| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
NEUROSCIENCE |
1 Department of Physiology, Feinberg School of Medicine
2 Institute for Neuroscience, Northwestern University, Chicago, IL 60611, USA
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
(Received 25 October 2006;
accepted after revision 10 November 2006;
first published online 16 November 2006)
Corresponding author G. Maccaferri: Department of Physiology, Northwestern University, Feinberg Medical School, 303 E Chicago Ave, Tarry Blg Rm 5-707 M211, Chicago, IL 60611, USA. Email: g-maccaferri{at}northwestern.edu
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Interneuronal networks in the stratum lacunosum moleculare of the hippocampus are especially interesting because they have been shown to react to exogenous GABA application with an all-or-none synchronized response (Perkins, 2002), suggesting that electrical coupling may be especially important in the recruitment of the elements of the assembly. Intriguingly, stratum lacunosum moleculare interneurons appear to have a diffuse electrical connectivity. For example, the percentage of double recordings showing electrical coupling in slices has been reported to range between 33% (Zsiros & Maccaferri, 2005) and 85% (Price et al. 2005).
Here, we have designed experiments with three main purposes: (i) to provide evidence for the propagation of unambiguously identified chemical input across coupled elements of the network; (ii) to quantify the degree of propagation in pairs of neurons; and (iii) to assess the role of electrical coupling during a dynamic model of network activity, which is driven by GABAA receptor-mediated excitatory input (Perrault & Avoli, 1989, 1992; Lamsa & Kaila, 1997; Perkins, 2002).
We propose that the interplay between chemical input and electrical synapses enhances divergence within the network and acts as a recurrent circuit, which is an important determinant of population activity in the hippocampus.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Slices were prepared from young rats (P12–P23). Rats were first deeply anaesthetized using isoflurane (
0.5–1 ml and inhalation to effect) in an induction chamber, in compliance with the guidelines provided by the IACUC of Northwestern University and the National Institutes of Health (NIH). The level of anaesthesia was assessed by monitoring the pedal withdrawal reflex and by pinching the tail and ears. Following deep anaesthesia, rats were quickly decapitated and the brain removed from the skull in a small container filled with chilled solution of the following composition (mM): 234 sucrose, 28 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 0.5 CaCl2, 7 MgSO4, 7 glucose, 1 ascorbic acid, 3 pyruvic acid saturated with 95% O2, 5% CO2 at pH 7.4. Both hemispheres of the brain were glued onto the stage of a vibrating microtome (Leica, VT 1000S), submerged with chilled artificial cerebrospinal fluid (ACSF), and sections of 300–400 µm were cut and stored in an incubation chamber at 34–35°C for at least 30 min, then stored at room temperature until use.
Whole-cell recordings
Conventional patch-clamp recordings were performed. Slices were superfused with preheated ACSF maintained at a constant temperature (32–35°C). ACSF was of the following composition: (mM) 130 NaCl, 24 NaHCO3, 3.5 KCl, 1.25 NaH2PO4, 2 CaCl2, 1 MgSO4, 10 glucose saturated with 95% O2, 5% CO2 at pH = 7.4. In order to increase the frequency of spontaneous events, in nine current-clamp experiments with connected pairs, external potassium was raised up to to 7.5 mM. Cells were observed and selected for recording by means of 40x IR immersion DIC objective applied to a direct microscope (Olympus, Japan) equipped with an infrared camera system (DAGE-MTI, Michigan City, IN, USA). Interneurons were selected in the CA1 stratum lacunosum moleculare, according to the same criteria as in Zsiros & Maccaferri (2005). Pipettes were pulled from borosilicate thin glass capillaries (MTW150F-3, WPI) and filled with the appropriate filtered intracellular solution to a 1.5–5 M
final resistance, as detailed below. Recordings were carried out using a Multiclamp 700 amplifier (Molecular Devices, Sunnyvale, CA, USA). Data were filtered at 3 KHz and digitized at 10–20 kHz using a Digidata A/D board and the Clampex 9 program suite (Molecular Devices). Series resistances were constantly monitored by injecting a- 5/10 mV step in voltage-clamp or a –50/100 pA pulse in current-clamp configuration. Series resistances were not compensated in voltage-clamp configuration, whereas they were balanced via a bridge circuit in current-clamp mode. Liquid junction potential (Vjp; by convention in the direction of the batch relative to the pipette: Vjp
=
Vbath
–
Vpipette) was experimentally estimated (Neher, 1992; Figl et al. 2004) and subtracted offline, so that for voltage-clamp experiments Vholding
=
Vcommand
–
Vjp and for current-clamp experiments Vmembrane
=
Vrecorded
– Vjp.
Low-chloride solution for current-clamp recordings.
The composition of this solution was (mM): 125 K-methylsulphate, 4 ATP-Mg2, 4 NaCl, 0.3 GTP, 16 KHCO3 equilibrated with 95% O2, 5% CO2 to a pH 7.3. Estimated reversal potential for GABAA (as detailed in Aradi & Maccaferri, 2004) was –80 mV. Vjp was +10 mV.
High-chloride solution for current-clamp recordings.
The compostition of this solution was (mM): 125 KCl, 4 ATP-Mg2, 10 NaCl, 0.3 GTP, 16 KHCO3 equilibrated with 95% O2, 5% CO2 to a pH 7.3. Estimated reversal potential for GABAA was –2 mV. For the purpose of simplicity and comparison with voltage-clamp recordings reversal potential for GABAA is considered to be 0 mV in the Result section. Vjp was +5 mV.
Low-chloride solution for voltage-clamp recordings.
The composition of this solution was (mM): 125 K-methylsulphate, 4 ATP-Mg2, 4 NaCl, 0.3 GTP, 16 KHCO3, 5 N-2(2,6-dimethylphenylcarbamoylmethyl) triethylammonium chloride (QX-314-Cl) and 0.2% biocytin equilibrated with 95% O2, 5% CO2 to a pH 7.3. Estimated reversal potential for GABAA was –66 mV. QX-314 was included in the intracellular solution at high concentration in order to block voltage-dependent conductances and GABAB receptor-operated potassium currents. Vjp was +10 mV.
High-chloride solution for voltage-clamp recordings.
The composition of this solution was (mM): 125 KCl (or CsCl), 4 ATP-Mg2, 10 NaCl, 0.3 GTP, 16 KHCO3, 5 N-2(2,6-dimethylphenylcarbamoylmethyl) triethylammonium chloride (QX-314-Cl) and 0.2% biocytin equilibrated with 95% O2, 5% CO2 to a pH 7.3. Estimated reversal potential for GABAA was 0 mV. QX-314 was included in the intracellular solution at high concentration in order to block voltage-dependent conductances and GABAB receptor-operated potassium currents. Vjp was +5 mV.
Holding potential for voltage-clamp recordings. The holding potential for voltage-clamp experiments was set to –15 mV in the experiments shown in Figs 1, 4, and 6D.
|
|
|
|
In order to avoid the potential disruption of intracellular gradients caused by whole-cell recording configuration, non-invasive measurement of membrane potential was performed in cell-attached mode, similarly to what is described in Morikawa et al. (2003). Briefly, low-resistance electrodes (1.5 M) when filled with ACSF were used in cell-attached configuration. After verification of the formation of a tight seal (>2 G) in voltage-clamp configuration, the amplifier was switched to I
= 0 mode, and the membrane potential was recorded. Cell-attached configuration was then monitored by switching back to voltage-clamp mode and by delivering voltage steps (250 ms duration, –10 mV amplitude) from a holding potential of –60 mV. Pressure-induced or spontaneous breakthrough of this configuration was also easily detected in current-clamp mode as an abrupt depolarization of the membrane potential.
Evaluation of electrical coupling and junctional conductance in paired current-clamp experiments
We tested for the presence of gap junctions between pairs by repetitively injecting a 500 ms current step of –100 pA. Cells were considered coupled if they had a DC coefficient coupling (ratio of the voltage deflection in the non-injected cell to the voltage deflection in the injected cell measured during the last 50 ms of the stimulus) value higher than 0.005, and the shape of the electrotonic response was recognizable in the recording from the non-injected cell. The synaptic coupling coefficient was calculated as the ratio of the peak of the postsynaptic potential recorded in the cell with the low-chloride solution to the peak of the synchronous event measured in the chloride-loaded neuron. The effective junctional conductance (Gj) was calculated according to the equation
|
| (1) |
Drug preparation and application in slices
All the following drugs were from Tocris Biosciences. SR-95531 (gabazine) was used at 12.5 µM. The powder was dissolved in water as a stock solution at 25 mM, aliquoted in 100 µl vials, and frozen at –20°C. D–AP5 was dissolved in water, stocked frozen in 100 µl aliquots at 50 mM and used at 50 µM. Stock solutions of NBQX were dissolved in DMSO at 100 mM, aliquoted in 40 µl vials and frozen at –20C. The final concentration used was 20 µM. CGP55845 was dissolved in DMSO in stock solutions at 100 mM, then aliquoted and used at a final concentration of 1–5 µM. Carbenoxolone was obtained from Sigma as a disodium salt, and was dissolved directly into the recording solution to a final concentration of either 100 or 500 µM. 4-Aminopyridine was from Sigma and was also dissolved directly into the recording solution to its final concentration of 50–100 µM.
Analysis of spontaneous events in single cell recordings
Spontaneous currents recorded from single neurons held at depolarized potentials were analysed using the Clampfit 9.0 (Molecular Devices), Origin Pro7.0 (OriginLab, Northampton, MA, USA), and Microsoft Excel suites of programs. Events were first collected using the template-based analysis feature of Clampfit, and then reviewed by visual inspection. However, under our experimental conditions at depolarized potentials, a significant fraction of the events were below detection levels, leading to potential ambiguities in the interpretation of measured amplitude and frequency (Stell & Mody, 2002). Therefore, we used the ‘largest-amplitude count matching’ method developed by Stell & Mody (2002) to build the I–V relationship in Fig. 2. The number of events matched for inward currents was 146 ± 40 (n = 7) and 10 ± 1 (n = 7) for outward events. Largest-amplitude count matching could not be used in the pharmacological analysis of Figs 5 and 6 because of the reduced number of observations due to the effect of the drugs themselves. Therefore, we combined amplitude and frequency in a single index as ‘synaptic impact’ (event amplitude multiplied by its frequency, see Galarreta & Hestrin, 1998), which is related to the synaptic effect on average membrane potential, provided that the kinetics of events are unchanged (Tsodyks & Markram, 1997; Abbott et al. 1997).
|
Analysis of synchronous postsynaptic potentials was performed by using the WinEDR and WinWCP software package (courtesy of Dr J. Dempster, University of Strathclyde, UK). The rising phase of the depolarizing events in the cell recorded with high-chloride solution was used to trigger acquisition in both channels. Events were then inspected visually; obviously distorted records were rejected and not used for the analysis.
Analysis of network-driven events in slices exposed to 4-aminopyridine
Spontaneous network currents recorded from single neurons were analysed using the Clampfit 9.0, OriginPro7.0, and Microsoft Excel suites of programs. Events were first collected using the threshold-based analysis feature of Clampfit, reviewed by visual inspection, and aligned by their peaks. I–V plots shown in Fig. 9 were built from spontaneous events collected in 5 min epochs (for example, 8 ± 2 events at holding potential (Vh) =
–65 mV in Fig. 9A, n
= 5 cells, and 7 ± 1 events at Vh
=
–70 mV for Fig. 9C, n
= 5 cells). Analysis of the time course of the effect of carbenoxolone (and long recording) presented in Fig. 11 was performed first by averaging the amplitudes of spontaneous events within the same neuron in 1 min bins and then by averaging each bin across different cells.
|
|
Trains of spikes were evoked by current steps injected through the whole-cell electrode (1 s duration, 100–350 pA amplitude). Spikes collected from 20 sweeps were aligned and averaged in control conditions and after application of gabazine (12.5 µM). Afterhyperpolarizations were measured from action potential threshold (defined as the membrane potential at which dV/dt exceeds 10 V s–1 (Fricker et al. 1999)) to the negative peak of the hyperpolarization. Cell membrane potential was maintained close to the resting potential of the cells between –65 and –70 mV.
Statistical methods
Statistical analysis was performed using the following software packages: Clampfit 9.0, Excel, Origin, Prism (Graphpad, San Diego, CA, USA), Win EDR, and Win WCP. Non-parametric tests (Wilcoxon signed test, abbreviated as W, Mann–Whitney test, abbreviated as M–W, and Kruskal–Wallis test, abbreviated as K–W) were used, as appropriate. Data are shown as mean ± S.E.M.
Computational methods
We performed simulations using the NEURON 5.4 software (Hines & Carnevale, 1997) running on Linux. The model interneurons had a soma and four dendrites to resemble the morphology of multipolar interneurons located in the stratum lacunosum moleculare. We included voltage-activated channels Na+, delayed rectifier K+, and A-type K+ channels (the kinetics of the channels were taken from Aradi & Holmes, 1999; and Yamada et al. 1989). The morphological and biophysical parameters were as follows: (i) soma: diameter = 15 µm, number of compartments = 2, maximal conductance values for sodium conductance (gNa) = 150 mS cm–2; delayed rectifier potassium conductance (gK-DR) = 40 mS cm–2 and A-type potassium channel conductance (gK-A) = 2.55 mS cm–2; (ii) dendrites: length = 100 µm, diameter: proximal= 1.5 µm, distal = 0.2 µm, number of compartments = 20, maximal conductance values for gNa
= 13 mS cm–2, gK-DR
= 4 mS cm–2. Resting membrane potential was set at –65 mV, and membrane specific capacitance was 3 µF cm–2. We used a relatively high value of specific capacitance to compensate for the oversimplified anatomical structure of the model, which includes only four dendrites and is likely to underestimate the global membrane capacitance of the neuron. As shown in Supplementary Fig. 1, the selected value of specific capacitance produces DC and spike coefficient coupling that are similar to those reported experimentally in the literature (compare with Zsiros & Maccaferri, 2005 and Price et al. 2005). Furthermore, our approach should be considered conservative because a more standard value of specific capacitance such as 1 µF cm–2 would enhance coupling in the model (see Supplementary Fig. 1). Voltage-dependent conductances were set to 0 mS cm–2 in the model cell under simulated voltage-clamp conditions to mimic the effect of QX-314. The input resistance for these model neurons was 250 M (Price et al. 2005). Specific and axial resistance were set to 4347.8 cm2 and 210 cm, respectively. Two interneurons were coupled by a junctional conductance (Gj, 0.5 or 2 nS), which connected the soma of cell 1 to the mid-dendrite of cell 2. Gap junction coupling was modelled by a current (IGj) proportional to the potential difference between the two cells at the gap junction site, i.e. IGj1
=
Gj(V2
–
V1) in the first cell, where Gj is the junctional conductance, and IGj2
=
–IGj1 in the second cell. The kinetics of the postsynaptic conductance were described by the following equation:
|
| (2) |
rise) = 0.5 ms and decay time constant (decay) = 15 ms. A time step of 0.01 ms was used in the simulations. |
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
In order to explore potential experimental scenarios, we built a computational model of a small network consisting of two multipolar interneurons (cell 1 and cell 2, Fig. 1A). The two model cells were connected by a gap junction, and both received exclusively GABAA receptor-mediated synaptic input. The basic characteristics of the model neurons were mostly tailored on the available data in the literature for neurogliaform cells (Price et al. 2005; Zsiros & Maccaferri, 2005). Provided that the reversal potential for GABAergic input (EGABA) in the two cells could be set to different values, then voltage clamping one cell of the pair (cell 1) at an intermediate holding potential (set between the two different EGABA values) should distinguish direct synaptic input from gap junction-propagated events because of their different polarities. Figure 1B shows the results of a set of simulations with a holding potential of –15 mV. EGABA was set to 0 mV in cell 1, whereas it was set, alternatively, to –50 mV, –60 mV, and –70 mV in cell 2. Sequential activation of identical GABAergic conductances in the two different model cells resulted in postsynaptic currents in cell 1, but with opposite polarities and different amplitudes.
We directly tested this prediction by recording spontaneous GABAA receptor-mediated synaptic activity from stratum lacunosum moleculare interneurons in hippocampal slices, using a medium containing blockers of glutamatergic and GABAB receptor-mediated synaptic transmission (Fig. 1C). We took advantage of the whole-cell recording configuration, and used a chloride-loaded pipette to manipulate EGABA in the recorded neuron (calculated value: 0 mV, see Methods for details). We reasoned that chloride loading would remain mostly restricted to the cell under voltage clamp, and should not spread significantly to other neurons that are electrically coupled to it. This assumption relies on work showing that: (i) various connexin channels or hemichannels have low chloride relative permeability (Beblo & Veenstra, 1997; Eskandari et al. 2002); (ii) Cx36 channels have undetermined anionic selectivity; and (iii) the lowest conductance reported within the connexin family (Srinivas et al. 1999); and, lastly (iv) that interneurons possess chloride clearance mechanisms, for example, via the KCC2 cotransporter (Gulyas et al. 2001). Therefore, in our experimental conditions, different EGABA values should exist in the recorded cell compared to the unperturbed interneurons of the slice.
The results from 17 recordings confirmed the presence of both inward and outward spontaneous currents (Fig. 1D). When compared to inward currents, outward events were smaller (7.1 ± 0.3 pA versus 12.8 ± 1.0 pA, n = 17, P < 0.05, W-test) and occurred much less frequently (0.4 ± 0.1 Hz versus 4.0 ± 0.7 Hz, n = 17, P < 0.05, W-test). We further quantified these differences by calculating the synaptic impact (event amplitude multiplied by its frequency) for the different polarities. At a holding potential of –15 mV, the synaptic impact of the outward events was 3.4 ± 0.8 pA s–1, compared to 47.8 ± 7.3 pA s–1 for inward currents (n = 17, P < 0.05, W-test). Thus, our experimental results were consistent with the prediction that voltage-clamp recording from a single neuron can reveal synaptic activity spreading from other elements of the network, in addition to direct chemical input.
Next, in order to assess the voltage dependency of inward and outward events, we recorded pharmacologically isolated GABAA receptor-mediated spontaneous activity at different membrane potentials. The frequencies of both inward and outward events were highly voltage dependent in the range tested (Fig. 2A and B, P < 0.05, n = 7 cells, K–W-test). This indicated that a significant proportion of the events fell below threshold detection, and that variations of the driving force at different holding potentials had a relevant impact on the fraction of detectable events (Stell & Mody, 2002). The extrapolated reversal potential of inward currents was –1 mV (Fig. 2A), which was very close to the predicted value of 0 mV. In addition, as expected for direct GABAergic input, the amplitude of inward events was larger at more hyperpolarized membrane potentials. In contrast, outward currents had an estimated reversal potential of –70 mV, and displayed opposite voltage dependency, so that events grew larger at more depolarized holding potentials (Fig. 2B). These experiments suggest that inward currents are the result of direct neurotransmitter release onto the chloride-loaded cell, whereas outward currents reflect events originating from cells with different EGABA values (Lamsa & Taira, 2003; Vida et al. 2006), but electrically coupled to the recorded neuron.
If outward currents were events propagated via gap junctions, then their kinetics would be expected to undergo some form of filtering. We used the same configuration introduced in Fig. 1A and examined the location dependency of the kinetics of the propagated events by varying the position of the chemical synapse along the dendrites. As shown in Fig. 3A–C, location had a strong effect both on amplitude and rise time kinetics. Decay kinetics were also affected, but to a lesser degree. Therefore, we next analysed the kinetic properties of experimentally recorded spontaneous inward and outward currents (Fig. 4A). As expected, we found that inward currents had much faster rise and decay time kinetics. The 20–80% rise time was 0.6 ± 0.0 ms for inward currents, compared to 8.1 ± 1.1 ms for outward events (Fig. 4B, P < 0.05, n = 17 W-test) and the 100–37% decay time was 15.1 ± 1.0 ms for inward events compared to 29.3 ± 2.4 ms for outward currents (Fig. 4B, P < 0.05, n = 17, W-test). The two types of events appeared separated in two distinct clusters, when plotted along their kinetic axis (Fig. 4C). Although the kinetics of averaged outward and inward currents were clearly different, spontaneous outward events with relatively fast time could be occasionally detected (see for example Fig. 6A and D). Possible explanations are: (i) that these events reflected release of neurotransmitter at sites closest to the gap junction, or that (ii) fast events propagated across intersomatic gap junctions, whereas slower events propagated across gap junctions located at more distal dendritic sites.
|
Next, we tested the pharmacological properties of outward currents recorded at depolarized potentials, and compared them to inward currents (Fig. 5A). Both inward and outward events were fully blocked by the GABAA receptor antagonist gabazine. Synaptic impact decreased from 28.4 ± 9.0 pA s–1 in control to 0.0 ± 0.0 pA in the presence of the drug for inward currents (n = 5, P < 0.05, W-test), and from 4.2 ± 1.7 pA s–1 to 0.0 ± 0.0 pA for outward currents (n = 5, P < 0.05, W-test). The sensitivity to gabazine suggests that it is unlikely that outward currents reflect the propagation of action potential afterhyperpolarization from coupled neurons. Nevertheless, we directly tested the gabazine sensitivity of current-evoked spike afterhyperpolarizations recorded from interneurons. Neither amplitude (16.1 ± 1.0 mV in control versus 15.8 ± 1.1 mV in the presence of the drug, n = 10, P > 0.2, W-test), nor the frequency (7.4 ± 1.8 Hz in control versus 8.2 ± 2.0 Hz, n = 10, P > 0.2, W-test) of current-evoked spike afterhyperpolarization was affected by gabazine (Fig. 5B). If inward and outward events truly reflected GABAergic input originating from different electrically coupled cells, then gap junction blockers would be expected to predominantly affect the propagated events. Indeed, as shown in Fig. 6A and B, the gap junction uncoupler carbenoxolone (100 µM) had no significant effect on the synaptic impact of inward events (from 34.9 ± 12.0 pA s–1 in control to 34.8 ± 5.5 pA s–1 after 25 min in the presence of the drug (n = 8, P > 0.2, W-test), whereas outward currents were reduced from 3.3 = 8 ± 1.0 pA s–1 in control to 0.5 ± 0.2 pA s–1 after 25 min of carbenoxolone application (n = 8, P < 0.05, W-test). We also performed control experiments to exclude the possibility of time-dependent run-down (Fig. 6C). After the control period, 25 additional minutes in the absence of any drug did not change significantly the synaptic impact of either inward or outward currents (inward currents: from 69.9 ± 18.2 pA s–1 in control to 76.1 ± 21.9 pA s–1 after the 25 additional minutes (n = 6, P > 0.2, W-test), outward currents: from 1.9 ± 0.6 pA s–1 in control to 1.9 ± 0.4 pA s–1 after the 25 additional minutes (n = 6, P > 0.2, W-test)). Lastly, we studied the effect of carbenoxolone applied to the recorded neuron via the whole-cell pipette, in order to reduce potential unspecific effects in the rest of the network. We compared the synaptic impact of inward and outward currents in two populations of cells recorded with standard electrodes versus carbenoxolone-containing pipettes (100 µM, Fig. 6D). Synaptic impact of inward currents was 46.9 ± 7.0 pA s–1 (n = 15) in cells recorded with standard pipettes versus 37.5 ± 7.6 pA s–1 (n = 12) in neurons recorded with carbenoxolone-loaded electrodes (P > 0.2, M–W-test). In contrast, synaptic impact of outward currents was 3.1 ± 1.0 pA s–1 (n = 15) in cells recorded with control electrodes, but smaller (P < 0.05, M–W-test) in neurons recorded with drug included in the electrodes (0.9 ± 0.5 pA s–1, n = 12). In conclusion, our voltage-clamp experiments show for the first time that, under the appropriate experimental conditions, propagated spontaneous synaptic currents can be identified in single-cell recordings.
Propagation of postsynaptic potentials in electrically coupled neurons
Next, we addressed the question of gap junction-mediated propagation of synaptic events in pairs of electrically coupled interneurons. After blockade of glutamatergic and GABAB receptor-mediated synaptic transmission, we took advantage of experimental conditions that allowed us to distinguish direct synaptic input from events propagated via gap junctions (Fig. 7A–C). The first cell of the pair (cell 1) was recorded with a chloride-loaded electrode (estimated EGABA
= 0 mV, see Methods for details), whereas a low-chloride pipette was used for the second neuron (cell 2, estimated EGABA
=
–80 mV). Both cells were held at –70 mV. Under these conditions, large spontaneous depolarizing events were expected in cell 1 because of the large driving force (i.e. Vm
–
EGABA
=
–70 mV). In contrast, the driving force in cell 2 was expected to be smaller and of opposite direction (+10 mV), hence the observation of low-amplitude hyperpolarizing activity was predicted. In order to study whether events from cell 1 could propagate to cell 2, we used spontaneous events in cell 1 as a trigger to align the corresponding voltage recorded in cell 2. The degree of coupling of synaptic events between the two neurons (syn coupling coefficient) was 0.041 ± 0.006, which compared to 0.059 ± 0.008 for the DC coupling coefficient (Fig. 7D). Thus, paired recordings from electrically coupled interneurons apparently suggested that postsynaptic events can propagate from one neuron to another via gap junctions. Furthermore, we decided to compare propagation of the events in control conditions versus in the presence of the gap junction uncoupler carbenoxolone. As shown in Fig. 8A–C, we verified that external application of carbenoxolone (100 or 500 µM to speed up its blockade) reduced steady-state coupling. DC coupling decreased from 0.056 ± 0.010 to 0.001 ± 0.000 in the presence of the drug (n
= 10, P < 0.05, W-test), and reduced the junctional conductance. The effective junctional conductance in control conditions was estimated to be 422.1 ± 58.7 pS (n
= 13) compared to 16.8 ± 6.6 pS (n
= 10) in the presence of the drug (P < 0.05, M–W-test). After having verified that carbenoxolone was effective in reducing electrical coupling, we observed that propagation of synaptic events was severely depressed (Fig. 8D and F), indicating that the synchronous activity observed in cell 2 was due to spread of postsynaptic potentials via gap junctions. The synaptic coupling coefficient decreased from 0.042 ± 0.006 in control conditions to 0.008 ± 0.004 in the presence of carbenoxolone (Fig. 8F, n
= 10, P < 0.05, W-test). As a further control, we also verified that no propagated activity was detected in pairs of cells that were not electrically coupled. In 19 paired recordings from uncoupled neurons, no propagation was observed (Fig. 8E).
|
|
Next, we decided to study the significance of the propagation of synaptic input in stratum lacunosum moleculare interneuron networks during GABAergic input-driven activity in the absence of ionotropic glutamatergic transmission (Perreault & Avoli, 1989, 1992; Lamsa & Kaila, 1997; Perkins, 2002). In the presence of glutamatergic ionotropic and GABAB receptor antagonists, voltage-clamp recordings from stratum lacunosum moleculare interneurons in slices exposed to 4-aminopyridine (50–100 µM) revealed the spontaneous occurrence of large (amplitude at –65 mV was –3.8 ± 0.8 nA when high-chloride intracellular solution was used, n
= 5 cells) and long-lasting (half-width at –65 mV was 843 ± 103 ms, n
= 5 cells) events, which were strongly depressed by gabazine (Fig. 9A and B). We will refer to these currents as giant postsynaptic currents (giant PSCs). Linear fitting of the experimental current–voltage relationship of giant PSCs obtained from five cells recorded with high-chloride intracellular solution (identical to the one used for the recordings of Fig. 2) yielded a conductance of 61.5 nS (which compared to 1.6 nS for the largest spontaneous events recorded in quiescent slices, as shown in Fig. 2). Thus, a rather conservative estimate indicates that at the peak of 4-aminopyridine-induced network activity, synchronized GABAergic input is 40 times larger than the largest putative unitary input. The reversal potential estimated from the fitting was –3 mV, in reasonable agreement with a predicted value of 0 mV. When we repeated the same experiment using low-chloride pipettes, biphasic currents were recorded close to the reversal potential, similarly to what described in pyramidal neurons (Perkins & Wong, 1996). The estimated reversal potential from five recordings was –35 mV (peaks of biphasic waveforms were averaged), which was largely different from the predicted value of –66 mV (Fig. 9C). Both components of the biphasic currents were strongly reduced by gabazine, as shown in Fig. 9D, n
= 5 cells).
In order to confirm that giant GABAergic events were associated with excitation in interneurons, we performed simultaneous double recordings in whole-cell voltage- and current-clamp configurations. As shown in Fig. 10A, the occurrence of giant spontaneous events in the cell under voltage clamp was mirrored by strong bursting activity in the neuron recorded in current-clamp (n = 5 double recordings). Similar results were obtained using non-invasive techniques, which preserve intracellular chloride gradients (Morikawa et al. 2003). Coupling whole-cell recording from one neuron to cell-attached current-clamp recording in the second cell of the pair confirmed the results obtained with double whole-cell recordings (Fig. 10B). The average resting membrane potential recorded in cell-attached configuration was –67.5 ± 1.9 mV (n = 9). In eight out of nine double recordings, giant PSCs recorded in the cell under voltage clamp were associated with depolarizing responses in the cell recorded in cell-attached configuration, whereas in one cell a biphasic hyperpolarizing–depolarizing response was observed (Fig. 10C).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Mixed polysynaptic pathways
Despite its importance and indication from previous studies in invertebrate and non-cortical systems that this may be the case (for example, Decima, 1969; Martin & Pilar, 1963; Zipser & Bennett, 1976; Marder & Eisen, 1984; Slesinger & Bell, 1985; Pereda et al. 1995; Garcia-Perez et al. 2004; for a review of electrical coupling in the mammalian brain see Connors & Long, 2004; for a general review including older work see Bennett, 1977), the propagation of synaptic signals across gap junctions has yet not been studied in pairs of mammalian cortical neurons, and work has concentrated only on the transmission of spikes or steady-state electrotonic signals. The efficiency of propagation of synaptic potentials in cortical neurons has not been measured directly, and quantification is still lacking. It is important to underscore that the spreading of synaptic events differs from the more commonly studied propagation of artificially generated electrotonic signals or spikes, for at least three reasons. First, gap junction-mediated propagation of electrical signals is influenced by the kinetics of the signal itself (Galarreta & Hestrin, 1999; Price et al. 2005). Therefore, synaptic signals are likely to propagate less efficiently than steady-state voltages, but more efficiently than spikes. Second, the study of synaptic events offers insights on the propagation of electrical signals generated on the entire somatodendritic structure of the neuron, in contrast to artificial signals generated by strictly somatic current injections. Third, artificial electrotonic pulses are not associated with any change in membrane conductance, whereas synaptic events require the opening of synaptic channels and may affect dendritic voltage-dependent conductances.
Our results show that part of the spontaneous synaptic events recorded from a single hippocampal interneuron is the result of neurotransmitter release occurring on another cell. This type of polysynaptic transmission could be generated by the sequential chain of a chemical input that is directly targeting a neuron electrically coupled to another cell (Fig. 12A). We propose to term this network arrangement mixed polysynaptic pathway, because of the cooperation between chemical and electrical synapses. Studies in the neocortex (Simon et al. 2005) and in the hippocampus (Fukuda & Kosaka, 2000) have already demonstrated the close proximity of chemical synapses and gap junctions on the membrane of interneurons, thus identifying the potential structural basis underlying mixed polysynaptic pathways.
|
Our experiments have taken advantage of the biophysical properties of the GABAA receptor, which is mostly permeable to chloride ions (Bormann et al. 1987), and of the whole-cell variation of the patch-clamp techniques, which allows the manipulation of the ionic conditions of the recorded neurons. It is important to emphasize that chloride manipulation was essential to allow the unambiguous dissociation of direct versus propagated input. Indeed, the identification of propagated events was strengthened by the simultaneous requirements of: (i) a polarity incompatible with the one predicted for synaptic events generated by direct neurotransmitter release; and (ii) pharmacological sensitivity to the gap junction uncoupler carbenoxolone. Although the specificity of carbenoxolone for gap junctions is still controversial (for work reporting: (i) lack of gap junction-unrelated effects onto the excitability of pyramidal cells in hippocampal slices see Kohling et al. 2001; Margineanu & Klitgaard, 2001; Schmitz et al. 2001; (ii) lack of direct effects on membrane properties and excitability of hippocampal interneurons see Yang & Michelson, 2001; (iii) increased excitability of pyramidal cells in hippocampal slices see Jahromi et al. 2002; (iv) depression of excitability of pyramidal cells in hippocampal cultures see Rouach et al. 2004), we believe that unspecific effects on neuronal membrane excitability, if present at all, should not significantly affect the interpretation of our results for the following reasons. First, carbenoxolone does not change the synaptic impact of GABAergic inward currents recorded at depolarized potentials, whereas it strongly depresses outward currents. Indeed, although bath application of carbenoxolone slightly increased the synaptic impact of inward currents, this increase was not statistically significant. If carbenoxolone depressed the excitability of the GABAergic network under study, this should be reflected by a decrease in the synaptic impact of all the spontaneous events recorded (both inward and outward), in contrast with our observation. Second, if carbenoxolone affected the intrinsic properties of the GABAergic network under study, external application to the entire slice would be expected to have largely different effects compared to a more limited exposure such as intracellular loading via a recording pipette containing QX-314. In contrast, our experimental results show similar results (no effect on synaptic impact of inward current, strong depression of synaptic impact of outward currents). Third, the time course of the effect of carbenoxolone on outward currents is very similar to the one reported by Zsiros & Maccaferri (2005), which was associated with the direct demonstration of blockade of electrical coupling in paired recordings.
Propagated GABAergic input to the stratum lacunosum moleculare in quiescent and 4-aminopyridine-exposed slices
In order to boost the signal-to-noise ratio and to unambiguously distinguish propagated from direct input, our observations were made under elevated intracellular chloride conditions, and at depolarized potentials in voltage-clamp experiments, which may overestimate the importance of propagated input in a quiescent slice where EGABA is potentially close to resting potentials. We did not directly address this point because of the promiscuity of electrical coupling in the stratum lacunosum moleculare network (Zsiros & Maccaferri, 2005), which would require measurements of EGABA for every different subtype of interneuron participating in the network. Nevertheless, the observation of outward currents places limit to EGABA in a subpopulation of coupled neurons, suggesting that, at least in a fraction of the coupled cells, EGABA is likely to be close to resting potentials (estimated reversal for outward currents was –70 mV). In conclusion, the significance of propagated events in quiescent slices appears to be somewhat modest, as indicated by the fact that the synaptic impact of outward events recorded under favourable conditions was much smaller than the one of inward currents. However, three main considerations need to be highlighted. First, we evaluated gap junction-mediated propagation of synaptic events by somatic recordings, therefore potentially underestimating propagation to domains directly adjacent to dendritic gap junctions (Simon et al. 2005). Therefore, the degree of attenuation that we observed at the soma may not reflect propagated potentials at the dendrites, which could still play a significant role in the integration of local signals (Polsky et al. 2004). Technical limitations prevent a direct exploration of this issue: paired recording from membrane domains directly adjacent to the unknown site of the gap junction(s) would be required. Second, EGABA is a dynamic variable, which can be modulated by the state of the network (Thompson & Gahwiler, 1989; Lamsa & Taira, 2003; but see also Perkins, 1999). Our experimental determination of the reversal potential for giant PSCs during dynamic network states suggests that synchronized giant GABAergic input onto stratum lacunosum moleculare interneurons is clearly excitatory. This interpretation was further strengthened by the converging results obtained with whole-cell and non-invasive cell-attached recordings, showing the synchronous occurrence of bursting/depolarizing responses and giant PSCs in the interneuronal network. These results explain very well the finding of Perkins (2002) that exogenous GABA puffed onto the stratum lacunosum moleculare activates an interneuron network. Third, although no quantitative work is available for interneurons located in the stratum lacunosum-moleculare of the hippocampus, studies in the neocortex suggest that individual interneurons may be electrically coupled to at least cells (Amitai et al. 2002; Fukuda et al. 2006). Therefore, the summation of depolarizing propagated events from many cells during synchronous activity could further enhance their impact on the target neuron.
We have directly addressed this point in a model of network synchronization that depends predominantly on GABAA receptor-mediated synaptic transmission. It is important to note that blockade of gap junctions under these dynamic conditions would be expected to eliminate an electrical coupling-based recurrent pathway spreading excitation (Fig. 12B and C). Indeed, our results show that synchronized giant PSCs recorded from stratum lacunosum moleculare interneurons are critically regulated by electrical coupling. Thus, it is important to underscore that our ‘quiescent slice’ approach was essential for the precise quantification and study of the biophysical properties underlying propagation of synaptic potentials, but that the significance of this mechanism may be especially related to dynamic network states. For example, a similar this type of network synchronization has been suggested to trigger the initiation of seizures in patients affected by Taylor's-type focal cortical dysplasia (Taylor et al. 1971; Avoli, 1996; D'Antuono et al. 2004). Furthermore, GABAergic mediated network hyperexcitability could spread across brain regions affected by loss of principal neurons, and with disrupted excitatory synaptic input such as mesial limbic structures of temporal lobe epilepsy patients with Ammon's horn sclerosis (Gloor, 1991). Intriguingly, 4-amynopiridine-induced experimental seizures in vivo have been reported to increase the expression of messenger RNA for various types of connexins in primary epileptic foci (Gajda et al. 2003), suggesting the possibility that positive feedback between epileptiform activity and the strength of electrical coupling might contribute to maintaining epileptic conditions. However, this effect was described in vivo after at least one hour of epileptic activity (Gajda et al. 2003), whereas our in vitro baseline exposure to 4-aminopyridine was much more limited before application of carbenoxolone. This suggests that this mechanism is unlikely to have played a major role under our experimental conditions.
Conclusions
In summary, we have studied the integration of synaptic potentials and electrical coupling in stratum lacunosum moleculare interneurons of the hippocampus. The unambiguous identification of chemically mediated versus electrically propagated events has allowed us to give a quantitative estimate of the synaptic coupling in pairs of neurons. This mechanism is likely to play an important role in normal synchronized states of the network, and may contribute to epileptic seizures.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Amitai Y, Gibson JR, Beierlein M, Patrick S, Ho AM, Connors BW & Golomb D (2002). The spatial dimensions of electrically coupled networks of interneurons in the neocortex. J Neurosci 22, 4142–4152.
Aradi I & Holmes WR (1999). Role of multiple calcium and calcium-dependent conductances in regulation of hippocampal dentate granule cell excitability. J Comput Neurosci 6, 215–235.[CrossRef][Medline]
Aradi I & Maccaferri G (2004). Cell type-specific synaptic dynamics of synchronized bursting in the juvenile CA3 rat hippocampus. J Neurosci 24, 9681–9692.
distal 1.5 
