentials, coAPs) using extracellular wire electrodes (see Figure 4).

Most published studies of synaptic integration use single pulse stimulation to induce synaptic responses. However, meaningful motor outputs are achievable only through repetitive firing in motoneurons (see part I of this study for more details). Therefore, we used a 5-pulse train of stimulation to induce synaptic responses. The 5 pulses were

70 delivered at 25 and 50 Hz; since previous reports showed robust synaptic plasticity at these physiological frequencies (Jiang et al. 2015). Two stimulation intensities were used for each input; 1.5xT and 10xT (T is threshold). The threshold for the sensory or the descending pathway was defined as the minimum amount of current used to stimulate that particular pathway to trigger the smallest response in the ventral roots.

Specific aim 3A: Plasticity patterns of sensory and motor inputs in the isolated sacral cord.

Sensory inputs to spinal motoneurons exhibit depression with repeated stimulation

To trigger sensory synaptic inputs in motoneurons, the dorsal roots (containing the axons of sensory neurons) were stimulated electrically using 1.5xT or 10xT 5-pulse train at 25 or

50 Hz. The response to dorsal root stimulation was most consistent and stable at the sacral

S3-S4 as well as the caudal Ca1-Ca2 ventral roots, thus data collected from these segments were used for analysis.

At both intensities and frequencies, extracellular ventral root recordings showed that the amplitude of the compound action potential (coAP) becomes progressively smaller with repetitive stimulation of the dorsal roots (Figure 25, top panel). Figure 26A shows summary of data collected from multiple experiments (n=15/each, one-way ANOVA, p<0.0001 for all intensity/frequency combinations).

When synaptic responses were recorded intracellularly in single motoneurons, the cells initially fired action potentials but later failed with subsequent pulses (Figure 25, middle panel). To measure the exact amplitude of the synaptic potentials, QX-314 (blocker of the voltage-gated Na+ channels) was added to the internal microelectrode solution to block

71 spiking (Figure 25, bottom panel). Synaptic potentials in response to repetitive dorsal root stimulation at different frequencies and intensities exhibited gradual depression (Figure

26B, n=9/each, one-way ANOVA, p≤0.005 for all intensity/frequency combinations).

Interestingly, the depression seen in the amplitude of coAPs was much more pronounced compared to the depression in synaptic potentials (compare A and B in Figure 26). By the fifth pulse of stimulation, the amplitudes of coAPs decreased by 85-94%, while the corresponding EPSPs showed only 22-35% reduction.

Plasticity of the sensory input is partially due to short-term synaptic depression

The depression seen in the sensory signal could be the result of different factors: 1) the sensory axons fail to fire at the stimulation frequencies used, 2) the activation of polysynaptic inhibitory pathways in the cord which affects later pulses than earlier ones, and/or 3) depression of synaptic transmission caused by depletion of neurotransmitter in the terminal.

First, to investigate if the depression in the sensory input is caused by stimulation failure, we used another bipolar electrode on the dorsal root (labeled “C” in Figure 4) to record the root potential which is dependent on the number of activated axons. At 25 Hz, RM one- way ANOVA showed that dorsal root potentials at all pulses were not different when stimulated at 1.5xT (Figure 27A, p=0.3) or 10xT (Figure 27B, p=0.7). Similar results were obtained at 50 Hz of stimulation (not shown). This indicates that similar number of sensory axons are consistently recruited with every one of the five pulses of stimulation.

Interestingly, when stimulation intensity was increased to 100xT, no further increase in the root potential amplitude was seen indicating that 10xT recruited the maximum number of sensory axons.

Second, when inhibitory synaptic transmission is blocked using strychnine and picrotoxin, the sensory input shows less depression (see Figure 23), indicating that depression is partly caused by activation of inhibitory neurons. However, the fact that depression still happens while inhibition is blocked indicates that STD at the terminal contributes to the depression of the sensory input.

Figure 25: Motor pool and single cell responses to 25 Hz 5-pulse train of electrical stimulation of the dorsal roots in the sacral cord.

Top: Extracellular ventral root recording showing gradual depression of coAPs’ amplitude with repetitive stimulation of the dorsal roots at 1.5xT. Middle: Intracellular recording from a motoneuron that failed to fire action potentials with the last 3 pulses of stimulation. Bottom: When QX-314 is added to the internal microelectrode solution, action potential firing is inhibited and EPSPs show a similar pattern of depression to that seen in the ventral root.

8 1 .5 x T / 2 5 H z

1 .5 x T / 5 0 H z

1 0 xT / 2 5 H z

1 0 xT / 5 0 H z

) 6 V 

m  ( #  # e 

d #

t i

l 

p 4 

m 

A  # 

* * #  P * * #  A  #  o   

C # 2     # *  #  *  #   # * *  #  *  *  # *  # * *  * *  * 0

P 1 P 2 P 3 P 4 P 5 B

2 0 1 .5 x T / 2 5 H z

1 .5 x T / 5 0 H z

1 0 xT / 2 5 H z

1 0 xT / 5 0 H z 

) 1 5

m (

# # # e #  #  d # #  u * *  # t * * i  l * * 1 0 * * p * *

m *

S P

E 5

0 P 1 P 2 P 3 P 4 P 5

Figure 26: Plasticity pattern of the sensory inputs at the system and cellular levels.

A: Summary of the ventral root responses to repetitive dorsal root stimulation at different frequencies and intensities. The coAP amplitude consistently exhibits gradual depression with successive stimuli. B: The EPSPs recorded intracellularly from motoneurons shows a similar but less potent plasticity pattern. P1 to P5 indicate the coAP (A) or EPSP (B) at a particular pulse number. RM one-way ANOVA was used to study the pattern of adaptation at each intensity/frequency combination. Data represented as the mean ± SEM. *: different from P1 of 1.5xT / 25Hz, Ψ: different from P1 of 1.5xT / 50Hz, #: different from P1 of 10xT / 25Hz, and Ω: different from P1 of 10xT / 50Hz.

Figure 27: Depression of the sensory response is not caused by stimulation failure.

The dorsal root potentials recorded in response to electrical stimulation of the roots’ distal end at 1.5xT (A) and 10xT (B) do not show any adaptation with 5-pulse stimulation at 25 HZ (RM one-way ANOVA). Top: Example dorsal root potentials recorded in the same dorsal root at different intensities. Bottom: summary of responses from multiple experiments. The root potentials were normalized to the amplitude of the first pulse in each experiment. Data represented as the mean ± SD.

Descending (motor) inputs to motoneurons exhibit facilitation with repeated stimulation

To trigger the descending (motor) synaptic input in motoneurons, a custom-made concentric electrode (see methods for details) was placed on the cord ventral surface below the lumbosacral enlargement next to the midline to stimulate the descending fibers. At this part of the cord, the stimulation activates mainly the axons of the lateral vestibulospinal tract (LVST, see discussion for details) (Jiang et al. 2015). The response to descending stimulation was largest in amplitude at the upper segments of the sacral cord (not shown).

However, we focused our analysis on the response at the sacral S3-S4 as well as the caudal

Ca1-Ca2 ventral roots where the sensory response is also consistent and stable.

The descending fibers were stimulated electrically using 1.5xT or 10xT 5-pulse train at 25 or 50 Hz. Ventral root recordings showed that the amplitude of the compound action potential (coAP) gradually increases with repetitive stimulation at both intensities and frequencies (Figure 28, top panel). Data collected from multiple experiments show the same pattern of facilitation (Figure 29A, n=15/each, one-way ANOVA, p<0.01 for all intensity/frequency combinations).

In single motoneurons, stimulation of descending fibers generated synaptic potentials of increasing amplitude which resulted in generation of action potentials with the later pulses

(Figure 28, middle panel). When action potentials were blocked by QX-314 (Figure 28, bottom panel), synaptic potentials showed similar pattern of facilitation at different frequencies and intensities (Figure 29B, n=9/each, one-way ANOVA, p≤0.008 for all intensity/frequency combinations).

The depolarization in the baseline potential between the pulses (Figure 28, bottom panel) is probably due to excitatory interneuronal network activity (Jiang et al. 2015), and usually lasts for few seconds after the stimulation train (not shown).

Similar to sensory inputs, the degree of facilitation was more profound in the coAPs compared to synaptic potentials (Figure 29). The amplitudes of coAPs increased by 25-

200%, compared to only 22-40% increase in the EPSPs’ amplitudes.

Descending stimulation specifically activated descending fibers on one side of the cord

It is possible and manifestly unavoidable that our descending stimulation might also evoke local descending pathways in the sacral cord. However, we wanted to ensure that the current is not spreading non-specifically in the tissue. To test the specificity of stimulation, we monitored the response of the ventral roots on both sides of the cord. The response was mainly evoked on the same side of stimulation, and minimal or no response was seen on the contralateral side (Figure 30A). This indicates that the stimulation did not activate other local circuits that project to the other side of the cord. These data are also in agreement with anatomical data suggesting that the LVST projections are mostly ipsilateral in the lumbosacral cord (Liang et al. 2014).

To further confirm, we combined descending stimulation with dorsal root stimulation and monitored the ventral root output on both sides. The side of the cord which received only sensory input showed the typical pattern of synaptic depression (Figure 30B). On the contralateral side, however, the combined descending and sensory stimulation generated a steady motor output, characteristic of sensorimotor integration (discussed in details below in specific aim 3B).

Figure 28: Response to 25 Hz 5-pulse train of electrical stimulation of the descending axons in the sacral cord

Top: Extracellular ventral root recording showing gradual increase of coAPs’ amplitude with repetitive stimulation of the descending axons at 1.5xT. Middle: Intracellular recording from a motoneuron that fired action potentials with the last 3 pulses of stimulation and not the earlier two. Bottom: When QX-314 is used in the microelectrode to block action potential firing, EPSPs show a similar pattern of facilitation to that seen in the ventral root.

A 1 .5 x T / 2 5 H z 1 4 1 .5 x T / 5 0 H z

1 0 xT / 2 5 H z 1 2 1 0 xT / 5 0 H z # #  ) # 

V #  #  # 

m 1 0

(  #  

e  d

u 8

l p

m 6

  A

4   o  

C       2 * * * *

0 P 1 P 2 P 3 P 4 P 5

1 .5 x T / 2 5 H z 3 0 1 .5 x T / 5 0 H z 1 0 x T / 2 5 H z 1 0 x T / 5 0 H z 2 5  )      V    #  # #  

m # #  #  ( # # # 

#  # 2 0 # # # e # #   d   

u 

t  

i  l 1 5  

p 

*  * m *  * * *

a * * * * * * * *

P 1 0

P E

P 1 P 2 P 3 P 4 P 5

Figure 29: Plasticity pattern of the descending inputs at the system and cellular levels.

A: Summary of the ventral root responses to repetitive stimulation of descending axons at different frequencies and intensities. The coAP amplitude consistently exhibits gradual facilitation with successive stimuli. B: The EPSPs recorded intracellularly from motoneurons shows a similar plasticity pattern. P1 to P5 indicate the coAP (A) or EPSP (B) at a particular pulse number. RM one-way ANOVA was used to study the pattern of adaptation at each intensity/frequency combination. Data represented as the mean ± SEM. *: different from P1 of 1.5xT / 25Hz, Ψ: different from P1 of 1.5xT / 50Hz, #: different from P1 of 10xT / 25Hz, and Ω: different from P1 of 10xT / 50Hz.

Figure 30: Descending stimulation does not spread non-specifically in the tissue.

A: When descending stimulation is used on one side of the cord, a response is evoked in the ventral roots ipsilateral to stimulation but not on the contralateral side. B: In another experiment, the descending stimulation on one side was triggered simultaneously with dorsal stimulation on both sides of the cord. A steady motor output is generated in the left ventral root where both inputs are induced, while the contralateral side shows typical gradual depression which is characteristic of the sensory input.

Delay and decay of sensory and motor inputs

To better understand how the sensory and motor inputs would interact in the motoneurons, we measured different parameters of the EPSPs generated by each input (Figure 31); the data is summarized in Table 2. The synaptic potentials generated from each input had similar delays (p=0.46, paired t-test), which means that both reached the soma of motoneurons synchronously. However, the EPSPs generated by the motor input took longer to reach their peak amplitude (p= 0.04, paired t-test), and to decay (p=0.01, paired t-test) than those of the sensory input. These differences could be due to multiple anatomical differences between the two pathways (see discussion).

The slow decay of the descending motor inputs increases the chance for integration, which would be more influential when the other synaptic inputs have different onsets.

Figure 31: Measurements of timing parameters for the synaptic potentials.

The delay was measured as the time from the stimulus trigger to the onset of the rising phase of the EPSP. The time from the EPSP onset to its peak amplitude was defined as

“time to peak” (TPeak). Half decay time (T½) was measured from the peak to the point where the EPSP decayed to 50% of its peak amplitude. These parameters were measured from EPSPs triggered by single pulses to each individual pathway.

Table 2: Different timing parameters for the synaptic potentials of the sensory and motor inputs.

Time parameters Sensory Input Motor Input (Mean ± SD, milliseconds)

Delay from trigger 1.98 ± 0.6 1.85 ± 0.36

Time to peak (TPeak) 1.67 ± 0.63 2.19 ± 0.63

Half decay time (T½) 3.79 ± 1.75 5.58 ± 1.99

Specific aim 3B: Spinal sensorimotor integration at the cellular and system levels.

The data above showed that two of the main inputs to the motor pool result in wildly changing motor output. To investigate how a stable motor output can be generated from varying sensory and motor inputs, we measured the cellular and system motor outputs in response to simultaneous activation of the two inputs.

Integration of sensory and motor inputs at 1.5xT stimulation intensity

When dorsal root and descending fibers were stimulated simultaneously at 1.5xT/25 Hz, the resulting output in the ventral roots showed less adaptation (Figure 32, top panel and

“S&M” in Figure 33A). RM one-way ANOVA revealed that the amplitudes of the coAPs at the the first three pulses were not different (Figure 33A). At the fourth and fifth pulses, there was a decline of 15% and 28% respectively, compared to 50% and 59% decrease in the mathematical sum of separate inputs (S+M), 86% and 95% decrease when the sensory input is activated separately (S), and 122% and 110% increase when the motor input is induced separately (M).

Using RM two-way ANOVA, the amplitudes of the coAPs resulting from combined inputs

(S&M) were significantly larger (p=0.0006) than the mathematical linear summation of separate inputs (S+M) at all five pulses (Figure 33A). This indicates that integration of sensory and motor inputs does not only result in a steady motor output, but also follows a supralinear fashion.

Figure 32: Response to 25 Hz 5-pulse train of simultaneous stimulation of the dorsal roots and descending axons.

Recordings from the same ventral root and cells in Figure 25 and Figure 28 but with combined stimulation of the sensory and descending fibers at 1.5xT/25 Hz. Top: Ventral root recording showing stable coAPs’ amplitude with repetitive stimulation. Middle: Motoneurons consistently fired action potentials with rare failures at later pulses. Bottom: EPSPs, recorded with QX-314 in the internal microelectrode solution, have similar amplitudes throughout the stimulation train.

On the cellular level, the integration of the two inputs resulted in non-adapting synaptic potentials with the EPSP amplitudes not changing at any pulse (Figure 32, bottom panel and “S&M” in Figure 33B, p=0.38). On the other hand, the separate sensory or motor inputs exhibited about one third decrease or increase, respectively, by the fifth pulse.

Surprisingly, the profile of synaptic potential summation was opposite to the one observed with the ventral root coAPs. On the cellular level, the EPSPs showed sublinear summation

(Figure 33, S&M

Simultaneous stimulation of the dorsal roots and descending fibers at the same intensity

(1.5xT) but higher frequency (50 Hz), similarly resulted in more stable outputs. The output in the ventral roots showed less adaptation than the mathematical summation of individual coAPs (Figure 34A). The amplitudes of the last three S&M coAPs showed a reduction of

18%, 17%, and 33% as compared to 44%, 39%, and 57% in the repective S+M coAPs.

On the cellular level, the amplitudes of the integrated synaptic potentials did not change at any pulse (Figure 34B, “S&M”, p=0.44), while the separate sensory and motor inputs exhibited about one third decrease or two thirds increase, respectively, by the fifth pulse.

Additionally, the amplitudes of the S&M coAPs were significantly larger (RM two-way

ANOVA p<0.0001) than the linear summation (S+M) at all five pulses (Figure 34A), indicating supralinear summation. On the motoneuron level, however, the profile of synaptic potential summation was sublinear (p=0.001).

1 2 * * * * * * * * * * * * * * * * * * * *

) 1 0 V &

m & & (

e 8 &

d &

i l

p 6

m + +

A + + S + + $ + P 4 + + + M $ + A $ + S + M +

o m $ m $ m m $ $ S & M C $ 2 $ $ $ $ $ $ $ $ 0 P 1 P 2 P 3 P 4 P 5 B

3 0 * * * * * * * * * * * * * * * * * * * *

2 5 +

) +

V m

( +

2 0

t i

l m 1 5 m p m m m m

m $ m $ m m m m $ m $ $ a m m $ $ $ $ P $ S

1 0 $ S

M P

E S + M S & M 5

0 P 1 P 2 P 3 P 4 P 5

Figure 33: Integration of sensory and motor inputs at 1.5xT/25 Hz stimulation.

Summary of responses to 1.5xT/25 Hz electrical stimulation of sensory dorsal root inputs (S), descending motor inputs (M), or both simultaneously (S&M) at the motor pool level (A) and the cellular level (B). S+M represents the mathematically calculated linear summation of the two inputs. P1 to P5 indicate the coAP (A) or EPSP (B) at a particular pulse number. A: The integration of the two inputs results in a more steady motor output (S&M), and supralinear summation of the coAPs (S&M > S+M). B: The integration of the two inputs results in non-adapting synaptic potentials (S&M), and sublinear summation of the EPSPs (S+M > S&M). Data represented as the mean ± SEM. RM one- way ANOVA was used to study the pattern of adaptation of each input. RM two-way ANOVA was used to test the type of integration (S&M vs. S+M). $: different from S P1, m: different from M P1, +: different from S+M P1, &: different from S&M P1, *: S&M different from S+M.

1 2 * * * * * * * * * * * * * * * * * * * *

1 0

) V

& & m

( & & & 8 & &

e & d

& u

t &

i l

p 6

+ m $ + + A + $ m + m $ + m + + P 4 m + $ + m m A m + m m +

o m S $ m $ m

C $ $ M $ $ 2 $ S + M $ $ $ $ S & M $

0 P 1 P 2 P 3 P 4 P 5

4 0 * * * * * * * * * * * * * * * * * * * *

3 5 )

+ V 3 0 + + + +

m + + + + ( + + + + + + +

e 2 5

t i

l m m 2 0 m p m m m m m

m m

$ m m a

m 1 5 m

P m $ S $ S

P 1 0 M E S + M 5 S & M

0 P 1 P 2 P 3 P 4 P 5

Figure 34: Integration of sensory and motor inputs at 1.5xT/50 Hz stimulation.

Summary of responses to 1.5xT/50 Hz electrical stimulation of sensory dorsal root inputs (S), descending motor inputs (M), or both simultaneously (S&M) at the motor pool level (A) and the cellular level (B). S+M represents the mathematically calculated linear summation of the two inputs. P1 to P5 indicate the coAP (A) or EPSP (B) at a particular pulse number. A: The integration of the two inputs results in a more steady motor output (S&M), and supralinear summation of the coAPs (S&M > S+M). B: The integration of the two inputs results in non-adapting synaptic potentials (S&M), and sublinear summation of the EPSPs (S+M > S&M). Data represented as the mean ± SEM. RM one- way ANOVA was used to study the pattern of adaptation of each input. RM two-way ANOVA was used to test the type of integration (S&M vs. S+M). $: different from S P1, m: different from M P1, +: different from S+M P1, &: different from S&M P1, *: S&M different from S+M.

Sublinearity in the summation of EPSPs could be due to local reduction in driving force across the dendritic membrane caused by the activation of a large number of depolarizing conductances. To investigate this possibility, computer simulations using Neurons in

Action software (version 2) were done using the “interactions of synaptic potentials” simulation. In this model, two neighboring excitatory synapses were inserted in a membrane patch. The onset and time to peak for each synapse were matched to the experimental data (Table 2) for either the sensory or the motor input. The maximum conductances for each synapse were weighted based on the relative amplitudes of the first synaptic potential in the train produced by each input at 1.5xT stimulation intensity. The different parameters for the simulation are summarized in Table 3. All parameters were kept the same during each run of the simulations except for the reversal potential of the synapses which was varied to change the driving force (resting membrane potential kept at

-65 mV). Both the sensory and descending (motor) synaptic responses in our preparation are eliminated by DNQX (blocker of AMPARs, not shown) indicating that both synapses are glutamatergic and thus have equal equilibrium potential. Consequently, the value of the equilibrium potential was changed from one run to another but kept equal for both synapses during each run. Figure 35 shows the effect of changing the driving force of the synapses on the difference between the integrated EPSP and the mathematical summation of individual EPSPs. As the driving force increases, the magnitude of sublinearity is increased, and the gain of the relationship was determined by the conductances of the synapses. Thus, large amplitude EPSPs (i.e. strong synapses or high density of synapses) are more likely to summate sublinearly.

Table 3: Parameters for computer simulations of two excitatory synapses in a membrane patch.

Synapse 1 (S) Synapse2 (M)

EPSP delay (Onset, ms) 1.98 1.85

Time to peak (Tpeak, ms) 1.67 2.19 1X = 0.17 1X = 0.111 Maximum conductance (Gmax, µS) 2X = 0.34 2X = 0.222 Reversal potential (E, mV) Varied Varied

2X conductance 6 1X conductance

4 Loss from linearity from (mV) Loss 2

0 -60 -40 -20 0 20 40 60

Equilibrium Potential

Figure 35: Dependence of EPSP integration on driving force.

The deviation of EPSP integration from linearity plotted as a function of change in driving force (changing equilibrium potential at the same membrane potential). The slope of the relationship becomes steeper as the conductance of the synapses increase.

Integration of sensory and motor inputs at 10xT stimulation intensity

Our recordings of dorsal root potential during stimulation of the root’s distal end showed that 10xT intensity recruits the maximum number of sensory axons (not shown). This indicates that at this stimulation intensity, a maximal number of motoneurons will be activated, which correspond physiologically to maximum force generation by the innervated muscle. Importantly, the individual sensory and motor inputs show the same depression and facilitation patterns reported at the lower intensity (Figure 26, Figure 29, and Figure 36).

At the system level (ventral root recordings), the sensorimotor integration at 10xT tended to be more or less linear. At 25 Hz, there was no difference between the amplitudes of

S&M vs. S+M at the first three pulses (Figure 36A). The last two S&M pulses were about

20% higher than their equivalent S+M pulses. At 50 Hz, the last three S&M pulses were not different from those of S+M, and the first two were about 15% lower than respective

S+M responses (Figure 37A). Importantly, with both stimulation frequencies, the integrated inputs evoked a more stable motor output than the mathematical summation of individual inputs (Figure 36A and Figure 37A). In single motoneurons, the 10xT EPSPs integrated sublinearly at both 25 Hz and 50 Hz (Figure 36B and Figure 37B, two-way

ANOVA, p= 0.002 and 0.001, respectively).

The type of summation at the cellular and system levels with different frequencies and intensities of stimulation is summarized in Table 4.

A * 2 0 * n s n s n s * * *

1 8 )

V 1 6 +

+ m

( + 1 4 m + + & e m + &

d m 1 2 m + u m + +

t m + i

l m +

1 0 + p

m $

8 $ A

$ $ P 6

$ A

$ S o 4 $ $ M C $ $ $ $ $ S + M $ $ 2 $ S & M

P 1 P 2 P 3 P 4 P 5 B

4 0 * * * * * * * * * * * * * * * * * * * *

) 3 5 V +

m 3 0

(

e m m m m d 2 5 m m m m u m

t m m m i

l m

2 0 m m m p

m $ $

a 1 5 $ $ $

$ $ $ S P

S 1 0 M

P S + M E 5 S & M

0 P 1 P 2 P 3 P 4 P 5

Figure 36: Integration of sensory and motor inputs at 10xT/25 Hz stimulation.

Summary of responses to 10xT/25 Hz electrical stimulation of sensory dorsal root inputs (S), descending motor inputs (M), or both simultaneously (S&M) at the motor pool level (A) and the cellular level (B). S+M represents the mathematically calculated linear summation of the two inputs. P1 to P5 indicate the coAP (A) or EPSP (B) at a particular pulse number. A: The integration of the two inputs results in a more steady motor output (S&M), and linear summation of the coAPs except at P4 and P5 where it becomes supralinear. B: The integration of the two inputs results in non-adapting synaptic potentials (S&M), and sublinear summation of the EPSPs (S+M > S&M). Data represented as the mean ± SEM. RM one-way ANOVA was used to study the pattern of adaptation of each input. RM two-way ANOVA was used to test the type of integration (S&M vs. S+M). $: different from S P1, m: different from M P1, +: different from S+M P1, &: different from S&M P1, *: S&M different from S+M.

A * 2 0 * * * * n s n s n s * 1 8 *

) 1 6 + V + +

m m + ( 1 4 + + m + + + e m m & + & + d m m 1 2 m + & u m

t m

i l

p 1 0

$ m $

A 8

P $ S A 6 $

o $ $ $ M

C $ $ $ 4 $ $ S & M $ $ $ 2 S + M

P 1 P 2 P 3 P 4 P 5

B * * * * * 4 5 * * * * * * * * * * * * * * * 4 0 + +

) + + + + V 3 5 + + + + + +

m + (

3 0 e

d m m m m u m m m t 2 5 m i m m l m m m m p m m

2 0

P 1 5 $ S $ S $

P M 1 0 E S & M 5 S + M

P 1 P 2 P 3 P 4 P 5

100

Figure 37: Integration of sensory and motor inputs at 10xT/50 Hz stimulation.

Summary of responses to 10xT/50 Hz electrical stimulation of sensory dorsal root inputs (S), descending motor inputs (M), or both simultaneously (S&M) at the motor pool level (A) and the cellular level (B). S+M represents the mathematically calculated linear summation of the two inputs. P1 to P5 indicate the coAP (A) or EPSP (B) at a particular pulse number. A: The integration of the two inputs results in a more steady motor output (S&M), and linear summation of the coAPs except at P1 and P1 where it is sublinear. B: The integration of the two inputs results in non-adapting synaptic potentials (S&M), and sublinear summation of the EPSPs (S+M > S&M). Data represented as the mean ± SEM. RM one-way ANOVA was used to study the pattern of adaptation of each input. RM two-way ANOVA was used to test the type of integration (S&M vs. S+M). $: different from S P1, m: different from M P1, +: different from S+M P1, &: different from S&M P1, *: S&M different from S+M.

101

Table 4: Summary of the type of summation of sensorimotor inputs at different intensities/frequencies combinations.

Ventral root coAPs Motoneuron EPSPs

25 Hz 1.5xT Supralinear Intensity 50 Hz Sublinear 25 Hz 10xT Linear? Intensity 50 Hz

102

Effect of the neuromodulatory state on sensorimotor integration in the spinal cord

Our spinal cord preparation represents a basal state of activity, because it has a low level of neuromodulators. However, the neuromodulatory state of spinal networks can be tuned up via the administration of neurotransmitters and drugs such as serotonin and methoxamine (Li et al. 2007, Lee and Heckman 1999). The level of neuromodulation is thought to set the excitability of both presynaptic and postsynaptic neurons in the spinal network, therefore affecting patterns of plasticity (Nadim and Bucher 2014).

To test the effect of enhanced neuromodulatory state on sensorimotor integration, methoxamine (10 µM) was applied and the ventral root response to stimulation at 10xT/25

Hz was recorded (n=6). This intensity/frequency combination was chosen because it resulted in the most pronounced plasticity pattern in both inputs (see Figure 26 and Figure

29).

The addition of methoxamine resulted in reduced depression in the sensory response

(Figure 38A, RM-two way ANOVA, p<0.001) and more facilitation with the descending motor input (Figure 38B, p=0.007). Importantly, the response to integrated inputs (S&M) was more steady in presence of methoxamine, and became larger at the last two pulses

(Figure 38C, p=0.03).

To investigate the effect of methoxamine on the cellular level, the drug was tested in a small number of cells (n=4). When methoxamine was applied, the amplitudes of the synaptic potentials were consistently increased with both inputs (Figure 39). Interestingly, methoxamine slowed down the decay of the EPSPs; an effect which is thought to facilitate and favor integration of synaptic potentials (Magee 2000).

103

The results presented in this part of the dissertation shows opposite patterns of adaptation in the sensory vs. the descending motor inputs in the spinal cord. The data, however, suggest that integration of multiple, despite adapting, excitatory inputs help generate a stable motor output. The data also revealed contrasting fashions of summation of the synaptic effects at the cellular vs. the system levels. We believe that this discrepancy is due to the firing threshold of motoneurons acting as a high-pass filter. The summation of multiple excitatory inputs in motoneurons, despite being sublinear, it is enough to keep synaptic potentials above the firing threshold causing more cells in the pool to fire, and hence generating supralinear effect in the ventral roots. This explanation agrees with the data collected in presence of methoxamine where the firing threshold becomes more hyperpolarized.

104

S e n s o ry In p u t

* ) 1 5 * * * * V n s

* * * *

(

d 1 0

p m

A 5

P C o n tro l A

o M e th o x a m in e C 0

P 1 P 2 P 3 P 4 P 5

B M o to r in p u t

) 1 2 * * *

V n s

* * * *

(

d 8

p m

A 4

P C o n tro l A

o M e th o x a m in e C 0 P 1 P 2 P 3 P 4 P 5

C S & M in p u ts

) 2 0 * n s n s n s * *

V * *

m (

1 6

d u

t 1 2

l p

m 8

P 4 C o n tro l A

o M e th o x a m in e C 0

P 1 P 2 P 3 P 4 P 5

105

Figure 38: The effect of methoxamine on plasticity and integration of sensorimotor inputs at the system level.

The effect of methoxamine (10 µM) on the ventral root response to 10xT/25 Hz stimulation of the dorsal roots (A), descending axons (B), or both of them simultaneously (C). Methoxamine reduced the depression of the sensory inputs, increased the facilitation of the motor inputs, and helped generate a more stable motor output. Data represented as the mean ± SEM. RM two-way ANOVA was used to test the effect of methoxamine on each type of synaptic inputs.

106

Figure 39: Methoxamine increases the amplitude and slows the decay of sensory and motor synaptic potentials.

The application of methoxamine (10 µM) to the bath solution, seems to increase the amplitude of the EPSPS of both the sensory (top) and descending motor inputs (bottom). The drug also resulted in prolonged duration and slow decay of the EPSPs

107

DISCUSSION

The present study investigated the three main synaptic inputs that contribute to the generation of spinal motor output for locomotion. These inputs are: 1) premotor interneuronal inputs from the central pattern generators (CPGs), 2) peripheral sensory inputs from the muscle and cutaneous afferents, and 3) descending motor inputs from supraspinal structures.

First, we used a pharmacological approach to examine the contribution of some ionic conductances in generating spontaneous rhythmic activity from the CPGs in the spinal cord. We demonstrated a novel mechanistic role for SK channels in initiating and modulating the amplitude of spontaneous rhythmic motor outputs. One of the challenges with pharmacological manipulations of neuronal tissues is the specificity of reagents used.

To overcome this challenge, we used several very different pharmacological approaches – some involving direct and others indirect effects – and showed consistently that SK channel inhibition is required for initiation of different forms of spinal rhythmic outputs. We also showed that grading the level of SK inhibition modulates the burst amplitude in a dose- dependent manner.

Second, we investigated how variable sensory and descending inputs can generate stable motor outputs. We recorded the synaptic responses to each type of input from single motoneurons (intracellular recording) as well as from ventral roots (extracellular recording). Our recordings revealed opposite dynamics of synaptic plasticity for each input.

When the two inputs were combined, the summated response followed opposite fashions

108 at the cellular vs. the system levels. However, the two inputs generated a stable motor output because the integrated synaptic potentials were maintained above the firing threshold. This was confirmed when the motor response to separate and combined inputs were measured at a higher neuromodulatory state.

THE ROLE OF SK CHANNELS IN THE INITIATION AND

AMPLITUDE MODULATION OF RHYTHMIC MOTOR ACTIVITY

Induction of burst firing in the mature spinal cord tissue in vitro

Isolated spinal cord preparations have been long used to study alternating bursting using combinations of 5-HT/DA/NMDA (Smith and Feldman 1987, Kudo and Yamada 1987) and synchronized bursting using disinhibition (Bracci, Ballerini, and Nistri 1996b). Due to technical challenges (Jiang, Carlin, and Brownstone 1999), most of the published work about bursting in the spinal cord was done in neonatal animals less than a week old.

Although the tissue from these young animals is easier to maintain in vitro, it models a functionally-immature state in which some ionic conductances are still developing.

Throughout the current study, we have used spinal cord tissue from adult and functionally- mature young mice, to show the crucial role of SK channels in both the synchronized as well as the alternating bursting, respectively. The use of mature spinal cord tissue in our experiments guarantees that the role of SK channels in bursting continues during adult life.

Interestingly, our data show that adding methoxamine to a cocktail that normally induces alternating bursts, resulted in slow synchronous activity with a cycle and frequency that resembled disinhibited bursting (Mahrous and Elbasiouny 2017b). However, a fast

109 alternating bursting was found within these synchronous episodes (fig. 2A), which we referred to as episodic alternating pattern. The new drug combination containing methoxamine induced the episodic alternating pattern reliably in spinal cord preparations from functionally mature mice (up to postnatal day 15).

Evidence suggesting SK channel inhibition during bursting

As the motoneuron fires repetitively during the burst, Ca2+ would accumulate inside the cell and progressively activate SK channels, which would terminate motoneuron firing quickly (el Manira, Tegner, and Grillner 1994, Hill et al. 1992). Therefore, somatic SK channels would need to be inhibited to some degree for motoneurons to sustain firing for several seconds. In fact, there is mounting evidence that suggests the inhibition of motoneuronal SK channels during bursting behavior. First, most cocktails used to evoke motoneuron bursting in vitro contain one or more agents that inhibit SK channels, such as serotonin (Grunnet, Jespersen, and Perrier 2004, Wallen et al. 1989, Talley, Sadr, and

Bayliss 1997), dopamine (Han et al. 2007), methoxamine (Wagner, Ronnekleiv, and Kelly

2001), and bicuculline (Debarbieux, Brunton, and Charpak 1998, Khawaled et al. 1999).

Second, temporary reduction in AHP amplitude has been observed during bursting both in vivo and in vitro (Brownstone et al. 1992, Schmidt 1994), suggesting active inhibition of

SK channels during bursting. Third, cholinergic interneurons, which activate C-boutons on motoneurons, have been found to fire in phase with ventral root locomotor bursts

(Zagoraiou et al. 2009). This suggests that cholinergic inhibition of SK channels through type 2 muscarinic (M2) receptors (Miles et al. 2007) may regulate the ongoing bursting activity.

110

Although acetylcholine at the synapse acts through G-protein-mediated modulation of SK channels, it could serve as an integral modulator of the locomotor network that is activated during the high-excitability neuromodulatory states. This idea suggests that cocktails of excitatory neurotransmitters or high concentrations of synaptic inhibition blockers (Jiang et al. 2009) could also inhibit SK channels indirectly by activating the cholinergic neurons within the spinal network.

SK channel inhibition facilitates burst initiation

Previous studies suggested a role for SK channels in regulating and terminating ongoing bursting activity (el Manira, Tegner, and Grillner 1994, Zagoraiou et al. 2009), but not in its initiation or gradation.

In the current study, we show for the first time a crucial role for SK channels in the initiation of both alternating and synchronized spinal rhythmic activity. STR and PTX failed to produce the synchronized bursting induced by strychnine and bicuculline. Since

PTX is more selective to GABAA receptors than bicuculline (Pflieger, Clarac, and Vinay

2002), we hypothesized that the SK channel-blocking effect of bicuculline (Khawaled et al. 1999, Kleiman-Weiner et al. 2009) is important for burst initiation. In presence of STR and PTX, when SK channels were inhibited via apamin, d-tubocurarine, or oxotremorine, or by lowering extracellular free Ca2+, high amplitude bursting was induced.

In the younger animals, different SK activators completely inhibited the robust alternating bursting induced by our new cocktail of neuromodulators. The direct blockade of SK channels afterwards using apamin restored the bursting fully. These experiments show SK inhibition is needed to initiate burst firing in the spinal cord.

111

The possible mechanisms underlying burst initiation by SK inhibition

Because of their widely-distributed expression and multiple effects on neuronal excitability, SK channel inhibition can facilitate burst firing through multiple mechanisms.

First, SK channel inhibition has been shown to increase the firing rate of neurons and increase the gain of their input-output relationship (Manuel et al. 2006, Eccles, Eccles, and

Lundberg 1958, Deardorff et al. 2014). This effect is expected to happen all over the spinal network (motoneurons and interneurons) shifting the network excitability towards a state that favors bursting (Sharples and Whelan 2017, Marder, O'Leary, and Shruti 2014).

Although, blocking synaptic inhibition using STR and PTX amplifies the excitatory drive in network, it was not enough to initiate bursting until SK channels were inhibited.

Second, in CPG cells, spontaneous firing is thought to be triggered by post-inhibitory rebound generated by a low voltage-activated (LVA, Figure 40A) Ca2+ current, and terminated by a Ca2+-activated K current (Tegner and Grillner 1999). This LVA current might be also facilitated by SK inhibition.

Third, motoneuron bursting is triggered by synaptic inputs from premotor interneurons.

The release of neurotransmitters from the synaptic terminals is dependent on the activation of voltage-gated Ca2+ channels (Neher and Sakaba 2008, Katz and Miledi 1965).

Presynaptic SK channels have been shown to negatively regulate Ca2+ influx and neurotransmitter release from premotor synaptic terminals in the spinal cord (Nanou et al.

2013). Therefore, SK inhibition facilitates the release of excitatory neurotransmitters from the premotor interneurons (Figure 40B).

112

Figure 40: Possible mechanisms for regulation of motor rhythmic bursting by SK channels in the spinal cord.

A: SK channels negatively regulate the LVA Ca2+ current responsible for the spontaneous firing of the CPG neurons. Inhibition of SK would facilitate the LVA current facilitating the initiation of bursting activity. B: Glutamate release from the premotor interneurons requires Ca2+ entry through N-type Ca2+ channels. Presynaptic SK channels exert negative feedback on the Ca2+ current and suppress the neurotransmitter release; an effect that would be reversed by SK inhibition. C: Dendritic SK channels activated by Ca2+ entry through AMPA and NMDA receptors which results in reduction of the amplitude of the EPSPs. Inhibition of dendritic SK channel enhances the excitatory inputs to the motoneurons (and possibly interneurons).

113

Fourth, at the motoneuron, glutamatergic synaptic inputs from interneurons drive the burst firing (discussed below). Synaptic currents mediated by glutamatergic receptors (Figure

40C) have been shown to be amplified by inhibition of SK channels (Ngo-Anh et al. 2005,

Nanou et al. 2013, Faber, Delaney, and Sah 2005). In agreement with this, we have shown in this study that apamin enhances the low frequency/amplitude excitatory inputs to the motor pool. These low frequency/amplitude inputs are expected to predominate at the initiation phase of the burst.

SK inhibition and burst amplitude modulation

The data showed that the burst amplitude correlates with the concentration of SK channel modulators; where SK inhibitors increased the amplitude and SK activators decreased it in a dose-dependent fashion. The enhancement of the synaptic inputs to motoneurons caused by inhibition of dendritic SK channels is expected to facilitate the recruitment of more motoneurons in the pool. Together with increasing the firing rates of motoneurons and interneuron, these effects can set the burst amplitude. In fact, in our study, only SK activators and AMPA receptor (AMPAR) blockers reduced the burst amplitude.

Our intracellular recordings during synchronized bursting suggested a form of synchrony among motoneurons of the same motor pool. When spike-triggered averaging was used to average the ventral root activity triggered by individual spikes in motoneurons during bursting, a peak was always seen in the average. Although, these cells were recorded separately, multiple cells from the same pool triggered an averaged peak in the same root during bursting. Furthermore, all the cells changed their firing frequency parallel to changes in the ventral root amplitude. This synchrony suggested that motoneurons could be connected through gap junctions.

114

Gap junctions have been found to connect motoneurons (Kiehn and Tresch 2002) and premotor interneurons (Hinckley and Ziskind-Conhaim 2006) in prenatal and early postnatal preparations. Additionally, histological studies showed that CX36 gap junctions continue to be expressed in adult spinal motoneurons (Rash et al. 2001, Bautista and Nagy 2014), but whether they play the same role in adult animals is still unclear. If their role continues during adulthood, then the level of expression and activation of gap junctions would have had a great impact on the burst amplitude. We showed that blocking gap junctions did not change the burst amplitude, indicating that the synchrony among the motor pools was not mediated through gap junctions. The motoneurons are more likely synchronized by a common synaptic input (Sears and Stagg 1976, Negro, Sukru Yavuz, and Farina 2016).

Specific SK channel role vs. non-specific excitability modulation

We tested the alternate hypothesis that burst initiation/termination could result from modulating the excitation-inhibition balance, not necessarily via SK channels. First, we increased the network excitability using electrical stimulation in presence of STR and PTX, but bursting was not elicited. Second, the motoneuron excitability was suppressed by blocking the Ca2+ PIC during bursting; however this did not terminate nor reduce the burst amplitude. Ca2+ PIC blockade was chosen to suppress the motoneuronal excitability because PICs are known to amplify synaptic inputs, strongly increase motoneuron excitability, and regulate the motor output (Heckman, Gorassini, and Bennett 2005). The burst duration was shortened when the Ca2+ PIC was blocked, which is in agreement with the role of the Ca2+ PIC in maintaining self-sustained firing after excitatory inputs are terminated (Lee and Heckman 1998).

115

Acutely spinalized preparations may have reduced PICs (Li and Bennett 2003) due to the loss of facilitation by descending serotoninergic systems (Hounsgaard et al. 1988).

Nevertheless, the Ca2+ PIC is increased in our experiments due to blocking of the dendritic

SK channels by apamin (Li and Bennett 2007, Hounsgaard and Mintz 1988). Moreover, our intracellular recording showed that motoneurons fired at high rates (up to 200 Hz, see

Fig. 7B), similar to in vivo preparations with uncompromised PIC activation (Meehan et al. 2010, Turkin et al. 2010). However, to confirm that Ca2+ PIC was fully activated, we elicited bursting by replacing apamin with serotonin, which both inhibits SK channels

(Berger, Bayliss, and Viana 1992, Grunnet, Jespersen, and Perrier 2004) and activates the

Ca2+ PIC (Hounsgaard and Kiehn 1985, Li et al. 2007).

Blocking the Ca2+ PIC with high nimodipine concentrations under these conditions did not change the burst characteristics, whereas the SK activator CyPPA eliminated the bursting.

These data confirm a specific role for SK channels inhibition in burst initiation, as well as a greater influence of SK channels versus Ca2+ PIC on the motor output. The dual regulatory role of dendritic and somatic SK channels on the motor pool input and output renders them a powerful mechanism for shaping the motor output.

Synaptic conductances mediating bursting activity

In order to investigate the regulatory effect of SK channels in bursting, we needed to identify the ionic conductances that mediate the bursting. We opted to answer this question using synchronized bursting because the drugs used to induce it are fewer and more specific than the neurotransmitter cocktails used to induce the alternating bursting. Evidence indicates that the core of spinal locomotor networks is composed of glycine and glutamate neurons in most vertebrates (Grillner 2003, Kiehn 2011). During synchronized bursting,

116 glycine and GABAA receptors were blocked; hence synaptic inhibition is not involved in the regulation or rhythmicity of these synchronized bursts. The glutamatergic receptors,

AMPA and NMDA, are then the candidate for driving this behavior. When NMDA receptors (NMDARs) were blocked using APV, there was an initial silent period after which shorter bursts with longer cycle periods were restored. This indicated that NMDARs are not an exclusive mechanism for driving the bursting.

The shortening of bursts after blocking NMDARs agrees with previous research that showed that NMDARs confer long-lasting plateau depolarization which can be activated in response to brief synaptic excitation (Manuel et al. 2012, Wallen and Grillner 1987). A similar effect was seen upon blockade of the Ca2+ PIC, which mediates a similar plateau potential (discussed above).

The decrease in burst frequency upon NMDAR blockade by APV is thought to be caused by reduced activation of SK channels (caused by loss of Ca2+ from NMDARs) which should be also accompanied by prolonged burst duration (Grillner 2003). However this cannot be the case in our data because: 1) the SK channels are fully blocked by apamin, and 2) the burst duration was not prolonged in presence of APV; it was actually reduced.

Alternatively, the reduced frequency might be caused by loss of NMDAR-mediated facilitation of the rhythm generation in the CPGs.

When AMPARs were separately blocked using DNQX, bursting completely stopped, indicating that AMPARs play a significant role in mediating the bursts. However, bursts could still be restored by bath application of NMDA or evoked by dorsal root stimulation; this indicates that both types of receptors are involved in generating the bursts. These findings agree with earlier work showing that synchronized and alternating bursting in the

117 neonatal rat spinal cord preparation are mediated by either NMDA or AMPA receptors

(Bracci, Ballerini, and Nistri 1996b, Beato, Bracci, and Nistri 1997). However, a more recent study considered NMDARs to be the main drive of this behavior, because the synchronized bursts in their study were eliminated by blocking NMDA receptors using

APV (Jiang et al. 2009). One possible explanation for these apparently contradictory results is the occasional desensitization of AMPA receptors upon blockade of NMDA receptors

(Ballerini, Bracci, and Nistri 1995, Constals et al. 2015). Another possibility is that APV application in Jiang et al. (2009) did not last longer than the silent period observed in this study (see results).

Cholinergic modulation of motoneuron bursting through SK channels

Although the core of the CPG is built on excitatory and inhibitory ligand-gated ion channels (Grillner 2003, Kiehn 2011), G-protein mediated systems are also integrated into the normal operation of the locomotor network. Cholinergic signaling has been shown to increase neuronal network excitability through different mechanisms, including enhancement of excitatory glutamatergic synaptic responses (Markram and Segal 1990,

Marino et al. 1998) and suppression of the AHP (Madison, Lancaster, and Nicoll 1987,

Lape and Nistri 2000). These two effects have been shown to be mediated by muscarinic inhibition of SK channels (Ngo-Anh et al. 2005, Faber, Delaney, and Sah 2005, Miles et al. 2007).

The muscarinic M1 receptors inhibit postsynaptic SK channels in the hippocampus, an effect mediated by casein kinase 2 (CK2) (Giessel and Sabatini 2010, Bildl et al. 2004,

Allen et al. 2007, Maingret et al. 2008) or protein kinase C (PKC) (Buchanan et al. 2010), leading to increased glutamatergic responses. On the other hand, suppression of the AHP,

118 at least in motoneurons, has been shown to be due to SK inhibition via M2 receptors (Miles et al. 2007) at the cholinergic “C-boutons” (Hellstrom et al. 2003, Wilson, Rempel, and

Brownstone 2004, Miles et al. 2007). Both enhanced synaptic inputs and suppressed AHP were expected to contribute to the facilitation of bursting in our study.

When oxotremorine sesquifumarate salt (a muscarinic agonist with higher selectivity to

M2) was applied to the disinhibited spinal cord, synchronized bursting was initiated and the burst amplitude was graded with the drug concentration. This effect was only inhibited by combined, not individual, blockade of M1 and M2 receptors. One could argue that M1 receptor-mediated suppression of other outward currents, such as M-current mediated by

KCNQ K+ channel (Alaburda, Perrier, and Hounsgaard 2002, Lombardo and Harrington

2016), could be contributing to burst initiation. However, the selective blockade of M1 receptors by telenzepine did not inhibit oxotremorine-induced bursting. Additionally, the facts that burst firing was 1) not evoked by oxotremorine when M1 and M2 receptors were blocked by concurrent pre-application of telenzepine and methoctramine, 2) evoked by the application of apamin subsequent to telenzepine and methoctramine concurrent administration, and 3) eliminated by administration of several SK channel activators

(CyPPA, SKA-19 and NS13001) further support that the effect of oxotremorine is mainly through SK channels.

The effect of SK modulators on motoneurons vs. interneurons

In the spinal cord, SK channels are expressed by α-motoneurons as well as interneurons.

However, the expression of SK channels in motoneurons is more pronounced than interneurons (Johnson and Sears 1988, Yamamoto, Hertzberg, and Nagy 1991, Deardorff et al. 2013). SK channels are known to form large clusters, along with other ion channels

119 and signaling proteins, on the motoneuron soma at the cholinergic synapse known as the

C-bouton (Hellstrom et al. 2003, Wilson, Rempel, and Brownstone 2004, Deardorff et al.

2013). The unique morphology of the C-bouton with these channel clusters on the postsynaptic membrane are specific to α-motoneurons and can be used to differentiate them from other neurons in the spinal cord (Deardorff et al. 2013, Yamamoto, Hertzberg, and

Nagy 1991). This may explain the relatively large AHP and slow firing rates of motoneurons compared to most interneurons (Hunt and Kuno 1959).

Although the SK channel modulators used in this study are expected to enhance excitatory inputs and firing rates of both motoneurons and interneurons, we think that their effects in different cell types serve different functions. The fact that burst initiation is a feature of the

CPG neurons suggests a more influential role of SK inhibitors in interneurons for burst initiation (discussed above under “the mechanism of burst initiation by SK inhibition”).

Nonetheless, the burst amplitude modulation is probably caused by modulation of motoneuronal SK channels (discussed under “SK inhibition and burst amplitude modulation”) which regulate the recruitment and firing rates of motoneurons.

120

SENSORIMOTOR INTEGRATION IN THE SPINAL CORD

The spinal CPG neurons have the ability to maintain rhythmic stereotyped activity for long time without any extrinsic inputs. However, the regulation by both sensory inputs from the periphery and motor inputs from the descending tracts is crucial for generating the proper spinal motor output. The descending motor inputs (motor commands) from supraspinal structures are required to initiate the activity of the CPGs (Kiehn 2016, Musienko et al.

2012). On the other hand, the sensory (mainly proprioceptive) inputs control the timing and amplitude based on information from the environment (Rossignol, Dubuc, and Gossard

2006). The data in part-II of the results provides detailed insights about integration of sensorimotor signals in the spinal cord and explains several physiological phenomena such as the ability to produce stable motor output from varying inputs.

Short-term plasticity of sensory and motor inputs

Use-dependent (or short-term) plasticity is a hallmark of synaptic transmission in the nervous system (Zucker and Regehr 2002, Fioravante and Regehr 2011). Therefore, repeated activation of sensory and motor synaptic inputs during different motor tasks is expected to trigger short-term plasticity at physiological frequencies. Repeated activation of synapses results in gradual increase or decrease in the resulting synaptic current in the postsynaptic cell. These changes are related to the Ca2+ dynamics and the probabilistic vesicular release at the presynaptic terminal (Regehr 2012, Neher and Sakaba 2008). We studied the short-term changes in synaptic activity of both the sensory and the motor inputs to spinal motoneurons.

121

Depression of the sensory input

The sensory input was induced by electrical stimulation of the ipsilateral dorsal roots, which contain the axons of sensory neurons including cutaneous and muscle afferents.

However, in-vitro stimulation of the dorsal roots generates a response that contain a prominent monosynaptic component presumably from the Ia muscle afferents (Jiang et al.

2015, Pinco and Lev-Tov 1993). Ia afferents belong to proprioceptive afferents which encode changes in muscle length and lengthening velocity (Hunt 1951), and play a crucial role in the regulation of locomotion (Rossignol, Dubuc, and Gossard 2006). Ia fibers form excitatory glutamatergic monosynaptic connections with ipsilateral motoneurons (Eccles

1946). When dorsal roots were stimulated repetitively, the synaptic response in motoneurons as well as the coAPs in the ventral roots showed progressive decline in amplitude. These results agree with published data that showed a similar STD pattern for sensory inputs in vitro (Jiang et al. 2015, Barriere et al. 2008, Lev-Tov and Pinco 1992).

This depression is not simply caused by the activation of polysynaptic inhibitory pathways because: 1) it was only partially relieved when synaptic inhibition was blocked, and 2) it persists when polysynaptic transmission is blocked by mephenesin (Lev-Tov and Pinco

1992). This indicates that this depression is, at least partially, caused by gradual reduction in neurotransmitter release.

Reduced vesicular release could be due to several factors , most commonly depletion of the readily releasable pool (RRP) (Fioravante and Regehr 2011). The depletion of the RRP is commonly seen in synapses with high initial release probability (Alabi and Tsien 2012), indicated by the large-amplitude of the first EPSP in a train of stimuli. When such synapse is entrained, the majority of the RRP is depleted with the early activation, and the

122 subsequent synaptic potentials becomes smaller. In agreement with this explanation, the plasticity pattern of this Ia afferent-motoneuron synapse was reversed to facilitation by lowering the extracellular calcium (Pinco and Lev-Tov 1993). With low Ca2+, the first

EPSP of the train becomes smaller and the subsequent ones are progressively larger due to accumulation of Ca2+ in the terminal and progressive release of the RRP.

Facilitation of the descending motor input

The descending motor inputs were induced by local stimulation of the remaining descending axons in the sacral cord. Only few descending tracts reach the sacral cord, including the lateral vestibulospinal tract (LVST) (Liang et al. 2014). Fibers of the LVST originate in the lateral vestibular nucleus and travels the entire length of the spinal cord where they synapse mainly on ipsilateral ventral horn neurons in the mouse, rat, and cat

(Liang et al. 2014, Kuze et al. 1999, Bacskai, Szekely, and Matesz 2002).

A few other descending fibers originating in oral pontine reticular nucleus (PnO), and gigantocellular reticular nucleus (Gi) have been also traced down to the sacral cord of the mouse (Liang, Watson, and Paxinos 2015, 2016). However, the Gi fibers project bilaterally in the spinal cord (Liang, Watson, and Paxinos 2016) and the PnO sends only small number of fibers to the lower segments of the spinal cord (Liang, Watson, and Paxinos 2015); these properties are different from the descending response in our data.

The local stimulation of the descending fibers in our experiments triggered a response with a similar delay to the sensory response. However, the EPSPs of the descending input had longer time-to-peak and half decay which could be due to the synapses being located more

123 distally on the motoneuron dendrites (Magee 2000, Spruston 2008) and/or because this pathway has a pronounced polysynaptic component (Jiang et al. 2015).

Upon repetitive stimulation, the descending inputs showed facilitation in both the ventral root and cellular responses. Interestingly, the baseline in-between the EPSPs at different pulses was elevated, and usually remained elevated for few seconds after the train.

Computer models of a motoneuron pool attributed this behavior to increased background network excitation (Jiang et al. 2015). The facilitation of the descending response could be due to either: 1) short-term synaptic facilitation (STF), caused by gradual accumulation of

Ca2+ in the synaptic terminals, or 2) increased background network activity, or a combination of both.

Of note, with either input, the ventral root response showed larger changes than synaptic responses recorded in single motoneurons (discussed below).

Integration of sensory and descending (motor) inputs

When the two inputs were combined, there were two main features of the resulting integrated sensorimotor response. First, opposite fashions of summation between cellular and ventral root responses. Second, the motor output of the integrated inputs was more stable than either one generated by individual inputs.

Contrasting summation at cellular and system levels

The coAP summation in the ventral roots was supralinear at 1.5xT and more or less linear at 10xT, while at the cell level, the EPSP summation was always sublinear. Sublinearity with the summation of large-amplitude synaptic potentials has been reported in other neuronal types (Magee 2000, Wolf, Zhao, and Roberts 1998). Using a simple

124 computational tool, we have shown that this phenomenon is due to a reduction in the local driving force. Two main findings to be pointed out from these simulations: 1) the degree of sublinearity increases when the driving force is large (for instance, this effect would be more prominent with the summation of EPSPs than IPSPs at the resting potential), 2) sublinearity increases when the synaptic conductance is increased (i.e. the summation of higher-amplitude EPSPs is expected to show more sublinearity). In fact, the sublinearity in our experimental data was higher at 10xT than at 1.5xT.

Despite of this sublinear summation, the smaller increases in the synaptic potential amplitudes had larger effects on the motor pool output. Supralinear summation predominated the ventral root data at 1.5xT intensity (which represents physiologically- relevant activation of the motor pool). This indicates that the motor output generated by integrated excitatory inputs is larger than additive at the system level.

To better understand the source of this discrepancy, it is helpful to consider that the output of any neuron is high-pass filtered by its firing threshold. Any synaptic potential that does not pass this filter is not detected at the system level. For instance, in our data set, the sensory EPSP amplitude declines from an average 11 mV at the first pulse of the 1.5xT/25

Hz train to about 7.5 mV at the last pulse. Consider an average motoneuron with a membrane potential of -65 mV, firing threshold of -55 mV, and sensory EPSPs were 11,

10, 9, 8, and 7 mV, then only the first two EPSPs will trigger action potentials and contribute to the coAP in the ventral roots.

125

Stable motor output generated by integrated inputs and higher neuromodulatory state

In the typical cell described above, if the descending input was coactivated and added ≥ 3 mV to the last 3 pulses, the cell would fire throughout the train. When this effect happens in a larger number of cells in the pool, the motor output starts to be stable throughout the stimulation train. Therefore, the simultaneous activation of multiple excitatory inputs to the motor pool ensures the generation of stable motor output by maintaining the synaptic potentials above the threshold for firing in individual motoneurons. This effect is expected to be more pronounced at higher neuromodulatory states. Methoxamine, the α1-adrenergic receptor agonist, as shown in the first part of this study, increases the excitability of spinal networks. The methoxamine-induced neuromodulatory state resulted in less sensory depression, more descending facilitation, and even more stable integrated output.

In awake behaving animals, when the motor pool is activated, e.g. during locomotion, it is receives several excitatory inputs including inputs from the CPGs, peripheral sensory inputs, and descending inputs. The motor pool output generated by summation of more than two excitatory inputs might no longer be supralinear; but it will definitely be more stable. In addition, the neuromodulatory state of a behaving animal is typically higher than our in vitro preparation, which would also favor the generation of stable motor outputs.

126

REFERENCES

Adelman, J. P., J. Maylie, and P. Sah. 2012. "Small-conductance Ca2+-activated K+

channels: form and function." Annu Rev Physiol 74:245-69. doi:

10.1146/annurev-physiol-020911-153336.

Alabi, A. A., and R. W. Tsien. 2012. "Synaptic vesicle pools and dynamics." Cold Spring

Harb Perspect Biol 4 (8):a013680. doi: 10.1101/cshperspect.a013680.

Alaburda, A., J. F. Perrier, and J. Hounsgaard. 2002. "An M-like outward current regulates

the excitability of spinal motoneurones in the adult turtle." J Physiol 540

(Pt 3):875-81.

Allen, D., B. Fakler, J. Maylie, and J. P. Adelman. 2007. "Organization and regulation of

small conductance Ca2+-activated K+ channel multiprotein complexes." J

Neurosci 27 (9):2369-76. doi: 10.1523/JNEUROSCI.3565-06.2007.

Arata, A., H. Onimaru, and I. Homma. 1998. "The adrenergic modulation of firings of

respiratory rhythm-generating neurons in medulla-spinal cord preparation

from newborn rat." Exp Brain Res 119 (4):399-408.

Avery, M. C., and J. L. Krichmar. 2017. "Neuromodulatory Systems and Their

Interactions: A Review of Models, Theories, and Experiments." Front

Neural Circuits 11:108. doi: 10.3389/fncir.2017.00108.

Bacskai, T., G. Szekely, and C. Matesz. 2002. "Ascending and descending projections of

the lateral vestibular nucleus in the rat." Acta Biol Hung 53 (1-2):7-21.

127

Ballerini, L., E. Bracci, and A. Nistri. 1995. "Desensitization of AMPA receptors limits the

amplitude of EPSPs and the excitability of motoneurons of the rat isolated

spinal cord." Eur J Neurosci 7 (6):1229-34.

Barbeau, H., and S. Rossignol. 1987. "Recovery of locomotion after chronic spinalization

in the adult cat." Brain Res 412 (1):84-95.

Barriere, G., M. Tartas, J. R. Cazalets, and S. S. Bertrand. 2008. "Interplay between

neuromodulator-induced switching of short-term plasticity at sensorimotor

synapses in the neonatal rat spinal cord." J Physiol 586 (7):1903-20. doi:

10.1113/jphysiol.2008.150706.

Bautista, W., and J. I. Nagy. 2014. "Connexin36 in gap junctions forming electrical

synapses between motoneurons in sexually dimorphic motor nuclei in

spinal cord of rat and mouse." Eur J Neurosci 39 (5):771-87. doi:

10.1111/ejn.12439.

Beato, M., E. Bracci, and A. Nistri. 1997. "Contribution of NMDA and non-NMDA

glutamate receptors to locomotor pattern generation in the neonatal rat

spinal cord." Proc Biol Sci 264 (1383):877-84. doi:

10.1098/rspb.1997.0122.

Beato, M., and A. Nistri. 1999. "Interaction between disinhibited bursting and fictive

locomotor patterns in the rat isolated spinal cord." J Neurophysiol 82

(5):2029-38.

Bennett, D. J., Y. Li, and M. Siu. 2001. "Plateau potentials in sacrocaudal motoneurons of

chronic spinal rats, recorded in vitro." J Neurophysiol 86 (4):1955-71.

128

Berger, A. J., D. A. Bayliss, and F. Viana. 1992. "Modulation of neonatal rat hypoglossal

motoneuron excitability by serotonin." Neurosci Lett 143 (1-2):164-8.

Bevan, S., and D. Parker. 2004. "Metaplastic facilitation and ultrastructural changes in

synaptic properties are associated with long-term modulation of the lamprey

locomotor network." J Neurosci 24 (42):9458-68. doi:

10.1523/JNEUROSCI.3391-04.2004.

Bildl, W., T. Strassmaier, H. Thurm, J. Andersen, S. Eble, D. Oliver, M. Knipper, M.

Mann, U. Schulte, J. P. Adelman, and B. Fakler. 2004. "Protein kinase CK2

is coassembled with small conductance Ca(2+)-activated K+ channels and

regulates channel gating." Neuron 43 (6):847-58. doi:

10.1016/j.neuron.2004.08.033.

Bracci, E., L. Ballerini, and A. Nistri. 1996a. "Localization of rhythmogenic networks

responsible for spontaneous bursts induced by strychnine and bicuculline in

the rat isolated spinal cord." J Neurosci 16 (21):7063-76.

Bracci, E., L. Ballerini, and A. Nistri. 1996b. "Spontaneous rhythmic bursts induced by

pharmacological block of inhibition in lumbar motoneurons of the neonatal

rat spinal cord." J Neurophysiol 75 (2):640-7.

Brown, A. G., and R. E. Fyffe. 1981. "Direct observations on the contacts made between

Ia afferent fibres and alpha-motoneurones in the cat's lumbosacral spinal

cord." J Physiol 313:121-40.

Brown, K. E., J. L. Neva, S. J. Feldman, W. R. Staines, and L. A. Boyd. 2018.

"Sensorimotor integration in chronic stroke: Baseline differences and

129

response to sensory training." Restor Neurol Neurosci 36 (2):245-259. doi:

10.3233/RNN-170790.

Brown, T. Graham. 1911. "The Intrinsic Factors in the Act of Progression in the Mammal."

Proceedings of the Royal Society of London. Series B (572 ):12.

Brownstone, R. M., L. M. Jordan, D. J. Kriellaars, B. R. Noga, and S. J. Shefchyk. 1992.

"On the regulation of repetitive firing in lumbar motoneurones during

fictive locomotion in the cat." Exp Brain Res 90 (3):441-55.

Buchanan, K. A., M. M. Petrovic, S. E. Chamberlain, N. V. Marrion, and J. R. Mellor.

2010. "Facilitation of long-term potentiation by muscarinic M(1) receptors

is mediated by inhibition of SK channels." Neuron 68 (5):948-63. doi:

10.1016/j.neuron.2010.11.018.

Carlsson, A., T. Magnusson, and E. Rosengren. 1963. "5-Hydroxytryptamine of the Spinal

Cord Normally and after Transection." Experientia 19:359.

Cazalets, J. R., M. Borde, and F. Clarac. 1995. "Localization and organization of the central

pattern generator for hindlimb locomotion in newborn rat." J Neurosci 15

(7 Pt 1):4943-51.

Cazalets, J. R., Y. Sqalli-Houssaini, and F. Clarac. 1992. "Activation of the central pattern

generators for locomotion by serotonin and excitatory amino acids in

neonatal rat." J Physiol 455:187-204.

Coleman, N., H. M. Nguyen, Z. Cao, B. M. Brown, D. P. Jenkins, D. Zolkowska, Y. J.

Chen, B. S. Tanaka, A. L. Goldin, M. A. Rogawski, I. N. Pessah, and H.

Wulff. 2015. "The riluzole derivative 2-amino-6-trifluoromethylthio-

benzothiazole (SKA-19), a mixed KCa2 activator and NaV blocker, is a

130

potent novel anticonvulsant." Neurotherapeutics 12 (1):234-49. doi:

10.1007/s13311-014-0305-y.

Constals, A., A. C. Penn, B. Compans, E. Toulme, A. Phillipat, S. Marais, N. Retailleau,

A. S. Hafner, F. Coussen, E. Hosy, and D. Choquet. 2015. "Glutamate-

induced AMPA receptor desensitization increases their mobility and

modulates short-term plasticity through unbinding from Stargazin." Neuron

85 (4):787-803. doi: 10.1016/j.neuron.2015.01.012.

Deardorff, A. S., S. H. Romer, Z. Deng, K. L. Bullinger, P. Nardelli, T. C. Cope, and R. E.

Fyffe. 2013. "Expression of postsynaptic Ca2+-activated K+ (SK) channels

at C-bouton synapses in mammalian lumbar -motoneurons." J Physiol 591

(4):875-97. doi: 10.1113/jphysiol.2012.240879.

Deardorff, A. S., S. H. Romer, P. M. Sonner, and R. E. Fyffe. 2014. "Swimming against

the tide: investigations of the C-bouton synapse." Front Neural Circuits

8:106. doi: 10.3389/fncir.2014.00106.

Debarbieux, F., J. Brunton, and S. Charpak. 1998. "Effect of bicuculline on thalamic

activity: a direct blockade of IAHP in reticularis neurons." J Neurophysiol

79 (6):2911-8.

Di Pasquale, E., D. Morin, R. Monteau, and G. Hilaire. 1992. "Serotonergic modulation of

the respiratory rhythm generator at birth: an in vitro study in the rat."

Neurosci Lett 143 (1-2):91-5.

Drew, T., and D. S. Marigold. 2015. "Taking the next step: cortical contributions to the

control of locomotion." Curr Opin Neurobiol 33:25-33. doi:

10.1016/j.conb.2015.01.011.

131

Dun, N. J., Z. G. Jiang, and N. Mo. 1986. "Tubocurarine suppresses slow calcium-

dependent after-hyperpolarization in guinea-pig inferior mesenteric

ganglion cells." J Physiol 375:499-514.

Eccles, J. C. 1946. "Synaptic potentials of motoneurones." J Neurophysiol 9:87-120. doi:

10.1152/jn.1946.9.2.87.

Eccles, J. C., R. M. Eccles, and A. Lundberg. 1958. "The action potentials of the alpha

motoneurones supplying fast and slow muscles." J Physiol 142 (2):275-91. el Manira, A., J. Tegner, and S. Grillner. 1994. "Calcium-dependent potassium channels

play a critical role for burst termination in the locomotor network in

lamprey." J Neurophysiol 72 (4):1852-61.

Enomoto, K., K. Furuya, T. Maeno, C. Edwards, and T. Oka. 1991. "Oscillating activity of

a calcium-activated K+ channel in normal and cancerous mammary cells in

culture." J Membr Biol 119 (2):133-9.

Faber, E. S., A. J. Delaney, and P. Sah. 2005. "SK channels regulate excitatory synaptic

transmission and plasticity in the lateral amygdala." Nat Neurosci 8

(5):635-41. doi: 10.1038/nn1450.

Faber, E. S., and P. Sah. 2007. "Functions of SK channels in central neurons." Clin Exp

Pharmacol Physiol 34 (10):1077-83. doi: 10.1111/j.1440-

1681.2007.04725.x.

Fioravante, D., and W. G. Regehr. 2011. "Short-term forms of presynaptic plasticity." Curr

Opin Neurobiol 21 (2):269-74. doi: 10.1016/j.conb.2011.02.003.

132

Frigon, A., and S. Rossignol. 2006. "Experiments and models of sensorimotor interactions

during locomotion." Biol Cybern 95 (6):607-27. doi: 10.1007/s00422-006-

0129-x.

Gabbay, H., and A. Lev-Tov. 2004. "Alpha-1 adrenoceptor agonists generate a "fast"

NMDA receptor-independent motor rhythm in the neonatal rat spinal cord."

J Neurophysiol 92 (2):997-1010. doi: 10.1152/jn.00205.2004.

Giessel, A. J., and B. L. Sabatini. 2010. "M1 muscarinic receptors boost synaptic potentials

and calcium influx in dendritic spines by inhibiting postsynaptic SK

channels." Neuron 68 (5):936-47. doi: 10.1016/j.neuron.2010.09.004.

Gonzalez-Islas, C., and P. Wenner. 2006. "Spontaneous network activity in the embryonic

spinal cord regulates AMPAergic and GABAergic synaptic strength."

Neuron 49 (4):563-75. doi: 10.1016/j.neuron.2006.01.017.

Grillner, S. 1975. "Locomotion in vertebrates: central mechanisms and reflex interaction."

Physiol Rev 55 (2):247-304. doi: 10.1152/physrev.1975.55.2.247.

Grillner, S. 1996. "Neural networks for vertebrate locomotion." Sci Am 274 (1):64-9.

Grillner, S. 2003. "The motor infrastructure: from ion channels to neuronal networks." Nat

Rev Neurosci 4 (7):573-86. doi: 10.1038/nrn1137.

Grillner, S., and B. Robertson. 2015. "The basal ganglia downstream control of brainstem

motor centres--an evolutionarily conserved strategy." Curr Opin Neurobiol

33:47-52. doi: 10.1016/j.conb.2015.01.019.

Grunnet, M., T. Jespersen, and J. F. Perrier. 2004. "5-HT1A receptors modulate small-

conductance Ca2+-activated K+ channels." J Neurosci Res 78 (6):845-54.

doi: 10.1002/jnr.20318.

133

Guertin, P. A. 2012. "Central pattern generator for locomotion: anatomical, physiological,

and pathophysiological considerations." Front Neurol 3:183. doi:

10.3389/fneur.2012.00183.

Han, P., S. T. Nakanishi, M. A. Tran, and P. J. Whelan. 2007. "Dopaminergic modulation

of spinal neuronal excitability." J Neurosci 27 (48):13192-204. doi:

10.1523/JNEUROSCI.1279-07.2007.

Heckman, C. J., M. A. Gorassini, and D. J. Bennett. 2005. "Persistent inward currents in

motoneuron dendrites: implications for motor output." Muscle Nerve 31

(2):135-56. doi: 10.1002/mus.20261.

Heckman, C. J., J. J. Kuo, and M. D. Johnson. 2004. "Synaptic integration in motoneurons

with hyper-excitable dendrites." Can J Physiol Pharmacol 82 (8-9):549-

55. doi: 10.1139/y04-046.

Heckman, C. J., and R. H. Lee. 1999. "Synaptic integration in bistable motoneurons." Prog

Brain Res 123:49-56.

Hellstrom, J., A. L. Oliveira, B. Meister, and S. Cullheim. 2003. "Large cholinergic nerve

terminals on subsets of motoneurons and their relation to muscarinic

receptor type 2." J Comp Neurol 460 (4):476-86. doi: 10.1002/cne.10648.

Higley, M. J., and B. L. Sabatini. 2010. "Competitive regulation of synaptic Ca2+ influx

by D2 dopamine and A2A adenosine receptors." Nat Neurosci 13 (8):958-

66. doi: 10.1038/nn.2592.

Hill, R., T. Matsushima, J. Schotland, and S. Grillner. 1992. "Apamin blocks the slow AHP

in lamprey and delays termination of locomotor bursts." Neuroreport 3

(10):943-5.

134

Hille, Bertil. 2001. Ion Channels of Excitable Membranes third ed. Sunderland, MA:

Sinauer.

Hinckley, C. A., and L. Ziskind-Conhaim. 2006. "Electrical coupling between locomotor-

related excitatory interneurons in the mammalian spinal cord." J Neurosci

26 (33):8477-83. doi: 10.1523/JNEUROSCI.0395-06.2006.

Hougaard, C., B. L. Eriksen, S. Jorgensen, T. H. Johansen, T. Dyhring, L. S. Madsen, D.

Strobaek, and P. Christophersen. 2007. "Selective positive modulation of

the SK3 and SK2 subtypes of small conductance Ca2+-activated K+

channels." Br J Pharmacol 151 (5):655-65. doi: 10.1038/sj.bjp.0707281.

Hounsgaard, J., H. Hultborn, B. Jespersen, and O. Kiehn. 1988. "Bistability of alpha-

motoneurones in the decerebrate cat and in the acute spinal cat after

intravenous 5-hydroxytryptophan." J Physiol 405:345-67.

Hounsgaard, J., and O. Kiehn. 1985. "Ca++ dependent bistability induced by serotonin in

spinal motoneurons." Exp Brain Res 57 (2):422-5.

Hounsgaard, J., and I. Mintz. 1988. "Calcium conductance and firing properties of spinal

motoneurones in the turtle." J Physiol 398:591-603.

Hunt, C. C. 1951. "The reflex activity of mammalian small-nerve fibres." J Physiol 115

(4):456-69.

Hunt, C. C., and M. Kuno. 1959. "Properties of spinal interneurones." J Physiol 147:346-

63.

Ishii, T. M., J. Maylie, and J. P. Adelman. 1997. "Determinants of apamin and d-

tubocurarine block in SK potassium channels." J Biol Chem 272

(37):23195-200.

135

Izhikevich, E. M., N. S. Desai, E. C. Walcott, and F. C. Hoppensteadt. 2003. "Bursts as a

unit of neural information: selective communication via resonance." Trends

Neurosci 26 (3):161-7. doi: 10.1016/S0166-2236(03)00034-1.

Izhikevich, Eugene M. 1999. "Neural excitability, spiking and bursting." International

Journal of Bifurcation and Chaos 10 (6):1171-1266.

Jiang, M. C., S. M. Elbasiouny, W. F. Collins, 3rd, and C. J. Heckman. 2015. "The

transformation of synaptic to system plasticity in motor output from the

sacral cord of the adult mouse." J Neurophysiol 114 (3):1987-2004. doi:

10.1152/jn.00337.2015.

Jiang, M. C., and C. J. Heckman. 2006. "In vitro sacral cord preparation and motoneuron

recording from adult mice." J Neurosci Methods 156 (1-2):31-6. doi:

10.1016/j.jneumeth.2006.02.002.

Jiang, M., J. E. Schuster, R. Fu, T. Siddique, and C. J. Heckman. 2009. "Progressive

changes in synaptic inputs to motoneurons in adult sacral spinal cord of a

mouse model of amyotrophic lateral sclerosis." J Neurosci 29 (48):15031-

8. doi: 10.1523/JNEUROSCI.0574-09.2009.

Jiang, Z., K. P. Carlin, and R. M. Brownstone. 1999. "An in vitro functionally mature

mouse spinal cord preparation for the study of spinal motor networks."

Brain Res 816 (2):493-9.

Johnson, I. P., and T. A. Sears. 1988. "Ultrastructure of interneurons within motor nuclei

of the thoracic region of the spinal cord of the adult cat." J Anat 161:171-

85.

136

Johnson, S. M., J. C. Smith, and J. L. Feldman. 1996. "Modulation of respiratory rhythm

in vitro: role of Gi/o protein-mediated mechanisms." J Appl Physiol (1985)

80 (6):2120-33. doi: 10.1152/jappl.1996.80.6.2120.

Jordan, L. M. 1983. "Factors determining motoneuron rhythmicity during fictive

locomotion." Symp Soc Exp Biol 37:423-44.

Jordan, L. M., J. Liu, P. B. Hedlund, T. Akay, and K. G. Pearson. 2008. "Descending

command systems for the initiation of locomotion in mammals." Brain Res

Rev 57 (1):183-91. doi: 10.1016/j.brainresrev.2007.07.019.

Kaada, B. R. 1950. "Site of action of myanesin in the central nervous system." J

Neurophysiol 13 (1):89-104.

Kaczmarek, Leonard K., and Irwin B. Levitan. 1986. Neuromodulation: The Biochemical

Control of Neuronal Excitability. New York: Oxford University Press.

Katz, B., and R. Miledi. 1965. "The Effect of Calcium on Acetylcholine Release from

Motor Nerve Terminals." Proc R Soc Lond B Biol Sci 161:496-503.

Khawaled, R., A. Bruening-Wright, J. P. Adelman, and J. Maylie. 1999. "Bicuculline block

of small-conductance calcium-activated potassium channels." Pflugers

Arch 438 (3):314-21. doi: 10.1007/s004240050915.

Kiehn, O. 2006. "Locomotor circuits in the mammalian spinal cord." Annu Rev Neurosci

29:279-306. doi: 10.1146/annurev.neuro.29.051605.112910.

Kiehn, O. 2011. "Development and functional organization of spinal locomotor circuits."

Curr Opin Neurobiol 21 (1):100-9. doi: 10.1016/j.conb.2010.09.004.

Kiehn, O. 2016. "Decoding the organization of spinal circuits that control locomotion."

Nat Rev Neurosci 17 (4):224-38. doi: 10.1038/nrn.2016.9.

137

Kiehn, O., and O. Kjaerulff. 1998. "Distribution of central pattern generators for rhythmic

motor outputs in the spinal cord of limbed vertebrates." Ann N Y Acad Sci

860:110-29.

Kiehn, O., and M. C. Tresch. 2002. "Gap junctions and motor behavior." Trends Neurosci

25 (2):108-15.

Kim, L. H., S. Sharma, S. A. Sharples, K. A. Mayr, C. H. T. Kwok, and P. J. Whelan. 2017.

"Integration of Descending Command Systems for the Generation of

Context-Specific Locomotor Behaviors." Front Neurosci 11:581. doi:

10.3389/fnins.2017.00581.

King, A. E., J. A. Lopez-Garcia, and M. Cumberbatch. 1992. "Antagonism of synaptic

potentials in ventral horn neurones by 6-cyano-7-nitroquinoxaline-2,3-

dione: a study in the rat spinal cord in vitro." Br J Pharmacol 107 (2):375-

81.

Kleiman-Weiner, M., M. P. Beenhakker, W. A. Segal, and J. R. Huguenard. 2009.

"Synergistic roles of GABAA receptors and SK channels in regulating

thalamocortical oscillations." J Neurophysiol 102 (1):203-13. doi:

10.1152/jn.91158.2008.

Kudo, N., and T. Yamada. 1987. "N-methyl-D,L-aspartate-induced locomotor activity in a

spinal cord-hindlimb muscles preparation of the newborn rat studied in

vitro." Neurosci Lett 75 (1):43-8.

Kuze, B., K. Matsuyama, T. Matsui, H. Miyata, and S. Mori. 1999. "Segment-specific

branching patterns of single vestibulospinal tract axons arising from the

138

lateral vestibular nucleus in the cat: A PHA-L tracing study." J Comp

Neurol 414 (1):80-96.

Lanuza, G. M., S. Gosgnach, A. Pierani, T. M. Jessell, and M. Goulding. 2004. "Genetic

identification of spinal interneurons that coordinate left-right locomotor

activity necessary for walking movements." Neuron 42 (3):375-86.

Lape, R., and A. Nistri. 2000. "Current and voltage clamp studies of the spike medium

afterhyperpolarization of hypoglossal motoneurons in a rat brain stem slice

preparation." J Neurophysiol 83 (5):2987-95.

Lappin, S. C., T. J. Dale, J. T. Brown, D. J. Trezise, and C. H. Davies. 2005. "Activation

of SK channels inhibits epileptiform bursting in hippocampal CA3

neurons." Brain Res 1065 (1-2):37-46. doi:

10.1016/j.brainres.2005.10.024.

Lee, R. H., and C. J. Heckman. 1998. "Bistability in spinal motoneurons in vivo: systematic

variations in rhythmic firing patterns." J Neurophysiol 80 (2):572-82.

Lee, R. H., and C. J. Heckman. 1999. "Enhancement of bistability in spinal motoneurons

in vivo by the noradrenergic alpha1 agonist methoxamine." J Neurophysiol

81 (5):2164-74. doi: 10.1152/jn.1999.81.5.2164.

Lee, R. H., and C. J. Heckman. 2000. "Adjustable amplification of synaptic input in the

dendrites of spinal motoneurons in vivo." J Neurosci 20 (17):6734-40.

Lev-Tov, A., and M. Pinco. 1992. "In vitro studies of prolonged synaptic depression in the

neonatal rat spinal cord." J Physiol 447:149-69.

Lewis, G. N., and W. D. Byblow. 2002. "Altered sensorimotor integration in Parkinson's

disease." Brain 125 (Pt 9):2089-99.

139

Li, X., and D. J. Bennett. 2007. "Apamin-sensitive calcium-activated potassium currents

(SK) are activated by persistent calcium currents in rat motoneurons." J

Neurophysiol 97 (5):3314-30. doi: 10.1152/jn.01068.2006.

Li, X., K. Murray, P. J. Harvey, E. W. Ballou, and D. J. Bennett. 2007. "Serotonin

facilitates a persistent calcium current in motoneurons of rats with and

without chronic spinal cord injury." J Neurophysiol 97 (2):1236-46. doi:

10.1152/jn.00995.2006.

Li, Yunru, and David J Bennett. 2003. "Persistent Sodium and Calcium Currents Cause

Plateau Potentials in Motoneurons of Chronic Spinal Rats." J Neurophysiol

90 (2):857-869.

Liang, H., T. Bacskai, C. Watson, and G. Paxinos. 2014. "Projections from the lateral

vestibular nucleus to the spinal cord in the mouse." Brain Struct Funct 219

(3):805-15. doi: 10.1007/s00429-013-0536-4.

Liang, H., C. Watson, and G. Paxinos. 2015. "Projections from the oral pontine reticular

nucleus to the spinal cord of the mouse." Neurosci Lett 584:113-8. doi:

10.1016/j.neulet.2014.10.025.

Liang, H., C. Watson, and G. Paxinos. 2016. "Terminations of reticulospinal fibers

originating from the gigantocellular reticular formation in the mouse spinal

cord." Brain Struct Funct 221 (3):1623-33. doi: 10.1007/s00429-015-0993-

Lisman, J. E. 1997. "Bursts as a unit of neural information: making unreliable synapses

reliable." Trends Neurosci 20 (1):38-43. doi: 10.1016/S0166-

2236(96)10070-9.

140

Logsdon, S., A. F. Johnstone, K. Viele, and R. L. Cooper. 2006. "Regulation of synaptic

vesicles pools within motor nerve terminals during short-term facilitation

and neuromodulation." J Appl Physiol (1985) 100 (2):662-71. doi:

10.1152/japplphysiol.00580.2005.

Lombardo, J., and M. A. Harrington. 2016. "Nonreciprocal mechanisms in up- and

downregulation of spinal motoneuron excitability by modulators of

KCNQ/Kv7 channels." J Neurophysiol 116 (5):2114-2124. doi:

10.1152/jn.00446.2016.

Lu, B., Q. Zhang, H. Wang, Y. Wang, M. Nakayama, and D. Ren. 2010. "Extracellular

calcium controls background current and neuronal excitability via an

UNC79-UNC80-NALCN cation channel complex." Neuron 68 (3):488-99.

doi: 10.1016/j.neuron.2010.09.014.

Madison, D. V., B. Lancaster, and R. A. Nicoll. 1987. "Voltage clamp analysis of

cholinergic action in the hippocampus." J Neurosci 7 (3):733-41.

Magee, J. C. 2000. "Dendritic integration of excitatory synaptic input." Nat Rev Neurosci

1 (3):181-90. doi: 10.1038/35044552.

Mahrous, A. A., and S. M. Elbasiouny. 2017a. "Modulation of SK channels regulates

locomotor alternating bursting activity in the functionally-mature spinal

cord." Channels (Austin):0. doi: 10.1080/19336950.2017.1389825.

Mahrous, A. A., and S. M. Elbasiouny. 2017b. "SK channel inhibition mediates the

initiation and amplitude modulation of synchronized burst firing in the

spinal cord." J Neurophysiol 118 (1):161-175. doi: 10.1152/jn.00929.2016.

141

Maingret, F., B. Coste, J. Hao, A. Giamarchi, D. Allen, M. Crest, D. W. Litchfield, J. P.

Adelman, and P. Delmas. 2008. "Neurotransmitter modulation of small-

conductance Ca2+-activated K+ channels by regulation of Ca2+ gating."

Neuron 59 (3):439-49. doi: 10.1016/j.neuron.2008.05.026.

Manuel, M., Y. Li, S. M. Elbasiouny, K. Murray, A. Griener, C. J. Heckman, and D. J.

Bennett. 2012. "NMDA induces persistent inward and outward currents that

cause rhythmic bursting in adult rodent motoneurons." J Neurophysiol 108

(11):2991-8. doi: 10.1152/jn.00518.2012.

Manuel, M., C. Meunier, M. Donnet, and D. Zytnicki. 2006. "The afterhyperpolarization

conductance exerts the same control over the gain and variability of

motoneurone firing in anaesthetized cats." J Physiol 576 (Pt 3):873-86. doi:

10.1113/jphysiol.2006.117002.

Marder, E., T. O'Leary, and S. Shruti. 2014. "Neuromodulation of circuits with variable

parameters: single neurons and small circuits reveal principles of state-

dependent and robust neuromodulation." Annu Rev Neurosci 37:329-46.

doi: 10.1146/annurev-neuro-071013-013958.

Marino, M. J., S. T. Rouse, A. I. Levey, L. T. Potter, and P. J. Conn. 1998. "Activation of

the genetically defined m1 muscarinic receptor potentiates N-methyl-D-

aspartate (NMDA) receptor currents in hippocampal pyramidal cells." Proc

Natl Acad Sci U S A 95 (19):11465-70.

Markram, H., and M. Segal. 1990. "Acetylcholine potentiates responses to N-methyl-D-

aspartate in the rat hippocampus." Neurosci Lett 113 (1):62-5.

142

McNeill, David , Liesbet Quaeghebeur, and Susan Duncan. 2008. "IW—“the man who

lost his body”." In Handbook of Phenomenology and Cognitive Sciences,

edited by Shaun Gallagher and Daniel Schmickin. Dordrecht: Springer.

Meech, R. W. 1978. "Calcium-dependent potassium activation in nervous tissues." Annu

Rev Biophys Bioeng 7:1-18. doi: 10.1146/annurev.bb.07.060178.000245.

Meehan, C. F., L. Grondahl, J. B. Nielsen, and H. Hultborn. 2012. "Fictive locomotion in

the adult decerebrate and spinal mouse in vivo." J Physiol 590 (2):289-300.

doi: 10.1113/jphysiol.2011.214643.

Meehan, C. F., M. Moldovan, S. L. Marklund, K. S. Graffmo, J. B. Nielsen, and H.

Hultborn. 2010. "Intrinsic properties of lumbar motor neurones in the adult

G127insTGGG superoxide dismutase-1 mutant mouse in vivo: evidence for

increased persistent inward currents." Acta Physiol (Oxf) 200 (4):361-76.

doi: 10.1111/j.1748-1716.2010.02188.x.

Miles, G. B., R. Hartley, A. J. Todd, and R. M. Brownstone. 2007. "Spinal cholinergic

interneurons regulate the excitability of motoneurons during locomotion."

Proc Natl Acad Sci U S A 104 (7):2448-53. doi: 10.1073/pnas.0611134104.

Musienko, P. E., P. V. Zelenin, V. F. Lyalka, Y. P. Gerasimenko, G. N. Orlovsky, and T.

G. Deliagina. 2012. "Spinal and supraspinal control of the direction of

stepping during locomotion." J Neurosci 32 (48):17442-53. doi:

10.1523/JNEUROSCI.3757-12.2012.

Nadim, F., and D. Bucher. 2014. "Neuromodulation of neurons and synapses." Curr Opin

Neurobiol 29:48-56. doi: 10.1016/j.conb.2014.05.003.

143

Nanou, E., M. H. Alpert, S. Alford, and A. El Manira. 2013. "Differential regulation of

synaptic transmission by pre- and postsynaptic SK channels in the spinal

locomotor network." J Neurophysiol 109 (12):3051-9. doi:

10.1152/jn.00067.2013.

Negro, F., U. Sukru Yavuz, and D. Farina. 2016. "The human motor neuron pools receive

a dominant slow-varying common synaptic input." J Physiol. doi:

10.1113/JP271748.

Neher, E., and T. Sakaba. 2008. "Multiple roles of calcium ions in the regulation of

neurotransmitter release." Neuron 59 (6):861-72. doi:

10.1016/j.neuron.2008.08.019.

Ngo-Anh, T. J., B. L. Bloodgood, M. Lin, B. L. Sabatini, J. Maylie, and J. P. Adelman.

2005. "SK channels and NMDA receptors form a Ca2+-mediated feedback

loop in dendritic spines." Nat Neurosci 8 (5):642-9. doi: 10.1038/nn1449.

O'Donovan, M. J., A. Bonnot, G. Z. Mentis, N. Chub, A. Pujala, and F. J. Alvarez. 2010.

"Mechanisms of excitation of spinal networks by stimulation of the ventral

roots." Ann N Y Acad Sci 1198:63-71. doi: 10.1111/j.1749-

6632.2010.05535.x.

Perret, C. 1983. "Centrally generated pattern of motoneuron activity during locomotion in

the cat." Symp Soc Exp Biol 37:405-22.

Pflieger, J. F., F. Clarac, and L. Vinay. 2002. "Picrotoxin and bicuculline have different

effects on lumbar spinal networks and motoneurons in the neonatal rat."

Brain Res 935 (1-2):81-6.

144

Pinco, M., and A. Lev-Tov. 1993. "Modulation of monosynaptic excitation in the neonatal

rat spinal cord." J Neurophysiol 70 (3):1151-8. doi:

10.1152/jn.1993.70.3.1151.

Ransom, B. R., J. L. Barker, and P. G. Nelson. 1975. "Two mechanisms for poststimulus

hyperpolarisations in cultured mammalian neurones." Nature 256

(5516):424-5.

Rash, J. E., T. Yasumura, K. G. Davidson, C. S. Furman, F. E. Dudek, and J. I. Nagy. 2001.

"Identification of cells expressing Cx43, Cx30, Cx26, Cx32 and Cx36 in

gap junctions of rat brain and spinal cord." Cell Commun Adhes 8 (4-

6):315-20.

Regehr, W. G. 2012. "Short-term presynaptic plasticity." Cold Spring Harb Perspect Biol

4 (7):a005702. doi: 10.1101/cshperspect.a005702.

Riddle, C. N., S. A. Edgley, and S. N. Baker. 2009. "Direct and indirect connections with

upper limb motoneurons from the primate reticulospinal tract." J Neurosci

29 (15):4993-9. doi: 10.1523/JNEUROSCI.3720-08.2009.

Rossignol, S., R. Dubuc, and J. P. Gossard. 2006. "Dynamic sensorimotor interactions in

locomotion." Physiol Rev 86 (1):89-154. doi:

10.1152/physrev.00028.2005.

Rouchet, N., O. Waroux, C. Lamy, L. Massotte, J. Scuvee-Moreau, J. F. Liegeois, and V.

Seutin. 2008. "SK channel blockade promotes burst firing in dorsal raphe

serotonergic neurons." Eur J Neurosci 28 (6):1108-15. doi: 10.1111/j.1460-

9568.2008.06430.x.

145

Rozental, R., M. Srinivas, and D. C. Spray. 2001. "How to close a gap junction channel.

Efficacies and potencies of uncoupling agents." Methods Mol Biol 154:447-

76.

Salameh, A., and S. Dhein. 2005. "Pharmacology of gap junctions. New pharmacological

targets for treatment of arrhythmia, seizure and cancer?" Biochim Biophys

Acta 1719 (1-2):36-58. doi: 10.1016/j.bbamem.2005.09.007.

Schmidt, B. J. 1994. "Afterhyperpolarization modulation in lumbar motoneurons during

locomotor-like rhythmic activity in the neonatal rat spinal cord in vitro."

Exp Brain Res 99 (2):214-22.

Schumacher, M. A., A. F. Rivard, H. P. Bachinger, and J. P. Adelman. 2001. "Structure of

the gating domain of a Ca2+-activated K+ channel complexed with

Ca2+/calmodulin." Nature 410 (6832):1120-4. doi: 10.1038/35074145.

Sears, T. A., and D. Stagg. 1976. "Short-term synchronization of intercostal motoneurone

activity." J Physiol 263 (3):357-81.

Sharples, S. A., and P. J. Whelan. 2017. "Modulation of Rhythmic Activity in Mammalian

Spinal Networks Is Dependent on Excitability State." eNeuro 4 (1). doi:

10.1523/ENEURO.0368-16.2017.

Smith, J. C., and J. L. Feldman. 1987. "In vitro brainstem-spinal cord preparations for study

of motor systems for mammalian respiration and locomotion." J Neurosci

Methods 21 (2-4):321-33.

Spruston, N. 2008. "Pyramidal neurons: dendritic structure and synaptic integration." Nat

Rev Neurosci 9 (3):206-21. doi: 10.1038/nrn2286.

146

Standen, N. B., and P. R. Stanfield. 1978. "A potential- and time-dependent blockade of

inward rectification in frog skeletal muscle fibres by barium and strontium

ions." J Physiol 280:169-91.

Steeves, J. D., G. N. Sholomenko, and D. M. Webster. 1987. "Stimulation of the

pontomedullary reticular formation initiates locomotion in decerebrate

birds." Brain Res 401 (2):205-12.

Streit, J. 1993. "Regular oscillations of synaptic activity in spinal networks in vitro." J

Neurophysiol 70 (3):871-8.

Sun, X., Y. Zhao, and M. E. Wolf. 2005. "Dopamine receptor stimulation modulates

AMPA receptor synaptic insertion in prefrontal cortex neurons." J Neurosci

25 (32):7342-51. doi: 10.1523/JNEUROSCI.4603-04.2005.

Swash, M. 2018. "Sensorimotor integration is problematic in amyotrophic lateral

sclerosis." Clin Neurophysiol. doi: 10.1016/j.clinph.2018.01.005.

Talley, E. M., N. N. Sadr, and D. A. Bayliss. 1997. "Postnatal development of serotonergic

innervation, 5-HT1A receptor expression, and 5-HT responses in rat

motoneurons." J Neurosci 17 (11):4473-85.

Tegner, J., and S. Grillner. 1999. "Interactive effects of the GABABergic modulation of

calcium channels and calcium-dependent potassium channels in lamprey."

J Neurophysiol 81 (3):1318-29. doi: 10.1152/jn.1999.81.3.1318.

Tovar, K. R., B. J. Maher, and G. L. Westbrook. 2009. "Direct actions of carbenoxolone

on synaptic transmission and neuronal membrane properties." J

Neurophysiol 102 (2):974-8. doi: 10.1152/jn.00060.2009.

147

Turkin, V. V., D. O'Neill, R. Jung, A. Iarkov, and T. M. Hamm. 2010. "Characteristics and

organization of discharge properties in rat hindlimb motoneurons." J

Neurophysiol 104 (3):1549-65. doi: 10.1152/jn.00379.2010.

Vessey, J. P., M. R. Lalonde, H. A. Mizan, N. C. Welch, M. E. Kelly, and S. Barnes. 2004.

"Carbenoxolone inhibition of voltage-gated Ca channels and synaptic

transmission in the retina." J Neurophysiol 92 (2):1252-6. doi:

10.1152/jn.00148.2004.

Viana, F., D. A. Bayliss, and A. J. Berger. 1993. "Multiple potassium conductances and

their role in action potential repolarization and repetitive firing behavior of

neonatal rat hypoglossal motoneurons." J Neurophysiol 69 (6):2150-63.

Wagner, E. J., O. K. Ronnekleiv, and M. J. Kelly. 2001. "The noradrenergic inhibition of

an apamin-sensitive, small-conductance Ca2+-activated K+ channel in

hypothalamic gamma-aminobutyric acid neurons: pharmacology, estrogen

sensitivity, and relevance to the control of the reproductive axis." J

Pharmacol Exp Ther 299 (1):21-30.

Wallen, P., J. T. Buchanan, S. Grillner, R. H. Hill, J. Christenson, and T. Hokfelt. 1989.

"Effects of 5-hydroxytryptamine on the afterhyperpolarization, spike

frequency regulation, and oscillatory membrane properties in lamprey

spinal cord neurons." J Neurophysiol 61 (4):759-68.

Wallen, P., and S. Grillner. 1987. "N-methyl-D-aspartate receptor-induced, inherent

oscillatory activity in neurons active during fictive locomotion in the

lamprey." J Neurosci 7 (9):2745-55.

148

Whelan, P., A. Bonnot, and M. J. O'Donovan. 2000. "Properties of rhythmic activity

generated by the isolated spinal cord of the neonatal mouse." J

Neurophysiol 84 (6):2821-33.

Whelan, P. J. 1996. "Control of locomotion in the decerebrate cat." Prog Neurobiol 49

(5):481-515.

White, S. R., and S. J. Fung. 1989. "Serotonin depolarizes cat spinal motoneurons in situ

and decreases motoneuron afterhyperpolarizing potentials." Brain Res 502

(2):205-13.

Wikstrom, M. A., and A. El Manira. 1998. "Calcium influx through N- and P/Q-type

channels activate apamin-sensitive calcium-dependent potassium channels

generating the late afterhyperpolarization in lamprey spinal neurons." Eur

J Neurosci 10 (4):1528-32.

Wilson, J. M., J. Rempel, and R. M. Brownstone. 2004. "Postnatal development of

cholinergic synapses on mouse spinal motoneurons." J Comp Neurol 474

(1):13-23. doi: 10.1002/cne.20089.

Witham, C. L., K. M. Fisher, S. A. Edgley, and S. N. Baker. 2016. "Corticospinal Inputs

to Primate Motoneurons Innervating the Forelimb from Two Divisions of

Primary Motor Cortex and Area 3a." J Neurosci 36 (9):2605-16. doi:

10.1523/JNEUROSCI.4055-15.2016.

Wolf, E., F. Y. Zhao, and A. Roberts. 1998. "Non-linear summation of excitatory synaptic

inputs to small neurones: a case study in spinal motoneurones of the young

Xenopus tadpole." J Physiol 511 ( Pt 3):871-86.

149

Yamamoto, T., E. L. Hertzberg, and J. I. Nagy. 1991. "Subsurface cisterns in alpha-

motoneurons of the rat and cat: immunohistochemical detection with

antibodies against connexin32." Synapse 8 (2):119-36. doi:

10.1002/syn.890080206.

Zagoraiou, L., T. Akay, J. F. Martin, R. M. Brownstone, T. M. Jessell, and G. B. Miles.

2009. "A cluster of cholinergic premotor interneurons modulates mouse

locomotor activity." Neuron 64 (5):645-62. doi:

10.1016/j.neuron.2009.10.017.

Zhao, S., A. F. Sheibanie, M. Oh, P. Rabbah, and F. Nadim. 2011. "Peptide

neuromodulation of synaptic dynamics in an oscillatory network." J

Neurosci 31 (39):13991-4004. doi: 10.1523/JNEUROSCI.3624-11.2011.

Zucker, R. S., and W. G. Regehr. 2002. "Short-term synaptic plasticity." Annu Rev Physiol

64:355-405. doi: 10.1146/annurev.physiol.64.092501.114547.

150