Cortical oscillatory dynamics and benzodiazepine-site modulation of tonic inhibition in fast spiking interneurons

Tonic conductance mediated by extrasynaptic GABA A receptors has been implicated in the modulation of network oscillatory activity. Using an in vitro brain slice to produce oscillatory activity and a kinetic model of GABA A receptor dynamics, we show that changes in tonic inhibitory input to fast spiking interneurons underlie benzodiazepine-site mediated modulation of neuronal network synchrony in rat primary motor cortex. We found that low concentrations (10 nM) of the benzodiazepine site agonist, zolpidem, reduced the power of pharmacologically-induced beta-frequency (15 e 30 Hz) oscillatory activity. By contrast, higher doses augmented beta power. Application of the antagonist, ﬂ umazenil, also increased beta power suggesting endogenous modulation of the benzodiazepine binding site. Voltage-clamp experiments revealed that pharmacologically-induced rhythmic inhibitory postsynaptic currents were reduced by 10 nM zolpidem, suggesting an action on inhibitory interneurons. Further voltage-clamp studies of fast spiking cells showed that 10 nM zolpidem augmented a tonic inhibitory GABA A receptor mediated current in fast spiking cells whilst higher concentrations of zolpidem reduced the tonic current. A kinetic model of zolpidem-sensitive GABA A receptors suggested that incubation with 10 nM zolpidem resulted in a high proportion of GABA A receptors locked in a kinetically slow desensitized state whilst 30 nM zolpidem favoured rapid transition into and out of desensitized states. This was con ﬁ rmed experimentally using a challenge with saturating concentrations of GABA. Selective modulation of an interneuron-speci ﬁ c tonic current may underlie the reversal of cognitive and motor de ﬁ cits afforded by low-dose zolpidem in neuropathological states. © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http

As with gamma (35e80 Hz) oscillations (Whittington et al., 1995;Traub et al., 1996), beta oscillatory activity in M1 is generated as a consequence of sustained excitation of networks of inhibitory interneurons (Yamawaki et al., 2008).Inhibitory interneurons are able to entrain each other to fire in a synchronous manner and hence sculpt pyramidal cell activity through repetitive (phasic) inhibitory discharges, the frequency of which is dependent upon the kinetics of the inhibitory postsynaptic potentials (Whittington et al., 1995;Fisahn et al., 1998;Traub et al., 2003).Recent evidence suggests that a sustained inhibitory membrane conductance (I tonic ), arising from spillover of GABA, and mediated by high affinity extrasynaptic GABA A receptors (Farrant and Nusser, 2005;Bright et al., 2007) also plays a fundamental role in shaping network excitability (Semyanov et al., 2004;Mann and Mody, 2010).In thalamus (Belelli et al., 2005;Cope et al., 2005;Bright et al., 2007), dentate gyrus (Nusser and Mody, 2002) and cerebellum (Brickley et al., 1996), I tonic is maintained by the activity of GABA receptors containing benzodiazepine insensitive v and/or a4 or a6 subunits (Brickley et al., 2001).However, I tonic may also be mediated by receptors containing the a5 and v subunits (Clarkson et al., 2010).In addition, a benzodiazepine-sensitive, presumably non-v subunit mediated I tonic has been reported in hippocampal neurons (Bai et al., 2001;Semyanov et al., 2003Semyanov et al., , 2004) and somatosensory cortex (Yamada et al., 2007).As I tonic is more active in fast-spiking (FS) interneurons compared to pyramidal cells (Semyanov et al., 2003), and FS cells have been reported to express high levels of a1 subunit containing GABA A receptors (Bacci et al., 2003;Thomson et al., 2000), an effect of benzodiazepines on I tonic in a single FS cell could have a profound effect on synchronous activity in neuronal networks.
Using magnetoencephalographic techniques we have recently shown that cognitive and motor deficits in stroke patients are associated with slow wave (4e12 Hz) and beta (15e30 Hz) activity in cortical regions, including M1 (Hall et al., 2010(Hall et al., , 2014)).Subsedative doses of zolpidem reduce these, correlating with improved clinical outcomes (Hall et al., 2010(Hall et al., , 2014)).Similarly, in Parkinson's patients, we have shown that the power of interhemispheric beta oscillations is unbalanced compared to controls, such that beta power contralateral to Parkinsonian symptoms is raised above ipsilateral beta power.The imbalance is reversed by sub-sedative doses of zolpidem (Hall et al., 2014) through simultaneous augmentation and depression of contralateral and ipsilateral beta power, respectively.These observations suggest that zolpidem is capable of bidirectional modulation of neuronal network activity in vivo.Here, using an in vitro brain slice model of neuronal oscillatory activity and a kinetic model of GABA A receptor dynamics, we show that changes in tonic inhibitory input to FS interneurons produce bidirectional modulation of neuronal network synchrony in M1 mediated by benzodiazepine-site activation.

Ethical approval
All procedures were approved by Aston University's Local Ethical Review Panel and conducted in accordance with the Animals (Scientific Procedures) Act 1986 UK and European Communities Council Directive 1986 (86/609/EEC).

Extracellular recordings
For extracellular recordings, slices were transferred to an interface chamber (Scientific System Design Inc, Canada) and continuously perfused (1e2 ml/min) with glucose-based aCSF.The perfusate was maintained at 33e34 C using a PTC03 proportional temperature controller (Scientific System Design Inc., Canada).Kainic acid (KA, 100 nM) and carbachol (CCh, 5 mM) were added to the perfusate to promote network oscillatory activity.We measured oscillatory power every 5 min and waited until peak power had been within 10% variance for more than 30 min.Typically, stabilization occurred after 60e90 min.
Local field potential (LFP) recordings were made using borosilicate glass micropipettes filled with aCSF (1e3 MU; P-97, Sutter instrument Co, USA) and visually positioned in layer V of M1.Signals were recorded and low-pass filtered (200 Hz) using a programmable signal conditioner (CyberAmp 380, Molecular Devices, USA).Oscillatory activity was digitized at 10 kHz using an A-D converter (CED Micro-1401 mk II; Cambridge Electronic Design, UK) and recorded to disk.Spike2 software (CED, UK) was used for all acquisition and analysis.

Whole-cell recording
Slices were transferred to a submersion recording chamber perfused at 5e7 ml/ min with glucose-based aCSF at 32 ± 0.5 C, on the stage of an Olympus BX51WI microscope.Layer V in M1 was readily identified by the presence of large pyramidal (Betz) cells visualized using infrared video-microscopy with differential interference contrast optics.Whole-cell recordings were made from cells visually identified as non-pyramidal, using borosilicate glass pipettes (3e4 MU).

Data analysis
All data were analysed off-line using Clampfit 10.2 (Molecular Devices, USA), Spike2 (CED, UK) or Mini-Analysis (Synaptosoft, USA).Spontaneous IPSCs (sIPSCs) were analysed using Mini-analysis.The KolmogoroveSmirnov (KS) test was used to assess changes in the cumulative probability distribution of inter-event intervals (IEI) and Student's t-test was used to assess changes in IEI, decay time and amplitude.To determine the effects of drugs on spontaneous release, 200 consecutive sIPSCs were analysed from each recording in each condition (control v drug) and the median value for interval, amplitude or decay calculated prior to finding a meanmedian value across multiple recordings.
In experiments involving application of GABA during whole-cell voltage clamp recording, the rate and extent of desensitization of GABA-induced inward currents was calculated by exponential curve fitting, performed using a simplex algorithm combined with a Chebyshev routine.
The fit solves for the amplitude A, the time constant t, and the constant y-offset C for each component i.The reduction in current after a long application of GABA was taken as the steady-state level and calculated as a normalized value of the peak current achieved.
For tonic current measurements, to exclude confounding effects of overlapping sIPSCs, the baseline current was measured as described by Nusser and Mody (2002).Briefly, the mean of a 5-ms epoch was taken every 100 ms.The mean and SD of the averaged baseline points were calculated for 10 s (~100 averaged baseline points) at three distinct times of the recordings (periods A, B, and C).Two changes in the baseline current were calculated between the three periods.The first (DBL) was the value of the difference during recording periods A and B, both taken during the control period and reflecting baseline fluctuation.The second (DDrug) was the value of the difference during recording periods C and B, and reflecting drug induced changes to the holding current.The two baseline changes (DBL and DDrug) were then statistically compared (paired t-test).
The frequencyedomain method (Fast fourier transform (FFT)) was used to generate power spectra of LFP data.Unless otherwise stated, 60 s epochs of sampled data were analysed using Spike 2 (CED, Cambridge, UK).All statistical analyses were E.J. Prokic et al. / Neuropharmacology 95 (2015) 192e205 performed using mean peak values, and, unless otherwise stated, it is these values that are reported in the text.
All data are expressed as mean ± S.E.M. and * indicates significance at P < 0.05, ** at P < 0.01 and *** at P < 0.001.

Source of drugs
Salts and other agents used in preparation of aCSF and electrode solutions were obtained from Sigma (UK) and Fisher (UK).Bicuculline methobromide, gabazine, zolpidem, THIP (4,5,6,pyridin-3-ol) hydrochloride) and flumazenil 4]benzodiazepine-3-carboxylic acid) were obtained from Tocris, UK.Drugs were initially prepared in stock solutions of 1e10 mM and stored at À20 C prior to dilution to the final concentration in aCSF immediately before use.

Modelling of GABA A receptor dynamics
Kinetic models were extended from previously published models of GABA A receptor kinetics (Jones and Westbrook, 1995;Jones et al., 1998;Karayannis et al., 2010), and, where possible, we have remained faithful to these.Models were generated as ordinary differential equations (ODEs) using techniques described previously (Croft et al., 2013) and the system solved using Matlab R2013b (Mathsworks) as follows: To simulate pre-incubation of tissue with either zolpidem or ambient GABA, models were allowed to reach steady state (t 0 ¼ 150 s).GABA was then rapidly, but not discontinuously, increased to the specified concentration for each experiment.Simulation of GABA A channels was calculated by measuring the movement of Cl À into the cell (see reaction scheme, Fig. 10).To provide a comparison with experimental data, we inverted the response measured to mirror changes in tonic current observed within Cl À loaded neurons.The concentration profile was modelled for 1 mM GABA at time t 0 þ 200 s.To simulate the effects of adding zolpidem post GABA stimulation, 10 nM zolpidem was added at t 0 þ 250 s.Where the models previously generated by Jones and Westbrook (1995) have focused upon single channel kinetics, our models used as many of these parameters as possible and then adapted to show the possible mechanisms of the experimental data using Matlab R2013b.The kinetic rate constants for parameters previously determined experimentally were constrained, while the others were allowed to vary as the software required.Initial conditions and reaction rate constants used in the appropriate models are shown in Fig. 10C.

Model description e GABA
The model described here contains the same unbound state (U) and two transition states (B 1 and B 2 ) as published by Jones and Westbrook (1995) (see Fig. 10).However, to describe the precise kinetics observed in our experimental data we have modified and extended this model to contain additional open and desensitized states (depicted by the presence of the dotted boxes in Fig. 10).The Jones and Westbrook (1995) model contains two open states (O 1 and O 2 ) and two desensitized states (D 1 and D 2 ) both of which occur following GABA binding.Further to this, we have proposed the existence of an additional open (O 0 ) and desensitized state (D 0 ) both occurring from the unbound state (U; Wlodarczyk et al., 2013).To remain consistent with the structure of the Jones and Westbrook model we have allowed the single desensitized (D 1 ) state to transition with the unbound desensitized state (D 0 ).As with the Jones and Westbrook models, the two-agonist binding sites are assumed to be equal and independent.

Model description e zolpidem
To account for the actions of zolpidem binding, previous models of the GABA A receptor kinetics have simply increased the rate with which GABA binds to its receptor (K on ).Here, additional parameters have been fitted to ensure good agreement with experimental data (Karayannis et al., 2010).Our initial attempts at using this K on modelling approach proved unsuccessful in reproducing our experimental data shown in Fig. 11.As a consequence, primarily to reflect the non-intuitive data related to the 10 nM zolpidem effect, we have derived a new kinetic model whereby we allow zolpidem to be a reactant in the scheme.Based upon previous studies (Pritchett et al., 1990;McKernan et al., 1995), zolpidem has been shown to bind to GABA A receptors with higher or lower affinity, depending on subunit expression of the receptors.For simplicity, we have depicted this as a two-state binding event.The first is the higher affinity state (UZ1) and the second, the lower affinity state (UZ2).Consistent with previous work, we assume that the presence of the bound zolpidem decreases the K off rate of GABA binding and increases the apparent frequency of GABA A receptor channel opening (see Bianchi et al., 2009;Bianchi, 2010;see rate constants Fig 10C).We next assumed that zolpidem binding is effectively irreversible due to its constant presence in the media.Finally, we assume that the kinetic rate of GABA A channel opening is equal regardless of zolpidem or GABA bound states, although it is well established that the frequency of opening of GABA receptors may be altered by benzodiazepine binding.
The reaction scheme for the zolpidem model is depicted in Fig. 10B.With zolpidem bound to the low affinity site (UZ2), GABA is now able to bind and the channel transitions through the states analogous to that shown for the GABA binding model.Where possible, the rate constants have been conserved between the two models, however, to account for changes in receptor dynamics brought about by the binding of zolpidem, new rate constants were attributed as shown in Fig. 10C.

Initial conditions and rate constants used in simulating experimental conditions
The initial concentration of species in the model was, U 5 mM, Cl À out 130 mM, Cl À in 0.9 mM and ambient GABA 100 nM.Both concentrations of ambient GABA and Cl À out remained constant through all simulations to reflect experimental conditions.Zolpidem concentrations were varied as stipulated in each simulation.For the majority of simulations 1 mM GABA was used as the simulating ligand unless stated.The rate constants used were fixed parameters when simulating experimental conditions.

Involvement of GABA and benzodiazepine binding in betaoscillations
Previous studies indicate that GABA A receptors containing a1 and g2 subunits are present on inhibitory interneurons, including FS cells.Since the benzodiazepine-binding site lies at the interface between these subunits, and GABAergic interneurons are subject to cell-type specific GABA tone, we hypothesized that benzodiazepine-site activation would act, via GABAergic interneurons, to modulate oscillatory activity.
To investigate the role of the benzodiazepine site in modulation of beta oscillations, we tested the effects of the benzodiazepine-site agonist, zolpidem.In previous in vitro studies, zolpidem has often been used at mM concentrations (e.g.P alhalmi et al., 2004;Koniaris et al., 2011).We have reported that zolpidem depresses neuronal oscillatory activity in vivo in the cortex of stroke (Hall et al., 2010) and Parkinson's disease (Hall et al., 2014) patients at sub-sedative doses (5 mg).Plasma concentration after oral administration of 5e10 mg of zolpidem is in the range of 14e93 ng/mL (Olubodun et al., 2003;Greenblatt et al., 2006).Zolpidem is approximately 92% bound by plasma proteins (Salv a and Costa, 1995), suggesting that the effective concentration of zolpidem in vivo may be in the nM range; hence we chose to explore the effects of low nanomolar concentrations of zolpidem in vitro.
It is generally accepted that benzodiazepine site agonism causes an enhancement of beta activity or 'beta buzz' (Glaze, 1990), so we explored the dose-dependency of benzodiazepine-site modulation by constructing cumulative dose response curves (30e500 nM).In contrast to the depression of beta power with 10 nM zolpidem, an enhancement was seen at 30 nM (319.2 ± 119.9%, P < 0.05, n ¼ 7; Fig. 3).We repeatedly attempted to construct concentrationeresponse curves for the complete range of concentrations from 10 to 500 nM.However, we found that once oscillatory power had been depressed by 10 nM zolpidem it was not possible to augment oscillatory power with higher concentrations.
Zolpidem clearly had two discernible effects, with reduction of oscillatory power at 10 nM and clear augmentation at 30 nM.At higher concentrations (100e500 nM), the drug began to depress beta power again, but this effect was quite variable, so we focused further experiments on the repeatable effects of zolpidem at 10 and 30 nM.
In view of the narrow concentration threshold between inhibition and augmentation of oscillatory activity, we examined the benzodiazepine-site specificity of the different effects of zolpidem, by repeating the previous experiments in the presence of a benzodiazepine-site antagonist.Flumazenil (500 nM), prevented both the depressing and potentiating effects of zolpidem (Fig. 4Ai  and Aii).This indicates that zolpidem exerts both effects on beta power through the benzodiazepine binding site.Interestingly, application of flumazenil alone significantly augmented oscillatory power (276.7 ± 111.4% of control at 100 nM, 292.1 ± 90.8% at 200 nM and 273.2 ± 90.0% at 500 nM, n ¼ 6, P < 0.05, Fig. 4Bi and Bii).

Effects of benzodiazepine binding on phasic and tonic GABA A R currents
We have recently reported a depression of pathological theta/ beta oscillations associated with alleviation of stroke related deficits in cognitive function (Hall et al., 2010), by sub-sedative doses of zolpidem in a stroke patient.Similarly, we have described benefits of low-dose zolpidem treatment in patients with Parkinson's disease (Hall et al., 2014).Thus, the inhibition of beta activity by zolpidem at a low concentration (10 nM) was of particular interest.Zolpidem could alter phasic and/or tonic inhibition GABAergic inhibition (Nusser and Mody, 2002;Semyanov et al., 2003;Gao and Smith, 2010), so we investigated these possibilities in pyramidal and FS interneurons in M1.

Tonic and phasic inhibition in pyramidal cells
We made whole-cell voltage clamp experiments from pyramidal neurons in layer V. To establish whether the effects of zolpidem we saw in LFP recordings of network activity were related to GABAergic neuronal activity we recorded sIPSCs, which reflect inhibitory activity from local networks of GABAergic neurons.Fig. 5 shows that application of KA and CCh elicited rhythmic sIPSCs at beta (and lower) frequency, evidenced as a peak at 15e20 Hz in the fast Fourier transform of the data.Addition of zolpidem (Fig. 5B) significantly depressed peak beta power, (P < 0.05; n ¼ 6; Fig. 5C).Mean area power (between 13 and 29 Hz) was 64.7 ± 27.9 pA s, increasing to 521.4 ± 302.1 pA s in KA and CCh and decreasing to 327.9 ± 86.1 pA s in zolpidem.Although beta frequency in submerged recording conditions was lower than that seen in interface LFP recordings, it was clear that zolpidem had a similar effect at low concentrations, and that these effects were related to rhythmic inhibitory activity generated by GABAergic neurons impinging on principal cells.
A change in rhythmic sIPSCs could be attributable to a direct reduction in presynaptic release of GABA, inhibition of postsynaptic receptors, or indirect reduction of GABA release via depression of interneuronal excitability, and might be mediated via phasic or tonic inhibition on principal cells or interneurons.We explored these possibilities using zolpidem at concentrations found to either depress or augment beta oscillatory power.
In voltage clamp recordings from pyramidal cells, a tonic current (mean 14.7 ± 4.8 pA, n ¼ 13) was revealed by blockade of GABA A receptors with bicuculline (20 mM).Zolpidem (10 nM) had no significant effect on I tonic (2.4 ± 6.6 pA, P > 0.05, n ¼ 6).Zolpidem also failed to modulate I tonic at 30 nM (mean change 2.3 ± 7.4 pA, P > 0.05, n ¼ 7).These data suggested that pyramidal cell I tonic was zolpidem-insensitive.We also examined the kinetics of sIPSCs recorded in pyramidal neurons during application of 10 and 30 nM zolpidem, and found no significant change in mean-median IEI, amplitude or decay time constant, indicating that phasic inhibition in pyramidal cells is also unaffected by zolpidem at the concentrations used (See Supplementary figures for analyses of tonic and phasic currents in pyramidal cells).

Phasic inhibition in FS cells
In order to address similar questions in FS cells, Betz cells in layer V were first located on the basis of size, orientation and morphology, and FS cells then identified through their own characteristic morphology and location adjacent to Betz cells (Fig. 6A).The stereotypical arrangement between Betz and GABAergic interneurons was confirmed, as previously reported (Ali et al., 2001;Ali, 2003).In FS cells (n ¼ 6) depolarizing current injection, elicited high frequency (80e198 Hz) non-adapting spikes with short duration at half amplitude (0.76 ± 0.03 ms), a pronounced medium after-hyperpolarization (AHP; 17.4 ± 1.6 mV) with no slow AHP, and a mean input resistance of 99.0 ± 6.0 MU (Fig. 6B).There was no evidence of a depolarizing sag in the response to hyperpolarizing current injection.These characteristics are similar to those described previously for FS cells in M1 (Kawaguchi, 1995;Cauli et al., 1997;Kawaguchi and Kubota, 1997;Ali et al., 2001).In subsequent studies we used voltage-clamp recording to assess phasic and tonic GABAergic inhibitory currents in FS neurons.

Effects of 10 nM zolpidem on I tonic in FS cells
Given that we could find no effect of 10 nM zolpidem on phasic inhibition impinging on GABAergic neurons in M1, we analysed the effects of zolpidem application on tonic inhibition.Since tonic inhibition has not previously been demonstrated in FS cells in M1, we first tested the effects of the competitive GABA A receptor antagonist, gabazine (250 nM).As Fig. 6C shows, gabazine revealed a tonic current (mean 10.3 ± 4.4 pA, P < 0.05 compared to baseline fluctuation; n ¼ 5).Subsequent co-application of bicuculline (20 mM) removed the small residual current, although overall there was no significant change compared to gabazine alone (2.5 ± 2.3 pA).
Application of bicuculline alone (20 mM), as Fig. 6D shows, also caused a decrease in holding current (22.66 ± 3.55 pA, n ¼ 6, P < 0.001).These data indicated that a tonic current was indeed  present in inhibitory neurons in M1.We next assessed the sensitivity of I tonic to zolpidem.At 10 nM, zolpidem itself generated a slow, inward current (mean amplitude 15.0 ± 3.5 pA; n ¼ 6; Fig. 7A,D).Subsequent application of bicuculline (20 mM) abolished I tonic suggesting that the increase in inward current by zolpidem was due to augmentation of existing I tonic .Indeed, bicuculline always reduced I tonic beyond the pre-zolpidem baseline, showing the presence of a non-zolpidem induced component of tonic current in the inhibitory neurons.To determine whether two components could be differentiated, we attempted to modulate I tonic through alteration of ambient GABA concentration.We pre-applied GABA (1 mM), which induced in a large inward current (70.0 ± 14.1 pA; n ¼ 6, P < 0.05; Fig. 7B,D).Subsequent addition of zolpidem (10 nM) generated a further inward current, of the same magnitude as that seen without addition of GABA (14.9 ± 5.1 pA; P < 0.05; Fig. 7B,D), suggesting that at 10 nM, the I tonic induced by zolpidem did not depend on ambient GABA concentration.
Previous reports have indicated that I tonic is often linked to the presence of the delta (d) subunit within extrasynaptic GABA receptor complexes (Nusser et al., 1998) so we studied the subunit composition of receptors mediating I tonic in FS cells using the dsubunit-selective agonist, THIP.THIP has been reported to be selective for d-subunit containing receptors at up to 800 nM (Meera et al., 2011), so we used a slightly higher concentration of THIP (1 mM) look for occlusion of the zolpidem mediated augmentation of I tonic .THIP alone had no effect on phasic inhibition (data not shown) but did induce a slow inward current in FS cells (Fig. 7C,D).After stabilization of the THIP-induced current (mean 17.1 ± 11.6 pA; n ¼ 6), a further increase in I tonic was induced by coapplication of zolpidem at 10 nM (14.9 ± 6.7 pA), and the difference was significant (Fig. 7C,D; P < 0.05).These data suggest that a pool of THIP-sensitive, d-subunit containing receptors mediates a component of I tonic in FS cells, but that these receptors are not the same as those mediating the effects of zolpidem.
Finally, it should be noted that when bicuculline (20 mM) was applied prior to zolpidem, both phasic and tonic inhibition were abolished, and changes in holding current in response to zolpidem were not seen (n ¼ 6, data not shown), indicating that both forms of I tonic were mediated by bicuculline-sensitive GABA A receptors.

Effects of benzodiazepine site antagonism on tonic and phasic inhibition in FS cells
We further examined the role of the benzodiazepine site on the effects of zolpidem using the antagonist flumazenil.Flumazenil (500 nM) had no effect on phasic inhibition (Fig. 8A,B).The mean IEI of sIPSCs in control was 416.5 ± 14.7 ms compared to 488.5 ± 19.6 ms in flumazenil (n ¼ 6; P > 0.05 KS), and sIPSC amplitude was also unaltered (81.8 ± 1.4 pA v 80.0 ± 1.4 pA; P > 0.05 KS).However, when applied alone, flumazenil decreased the holding current (31.3 ± 13.9 pA; P < 0.01, n ¼ 6), demonstrating that I tonic in FS cells depends on benzodiazepine-site activity.If the inhibitory effects of zolpidem on beta oscillations were mediated via the zolpidem-sensitive tonic current in FS cells, then we may expect the zolpidem (10 nM) induced augmentation of I tonic to be blocked by flumazenil.This was indeed the case, as when zolpidem (10 nM) was applied in the presence of flumazenil no further change in holding current was observed (Fig. 8C).

Effects of 30 nM zolpidem on phasic and tonic inhibition in FS cells
As described above, 30 nM zolpidem elicited an increase in oscillatory power, suggesting that the higher concentration may have an additional effect on phasic inhibition, and/or a different effect on I tonic., so we explored the effects of zolpidem at 30 nM on phasic and tonic inhibition in FS cells.Zolpidem at 30 nM, as at 10 nM, had no effect on IEI, amplitude or decay time of sIPSCs (data not shown).However, in contrast to its effects at 10 nM where it augmented I tonic , at 30 nM zolpidem elicited a decrease in I tonic (17.8 ± 20.4 pA; P < 0.05; n ¼ 6; Fig. 9A,C).

The role of desensitization in the differential effects of zolpidem in FS cells
Since the effects of benzodiazepine site ligands might be expected to differ depending on receptor occupancy, and since both occupancy and benzodiazepine binding might alter GABA A receptor dynamics such as desensitization and resensitization, we recorded I tonic under conditions in which GABA concentration was fixed at a sub-saturating level of 1 mM.At 1 mM (Fig. 9B,C), GABA induced a maximum inward current (78.62 ± 47.5; P < 0.05; n ¼ 6) and steady state desensitization (28.3 ± 7.6; P < 0.05).In contrast to its effects at ambient GABA levels, application of 30 nM zolpidem in the presence of 1 mM GABA, elicited an increase in I tonic (82.3 ± 37.6 pA; P < 0.05; n ¼ 6; Fig. 9B,C) and this was followed by rapid desensitization of the zolpidem-induced response (28.3 ± 7.6 pA).The large increase in tonic current and the desensitization profile observed in the presence of 1 mM GABA suggested that a robust desensitization process was taking place.
Using the kinetic simulation model described in the methods we have simulated the experimental data described above.The model contains the presence of open and desensitized states for the unbound receptor and clearly shows the opposing effects of low and higher dose zolpidem.Fig. 10 shows the Markov models for GABA (Fig. 10A) and zolpidem (Fig. 10B) binding; the rate constants used (Fig. 10C) were first estimated using the parameters of Jones and Westbrook (1995).
As Fig. 7A showed, zolpidem at 10 nM caused an increased I tonic .In the kinetic model, this behaviour was also seen (Fig. 10Di).By contrast, zolpidem caused a decrease in I tonic at 30 nM (Fig. 9A), and again this was captured in the behaviour of the model (Fig. 10Dii), with the distribution of receptor-ligand states suggesting that this is related to enhanced entry into rapidly desensitized states.When we 'applied' zolpidem in the presence of 1 mM GABA in the model (Fig. 10Diii), 30 nM zolpidem caused a clear (and relatively large) increase in I tonic followed by a clear desensitization, analogous to the situation shown in Fig. 9B.Both experiment and the model indicated that at a concentration of 10 nM, zolpidem induces a consistent, low amplitude I tonic that is independent of ambient GABA concentration.By contrast, at 30 nM, zolpidem induces changes in I tonic whose magnitude and even direction is highly dependent on ambient GABA.We interpreted these data to indicate that at 10 nM, zolpidem falls below the threshold for exerting effects on GABA receptor affinity, instead working to enhance channel opening through effects on gating in a subpopulation of GABA receptors.At 30 nM, zolpidem begins to affect the GABA unbinding rate, and this is manifest as a greatly enhanced I tonic in the presence of 1 mM GABA, which was as large as the current induced by 1 mM GABA itself.These affinity effects are also manifest as rapid desensitization events, seen clearly at 30 nM in the presence or absence of applied GABA.
The experiments with 30 nM zolpidem indicated that at different concentrations of zolpidem and ambient GABA, we were observing differences in the rate and extent of steady state desensitization of GABA receptors on FS cells.Entry into and out of desensitized states is controlled, at least in part, through the degree of binding of the GABA receptor.For example, at synaptic levels of GABA, desensitization is rapid, and exit and entry to and from open and desensitized states play a major role in shaping the decay phase of the GABA IPSC.At lower levels of GABA, however, receptor occupation is unlikely to be high, and channel kinetics are altered.Recently, an elegant study of receptors modified to disable dual binding (Petrini et al., 2011) reported that mono-liganded GABA receptors showed dramatically slowed kinetics compared to dual bound receptors, and that entry and exit from desensitization was particularly slow at very low levels of receptor occupancy.We hypothesized that, in the presence of nanomolar ambient GABA and 10 nM zolpidem, receptor binding would be low and effects of zolpidem on K off would be negligible, such that slow entry and exit from a desensitized state similar to that described by Petrini et al. (2011) would be favoured.By contrast, this would not be so at 30e1000 nM, when effects on K off dominate, and when zolpidem induced rapid desensitization would be readily observed (as in Fig. 9B) and may even be enhanced by benzodiazepine binding (Mellor and Randall, 1997).Benzodiazepine site mediated effects on K off cannot affect the maximal rate and extent of responses to saturating concentrations of GABA (Bianchi et al., 2009;Fig. 8. Flumazenil (500 nM) reduces the tonic inhibitory activity and blocks the zolpidem-induced tonic current in FS interneurons.A) Representative traces from single experiment showing sIPSCs in control and flumazenil (500 nM).Scale bars 50 pA Â 500 ms.Bi) Cumulative probability plot showing there was no significant change in the frequency of sIPSCs with the addition of flumazenil (500 nM).Bii) Cumulative probability plot showing there was no significant difference to sIPSC amplitude with the addition of flumazenil (500 nM).C) Representative voltage-clamp recording from a FS interneuron at a holding potential of À70 mV.Addition of flumazenil (500 nM) slowly reduced a tonic current.Subsequent addition of zolpidem (10 nM) was without effect.Scale bars 100 pA, 5 min.Fig. 9. Zolpidem (30 nM) inhibits tonic current without effect on phasic IPSCs in FS interneurons.A) Voltage-clamp recording from an FS interneuron at a holding potential of À70 mV.Addition of zolpidem (30 nM) reduced a tonic current.Scale bar 100 pA, 5 min.B) Voltage-clamp recording from an FS interneuron.Addition of GABA (1 mM) induced a large inward current, which was increased by further addition of zolpidem (30 nM).Note the desensitization of the tonic current that takes place in both GABA and zolpidem.Scale bar 100 pA, 1 min.C) Bar chart showing change in holding current induced by the addition of zolpidem (30 nM), GABA (1 mM) and GABA (1 mM) þ zolpidem (30 nM).*P < 0.05.Bianchi, 2010).However, if, at 10 nM, zolpidem was mediating effects on GABA receptors primarily through effects on mono-or unliganded, kinetically slow entry and exit from desensitized states, this would be revealed as a submaximal response to challenge with saturating GABA.
We therefore tested for a population of kinetically slow desensitized GABA A receptors by pre-incubating in 10 nM, 30 nM or 1 mM zolpidem and subsequently challenging with saturating (1 mM) GABA in the continued presence of zolpidem.Application of 1 mM GABA alone typically elicited a large inward current (976.9 ± 184.1 pA, n ¼ 8), followed by desensitization (t 19.74 ± 1.37 s, extent 366.4 ± 96.50 pA, n ¼ 8), to a steady-state (478.7 ± 197.3 pA, n ¼ 8).We then measured the maximal and steady-state current, and the rate and extent of desensitization in a series of experiments in which GABA (1 mM) was applied after pre-incubation (for 60 min) in different concentrations of zolpidem.Zolpidem was also present in the bath at the pre-incubation level for the duration of the experiment.As Fig. 11 shows, at both 30 nM and 1 mM, the effects of saturating GABA were not significantly different to control (n ¼ 6 in each group), with large maximal current, rate and extent of desensitization and steady-state values.Indeed, discernible changes to the desensitization parameters were only visible after pre-incubation with 10 nM zolpidem.Here, the application of GABA resulted in much reduced maximal and steady-state currents and a lesser rate   11B).These data strongly suggest that a population of receptors was sequestered into a kinetically slow desensitized state by incubation with 10 nM zolpidem and that such receptors were then unavailable to participate in further desensitization processes.We tested this hypothesis using an in silico approach, applying a 'saturating concentration' of GABA in a model that allowed analysis of the individual receptor species concentrations after pre-incubation with different 'zolpidem concentrations'.
When we repeated the experiments in silico, the model showed a profile consistent with the in vitro data, such that GABA induced a large peak and desensitization to a steady-state.Hence, control, 30 nM and 1 mM zolpidem simulations showed extensive desensitization, with large 'currents' and rapid kinetics.In excellent agreement with the in vitro experiments, 10 nM zolpidem occluded these effects to a significant degree (Fig. 11Aii).The model predicted a reduction in the population of receptors immediately available for maximal activation and desensitization in response to 1 mM GABA.Analysis of the individual receptor species concentrations revealed that after incubation with 10 nM zolpidem, 54.9% of GABA receptors were unavailable for binding in 1 mM GABA, compared to only 12.1% of receptors following incubation in 1000 nM zolpidem.
Hence, the kinetically slow desensitization produced by 10 nM zolpidem incubation appeared to have prevented GABA binding during the challenge with saturating GABA, whilst higher concentrations of zolpidem favoured kinetically rapid transitions between states and allowed z90% of receptors to bind GABA and become rapidly desensitized.

Beta oscillations and FS interneurons in M1
FS cells preferentially form synaptic connections at perisomatic and proximal dendritic regions of numerous neighbouring pyramidal cells, and are thus able to exert powerful control over large assemblies of neurons (Cobb et al., 1995).The characteristic features of FS cells, which include short duration action potentials, high frequency firing with no adaptation, reliable synchronous GABA release and rapidly decaying IPSPs onto principal cells, mediated by receptors containing the a1 subunit (Thomson et al., 2000), appear crucial for both coherence of network oscillations and frequency control (Bartos et al., 2001(Bartos et al., , 2002)).Recently, Manseau et al. (2010) have reported the asynchronous release of GABA at autaptic and synaptic contacts from FS interneurons, highlighting the fact that these cells control both synchronization and desynchronization processes in cortical networks and underlying their pivotal position in network function.

FS interneurons and I tonic
In addition to phasic synaptic inhibition, CNS neurons are also subject to persistent, tonic inhibition.Due to the substantial charge transfer associated with tonic inhibition, its effects are profound and include altering excitability of individual neurons as well as influences on excitatory synaptic efficacy and integration (Farrant and Nusser, 2005).It has been demonstrated that differential regulation of GABA uptake (Semyanov et al., 2003) enhances I tonic FS cells compared to that seen in pyramidal cells.Hence, I tonic in FS cells appears likely to exert powerful control over network activity, and, indeed, eliminating tonic inhibition in interneurons has been shown to increase the power of gamma activity in CA1/CA3 of the hippocampus (Mann and Mody, 2010).
In agreement with studies in other areas, we found I tonic in all our recordings from FS cells in M1.Our results show that I tonic in FS cells is increased by 10 nM zolpidem, but that the opposite effect is seen at 30 nM.At these two concentrations, zolpidem has opposing effects on the power of beta oscillations in M1 and, given the strong relationship between FS cell function and network activity, it seems likely that the bidirectional effects on I tonic and on network oscillatory power are related.The actions of flumazenil in blocking I tonic and enhancement of oscillatory power in M1 further supports this interpretation.The benzodiazepine site has been implicated in modulation of neuronal network oscillations (P alhalmi et al., 2004) and a zolpidem-sensitivity of I tonic has also been reported in other areas, albeit at considerably higher concentrations (Semyanov et al., 2003;Gao and Smith, 2010).The actions of zolpidem at the low concentrations described above may involve subunit-specific effects.When tested on recombinant receptors, zolpidem displayed a high potency at a1-containing GABA A receptors (a1b2g2, a1b3g2: K i ¼ 20 nM).Zolpidem had medium potency at a2 and a3 containing GABA A receptors (e.g.a121g2, a3b1g2: K i ¼ 400 nM) and was ineffective at a5 subunit containing receptors (a5b3g2, a5b2g2: K i 5000 nM) (Langer et al., 1992;Pritchett and Seeburg, 1990).The effects of very low concentrations of zolpidem on I to- nic , in our data strongly suggest that the GABA A receptors mediating the current in FS cells in M1 contain the a1 subunit (Pritchett and Seeburg, 1990;Langer et al., 1992;Crestani et al., 2000).This does not preclude the inclusion of d subunits, believed to account for most of the I tonic observed in other neurons (Brickley et al., 1996;Nusser and Mody, 2002;Belelli et al., 2005;Cope et al., 2005;Bright et al., 2007).Indeed, when we pre-applied the d-subunit selective agonist, THIP, we found no occlusion of zolpidem induced augmentation of I tonic , suggesting that non-desensitizing, d-subunit containing receptors were not primarily involved in the response to zolpidem.This is consistent with our earlier experiments (Fig. 5) in which bicuculline blocked both the tonic current induced by 10 nM zolpidem and also a further, presumably zolpidem-insensitive, component.
Our data showed that application of zolpidem augmented an existing I tonic in FS cells, indicated by the reduction of I tonic beyond baseline by bicuculline.These data confirm previous reports (Semyanov et al., 2003) of a strong I tonic at ambient GABA levels in FS cells in CA1 in the absence of uptake inhibitors.Moreover, the effect of flumazenil on both zolpidem-induced and baseline I tonic suggests that in these cells the current depends on endogenous activation of the benzodiazepine site, as described recently by Christian et al. (2013).
That we found no effects of low concentrations of zolpidem on phasic inhibition indicates that the dual effects of zolpidem on neuronal network oscillations in M1 may be mediated via tonic inhibition alone.In support of this interpretation, Wulff et al. (2009) have recently reported that knockout of GABA A receptors on parvalbumin positive FS cells in hippocampus does not alter gamma activity, suggesting that extrasynaptic inhibitory drive may be more important than previously thought.The reduction in I tonic seen with 30 nM zolpidem might be predicted to increase the number or effectiveness of FS cells participating in network activity, suggesting that the benzodiazepine-site may be involved in scaling the size of the active network.
What receptor states may underlie the behaviour of I tonic in response to different degrees of benzodiazepine-site agonism?In our basic model, a population of closed, unbound receptors (U) exists in equilibrium with a population of closed, bound receptors (B 1 and B 2 ), and the latter population may transition between open, bound receptors (O 1 and O 2 ) and desensitized, receptors (D).There is also an open, unbound state (O 0 ).At low ambient [GABA], the lowest concentration of zolpidem (10 nM) has little effect on K off and the probability of transition to O 1 and O 2 is low.However, the O 0 state is stabilized by effects of benzodiazepine binding on channel gating properties, which dominate over subtle effects on K off (Bianchi, 2010), and this leads to increased charge transfer expressed macroscopically as enhancement of the tonic, unbound GABA A R conductance (Wlodarczyk et al., 2013).At higher concentrations of zolpidem (30e1000 nM), zolpidem begins to depress K off , causing an apparent increase in affinity of the GABA A R for GABA and shifting the equilibrium toward an increase in C b .The probability of moving from B into the O states at low ambient [GABA] remains low, as most receptors will be singly bound, and consequently the desensitized state, D, remains the highest affinity bound state.This leads to a shift in the equilibrium towards D, and macroscopic desensitization in the absence of significant increased charge transfer.Hence, at 10 nM, zolpidem acts to stabilize an open, unbound GABA A R through effects on gating leading to augmented I tonic , and at 30e1000 nM, effects on affinity dominate and desensitization leads to a decreased macroscopic I tonic .When [GABA] is raised artificially to 1000 nM most GABA A R are dual-bound, and hence effects on affinity dominate at all concentrations of zolpidem, because the population of receptors in C b is large.Here, the equilibrium allows for transitions from B towards O and D, leading to augmented macroscopic currents and/or slow desensitization.
This model is further supported by the effects of flumazenil, application of which reveals a persistent GABA current even in the absence of zolpidem, suggesting that an endogenous benzodiazepine-mediated stabilization of the O 0 state exists in neocortical slices in vitro (Christian et al., 2013).Furthermore, when the proposed O 0 state is maximally augmented through prolonged incubation in 10 nM zolpidem at low ambient [GABA], the population of receptors sequestered into this state would be predicted to depress the rate and extent of maximal desensitization, since a smaller receptor population would be available to participate due to sequestration of receptors into the O 0 state.This is observed experimentally upon challenge of pre-incubated slices with 1 mM GABA (Fig. 11).This sequestration of receptors into a kinetically slow state presumably underlies the lack of effect of higher concentrations of zolpidem when applied subsequent to 10 nM zolpidem.By contrast, pre-incubation of slices with higher concentrations of zolpidem affects K off , depressing the population of receptors in U, dominating subtle gating effects, and effectively depleting the pool of receptors in O 0 .This pushes the equilibrium towards a high proportion of receptors in the B/O state, leading to return of maximal rates and extent of desensitization when GABA concentration is raised to mM levels.

Functional significance
The differential concentration-dependent effects of zolpidem on phasic and tonic inhibition in FS cells and consequent modulatory effects on oscillatory activity highlight the complex interplay between these two inhibitory mechanisms.Indeed, the constitutively active I tonic, activated on FS cells by high affinity receptors may be a fundamental requirement for normal physiological functioning.It may act as a 'gatekeeper', having the ability to sculpt oscillatory power over a slower timescale than compared to phasic inhibition by thresholding the level of FS cell activity (Pavlov et al., 2009) and/or GABA before there is FS cell recruitment to an oscillation.Recent reports of 'endozepine' activity in the thalamus (Christian et al., 2013) confirm the importance of the benzodiazepine site in control of GABA A receptor function.Our data highlight how subtle alterations in both GABA and benzodiazepine site activation may confer radically different desensitization kinetics onto GABA A receptors, and how this may alter network activity.The origin of the GABA that activates receptors on FS cells is presumably input from other GABA neurons.Such disinhibitory interneuronal networks are currently a topic of great interest in understanding cortical circuit function (Lee et al., 2013;Pi et al., 2013 see also Klausberger et al., 2005).
The oscillations we have reported here are high beta/low gamma.Others have shown that oscillations in vitro depend linearly on temperature (Dickinson et al., 2003) and it is not possible to state with confidence that they relate to pathological beta activity as seen in Parkinson's, or indeed in vivo gamma activity as seen in healthy brain in vivo.However, our previous clinical studies in M1 have shown that pathological slow wave and beta frequency oscillations can be suppressed by low-dose zolpidem (Hall et al., 2010(Hall et al., , 2014)), effects which correlate with improved cognitive and motor function.It seems plausible that the effects of low-dose zolpidem on FS interneurons and the modulation of tonic inhibition may provide the mechanism underlying these clinical observations.

Fig. 1 .
Fig. 1.Beta oscillations in M1 are dependent on GABAergic inhibition.Ai) Mean frequency and power of beta oscillations in M1.Aii) Representative power spectrum of beta activity in M1 in vitro.Bi) Raw data traces showing typical effects of gabazine (250 nM and 2 mM) on oscillatory activity in M1.Bii) Pooled data showing change in peak beta power in the presence of gabazine and representative power spectra.Ci) Raw data traces showing typical effects of tiagabine (1 mM and 10 mM) on oscillatory activity in M1.Cii) Pooled data showing change in peak beta power in the presence of tiagabine and representative power spectra.Scale bars 200 ms Â 50 mV.*P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 2 .
Fig. 2. 10 nM zolpidem reduces beta oscillatory power.A) Raw data traces showing the reduction in oscillatory activity induced by bath application of zolpidem at 10 nM.B) Bar charts of pooled data showing a reduction in beta oscillatory power induced by zolpidem (10 nM).C) Representative power spectra.Scale bars 200 ms Â 50 mV.***P < 0.001.

Fig. 3 .
Fig. 3. Concentration-dependent modulation of oscillatory power by zolpidem.A) Raw data traces showing the effect of different doses of zolpidem (30e500 nM) on beta oscillatory activity.B) Bar chart of pooled data showing a reduction in beta oscillatory power induced by zolpidem (30e500 nM).C) Representative power spectra.Scale bars 200 ms Â 50 mV.*P < 0.05.

Fig. 4 .
Fig. 4. Flumazenil (500 nM) blocks zolpidem induced depression and augmentation of beta power.Ai) Pooled data showing flumazenil blockade of the effects of zolpidem at 10 and 30 nM on modal peak beta oscillatory power.Aii) Power spectra of typical data recorded during flumazenil blockade of zolpidem responses.Bi) Bar chart showing effects of increasing concentration of flumazenil on beta power.Bii) Representative power spectrum showing increase in oscillatory power obtained in flumazenil (100e500 nM).Scale bars 200 ms Â 50 mV.*P < 0.05.

Fig. 5 .
Fig. 5. Zolpidem reduces rhythmic IPSC activity in FS cells.A) Raw traces showing the effects of zolpidem on KA þ CCh induced sIPSC activity in an FS cell.Scale bars 500 ms Â 200 pA.B) Power spectrum (30 data segments) from the recording in A), showing effects of zolpidem on spectral power.C) Pooled data showing effects of zolpidem (10 nM) on area power at 15e29 Hz. *P < 0.05.

Fig. 6 .
Fig. 6.Identification of FS interneurons.A) Photo-micrograph showing recording electrode and putative FS interneuron (asterisk) in close proximity to Betz cells (arrows) in LV of M1.Scale bar 25 mm.B) Voltage traces of the FS cell identified in A), in response to À400 pA and þ100 pA current steps.Scale bars 25 mV, 50 ms.Note the high firing frequency (>100 Hz) and action potentials with large AHPs.C) Representative voltage-clamp recording from a FS interneuron at a holding potential of À70 mV during application of gabazine (250 nM) and bicuculline (20 mM).Scale bar 100 pA, 2 min.D) Representative voltage-clamp recording from a FS interneuron at a holding potential of À70 mV during application of bicuculline alone.Scale bar 100 pA, 2 min.

Fig. 7 .
Fig. 7. Zolpidem (10 nM) augments a tonic inhibitory current in FS interneurons.A) Representative recording showing zolpidem (10 nM) induction of I tonic .Subsequent addition of bicuculline (20 mM) abolished both phasic IPSCs and zolpidem-induced tonic current.Scale bar 50 pA, 2 min.B) Representative recording showing a GABA and zolpidem induced tonic current.Note that the GABA current is desensitizing, but the zolpidem current is not.Scale bars 100 pA, 5 min.C) Voltage-clamp recording from an FS interneuron at a holding potential of À70 mV.Addition of THIP (1 mM) induced a tonic current.Subsequent addition of zolpidem (10 nM) augmented the current and the effect of both drugs was blocked by bicuculline (20 mM).Scale bar 100 pA, 2 min D) Histogram showing mean change in holding current induced by zolpidem (10 nM), GABA (1 mM), GABA (1 mM) þ zolpidem (10 nM), bicuculline (20 mM) alone, THIP and further changes induced by co-addition of zolpidem.*P < 0.05, **P < 0.01.

Fig. 10 .
Fig. 10.Simulation of experimental data.A) Markov model containing unbound state (U) and two transition states (B 1 and B 2 ), two open states (O 1 and O 2 ) and two desensitized states (D 1 and D 2 ) both of which occur following GABA binding.Further to this, we have proposed the existence of an additional open (O 0 ) and desensitized state (D 0 ) both occurring from the unbound state (dashed boxes in scheme).B) Model accounting for zolpidem binding depicted this as a two-state binding event.The first is a higher affinity state (UZ1) and the second, a lower affinity state (UZ2).O and D refer to open and desensitized states, respectively.C) Rate constant table for the schemes in A and B. Di) Simulation of experimental data in Fig. 3B (10 nM zolpidem).Dii) Simulation of experimental data in Fig. 5A (30 nM zolpidem).Diii) Simulation of experimental data in Fig. 5B (30 nM zolpidem in 1 mM GABA).

Fig. 11 .
Fig. 11.Zolpidem (10 nM) blocks fast desensitization processes in FS interneurons.Ai) Voltage-clamp traces showing pre-incubation of slices in 10 nM (grey), 30 nM (red) and 1 mM zolpidem (blue) have differing effects on desensitization processes.Aii) Simulation of experimental data in Bi.Bi) Bar chart showing maximum current induced by addition of 1 mM GABA after pre-incubation with zolpidem (10, 30, 1000 nM).Bii) The steady state response is reduced after pre-incubation with zolpidem (10 nM).Biii) The decay time constant of desensitization is significantly increased and Biv) the amplitude of desensitization is significantly reduced after pre-incubation with zolpidem (10 nM).*P < 0.05, **P < 0.01.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) ± 54.93 pA, t 71.19 ± 21.43 s, extent 127.3 ± 29.99 pA, steady-state 142.3 ± 26.56 pA, P < 0.05, n ¼ 6).Summary data are presented in Fig.