| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Anatomy and Neurobiology, University of California, Irvine, California, 92697-1280
Submitted 22 September 2003; accepted in final form 9 January 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
-modulated depolarizing current inputs to CA1 pyramidal cells demonstrated that the principles underlying the modulation of pyramidal cell excitability by heterogeneous IPSC populations also apply during membrane potential oscillations. Taken together, these experimental results and the computational modeling data show the existence of simple rules governing the interactions of heterogeneous interneuronal inputs and principal cells. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
; McBain and Fisahn 2001). However, in most cases, the functional roles of the specific interneuronal classes are still not fully understood, and modeling studies indicated that the degree of heterogeneity itself is an important functional determinant of neuronal networks (Aradi and Soltesz 2002). For example, alterations in the cell-to-cell variance of a parameter, such as the degree of spike frequency adaptation or resting membrane potential, within a given interneuronal subtype, can modulate a wide range of network properties, including firing rates, oscillation frequencies, and network synchrony (Aradi and Soltesz 2002; Tiesinga and Jose 2000; Wang and Buzsaki 1996). Similarly, altering the variance of inhibitory postsynaptic current (IPSC) peak conductances injected into simulated neurons has been demonstrated to powerfully alter the excitability of the postsynaptic cells, even if the scatter in the IPSC populations was introduced without alterations in the mean peak conductance of the GABAergic inputs (Aradi et al. 2002).
Previous computational modeling studies suggested that the effects of heterogeneity in GABAergic inputs on firing rates of postsynaptic cells can be deduced from a few simple rules (Aradi et al. 2002; see also Fig. 1). Here we tested these ideas using dynamic clamp techniques (Sharp et al. 1993), which, similar to modeling methods, allow the exact modulation of the variance in IPSC peak conductances or conductance decays around unaltered means. Since heterogeneous perisomatic GABAergic inputs play especially important roles in the regulation of firing rates and synchrony of postsynaptic neurons (Cobb et al. 1995; Freund 2003; Miles et al. 1996), we examined the effects of variability using somatically injected IPSC conductances. These dynamic clamp results from real neurons fully confirm and extend the previous computational modeling data showing the existence of simple rules determining the modulation of neuronal excitability by the heterogeneity of GABAergic inputs.
|
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Brain slices were prepared as previously described (Chen et al. 1999). Anesthetized Wistar rats 18–22 days of age were decapitated, and the brains were removed and cooled in 4°C oxygenated (95% O2-5% CO2) artificial cerebrospinal fluid (ACSF) composed of (mM) 126 NaCl, 2.5 KCl, 26 NaHCO3, 2 CaCl2, 2 MgCl2, 1.25 NaH2PO4, and 10 glucose. Horizontal brain slices (400 µm) were prepared with a vibratome tissue sectioner (Lancer Series 1000). The brain slices were sagittally bisected into two hemispheric components and were incubated by submerging in 32°C ACSF for 
1 h in a holding chamber before recording.
Electrophysiology
Individual slices were transferred to a recording chamber (Chen et al. 1999) and perfused with ACSF containing 20 µM bicuculline methiodide (BMI), 20 µM 2-amino-5-phosphonovaleric acid (APV), and 10 µM 2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide (NBQX). The brain slices rested on filter paper and were stabilized with platinum wire weights. The tissue was continuously superfused with humidified 95% O2-5% CO2, and the temperature of the perfusion solution was maintained at 34° C. All salts were obtained from Fluka; APV and NBQX were purchased from Tocris; and BMI was from Sigma Research Biochemicals International. Patch pipettes were pulled from borosilicate (KG-33) glass capillary tubing (1.5 mm OD; Garner Glass) with a Narishige PP-83 two-stage electrode puller. Pipette solutions consisted of (in mM) 140 potassium gluconate, 2 MgCl2, and 10 HEPES. "Blind" whole cell recordings were obtained as previously described (e.g., Chen et al. 1999). The membrane voltage was recorded with Axoclamp 2B or Axopatch 200A amplifiers (Axon Instruments; since the data obtained with the 2 amplifiers were similar, the data were pooled) and digitized at 88 kHz (Neurocorder, NeuroData) before being stored in PCM form on videotape. The series resistance was monitored throughout the recordings, and the data were rejected if it significantly increased.
Dynamic clamp
The hardware-based dynamic clamp system Synaptic Module 1 (SM-1) Conductance Injection Amplifier (Cambridge Conductance) (Kleppe and Robinson 1999) was used to inject trains of artificial inhibitory synaptic conductance pulses of variable amplitudes and decay time constants. The "real-time" membrane potential [Vm(t)] was measured and fed into SM-1 together with the conductance-command waveform [GIPSC(t)]. The inhibitory synaptic conductances were generated with the NEURON simulator as detailed in the model description (see Model cells). DigiData 1332 digitizer (Axon Instruments) was used as an interface to transfer the conductance commands to the SM-1. The electrical circuit of SM-1 computed the current to be injected into the cell based on the following equation: IPSC(t) = GIPSC(t) x [Vm(t) - EGABA], where the synaptic reversal potential (EGABA) was set to specific values on SM-1 (most often to -65 mV, unless stated otherwise in specific experiments).
Analysis
Recordings were filtered at 3 kHz before digitization at 20 kHz, using a personal computer for analysis, by Strathclyde Electrophysiology Software (courtesy of J. Dempster, University of Strathclyde, Glasgow, UK), and the Synapse Software (courtesy of Dr. Y. De Koninck, Laval University, Quebec City, Canada), and with Sigma-Plot. Data are presented as means ± SE.
Model cells
The multicompartmental models of CA3 and CA1 pyramidal cells, constructed with the NEURON software (Hines and Carnevale 1997), were described previously in detail (Aradi and Soltesz 2002; Aradi et al. 2002; Chen et al. 2001). Briefly, the CA1 model cells were constructed with 12 compartments and included Na+, delayed rectifier K+, and h-currents. The passive parameters, as well as the properties and distributions of the voltage-gated conductances, were determined from experiments and from published values (Chen et al. 2001). For further details, see http://www.ucihs.uci.edu/anatomy/soltesz/supp.htm.1 The model of bursting CA3 neurons by Traub et al. (1991) was implemented with 19 compartments and six voltage-activated channels. Briefly, the CA3 model included Na+ channels, delayed rectifier K+ channels, Ca2+ channels, SK- and BK-type Ca2+-dependent K+ channels, and A-type K+ channels (Traub et al. 1991).
Modeling synaptic currents
In light of previous studies showing that most mIPSCs originate proximal to the soma (Cossart et al. 2000; Soltesz et al. 1995), the perisomatic IPSCs were modeled using the peak amplitude (
) and kinetics (rise and decay time constants, 
rise and decay, respectively) of miniature IPSCs (mIPSCs) measured experimentally from CA1 pyramidal cells (Aradi et al. 2002; Chen et al. 1999)
 |
 |
The simulated GIPSC(t) waveforms were used in both the modeling and dynamic clamp experiments. The IPSC commands were computed by the analog circuit (SM-1) in the dynamic clamp experiments using the simulated GIPSC(t) command waveform.
The experimentally determined mean and SD of the mIPSCs (meanexp and SDexp) for the amplitude and decay kinetics used in this study were the same as reported in Aradi et al. (2002) (IPSC conductances: meanexp = 0.896 nS; SDexp = 0.459 nS; decay time constants: meanexp = 6.8 ms; SDexp = 3.4 ms). The IPSC rise time constant was 1.2 ms. Note that SDexp reflects the variance of the IPSC conductance and decay in individual cells and not the SD of the means between neurons (Aradi et al. 2002). From these experimental data, it is apparent that CVexp= SDexp/meanexp is around 0.5 for both the IPSC conductance amplitudes and decay time constants. In the simulations and dynamic clamp experiments, the amplitude (or the decay time constant) was either kept the same across the population of IPSCs arriving at the model cell (i.e., CV = 0), or the values were increasingly scattered around the population mean, with CV = 0.25, 0.5, or 0.75. It is important to emphasize that changes in the variance of the conductance or decay in a population of IPSCs (introduced without changes in the mean conductance or mean conductance decay) do not alter the mean charge transfer (for details, see Aradi et al. 2002).
The IPSC conductances were injected at 20 Hz in Figs. 2, 3, and 5 with SD = 0, at the experimentally observed spontaneous IPSC (sIPSC) frequency (meanexp = 16.5 Hz and SDexp = 7.7 Hz) in Fig. 4, A–D, or at 40 Hz in Fig. 6. Note that, although the total GABAergic input frequency across the soma and the dendritic tree is likely to be higher (especially in vivo; Pare et al. 1998), these frequencies around 20–40 Hz most likely represent the inputs arriving from perisomatic synapses (i.e., from basket- and axo-axonic cells; Cossart et al. 2000; Soltesz et al. 1995; Ylinen et al. 1995), which is what this study focused on, using somatic conductance injections.
|
|
|
|
|
To examine the effects of different populations of IPSC conductances (e.g., groups of IPSCs with or without variance) on action potential firing rates, the model and biological pyramidal cells were depolarized using a somatically injected positive current (Idepol) so that the CA1 cells fired tonically at around 10–20 Hz, and the CA3 cells produced bursts at around 1 Hz (Traub et al. 1991). The steady-state depolarizations lasted for 2 s in Figs. 2, 3, 4 and 8 s in Fig. 5. The frequency intracellular membrane voltage oscillations (Soltesz and Deschênes 1993) were mimicked by injecting a 2-s-long, 5-Hz sinusoidal function (Idepol = I0 x [1 + sin(5 Hz x t)], where I0 was between 0.05 and 0.5 nA).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Previous computational modeling studies suggested that the effects of altered variances in the peak conductance and decay of populations of IPSCs on firing rates could be predicted from a few simple rules (Aradi et al. 2002). Since these rules form the basis of this paper, they are schematically illustrated in Fig. 1. Briefly, the rules state that the effect of heterogeneous populations of GIPSCs on the average firing rate of the postsynaptic cell depends on 1) the degree of heterogeneity (i.e., the variance); 2) the mean around which the heterogeneity takes place; 3) the excitation that the postsynaptic cell receives; and 4) EGABA. Central to predicting the effects of heterogeneous GIPSCs on the cell's firing rate using these rules is the input-output plot (generated by testing the effects of homogeneous GIPSC populations with increasing means and CV = 0 while the cell is given a suprathreshold current injection; Fig. 1A). The region on the input-output plot where the GIPSCs become large enough to suppress cell firing (the hatched region in Fig. 1A) is referred to as the "critical region." The larger the variance and the closer the mean GIPSC to the critical region (indicated in Fig. 1B by the location of the distribution relative to the hatched region), the more pronounced the effects of heterogeneous populations of inputs will be on the firing rate (Fig. 1C). The direction of variance effects depends on the relationship of the mean GIPSC to the critical region. Namely, if the mean is smaller (larger) than the critical region, heterogeneity will decrease (increase) the average firing rate (Fig. 1, B and C). Introduction of stochastic processes (e.g., nonregular inter-event intervals) causes a decrease in the slope of the critical region (i.e., a multiplicative change, see Chance et al. 2002 and Fig. 1D, dashed line), whereas alterations in excitation (depolarization), or changes in EGABA, cause parallel shifts in the critical region (i.e., an additive change; Fig. 1D). For further explanation, see Supplementary material.
Effects of variance in the peak conductance in biological CA1 pyramidal cells
The rules suggested by the computational studies were derived from simulations carried out on model neurons that had only a few ion channel types. Therefore it was important to test the applicability of these rules in complex biological neurons. Dynamic clamp techniques were used to inject inhibitory conductances into CA1 pyramidal cells recorded in the whole cell configuration with endogenous fast GABAergic and glutamatergic inputs blocked (see METHODS). In most experiments (with the exception of Fig. 4), two conductance values were injected around the mean in an alternating fashion, and the scatter was increased between them to test the effects of increased variance (e.g., with a mean value of 10 nS, 1 series of conductance injection could be 8, 12, 8, 12 nS..., the next could be 6, 14, 6, 14 nS, etc.).
The slope of critical region in the input-output plots from a biological neuron appeared less steep (Fig. 2A) than in the relatively simple model cells (Fig. 1A) (Aradi et al. 2002). Increasing the scatter in the peak conductance of the incoming IPSCs around mean values that were strong enough to completely suppress firing (point B in Fig. 2A) resulted in an elevation of the average firing rate (Fig. 2B; note that point B in Fig. 2A corresponds to CV = 0 in Fig. 2B), as predicted from the model neurons (distribution 4 in Fig. 1B and point 4 in Fig. 1C). These effects are also illustrated by a larger number of cells (Fig. 2, C and D; n = 8). The modulation of the firing rates by changes in variance in the peak conductances was studied at multiple mean conductance values (and not just at 1 point as in Fig. 2B), and the results are illustrated on color-coded contour plots (Fig. 2E).
As demonstrated by the contour plot in Fig. 2E, increasing variance in the inputs (plotted on the y axis as the CV of GIPSC) resulted in enhanced firing rates (e.g., 0–20 Hz at GIPSC = 15 nS, indicated on the x axis). Indeed, in all the results shown in Fig. 2, A–E, increased heterogeneity in the population of incoming inputs invariably enhanced the firing rate of the pyramidal cells. This is due to the fact that, in these experiments, the mean values around which the scatter was introduced were to the right of or in the critical region (Fig. 1, B and C), i.e., the means were large enough to decrease the cell firing rate. However, the modeling studies predicted that when the mean conductance is small and located to the left of the critical region (e.g., distribution 2 in Fig. 1B), scattering the peak conductances of the inhibitory inputs should decrease the firing rate. In agreement with the modeling studies, increasing the variance in the peak conductances around sufficiently small mean values caused a decrease in the discharge frequency (Fig. 2, F and G; same cell as in Fig. 2E). In the case of the illustrated cell in Fig. 2, F and G, although small (<3 nS) conductances had little effect on the pyramidal cell firing (Fig. 2F), scattering of the conductances around these small means resulted in some IPSC conductances that were large enough to reduce the firing of the CA1 cell (Fig. 2G). Since these results (n = 5) agreed with the predictions of the modeling studies on the variance increases around both small mean conductances (leading to decreased firing, as with GIPSC values < 3.4 nS in Fig. 2G) and around large means (resulting in increased firing, as in GIPSC values > 3.4 nS in Fig. 2G and for all GIPSC values in Fig. 2E), in the subsequent dynamic clamp experiments, we focused on studying the effects of alterations in GIPSC variances around larger means.
Alterations in postsynaptic excitation and in EGABA and effects of heterogeneity in IPSC conductance decay
The predictions of the modeling studies regarding the effects of Idepol and EGABA (Fig. 1D) were also verified in CA1 pyramidal cells using dynamic clamp methods. As expected (Fig. 1D), increased Idepol caused a parallel, rightward shift in the input-output plots, since with a greater excitatory input, larger GIPSCs were required to suppress the cell firing (Fig. 3A1). As a result of the rightward shift in the input-output plots, increasing the GIPSC variance around mean GIPSC values to the right of the critical region (10 nS in Fig. 3, A1 and A2) resulted in a larger increase in cell firing with stronger depolarizing inputs (Fig. 3A2). In other words, with the increase in depolarization, the critical region moved closer to the mean GIPSC value around which the scatter was introduced, and (as suggested in Fig. 1, B and C) the closer the mean is to the critical region, the larger the variance effects become. Figure 3B illustrates the influence of increasing tonic depolarization on variance effects around multiple peak conductances. Again, increased Idepol caused a rightward shift in the contour plots (note the changes in the x axis in Fig. 3B), since with increasing excitation, larger GIPSCs are required to suppress the cell firing (n = 12). In contrast to the rightward shift in the contours with increasing Idepol (Fig. 3B), a hyperpolarizing shift in EGABA (e.g., from -65 to -75 mV) caused a leftward shift in the contours (Fig. 3C; n = 10), since smaller conductance values could suppress the cell firing if EGABA was more hyperpolarized.
Next, the heterogeneity in the decay time constants of the conductances was also tested, with the IPSC peak conductance kept at a constant value. As expected, increasing the mean decay time constants (plotted on the x axis of Fig. 3D) decreased the firing of the cell (at a given Idepol). In addition, the scattering of the decay time constants around mean values that were large enough to decrease firing resulted in an increase in the discharge rate of the postsynaptic cell (Fig. 3D; n = 11), as predicted by the computational simulations (Aradi et al. 2002).
Effects of variance in the inter-event intervals and Gaussian-like amplitude distributions
In the experiments described so far, the IPSCs arrived at a constant inter-event interval, and the peak conductances (and the decay time constants) were represented by two values around the mean in an alternating fashion during a single period of depolarizing current injection. To determine if the rules outlined in Fig. 1 also applied to a more realistic scenario, the IPSC conductances were drawn from a Gaussian distribution (represented by 9 values; Fig. 4A), and they were introduced with the SD and mean of the experimentally observed sIPSC frequency (Fig. 4B; for meanexp and SDexp, see METHODS) (Aradi et al. 2002). Note that the Gaussian-like scatter in the peak amplitude was introduced without a change in the mean peak conductance values. As expected from the effects of introduced stochastic processes (Fig. 1D; see also Aradi et al. 2002 for a discussion and references), the critical region in the input-output plots obtained under these conditions had a shallow slope, with a long tail on the right (Fig. 4C; n = 5), and the plot did not reach a complete suppression of cell firing (as explained in the Supplementary material in connection with the dashed line in Fig. 1D, the lack of complete suppression of firing is because the cell can fire during the longer gaps between the IPSCs that occur when the inputs arrive with a relatively large SD in the inter-event intervals). As a result of the shallow slope, introduction of increasing variances in GIPSCs had the predicted small effects. Specifically, increasing the CV of GIPSCs around point D in Fig. 4C increased the illustrated cell's firing from 5 to 7 Hz in Fig. 4D. Note that the elevation of the firing frequency to 7 Hz in the illustrated cell was a relatively small effect (as expected); nevertheless, it was close to the predicted maximum value of 8 Hz [since the cell fired at 5 Hz at point D in Fig. 4C and the cell's maximal firing with GIPSC = 0 was 11 Hz, scattering the IPSC conductances around the mean value represented by point D in a symmetric manner should elevate the firing frequency to the theoretical maximum of (5 + 11)/2 = 8 Hz]. Note also that if the input-output plot of a particular cell is steeper (Fig. 4C, inset), the relative change in firing frequency with increasing variance is also larger (Fig. 4D, inset; the change with increasing variance in this cell goes from 2 to 7 Hz).
Taken together, these results showed that the introduction of IPSC conductances with irregular inter-event intervals and with Gaussian-like amplitude distributions still resulted in the predicted modulatory effects on the average firing rates of the postsynaptic CA1 pyramidal neurons in these dynamic clamp experiments.
Effects of transient alterations in the heterogeneity of GABAergic inputs
The efficacy of inhibition (e.g., in plasticity studies or in epilepsy research) is usually studied by focusing on the changes in the mean amplitude or mean decay of IPSCs. However, based on the previous simulation results (Aradi et al. 2002) and the dynamic clamp data presented above, we expected that neurons should also be able to change their firing rates in response to sudden alterations in the heterogeneity of inputs, even if the mean values are unchanged. Since the sensitivity of principal cells to step-wise alterations in the variance of IPSC peak conductances has not been studied, we performed both dynamic clamp and modeling experiments to study this question. Pyramidal cells received IPSCs conductances at 20 Hz with two peak amplitudes (small and large, around the mean), and the effects of changes in the variance in these IPSC conductances were tested with CV = 0.25, 0.5, or 0.75. In these experiments, the small and large conductances were injected in a random order but they appeared with equal probability to ensure that the mean peak conductance did not change over time (the CV was changed from low CV = 0.25 to high CV = 0.5 or 0.75 for 1 s). The cells received a tonic depolarizing current input so that, in the absence of GIPSCs, they fired repetitive action potentials (in the case of CA1 cells) or bursts of action potentials (in the case of CA3 cells; see insets in Fig. 5, C and D; note that the bursts were smaller in the model cells compared with the biological CA3 pyramidal cells). Injections of IPSC conductances decreased (or suppressed) the tonic firing in CA1 cells (see the 1st second of the traces in Fig. 5, A and B), whereas the injected GIPSCs switched the firing of CA3 cells from bursting to single action potential firing (see the 1st second of the traces in Fig. 5, C and D; naturally, if the GIPSCs were increased further, the firing of single action potentials was also suppressed in CA3 cells as well).
The results demonstrated that both the biological (Fig. 5A; n = 8) and the model (Fig. 5B) CA1 cells increased their firing rate during the transition from low to high levels of variability in the incoming IPSC peak conductances. These results, obtained in this more dynamic setting, agree with the results presented in Fig. 2, A–D, i.e., that if the mean IPSC conductance is large enough to decrease (or completely suppress) the postsynaptic firing, increased variance in the peak conductance values in a population of inputs will enhance firing rates (as in Fig. 2B). Interestingly, both the biological (Fig. 5C; n = 9) and the simulated (Fig. 5D) CA3 pyramidal cells showed an increase in their firing rates if the change in CV was small (1st step in CV in Fig. 5, C and D), similar to CA1 cells. However, when the step in CV was large enough (2nd step in CV in Fig. 5, C and D), CA3 cells responded with bursts of action potentials. The reason for the burst firing during the high-CV period is that small IPSC conductances (especially if they arrive in clusters) allow the bursts to re-appear (note that, as explained above, the CA3 cells fired recurrent bursts without the IPSCs; Fig. 5, C and D, insets). Taken together, these dynamic clamp and modeling results show that CA1 cells and CA3 cells can detect sudden changes in GABAergic heterogeneity.
Modulation of pyramidal cell discharges by variability in IPSC conductances during -
oscillations
GABAergic inputs to principal cells play a central role in the generation of (5–10 Hz) and (40 Hz) oscillations (for reviews, see Freund and Buzsaki 1996; McBain and Fisahn 2001; Traub et al. 1998). In the next series of experiments, the modulation of principal cell firing was investigated during - oscillations, simulated by the injection of a sinusoidal 5-Hz depolarizing current (Idepol in Fig. 6A, top left panel) and IPSC conductances with either constant (CV = 0 in Fig. 6A, middle left panel) or variable (CV = 0.75 in Fig. 6A, bottom left panel) peak amplitudes (note that the injection of the variable inputs was done using 2 peak conductance values that appeared in a random fashion, but the mean conductance with CV = 0 and 0.75 was unchanged). IPSC conductances were delivered at a constant frequency of 40 Hz to simulate frequency events (Traub et al. 1998). Similar to the previous experiments with steady-state depolarization, first the input-output plots were generated with CV = 0 to locate the critical region by determining the mean IPSC conductance at which the pyramidal cell firing was abolished (Fig. 6A, middle panel). Next, the variance in IPSC conductance was introduced within a range of GIPSC values spanning several nano Siemens (nS) (Fig. 6C) around the critical region.
Without the IPSCs, the sinusoidal depolarizing current injection produced clusters of spikes during the depolarizing phase of the oscillations in both the biological and the model CA1 pyramidal cells (Fig. 6A, top row). Addition of the IPSC conductances with CV = 0 abolished the firing of the pyramidal cell (Fig. 6A, middle row). Increased variance in the inputs (CV = 0.75) resulted in discharges of variable numbers of action potentials on the depolarizing phase of (Fig. 6A, bottom row; note that these recordings look remarkably similar to the intracellular - oscillations recorded in vivo) (Soltesz and Deschênes 1993; Ylinen et al. 1995). Figure 6B shows the input-output plot for the biological cell shown in Fig. 6A, i.e., the effect of the increasing mean GIPSC with CV = 0 on the cell's average firing rate. As predicted from the input-output plot, introducing variance around a sufficiently large mean IPSC conductance enhanced the average firing rate of the postsynaptic cell (Fig. 6C). These experiments showed that heterogeneity in the GABAergic inputs could powerfully modulate the output of pyramidal cells (n = 6) even in the dynamic setting provided by - oscillations. As shown in Fig. 6C, variance effects were seen within a range of mean GIPSC (5–12.5 nS in Fig 6C) and CV (0 to 0.75) values. In addition, since large or small IPSCs occur when more or fewer presynaptic inputs discharge together during one cycle, these results also suggest that CA1 pyramidal cells are sensitive to subtle changes in interneuronal synchrony during - oscillations.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
The functional importance of the large synapse-to-synapse and cell-to-cell variability in the GABAergic system is not well understood. Here we used dynamic clamp techniques to systematically change the variance and/or mean of the GIPSC peak and decay values in populations of GABAergic inputs to CA1 (and CA3) pyramidal cells under a variety of conditions, to determine if the rules derived from simulation studies also apply to biological neurons.
Changes in interneuronal heterogeneity may occur during development (with the arrival of new interneuronal subtypes), after seizures and trauma (loss of specific interneuronal groups), or during periods of GABAergic plasticity. While these alterations in GABAergic input variance are interesting and potentially important, it is likely that during the normal functioning of neuronal networks, the variance of the various input parameters (e.g., the GIPSCs of spontaneous events) does not change significantly. However, our dynamic clamp and modeling results show that the effects of input heterogeneity depend on the postsynaptic factors such as the level of excitation that the cell receives and on EGABA, both of which may undergo frequent, activity-dependent fluctuations (Kaila et al. 1997; Staley et al. 1995). Therefore even if the variance in the GABAergic inputs remains constant, the effects of the input heterogeneity on the principal cell's firing rates can still change on a moment-to-moment basis depending on the functional state of the postsynaptic neuron. For example, depolarization-hyperpolarization episodes cause shifting of the contours in the manner indicated by the series of plots shown in Fig. 3B, resulting in dramatic changes in the modulatory effects of GABAergic synaptic inputs on cell discharge rates, even if the mean and CV of GIPSCs remain unaltered.
The GIPSC mean and CV values around which the effects on firing rates were observed were within physiologically relevant ranges. For example, in the experiments in Fig. 2A, the GIPSC input was strong enough to suppress firing at around 14 nS, which is 
15 times the experimentally measured mean mIPSC peak conductance (meanexp = 0.896 nS; see METHODS and Aradi et al. 2002). If one considers meanexp to be the conductance of a single GABAergic synapse and takes into account the fact that a single basket cell makes about 7–15 synapses on the perisomatic region of a single pyramidal cell (Freund and Buzsaki 1996), the 14-nS conductance input could be induced by the firing of one or two basket cells. The CV values that were studied were also close to the experimentally observed GIPSC peak amplitude and decay CV values (CVexp for both is around 0.5; see METHODS). Furthermore, the variance effects were similar on both the tonic, sustained firing evoked by seconds-long depolarizing current pulses (Fig. 2) and on the intermittent, oscillatory firing patterns observable in intracellular recordings from CA1 pyramidal cells in vivo (Fig. 6) (Soltesz and Deschênes 1993).
Detection of changes in input heterogeneity by CA1 and CA3 cells
For simplicity, in the first set of experiments, we used alternating sequences of small and large GIPSCs. In the subsequent experiments, more realistic patterns of inputs were also studied, including small and large GIPSCs in random order, sudden alterations in the CV of inputs, GIPSCs drawn from Gaussian-like (discrete) distributions (note that although mIPSC and sIPSC distributions are skewed toward larger events, they can be adequately fitted with the sum of multiple Gaussians; e.g., Soltesz et al. 1995), as well as -frequency inputs. Although the basic principles outlined in Fig. 1 could predict and explain all the experimental results, our new data also revealed intriguing differences in the effects of alterations in input heterogeneity between CA1 and CA3 pyramidal cells. Specifically, the data showed that CA1 cells responded with alterations in the firing rates to changes in the CV of GIPSCs, whereas CA3 cells changed their firing mode from single spiking to bursting. It is an interesting possibility, to be followed up in future studies, that one of the functions of burst firing in CA3 cells (Kepecs et al. 2002) may be to act as variance detectors for incoming GABAergic inputs.
Functional implications
Recent studies have revealed important functional roles of neuronal "noise" (Chance et al. 2002; Fellous et al. 2003; Prescott and De Koninck 2003), originating from glutamatergic and GABAergic synaptic and extrasynaptic receptors, as well as other sources. This study complements these results by focusing on the heterogeneous nature of the fast, GABAA receptor–mediated, perisomatic inputs originating from basket and axo-axonic cells (Soltesz et al. 1995) in both simple and dynamic settings. Although the focus in this paper was on the modulation of cell firing rates (and, in the case of CA3 cells, firing patterns) by GABAergic inputs as a major physiologically and pathophysiologically relevant endpoint, the insights provided by these experiments and simulations will also be relevant for studies of variance effects on GABAergic functions other than pure inhibition (e.g., synchronization of populations of postsynaptic neurons) (Cobb et al. 1995; Lytton and Sejnowski 1991; Soltesz and Deschênes 1993; Traub et al. 1998; White et al. 2000), particularly since heterogeneity has been shown to modulate interneuronal synchrony (Aradi and Soltesz 2002; Bartos et al. 2001; Golomb and Rinzel 1993; Tiesinga and Jose 2000; Wang and Buzsaki 1996). Given the powerful nature of the perisomatic GABAergic inputs to principal cells (Geiger et al. 1997; Gulyas et al. 1993), understanding the role of cell-to-cell variability observed in perisomatically projecting interneuronal populations (Losonczy et al. 2002; Parra et al. 1998; Pawelzik et al. 2002) and the synapse-to-synapse variability described both in anatomical and physiological studies (Nusser et al. 1997; Soltesz et al. 1995) is likely to reveal new mechanisms underlying the complex dynamics of interneuronal-principal cell networks.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
GRANTS
This work was supported by National Institute of Neurological Disorders and Stroke Grant NS-35915 to I. Soltesz.
|
FOOTNOTES |
|---|
* I. Aradi and V. Santhakumar contributed equally to this study. 
1 See also Additional Supplementary Text available online at http://jn.physiology.org/cgi/content/full/00916.2003/DC1.
Address for reprint requests and other correspondence: V. Santhakumar, Dept. of Anatomy and Neurobiology, Univ. of California, Irvine, CA 92697-1280 (E-mail: vsanthak{at}uci.edu).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Aradi I and Soltesz I. Modulation of network behaviour by changes in variance in interneuronal properties. J Physiol 538: 227-251, 2002.
Bartos M, Vida I, Frotscher M, Geiger JR, and Jonas P. Rapid signaling at inhibitory synapses in a dentate gyrus interneuron network. J Neurosci 21: 2687-2698, 2001.
Chance FS, Abbott LF, and Reyes AD. Gain modulation from background synaptic input. Neuron 35: 773-782, 2002.[CrossRef][ISI][Medline]
Chen K, Aradi I, Thon N, Eghbal-Ahmadi M, Baram TZ, and Soltesz I. Persistently modified h-channels after complex febrile seizures convert the seizure-induced enhancement of inhibition to hyperexcitability. Nat Med 7: 331-337, 2001.[CrossRef][ISI][Medline]
Chen K, Baram TZ, and Soltesz I. Febrile seizures in the developing brain result in persistent modification of neuronal excitability in limbic circuits. Nat Med 5: 888-894, 1999.[CrossRef][ISI][Medline]
Cobb SR, Buhl EH, Halasy K, Paulsen O, and Somogyi P. Synchronization of neuronal activity in hippocampus by individual GABAergic interneurons. Nature 378: 75-78, 1995.[CrossRef][Medline]
Cossart R, Hirsch JC, Cannon RC, Dinoncourt C, Wheal HV, Ben-Ari Y, Esclapez M, and Bernard C. Distribution of spontaneous currents along the somato-dendritic axis of rat hippocampal CA1 pyramidal neurons. Neuroscience 99: 593-603, 2000.[CrossRef][ISI][Medline]
Fellous JM, Rudolph M, Destexhe A, and Sejnowski TJ. Synaptic background noise controls the input/output characteristics of single cells in an in vitro model of in vivo activity. Neuroscience 122: 811-829, 2003.[CrossRef][ISI][Medline]
Freund TF. Interneuron diversity series: rhythm and mood in perisomatic inhibition. Trends Neurosci 26: 489-495, 2003.[CrossRef][ISI][Medline]
Freund TF and Buzsaki G. Interneurons of the hippocampus. Hippocampus 6: 347-470, 1996.[CrossRef][ISI][Medline]
Geiger JR, Lubke J, Roth A, Frotscher M, and Jonas P. Submillisecond AMPA receptor-mediated signaling at a principal neuron-interneuron synapse. Neuron 18: 1009-1023, 1997.[CrossRef][ISI][Medline]
Golomb D and Rinzel J. Dynamics of globally coupled inhibitory neurons with heterogeneity. Phys Rev 48: 4810-4814, 1993.
Gulyas AI, Miles R, Hajos N, and Freund TF. Precision and variability in postsynaptic target selection of inhibitory cells in the hippocampal CA3 region. Eur J Neurosci 5: 1729-1751, 1993.[CrossRef][ISI][Medline]
Hines ML and Carnevale NT. The NEURON simulation environment. Neural Comput 9: 1179-1209, 1997.[Abstract]
Kaila K, Lamsa K, Smirnov S, Taira T, and Voipio J. Long-lasting GABA-mediated depolarization evoked by high-frequency stimulation in pyramidal neurons of rat hippocampal slice is attributable to a network-driven, bicarbonate-dependent K+ transient. J Neurosci 17: 7662-7672, 1997.
Kepecs A, Wang XJ, and Lisman J. Bursting neurons signal input slope. J Neurosci 22: 9053-9062, 2002.
Kleppe IC and Robinson HP. Determining the activation time course of synaptic AMPA receptors from openings of colocalized NMDA receptors. Biophys J 77: 1418-1427, 1999.
Losonczy A, Zhang L, Shigemoto R, Somogyi P, and Nusser Z. Cell type dependence and variability in the short-term plasticity of EPSCs in identified mouse hippocampal interneurones. J Physiol 542: 193-210, 2002.
Lytton WW and Sejnowski TJ. Simulations of cortical pyramidal neurons synchronized by inhibitory interneurons. J Neurophysiol 66: 1059-1079, 1991.
McBain CJ and Fisahn A. Interneurons unbound. Nat Rev Neurosci 2: 11-23, 2001.[ISI][Medline]
Miles R, Toth K, Gulyas AI, Hajos N, and Freund TF. Differences between somatic and dendritic inhibition in the hippocampus. Neuron 16: 815-823, 1996.[CrossRef][ISI][Medline]
Nusser Z, Cull-Candy S, and Farrant M. Differences in synaptic GABA(A) receptor number underlie variation in GABA mini amplitude. Neuron 19: 697-709, 1997.[CrossRef][ISI][Medline]
Pare D, Shink E, Gaudreau H, Destexhe A, and Lang EJ. Impact of spontaneous synaptic activity on the resting properties of cat neocortical pyramidal neurons In vivo. J Neurophysiol 79: 1450-1460, 1998.
Parra P, Gulyas AI, and Miles R. How many subtypes of inhibitory cells in the hippocampus? Neuron 20: 983-993, 1998.[CrossRef][ISI][Medline]
Pawelzik H, Hughes DI, and Thomson AM. Physiological and morphological diversity of immunocytochemically defined parvalbumin- and cholecystokinin-positive interneurones in CA1 of the adult rat hippocampus. J Comp Neurol 443: 346-367, 2002.[CrossRef][ISI][Medline]
Prescott SA and De Koninck Y. Gain control of firing rate by shunting inhibition: roles of synaptic noise and dendritic saturation. Proc Natl Acad Sci USA 100: 2076-2081, 2003.
Sharp AA, O'Neil MB, Abbott LF, and Marder E. The dynamic clamp: artificial conductances in biological neurons. Trends Neurosci 16: 389-394, 1993.[CrossRef][ISI][Medline]
Soltesz I and Deschênes M. Low- and high-frequency membrane potential oscillations during theta activity in CA1 and CA3 pyramidal neurons of the rat hippocampus under ketamine-xylazine anesthesia. J Neurophysiol 70: 97-116, 1993.
Soltesz I, Smetters DK, and Mody I. Tonic inhibition originates from synapses close to the soma. Neuron 14: 1273-1283, 1995.[CrossRef][ISI][Medline]
Staley KJ, Soldo BL, and Proctor WR. Ionic mechanisms of neuronal excitation by inhibitory GABAA receptors. Science 269: 977-981, 1995.
Tiesinga PH and Jose JV. Robust gamma oscillations in networks of inhibitory hippocampal interneurons. Network 11: 1-23, 2000.[ISI][Medline]
Traub RD, Spruston N, Soltesz I, Konnerth A, Whittington MA, and Jefferys GR. Gamma-frequency oscillations: a neuronal population phenomenon, regulated by synaptic and intrinsic cellular processes, and inducing synaptic plasticity. Prog Neurobiol 55: 563-575, 1998.[CrossRef][ISI][Medline]
Traub RD, Wong RK, Miles R, and Michelson H. A model of a CA3 hippocampal pyramidal neuron incorporating voltage-clamp data on intrinsic conductances. J Neurophysiol 66: 635-650, 1991.
Wang XJ and Buzsaki G. Gamma oscillation by synaptic inhibition in a hippocampal interneuronal network model. J Neurosci 16: 6402-6413, 1996.
White JA, Banks MI, Pearce RA, and Kopell NJ. Networks of interneurons with fast and slow gamma-aminobutyric acid type A (GABAA) kinetics provide substrate for mixed gamma-theta rhythm. Proc Natl Acad Sci USA 97: 8128-8133, 2000.
Ylinen A, Soltesz I, Bragin A, Penttonen M, Sik A, and Buzsaki G. Intracellular correlates of hippocampal theta rhythm in identified pyramidal cells, granule cells, and basket cells. Hippocampus 5: 78-90, 1995.[CrossRef][ISI][Medline]
This article has been cited by other articles:
|
|
 |
|
  D. Parker and S. Bevan Modulation of Cellular and Synaptic Variability in the Lamprey Spinal Cord J Neurophysiol, January 1, 2007; 97(1): 44 - 56. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.-M. Goaillard and E. Marder Dynamic Clamp Analyses of Cardiac, Endocrine, and Neural Function Physiology, June 1, 2006; 21(3): 197 - 207. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Foldy, J. Dyhrfjeld-Johnsen, and I. Soltesz Structure of cortical microcircuit theory J. Physiol., January 1, 2005; 562(1): 47 - 54. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. Aradi and G. Maccaferri Cell Type-Specific Synaptic Dynamics of Synchronized Bursting in the Juvenile CA3 Rat Hippocampus J. Neurosci., October 27, 2004; 24(43): 9681 - 9692. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||
