ORIGINAL ARTICLE
Mechanisms Underlying Midazolam-Induced Peripheral Nerve Block and Neurotoxicity Eser Yilmaz, MS,*† Karen A. Hough, AS, CVT, RLAT,*† Gerald F. Gebhart, PhD,*† Brian A. Williams, MD, MBA,*†‡ and Michael S. Gold, PhD*† Background and Objectives: The benzodiazepine midazolam has been reported to facilitate the actions of spinally administrated local anesthetics. Interestingly, despite the lack of convincing evidence for the presence of γ-aminobutyric acid type A (GABAA) receptors along peripheral nerve axons, midazolam also has been shown to have analgesic efficacy when applied alone to peripheral nerves. These observations suggest midazolam-induced nerve block is due to another site of action. Furthermore, because of evidence indicating that midazolam has equal potency at the benzodiazepine site on the GABAA receptor and the 18-kd translocator protein (TSPO), it is possible that at least the nerve-blocking actions of midazolam are mediated by this alternative site of action. Methods: We used the benzodiazepine receptor antagonist flumazenil, and the TSPO antagonist PK11195, with midazolam on rat sciatic nerves and isolated sensory neurons to determine if either receptor mediates midazolam-induced nerve block and/or neurotoxicity. Results: Midazolam (300 μM)–induced block of nerve conduction was reversed by PK11195 (3 μM), but not flumazenil (30 μM). Midazolaminduced neurotoxicity was blocked by neither PK11195 nor flumazenil. Midazolam also causes the release of Ca2+ from internal stores in sensory neurons, and there was a small but significant attenuation of midazolaminduced neurotoxicity by the Ca2+ chelator, BAPTA. BAPTA (30 μM) significantly attenuated midazolam-induced nerve block. Conclusions: Our results indicate that processes underlying midazolaminduced nerve block and neurotoxicity are separable, and suggest that selective activation of TSPO may facilitate modality-selective nerve block while minimizing the potential for neurotoxicity.
actions of local anesthetic–induced peripheral nerve block5 and have analgesic efficacy when applied alone to peripheral nerves.6 Consistent with a direct action on the peripheral nerve, we recently showed that midazolam can produce a small but significant block of compound action potential (CAP) propagation in an isolated sciatic nerve preparation.7 Importantly, we focused on relatively low concentrations in this nerve block study because we had previously demonstrated that higher concentrations of midazolam are neurotoxic.8 Together, these observations suggest that midazolam-induced nerve block is due to an action at site(s) other than the benzodiazepine site on the GABAA receptor, raising the intriguing possibility that identification of this target may suggest a way to maximize the therapeutic utility of midazolam, or better yet, a more effective way to achieve the same therapeutic action while minimizing the potential for neurotoxicity. Furthermore, because of evidence indicating that midazolam has equal potency at both the benzodiazepine receptor found on GABAA receptors and the “peripheral benzodiazepine receptor” or 18-kd translocator protein (TSPO), it is possible that at least the nerve block actions of midazolam are mediated by this alternative site of action. To assess this possibility, we used benzodiazepine receptor and TSPO antagonists to the study of rat sciatic nerve and sensory neurons in isolation to determine whether either receptor was responsible for nerve block and/or neurotoxicity. Our results suggest that TSPO binding of midazolam mediates its nerve block effects, whereas midazolam toxicity is due to a midazolam-induced release of Ca2+ from internal stores via a mechanism independent of either the central benzodiazepine receptor or TSPO.
(Reg Anesth Pain Med 2014;39: 525–533)
W
hile midazolam, a benzodiazepine receptor agonist, is most widely used in the context of general anesthesia for its sedative and amnestic properties, it also has been used off-label in regional anesthesia. Midazolam was first applied intrathecally, based on preclinical data indicating that spinal γ-aminobutyric acid type A (GABAA) receptor activation with exogenous agonists was analgesic.1 The clinical data in support of the analgesic efficacy of spinal benzodiazepine administration is far less compelling.2,3 Nevertheless, midazolam appears to facilitate the actions of spinal administration of local anesthetics.4 Interestingly, despite the absence of evidence in support of a source of GABA along peripheral nerves, perineural midazolam has been shown to facilitate the From the *Center for Pain Research and †Department of Anesthesiology, University of Pittsburgh School of Medicine; and ‡VA Pittsburgh Health System, Pittsburgh, PA. Accepted for publication August 25, 2014. Address correspondence to: Michael S. Gold, PhD, 3500 Terrace St, Room E1440 BST, Pittsburgh, PA 15139 (e‐mail: [email protected]). The authors declare no conflict of interest. This research is supported by the Department of Defense grant OR090012 to B. A.W., M.S.G., and G.F.G. Copyright © 2014 by American Society of Regional Anesthesia and Pain Medicine ISSN: 1098-7339 DOI: 10.1097/AAP.0000000000000176
METHODS Animals Adult male Sprague-Dawley rats (275–350 g; Harlan, Indianapolis, Indiana) were housed in the University of Pittsburgh Association for Assessment and Accreditation of Laboratory Animal Care International–accredited animal care facility with unrestricted food and water on a 12:12-hour light-dark cycle until use. All procedures involving animals were reviewed and approved by the University of Pittsburgh Institutional Animal Care and Use Committee. Animal care and handling were in accordance with the Guide for the Care and Use of Laboratory Animals as adopted and promulgated by the National Institutes of Health.
Compound Action Potential Recording From the Isolated Sciatic Nerve Rat sciatic nerves were harvested and prepared for CAP recording as previously described.7 Briefly, rats were anesthetized with a subcutaneous injection of 1 mL/kg of a mixture of ketamine (55 mg/mL)/xylazine (5.5 mg/mL)/acepromazine (1.1 mg/mL). Sciatic nerves (~30 mm) were quickly dissected and transferred to a dish containing ice-cold Locke solution of the following composition (in mM): 136 NaCl, 5.6 KCl, 14.3 NaHCO3, 1.2 NaH2PO4, 2.2 CaCl2, 1.2 MgCl2, and 11 dextrose, equilibrated continuously
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
525
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Yilmaz et al
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
with 95% O2 and 5% CO2, to pH 7.2 to 7.4. Under a dissecting microscope, nerves were trimmed of excess connective tissue and kept in ice-cold oxygenated Locke solution until use. Nerves were mounted in a recording chamber such that the distal end was laid over 2 platinum stimulating electrodes isolated from the central portion of the nerve with a grease gap. A glass suction electrode connected to a differential preamplifier (0.1–10 kHz; model DAM-80; WPI, Sarasota, Florida) was used to isolate the central end of the nerve. The remainder of the nerve (~15–20 mm) was superfused continuously (2–5 mL/min) with oxygenated Locke solution with and without drugs delivered via a gravity-driven perfusion system. Unless otherwise indicated, CAP experiments were performed at room temperature. Compound action potentials were evoked with electrical pulses delivering constant voltage of 0.2 to 0.5 millisecond in duration and applied at 0.05 Hz, where the stimulus intensity was ~2 times that needed to evoke a maximal amplitude C-fiber component of the CAP. This stimulus intensity was used to ensure the stability of the CAP by eliminating the impact of small changes in action potential threshold on the CAP. In addition, using the same stimulus intensity to evoke both A and C waves enabled simultaneous monitoring of changes in both waveforms.7 Compound action potential data were filtered at 2 kHz (low-pass) and sampled at 20 kHz via a CED 1401 Micro A/D converter and acquired and analyzed using CED Spike 2 version 5 for MS Windows (CED, Cambridge, England). Compound action potential data were rectified, and the average of 6 consecutive CAPs was integrated to quantify A- and C-fiber components of the CAP as area under the curve (AUC). The A-fiber deflection of the CAP (A wave) was easily distinguished from that associated with the C-fiber deflection (C wave) because of the time delay between the arrivals of the 2 waves at the recording electrode. Weighted conduction velocities were calculated by the following formula: CV [m/s] = (nerve length [m]) Σ(amplitude [mV]) / Σ (AOT [mV-s]), where amplitude is the CAP signal at each sampling interval (of 0.06 millisecond) and AOT is amplitude over time, calculated such that AOT = (average amplitude at each sampling interval [mV]) (time(at interval) − time0 [s]).
Cell Culture Isolated sensory neurons were used for assays of either neurotoxicity based on trypan blue exclusion or intracellular Ca2+ experiments using fura-2–based microfluorimetry. Dorsal root ganglia (DRG, L3-L5, bilateral) were harvested from deeply anesthetized rats, enzymatically treated, mechanically dissociated, and plated on poly-D-lysine–coated coverslips as previously described.8 For the neurotoxicity experiments, cells were cultured overnight at 37°C/3% CO2 in a minimal essential media (MEM)–based solution containing ~90% MEM, 10% fetal bovine serum (FBS), 1 MEM-vitamins, 1 penicillin-streptomycin, and 35 ng/mL nerve growth factor (MEM-FBS). For Ca2+-imaging experiments, neurons were cultured at 37°C/3% CO2 in MEM-FBS for 2 hours prior to transfer to an L-15–based media (90% L-15, 10% FBS, 10 mM glucose, and 5 mM HEPES buffered to 7.4 pH), at room temperature prior to study.
Neurotoxicity The neurotoxicity experiments were conducted as described previously.8 Briefly, a 24-hour drug exposure protocol was used. In these experiments, ~24 hours after plating, coverslips were washed and flooded with MEM-FBS–containing test agents. In the first series of experiments, coverslips were treated with test agents that consisted of choline (isotonic/osmotic control), midazolam (300 μM), flumazenil (30 μM), PK11195 (3 μM), midazolam + flumazenil,
526
or midazolam + PK11195 and returned to the CO2 incubator for 24 hours. In a second series of experiments, the Ca2+ chelator, BAPTA-AM, was used as the test agent alone or in combination with midazolam. While 10 μM BAPTA-AM is routinely used in physiological experiments in vitro to block all but the most robust increases in intracellular Ca2+, preliminary results indicated that 10 and even 3 μM BAPTA-AM alone increased DRG neuron cell death after 24 hours in culture (not shown). Consequently, BAPTAAM was used at 1.0 and 0.3 μM alone or in combination with midazolam (300 μM). After incubating coverslips with test agents, neurons were washed twice with fresh MEM-FBS and then incubated with trypan blue (0.1%) for 10 minutes in a HEPES-buffered bath solution (HBS: 130 mM NaCl, 3 mM KCl, 2.5 mM CaCl2, 0.6 mM MgCl2, 10 mM HEPES, 10 mM glucose; 325 mOsm, pH 7.4). Trypan blue was washed twice with fresh bath solution, and neurons were fixed for 30 minutes in ice-cold 4% paraformaldehyde made in 1 phosphate-buffered saline. Coverslips were then transferred to fresh phosphate-buffered saline and stored at 4°C until analysis. Neurotoxicity was quantified as described previously8 by counting all trypan blue–positive and –negative neurons in a single pass through each coverslip at 40x magnification on an inverted microscope (Olympus CK2 Phase, Olympus Optical Co., Tokyo, Japan). Fractional cell death was determined by dividing the number of trypan blue stained neurons by the total number of neurons counted. A single rat was used for each preparation of neurons and each test agent was tested on 2 coverslips from each rat. The average fractional cell death from these 2 coverslips was used as the value for each animal, so that the “n” for each test agent was the number of animals studied.
Ca2+ Imaging Imaging experiments were conducted as described previously.9 Briefly, DRG neurons were loaded with 2.5 μM fura-2 AM ester (TEF Labs, Austin, Texas) with 0.025% pluronic (TEF Labs) for 20 minutes and washed in HBS for 20 minutes at room temperature. Coverslips were then placed in a recording chamber where they were continuously superfused with HBS. In some experiments, coverslips were superfused with HBS followed by Ca2+-free HBS (in mM: 130 NaCl, 3 KCl, 5 MgCl2, 2 EGTA, 10 HEPES, 10 glucose, pH 7.4, osmolality 325 mOsm). Fluorescence data were acquired at room temperature on a PC running Metafluor Software (Molecular Devices, Sunnyvale, California) via a charge-coupled device camera (model RTE/CCD 1300; Roper Scientific, Trenton, New Jersey). The ratio (R) of fluorescence emission at 510 nm in response to 340/380-nm excitation that was controlled by a lambda 10-2 filter changer (Sutter Instrument Co, Novato, California) was acquired at 0.2 Hz during baseline recordings and interstimulus periods and 1 Hz during the application of test compounds. Data for each neuron were analyzed as a change in R in response to a test compound relative to baseline, where an increase of greater than 20% above baseline was considered a response. Because the heterogeneity between neurons was larger than the heterogeneity between preparations, the “n” for the imaging data was neurons rather than rats, although each test agent was tested on neurons from at least 3 rats. The concentration-dependent increase in intracellular Ca2+ in response to midazolam was fitted with a modified Hill equation: Emax * Dn / (Dn + ECn50), where Emax is the maximal increase in fluorescence ratio, D is the concentration of midazolam, EC50 is the concentration of midazolam producing a response 50% of maximal, and n is the Hill coefficient. The EC50 concentration © 2014 American Society of Regional Anesthesia and Pain Medicine
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
of midazolam was used for subsequent experiments with additional test agents.
Test Agents Midazolam was purchased as a preservative-free injectable solution (5 mg/mL; Henry-Schein, Melville, New York) and diluted in culture media or HBS immediately prior to use. Diazepam, another benzodiazepine receptor agonist, was purchased in powder form (Sigma-Aldrich, St Louis, Missouri), dissolved in dimethyl sulfoxide (DMSO) as 1000 stock solution and stored at −20°C until the day of use. PK11195 (Tocris, Minneapolis, Minnesota) and flumazenil (Tocris) were dissolved in DMSO as 100 stock solutions and stored at −20°C until the day of use. BAPTAAM (Life Technologies, Carlsbad, California) was diluted in culture media and HBS as a 100 stock and stored at −20°C until the day of use. The concentration of midazolam used in this study (300 μM) was based on our previous estimate,8 which on a log scale, is close to the high end of the concentration achieved clinically (~460 μM).5 Furthermore, because we sought to determine whether the nerve-blocking or neurotoxic effects of midazolam were due to actions at either the benzodiazepine receptor or TSPO, we used the lowest clinically relevant concentration to produce a significant increase in both nerve block and neurotoxicity.7,8 The concentrations of PK11195 and flumazenil used were based on literature.10,11 The final concentrations of BAPTA-AM (30,
Mechanisms of Midazolam
10, 3, 1, and 0.3 μM) used were also based on the literature and results of initial neurotoxicity and Ca2+-imaging studies on isolated DRG neurons.
Statistical Analysis Data are presented as mean ± SEM. The fractional block of the CAP associated with midazolam with and without antagonists, neurotoxicity, and Ca2+-imaging data were analyzed with a 1-way analysis of variance with the Holm-Sidak test used for post hoc analysis. P < 0.05 was considered statistically significant.
RESULTS Midazolam-Induced CAP Block Midazolam binds to both the benzodiazepine site on the GABAA receptor and TSPO with comparable affinity.12 We therefore sought to determine whether either of these binding sites contributed to midazolam-induced peripheral nerve block. The benzodiazepine receptor antagonist flumazenil (30 μM) and the TSPO ligand PK11195 (3 μM) were used for this purpose. Consistent with the results of our previous study,7 midazolam (300 μM) produced a small but significant reduction in the AUC of the C wave, but not the A wave of the CAP (Fig. 1), and while PK11195 (n = 5) alone had no significant influence on either A or C wave of the
FIGURE 1. Effect of flumazenil and PK11195 on midazolam-induced block of the CAP in isolated rat sciatic nerve. A, Voltage traces from a representative recording illustrating changes in rectified A-wave (left) and C-wave (right) of the CAP evoked before and after application of midazolam (300 μM). Inset: Raw voltage trace illustrating the timing of the stimulus artifact (S) relative to the A-wave and the C-wave of the CAP. Note changes in scale for both amplitude and duration. Pooled data analyzed as a percent change from baseline for the AUC of the A-wave (B) and C-wave (C) in the presence of vehicle (V, n = 5), midazolam (M, n = 6), flumazenil (F, n = 5), PK11195 (P, n = 5), midazolam + flumazenil (M + F, n = 5), midazolam + PK11195 (M + P, n = 9). Insets are pooled data from nerve studied at 32°C to 34°C, M (n = 6), P (n = 6), M + P (n = 7). *P < 0.05, **P < 0.01. © 2014 American Society of Regional Anesthesia and Pain Medicine
527
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
18.68 ± 0.79 17.79 ± 0.72 0.319 ± 0.012 0.291 ± 0.010 18.06 ± 0.43 18.62 ± 0.44 0.261 ± 0.021 0.255 ± 0.019 14.27 ± 0.31 13.88 ± 0.34 0.294 ± 0.004 0.294 ± 0.010 14.18 ± 0.25 13.45 ± 0.39 0.269 ± 0.018 0.271 ± 0.017 A wave (before) A wave (after) C wave (before) C wave (after)
14.37 ± 0.25 14.74 ± 0.40 0.248 ± 0.007 0.261 ± 0.007
14.12 ± 0.66 13.58 ± 0.71 0.293 ± 0.032 0.280 ± 0.032
14.36 ± 0.58 13.93 ± 0.78 0.330 ± 0.009 0.312 ± 0.010
14.24 ± 0.56 14.49 ± 0.47 0.290 ± 0.006 0.309 ± 0.015
14.14 ± 0.22 13.85 ± 0.26 0.280 ± 0.012 0.276 ± 0.011
18.26 ± 0.67 16.71 ± 0.76 0.325 ± 0.018 0.271 ± 0.011*
M + P (n = 7) P (n = 6) M (n = 6) B (n = 4) P + M (n = 9) F + M (n = 5) P (n = 5) F (n = 5) M (n = 6) V (n = 5)
TABLE 1. Impact of Test Agents on Nerve Conduction Velocity
528
Conduction velocity (m/s) as calculated from the time-weighted average of the rectified A- and C-waves of the CAP. Data are average ± SEM values from nerves before and after application of test compounds, where the number of nerves in each group is indicated in parenthesis in the column heading. V is vehicle, M is midazolam (300 μM), F is flumazenil (30 μM), P is PK11195 (3 μM), F + M is flumazenil + midazolam, P + M is PK1119 + midazolam, and B is BAPTA-AM (30 μM). As nerves were pretreated with BAPTA-AM, “before” and “after” are before and after application of midazolam. The last three columns are from nerves studied at 32°C to 34°C. The only statistically significant impact of a test compound on conduction velocity as assessed with a paired t test was the decrease in C-wave conduction velocity following midazolam application in nerves studied at 32°C to 34°C (P < 0.01).
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
Yilmaz et al
CAP, flumazenil (n = 5) produced a significant increase in the C wave of the CAP. Furthermore, while coapplication of midazolam and flumazenil (n = 5) resulted in a change in the AUC of the CAP C-wave, no different than the algebraic sum of the change in the C wave in the presence of midazolam and flumazenil alone, coapplication of midazolam and PK11195 (n = 9) resulted in a significant attenuation of the midazolam-induced reduction in the AUC of the CAP C wave (Fig. 1). None of the test agents significantly affected conduction velocity (Table 1). To confirm the potential involvement of TSPO under more physiological conditions, we repeated the PK11195 experiments at 32°C to 34°C, temperatures closer to those observed in a distal appendage in vivo (insets to Figs. 1B, C). At this higher temperature, the magnitude of the midazolam-induced block of the Cwave CAP (n = 6), assessed as a percent change in AUC from baseline, was even greater than that observed at room temperature. PK11195 alone had no significant influence on the AUC of the CAP (n = 5). Finally, consistent with results obtained at room temperature, the midazolam-induced reduction in the AUC of the CAP C-wave was significantly attenuated by PK11195 (n = 8). However, in contrast to the negative results obtained with midazolam on conduction velocity at room temperature, midazolam produced a small but significant decrease in C-wave conduction velocity, which was only partially attenuated with PK11195 (Table 1).
Midazolam-Induced Neurotoxicity Neither PK11195 (3 μM) nor flumazenil (30 μM) alone affected neuron survival as neuron death rates were comparable to vehicle-treated controls (Fig. 2). As we previously demonstrated,8 midazolam (300 μM) was toxic to sensory neurons, with ~40% of all neurons stained with trypan blue after a 24-hour exposure (n = 6, Fig. 2). This increase in cell death was not significantly altered by the presence of either flumazenil (n = 6) or PK11195 (n = 6, Fig. 2).
Midazolam-Induced Increase in Intracellular Ca2+ Previous studies of smooth muscle cells indicate that midazolam effectively attenuates evoked Ca2+ transients in these cells via several mechanisms.13,14 In astrocytes, however, there is evidence that the opposite is true.15 Furthermore, because an increase in intracellular Ca2+ is often responsible for the neurotoxicity associated with receptor activation, we assessed the impact of midazolam on intracellular Ca2+ in DRG neurons. In contrast to smooth muscle cells, application of midazolam alone was sufficient to drive a rapid increase in intracellular Ca2+ in DRG neurons (Figs. 3A, B). This increase was concentration-dependent, with a threshold between 30 and 100 μM and an EC50 of ~200 μM (n = 33 neurons from 3 different rats, Fig. 3B). All neurons that responded to midazolam with an increase in intracellular Ca2+ also responded to application of 30 mM K+ as evidence of healthy neurons with a resting membrane potential sufficient to enable a depolarization-induced influx of Ca2+ via voltage-gated Ca2+ channels.16 To determine whether the impact of midazolam on the concentration of intracellular Ca2+ in DRG neurons was a drug class effect, we assessed the impact of another benzodiazepine receptor agonist, diazepam, on intracellular Ca2+ levels. Diazepam also led to increases in intracellular Ca2+ in DRG neurons (Fig. 3A). However, the potency of diazepam was significantly less than that of midazolam, with only ~50% of neurons responsive to 1 mM diazepam (Fig. 3C). Furthermore, the diazepam-induced Ca2+ transient was desensitizing such that after a neuron responded to © 2014 American Society of Regional Anesthesia and Pain Medicine
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
Mechanisms of Midazolam
FIGURE 2. Flumazenil and PK11195 do not influence midazolam-induced neurotoxicity. Neurotoxicity was assessed with trypan blue staining. Pooled data are from 6 animals, where V is vehicle containing the combination of choline (to control for osmolality) and DMSO (0.1%) used for the preparation of test agents; M is midazolam (300 μM); F is flumazenil (30 μM); P is PK11195 (3 μM); M + F, midazolam + flumazenil; M + P, midazolam + PK11195. n = 6, **P < 0.01.
diazepam at 1 concentration, subsequent diazepam applications of equal to or greater concentration were unable to induce another Ca2+ transient (Fig. 3A). To rule out a role for either the central benzodiazepine receptor or TSPO in the midazolam-induced increase in intracellular Ca2+, we studied 2 more groups of neurons in which flumazenil (30 and 300 μM) or PK11195 (3 and 30 μM) was applied with an EC50 concentration of midazolam (200 μM). Neither flumazenil (n = 62) nor PK11195 (n = 72) was associated with an increase in intracellular Ca2+ on its own, and neither flumazenil (n = 53) nor PK11195 (n = 52) had any detectable influence on the amplitude of the midazolam-induced increase in intracellular Ca2+ (Fig. 3D) or number of neurons that responded to midazolam with an increase in intracellular Ca2+ (not shown). Finally, to determine whether the midazolam-induced increase in intracellular Ca2+ was due to influx or release from intracellular stores, a group of neurons was challenged with 200 μM midazolam in the presence and absence of Ca2+ in the bath solution. The results of this experiment revealed no detectable difference in the midazolaminduced Ca2+ transient evoked in the presence or absence of extracellular Ca2+ (Figs. 3E, F). While a midazolam-induced increase in intracellular Ca2+ suggests a likely mechanism for the midazolam-induced neurotoxicity, we next tested this suggestion directly. However, because the midazolam-induced Ca2+ transient was not due to the activation of an influx pathway, it was necessary to use an approach that would enable us to prevent the increase in intracellular Ca2+ associated with the release of Ca2+ from internal stores. For this, we used the intracellular Ca2+ chelator BAPTA-AM. In an initial set of Ca2+-imaging experiments, we were able to confirm that preincubating DRG neurons with 30 μM BAPTA-AM for 30 minutes resulted in a complete block of the midazolam-induced Ca2+ transient (n = 73, Fig. 4A). The midazolam-induced Ca2+ transient was blocked by greater than 80%, 70%, and 60% in neurons preincubated with 10, 3, and 1 μM BAPTA-AM, respectively (Fig. 4A). © 2014 American Society of Regional Anesthesia and Pain Medicine
For the neurotoxicity experiments, DRG neurons were incubated either in the presence or absence of midazolam, with or without BAPTA-AM. Normal resting levels of Ca2+ are essential for a variety of cellular processes. Therefore, we examined a range of concentrations of BAPTA-AM in the absence of midazolam to determine an appropriate concentration(s) of BAPTAAM for these experiments. We found a significant increase in trypan blue staining in DRG neurons (cultured for 24 hours) following a 30-minute incubation in either 30 μM (66.8 ± 8.5%, n = 4) or 10 μM (27.15 ± 2.86%, n = 3) BAPTA-AM, but not after 1 or 0.3 μM BAPTA-AM (Fig. 4B). Whereas 1 and 0.3 μM BAPTA-AM were without effect on trypan blue staining relative to vehicle-treated controls, 1 μM BAPTA-AM produced a small but significant decrease in midazolam-induced neurotoxicity (Fig. 4B). Finally, because it is also possible that an increase in intracellular Ca2+ contributes to the midazolam-induced nerve block, we recorded CAPs in the presence or absence of midazolam with or without a 30-minute preincubation with BAPTA-AM (30 μM). BAPTA-AM had no significant influence on the CAP AUC (10.06 ± 0.81 mV-s with BAPTA-AM, 11.10 ± 0.78 mV-s without BAPTA-AM for A-CAP, and 21.55 ± 5.21 μV-s with BAPTA-AM, 26.83 ± 3.22 μV-s without BAPTA-AM for C-CAP), conduction velocity (14.27 ± 0.31 m/s with BAPTA-AM, 14.37 ± 0.25 m/s without BAPTA-AM for A-CAP, 0.294 ± 0.004 m/s with BAPTA-AM, 0.248 ± 0.007 m/s without BAPTA-AM for C-CAP), or amplitude (7.8 ± 1.25 mV with BAPTA-AM, 9.3 ± 1.05 mV without BAPTA-AM for A-CAP, 111.25 ± 13.9 μV with BAPTAAM, 123.75 ± 28.16 μV without BAPTA-AM for C-CAP). However, midazolam-induced block of the C-CAP was attenuated significantly (P < 0.01) in the presence of BAPTA (n = 4, Fig. 4C).
DISCUSSION The purpose of this study was to determine whether midazolaminduced nerve block and/or neurotoxicity involves the activation
529
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Yilmaz et al
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
of the central benzodiazepine receptor or TSPO. Our results indicate that while midazolam-induced nerve block was significantly attenuated by the TSPO ligand PK11195, midazolam neurotoxicity was not. The midazolam-induced neurotoxicity was significantly attenuated by a concentration of BAPTA-AM that produced no detectable increase in toxicity alone (1 μM), but that significantly attenuated the midazolam-induced increase in intracellular Ca2+. The midazolam-induced increase in intracellular Ca2+ was not attenuated by either flumazenil or PK11195 and also was not attenuated in the absence of extracellular Ca2+. Finally, BAPTA-AM produced a significant attenuation of midazolaminduced block of the CAP C-wave. These results suggest that different mechanisms underlie midazolam-induced nerve block and neurotoxicity; the nerve block is dependent on the activation of TSPO and an increase in intracellular Ca2+, whereas neurotoxicity is dependent, at least in part, on an increase in intracellular Ca2+.
Our results with PK11195 suggest that midazolam-induced block of the C-wave is due, at least in part, to an action at TSPO; TSPO has been implicated in numerous mitochondrial processes including the induction of apoptosis.17–19 We suggest that midazolam is unlikely to engage TSPO-dependent apoptotic pathways in sensory neurons, given the absence of a detectable influence of PK11195 on the midazolam-induced increase in trypan blue staining. However, it is possible that binding of midazolam to TSPO leads to a reduction in mitochondrial respiration, hence reduced ATP production. This, in turn, could lead to inhibition of proteins essential for the maintenance of membrane potential after action potential firing (eg, sodium/potassium ATPase) and consequently a depolarization-induced block of action potential propagation. Alternatively, an increase in intracellular Ca2+ downstream of TSPO activation could account for the differential block of C- over A-fibers, if the Ca2+-dependent processes responsible for conduction block
FIGURE 3. Midazolam induces increases in intracellular Ca2+ in DRG neurons. A, Representative Ca2+ transients as assessed by changes in the fura-2 fluorescence ratio in response to increasing concentrations of midazolam (top traces) or diazepam (bottom traces). B, Pooled concentration-response data (n = 33 neurons from 3 rats) for midazolam were fitted with a Hill equation to determine the concentration associated with an peak increase in Ca2+ that was 50% of the maximal increase (EC50), as well as the maximal increase (Emax). C, Because the responses to diazepam (n = 42 to 68 cells per diazepam concentration) were desensitizing, we estimated differences in the potency of midazolam and diazepam with cumulative response plots, where the percentage of neurons responding to each concentration of test agent is plotted. There was a concentration of both compounds that produced a response in ~100% of the neurons tested. Consequently, both cumulative distributions were also well fitted with a Hill equation used to estimate the concentration at which 50% of the neurons tested were responsive. D, Pooled data from neurons treated with an EC50 concentration of midazolam alone (200 μM, n = 72), flumazenil (F, 30 μM, n = 53), and PK11195 (P, 3 μM, n = 52), followed by the combination of M + F (n = 53) and M + P (n = 52) applied to the same F and P neurons, respectively, indicate that neither antagonist had a detectable influence on the midazolam-induced Ca2+ transient. E, An example of midazolam-induced Ca2+ transients evoked from a neuron in the absence (− Ca2+) and presence (+Ca2+) of Ca2+ in the bath solution. F, Pooled data (n = 18) indicate that there was no detectable influence of extracellular Ca2+ on the midazolam-induced Ca2+ transient.
530
© 2014 American Society of Regional Anesthesia and Pain Medicine
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
Mechanisms of Midazolam
FIGURE 4. Effect of intracellular Ca2+ chelation on the actions of midazolam on sensory neurons. The Ca2+ chelator BAPTA, applied as the membrane-permeable AM ester, was used to block midazolam (300 μM)–induced increase in intracellular Ca2+. A, Pooled data from isolated sensory neurons challenged with 300 μM midazolam indicate that block of the transient was concentration dependent, with the transient completely blocked in close to 100% of the neurons tested at 30 μM BAPTA (n = 28 to 73 neurons per BAPTA concentration). B, Pooled neurotoxicity data from neurons treated with BAPTA alone or in combination with midazolam (n = 4 per group). V is vehicle; .3B is 0.3 μM BAPTA; 1B is 1 μM BAPTA; M is midazolam (300 μM). C, Pooled C-wave CAP data from nerves treated with midazolam (M, 300 μM) alone (n = 6) and a separate group of nerves (n = 4) treated with 30 μM BAPTA prior to the application of midazolam (300 μM, n = 4). The midazolam-alone data group are from Figure 1 and are replotted here for comparison. *P < 0.05, **P < 0.01.
were differentially distributed between C- and A-fibers. It is also possible that processes associated with TSPO activation and the increase in intracellular Ca2+ occur in parallel. Thus, additional experiments will be needed to define the mechanisms mediating the link between midazolam, TSPO, and nerve block. The results of the present study suggest midazolam-induced nerve block is relatively selective for C-fibers. However, a ~20% block of the A-fiber CAP was observed at 32°C to 34°C in the present study, and we previously reported a midazolam-induced A-fiber block roughly half that of the C-fiber block.7 Careful analysis of both sets of data revealed what appear to be 2 groups of nerves: one in which 300 μM midazolam blocks between 10% and 20% of the A-wave and another in which little if any block is detected. In our previous study, 5 of the 6 nerves studied had a block between 10% and 20%, whereas in the present study, only 3 of the 6 nerves studied had an A-wave block between 10% and 20%, the block in 1 nerve was less than 5%, and in 2 nerves there was a slight increase in amplitude with time that was not corrected. If 300 μM midazolam is close to the threshold concentration necessary to activate processes responsible for the A-fiber block, it could account for the variability observed between nerves. The observation that flumazenil selectively increased the amplitude of the C-CAP was unexpected. While off-target effects of even the most pharmacologically selective compounds have been described, this observation raises the intriguing possibility that action potential propagation in peripheral C-fibers is regulated by GABAA receptor activation. The presence of functional GABAA receptors in peripheral nerves is suggested by the previous observations that peripheral administration of GABAA receptor agonists is antinociceptive.20,21 However, while resident immune cells would be one of the only endogenous sources of GABA22 needed for such a mechanism to explain the impact of flumazenil, the density of immune cells in peripheral nerves in the absence of injury is low. Furthermore, flumazenil should have blocked not only endogenous GABA signaling, but also the actions of midazolam, resulting in a change in the amplitude of the C-CAP comparable to that observed with flumazenil alone. Thus, we suggest that an off-target effect is more likely. Regardless, the negative results with flumazenil should still be interpreted with caution, given evidence of benzodiazepine binding to GABAA receptors through flumazenil-independent mechanisms (ie, independent of the γ2 subunit), as well as the possibility that the concentration used was insufficient to completely block the relatively high concentration of midazolam used, at the benzodiazepine site. © 2014 American Society of Regional Anesthesia and Pain Medicine
The midazolam-induced increase in intracellular Ca2+ via release of Ca2+ from internal stores not only provides a mechanism for midazolam-induced neurotoxicity, but also suggests a novel site of action for this compound. The observation that midazolaminduced Ca2+ transients were present in all neurons tested suggests that the mechanism(s) responsible for the Ca2+ transient are ubiquitously expressed among DRG neurons. Our previous results indicate that the relative contribution of various Ca2+ regulatory elements to the magnitude and duration of the depolarizationinduced Ca2+ transient varies among subpopulations of DRG neurons.16 Nevertheless, there are 2 internal sources of Ca2+ that are ubiquitous among sensory neurons. One source is mitochondria. Interestingly, a Ca2+ regulatory protein in mitochondria, the Na+/ Ca2+/Li+-exchanger (NCLX), is blocked by benzodiazepines.23 However, we consider block of NCLX an unlikely mechanism for the midazolam-induced increase in intracellular Ca2+ because midazolam NCLX block would be expected to decrease the release of Ca2+ from mitochondria. Furthermore, while a nonspecific action on mitochondria leading to a breakdown of the proton gradient will result in an increase in intracellular Ca2+ in all DRG neurons, our previous results16 suggest that the transient associated with the breakdown of the mitochondria proton gradient is considerably smaller than that associated with midazolam. A second ubiquitous source of Ca2+ in sensory neurons is the endoplasmic reticulum,16 which is capable of sustaining robust and reproducible Ca2+ transients in sensory neurons comparable to midazolam-evoked transients, even in the presence of Ca2+-free bath solution.9 Thus, while additional experiments are needed to confirm the source of the midazolam-induced Ca2+ transient, the endoplasmic reticulum is the most likely source. Moreover, our observation that diazepam can also drive an increase in intracellular Ca2+ suggests that this might be a drug class effect not previously described. While attenuation of the midazolam-induced neurotoxicity by 1 μM BAPTA was significant, this attenuation was far from a complete block. This observation raises the possibility that midazolam activated a Ca2+-independent apoptotic pathway, such as phospholipase A2–mediated apoptosis.24 Nevertheless, consistent with the observation that 1 μM BAPTA was insufficient to completely block the midazolam-induced Ca2+ transient, it appears more likely that the residual toxicity was due to BAPTA-insensitive increase in intracellular Ca2+. To directly test this possibility, however, it will likely be necessary to identify the mechanism(s) underlying the increase in intracellular Ca2+ so that this mechanism
531
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Yilmaz et al
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
can be specifically manipulated in a way that does not produce toxic changes in resting intracellular Ca2+ concentrations. Identification of this mechanism(s) may also enable ways to further maximize the therapeutic actions of midazolam while minimizing toxic adverse effects. While TSPO appears to be the primary mechanism underlying the nerve-blocking actions of midazolam, our results with BAPTA-AM suggest that an increase in intracellular Ca2+ may also contribute to this effect. This begs the question as to the basis of the Ca2+-induced attenuation of action potential propagation in C-fibers. One possibility is a depolarization-induced inactivation of voltage-gated Na+ channels necessary for the propagation of action potentials secondary to the increase in intracellular Ca2+. Although there is evidence for functional Ca2+-dependent K+ channels in peripheral axons, we believe that an increase in K+ channel activity is unlikely to account for the actions of midazolam. That is, an increase in K+ channel activity will result in membrane hyperpolarization, which should relieve Na+ channels from an inactivated state and thereby increase AP conduction velocity, if not also CAP amplitude. Consistent with this expectation, our preliminary results indicate that the Ca2+-dependent K+ channel opener NS1619 (100 μM) results in an increase in the amplitude of both A (39% ± 12%) and C (47% ± 7.4%) waves of the CAP (n = 3). Mechanisms that could account for a Ca2+-induced depolarization of the sciatic nerve include the activation of the electrogenic plasma membrane Na+/Ca2+ exchanger and/or the activation of Ca2+-dependent Cl− channels. However, the limited impact of midazolam on conduction velocity argues against a depolarizationinduced inactivation of Na+ channels, given evidence that local anesthetic–induced CAP block is preceded by conduction velocity slowing.25 Additional experiments will be needed to confirm mechanism(s) linking increased intracellular Ca2+ to CAP block. While there are several interesting and potentially important implications of the results of this study, there are also important limitations. First, data were generated in rat tissue. Therefore, species differences may influence results and conclusions. For example, midazolam-induced toxicity and drug kinetics may differ between rats and humans; thus, relatively limited neurotoxicity observed in patients may be due to a species difference and/or a concentration difference, where toxic doses are rarely, if ever, achieved clinically. Second, in our CAP recording experiments, we used bath application of drugs to the entire sciatic nerve. Given the limited length of nerve directly impacted by clinical manipulations and the extent of the safety factor in the intact nerve, the relative drug concentrations we used in these experiments may be considerably higher than those obtained clinically. In addition, diffusion will also significantly reduce the magnitude of effects observed clinically. Third, we have assessed neurotoxicity in isolated sensory neurons. These neurons might be more vulnerable than neurons in an intact system because they are devoid of sources of protection found in vivo including support cells and vasculature. And fourth, while we have interpreted both neurotoxicity and nerve block data in the context of results obtained with Ca2+ imaging, the imaging data were collected at room temperature, and the toxicity data were collected at 37°C. As there is evidence to suggest that the Ca2+ release from intracellular stores is temperature-dependent, where, for example, low-amplitude Ca2+ oscillations at room temperature become spikes at 37°C in T lymphocytes, we may have underestimated the impact of midazolam on Ca2+ release in DRG neurons. Given that ~184 μM midazolam is a threshold concentration for neurotoxicity,8 and ~100 μM is the threshold concentration for an increase in intracellular Ca2+ at room temperature, such an underestimate would add further support to the suggestion that Ca2+-independent mechanisms contribute to midazolam-induced neurotoxicity.
532
The present results are important for multiple reasons. First, the evidence that a C-fiber–selective midazolam-induced nerve block is mediated by TSPO suggests a new direction for development of approaches that could provide a modality-selective nerve block. Second, our study provides evidence that different mechanisms mediate midazolam-induced nerve block and neurotoxicity. This finding may suggest approaches to both minimize neurotoxicity and maximize therapeutic utility. And third, with respect to clinical use of peripheral nerve blocks, our results continue to argue against the clinical perineural use of midazolam at concentrations even close to that that may be neurotoxic alone or in combination with local anesthetics that are also able to drive toxic increases in intracellular Ca2+.8
REFERENCES 1. Yamamoto T, Yaksh TL. Spinal pharmacology of thermal hyperesthesia induced by incomplete ligation of sciatic nerve. I. Opioid and nonopioid receptors. Anesthesiology. 1991;75:817–826. 2. Dirig DM, Yaksh TL. Intrathecal baclofen and muscimol, but not midazolam, are antinociceptive using the rat-formalin model. J Pharmacol Exp Ther. 1995;275:219–227. 3. Goodchild CS, Serrao JM. Intrathecal midazolam in the rat: evidence for spinally-mediated analgesia. Br J Anaesth. 1987;59:1563–1570. 4. Bharti N, Madan R, Mohanty PR, et al. Intrathecal midazolam added to bupivacaine improves the duration and quality of spinal anaesthesia. Acta Anaesthesiol Scand. 2003;47:1101–1105. 5. Jarbo K, Batra YK, Panda NB. Brachial plexus block with midazolam and bupivacaine improves analgesia. Can J Anaesth. 2005;52:822–826. 6. Williams BA, Murinson BB, Grable BR, et al. Future considerations for pharmacologic adjuvants in single-injection peripheral nerve blocks for patients with diabetes mellitus. Reg Anesth Pain Med. 2009; 34:445–457. 7. Yilmaz-Rastoder E, Gold MS, Hough KA, et al. Effect of adjuvant drugs on the action of local anesthetics in isolated rat sciatic nerves. Reg Anesth Pain Med. 2012;37:403–409. 8. Williams BA, Hough KA, Tsui BY, et al. Neurotoxicity of adjuvants used in perineural anesthesia and analgesia in comparison with ropivacaine. Reg Anesth Pain Med. 2011;36:225–230. 9. Scheff NN, Lu SG, Gold MS. Contribution of endoplasmic reticulum Ca2+ regulatory mechanisms to the inflammation-induced increase in the evoked Ca2+ transient in rat cutaneous dorsal root ganglion neurons. Cell Calcium. 2013;54:46–56. 10. Awad M, Gavish M. Binding of [3H]Ro 5-4864 and [3H]PK 11195 to cerebral cortex and peripheral tissues of various species: species differences and heterogeneity in peripheral benzodiazepine binding sites. J Neurochem. 1987;49:1407–1414. 11. Weiss M, Tikhonov D, Buldakova S. Effect of flumazenil on GABAA receptors in isolated rat hippocampal neurons. Neurochem Res. 2002;27:1605–1612. 12. Bender AS, Hertz L. Pharmacological characteristics of diazepam receptors in neurons and astrocytes in primary cultures. J Neurosci Res. 1987;18:366–372. 13. Yoshimura H, Kai T, Nishimura J, et al. Effects of midazolam on intracellular Ca2+ and tension in airway smooth muscles. Anesthesiology. 1995;83:1009–1020. 14. Yang M, Ding X, Murray PA. Differential effects of intravenous anesthetics on capacitative calcium entry in human pulmonary artery smooth muscle cells. Am J Physiol Lung Cell Mol Physiol. 2008;294: L1007–L1012. 15. Code WE, Hertz L, White HS. Midazolam potently affects free calcium concentration in astrocytes. Can J Anaesth. 1990;37:S27.
© 2014 American Society of Regional Anesthesia and Pain Medicine
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
16. Lu SG, Zhang X, Gold MS. Intracellular calcium regulation among subpopulations of rat dorsal root ganglion neurons. J Physiol. 2006;577:169–190. 17. Veenman L, Gavish M. The role of 18 kDa mitochondrial translocator protein (TSPO) in programmed cell death, and effects of steroids on TSPO expression. Curr Mol Med. 2012;12:398–412. 18. Rupprecht R, Papadopoulos V, Rammes G, et al. Translocator protein (18 kDa) (TSPO) as a therapeutic target for neurological and psychiatric disorders. Nat Rev Drug Discov. 2010;9:971–988.
Mechanisms of Midazolam
21. Reis G, Pacheco D, Francischi J, et al. Involvement of GABA A receptor–associated chloride channels in the peripheral antinociceptive effect induced by GABA A receptor agonist muscimol. Eur J Pharmacol. 2007;564:112–115. 22. Bhat R, Axtell R, Mitra A, et al. Inhibitory role for GABA in autoimmune inflammation. Proc Natl Acad Sci U S A. 2010;107:2580–2585. 23. Palty R, Sekler I. The mitochondrial Na(+)/Ca(2+) exchanger. Cell Calcium. 2012;52:9–15.
19. Gatliff J, Campanella M. The 18 kDa translocator protein (TSPO): a new perspective in mitochondrial biology. Curr Mol Med. 2012;12:356–368.
24. Sun GY, Shelat PB, Jensen MB, et al. Phospholipases A2 and inflammatory responses in the central nervous system. Neuromolecular Med. 2010;12: 133–148.
20. Carlton SM, Zhou S, Coggeshall RE. Peripheral GABA(A) receptors: evidence for peripheral primary afferent depolarization. Neuroscience. 1999;93:713–722.
25. Gokin AP, Philip B, Strichartz GR. Preferential block of small myelinated sensory and motor fibers by lidocaine: in vivo electrophysiology in the rat sciatic nerve. Anesthesiology. 2001;95:1441–1454.
© 2014 American Society of Regional Anesthesia and Pain Medicine
533
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Mechanisms Underlying Midazolam-Induced Peripheral Nerve Block and Neurotoxicity Eser Yilmaz, MS,*† Karen A. Hough, AS, CVT, RLAT,*† Gerald F. Gebhart, PhD,*† Brian A. Williams, MD, MBA,*†‡ and Michael S. Gold, PhD*† Background and Objectives: The benzodiazepine midazolam has been reported to facilitate the actions of spinally administrated local anesthetics. Interestingly, despite the lack of convincing evidence for the presence of γ-aminobutyric acid type A (GABAA) receptors along peripheral nerve axons, midazolam also has been shown to have analgesic efficacy when applied alone to peripheral nerves. These observations suggest midazolam-induced nerve block is due to another site of action. Furthermore, because of evidence indicating that midazolam has equal potency at the benzodiazepine site on the GABAA receptor and the 18-kd translocator protein (TSPO), it is possible that at least the nerve-blocking actions of midazolam are mediated by this alternative site of action. Methods: We used the benzodiazepine receptor antagonist flumazenil, and the TSPO antagonist PK11195, with midazolam on rat sciatic nerves and isolated sensory neurons to determine if either receptor mediates midazolam-induced nerve block and/or neurotoxicity. Results: Midazolam (300 μM)–induced block of nerve conduction was reversed by PK11195 (3 μM), but not flumazenil (30 μM). Midazolaminduced neurotoxicity was blocked by neither PK11195 nor flumazenil. Midazolam also causes the release of Ca2+ from internal stores in sensory neurons, and there was a small but significant attenuation of midazolaminduced neurotoxicity by the Ca2+ chelator, BAPTA. BAPTA (30 μM) significantly attenuated midazolam-induced nerve block. Conclusions: Our results indicate that processes underlying midazolaminduced nerve block and neurotoxicity are separable, and suggest that selective activation of TSPO may facilitate modality-selective nerve block while minimizing the potential for neurotoxicity.
actions of local anesthetic–induced peripheral nerve block5 and have analgesic efficacy when applied alone to peripheral nerves.6 Consistent with a direct action on the peripheral nerve, we recently showed that midazolam can produce a small but significant block of compound action potential (CAP) propagation in an isolated sciatic nerve preparation.7 Importantly, we focused on relatively low concentrations in this nerve block study because we had previously demonstrated that higher concentrations of midazolam are neurotoxic.8 Together, these observations suggest that midazolam-induced nerve block is due to an action at site(s) other than the benzodiazepine site on the GABAA receptor, raising the intriguing possibility that identification of this target may suggest a way to maximize the therapeutic utility of midazolam, or better yet, a more effective way to achieve the same therapeutic action while minimizing the potential for neurotoxicity. Furthermore, because of evidence indicating that midazolam has equal potency at both the benzodiazepine receptor found on GABAA receptors and the “peripheral benzodiazepine receptor” or 18-kd translocator protein (TSPO), it is possible that at least the nerve block actions of midazolam are mediated by this alternative site of action. To assess this possibility, we used benzodiazepine receptor and TSPO antagonists to the study of rat sciatic nerve and sensory neurons in isolation to determine whether either receptor was responsible for nerve block and/or neurotoxicity. Our results suggest that TSPO binding of midazolam mediates its nerve block effects, whereas midazolam toxicity is due to a midazolam-induced release of Ca2+ from internal stores via a mechanism independent of either the central benzodiazepine receptor or TSPO.
(Reg Anesth Pain Med 2014;39: 525–533)
W
hile midazolam, a benzodiazepine receptor agonist, is most widely used in the context of general anesthesia for its sedative and amnestic properties, it also has been used off-label in regional anesthesia. Midazolam was first applied intrathecally, based on preclinical data indicating that spinal γ-aminobutyric acid type A (GABAA) receptor activation with exogenous agonists was analgesic.1 The clinical data in support of the analgesic efficacy of spinal benzodiazepine administration is far less compelling.2,3 Nevertheless, midazolam appears to facilitate the actions of spinal administration of local anesthetics.4 Interestingly, despite the absence of evidence in support of a source of GABA along peripheral nerves, perineural midazolam has been shown to facilitate the From the *Center for Pain Research and †Department of Anesthesiology, University of Pittsburgh School of Medicine; and ‡VA Pittsburgh Health System, Pittsburgh, PA. Accepted for publication August 25, 2014. Address correspondence to: Michael S. Gold, PhD, 3500 Terrace St, Room E1440 BST, Pittsburgh, PA 15139 (e‐mail: [email protected]). The authors declare no conflict of interest. This research is supported by the Department of Defense grant OR090012 to B. A.W., M.S.G., and G.F.G. Copyright © 2014 by American Society of Regional Anesthesia and Pain Medicine ISSN: 1098-7339 DOI: 10.1097/AAP.0000000000000176
METHODS Animals Adult male Sprague-Dawley rats (275–350 g; Harlan, Indianapolis, Indiana) were housed in the University of Pittsburgh Association for Assessment and Accreditation of Laboratory Animal Care International–accredited animal care facility with unrestricted food and water on a 12:12-hour light-dark cycle until use. All procedures involving animals were reviewed and approved by the University of Pittsburgh Institutional Animal Care and Use Committee. Animal care and handling were in accordance with the Guide for the Care and Use of Laboratory Animals as adopted and promulgated by the National Institutes of Health.
Compound Action Potential Recording From the Isolated Sciatic Nerve Rat sciatic nerves were harvested and prepared for CAP recording as previously described.7 Briefly, rats were anesthetized with a subcutaneous injection of 1 mL/kg of a mixture of ketamine (55 mg/mL)/xylazine (5.5 mg/mL)/acepromazine (1.1 mg/mL). Sciatic nerves (~30 mm) were quickly dissected and transferred to a dish containing ice-cold Locke solution of the following composition (in mM): 136 NaCl, 5.6 KCl, 14.3 NaHCO3, 1.2 NaH2PO4, 2.2 CaCl2, 1.2 MgCl2, and 11 dextrose, equilibrated continuously
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
525
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Yilmaz et al
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
with 95% O2 and 5% CO2, to pH 7.2 to 7.4. Under a dissecting microscope, nerves were trimmed of excess connective tissue and kept in ice-cold oxygenated Locke solution until use. Nerves were mounted in a recording chamber such that the distal end was laid over 2 platinum stimulating electrodes isolated from the central portion of the nerve with a grease gap. A glass suction electrode connected to a differential preamplifier (0.1–10 kHz; model DAM-80; WPI, Sarasota, Florida) was used to isolate the central end of the nerve. The remainder of the nerve (~15–20 mm) was superfused continuously (2–5 mL/min) with oxygenated Locke solution with and without drugs delivered via a gravity-driven perfusion system. Unless otherwise indicated, CAP experiments were performed at room temperature. Compound action potentials were evoked with electrical pulses delivering constant voltage of 0.2 to 0.5 millisecond in duration and applied at 0.05 Hz, where the stimulus intensity was ~2 times that needed to evoke a maximal amplitude C-fiber component of the CAP. This stimulus intensity was used to ensure the stability of the CAP by eliminating the impact of small changes in action potential threshold on the CAP. In addition, using the same stimulus intensity to evoke both A and C waves enabled simultaneous monitoring of changes in both waveforms.7 Compound action potential data were filtered at 2 kHz (low-pass) and sampled at 20 kHz via a CED 1401 Micro A/D converter and acquired and analyzed using CED Spike 2 version 5 for MS Windows (CED, Cambridge, England). Compound action potential data were rectified, and the average of 6 consecutive CAPs was integrated to quantify A- and C-fiber components of the CAP as area under the curve (AUC). The A-fiber deflection of the CAP (A wave) was easily distinguished from that associated with the C-fiber deflection (C wave) because of the time delay between the arrivals of the 2 waves at the recording electrode. Weighted conduction velocities were calculated by the following formula: CV [m/s] = (nerve length [m]) Σ(amplitude [mV]) / Σ (AOT [mV-s]), where amplitude is the CAP signal at each sampling interval (of 0.06 millisecond) and AOT is amplitude over time, calculated such that AOT = (average amplitude at each sampling interval [mV]) (time(at interval) − time0 [s]).
Cell Culture Isolated sensory neurons were used for assays of either neurotoxicity based on trypan blue exclusion or intracellular Ca2+ experiments using fura-2–based microfluorimetry. Dorsal root ganglia (DRG, L3-L5, bilateral) were harvested from deeply anesthetized rats, enzymatically treated, mechanically dissociated, and plated on poly-D-lysine–coated coverslips as previously described.8 For the neurotoxicity experiments, cells were cultured overnight at 37°C/3% CO2 in a minimal essential media (MEM)–based solution containing ~90% MEM, 10% fetal bovine serum (FBS), 1 MEM-vitamins, 1 penicillin-streptomycin, and 35 ng/mL nerve growth factor (MEM-FBS). For Ca2+-imaging experiments, neurons were cultured at 37°C/3% CO2 in MEM-FBS for 2 hours prior to transfer to an L-15–based media (90% L-15, 10% FBS, 10 mM glucose, and 5 mM HEPES buffered to 7.4 pH), at room temperature prior to study.
Neurotoxicity The neurotoxicity experiments were conducted as described previously.8 Briefly, a 24-hour drug exposure protocol was used. In these experiments, ~24 hours after plating, coverslips were washed and flooded with MEM-FBS–containing test agents. In the first series of experiments, coverslips were treated with test agents that consisted of choline (isotonic/osmotic control), midazolam (300 μM), flumazenil (30 μM), PK11195 (3 μM), midazolam + flumazenil,
526
or midazolam + PK11195 and returned to the CO2 incubator for 24 hours. In a second series of experiments, the Ca2+ chelator, BAPTA-AM, was used as the test agent alone or in combination with midazolam. While 10 μM BAPTA-AM is routinely used in physiological experiments in vitro to block all but the most robust increases in intracellular Ca2+, preliminary results indicated that 10 and even 3 μM BAPTA-AM alone increased DRG neuron cell death after 24 hours in culture (not shown). Consequently, BAPTAAM was used at 1.0 and 0.3 μM alone or in combination with midazolam (300 μM). After incubating coverslips with test agents, neurons were washed twice with fresh MEM-FBS and then incubated with trypan blue (0.1%) for 10 minutes in a HEPES-buffered bath solution (HBS: 130 mM NaCl, 3 mM KCl, 2.5 mM CaCl2, 0.6 mM MgCl2, 10 mM HEPES, 10 mM glucose; 325 mOsm, pH 7.4). Trypan blue was washed twice with fresh bath solution, and neurons were fixed for 30 minutes in ice-cold 4% paraformaldehyde made in 1 phosphate-buffered saline. Coverslips were then transferred to fresh phosphate-buffered saline and stored at 4°C until analysis. Neurotoxicity was quantified as described previously8 by counting all trypan blue–positive and –negative neurons in a single pass through each coverslip at 40x magnification on an inverted microscope (Olympus CK2 Phase, Olympus Optical Co., Tokyo, Japan). Fractional cell death was determined by dividing the number of trypan blue stained neurons by the total number of neurons counted. A single rat was used for each preparation of neurons and each test agent was tested on 2 coverslips from each rat. The average fractional cell death from these 2 coverslips was used as the value for each animal, so that the “n” for each test agent was the number of animals studied.
Ca2+ Imaging Imaging experiments were conducted as described previously.9 Briefly, DRG neurons were loaded with 2.5 μM fura-2 AM ester (TEF Labs, Austin, Texas) with 0.025% pluronic (TEF Labs) for 20 minutes and washed in HBS for 20 minutes at room temperature. Coverslips were then placed in a recording chamber where they were continuously superfused with HBS. In some experiments, coverslips were superfused with HBS followed by Ca2+-free HBS (in mM: 130 NaCl, 3 KCl, 5 MgCl2, 2 EGTA, 10 HEPES, 10 glucose, pH 7.4, osmolality 325 mOsm). Fluorescence data were acquired at room temperature on a PC running Metafluor Software (Molecular Devices, Sunnyvale, California) via a charge-coupled device camera (model RTE/CCD 1300; Roper Scientific, Trenton, New Jersey). The ratio (R) of fluorescence emission at 510 nm in response to 340/380-nm excitation that was controlled by a lambda 10-2 filter changer (Sutter Instrument Co, Novato, California) was acquired at 0.2 Hz during baseline recordings and interstimulus periods and 1 Hz during the application of test compounds. Data for each neuron were analyzed as a change in R in response to a test compound relative to baseline, where an increase of greater than 20% above baseline was considered a response. Because the heterogeneity between neurons was larger than the heterogeneity between preparations, the “n” for the imaging data was neurons rather than rats, although each test agent was tested on neurons from at least 3 rats. The concentration-dependent increase in intracellular Ca2+ in response to midazolam was fitted with a modified Hill equation: Emax * Dn / (Dn + ECn50), where Emax is the maximal increase in fluorescence ratio, D is the concentration of midazolam, EC50 is the concentration of midazolam producing a response 50% of maximal, and n is the Hill coefficient. The EC50 concentration © 2014 American Society of Regional Anesthesia and Pain Medicine
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
of midazolam was used for subsequent experiments with additional test agents.
Test Agents Midazolam was purchased as a preservative-free injectable solution (5 mg/mL; Henry-Schein, Melville, New York) and diluted in culture media or HBS immediately prior to use. Diazepam, another benzodiazepine receptor agonist, was purchased in powder form (Sigma-Aldrich, St Louis, Missouri), dissolved in dimethyl sulfoxide (DMSO) as 1000 stock solution and stored at −20°C until the day of use. PK11195 (Tocris, Minneapolis, Minnesota) and flumazenil (Tocris) were dissolved in DMSO as 100 stock solutions and stored at −20°C until the day of use. BAPTAAM (Life Technologies, Carlsbad, California) was diluted in culture media and HBS as a 100 stock and stored at −20°C until the day of use. The concentration of midazolam used in this study (300 μM) was based on our previous estimate,8 which on a log scale, is close to the high end of the concentration achieved clinically (~460 μM).5 Furthermore, because we sought to determine whether the nerve-blocking or neurotoxic effects of midazolam were due to actions at either the benzodiazepine receptor or TSPO, we used the lowest clinically relevant concentration to produce a significant increase in both nerve block and neurotoxicity.7,8 The concentrations of PK11195 and flumazenil used were based on literature.10,11 The final concentrations of BAPTA-AM (30,
Mechanisms of Midazolam
10, 3, 1, and 0.3 μM) used were also based on the literature and results of initial neurotoxicity and Ca2+-imaging studies on isolated DRG neurons.
Statistical Analysis Data are presented as mean ± SEM. The fractional block of the CAP associated with midazolam with and without antagonists, neurotoxicity, and Ca2+-imaging data were analyzed with a 1-way analysis of variance with the Holm-Sidak test used for post hoc analysis. P < 0.05 was considered statistically significant.
RESULTS Midazolam-Induced CAP Block Midazolam binds to both the benzodiazepine site on the GABAA receptor and TSPO with comparable affinity.12 We therefore sought to determine whether either of these binding sites contributed to midazolam-induced peripheral nerve block. The benzodiazepine receptor antagonist flumazenil (30 μM) and the TSPO ligand PK11195 (3 μM) were used for this purpose. Consistent with the results of our previous study,7 midazolam (300 μM) produced a small but significant reduction in the AUC of the C wave, but not the A wave of the CAP (Fig. 1), and while PK11195 (n = 5) alone had no significant influence on either A or C wave of the
FIGURE 1. Effect of flumazenil and PK11195 on midazolam-induced block of the CAP in isolated rat sciatic nerve. A, Voltage traces from a representative recording illustrating changes in rectified A-wave (left) and C-wave (right) of the CAP evoked before and after application of midazolam (300 μM). Inset: Raw voltage trace illustrating the timing of the stimulus artifact (S) relative to the A-wave and the C-wave of the CAP. Note changes in scale for both amplitude and duration. Pooled data analyzed as a percent change from baseline for the AUC of the A-wave (B) and C-wave (C) in the presence of vehicle (V, n = 5), midazolam (M, n = 6), flumazenil (F, n = 5), PK11195 (P, n = 5), midazolam + flumazenil (M + F, n = 5), midazolam + PK11195 (M + P, n = 9). Insets are pooled data from nerve studied at 32°C to 34°C, M (n = 6), P (n = 6), M + P (n = 7). *P < 0.05, **P < 0.01. © 2014 American Society of Regional Anesthesia and Pain Medicine
527
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
18.68 ± 0.79 17.79 ± 0.72 0.319 ± 0.012 0.291 ± 0.010 18.06 ± 0.43 18.62 ± 0.44 0.261 ± 0.021 0.255 ± 0.019 14.27 ± 0.31 13.88 ± 0.34 0.294 ± 0.004 0.294 ± 0.010 14.18 ± 0.25 13.45 ± 0.39 0.269 ± 0.018 0.271 ± 0.017 A wave (before) A wave (after) C wave (before) C wave (after)
14.37 ± 0.25 14.74 ± 0.40 0.248 ± 0.007 0.261 ± 0.007
14.12 ± 0.66 13.58 ± 0.71 0.293 ± 0.032 0.280 ± 0.032
14.36 ± 0.58 13.93 ± 0.78 0.330 ± 0.009 0.312 ± 0.010
14.24 ± 0.56 14.49 ± 0.47 0.290 ± 0.006 0.309 ± 0.015
14.14 ± 0.22 13.85 ± 0.26 0.280 ± 0.012 0.276 ± 0.011
18.26 ± 0.67 16.71 ± 0.76 0.325 ± 0.018 0.271 ± 0.011*
M + P (n = 7) P (n = 6) M (n = 6) B (n = 4) P + M (n = 9) F + M (n = 5) P (n = 5) F (n = 5) M (n = 6) V (n = 5)
TABLE 1. Impact of Test Agents on Nerve Conduction Velocity
528
Conduction velocity (m/s) as calculated from the time-weighted average of the rectified A- and C-waves of the CAP. Data are average ± SEM values from nerves before and after application of test compounds, where the number of nerves in each group is indicated in parenthesis in the column heading. V is vehicle, M is midazolam (300 μM), F is flumazenil (30 μM), P is PK11195 (3 μM), F + M is flumazenil + midazolam, P + M is PK1119 + midazolam, and B is BAPTA-AM (30 μM). As nerves were pretreated with BAPTA-AM, “before” and “after” are before and after application of midazolam. The last three columns are from nerves studied at 32°C to 34°C. The only statistically significant impact of a test compound on conduction velocity as assessed with a paired t test was the decrease in C-wave conduction velocity following midazolam application in nerves studied at 32°C to 34°C (P < 0.01).
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
Yilmaz et al
CAP, flumazenil (n = 5) produced a significant increase in the C wave of the CAP. Furthermore, while coapplication of midazolam and flumazenil (n = 5) resulted in a change in the AUC of the CAP C-wave, no different than the algebraic sum of the change in the C wave in the presence of midazolam and flumazenil alone, coapplication of midazolam and PK11195 (n = 9) resulted in a significant attenuation of the midazolam-induced reduction in the AUC of the CAP C wave (Fig. 1). None of the test agents significantly affected conduction velocity (Table 1). To confirm the potential involvement of TSPO under more physiological conditions, we repeated the PK11195 experiments at 32°C to 34°C, temperatures closer to those observed in a distal appendage in vivo (insets to Figs. 1B, C). At this higher temperature, the magnitude of the midazolam-induced block of the Cwave CAP (n = 6), assessed as a percent change in AUC from baseline, was even greater than that observed at room temperature. PK11195 alone had no significant influence on the AUC of the CAP (n = 5). Finally, consistent with results obtained at room temperature, the midazolam-induced reduction in the AUC of the CAP C-wave was significantly attenuated by PK11195 (n = 8). However, in contrast to the negative results obtained with midazolam on conduction velocity at room temperature, midazolam produced a small but significant decrease in C-wave conduction velocity, which was only partially attenuated with PK11195 (Table 1).
Midazolam-Induced Neurotoxicity Neither PK11195 (3 μM) nor flumazenil (30 μM) alone affected neuron survival as neuron death rates were comparable to vehicle-treated controls (Fig. 2). As we previously demonstrated,8 midazolam (300 μM) was toxic to sensory neurons, with ~40% of all neurons stained with trypan blue after a 24-hour exposure (n = 6, Fig. 2). This increase in cell death was not significantly altered by the presence of either flumazenil (n = 6) or PK11195 (n = 6, Fig. 2).
Midazolam-Induced Increase in Intracellular Ca2+ Previous studies of smooth muscle cells indicate that midazolam effectively attenuates evoked Ca2+ transients in these cells via several mechanisms.13,14 In astrocytes, however, there is evidence that the opposite is true.15 Furthermore, because an increase in intracellular Ca2+ is often responsible for the neurotoxicity associated with receptor activation, we assessed the impact of midazolam on intracellular Ca2+ in DRG neurons. In contrast to smooth muscle cells, application of midazolam alone was sufficient to drive a rapid increase in intracellular Ca2+ in DRG neurons (Figs. 3A, B). This increase was concentration-dependent, with a threshold between 30 and 100 μM and an EC50 of ~200 μM (n = 33 neurons from 3 different rats, Fig. 3B). All neurons that responded to midazolam with an increase in intracellular Ca2+ also responded to application of 30 mM K+ as evidence of healthy neurons with a resting membrane potential sufficient to enable a depolarization-induced influx of Ca2+ via voltage-gated Ca2+ channels.16 To determine whether the impact of midazolam on the concentration of intracellular Ca2+ in DRG neurons was a drug class effect, we assessed the impact of another benzodiazepine receptor agonist, diazepam, on intracellular Ca2+ levels. Diazepam also led to increases in intracellular Ca2+ in DRG neurons (Fig. 3A). However, the potency of diazepam was significantly less than that of midazolam, with only ~50% of neurons responsive to 1 mM diazepam (Fig. 3C). Furthermore, the diazepam-induced Ca2+ transient was desensitizing such that after a neuron responded to © 2014 American Society of Regional Anesthesia and Pain Medicine
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
Mechanisms of Midazolam
FIGURE 2. Flumazenil and PK11195 do not influence midazolam-induced neurotoxicity. Neurotoxicity was assessed with trypan blue staining. Pooled data are from 6 animals, where V is vehicle containing the combination of choline (to control for osmolality) and DMSO (0.1%) used for the preparation of test agents; M is midazolam (300 μM); F is flumazenil (30 μM); P is PK11195 (3 μM); M + F, midazolam + flumazenil; M + P, midazolam + PK11195. n = 6, **P < 0.01.
diazepam at 1 concentration, subsequent diazepam applications of equal to or greater concentration were unable to induce another Ca2+ transient (Fig. 3A). To rule out a role for either the central benzodiazepine receptor or TSPO in the midazolam-induced increase in intracellular Ca2+, we studied 2 more groups of neurons in which flumazenil (30 and 300 μM) or PK11195 (3 and 30 μM) was applied with an EC50 concentration of midazolam (200 μM). Neither flumazenil (n = 62) nor PK11195 (n = 72) was associated with an increase in intracellular Ca2+ on its own, and neither flumazenil (n = 53) nor PK11195 (n = 52) had any detectable influence on the amplitude of the midazolam-induced increase in intracellular Ca2+ (Fig. 3D) or number of neurons that responded to midazolam with an increase in intracellular Ca2+ (not shown). Finally, to determine whether the midazolam-induced increase in intracellular Ca2+ was due to influx or release from intracellular stores, a group of neurons was challenged with 200 μM midazolam in the presence and absence of Ca2+ in the bath solution. The results of this experiment revealed no detectable difference in the midazolaminduced Ca2+ transient evoked in the presence or absence of extracellular Ca2+ (Figs. 3E, F). While a midazolam-induced increase in intracellular Ca2+ suggests a likely mechanism for the midazolam-induced neurotoxicity, we next tested this suggestion directly. However, because the midazolam-induced Ca2+ transient was not due to the activation of an influx pathway, it was necessary to use an approach that would enable us to prevent the increase in intracellular Ca2+ associated with the release of Ca2+ from internal stores. For this, we used the intracellular Ca2+ chelator BAPTA-AM. In an initial set of Ca2+-imaging experiments, we were able to confirm that preincubating DRG neurons with 30 μM BAPTA-AM for 30 minutes resulted in a complete block of the midazolam-induced Ca2+ transient (n = 73, Fig. 4A). The midazolam-induced Ca2+ transient was blocked by greater than 80%, 70%, and 60% in neurons preincubated with 10, 3, and 1 μM BAPTA-AM, respectively (Fig. 4A). © 2014 American Society of Regional Anesthesia and Pain Medicine
For the neurotoxicity experiments, DRG neurons were incubated either in the presence or absence of midazolam, with or without BAPTA-AM. Normal resting levels of Ca2+ are essential for a variety of cellular processes. Therefore, we examined a range of concentrations of BAPTA-AM in the absence of midazolam to determine an appropriate concentration(s) of BAPTAAM for these experiments. We found a significant increase in trypan blue staining in DRG neurons (cultured for 24 hours) following a 30-minute incubation in either 30 μM (66.8 ± 8.5%, n = 4) or 10 μM (27.15 ± 2.86%, n = 3) BAPTA-AM, but not after 1 or 0.3 μM BAPTA-AM (Fig. 4B). Whereas 1 and 0.3 μM BAPTA-AM were without effect on trypan blue staining relative to vehicle-treated controls, 1 μM BAPTA-AM produced a small but significant decrease in midazolam-induced neurotoxicity (Fig. 4B). Finally, because it is also possible that an increase in intracellular Ca2+ contributes to the midazolam-induced nerve block, we recorded CAPs in the presence or absence of midazolam with or without a 30-minute preincubation with BAPTA-AM (30 μM). BAPTA-AM had no significant influence on the CAP AUC (10.06 ± 0.81 mV-s with BAPTA-AM, 11.10 ± 0.78 mV-s without BAPTA-AM for A-CAP, and 21.55 ± 5.21 μV-s with BAPTA-AM, 26.83 ± 3.22 μV-s without BAPTA-AM for C-CAP), conduction velocity (14.27 ± 0.31 m/s with BAPTA-AM, 14.37 ± 0.25 m/s without BAPTA-AM for A-CAP, 0.294 ± 0.004 m/s with BAPTA-AM, 0.248 ± 0.007 m/s without BAPTA-AM for C-CAP), or amplitude (7.8 ± 1.25 mV with BAPTA-AM, 9.3 ± 1.05 mV without BAPTA-AM for A-CAP, 111.25 ± 13.9 μV with BAPTAAM, 123.75 ± 28.16 μV without BAPTA-AM for C-CAP). However, midazolam-induced block of the C-CAP was attenuated significantly (P < 0.01) in the presence of BAPTA (n = 4, Fig. 4C).
DISCUSSION The purpose of this study was to determine whether midazolaminduced nerve block and/or neurotoxicity involves the activation
529
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Yilmaz et al
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
of the central benzodiazepine receptor or TSPO. Our results indicate that while midazolam-induced nerve block was significantly attenuated by the TSPO ligand PK11195, midazolam neurotoxicity was not. The midazolam-induced neurotoxicity was significantly attenuated by a concentration of BAPTA-AM that produced no detectable increase in toxicity alone (1 μM), but that significantly attenuated the midazolam-induced increase in intracellular Ca2+. The midazolam-induced increase in intracellular Ca2+ was not attenuated by either flumazenil or PK11195 and also was not attenuated in the absence of extracellular Ca2+. Finally, BAPTA-AM produced a significant attenuation of midazolaminduced block of the CAP C-wave. These results suggest that different mechanisms underlie midazolam-induced nerve block and neurotoxicity; the nerve block is dependent on the activation of TSPO and an increase in intracellular Ca2+, whereas neurotoxicity is dependent, at least in part, on an increase in intracellular Ca2+.
Our results with PK11195 suggest that midazolam-induced block of the C-wave is due, at least in part, to an action at TSPO; TSPO has been implicated in numerous mitochondrial processes including the induction of apoptosis.17–19 We suggest that midazolam is unlikely to engage TSPO-dependent apoptotic pathways in sensory neurons, given the absence of a detectable influence of PK11195 on the midazolam-induced increase in trypan blue staining. However, it is possible that binding of midazolam to TSPO leads to a reduction in mitochondrial respiration, hence reduced ATP production. This, in turn, could lead to inhibition of proteins essential for the maintenance of membrane potential after action potential firing (eg, sodium/potassium ATPase) and consequently a depolarization-induced block of action potential propagation. Alternatively, an increase in intracellular Ca2+ downstream of TSPO activation could account for the differential block of C- over A-fibers, if the Ca2+-dependent processes responsible for conduction block
FIGURE 3. Midazolam induces increases in intracellular Ca2+ in DRG neurons. A, Representative Ca2+ transients as assessed by changes in the fura-2 fluorescence ratio in response to increasing concentrations of midazolam (top traces) or diazepam (bottom traces). B, Pooled concentration-response data (n = 33 neurons from 3 rats) for midazolam were fitted with a Hill equation to determine the concentration associated with an peak increase in Ca2+ that was 50% of the maximal increase (EC50), as well as the maximal increase (Emax). C, Because the responses to diazepam (n = 42 to 68 cells per diazepam concentration) were desensitizing, we estimated differences in the potency of midazolam and diazepam with cumulative response plots, where the percentage of neurons responding to each concentration of test agent is plotted. There was a concentration of both compounds that produced a response in ~100% of the neurons tested. Consequently, both cumulative distributions were also well fitted with a Hill equation used to estimate the concentration at which 50% of the neurons tested were responsive. D, Pooled data from neurons treated with an EC50 concentration of midazolam alone (200 μM, n = 72), flumazenil (F, 30 μM, n = 53), and PK11195 (P, 3 μM, n = 52), followed by the combination of M + F (n = 53) and M + P (n = 52) applied to the same F and P neurons, respectively, indicate that neither antagonist had a detectable influence on the midazolam-induced Ca2+ transient. E, An example of midazolam-induced Ca2+ transients evoked from a neuron in the absence (− Ca2+) and presence (+Ca2+) of Ca2+ in the bath solution. F, Pooled data (n = 18) indicate that there was no detectable influence of extracellular Ca2+ on the midazolam-induced Ca2+ transient.
530
© 2014 American Society of Regional Anesthesia and Pain Medicine
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
Mechanisms of Midazolam
FIGURE 4. Effect of intracellular Ca2+ chelation on the actions of midazolam on sensory neurons. The Ca2+ chelator BAPTA, applied as the membrane-permeable AM ester, was used to block midazolam (300 μM)–induced increase in intracellular Ca2+. A, Pooled data from isolated sensory neurons challenged with 300 μM midazolam indicate that block of the transient was concentration dependent, with the transient completely blocked in close to 100% of the neurons tested at 30 μM BAPTA (n = 28 to 73 neurons per BAPTA concentration). B, Pooled neurotoxicity data from neurons treated with BAPTA alone or in combination with midazolam (n = 4 per group). V is vehicle; .3B is 0.3 μM BAPTA; 1B is 1 μM BAPTA; M is midazolam (300 μM). C, Pooled C-wave CAP data from nerves treated with midazolam (M, 300 μM) alone (n = 6) and a separate group of nerves (n = 4) treated with 30 μM BAPTA prior to the application of midazolam (300 μM, n = 4). The midazolam-alone data group are from Figure 1 and are replotted here for comparison. *P < 0.05, **P < 0.01.
were differentially distributed between C- and A-fibers. It is also possible that processes associated with TSPO activation and the increase in intracellular Ca2+ occur in parallel. Thus, additional experiments will be needed to define the mechanisms mediating the link between midazolam, TSPO, and nerve block. The results of the present study suggest midazolam-induced nerve block is relatively selective for C-fibers. However, a ~20% block of the A-fiber CAP was observed at 32°C to 34°C in the present study, and we previously reported a midazolam-induced A-fiber block roughly half that of the C-fiber block.7 Careful analysis of both sets of data revealed what appear to be 2 groups of nerves: one in which 300 μM midazolam blocks between 10% and 20% of the A-wave and another in which little if any block is detected. In our previous study, 5 of the 6 nerves studied had a block between 10% and 20%, whereas in the present study, only 3 of the 6 nerves studied had an A-wave block between 10% and 20%, the block in 1 nerve was less than 5%, and in 2 nerves there was a slight increase in amplitude with time that was not corrected. If 300 μM midazolam is close to the threshold concentration necessary to activate processes responsible for the A-fiber block, it could account for the variability observed between nerves. The observation that flumazenil selectively increased the amplitude of the C-CAP was unexpected. While off-target effects of even the most pharmacologically selective compounds have been described, this observation raises the intriguing possibility that action potential propagation in peripheral C-fibers is regulated by GABAA receptor activation. The presence of functional GABAA receptors in peripheral nerves is suggested by the previous observations that peripheral administration of GABAA receptor agonists is antinociceptive.20,21 However, while resident immune cells would be one of the only endogenous sources of GABA22 needed for such a mechanism to explain the impact of flumazenil, the density of immune cells in peripheral nerves in the absence of injury is low. Furthermore, flumazenil should have blocked not only endogenous GABA signaling, but also the actions of midazolam, resulting in a change in the amplitude of the C-CAP comparable to that observed with flumazenil alone. Thus, we suggest that an off-target effect is more likely. Regardless, the negative results with flumazenil should still be interpreted with caution, given evidence of benzodiazepine binding to GABAA receptors through flumazenil-independent mechanisms (ie, independent of the γ2 subunit), as well as the possibility that the concentration used was insufficient to completely block the relatively high concentration of midazolam used, at the benzodiazepine site. © 2014 American Society of Regional Anesthesia and Pain Medicine
The midazolam-induced increase in intracellular Ca2+ via release of Ca2+ from internal stores not only provides a mechanism for midazolam-induced neurotoxicity, but also suggests a novel site of action for this compound. The observation that midazolaminduced Ca2+ transients were present in all neurons tested suggests that the mechanism(s) responsible for the Ca2+ transient are ubiquitously expressed among DRG neurons. Our previous results indicate that the relative contribution of various Ca2+ regulatory elements to the magnitude and duration of the depolarizationinduced Ca2+ transient varies among subpopulations of DRG neurons.16 Nevertheless, there are 2 internal sources of Ca2+ that are ubiquitous among sensory neurons. One source is mitochondria. Interestingly, a Ca2+ regulatory protein in mitochondria, the Na+/ Ca2+/Li+-exchanger (NCLX), is blocked by benzodiazepines.23 However, we consider block of NCLX an unlikely mechanism for the midazolam-induced increase in intracellular Ca2+ because midazolam NCLX block would be expected to decrease the release of Ca2+ from mitochondria. Furthermore, while a nonspecific action on mitochondria leading to a breakdown of the proton gradient will result in an increase in intracellular Ca2+ in all DRG neurons, our previous results16 suggest that the transient associated with the breakdown of the mitochondria proton gradient is considerably smaller than that associated with midazolam. A second ubiquitous source of Ca2+ in sensory neurons is the endoplasmic reticulum,16 which is capable of sustaining robust and reproducible Ca2+ transients in sensory neurons comparable to midazolam-evoked transients, even in the presence of Ca2+-free bath solution.9 Thus, while additional experiments are needed to confirm the source of the midazolam-induced Ca2+ transient, the endoplasmic reticulum is the most likely source. Moreover, our observation that diazepam can also drive an increase in intracellular Ca2+ suggests that this might be a drug class effect not previously described. While attenuation of the midazolam-induced neurotoxicity by 1 μM BAPTA was significant, this attenuation was far from a complete block. This observation raises the possibility that midazolam activated a Ca2+-independent apoptotic pathway, such as phospholipase A2–mediated apoptosis.24 Nevertheless, consistent with the observation that 1 μM BAPTA was insufficient to completely block the midazolam-induced Ca2+ transient, it appears more likely that the residual toxicity was due to BAPTA-insensitive increase in intracellular Ca2+. To directly test this possibility, however, it will likely be necessary to identify the mechanism(s) underlying the increase in intracellular Ca2+ so that this mechanism
531
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Yilmaz et al
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
can be specifically manipulated in a way that does not produce toxic changes in resting intracellular Ca2+ concentrations. Identification of this mechanism(s) may also enable ways to further maximize the therapeutic actions of midazolam while minimizing toxic adverse effects. While TSPO appears to be the primary mechanism underlying the nerve-blocking actions of midazolam, our results with BAPTA-AM suggest that an increase in intracellular Ca2+ may also contribute to this effect. This begs the question as to the basis of the Ca2+-induced attenuation of action potential propagation in C-fibers. One possibility is a depolarization-induced inactivation of voltage-gated Na+ channels necessary for the propagation of action potentials secondary to the increase in intracellular Ca2+. Although there is evidence for functional Ca2+-dependent K+ channels in peripheral axons, we believe that an increase in K+ channel activity is unlikely to account for the actions of midazolam. That is, an increase in K+ channel activity will result in membrane hyperpolarization, which should relieve Na+ channels from an inactivated state and thereby increase AP conduction velocity, if not also CAP amplitude. Consistent with this expectation, our preliminary results indicate that the Ca2+-dependent K+ channel opener NS1619 (100 μM) results in an increase in the amplitude of both A (39% ± 12%) and C (47% ± 7.4%) waves of the CAP (n = 3). Mechanisms that could account for a Ca2+-induced depolarization of the sciatic nerve include the activation of the electrogenic plasma membrane Na+/Ca2+ exchanger and/or the activation of Ca2+-dependent Cl− channels. However, the limited impact of midazolam on conduction velocity argues against a depolarizationinduced inactivation of Na+ channels, given evidence that local anesthetic–induced CAP block is preceded by conduction velocity slowing.25 Additional experiments will be needed to confirm mechanism(s) linking increased intracellular Ca2+ to CAP block. While there are several interesting and potentially important implications of the results of this study, there are also important limitations. First, data were generated in rat tissue. Therefore, species differences may influence results and conclusions. For example, midazolam-induced toxicity and drug kinetics may differ between rats and humans; thus, relatively limited neurotoxicity observed in patients may be due to a species difference and/or a concentration difference, where toxic doses are rarely, if ever, achieved clinically. Second, in our CAP recording experiments, we used bath application of drugs to the entire sciatic nerve. Given the limited length of nerve directly impacted by clinical manipulations and the extent of the safety factor in the intact nerve, the relative drug concentrations we used in these experiments may be considerably higher than those obtained clinically. In addition, diffusion will also significantly reduce the magnitude of effects observed clinically. Third, we have assessed neurotoxicity in isolated sensory neurons. These neurons might be more vulnerable than neurons in an intact system because they are devoid of sources of protection found in vivo including support cells and vasculature. And fourth, while we have interpreted both neurotoxicity and nerve block data in the context of results obtained with Ca2+ imaging, the imaging data were collected at room temperature, and the toxicity data were collected at 37°C. As there is evidence to suggest that the Ca2+ release from intracellular stores is temperature-dependent, where, for example, low-amplitude Ca2+ oscillations at room temperature become spikes at 37°C in T lymphocytes, we may have underestimated the impact of midazolam on Ca2+ release in DRG neurons. Given that ~184 μM midazolam is a threshold concentration for neurotoxicity,8 and ~100 μM is the threshold concentration for an increase in intracellular Ca2+ at room temperature, such an underestimate would add further support to the suggestion that Ca2+-independent mechanisms contribute to midazolam-induced neurotoxicity.
532
The present results are important for multiple reasons. First, the evidence that a C-fiber–selective midazolam-induced nerve block is mediated by TSPO suggests a new direction for development of approaches that could provide a modality-selective nerve block. Second, our study provides evidence that different mechanisms mediate midazolam-induced nerve block and neurotoxicity. This finding may suggest approaches to both minimize neurotoxicity and maximize therapeutic utility. And third, with respect to clinical use of peripheral nerve blocks, our results continue to argue against the clinical perineural use of midazolam at concentrations even close to that that may be neurotoxic alone or in combination with local anesthetics that are also able to drive toxic increases in intracellular Ca2+.8
REFERENCES 1. Yamamoto T, Yaksh TL. Spinal pharmacology of thermal hyperesthesia induced by incomplete ligation of sciatic nerve. I. Opioid and nonopioid receptors. Anesthesiology. 1991;75:817–826. 2. Dirig DM, Yaksh TL. Intrathecal baclofen and muscimol, but not midazolam, are antinociceptive using the rat-formalin model. J Pharmacol Exp Ther. 1995;275:219–227. 3. Goodchild CS, Serrao JM. Intrathecal midazolam in the rat: evidence for spinally-mediated analgesia. Br J Anaesth. 1987;59:1563–1570. 4. Bharti N, Madan R, Mohanty PR, et al. Intrathecal midazolam added to bupivacaine improves the duration and quality of spinal anaesthesia. Acta Anaesthesiol Scand. 2003;47:1101–1105. 5. Jarbo K, Batra YK, Panda NB. Brachial plexus block with midazolam and bupivacaine improves analgesia. Can J Anaesth. 2005;52:822–826. 6. Williams BA, Murinson BB, Grable BR, et al. Future considerations for pharmacologic adjuvants in single-injection peripheral nerve blocks for patients with diabetes mellitus. Reg Anesth Pain Med. 2009; 34:445–457. 7. Yilmaz-Rastoder E, Gold MS, Hough KA, et al. Effect of adjuvant drugs on the action of local anesthetics in isolated rat sciatic nerves. Reg Anesth Pain Med. 2012;37:403–409. 8. Williams BA, Hough KA, Tsui BY, et al. Neurotoxicity of adjuvants used in perineural anesthesia and analgesia in comparison with ropivacaine. Reg Anesth Pain Med. 2011;36:225–230. 9. Scheff NN, Lu SG, Gold MS. Contribution of endoplasmic reticulum Ca2+ regulatory mechanisms to the inflammation-induced increase in the evoked Ca2+ transient in rat cutaneous dorsal root ganglion neurons. Cell Calcium. 2013;54:46–56. 10. Awad M, Gavish M. Binding of [3H]Ro 5-4864 and [3H]PK 11195 to cerebral cortex and peripheral tissues of various species: species differences and heterogeneity in peripheral benzodiazepine binding sites. J Neurochem. 1987;49:1407–1414. 11. Weiss M, Tikhonov D, Buldakova S. Effect of flumazenil on GABAA receptors in isolated rat hippocampal neurons. Neurochem Res. 2002;27:1605–1612. 12. Bender AS, Hertz L. Pharmacological characteristics of diazepam receptors in neurons and astrocytes in primary cultures. J Neurosci Res. 1987;18:366–372. 13. Yoshimura H, Kai T, Nishimura J, et al. Effects of midazolam on intracellular Ca2+ and tension in airway smooth muscles. Anesthesiology. 1995;83:1009–1020. 14. Yang M, Ding X, Murray PA. Differential effects of intravenous anesthetics on capacitative calcium entry in human pulmonary artery smooth muscle cells. Am J Physiol Lung Cell Mol Physiol. 2008;294: L1007–L1012. 15. Code WE, Hertz L, White HS. Midazolam potently affects free calcium concentration in astrocytes. Can J Anaesth. 1990;37:S27.
© 2014 American Society of Regional Anesthesia and Pain Medicine
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
Regional Anesthesia and Pain Medicine • Volume 39, Number 6, November-December 2014
16. Lu SG, Zhang X, Gold MS. Intracellular calcium regulation among subpopulations of rat dorsal root ganglion neurons. J Physiol. 2006;577:169–190. 17. Veenman L, Gavish M. The role of 18 kDa mitochondrial translocator protein (TSPO) in programmed cell death, and effects of steroids on TSPO expression. Curr Mol Med. 2012;12:398–412. 18. Rupprecht R, Papadopoulos V, Rammes G, et al. Translocator protein (18 kDa) (TSPO) as a therapeutic target for neurological and psychiatric disorders. Nat Rev Drug Discov. 2010;9:971–988.
Mechanisms of Midazolam
21. Reis G, Pacheco D, Francischi J, et al. Involvement of GABA A receptor–associated chloride channels in the peripheral antinociceptive effect induced by GABA A receptor agonist muscimol. Eur J Pharmacol. 2007;564:112–115. 22. Bhat R, Axtell R, Mitra A, et al. Inhibitory role for GABA in autoimmune inflammation. Proc Natl Acad Sci U S A. 2010;107:2580–2585. 23. Palty R, Sekler I. The mitochondrial Na(+)/Ca(2+) exchanger. Cell Calcium. 2012;52:9–15.
19. Gatliff J, Campanella M. The 18 kDa translocator protein (TSPO): a new perspective in mitochondrial biology. Curr Mol Med. 2012;12:356–368.
24. Sun GY, Shelat PB, Jensen MB, et al. Phospholipases A2 and inflammatory responses in the central nervous system. Neuromolecular Med. 2010;12: 133–148.
20. Carlton SM, Zhou S, Coggeshall RE. Peripheral GABA(A) receptors: evidence for peripheral primary afferent depolarization. Neuroscience. 1999;93:713–722.
25. Gokin AP, Philip B, Strichartz GR. Preferential block of small myelinated sensory and motor fibers by lidocaine: in vivo electrophysiology in the rat sciatic nerve. Anesthesiology. 2001;95:1441–1454.
© 2014 American Society of Regional Anesthesia and Pain Medicine
533
Copyright © 2014 American Society of Regional Anesthesia and Pain Medicine. Unauthorized reproduction of this article is prohibited.
