| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1Laboratoire de Neurobiologie des Réseaux Sensorimoteurs, Unité Mixte de Recherche 7060, and 2Département de Mathématiques Appliquées, Université Paris Descartes (Paris 5), Centre National de la Recherche Scientifique, Paris, France; and 3Department of Biomedical Engineering, Boston University, Boston, Massachusetts
Submitted 14 May 2008; accepted in final form 14 July 2008
 |
ABSTRACT |
|---|
|
|
|
INTRODUCTION |
|---|
|
). The integrator of the oculomotor system is an excellent system to study persistent activity, especially because the overall function of its neural network is well defined by direct measurements of eye movements. The prepositus hypoglossi nucleus (PHN) is the common integrator for all horizontal eye movements (Goldman et al. 2002), including orientating saccades and the gaze-stabilizing vestibuloocular reflex. Both network and intrinsic properties have been proposed as a substrate for the mechanisms involved in implementing the neural integrator (Major and Tank 2004). An analysis of the intrinsic membrane properties of the PHN neurons (Idoux et al. 2006) has shown that there are three cell types in this nucleus (A, B, and D). Neurons of each cell type differ in their location, shape of the interspike interval, voltage-dependent filter properties, and response to neuromodulation. Furthermore, both experimental data (Rekling and Laursen 1989) and realistic cell models with active dendritic conductances suggest a significant involvement of the persistent sodium conductance (gNaP) in determining the cell discharge pattern (Idoux et al. 2006). In addition, Delgado-García et al. (2006) demonstrated that stimulation of saccadic afferents from excitatory burst neurons projecting on type B PHN neurons can trigger sustained activity in the presence of cholinergic agonists.
The gNaP, like the usual spike-generating gNa, activates rapidly; however, gNaP is activated at more negative potentials (
10 mV below spike threshold) and shows minimal time-dependent inactivation (Crill 1996). In numerous experiments, gNaP has been proposed to be instrumental in persistent activity (Major and Tank 2004). For example, gNaP has been shown to trigger plateau potentials, increase the amplitudes of inhibitory or excitatory postsynaptic potentials (EPSPs), and interact with conductances responsible for the afterhyperpolarization (AHP), which in turn control the excitability and the discharge rate of neurons (Vervaeke et al. 2006).
The present set of experiments focuses on the function of gNaP both in the soma and in the dendrites, using a combination of voltage- and current-clamp techniques, together with pharmacological block of gNaP and dynamic-clamp injection of virtual gNaP. The subsequent analysis fits compartmental neuronal models to data; and demonstrates the different roles of gNaP in the soma and the dendrites of the different cell types as well as the impact of voltage-dependent ionic conductances on the effective electrotonic length, which dynamically modulates the response to synaptic input. These results support the hypothesis that active dendritic conductances play a role in neural integration (Koulakov et al. 2002) and persistent activity in general, using mechanisms involving both synaptic and intrinsic properties of neurons.
|
|
METHODS |
|---|
|
Sixty-six neurons were recorded from adult (25- to 35-day-old) Wistar rats supplied by CERJ (Le-Genest-Saint-Isle, France). The animals were handled in accordance with the European Community Council Directive of November 24, 1986, and following the procedures issued by the French Ministère de l'Agriculture. Brain dissections were performed as described elsewhere (Beraneck et al. 2003). The entire dissection was carried out in sucrose artificial cerebrospinal fluid (aCSF, pH 7.4), composed of (in mM): NaCl, 0; NaHCO3, 26; NaH2PO4, 1; KCl, 5; MgCl2, 1.3; glucose, 11; sucrose, 225; CaCl2, 2.5; and oxygenated with 95% O2-5% CO2. Four to five slices (350 µm) were taken from each animal and incubated, for 
1 h, at room temperature in oxygenated regular aCSF (pH 7.4), which was the same as sucrose-aCSF except with NaCl (124 mM) and no sucrose.
Whole cell patch-clamp recordings: voltage- and current-clamp
Slices were transferred into a temperature-controlled recording chamber (32°C, Warner Instruments). Neurons of the PHN were patched under visual guidance with an infrared differential interference contrast setup. The PHN is located in the brain stem, below the floor of the fourth ventricle, and is limited rostrally by the abducens nucleus, laterally by the medial vestibular nucleus, ventrally by the paragigantocellularis nucleus, caudally by the hyglossal nucleus and the perihypoglossal complex (McCrea and Horn 2005).
Regular patch electrodes (6–8 M
) were filled with intracellular-like solution (pH 7.3) consisting of (in mM): potassium gluconate, 140; MgCl2, 2; EGTA, 0.1; HEPES, 10; ATP, 4; and GTP, 0.4 (Sekirnjak and du Lac 2002). This induces a calculated junction potential of –16 mV, which was not subtracted in the model fits or presentation of the data. All data were corrected for the electrode offset measured at the end of each experiment. Both voltage- and current-clamp experiments were performed with an AxoClamp 2B and a National Instruments acquisition card (PCI-6052E). Stimulations and recordings were done with custom Matlab (7.0.4.365 R14) programs; the signals were low-pass filtered at 2 kHz and sampled at 5 kHz.
Dynamic current-clamp recordings
Dynamic current-clamp stimulations were implemented using a second National Instruments data acquisition card (PCI-6052E) with the RTLDC software (version 2.2.6, http://www.rtxi.org) at a sampling rate of 13,333 kHz (i.e., an update period of 75 µs, whereas the fastest time constant is 150 µs for gNaP; cf. Table A1). A dynamic-clamp model of the persistent sodium conductance (gNaP) has been successfully used in previous studies to functionally add back a pharmacologically blocked conductance (Dorval Jr and White 2005). Since gNaP was added or subtracted, the deterministic version of this conductance was chosen to counteract or mimic the effects of riluzole.
To block the endogenous gNaP, riluzole (Sigma) was applied in the bath between 5 and 20 µM. The chamber was thoroughly rinsed between each slice to prevent gNaP block by residual riluzole.
Cell classification
Quantitative parameters summarizing the spike shape and discharge were extracted as previously described (see Fig. 1 and Beraneck et al. 2003). The following list is provided as a brief summary: Vm, the resting membrane potential (mV); F, the spontaneous discharge rate (spikes/s) and CV, its coefficient of variation; width, the width of the action potential taken at spike threshold (ms); dAHP, the index that quantifies the double AHP (V/s); and AHPR, the AHP rectification (V/s), both of which are extensively described in Beraneck et al. (2003); Fig. 1, as well as the concavity and the convexity (mV).
|
) using 10,000 bootstrap samples. White-noise measurements and complex admittance plots
Responses to discrete white-noise (WN) stimuli were measured to determine somatic and dendritic conductances and are shown as complex admittance plots, which are compact representations of Bode plots (split in magnitude and phase) typically used. Admittance is a generalized conductance and the inverse of impedance. In the complex admittance plots, a line can be drawn between each point and the (0, 0) point. The length of this segment corresponds to the magnitude of the admittance at a given frequency, whereas the angle between this segment and the horizontal x-axis is the phase shift between the response and the stimulation. Therefore the units of both axes of the complex admittance plots are micro-Siemens (µS).
To determine this frequency-dependent complex admittance and the properties of the endogenous gNaP, neurons were stimulated with a discrete WN stimulus in a voltage-clamp mode. In the frequency domain, the WN stimulus was composed of 36 or 60 frequencies, ranging from 0.2 Hz to 2 kHz, having identical magnitudes and random phases. No significant differences were found in the admittance obtained with either WN sources and thus no distinction will be made between them. The WN amplitude in the time domain was about 10 mV peak to peak. These stimuli were applied at a variety of membrane potentials required to estimate both the passive and active properties of the neuron, especially including the dendritic structure.
Data-fitting parameters using compartmental dendritic neuronal models
Methods previously published were used to obtain a data-fitted, conductance-based compartmental model of the cells (see APPENDIX A and Saint-Mleux and Moore 2000a,b). Ten compartments were defined: one for the electrode with or without compensation, one for the soma, and eight for the dendritic tree collapsed into one equivalent cylinder (see Fig. 2A, inset). In each compartment, in addition to the capacitor and the leak conductance representing the passive properties of the membrane, three voltage-dependent conductances were added: a slow potassium conductance, a gNaP, and a conductance responsible for an h-current (Idoux et al. 2006). For each cell, the fit was used to quantitatively determine the electrotonic structure of the neuron, the parameters of the conductances (see Table A1 in APPENDIX A), and the effect of riluzole on the endogenous gNaP.
|
 |
Statistical analysis
The normality of the distribution of each parameter was tested with a Lilliefors test, the robust version of the Kolmogorov–Smirnov test, where the parameters of the normal distribution have to be determined from the sample. Since most of the parameters had a normal distribution, means and SD will be given for the sake of consistency; however, significance was tested with nonparametric Kruskal–Wallis or Wilcoxon test, for respectively pre/postriluzole test and intracell-type comparison. Significance threshold was set at 5%.
|
|
RESULTS |
|---|
|
The classification of cells in the oculomotor brain stem has often been a matter of debate (Beraneck et al. 2003; Sekirnjak and du Lac 2002; Serafin et al. 1991); therefore a new classification algorithm with no a priori requirements (see METHODS) was applied to quantitative parameters measured from PHN adult rat neurons during spontaneous discharge. This algorithm shows that a two-type classification is the most likely distribution 99.85% of the time, whereas one-, three-, and four-type classifications are, respectively, 0.07, 0.04, and 0.04%.
The parameters of both cell types (Fig. 1 and Table 1) are similar to those obtained in a previous classification (Idoux et al. 2006): type A-like neurons do not have a biphasic AHP (dAHP = 0), their spikes are broader, and their AHPR is more pronounced than that for type B-like neurons. Furthermore, if the presence of oscillations, a qualitative parameter, is used, type A-like neurons can be further subdivided, as in the guinea pig neurons (Idoux et al. 2006), into genuine type A without oscillations (n = 4/18) or type D with subthreshold oscillations (n = 14/18). No oscillatory cells were found, under control conditions, in type B-like neurons, which will therefore be designated type B.
|
All PHN neurons have a persistent sodium conductance
In our previous studies (Idoux et al. 2006; Serafin et al. 1991), the existence of a persistent sodium conductance (gNaP) in the cell was assessed by detecting the presence or absence of a plateau potential triggered by a short subthreshold pulse. Discrete WN voltage-clamp measurements indicate that gNaP is actually present in all neurons, in contrast to the "plateau method," which fails to detect it in 30% of the neurons (
43% for type D).
During a voltage clamp, the net current can be inward, when sodium or calcium currents are greater than the outward potassium currents. In such a case, the real part of the complex admittance becomes negative at low frequencies (leftward shift between –62 and –41 mV in Fig. 2A); therefore the conductance responsible for the observed inward current can be considered as a functional "negative conductance." Since the shift is blocked in the presence of low concentrations of riluzole (Fig. 2B), this conductance can be defined as a gNaP, and was found in all cells tested (negative conductance: type B, 20/20; type D, 12/12; partially blocked by riluzole: type B, 10/10; type D, 9/9).
|
The wide frequency range used in complex admittance measurements provides a probe of remote dendritic properties, which allows one to quantitatively estimate both the somatic and dendritic conductances of compartmental neuronal models (see APPENDIX A for details). Since the somatic and dendritic distributions of ion channels are not known, all fits of the control data were done using a uniform distribution of all conductances throughout the soma and dendritic cable. The MSE (see METHODS) measures the error between the fit and the actual data. Therefore it represents the accuracy of the fits, with the lower value of the MSE being the most accurate fit (MSE = 0 fS2 would indicate that the fit is perfectly superimposed on the data). Since MSE is similar for type B and type D in both control (ctl) and riluzole cases (ril) (Bctl: 1.63 ± 0.54 fS2 and Dctl: 1.30 ± 0.34 fS2; Bril: 2.08 ± 0.77 fS2 and Dril: 2.03 ± 0.87 fS2; pB vs. D,ctl, pB vs. D,ril, pB,ctl vs. ril, pD,ctl vs. ril were all >5%), all the fits of both cell types in both conditions have similar accuracy: the subsequent differences are due to cell type and not to differences in the quality of the fits.
The active conductances differ between cell types: gh, the conductance responsible for the h-current, is negligible in type D neurons but clearly present in type B (B: 5.91 ± 9.91 nS vs. D: 0.03 ± 0.10 nS, P < 0.1%). Furthermore, although not significantly different because of a large variance, the gNaP of type D is half that of type B neurons (B: 1.31 ± 1.62 nS vs. D: 0.64 ± 0.48 nS, P = 12.9%).
The most significantly different parameter between type B and type D neurons is the electronic length (EL) of the dendrite. Typically, the EL represents the passive morphological structure of the dendrite; however, the presence of ionic conductances in the dendrites requires a consideration of an active electrotonic length (AEL), which actually controls the attenuation of the signal along the dendritic tree (see APPENDIX B and Moore et al. 1999). Thus not only is the passive EL significantly smaller for type B neurons (B: 0.37 ± 0.27; D: 0.54 ± 0.16; P = 0.2%), but also quantitative features of the voltage-dependent conductances make the AEL of the two cell types even more different in the subthreshold range. The AEL of type B neurons decreases dramatically to virtually zero as the neuron is depolarized toward the spike threshold (Fig. 3A). This minimum point occurs as a result of the balance of inward sodium and outward currents, which represents a maximal or resonant value for the active impedance of the neuron. Thus the signal is more attenuated along the dendritic tree in type D neurons, making it more independent of the soma, whereas the dendrites of type B can dynamically regulate the amount of postsynaptic potential reaching the soma.
|
In both cell types, the uniformly distributed gNaP is significantly reduced by riluzole in the bath (Fig. 2; B: 0.53 ± 0.47 nS, P = 0.5%, n = 9; D: 0.18 ± 0.17 nS, P = 0.8%, n = 9). To assess the relative importance of dendritic versus somatic gNaP, additional fits using a nonuniform gNaP distribution of the riluzole data were done. In these fits, the somatic and dendritic gNaP are now considered as independent parameters, where only one of them is reduced by the fitting procedure. As illustrated and demonstrated in the following paragraphs, parameter estimation with these constraints shows that the riluzole data are much better fitted with a dendritic compared to a somatic decrease of gNaP.
The inset in Fig. 3B shows the four conditions used in the fitting procedure presented here. Figure 3B, a and d, like Fig. 2 before, were obtained with a model where gNaP was considered as uniformly distributed, i.e., the dendritic and somatic gNaP have the same density. All parameters were the same for Fig. 3B, a and d except for the value of gNaP, which is decreased in Fig. 3Bd (cf. Fig. 2). On the other hand, for Fig. 3B, b and c, the somatic and dendritic gNaP were considered as independent parameters, respectively. Therefore to obtain Fig. 3Bb, only the somatic gNaP was reduced, whereas the dendritic gNaP was kept at its control value reached in Fig. 3Ba. Conversely for Fig. 3Bc, the dendritic gNaP was the free parameter and the somatic gNaP was maintained at its control value of Fig. 3Ba. The complex admittance plot of the estimated model with a reduced dendritic gNaP (lines, Fig. 3Bc) is always a better match to the riluzole data (dots) than that with just a reduced somatic gNaP (lines, Fig. 3Bb).
Furthermore, to quantify how significant the differences between the fits on somatic versus dendritic gNaP were, two numeric indices were calculated: 1) the mean squared error (MSE) between the fit and the data, 2) the fraction of gNaP blocked
 |
No correlation was found in either cell type between %gNaP and %gNaPsoma (B: r2 = 14.2%, P = 28.3%, n = 9; D: r2 = 17.1%, P = 38.9% n = 9, cf. Table A1 in APPENDIX A). Conversely, %gNaP and %gNaPdendrite are strongly and significantly correlated (B: r2 = 67.1%, P = 0.4%, n = 9; D: r2 = 78.1%, P = 0.2%, n = 9). Furthermore, the fits are less accurate, i.e., the MSE is significantly higher, when the free parameter is the somatic gNaP (B: MSEsoma 2.50 ± 1.00 vs. MSE = 2.08 ± 0.77, P = 2.8%, n = 9; D: 2.29 ± 0.91 vs. 2.03 ± 0.87 P = 1.1%, n = 9), but not when the free parameter is the dendritic gNaP (MSEdendrite B: 2.03 ± 0.79, n = 9; D: 2.10 ± 0.89, n = 9; both P values >5%). Thus most of the pharmacological block of gNaP by riluzole can be reproduced in the model by just reducing the value of the dendritic (Fig. 3B, c vs. d) but not the somatic gNaP (Fig. 3B, b vs. d).
Since the complex admittance is a transfer function between the input and the output of the neuron and most of the leftward shift in the complex admittance plot is provided by the dendritic gNaP, the dendritic gNaP is likely to play a more important role than the somatic gNaP in determining the computational properties (i.e., input/output relationships) of the neuron.
Perisomatic gNaP is involved in the spontaneous discharge
In the presence of riluzole (Fig. 4, A vs. B), type B neurons lose their spontaneous discharge (n = 10/10) as they undergo a steady depolarization (7/10) or no change in membrane potential (2/10). A slight hyperpolarization was observed only once. In riluzole, the membrane potential becomes very erratic, showing fast oscillations and frequent spike clusters. Although counterintuitive, these effects are reversed by dynamic-clamp injection in the soma of a virtual gNaP, that is, both the discharge profile and the membrane potential of the neuron are restored to their preriluzole state (10/10, Fig. 4C).
|
As with type B, the type D neuron's spontaneous discharge is abolished in the presence of riluzole (n = 8/8, Fig. 5, A vs. B). The mechanisms are likely to be different, since the membrane potential of type D neurons is either slightly hyperpolarized (Fig. 5B) or not changed (7/8). However, in all cases, the membrane potential is significantly more irregular and does not show obvious subthreshold oscillations. The spike discharge, unlike the oscillations, is restored with the dynamic-clamp injection of gNaP (7/7, Fig. 5B).
|
In the presence of TTX (0.5 µM; data not shown), spikes cannot be restored by a virtual gNaP. Thus somatic gNaP appears to be working in concert with the fast gNa to maintain normal spontaneous discharge in both cell types. The dynamic-clamp results for both cell types are consistent with their electrotonic structure: the more electrotonically compact the cell, the better the soma can control the dendrites.
|
|
DISCUSSION |
|---|
|
The gNaP is present in all PHN neurons, as shown by the white-noise experiments and confirmed by its blockage with riluzole and reinsertion with the dynamic-clamp. Although at high concentrations riluzole can block other channels (IC50 = 40 µM for Kv1.5, IC50 = 120 µM for Kv3.1, Ahn et al. 2005; IC50 = 70 µM for Kv1.4 ,Xu et al. 2001), at low concentrations (5 to 20 µM), it has been used to specifically target the persistent sodium conductance (Dorval Jr and White 2005). Furthermore, the ability to reproduce the same changes in the membrane potential dynamics with a pharmacological (riluzole) or a functional (with –gNaP) block confirms that 1) gNaP was present in all PHN neurons, 2) riluzole can block gNaP, and 3) the dynamic-clamp model is close enough to the endogenous gNaP to cancel it in the somatic region.
Both riluzole and injection of –gNaP induce an apparent depolarization of PHN type B neurons, although the injection of +gNaP in the presence of riluzole restores the mean membrane potential and firing rate of these neurons. These counterintuitive results would appear to be the consequence of interactions between multiple voltage-dependent conductances. Such interactions have been shown in CA1 neurons where modeling and dynamic-clamp experiments have shown that gNaP enhances afterhyperpolarizations (Vervaeke et al. 2006). As mentioned earlier, gNaP contributes to a depolarization required to trigger spikes, which would in turn activate hyperpolarizing potassium currents (Liu and Shipley 2008). When riluzole is applied to the bath (or with –gNaP), such a trigger would be missing, which may explain the disappearance of the spontaneous discharge, thus preventing the complete repolarization. This depolarization does not occur in type D neurons, which have a quantitatively different conductance profile.
Perisomatic gNaP is responsible for the spontaneous activity
In both type B and type D neurons, both the pharmacological block and functional block of gNaP lead to a disruption of the spontaneous firing rate while its recovery restores the discharge pattern. This conductance acts as a rapid positive feedback on the membrane potential (Vm) and because of its steep dependence on Vm and the 150-µs maximal time constant (tp; see Table A1 in APPENDIX A), it allows fast changes in the membrane potential needed to trigger the spike. The current-clamp and dynamic-clamp results fit well with the conclusions drawn from the WN experiments: the somatic injection of a conductance is able to control the type B neurons, which are electrotonically compact. Conversely, for the less electrotonically compact type D neurons, the cancellation of the riluzole effects seems to be restricted to the spike-triggering mechanism, which is mainly a perisomatic phenomenon (Khaliq and Raman 2006).
Density of gNaP in the PHN neurons
Somatic gNaP seems to be related to spike triggering in all cell types. However, the discharge rate of type D is significantly lower than that of type B, which might be explained by the lower value of gNaP of type D neurons. Furthermore, since riluzole cancellation of subthreshold oscillations cannot be counteracted nor mimicked by somatic injection of ±gNaP in type D cells, the dendritic gNaP may induce relatively soma-independent bistabilities that could play a critical role in neural integration.
Functional implications of different cell types
As shown by the present work, gNaP has two different functions depending on its location in different cell types, both of which are critical for the oculomotor integration: the somatic gNaP appears to maintain the spontaneous activity of these neurons, whereas the dendritic gNaP may dynamically alter the active electrotonic length and/or contribute to bistable processes involved in persistent activity.
The effect of the dendritic conductances may well be a key point in the cellular component of the integration and shows the importance of the cell classification. Since the active electrotonic length of type B cells decreases as the membrane potential gets closer to the spiking threshold, the attenuation along the dendritic tree plummets as well (Fig. 3A). Therefore increasingly more distal excitatory postsynaptic potentials would reach the soma and increase the probability of triggering an action potential. Thus such a potential-induced decrease of AEL provides a possible explanation of how the depolarizing cholinergic inputs can affect the PHN integrative properties of type B neurons seen both in vivo and in vitro (de Dios Navarro-López et al. 2005), that is, by increasing the number of EPSPs from distal synapses reaching the soma during a constant input.
Furthermore, type B neurons are bistable in the presence of N-methyl-D-aspartate (NMDA), which also induces a negative conductance similar to that of gNaP (Idoux et al. 2006). NMDA also produces bistability that would maintain persistent activity needed for the neural integrator and could provide an extrinsic modulation of the AEL on top of its endogenous changes, Thus both the network and intrinsic properties would play important roles in operation of the neural integrator (Major et al. 2008).
Finally, the modulation of the AEL by the set of available voltage- and/or ligand-dependent conductances would act as an intrinsic positive feedback loop, allowing a cell to integrate a constant input into a ramp of discharge rate and then hold it, as seen in the neural integrator, working memory, and persistent activity in general.
|
|
APPENDIX A: MODEL AND DATA-FITTING PROCEDURE |
|---|
|
; Murphey et al. 1995). As a summary, admittance of the ith compartment is given by the following formula, with i between 0 (soma) and the number of compartments (Nbcomp, i.e., eight in the present case)
 |
 |
To calculate admittance of the soma, Ycomp(V0, f), it is necessary to recursively add the individual Ycomp(Vi, f) starting at the end compartment. The complex admittance of the neuron is the one recorded at the soma; therefore Yneuron(V, f) = Ycomp(V0, f).
The gating variable xp is given by the set of traditional Hodgkin–Huxley voltage dependence
 |
 |
The Vi values are determined at steady state.
Definition of the parameters
Glossary
Each of the voltage-dependent conductances is defined by the four following items. For the pth conductance:
p
The variables xp and p are controlled by three parameters:
xp = 0.5) (Vi) at Vi = vp p = tp) The electrode is defined as a resistor–capacitor (RC) circuit, with a resistance Re, in series with the cell, and a capacitance Ce, parallel to the cell. Re can be compensated by a negative resistance, Rss.
Therefore the actual complex admittance at the recorded voltage V, Yt(V, f), recordable by the amplifier is
 |
 |
|
|
APPENDIX B: ACTIVE ELECTROTONIC LENGTH CALCULATION |
|---|
|
)
 |
 |
The second formulation makes it easier to see that, as for the analytical model, AEL = elength when the compartment has no voltage-dependent conductance because, in that case, |Y(Vi, 0)| = gleak.
|
|
GRANTS |
|---|
|
|
|
ACKNOWLEDGMENTS |
|---|
|
|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: L. E. Moore, Laboratoire de Neurobiologie des Réseaux Sensorimoteurs, UMR 7060, Université Paris Descartes (Paris 5), CNRS, 45 rue des Saints-Pères, 75270 Paris Cedex 06, France (E-mail: lee.e.moore{at}gmail.com)
|
|
REFERENCES |
|---|
|
Beraneck M, Hachemaoui M, Idoux E, Ris L, Uno A, Godaux E, Vidal PP, Moore LE, Vibert N. Long-term plasticity of ipsilesional medial vestibular nucleus neurons after unilateral labyrinthectomy. J Neurophysiol 90: 184–203, 2003.
Crill WE. Persistent sodium current in mammalian central neurons. Annu Rev Physiol 58: 349–362, 1996.[CrossRef][Web of Science][Medline]
de Dios Navarro-López J, Delgado-García JM, Yajeya J. Cooperative glutamatergic and cholinergic mechanisms generate short-term modifications of synaptic effectiveness in prepositus hypoglossi neurons. J Neurosci 25: 9902–9906, 2005.
Delgado-García JM, Yajeya J, de Dios Navarro-López J. A cholinergic mechanism underlies persistent neural activity necessary for eye fixation. Prog Brain Res 154: 211–224, 2006.[Web of Science][Medline]
Dorval AD Jr, White JA. Channel noise is essential for perithreshold oscillations in entorhinal stellate neurons. J Neurosci 25: 10025–10028, 2005.
Goldman MS, Kaneko CR, Major G, Aksay E, Tank DW, Seung HS. Linear regression of eye velocity on eye position and head velocity suggests a common oculomotor neural integrator. J Neurophysiol 88: 659–665, 2002.
Idoux E, Chambaz A, Eugene D, Magnani C, White JA, Moore LE. The function of persistent sodium conductance in prepositus hypoglossi nucleus neurons. Program 718.14. Abstract Viewer/Itinerary Planner. Washington, DC: Society for Neuroscience, 2007. Online.
Idoux E, Serafin M, Fort P, Vidal PP, Beraneck M, Vibert N, Muhlethaler M, Moore LE. Oscillatory and intrinsic membrane properties of guinea pig nucleus prepositus hypoglossi neurons in vitro. J Neurophysiol 96: 175–196, 2006.
Khaliq ZM, Raman IM. Relative contributions of axonal and somatic Na channels to action potential initiation in cerebellar Purkinje neurons. J Neurosci 26: 1935–1944, 2006.
Koulakov AA, Raghavachari S, Kepecs A, Lisman JE. Model for a robust neural integrator. Nat Neurosci 5: 775–782, 2002.[CrossRef][Web of Science][Medline]
Liu S, Shipley MT. Multiple conductances cooperatively regulate spontaneous bursting in mouse olfactory bulb external tufted cells. J Neurosci 28: 1625–1639, 2008.
Major G, Polsky A, Denk W, Schiller J, Tank DW. Spatio-temporally graded NMDA spike/plateau potentials in basal dendrites of neocortical pyramidal neurons. J Neurophysiol 99: 2584–2601, 2008.
Major G, Tank D. Persistent neural activity: prevalence and mechanisms. Curr Opin Neurobiol 14: 675–684, 2004.[CrossRef][Web of Science][Medline]
McCrea RA, Horn AK. Nucleus prepositus. Prog Brain Res 151: 205–230, 2005.[Web of Science][Medline]
Moore LE, Chub N, Tabak J, O'Donovan M. NMDA-induced dendritic oscillations during a soma voltage clamp of chick spinal neurons. J Neurosci 19: 8271–8280, 1999.
Murphey CR, Moore LE, Buchanan JT. Quantitative analysis of electrotonic structure and membrane properties of NMDA-activated lamprey spinal neurons. Neural Comput 7: 486–506, 1995.[Web of Science][Medline]
Rekling JC, Laursen AM. Evidence for a persistent sodium conductance in neurons from the nucleus prepositus hypoglossi. Brain Res 500: 276–286, 1989.[CrossRef][Web of Science][Medline]
Saint-Mleux B, Moore LE. Active dendritic membrane properties of Xenopus larval spinal neurons analyzed with a whole cell soma voltage clamp. J Neurophysiol 83: 1381–1393, 2000a.
Saint-Mleux B, Moore LE. Firing properties and electrotonic structure of Xenopus larval spinal neurons. J Neurophysiol 83: 1366–1380, 2000b.
Sekirnjak C, du Lac S. Intrinsic firing dynamics of vestibular nucleus neurons. J Neurosci 22: 2083–2095, 2002.
Serafin M, de Waele C, Khateb A, Vidal PP, Muhlethaler M. Medial vestibular nucleus in the guinea-pig. I. Intrinsic membrane properties in brainstem slices. Exp Brain Res 84: 417–425, 1991.[Web of Science][Medline]
Vervaeke K, Hu H, Graham LJ, Storm JF. Contrasting effects of the persistent Na+ current on neuronal excitability and spike timing. Neuron 49: 257–270, 2006.[CrossRef][Web of Science][Medline]
Xu L, Enyeart JA, Enyeart JJ. Neuroprotective agent riluzole dramatically slows inactivation of Kv1.4 potassium channels by a voltage-dependent oxidative mechanism. J Pharmacol Exp Ther 299: 227–237, 2001.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
TOP
ABSTRACT
INTRODUCTION
