MODULAR ORGANIZATION OF THE OLIVO-CORTICO-NUCLEAR SYSTEM

Get Complete Project Material File(s) Now! »

Principal neurons excitability and their electrophysiological properties

The intrinsic electrophysiological properties of the nuclear principal neurons have been studied in a variety of different preparations (for review, (Sastry et al., 1997)), including isolated brain stem-cerebellum slices (Llinas and Muhlethaler, 1988), cerebellar slices (Jahnsen, 1986b, a), organotypic cultures (Mouginot and Gahwiler, 1995) or dissociated neurons (Raman et al., 2000).

Spontaneous firing

The nuclear principal neurons, like the Purkinje cells, discharge spontaneous action potentials at rates close to 30 Hz (Thach, 1968; Jahnsen, 1986a; Llinas and Muhlethaler, 1988; Mouginot and Gahwiler, 1995; Aizenman and Linden, 1999; Raman et al., 2000). The firing rate can reach 300 Hz when a depolarizing current is injected to the neurons (Jahnsen, 1986a). Spontaneous firing is generated and maintained by the intrinsic properties of the nuclear neurons, as suggested by studies on dissociated neurons (Raman et al., 2000) and after blockade of all synaptic inputs (Mouginot and Gahwiler, 1995; Aizenman and Linden, 1999). Expression of a variety of different ionic channels allow the generation of pace-maker like currents to maintain the cycle of firing (Jahnsen, 1986a, b; Llinas and Muhlethaler, 1988; Sastry et al., 1997; Raman et al., 2000; Alvina and Khodakhah, 2008; Ovsepian et al., 2013).
The firing rate of the nuclear neurons is modulated by both their excitatory and inhibitory synaptic inputs (Mouginot and Gahwiler, 1995; Zhang et al., 2004). Consistently, the principal neuron discharge rates are modulated during motor behaviors (Thach, 1968; Armstrong and Edgley, 1984). In vivo, the influence of the inhibitory inputs may be more complex than a simple decrease inthe nuclear neuron firing rate. When McDevitt and colleagues (Mcdevitt et al., 1987) recorded simultaneously from Purkinje cells and their related nuclear neuron, their firing rates did not always vary inversely suggesting that the Purkinje cells and the principal neurons do not have necessarily a reciprocal relationship. This result indicates that a single Purkinje cell does not dominate the discharge activity of the principal neurons it contacts, and implies more heterogeneous and complex interactions with others afferents to the cerebellar nuclei. Intrinsic electrophysiological properties of nuclear neurons, such as their ability to generate rebound firing, may also explain this nonreciprocity.

Characteristics of the rebound discharge

In addition to their spontaneous firing, a characteristic of principal nuclear neurons is their ability to show pronounced rebound depolarization, often accompanied by a high-frequency burst of spikes or just by a more prolonged period of increased firing, immediately after an hyperpolarization period (Jahnsen, 1986a, b; Llinas and Muhlethaler, 1988; Aizenman and Linden, 1999; Pedroarena, 2010).The rebound depolarization is dependent on the membrane potential of the cell and on the duration and amplitude of the hyperpolarizing pulse: it is more prominent following longer and deeper hyperpolarization steps, or when the holding membrane potential is between -60 and -70 mV(Aizenman and Linden, 1999; Pedroarena, 2010). The rebound is readily induced by hyperpolarizing current injection at the soma, and can also be evoked by more physiological manipulations like the local uncaging of GABA or the electrical stimulation of hyperpolarizing IPSPs originating from Purkinje cell inputs (Llinas and Muhlethaler, 1988; Aizenman et al., 1998; Aizenman and Linden, 1999; Alvina et al., 2009). Interestingly, high-frequency trains of IPSPs were moreeffective at inducing a rebound discharge than a single IPSP or hyperpolarizing current pulses of similar amplitude and duration. This could be explained by the wide distribution of the Purkinje cell inputs on the dendritic arbor of the principal neurons. Distal inhibitory inputs will produce effective dendritic hyperpolarization which is most likely not achieved by hyperpolarization delivered from the soma. Diverse dynamic parameters of these synapses (as discussed in the section 1.3.1.1.) may also be involved in the efficiency of synaptic hyperpolarization to trigger rebounds. During the rebound, the increased excitability and associated spike burst induce large intracellular calcium transients in the principal neurons (Aizenman et al., 1998; Zhang et al., 2004; Schneider et al., 2013). By this way, inhibitory Purkinje cell inputs can drive postsynaptic excitation and calcium entry in the nuclear neurons, which mechanism may be the basis for plasticity at the Purkinje cell to principal neurons synapse (Aizenman et al., 1998)7.
Several attempts have been made to link the different firing and bursting phenotypes of nuclear cells to their morphological characteristics, with no clear success (Aizenman et al., 2003). Bursting phenotypes seems to be specific of principal neurons (Czubayko et al., 2001), however other cell types in the cerebellar nuclei also exhibit spontaneous firing (Uusisaari and Knopfel, 2011)8.

Does the rebound discharge exist in vivo?

Despite the biophysical robustness of rebound firing in the principal neurons, its prevalencein response to physiological stimuli is still debated. After sensory stimulation, the responses of nuclear neurons in vivo do not present the expected rebound discharge after inhibition periods, but rather consist in a sequence of early excitation – intermediate inhibition – late increase in firing rate (Armstrong et al., 1975; Rowland and Jaeger, 2005). The late excitation, which could perhaps be the consequence of a rebound depolarization after the intermediate inhibition from the Purkinje cell inputs, was however the least reliable component of the response (Rowland and Jaeger, 2005). As previously discussed in section 1.3.1.3., this late excitation may be due in part to the nuclear disinhibition resulting from pauses in the Purkinje cell firing (Ito et al., 1970; Armstrong et al., 1975; Witter et al., 2013). Moreover, direct and strong activation of the Purkinje cell inputs upon a nuclear neuron in vivo does not result in rebound firing in the majority of the recorded cells (Alvina et al., 2008; Chaumont et al., 2013). Occurrence of the rebound firing also differs between the experimental paradigm used (for reviews of protocols, Alvina et al., 2008; De Zeeuw et al., 2011). As the rebound discharge phenomenon is entirely based on the occurrence of an inhibition period and on its efficiency to activate several conductances like the T-Type channels, the main open 􃻞􃻜􃻌 􃻌 question is the extent to which in vivo synaptic inhibition can efficiently recruit T-type currents and lead to observable rebound firing. Several discrepancies have been pointed out. First, in vivo activation of GABAA receptors cannot hyperpolarize cells further than the membrane reversal potential for chloride ions (ECl). This later has been repeatedly measured to be near -75 mV in the principal neurons (Jahnsen, 1986b; Aizenman and Linden, 1999; Alvina et al., 2008; Zheng and Raman, 2009).
The known voltage-dependence of T-type channels makes it unlikely that they recover substantially even at ECl under those conditions (Cain and Snutch, 2010). Therefore, only the neurons which are hyperpolarized (by current injections for example) beyond ECl will generate a rebound burst. Modelling studies suggest that the ECl value for nuclear neurons was in a critical region where small changes on the order of 5–10 mV only would have a strong influence on rebound strength (Steuber et al., 2011). Secondly, Zheng et al. (Zheng and Raman, 2009) reported that high- frequency IPSPs evoke little post-inhibitory current through T-type channels, without evidence of T-type mediated large and brief calcium transients even in dendrites during imaging studies. Additionally, the late elevation in the firing rate observed after strong inhibition is very slow and persists for several hundred milliseconds (Armstrong et al., 1975; Rowland and Jaeger, 2005; Alvina et al., 2009; Zheng and Raman, 2009), far outlasting the duration of T-type current (Cain and Snutch, 2010). An interesting alternative hypothesis is that the T-type channel currents mediating the rebound firing are highly modulated in vivo and act synergically with other conductances. In the cerebellum, complex interactions between T-type channels and other conductances influence the net effect of T-type channels on neuronal excitability (Engbers et al., 2013). Window current, which relies on the few channels that are stochastically opened at approximately −60 mV thus creating a steady-state conductance (Cain and Snutch, 2010), may also play an additional role in the generation of rebound firing by modulating the basal intracellular calcium level and therefore act on intracellular machinery to modulate the neuronal excitability (Engbers et al., 2012; Engbers et al., 2013).
The post-inhibitory propensity to burst has been shown to increase with the strength of stimulation of Purkinje afferents (Aizenman and Linden, 1999; Tadayonnejad et al., 2009), but as the depth of hyperpolarization won’t change significantly with increased synaptic inhibition, this raises the possibility that stronger stimulations facilitate the rebound discharge by engaging additional mechanisms. First inhibition may activate other ionic conductances. Indeed, several channels expressed by the nuclear neurons have been shown to play a role in the rebound discharge (highvoltage- activated calcium current (HVA) (Muri and Knopfel, 1994; Gauck et al., 2001), potassium channels (Molineux et al., 2008), hyperpolarization-activated cyclic-nucleotide (HCN) (Engbers et al., 2011)). Moreover, interactions between diverse background synaptic conductances in the nuclear neurons have been found to exert a great influence on the appearance of the rebound phenotype and its strength (Steuber et al., 2011). Second, neuromodulatory systems may also have a strong influence on the rebound bursting phenotypes in vivo (Gould et al., 1997; Saitow et al., 2009; Murano et al., 2011; Schneider et al., 2013), but have been poorly studied. Modulations in the ECl itself, which may become more negative under certain conditions such as modulations in the expression of the chloride co-transporters NKCC1 and KCC2 known to control the ECl in neurons (Rivera et al., 1999; Banke and McBain, 2006), could also explain the variable occurrence of rebound discharge in vivo.

Principal neuron output activity: Rate coding versus temporal coding

To predict the nuclear neuron output spiking activity, one needs to understand their mechanisms of synaptic integration and their interactions with the intrinsic membrane properties of these principal neurons. Understanding which information is transmitted to the nuclear neurons and how they encode it into output activity is fundamental (Person and Raman, 2012b). Neuronal coding can be broadly categorized as “rate coding” or “temporal coding”. The number or rate of spikes in a particular time window carries information in the rate coding mode, whereas for temporal coding, the information is represented by the timing of individual spikes or bursts of spikes. Both rate coding
and temporal coding seem to occur in the cerebellar nuclei (Steuber and Jaeger, 2013). Rate coding in the nuclear neurons would imply that the relative firing rates coming from excitatory and inhibitory inputs determine the output spike frequency of the cerebellar nuclear cells. Indeed, dynamic clamp studies have suggested that the rate of Purkinje cell inputs results in a rate code in the nuclear cells (Gauck and Jaeger, 2000, 2003) and that increases in excitatory mossy fiber input rates or decreases in inhibitory Purkinje cell input rates are translated into smooth increases of the nuclear spike rates (Steuber et al., 2011). Modulation of Purkinje cell spike rates has been proposed to encode information in the cerebellar nuclei by linear summation more efficiently than pauses (Walter and Khodakhah, 2009).
The ability of nuclear neurons to exhibit rebound firing make them good candidates for temporal coding, as the rebound discharge may create well-timed spike burst following certain patterns of Purkinje cell inputs. Rebound firing has been incorporated into recent theories of cerebellar function (Kistler and De Zeeuw, 2003; Wetmore et al., 2008) and several functional roles have been assigned to it, such as timing and encoding information in association with plasticity mechanisms both at cortical and nuclear levels (Aizenman et al., 1998; Kistler and De Zeeuw, 2003; Pugh and Raman, 2006; Wetmore et al., 2008).
One of those models involving rebound discharge was described by Wetmore and colleagues (Wetmore et al., 2008) who defined a model of cerebellar motor memory and learning, called the “lock and key” hypothesis. In this model, the mechanisms of plasticity in the cerebellar cortex were necessary but not sufficient to generate a desired cerebellar output. The cortical output activity arising from those operations has to result in the appropriate temporal patterns to elicit rebound in the downstream nuclear cells, i.e. an increase (to hyperpolarize the neurons) followed by a decrease in the Purkinje cell spike rate (releasing the inhibition and allowing the neurons to fire). According to Wetmore et al., the cortical temporal spike patterns represent a ‘‘key’’, while the temporal filtering properties of the nuclear neurons that determine whether or not a rebound response occurs is a ‘‘lock’’. Therefore, successful learning shapes neural activity to match a temporal filter that prevents expression of stored but inappropriate motor responses.

Table of contents :

INTRODUCTION
CHAPTER 1: THE MAIN LOOP THROUGH THE CEREBELLAR NUCLEI: PRINCIPAL NEURONS AND MOTOR CONTROL 
1.1. SEGMENTATION AND TOPOGRAPHY OF THE CEREBELLUM
1.1.1. Cerebellar cortex
1.1.2. Cerebellar nuclei
1.2. SENSORI-MOTOR CIRCUITRY IN THE CEREBELLUM: MOSSY FIBERS SYSTEM AS EXTRACEREBELLAR INPUTS – CEREBELLAR
NUCLEI AS FINAL CEREBELLAR OUTPUT
1.2.1. Mossy fibers system: sensori-motor inputs
1.2.1.1. Topography and somatotopy in the cerebellar cortex
1.2.1.2. Mossy fibers innervation of the cerebellar nuclei
1.2.2. The cerebellar nuclei: the cerebellar output to control motor function
1.3. CN PRINCIPAL NEURONS: A KEY SYNAPTIC INTEGRATOR FOR THE MOTOR FUNCTION
1.3.1. Synaptic inputs in principal neurons: Excitation versus Inhibition
1.3.1.1. Excitatory inputs
1.3.1.2. Purkinje cell inhibitory inputs
1.3.1.3. Sequential integration of excitatory and inhibitory inputs
1.3.2. Principal neurons excitability and their electrophysiological properties
1.3.2.1. Spontaneous firing
1.3.2.2. Characteristics of the rebound discharge
1.3.2.3. Does the rebound discharge exist in vivo?
1.3.3. Principal neuron output activity: Rate coding versus temporal coding
CHAPTER 2: THE OLIVO-CORTICO-NUCLEAR LOOP
2.1. INFERIOR OLIVE PROJECTION TO THE CEREBELLAR CORTEX
2.1.1. Olivo-cortical innervation
2.1.1.1. One-to-one innervation of the Purkinje cells
2.1.1.2. Climbing fiber input elicit complex spike
2.1.2. Parasagittal segmentation of the climbing fibers inputs
2.2. THE CEREBELLO-OLIVARY FEEDBACK LOOP
2.2.1. The nucleo-olivary cells
2.2.2. Cerebellar control of the inferior olive activity
2.2.2.1. Structural organization of the inferior olive: electrotonic coupling and subthreshold oscillations
2.2.2.2. Inhibitory action of the nucleo-olivary neurons on the inferior olive physiology
2.3. MODULAR ORGANIZATION OF THE OLIVO-CORTICO-NUCLEAR SYSTEM
2.3.1. Olivary-nuclear innervation
2.3.2. The olivo-cerebellar module
2.3.2.1. The olivo-cerebellar module: a functional unit to control movement?
2.3.2.2. Mossy fibers inputs with respect to the cerebellar modules
2.3.3. Functional impact of the modular organization of the olivo-cerebellar system
2.3.3.1. Nuclear neurons activity within the modules
2.3.3.2. Homeostasis of the olivary, cortical and nuclear activities in the cerebellar feedback loops
2.4. CLIMBING FIBERS GIVE INSTRUCTIONS: SUPERVISED LEARNING IN THE CEREBELLUM
2.4.1. Synaptic plasticity and error signaling in the cerebellar cortex by climbing fiber inputs
2.4.2. Synaptic plasticity in the cerebellar nuclei: another degree of freedom for olivo-cerebellar mediated motor learning? . 34
CHAPTER 3: INHIBITORY NEURONS OF THE CEREBELLAR NUCLEI, A THIRD NUCLEAR CIRCUIT
3.1. INHIBITORY NEURONS OF THE CEREBELLAR NUCLEI: A HETEROGENEOUS POPULATION
3.1.1. Evidence for the presence of a third nuclear cell type and the question of their neurotransmitter contents
3.1.2. Electrophysiological properties of the glycinergic neurons
3.2. EVIDENCE FOR NON-PURKINJE CELLS INHIBITORY TRANSMISSION IN THE PRINCIPAL NEURONS
3.2.1. GABAA and glycine receptors: common features and specific characteristics
3.2.2. Interaction between GABAergic and glycinergic transmission: the case of mixed transmission
3.2.3. Searching for a glycinergic component of synaptic transmission in the cerebellar nuclei
3.3. CEREBELLAR NUCLEO-CORTICAL PATHWAY: ANOTHER FEEDBACK FOR FINE MODULATION OF PRINCIPAL NEURON ACTIVITY
3.3.1. The nucleo-cortical pathway: a forgotten feedback loop
3.3.2. Cell type identity of the nucleo-cortical neurons: a role for the inhibitory neurons
MATERIALS AND METHODS
RESULTS
CHAPTER 4: DIFFERENTIAL GABAERGIC AND GLYCINERGIC INPUTS OF INHIBITORY INTERNEURONS AND PURKINJE CELLS TO PRINCIPAL NEURONS OF THE CEREBELLAR NUCLEI
4.1. ARTICLE
4.2. SUPPLEMENTARY DATA
4.2.1. Transient expression of GlyT2 in nucleo-olivary neurons
4.2.2. Extracellular recordings of the GlyT2-expressing interneurons
4.3. CONCLUDING REMARKS
CHAPTER 5: BEYOND PRINCIPAL CELLS: THE EXTENDED CONNECTIVITY OF THE INHIBITORY GLYCINERGIC NEURONS OF THE CEREBELLAR NUCLEI
5.1. INTRA-CEREBELLAR OUTPUT OF THE CEREBELLAR NUCLEI: THE INHIBITORY NUCLEO-CORTICAL PATHWAY
5.2. INPUTS TO INHIBITORY NEURONS OF THE CEREBELLAR NUCLEI
5.2.1. Purkinje cells inputs onto inhibitory neurons
5.2.2. Excitatory inputs onto glycinergic neurons
5.2.3. Excitatory inputs onto the principal neurons: mossy fiber versus climbing fiber inputs
5.3. CONCLUDING REMARKS
DISCUSSION
BIBLIOGRAPHY

GET THE COMPLETE PROJECT