Voltage dependent allosteric modulation of IPSCs by benzodiazepines

Abstract

GABA_A receptors (GABA_AR) are inhibitory ion channels ubiquitously expressed in the central nervous system and play critical roles in brain development and function. Benzodiazepines are positive allosteric modulators of GABA_AR, by enhancing channel opening frequency when GABA is bound to the receptor. Midazolam is a commonly used benzodiazepine. It is frequently used for premature infants, but the long-term consequences of its use in this patient population are not well established. Here we studied the acute effects of midazolam on immature synapses. Using a rodent organotypic hippocampal slice preparation, we evaluated how midazolam affects inhibitory synaptic transmission onto CA1 pyramidal neurons. We find that 1 μM midazolam enhances evoked inhibitory post synaptic currents (eIPSCs) at a holding potential of −60 mV. Similarly, 1 μM midazolam enhances miniature IPSCs (mIPSCs) in CA1 pyramidal neurons at holding potentials of −60 mV and −30 mV. At depolarized holding potentials, however, midazolam no longer enhances mIPSCs. Depolarization of the postsynaptic cell by itself increases mIPSC decay, which apparently occludes the allosteric effects of midazolam. These results provide insight into how a benzodiazepine and membrane voltage may modulate GABA_AR function in developing circuits.

1. INTRODUCTION

Benzodiazepines are ion channel modulators which are used for their sedative, anxiolytic, and antiepileptic effects. The site of action for these drugs is through the GABA subtype A receptors (GABA_AR), which are ligand-gated ion channels ubiquitously expressed in the central nervous system (CNS) (McGoldrick and Galanopoulou, 2014). Benzodiazepines bind with high affinity to GABA_ARs and allosterically enhance channel opening frequency when GABA is simultaneously bound to the receptor (Sigel and Ernst, 2018). GABA_ARs selectively conduct chloride (Cl⁻) currents and subsequently hyperpolarize cells to decrease overall neuronal activity. This property makes GABA_ARs the predominant inhibitory mediator of the CNS. These physiological functions play major roles in the development and refinement of cortical circuits (Pelkey et al., 2017).

Midazolam, a commonly used benzodiazepine, is an ideal agent for sedation and induction of anesthesia because of its rapid onset and metabolism and its ability to induce retrograde amnesia (Pacifici, 2014). There are numerous studies demonstrating the safety of midazolam in the operating room and intensive care unit for both adult and pediatric populations (Duceppe et al., 2019). However, recent studies have shown negative outcomes and neuroanatomical changes in patients who received repeated exposures to midazolam in the neonatal intensive care unit (Duerden et al., 2016; Ng et al., 2017). This parallels large database studies which show that repeated and prolonged general anesthetic exposure in young children is associated with developmental disorders, learning disabilities, and behavioral problems (Hu et al., 2017; Wilder et al., 2009).

The brain of a premature infant is characterized by the rapid rate of neurogenesis and synaptogenesis, which is analogous to that seen in the first seven postnatal days of rodent development (Durrmeyer et al., 2010). During this period of rapid neuronal development, stem cells migrate, develop new synapses, and integrate into existing circuits in anticipation for further refinement during ex utero experience (Egawa and Fukuda, 2013; Sultan and Shi, 2018). In parallel, subunits of many ligand gated channels, including the GABA receptor, switch during different stages of development (Wisden et al., 1992). Developmental switches not only exist for receptors, but for ion transporters (i.e. KCCN transporter) as well, which can modify resting membrane potential and driving force for ion currents (Rivera et al., 1999). During this period of dynamic neural activity and development, perturbation of the global inhibitory circuit by benzodiazepine exposure may disrupt normal development.

In this study, we investigated how acute midazolam exposure modulates inhibitory synaptic transmission in immature synapses using a rodent hippocampal slice culture model. Ex vivo slice culture models allows simultaneous precise drug application and patch clamp recording of neurons within an intact circuit. Furthermore, neonatal rodent cultures have been shown to be analogous in synaptogenesis to third trimester human fetuses, allowing for comparison of midazolam exposure in premature and neonatal humans (Durrmeyer et al., 2010). We find that acute midazolam application enhances both evoked and miniature inhibitory postsynaptic currents (IPSCs) at hyperpolarized resting potentials, which is consistent with published in vitro literature (Rovira and Ben-Ari, 1993). However, midazolam no longer affects IPSC characteristics at depolarized potentials. Our results indicate that the effect of midazolam on GABA_AR-mediated currents will vary at different physiological potentials. These results suggest one pathway how repeated and prolonged midazolam exposure can have detrimental effects on early neural circuit development

2. RESULTS

2.1. Midazolam enhances evoked inhibitory postsynaptic currents in organotypic hippocampal slice cultures

To characterize the acute actions of midazolam in organotypic hippocampal slices, we recorded evoked inhibitory postsynaptic currents (eIPSCs) from CA1 pyramidal neurons during acute midazolam application. The bath solution included APV (50 μM) and DNQX (20 μM) to block glutamate-mediated synaptic transmission. We used a bipolar stimulating electrode to evoke currents every 20 seconds. Based on eIPSC latency after the stimulus artifact (<5 ms), these currents were monosynaptic responses. In some experiments, picrotoxin was included at the end of the recordings and abolished the currents, which indicated that the eIPSCs were mediated by GABA_ARs (Supplemental Figure 1).

After establishing a stable eIPSC baseline, vehicle (DMSO) or midazolam (1 μM) was bath applied (Figure 1A & 1B). After approximately 15 minutes of midazolam exposure, eIPSC charge (Figure 1B, upper panel) and amplitude (Figure 1B, lower panel) reached steady-state values. The steady-state values for vehicle and midazolam were quantified by averaging traces recorded 22–27 minutes after the start of drug application (Figure 1B).

Figure 1: Midazolam enhances evoked inhibitory post-synaptic currents (eIPSCs) in organotypic hippocampal slice cultures.

(A) Sample average current traces of monosynaptic eIPSCs recorded from individual CA1 pyramidal neurons during baseline (average of currents over 0–5 min) and steady-state (average over 22–27 min) for either vehicle (DMSO) or midazolam (1 μM) and overlay of traces. Holding potential, −60 mV.

(B) Time course for wash-in of vehicle (○) or midazolam (●) showing normalized eIPSC charge (upper panel) and amplitude (lower panel). Time windows averaged for baseline and steady-state are highlighted with gray bars. Wash-in period, which begins at 5 minutes, is highlighted with an open box. Values shown are mean ± SEM.

(C) Summary of normalized eIPSC charge versus drug condition. Values are shown as mean ± SEM. For each individual recording, steady-state values were normalized to that recording’s baseline. These normalized values were then pooled for each drug concentration for comparison with vehicle control. Charge reached maximum values in 1 μM midazolam. Number of cells tested are indicated on each column (one-way ANOVA, * p < 0.05, p-values for baseline charge vs. 300 nM: 0.052; 1 μM: 0.019; or 10 μM: 0.090, for charge values, see Table 1).

(D) Summary of normalized eIPSC amplitude versus drug condition. Values are shown as described in (C) (one-way ANOVA, * p < 0.05, p-values for baseline amplitude vs. 300 nM: 0.10; 1 μM: 0.039; or 10 μM: 0.095, for amplitude data, see Table 1).

Midazolam concentration of 1 μM significantly increased eIPSC charge and amplitude when compared to baseline (Figure 1C, 1D, Table 1). 10 μM of midazolam enhanced neither charge nor amplitude and this decreasing effect of higher concentrations of midazolam has been described previously (Khom et al., 2006; Rovira and Ben-Ari, 1993). Diazepam (2 μM), a widely used benzodiazepine in the clinic and in the GABA_AR literature, was used as a benzodiazepine comparison for midazolam (Georgopoulos et al., 2008; Jones-Davis et al., 2005). Diazepam also increased eIPSC charge and amplitude when compared to baseline values (Supplemental Figure 2, Table 1).

Table 1:

Midazolam increases evoked IPSCs (eIPSCs)

Vehicle (DMSO)
		Baseline (raw values)		Steady-state (raw values)			Normalized Values
	n	Charge (pA*ms)	Amp (pA)	Charge (pA*ms)	Amp (pA)	Charge (pA*ms)		Amplitude (pA)
	10	−4.4 ± 0.4	−120 ± 13	−4.6 ± 0.3	−105 ± 9	108 ± 7		95 ± 7
Midazolam
		Baseline (raw values)		Steady-state (raw values)			Normalized Values
Conc (μM)	n	Charge (pA*ms)	Amp (pA)	Charge (pA*ms)	Amp (pA)	Charge (pA*ms)	p-value Charge	Amplitude (pA)	p-value Amplitude
0.3	11	−7.0 ± 1.3	−170 ± 26	−8.6 ± 0.9*	−180 ± 20	140 ± 14	0.052	120 ± 10	0.10
1	17	−9.0 ± 1.2	−180 ± 16	−13 ± 2.3**	−230 ± 30	160 ± 14#	0.019	130 ± 10#	0.039
10	6	−7.7 ± 1.2	−200 ± 29	−9.8 ± 1.1*	−190 ± 20	130 ± 12	0.090	110 ± 6.7	0.090
Diazepam
		Baseline (raw values)		Steady-state (raw values)			Normalized Values
Conc (μM)	n	Charge (pA*ms)	Amp (pA)	Charge (pA*ms)	Amp (pA)	Charge (pA*ms)	p-value Charge	Amplitude (pA)	p-value Amplitude
2	9	−8.0 ± 1.4	−230 ± 39	−13 ± 2.0**	−300 ± 58*	170 ± 14##	0.0010	130 ± 9.5##	0.0099

Raw values are averages of baseline and steady-state values from eIPSCs. Values shown are mean ± SEM. Data were collected and analyzed as described in Figure 1. Steady-state values are significantly different in the steady state period than in baseline if labelled (paired Student’s t-test, * p < 0.05, ** p < 0.01). Normalized values for midazolam and diazepam are significantly different in comparison to vehicle control (one-way ANOVA, ^# p < 0.05, ^## p < 0.01). p-values are shown for normalized drug conditions compared to vehicle conditions.

2.2. Midazolam enhances miniature inhibitory post synaptic currents at hyperpolarized potentials

To further characterize the acute actions of midazolam, we investigated the effect of midazolam on global GABAergic inputs to CA1 pyramidal neurons. Miniature inhibitory post synaptic currents (mIPSCs) were recorded in the presence of APV (50 μM), DNQX (20 μM), and TTX (1 μM) to block glutamate-mediated synaptic transmission and action potential propagation (Figure 2). Using a CsCl based internal solution, mIPSCs were recorded for 5 minutes to establish a baseline at −60 mV, −30 mV, and +30 mV (Figure 2A). Subsequently, vehicle or midazolam (1 μM) was washed in for 15 minutes near the resting membrane potential at −60 mV, followed by a 5 min recording of steady-state mIPSCs at either −60 mV, −30 mV, or +30 mV, respectively (Figure 2A). Example traces of baseline and steady-state mIPSC recordings for all three holding potentials in either vehicle (DMSO, left panels) or midazolam (right panels) are shown in Figure 2.

Figure 2: The effect of midazolam on miniature IPSCs (mIPSCs) in CA1 pyramidal neurons.

(A) Schematic of the experimental protocol. After 5 minutes of baseline recording (gray box) at holding potentials (−60, −30, or +30 mV), either vehicle (DMSO) or drug (midazolam, 1 μM) was washed in for 15 minutes at a holding potential of −60 mV. mIPSCs were subsequently recorded for 5 minutes at the holding potentials in vehicle or drug once steady-state was reached (steady-state, black box).

(B, C, D) Representative traces of GABA_AR-mediated mIPSCs from three CA1 pyramidal neurons at holding potentials of −60 mV (B), −30 mV (C), or +30 mV (D) in the presence of vehicle (DMSO, left panel) or midazolam (right panel). A CsCl-based internal solution with a Cl⁻ reversal potential of 0 mV was used for all traces. Top traces: Representative mIPSCs during baseline (gray) and steady-state (black) of vehicle or midazolam. Middle trace: Expanded trace to show 800 ms of recording. Bottom traces: Average baseline (gray), steady-state (black), and overlay of mIPSC recordings.

To quantify the effect of midazolam on mIPSCs at the various holding potentials, the steady-state mIPSC characteristics were compared to their values during the baseline period (Figure 3, Table 2). Midazolam application increased the steady-state mIPSC charge at hyperpolarized potentials (−60 mV, −30 mV) when compared to baseline mIPSC charge, but not at +30 mV (Figure 3). In addition, midazolam increased current amplitude at −60 mV but, not at other potentials (Supplemental Figure 3, Table 2). In contrast, vehicle wash-in did not affect mIPSC charge or amplitude at any of the holding potentials.

Figure 3: Midazolam increases mIPSC charge at hyperpolarized potentials.

(A) mIPSC charge versus holding potential for baseline (left), vehicle steady-state (middle), and overlay of curves (right). Statistical comparisons were made between baseline and steady-state values for each holding potential using a paired Student’s t-test (* p < 0.05, ** p < 0.01). p-values for −60 mV: 0.28; −30 mV: 0.38; and +30 mV, 0.18.

(B) mIPSC charge versus holding potential for baseline (left), midazolam steady-state (middle), and overlay of curves (right). Statistical comparisons were made between baseline and steady-state values for each potential using a paired Student’s t-test (* p < 0.05, ** p < 0.01). p-values for −60 mV: 2.1e⁻⁴ (**); −30 mV: 0.033 (*); and +30 mV: 0.218.

Table 2:

mIPSC properties during baseline and steady-state after vehicle or drug washin at different holding potentials.

Vehicle (DMSO)
		Baseline (raw values)				Steady-state (raw values)
HP (mV)	n	Frequency (Hz)	Amplitude (pA)	Decay (ms)	Charge (pA*ms)	Frequency (Hz)	Amplitude (pA)	Decay (ms)	Charge (pA*ms)
−60	6	2.7 ± 1.0	27.9 ± 2.2	17.5 ± 0.5	440 ± 40	2.9 ± 1.2	27.1 ± 2.1	18.7 ± 1.0*	470 ± 50
−30	6	0.8 ± 0.1	14.5 ± 0.6	17.3 ± 1.1	230 ± 20	1.1 ± 0.2	13.9 ± 0.5	18.6 ± 1.2	240 ± 20
+30	6	0.9 ± 0.3	25.5 ± 3.2	26.1 ± 2.5	620 ±100	1.1 ± 0.3	20.7 ± 2.1	27.6 ± 2.5*	530 ± 80
Midazolam
		Baseline (raw values)				Steady-state (raw values)
HP (mV)	n	Frequency (Hz)	Amplitude (pA)	Decay (ms)	Charge (pA*ms)	Frequency (Hz)	Amplitude (pA)	Decay (ms)	Charge (pA*ms)
−60	5	5.3 ± 2.5	25.1 ± 1.6	15.8 ± 1.6	350 ± 50	6.6 ± 3.0	30.5 ± 1.6**	21.9 ± 2.0**	500 ± 40**
−30	5	0.6 ± 0.1	14.5 ± 2.6	17.7 ± 1.3	240 ± 50	0.9 ± 0.2**	15.1 ± 1.7	23.5 ± 1.4**	320 ± 30*
+30	5	1.2 ± 0.2	26.2 ± 1.7	25.8 ± 0.6	600 ± 50	1.4 ± 0.3	24.9 ± 1.3	28.8 ± 0.9	650 ± 30
Diazepam
		Baseline (raw values)				Steady-state (raw values)
HP (mV)	n	Frequency (Hz)	Amplitude (pA)	Decay (ms)	Charge (pA*ms)	Frequency (Hz)	Amplitude (pA)	Decay (ms)	Charge (pA*ms)
−60	6	2.2 ± 0.3	30.3 ± 2.7	18.8 ± 1.5	510 ± 80	2.4 ± 0.4	29.5 ± 2.0	24.7 ± 2.2**	660 ± 90*
+30	5	1.0 ± 0.1	24.9 ± 1.8	29.8 ± 1.4	670 ± 60	1.2 ± 0.1	21.5 ± 1.9**	36.9 ± 1.4**	720 ± 70**

Values shown are mean ± SEM and were collected as described in Figure 2A. Holding potential (HP) during baseline and steady state potentials are shown as mean +/− SEM. Normalized values were derived as described in the legend of Figure 1. Values significantly different from baseline (paired Student’s t-test, * p < 0.05, ** p < 0.01).

2.3. Midazolam enhancement of mIPSCs is voltage dependent

To compare the effect of midazolam on mIPSC characteristics across voltages, we normalized steady-state currents for each cell to their respective baseline values (Figure 4). These normalized values were then collapsed across conditions and used to compare different conditions at the same voltage. At a holding potential of −60 mV, midazolam significantly increased mIPSC charge, decay, and amplitude over baseline values (Table 2) and compared to normalized vehicle control (Figure 4). Additionally, mIPSC charge at −30 mV was significantly increased over vehicle control (Figure 4A). In contrast, midazolam had no significant effect on charge at a holding potential of +30 mV when compared to either baseline or normalized vehicle control.

Figure 4: The effect of midazolam on mIPSCs is voltage dependent.

Normalized bar graphs for mIPSC charge (A), decay (B), amplitude (C), and frequency of events (D) for vehicle (DMSO), midazolam (1 μM), or diazepam (2 μM) (mean ± SEM) (Figure 2, Table 2). The different holding potentials are indicated. Data were collected as described in Figure 2A. Diazepam served as a benzodiazepine comparison. Number of cells are indicated in each column in (D). Wash-in values were normalized to the baseline for each cell. Statistical comparisons were made between normalized values for each condition using one-way ANOVA (* p < 0.05, ** p < 0.01).

(A) Charge p-values for vehicle vs. midazolam at a holding potential of: −60 mV: 6.6e⁻⁵; −30 mV: 0.045; or +30 mV: 0.095. Charge p-values for vehicle vs. diazepam at a holding potential of: −60 mV: 0.043; +30 mV: 0.086. Charge p-values for midazolam vs. diazepam at a holding potential of: −60 mV: 0.030; +30 mV: 0.73.

(B) Decay p-values for vehicle vs. midazolam at a holding potential of: −60 mV: 5.3e⁻⁴; −30 mV: 0.029; or +30 mV: 0.33. Decay p-values for vehicle vs. diazepam at a holding potential of: −60 mV: 0.014; or +30 mV: 0.0050. Decay p-values for midazolam vs. diazepam at a holding potential of: −60 mV: 0.53; or +30 mV: 0.13.

(C) Amplitude p-values for vehicle vs. midazolam at a holding potential of: −60 mV: 4.4e⁻⁴; −30 mV: 0.21; or +30 mV: 0.23. Amplitude p-values for vehicle vs. diazepam at a holding potential of: −60 mV: 0.73; or +30 mV: 0.86. Amplitude p-values for midazolam vs. diazepam at a holding potential of: −60 mV: 4.0e⁻³; or +30 mV: 0.064.

(D) Frequency p-values for vehicle vs. midazolam at a holding potential of: −60 mV: 0.18; −30 mV: 0.23; or +30 mV: 0.67. Frequency p-values for vehicle vs. diazepam at a holding potential of: −60 mV: 0.89; or +30 mV: 0.98. Frequency p-values for midazolam vs. diazepam at a holding potential of: −60 mV: 0.30; or +30 mV: 0.92.

To begin to understand the basis for these voltage dependent changes in charge, we characterized the decay and amplitude of mIPSCs. A similar pattern was seen for mIPSC decay as found for charge; midazolam significantly increased decay at hyperpolarized potentials (−60 mV, −30 mV), but not at the depolarized (+30 mV) potential (Figure 4B). Neither mIPSC amplitude nor frequency was significantly increased at either −30 mV or +30 mV when compared to vehicle (Figure 4C, D). Diazepam increased mIPSC charge and decay at −60 mV holding potential whereas it increased only decay at +30 mV (Figure 4). To ensure that this voltage dependent effect by midazolam was not dependent upon the internal solution used, we tested a CsGlu internal solutions. With CsGlu internal solution, mISPC charge, decay, and amplitude were significantly increased at the hyperpolarized (−30 mV) holding potential, but not at 0 or +30 mV (Supplemental Figure 4).

2.4. Depolarized voltages alone increase mIPSCs decay

To investigate the lack of effect of midazolam at the depolarized potential the baseline data containing no vehicle or drug was combined into one data set (Figure 5, Table 2). In Figure 5, the non-normalized mIPSC decay data are shown in comparison across different voltages for the baseline, vehicle wash-in, and midazolam wash-in conditions. Increasing holding potentials from hyperpolarized (−60 mV, −30 mV) to depolarized (+30 mV) potentials significantly increased decay in all three conditions, baseline, vehicle, and midazolam (Figure 5). Therefore, postsynaptic depolarization of the CA1 pyramidal cell is sufficient to increase the mIPSC decay to occlude any increase that would occur with midazolam application. The same pattern of effects on decay, where depolarized voltage increased decay was seen in the CsGlu internal solution (Supplemental Figure 5).

Figure 5: Non-normalized mIPSCs reveal a maximal decay rate at depolarized potentials.

Summary bar graph (mean ± SEM) of mIPSC decay for all baseline recordings compared to the vehicle (DMSO) and midazolam (1 μM) steady-state at −60 mV, −30 mV, and +30 mV holding potentials. Midazolam application does not increase decay more than depolarization of the cell. Statistical comparisons were made between baseline, vehicle steady-state, and midazolam steady-state (one-way ANOVA * p < 0.05, ** p < 0.01).

Statistics for baseline: p-values for comparison of −60 mV vs. −30 mV: 0.99; or +30 mV: 3.1e⁻⁹; comparison of −30 mV vs. +30 mV: 6.8e⁻⁷. For vehicle: p-values for comparison of −60 mV vs −30 mV: 0.93; or +30 mV: .0083; comparison of −30 mV vs. +30 mV: 0.014. For midazolam: p-values for comparison of −60 mV vs. −30 mV: 0.53; or +30 mV: 0.012; comparison of −30 mV vs. +30 mV: 0.014.

3. DISCUSSION

Our results show that midazolam application on organotypic hippocampal slice cultures enhances synaptic inhibitory current charge, decay, and amplitude at hyperpolarized holding potentials. At depolarized potentials, however, midazolam no longer changes these currents. GABA_A mediated currents exhibit a voltage dependent increase in decay which occludes midazolam’s effects. Midazolam and other benzodiazepines are widely used as sedatives and antiepileptics in pediatric and neonatal medicine. These results have implications on the clinical efficacy and neurodevelopmental consequences of midazolam use.

Clinical studies have reported that patient exposure to anesthetics and sedatives, such as midazolam, at a young age correlates with neuroanatomical changes, cognitive impairment, and adverse clinical outcomes (Diaz et al., 2016; Duerden et al., 2016; Wilder et al., 2009). Immature neuronal circuits are sensitive to changes in excitatory/inhibitory neurotransmitter balance (Fagiolini et al., 2004; Ferrer et al., 2018). These synapses undergo developmentally regulated GABA_A receptor (GABA_AR) subunit switching which is altered by benzodiazepine exposure (Jacob et al., 2012; Wisden et al., 1992). In animal models, early anesthetic exposure can cause long-term changes in synaptic development, neuronal survival, and changes in behavioral paradigms (Jevtovic-Todorovic et al., 2003; Xu et al., 2019). Given these findings, we investigated the effect of acute midazolam on postsynaptic currents in ‘immature’ neurons.

We used rodent organotypic hippocampal slice cultures as a model for neurons undergoing rapid synaptogenesis in an established circuit. Specifically, we characterized the effects of acute midazolam application on inhibitory post synaptic currents (IPSCs) in CA1 pyramidal neurons. At a −60 mV holding potential, midazolam increased evoked IPSC (eIPSC) and miniature (mIPSC) current charge, decay, and amplitude with an optimal concentration of 1 μM. This is consistent with the midazolam physiology literature (Rovira and Ben-Ari, 1993) and with pediatric clinical pharmacokinetic data where plasma levels range from 64 ng/mL to 1,000 ng/mL (0.12 – 3 μM) (Pacifici, 2014).

Interestingly, we found that midazolam did not affect mIPSC recorded from depolarized pyramidal cells. A voltage dependent increase in mIPSC decay prevented any further increase in decay by midazolam. Collingridge et al. demonstrated that postsynaptic depolarization results in an increase in mIPSC decay in neurons (Mellor and Randall, 1998). Mellor and Randall further showed that benzodiazepine application increases IPSC decay in hyperpolarized, but not depolarized, cultured cerebellar granule cells (Mellor and Randall, 1998). Other groups have shown that anesthetic agents (propofol, etomidate, and isoflurane) have reduced effects on GABA_AR currents at depolarized potentials because voltage itself increased the probability of channel opening (O’Toole and Jenkins, 2012). Perhaps midazolam does not increase decay above this voltage dependent effect because it uses a similar mechanism or there is an upper limit to GABA_AR decay.

Benzodiazepines may not be as effective in decreasing neuronal activity during pathologically depolarized status, such as status epilepticus. 30–40% of adult status epilepticus is resistant to benzodiazepines and require additional antiepileptic drugs (Brigo et al., 2019). These drugs have different GABA_AR mechanisms of action or affect other ion channels, such as phenobarbital and phenytoin (Glauser and Sankar, 2008). Phenobarbital, a GABA_AR agonist, increases channel open time as opposed to frequency of channel opening used by midazolam (Cooper et al., 1996). A meta-analysis on second line status epilepticus treatment showed that phenobarbital is efficacious (Brigo et al., 2019). Our results suggest that second line antiepileptic drug be used sooner once it appears that a patient has benzodiazepine resistant epilepsy.

These midazolam results may have implications for endogenous modulators of GABA_AR function, such as neurosteroids. Neurosteroids are an intriguing family of endogenous neurochemicals that increase mIPSC current decay in a protein kinase C dependent manner (Harney et al., 2003). Neurosteroid production is upregulated by systemic events, such as stress and developmental states, as well as exogenous drugs (Belelli and Lambert, 2005; Brown et al., 2015). Furthermore, midazolam mediates some of its effects on tonic inhibition of pyramidal neurons via activation of neurosteroid production (Tokuda et al., 2010). Application of midazolam on developing neurons will not only change the balance of inhibition at different neuronal potentials, but also may change GABAergic function by endogenous agonists.

Acute administration of midazolam in organotypic hippocampal slices increases IPSC decay, however, depolarization of the postsynaptic cell eliminates midazolam mediated effects. During non-drug conditions, a neuron has increased GABA_AR inhibition at depolarized polarized in comparison to resting membrane potential. Midazolam application would eliminate this voltage sensitive inhibition. Chronic benzodiazepine administration could eliminate voltage dependent changes in inhibition, which may be required for normal neuronal integration and circuit development. Endogenous neurosteroids increase inhibition for specific developmental or neuroprotective states, whereas exogenous midazolam effects are nonspecific. These results begin to explain how repeated exposure to sedatives and anesthetics, such as midazolam, to immature neuronal circuits can lead to neurodevelopmental changes and result in subsequent cognitive and behavioral impairments in children.

4. EXPERIMENTAL PROCEDURE

4.1. Organotypic Hippocampal Slice Culture Preparation

Hippocampal slice cultures were prepared from postnatal day 6 (P6) mouse pups as described in Opitz-Arya and Barria (Opitz-Araya and Barria, 2011). Briefly, the brain was removed and placed in ice-cold low sodium artificial cerebrospinal fluid (ACSF) containing (mM): 1 CaCl₂, 10 Glucose, 4 KCl, 5 MgCl₂, 26 NaHCO₃, and 234 Sucrose which was bubbled with 95% O₂/ 5% CO₂. The hippocampus was dissected out, cut into 400 μm transverse slices with a McIlwain tissue chopper, plated onto porous membranes (Millicell) and cultured (37°C, 5% CO₂). Recordings were performed after 3–5 days in vitro (DIV 3–5). Slice culture media contained (in mM if not indicated): 1 L-Glutamine, 1 CaCl₂, 2 MgSO₄, 13 Glucose, 5.2 NaHCO₃, 30 HEPES, 8.4 g/L MEM Eagle Medium, 20% Horse serum heat inactivated, 1 mg/L insulin, and 0.00125% ascorbic acid. All animal procedures were approved by the institutional animal care and usage committee (IACUC) at Stony Brook University and were in concordance with the guidelines established by the National Institutes of Health.

4.2. Electrophysiological Recordings

External Solutions:

Recording ACSF consisted of (mM): 125 NaCl, 2.5 KCl, 15 Glucose, 25 NaHCO₃, 1.25 NaH₂PO₄, 2 CaCl₂, 1.2 MgCl₂ and was bubbled with 95% O₂/ 5% CO₂. The following drugs were applied in ACSF: 50 μM APV (Sigma), 20 μM DNQX (Sigma), 1 μM TTX (Alomone Labs), 50 μM Picrotoxin (Tocris) 300 nM-10 μM midazolam (USP), and 2 μM diazepam (Sigma). DMSO was matched to drug conditions by volume in a 1:1 manner as a vehicle control.

Internal Solutions:

The KCl-based internal solution (Cl⁻ reversal potential 0 mV) used for evoked inhibitory post synaptic currents (eIPSC) and miniature IPSC (mIPSC) recordings at −60 mV. The KCl internal solution contained (mM): 120 KCl, 10 HEPES, 2 phosphocreatine, 4 Mg-ATP, 0.3 Na-GTP, and 0.5 EGTA, pH 7.3 (KOH). The CsCl-based internal solution (Cl⁻ reversal potential 0 mV) used for mIPSC recordings at −60, −30 and +30 mV contained (mM): 120 CsCl, 10 HEPES, 2.5 phosphocreatine, 4 Mg-ATP, 0.2 Na-GTP, 0.5 EGTA, and 3 QX314-Cl, pH 7.3 (CsOH). The Cs gluconate (CsGlu) based internal solution (Cl⁻ reversal potential −60 mV) was used for mIPSC recordings at −30, 0, +30 mV holding potentials. CsGlu internal contained (mM): 100 CsGlu, 15 CsCl, 10 HEPES, 20 phosphocreatine, 4 Mg-ATP, 0.3 Na-GTP, 0.5 EGTA, and 3 QX314-Cl, pH 7.3 (CsOH). Recordings were corrected for junction potential. Reversal potentials were calculated with Equation 1 (Hille, 1992), where RT/F = 26.55 at 30 °C:

$$E_{Cl}=\frac{RT}{F}\text{ln}\frac{{\left[Cl\right]}_i}{{\left[Cl\right]}_o}$$ Equation 1:

The reversal potential for CsGlu was −60 mV; because of this, we were not able to obtain Cl- currents while recording mIPSCs at −60mV (data not shown), therefore, KCl internal solution was substituted. Internal solutions used are summarized in Supplemental Table 1.

Electrophysiology:

An EPC 10/2 USB amplifier with PatchMaster software (HEKA Elektronik, Lambrecht, Germany) was used to record membrane potentials or currents at 32 – 34 °C. Recordings were sampled at 10 kHz and low pass filtered using a 4-pole Bessel filter at 3 kHz. Upon achieving the whole cell configuration, we measured the resting membrane potential in current clamp (data not shown). The amplifier mode was changed to voltage clamp and the baseline holding potential set to −60 mV. Input resistance (R_in) and series resistance (R_s) were frequently monitored throughout the experiment with 5 mV hyperpolarizing pulses. Recordings included in analysis showed less than 15% change in R_in or R_s during the experiment.

4.3. Recording Procedures

Evoked IPSC (eIPSC) recordings:

Using the KCl-based internal solution, we recorded eIPSCs from CA1 pyramidal cells at −60 mV. Cells were recorded in standard ACSF containing 50 μM APV to block NMDAR currents and 20 μM DNQX to block AMPAR currents. Bipolar stimulating electrodes, pulled from theta glass, were placed approximately 200 μm away and stimulus intensity was adjusted so that monosynaptic responses were consistently evoked. Interstimulus interval was 20 s. After a baseline interval of 15 sweeps, vehicle (DMSO) or drug (midazolam or diazepam) was washed in continuously. Steady-state of eIPSC occurred after 15 minutes of wash-in.

Miniature IPSC (mIPSC) recordings:

Using CsCl, CsGlu, and KCl-based internal solutions, we recorded mIPSCs from CA1 pyramidal cells at various holdings potentials in standard ACSF containing 50 μM APV, 20 μM DNQX, and 1 μM TTX to block action potentials. After recording mIPSCs for 5 minutes, vehicle (DMSO) or drug (midazolam or diazepam) was washed in for 15 minutes at a holding potential of −60 mV. Steady-state mIPSCs were recorded in vehicle or drug for an additional 5 minutes (22–27 min). When different holding potentials were tested, baseline recordings were performed at the test potential (−60, −30, 0, +30 mV), the holding potential was switched to −60 mV during wash-in, and then the steady-state mIPSCs were recorded at the test potential.

4.4. Data Analysis and Statistics

eIPSC analysis:

Evoked responses were analyzed using custom written programs in Igor Pro (WaveMetrics, Lake Oswego, OR). For the analysis of the individual recording traces, we defined baseline as the current right before the start the stimulation artifact and measured current peak amplitudes in the following 100 milliseconds. Amplitude was defined as the peak current amplitude minus recording baseline post stimulus artifact. The charge was the integral of the current after the stimulus artifact.

The average of the first 15 sweeps was used as baseline for analysis of eIPSCs, while the average of sweeps 66–80 (17–22 minutes after start of wash-in) was used as the steady-state value. The steady-state value was then normalized to the baseline average value for each cell.

mIPSC analysis:

mIPSCs were digitally refiltered at 2 kHz prior to being analyzed using the MiniAnalysis program (Synaptosoft). mIPSCs were analyzed as described before (Akgul and Wollmuth, 2013; Ferrer et al., 2018). Amplitude threshold levels were set at approximately 2.5 times the baseline root mean squared (RMS) noise. Recordings with baseline noise RMS > 5 pA were not included in analysis. Segments with high levels of noise that obscured the baseline were omitted, and event detection resumed when the baseline leveled. After automated detection, high-resolution windows of the recordings were visually inspected to remove false events and to test for possible positive events.

To allow for comparison across cells, we normalized steady-state mIPSC amplitude, decay, and area to values obtained during the baseline period for each cell individually. For analysis of mean value differences across test voltages, the difference was taken between normalized averages for mIPSC data.

Data analysis:

Statistical analysis was done using Minitab (Minitab Inc.). Data sets were tested for normality and homoscedasticity. We used a two-tailed Student’s t-test to test for significant differences (p < 0.05) between baseline and steady-state within the same cell. One-way ANOVA was used to compare across drug and vehicle conditions for normalized values. Error bars were calculated for the standard of the means, with the equation: square root ((σ_a²/n_a) + (σ_b²/n_b)). All raw data and analysis are available upon request.