Benzodiazepines, especially midazolam, are analgesic after intrathecal injection in laboratory animals (1–3). Although not a widely accepted treatment, several human studies have claimed analgesic efficacy for midazolam in patients after epidural administration (4–6). Midazolam appears to be effective as an analgesic when given alone (5) or when combined with more traditional spinal analgesics (4–6). Spinally administered midazolam attenuates postsurgical pain (5,7,8) and chronic low back pain (9). The mechanisms of the spinal analgesic effects of benzodiazepines are not well understood. Some studies support the expected interaction of midazolam with classic benzodiazepine receptors with subsequent analgesia attributed to an indirect effect on spinal opioid transmission (1,10). Others have presented evidence for a more direct effect on spinal opioid receptors. For example, Rattan et al. (2,11) have shown that midazolam directly displaces opioid receptor binding in membranes from rat spinal cord. We noted those results with interest and decided to examine the direct effects of midazolam and three other benzodiazepines on radioligand binding to cloned human opioid receptor subtypes. In addition, we examined the effects of midazolam, diazepam, and chlordiazepoxide in an in vitro assay of opioid receptor activation, 35S-GTPγS binding. Our data support a direct interaction of these benzodiazepines as κ receptor agonists.
Chinese hamster ovary (CHO) cells expressing cloned human δ- and κ-opioid receptor subtypes were obtained from Glaxo Wellcome, Stevenage, UK. Receptor cDNA cloned into pCIN vectors were transfected into CHO cells by using a calcium phosphate method. High expressing cell lines were selected by using [3H]-diprenorphine binding. CHO cells containing human μ receptors were prepared at Affymax Research Institute, Palo Alto, CA. N-terminal, hemagglutinin-tagged receptors were expressed in CHO cells and selected by fluorescence-activated cell sorting by using antihemagglutinin antibody followed by phycoerythrin-conjugated goat antimouse immunoglobulin G. High expressing clones were obtained after several rounds of fluorescence-activated cell sorting.
Opioid receptor-transfected CHO cells were grown in 225-cm2 flasks in media consisting of Dulbecco’s modified eagle’s medium/F12 with fetal bovine serum 5%. Cells were scraped from the flasks and homogenized in HEPES 50 mM, MgCl2 10 mM, and EDTA 1 mM, pH 7.4, using a glass douncer. Crude membrane homogenates were centrifuged at 20,000 rpm for 30 min, rehomogenized, centrifuged again, resuspended in buffer, and frozen at −80°C. Protein determinations were made by using the BioRad method.
Opioid receptor binding assays were performed by using [3H]-diprenorphine 0.75 nM in 96-well microtiter plates. The assay buffer contained HEPES 50 mM, MgCl2 10 mM, EDTA 1 mM, and protease inhibitors, pH 7.4. Membrane homogenate protein concentrations were 2, 3, and 18 μg per well for μ, δ, and κ receptors, respectively, in a final volume of 150 μL. Protein amounts were selected from results of saturation binding studies (data not shown) to provide similar levels of receptor concentrations. Incubations were performed for 60 min at room temperature, terminated by rapid filtration through a 96-well GF/B filter, and washed with ice-cold buffer. After drying, scintillation fluid was added to each well, plates were sealed and then counted on a Packard Instruments TopCount (Packard Instruments, Meriden, CT). Test compounds were prepared in 100% dimethylsulfoxide and diluted in assay buffer immediately before use. Final concentrations of dimethylsulfoxide did not exceed 1%. Nonspecific binding was determined in the presence of 2.5 μM of GR 89696A, GW 357476B (BW 373 U86), and GI 103996A (a hydrolysis-resistant analog of remifentanil), for κ-, δ-, and μ-receptor subtypes, respectively.
35S-GTPγS binding assays were performed by using modifications of published assays (12–14). Membrane homogenates were prepared and stored as described above, thawed just before use, and maintained on ice. Membrane homogenates were diluted in assay buffer (2 μg/well for δ and 8 μg/well for κ and μ) and incubated in duplicate wells of 96-well microtiter plates with compounds and 35S-GTPγS (200 pM) for 60 min at 30°C in a final volume of 210 μL. For κ antagonist experiments, membranes were incubated in triplicate with nor-binaltorphimine (nor-BIN) for 30 min, then exposed to midazolam and 35S-GTPγS for an additional 30 min at 30°C. The reactions were terminated by rapid filtration through a 96-well GF/B filter plate and washed with ice-cold HEPES 20 mM. Protein concentrations were selected to give optimal responses to standard opioid compounds. The assay buffer was composed of HEPES 50 mM, MgCl2 5 mM, NaCl 120 mM, EDTA 1 mM, 1,4-dithiothreitol 0.5 mM, guanosine 5’-diphosphate 30 μM, and protease inhibitors, at pH 7.4. Nonspecific binding was determined in the presence of guanosine triphosphate (GTP) 100 μM.
Data were analyzed by using nonlinear curve-fitting functions in Prism version 2.01 (GraphPad Software, San Diego, CA). Responses at each concentration of compound were averaged, curves were fit, and appropriate response variables were derived.
35S-GTPγS and [3H]-diprenorphine were purchased from New England Nuclear (Boston, MA) and Amersham (Amersham Pharmacia Biotech (Piscataway, NJ), respectively. Midazolam, diazepam, chlordiazepoxide, nor-BIN, 1,4-dithiothreitol, guanosine diphosphate, and GTP were purchased from Sigma (St. Louis, MO). Flumazenil was a gift from Hoffman-LaRoche (Nutley, NJ). HEPES, fetal bovine serum, and Dulbecco’s modified eagle’s medium/F12 were purchased from GIBCO BRL (Rockville, MD). MgCl2, NaCl, and EDTA were purchased from JT Baker (Phillipsburg, NJ). [D-Ala2,MePhe4,Glyol5]enkephalin (DAMGO), [D-Ala2, D-Leu5]enkephalin, and naltrindole HCl were purchased from RBI (Natick, MA). GR 89696A, GW 357476B (BW 373 U86), and GI 103996A were synthesized at Glaxo Wellcome. Protease inhibitors were Pefabloc SC (Boehringer-Mannheim) 2.5 μg/mL, pepstatin A (Sigma) 0.1 μg/mL, leupeptin (Sigma) 0.1 μg/mL, and aprotinin (Sigma) 0.1 μg/mL.
[3H]-diprenorphine binding curves are shown in Figure 1 and the binding variables are summarized in Table 1. All three benzodiazepine agonists, midazolam, diazepam, and chlordiazepoxide, inhibited the binding of [3H]-diprenorphine to human κ-opioid receptors in a concentration-dependent manner. Diazepam inhibited δ-receptor binding with a 50% inhibitory concentration of 87 μM, whereas midazolam and chlordiazepoxide only partially inhibited at this receptor. None of the benzodiazepines significantly affected [3H]-diprenorphine binding to μ receptors. The benzodiazepine antagonist, flumazenil, had only minor effects at any of the three opioid receptor subtypes and only at 50% effective concentrations of >250 μM. For comparison, known opioid receptor ligands potently inhibited [3H]-diprenorphine binding at the various subtypes.
Similar to their effects in [3H]-diprenorphine binding assays, the three benzodiazepine agonists stimulated 35S-GTPγS binding in membrane homogenates from CHO cells expressing human κ receptors (Fig. 2). The benzodiazepines were weak, but fully effective in that they stimulated 35S-GTPγS binding to near the same extent as the known κ agonist GR 89696A (Table 1). In a separate experiment, flumazenil 100 μM had no effect on the stimulation of 35S-GTPγS binding by benzodiazepines or the opioid agonist GR 89696A (data not shown). In contrast, the selective κ receptor antagonist, nor-BIN, inhibited midazolam-induced binding of 35S-GTPγS (Fig. 3). Despite its very weak affinity, midazolam stimulated 35S-GTPγS binding to the human δ receptor by 227% at 200 μM. Diazepam and chlordiazepoxide showed no activity at δ receptors. None of the benzodiazepines stimulated 35S-GTPγS binding in membranes containing μ receptors. In contrast, known μ agonists, DAMGO, and GI 103996A were potent stimulators of μ receptor 35S-GTPγS binding.
Midazolam is analgesic in animal and human studies after spinal, but not systemic, injection (1–6,15). The mechanism by which midazolam exerts an analgesic effect after spinal injection is somewhat controversial. Midazolam exerts an indirect effect on pain transmission downstream from its expected effects on benzodiazepine-γ aminobutyric acid (GABA)A receptors (1,3,10). Goodchild and Serrao (1) showed that intrathecal midazolam increased the pain threshold in rats, and that this effect was blocked by prior injection of the benzodiazepine antagonist flumazenil. Likewise, Kohno et al. (10) recently demonstrated that midazolam augmented GABA-mediated responses in substantia gelatinosa neurons from rat spinal cord, an effect that would increase inhibitory neurotransmission. Rattan et al. (2) have shown that the in vivo antinociceptive effects of intrathecal midazolam can be reversed by the opioid antagonist naloxone, indicating the involvement of opioid receptors. Other studies, using selective opioid antagonists, support the downstream activation of spinal δ-opioid receptors in the analgesia produced by intrathecal midazolam (16). In the whole animal or patient, it is likely that the mechanism is complex, involving both direct and indirect effects.
Two studies using rat tissues have demonstrated that benzodiazepines can displace radiolabeled opioid ligands from binding sites in the spinal cord (2,11). Midazolam displaced tritiated subtype selective opioid radioligands from endogenous receptors in rat spinal cord with a rank order of potency of κ > δ > μ(2). This is the same rank order that we observed in the current study displacing [3H]-diprenorphine from cloned human opioid receptor subtypes.
Although informative about affinity for a receptor, traditional radioligand binding studies do not disclose the efficacy of a ligand. To determine the efficacy of benzodiazepines at cloned human opioid receptors, we used a functional binding assay, 35S-GTPγS binding. On binding of an agonist, opioid receptors couple to inhibitory G-proteins (Gi or Go) and bound guanosine diphosphate is exchanged for GTP. The efficacy of an agonist can be determined by measuring the binding of a stable GTP analog, 35S-GTPγS. The amount of 35S-GTPγS bound is directly proportional to the efficacy of the agonist (12–14). By using this assay, we found that midazolam, diazepam, and chlordiazepoxide each stimulated 35S-GTPγS binding to κ receptor-containing cell membranes. Although the compounds are weak agonists, they appear to be fully effective, in that they stimulated 35S-GTPγS binding to the same extent as the known κ agonist GR 89696A. The receptor specificity of midazolam’s effect on κ receptors is supported by the inhibition of its agonist effect by the nonspecific opioid antagonist naloxone, as well as by the selective κ antagonist nor-BIN. Furthermore, the absence of binding or functional activity in μ-expressing CHO cell membranes is evidence that the effect is selective for κ receptors and not a nonspecific effect (e.g., an interaction of midazolam with a G-protein). The affinity and activity of benzodiazepines at κ receptors are not entirely surprising. The novel κ agonist tifluadom is a substituted 1,4-benzodiazepine (17). Curiously, despite only partial displacement of [3H]-diprenorphine from δ receptors, midazolam stimulated 35S-GTPγS binding in membranes from cells expressing δ receptors.
Although benzodiazepines are only weak opioid agonists, large micromolar local concentrations are achieved after intrathecal injection. For example, Naguib et al. (5) administered caudal injections of midazolam to children at 50 μg/kg in 1 mL/kg of saline. Before dilution, this would expose local tissue to approximately 150 μM. In another study (8), intrathecal doses up to 2 mg midazolam in 3 mL were given to adults, an initial concentration of 2 mM.
Our results show that three common benzodiazepine agonists can bind to, and activate, human κ-opioid receptors, and one, midazolam, can also modestly activate δ receptors. These data suggest that part of the analgesia seen with intrathecal benzodiazepines may be attributed, in part, to direct stimulation of opioid receptors.
The authors extend their thanks to Nicola Bevan, Glaxo Wellcome, Stevenage, UK, for cell lines expressing human δ and κ receptor subtypes. We also thank Emily Tate and Andrea Dobbs, Affymax Research Institute, Palo Alto, CA, for the cell line expressing the human μ-receptor subtype.
1. Goodchild CS, Serrao JM. Intrathecal midazolam in the rat: evidence for spinally mediated analgesia. Br J Anaesth 1987; 59: 1563–70.
2. Rattan AK, McDonald JS, Tejwani GA. Differential effects of intrathecal midazolam on morphine-induced antinociception in the rat: role of spinal opioid receptors. Anesth Analg 1991; 73: 124–31.
3. Moreau J-L, Pieri L. Effects of an intrathecally administered benzodiazepine receptor agonist, antagonist and inverse agonist on morphine-induced inhibition of a spinal nociceptive reflex. Br J Pharmacol 1988; 93: 964–8.
4. Nishiyama T. The post-operative analgesic action of midazolam following epidural administration. Eur J Anaesthesiol 1995; 12: 369–74.
5. Naguib M, el Gammal M, Elhattab YS, Seraj M. Midazolam for caudal analgesia in children: comparison with caudal bupivacaine. Can J Anaesth 1995; 42: 758–64.
6. Nishiyama T, Matsukawa T, Hanaoka K. Continuous epidural administration of midazolam and bupivacaine for postoperative analgesia. Acta Anaesthesiol Scand 1999; 43: 568–72.
7. Valentine JM, Lyons G, Bellamy MC. The effect of intrathecal midazolam on post-operative pain. Eur J Anesthesiol 1996; 13: 589–93.
8. Goodchild CS, Noble J. The effects of intrathecal midazolam on sympathetic nervous system reflexes in man: a pilot study. Br J Clin Pharmacol 1987; 23: 279–85.
9. Serrao JM, Marks RL, Morley SJ, Goodchild CS. Intrathecal midazolam for the treatment of chronic mechanical low back pain: a controlled comparison with epidural steroid in a pilot study. Pain 1992; 48: 5–12.
10. Kohno T, Kumamoto E, Baba H, et al. Actions of midazolam on GABAergic transmission in substantia gelatinosa neurons of adult rat spinal cord slices. Anesthesiology 2000; 91: 507–15.
11. Rattan AK, Tejwani GA. Sodium ions modulate differentially the effect of a benzodiazepine agonist on rat spinal μ-, δ-, and κ-opioid receptors. Pharmacology 1994; 48: 30–40.
12. Szekeres PG, Traynor JR. Delta opioid modulation of the binding of guanosine-5′-O-(3-[35S]thio)triphosphate to NG108-15 cell membranes: characterization of agonist and inverse agonist effects. J Pharmacol Exp Ther 1997; 283: 1276–84.
13. Breivogel CS, Selley DE, Childers SR. Acute and chronic effects of opioids on δ and μ receptor activation of G proteins in NG108-15 and SK-N-SH cell membranes. J Neurochem 1997; 68: 1462–72.
14. Zhu J, Luo L-Y, Li J-G, et al. Activation of the cloned human kappa opioid receptor by agonists enhances [35
S]GTPγS binding to membranes: determination of potencies and efficacies of ligands. J Pharmacol Exp Ther 1997; 282: 676–84.
15. Kuger TJ, Hayashi T, Weiss CG, Hill HF. The spinal potentiating effects and the supraspinal inhibitory effect of midazolam on opioid-induced analgesia in rats. Eur J Pharmacol 1995; 275: 153–62.
16. Goodchild CS, Guo Z, Musgreave A, Gent JP. Antinociception by intrathecal midazolam involves endogenous neurotransmitters acting at spinal delta opioid receptors. Br J Anaesth 1996; 77: 758–63.
17. Cappelli A, Anzini M, Vomero S, et al. Synthesis, biological evaluation, and quantitative receptor docking simulations of 2-[(acylamino)ethyl]-1,4-benzodizepines as novel tifluadom-like ligands with high affinity and selectivity for κ-opioid receptors. J Med Chem 1996; 39: 860–72.