Enhancement of GABAergic Tonic Currents by Midazolam and Noradrenaline in Rat Substantia Gelatinosa Neurons In Vitro 

• ❖ Intrathecal midazolam and spinally released norepinephrine may produce analgesia by stimulating γ-aminobutyric acid (GABA) receptors

• ❖ In single-cell recordings in the substantia gelatinosa in the spinal cord slices from rats, midazolam enhanced tonic, extrasynaptic GABA^Areceptor currents, and coapplication of norepinephrine, which increases GABA release, further enhanced these inhibitory currents

SUBSTANTIA gelatinosa (SG), the lamina II of the spinal dorsal horn, is one of the crucial sites for the transmission and modification of noxious stimuli. Noxious stimuli are delivered particularly to the superficial dorsal horn through fine myelinated Aδ- and unmyelinated C-primary afferent fibers from the periphery and then transmitted to the upper central nervous system.1–3In the SG of the spinal cord, the noxious stimuli are modified by at least two inhibitory systems. One is γ-aminobutyric acid (GABA)-containing interneurons whose terminals are abundant in the superficial dorsal horn.4–7It has been shown that the primary afferent fibers activate GABAergic and/or glycinergic interneurons through the glutamatergic receptors, resulting in the suppression of nearby SG neurons.8Another is the descending inhibitory system projecting from the brainstem, in which noradrenaline is one of the representative neurotransmitters.9–12 

GABA, an important inhibitory neurotransmitter in the SG, has been known to affect neuronal excitability at the synaptic clefts. However, recent electrophysiologic studies have shown that GABA acts on extrasynaptic GABA type A (GABA^A) receptors, which causes extrasynaptic or tonic currents (inhibition),13,14whereas synaptic GABA^Areceptors produce synaptic or phasic currents. Thus, a clinical importance of GABA^Areceptor-mediated tonic inhibition has been suggested as a target of anesthetic or sedative drugs. It has been reported that an anticonvulsant that increases GABA concentration exerts its action by potentiating tonic rather than phasic inhibition in the hippocampus.15Despite the important role of extrasynaptic inhibitory currents, the effects of anesthetics on extrasynaptic GABA^Areceptors in the SG neurons remain to be elucidated.

It is well known that benzodiazepines enhance the affinity of GABA-GABA^Areceptors through allosteric modulation at benzodiazepine-binding sites.16,17Midazolam, an agonist of benzodiazepine receptors, has been shown to prolong the decay time of the inhibitory postsynaptic currents (IPSCs) in the rat SG neurons.18However, little is known about a direct effect of midazolam on extrasynaptic GABA^Areceptors in the SG neurons. Because midazolam does not activate GABA^Areceptors directly in the absence of GABA,19it is possible that ambient GABA concentration is important for the action of midazolam. Conversely, noradrenaline has been shown to increase GABA release through α¹-adrenergic receptors located at the cell body or presynaptic terminals of GABAergic neurons in the SG.20,21Therefore, noradrenaline may induce spillover of GABA from the synapses to act on extrasynaptic GABA^Areceptors, thereby influencing the effects of midazolam. In the current study, we first divided GABAergic inhibitory currents into synaptic and extrasynaptic ones and then examined the effects of midazolam on each current using whole cell patch-clamp technique in the adult rat spinal cord slices. Furthermore, the effects of coapplication of midazolam and noradrenaline were investigated to determine the interaction of the drugs.

All the experimental procedures involving the use of animals were approved of by the Ethics Committee on Animal experiments, Kyushu University (Fukuoka, Japan) and were in accordance with the United Kingdom Animals (Scientific Procedures) Act 1986 and associated guidelines.

Methods for obtaining adult rat spinal cord slice preparations have been described previously.20In brief, male adult Sprague-Dawley rats (6-8 weeks old) were deeply anesthetized with intraperitoneal urethane (1.5 g/kg), and then lumbosacral laminectomy was performed. The lumbosacral spinal cord (L1-S3) was removed and immersed in preoxygenated ice-cold Krebs. Immediately after the removal of spinal cord, the rats were given an overdose of urethane and were killed by exsanguination. After the dura mater, the ventral and dorsal roots, and the pia-arachnoid membrane were removed, a 500- to 600-μm-thick transverse slice was made using vibrating microslicer (DTK 1500; Dosaka Co. Ltd., Kyoto, Japan). The slice was placed on the nylon mesh in the recording chamber, which had a volume of 0.5 ml, and perfused with Krebs solution saturated with 95% O²and 5% CO²at a rate of 15 ml/min and maintained at 36°± 1°C. The Krebs solution contained 117 mm NaCl, 3.6 mm KCl, 2.5 mm CaCl², 1.2 mm MgCl², 1.2 mm NaH²PO⁴, 25 mm NaHCO³, and 11 mm glucose.

Blind whole cell patch-clamp recordings were made from SG neurons with patch-pipette electrodes having the resistance of 5-10 MΩ. The SG was clearly discernible as a distinct translucent band across the superficial dorsal horn under a dissecting microscope with transmitted illumination as reported previously.1,22The patch-pipette solution was composed of 110 mm Cs²SO⁴, 0.5 mm CaCl², 2 mm MgCl², 5 mm EGTA, 5 mm HEPES, 5 mm tetraethylammonium, and 5 mm ATP-Mg (pH 7.2) for inhibiting postsynaptic effects on K⁺channels. In addition, 2 mm guanosine-5′-O-(2-thiodiphosphate)-β-S, which inhibits the activation of guanosine triphosphate-binding proteins through GABA type B receptors in postsynaptic neurons, was added to the patch-pipette solution. Recording of signals was performed at least 20 min after the establishment of whole cell patch clamp. Signals were acquired with patch-clamp amplifier (Axopatch 200B; Axon Instruments, Union City, CA) at an acquisition rate of 10 kHz. Currents obtained in the voltage clamp mode were low-pass-filtered at 5 kHz and digitized with an A/D converter (Digidate 1440; Axon Instruments). Data were stored and analyzed with a personal computer using the pClamp acquisition program (version 10.1, Axon Instruments).

Drugs were dissolved in Krebs solution and applied by perfusion via  a three-way stopcock without any change in either the perfusion rate or the temperature. The time necessary for the solution to flow from the stopcock to the surface of the spinal cord slice was approximately 20 s. Drugs perfused in this study were midazolam (Wako, Osaka, Japan), flumazenil (Wako), noradrenaline (Wako), strychnine (Sigma-Aldrich, St. Louis, MO), bicuculline (Sigma-Aldrich), guanosine-5′-O-(2-thiodiphosphate)-β-S (Sigma-Aldrich), gabazine (Tocris Cookson, Bristol, United Kingdom), and GABA (Sigma-Aldrich).

To measure the baseline current of each trace, 30 s of the digitized current trace was plotted as all-points in 0.1-pA bins, and the peak of the histogram was determined by pClamp 10.1 software (Axon Instruments). Frequency, amplitude, decay-time constant, and averaged charge transfer of IPSCs (Q^IPSCs, area between the averaged IPSC and the baseline) were analyzed from recordings for 60 s using Mini analysis program (Synaptsoft, Decatur, GA). The change in charge transfer for 1 s produced by drugs (δQ^PC) was calculated by the equation: δQ^PC= F × (Q′^IPSCs− Q^IPSCs) × t, where F is the frequency (Hz) of IPSCs, Q′^IPSCsand Q^IPSCsare the averages of charge transfer per IPSC in the presence and absence of drugs, and t is 1 s. An increased charge transfer (δQ^TC) produced by drugs was calculated according to the equation: δQ^TC= I^TC× t, where I^TCis the average amplitude of the baseline current calculated from 60-s recordings after drug application, and t is 1 s. The average of the total area under IPSCs was calculated by recordings for 60 s before drug application using pClamp 10.1 software. Then the area was again measured for 60 s during peak of the response after drug administration. All numerical data were expressed as the mean ± SE (SEM). Statistical significance was determined as P < 0.05 using Student paired t  test to compare the frequency, amplitude, decay-time constant, and baseline shift. Student unpaired t  test was used to compare the charge transfers of δQ^PCand δQ^TC. Significant difference in the total area was analyzed by one-way analysis of variance followed by Student-Newman-Keuls test. In all cases, n refers to the number of neurons studied.

Stable whole cell recordings were obtained from 128 SG neurons, of which membrane potentials were more negative than −55 mV (average, 68.7 ± 1.6 mV). All the SG neurons exhibited IPSCs with an average frequency of 3.8 ± 0.5 Hz and amplitude of 16.9 ± 1.3 pA, and the holding current was 38.2 ± 4.8 pA at the membrane holding potential of 0 mV. Spontaneous excitatory postsynaptic currents were invisible because of the reversal potential for excitatory postsynaptic currents to be close to 0 mV.

First effects of two GABA^Areceptor antagonists, bicuculline and gabazine, on IPSCs and baseline membrane currents were examined. To block the glycine-evoked IPSCs, a glycine receptor antagonist, strychnine (2 μm) was always present in the following experiments. As shown in figure 1A, bath application of gabazine (1 μm) suppressed IPSCs without significant changes in basal current (boxes a and b). Expanded time scale traces before (box a) and after (box b) application of gabazine are shown in figure 1B, which correspond to figures 1Aa and Ab, respectively. In six of eight (75%) neurons, examined coadministration of bicuculline (20 μm, for 3 min) induced a slow inward current as shown in figure 1Ac. The mean current shown in figures 1Ca–ccorresponds to figures 1Aa–c, respectively. The peak of all- points histogram shifted to more negative values (fig. 1Cd). The baseline shift after additional application of bicuculline (9.5 ± 3.2 pA, P < 0.05, Student t  test, n = 6) was significantly larger than that induced by gabazine only (1.5 ± 1.3 pA, P > 0.05; fig. 1D). When bicuculline (20 μm) was applied without gabazine, IPSCs were abolished as gabazine application, and a baseline current also shifted inwardly in 15 of 19 (79%) SG neurons (8.0 ± 1.1 pA, P < 0.05, n = 15). In addition, the GABA-induced (1 mm) outward current (65.92 ± 21.6 pA) was suppressed by bicuculline (20 μm) but not by gabazine (1 μm; 64.6 ± 20.0 pA, n = 5), suggesting different effects of bicuculline and gabazine at this concentration (<10 μm) on the baseline current, as reported previously.23–25 

Application of midazolam (5 μm, for 3 min) induced slow outward current at the membrane holding potential of 0 mV in the presence of strychnine (fig. 2A). The baseline current and all-points count showed a significant outward shift from the control in 7 of 13 (54%) neurons (4.4 ± 0.8 pA, P < 0.05, n = 7; figs. 2Ba–c). It has been reported that midazolam induces prolongation of decay-time constant of IPSCs in the SG neurons at concentrations more than 1 μm.18,26Therefore, although IPSCs did not seem to be affected by midazolam in the time scale of figures 2A and B, upward shift of all-points count might include changes in the decay time of IPSCs. Thus, we again examined the effects of midazolam in the presence of gabazine (1 μm) to block IPSCs evoked by synaptic GABA^Areceptors. As shown in figure 3A, application of midazolam (5 μm, for 3 min) again evoked an outward current in the presence of gabazine (1 μm) in 7 of 12 (58%) neurons. The upward shift of baseline current shown in figures 3Ba and bcorresponds to figures 3Aa and b, respectively. All-points count is demonstrated in figure 3Bc. The average of the shift was 4.1 ± 0.9 pA (n = 7), which was not different from that induced by midazolam in the absence of gabazine (4.4 ± 0.8 pA; fig. 2). Figure 3Cshows the dose-dependent curve for the midazolam-induced outward current, in which the effective concentration producing half-maximal response (EC⁵⁰) was 2.1 ± 0.03 μm. In the following experiments, the slow outward current or baseline shift recorded in the presence of gabazine (1 μm) was designated as tonic inhibition or current and IPSCs as phasic inhibition. The midazolam-induced outward current was abolished in the presence of flumazenil (1 μm), an antagonist of benzodiazepine receptor, suggesting an involvement of GABA^A-benzodiazepine receptors (n = 10, data not shown).

In the presence of strychnine (2 μm), the frequency and amplitude of GABAergic IPSCs were 2.1 ± 0.2 Hz and 17.1 ± 1.1 pA, respectively, at the membrane holding potential of 0 mV. As shown in figure 4A, neither frequency nor amplitude of IPSCs (fig. 4Aa, control) was affected by perfusion with midazolam (5 μm, for 3 min, fig. 4Ab). The decay-time constant was, however, increased from 28.7 ± 2.5 to 39.3 ± 2.9 ms (P < 0.01, n = 18, paired t  test, 144.1 ± 9.3%) by the application of midazolam (fig. 4B), as reported previously.18,26 Figure 4Cshows the summary of the relative frequency, amplitude, and decay time after the administration of midazolam.

To differentiate the effects of midazolam on phasic and tonic inhibition, charge transfers through GABA^Areceptors for IPSCs and baseline currents were measured separately. The midazolam-induced increase in charge transfers for phasic currents (δQ^PC) was calculated by averaged charge transfers of control (Q^IPSCs) and during midazolam (Q′^IPSCs), frequency (F), and duration (1 s, t; fig. 5Aa), whereas that for tonic current (δQ^TC) was from baseline shift of the tonic currents (I^TC) and 1 s (t) (fig. 5Ab). As shown in figure 5B, the effect of midazolam on tonic currents was much larger than that on phasic currents (δQ^PC, 0.14 ± 0.02 pC, n = 18 and δQ^TC, 4.3 ± 0.7 pC, n = 7, P < 0.01, Student t  test).

Effects of noradrenaline were examined in 10 of 14 (71%) SG neurons that showed the midazolam-induced tonic inhibitory currents. Administration of noradrenaline (20 μm, for 1 min) increased both frequency and amplitude of GABAergic IPSCs (fig. 6Aa, control frequency and amplitude, 3.8 ± 0.5 Hz and 16.9 ± 1.3 pA, respectively; noradrenaline, 14.6 ± 1.8 Hz and 20.6 ± 2.1 pA) in the presence of strychnine. Lower traces show IPSCs in the expanded time scale before (fig. 6Aa1, control) and during noradrenaline application (fig. 6Aa2). After the responses were completely abolished, midazolam (5 μm) was applied for 3 min, and then noradrenaline (20 μm) was again administered for 1 min together with midazolam (fig. 6Ab). The interval of noradrenaline applications was more than 10 min. As shown in figure 6Ab, noradrenaline again increased frequency and amplitude of IPSCs (14.8 ± 1.1 Hz and 20.7 ± 1.6 pA, n = 10), but there were no differences from those during noradrenaline only. Nevertheless, baseline current shifted upward by the simultaneous application of noradrenaline and midazolam (figs. 6Ab1 and b2). The total areas of the currents during control and after noradrenaline application were 0.11 ± 0.02 and 0.86 ± 0.15 × 10⁶pA· ms (P < 0.01), respectively. Conversely, the area after simultaneous application of noradrenaline and midazolam was 1.5 ± 0.22 × 10⁶pA·ms, which was significantly larger than noradrenaline only (P < 0.01) and much greater than the sum of areas after the administration of noradrenaline (0.86 × 10⁶) and midazolam (0.27 ± 0.05 × 10⁶pA·ms) separately (fig. 6B, n = 10). The significant difference between the total areas following the coapplication and noradrenaline only was abolished in the presence of flumazenil (1 μm, n = 4, data not shown), indicating that the effect of midazolam was mediated by GABA^Areceptors.

The total area under the currents includes both IPSCs (phasic currents) and tonic currents. Therefore, to confirm that the increase in the area after simultaneous application of noradrenaline and midazolam was due to an activation of GABAergic tonic current, noradrenaline and midazolam were again administered in the presence of gabazine and strychnine. As shown in figure 7Aa, an application of noradrenaline, which was performed for 1 min after IPSCs were completely blocked by gabazine, induced an outward shift of the baseline current (tonic current). When noradrenaline was administered together with midazolam, the outward current was larger than after the administration of noradrenaline only (fig. 7Ab). There was a significant difference in areas between noradrenaline only (0.41 ± 0.10 × 10⁶) and coapplication of noradrenaline and midazolam (0.64 ± 0.16 × 10⁶pA · ms, n = 9, P < 0.05; fig. 7B).

To investigate the direct effects of midazolam on extrasynaptic GABAergic currents, we first sought to distinguish pharmacologically the synaptic and extrasynaptic currents. GABA^Areceptors are known to be pentameric assemblies of subunits that form a central ion channel.27Nineteen GABA^Areceptor subunits (α1-6, β1-3, γ1-3, δ, ε, θ, π, and ρ1-3) have been cloned from the mammalian central nervous system,13,28consequently numerous combinations of GABA^Areceptors have been synthesized. However, GABA^Areceptors of only a few subunit combinations have been shown to exist.13,29,30Particularly, the receptors containing a γ2 subunit with α1, α2, or α3 subunit are predominantly present in the synaptic clefts, whereas receptors that contain α4, α5, or α6 subunit predominantly or exclusively exist in extrasynaptic sites.13The composition of subunits, especially the difference of α subunits, determines their pharmacologic and electrophysiologic properties.31–34It has been reported that gabazine is a competitive drug for bicuculline, a nonselective GABA^Areceptor antagonist, but a submicromolar concentration of gabazine selectively blocks synaptic currents in the hippocampal and cortical neurons.23–25Therefore, we tested whether a low concentration of gabazine blocked only synaptic currents or not in SG neurons. As shown in figure 1, gabazine (1 μm) completely abolished synaptic currents (IPSCs) without any changes in baseline current. However, the simultaneous application of bicuculline (20 μm) produced an inward current, indicating the presence of intrinsic GABA acting on extrasynaptic GABA^Areceptors. The bicuculline-induced inward current might result from a blockade of the outward GABA current. Furthermore, bath application of GABA (1 mm) evoked an outward current in the presence of gabazine (1 μm) but not in the presence of 20 μm bicuculline. These findings suggest that this concentration of gabazine blocks only synaptic IPSCs without affecting extrasynaptic GABAergic current. The difference in the effects of gabazine and bicuculline is considered to be due to the different subunit compositions between synaptic and extrasynaptic GABA^Areceptors.13,35Thus, the effects of midazolam and noradrenaline on extrasynaptic currents were investigated in the presence of 1 μm gabazine in the current study.

Several studies have demonstrated that intrathecal administration of midazolam produces analgesic effects in human36–38and rats.39–41In the current study, it was demonstrated that midazolam evoked an outward current in more than half of the SG neurons tested in the presence of strychnine, an antagonist of glycine receptors (fig. 2), suggesting an inhibitory influence of midazolam on noxious transmission. The similar amplitude of the midazolam-induced outward current was again observed in the presence of gabazine (fig. 3), indicating that the outward baseline shift was not due to the summation IPSCs but due to an activation of extrasynaptic GABA^Areceptors. Because midazolam does not directly activate GABA^Areceptors but potentiates GABA-GABA^Areceptor affinity,19it is suggested that the midazolam-induced outward current is caused by GABA in the extrasynaptic space. Several reports have shown that ambient GABA concentration was controlled by spillover from inhibitory synaptic cleft42,43or reverse operation of GABA cotransporters located on the neurons and astrocytes.44,45The presence of intrinsic GABA in the extrasynaptic space in the spinal cord slices has been demonstrated by the bicuculline-induced inward currents in mice14,35and rats in the current study (fig. 1).

Midazolam prolonged the decay-time constant but did not affect the frequency or amplitude of synaptic GABAergic currents (IPSCs; fig. 4). It is suggested that this effect is also due to a potentiation of GABA-GABA^Areceptor affinity at the benzodiazepine sites of synaptic GABA^Areceptors, as reported previously.18,26However, the midazolam-induced increase in charge transfers through the extrasynaptic GABAergic receptors (δQ^TC, average, 4.3 pC) was about 30 times greater than that through the synaptic receptors (δQ^PC, 0.14 pC; fig. 5). Therefore, it is possible that the effect of midazolam on extrasynaptic GABA^Areceptors may play an important role in analgesia induced by the intrathecal midazolam, which has been shown in clinical36–38and experimental39–41studies. It seems to be difficult to maintain a 5 μm or more concentration of midazolam in the spinal cord by intravenous injection because of its effects on the cardiovascular system. According to the previous studies, however, intrathecal injection of midazolam can achieve this concentration without hemodynamic change or other marked side effect.36–38As far as our results, the EC⁵⁰for the midazolam-induced outward current was approximately 2 μm (fig. 3). Further studies may be required to define the appropriate dose for the clinical intrathecal analgesia.

Noradrenaline is a representative neurotransmitter of the descending inhibitory system that acts on presynaptic and postsynaptic neurons in the dorsal horn of the spinal cord, thereby modulating nociceptive transmission. In the current study, the application of noradrenaline increased both frequency and amplitude of IPSCs (fig. 6Aa). These findings are consistent with previous studies showing that noradrenaline excites GABAergic neurons through somatodendritic α1-adrenergic receptors21and increases frequency of miniature IPSCs also through α¹receptors located at presynaptic terminals.20In addition, the application of noradrenaline induced an outward current in the presence of gabazine (fig. 7Aa). Because the noradrenaline-induced outward current was observed in the presence of blockers of K⁺channels and G protein-coupled receptors in the current study, it is likely that the outward current was not induced by postsynaptic α²-adrenergic receptors in the SG neurons as reported previously,46but because of an activation of extrasynaptic GABAergic receptors such as midazolam. However, the total area of the noradrenaline-induced outward current in the absence of gabazine, which included both synaptic and extrasynaptic currents, was 0.86 × 10⁶pA · ms (fig. 6B), whereas that of extrasynaptic current in the presence of gabazine was 0.41 × 10⁶pA·ms (fig. 7B). Thus, the activation ratio of extrasynaptic to synaptic currents during noradrenaline application was 0.41/(0.86 − 0.41) = 0.91, suggesting that the effects of noradrenaline on synaptic and extrasynaptic currents were almost equal. Conversely, the effects of midazolam on extrasynaptic and synaptic currents calculated by changes in charge transfer were 4.3 pC (δQ^TC) and 0.14 pC (δQ^PC), respectively, and the ratio was more than 30 as mentioned in the previous section.

Finally, we applied noradrenaline in the presence of midazolam and found that the total area under the current (1.5 × 10⁶pA·ms) was much greater than the sum of areas during application of noradrenaline (0.86 × 10⁶) and midazolam (0.27 × 10⁶pA · ms), separately. Because the increases in frequency and amplitude were not affected by the simultaneous application of midazolam (fig. 6), it was suggested that an application of noradrenaline enhanced the midazolam-induced extrasynaptic inhibitory current. This was confirmed by the finding that the outward current induced by noradrenaline was increased significantly by the simultaneous application of midazolam in the presence of gabazine, which blocked synaptic GABAergic IPSCs (fig. 7). Previous studies have shown that the concentration of extrasynaptic GABA can be changed by synaptic42,43,47and nonsynaptic (activity of neuron-astrocyte unit)44,45,48,49origin of GABA. It is likely that noradrenaline excites GABAergic neurons and increases the possibility of GABA release from the presynaptic terminals through α¹-adrenergic receptors,20,21thereby increasing the ambient GABA concentration because of more frequent spillover of GABA from inhibitory synapses. In addition, it is suggested that monoamines such as noradrenaline can induce GABA release from the astrocytes into the extracellular space.49Either way, noradrenaline increases ambient GABA concentration, resulting in an enhancement of extrasynaptic inhibitory current in the presence of midazolam.

In conclusion, our results demonstrated that midazolam had greater effects on extrasynaptic GABA^Areceptors than the synaptic ones, suggesting a predominant role of the enhancement of GABAergic extrasynaptic currents in midazolam-induced analgesia. In addition, the noradrenaline-induced increase in ambient GABA concentration acts to enhance the extrasynaptic GABAergic currents in the SG of the spinal cord. Although the interaction of midazolam with noradrenaline should be further examined using in vivo  preparations, the current findings suggest a clinical availability of intrathecal administration of midazolam and a drug that increases GABA release such as noradrenaline.