Introduction

Top Abstract Introduction Methods Results Discussion References

The benzodiazepines are a family of hypnotic, anxiolytic and anticonvulsant drugs that have major clinical significance. The development of tolerance is a common outcome of prolonged exposure to these compounds, a phenomenon that can be attributed to functional rather than pharmacokinetic accommodation (Greenblatt and Shader, 1986; Rosenberg and Chiu, 1985). The major site of benzodiazepine action in the central nervous system is on GABA_A receptors. Benzodiazepines bind with high affinity to a major subgroup of GABA_A receptors, potentiating the gating of chloride currents by GABA and thus enhancing postsynaptic inhibition (Macdonald and Olsen, 1994). Chronic administration of benzodiazepines to rodents produces a decline both in GABA_A receptor-mediated currents (Gallager et al., 1984) and GABA-gated ³⁶Cl flux in certain brain regions that coincides with the onset of tolerance (Lewin et al., 1989; Marley and Gallager, 1989; Miller et al., 1988). However, the molecular events that underlie this loss of receptor function are not well defined. In benzodiazepine-treated animals, reductions have been detected in the number of binding sites (down-regulation) for GABA_A receptor ligands (Miller et al., 1988; Tietz et al., 1986; Wu et al., 1994) or in the allosteric coupling between the benzodiazepine and GABA binding sites on receptors (Tietz et al., 1989; Xie and Tietz, 1992). Although reductions in GABA_A receptor subunit mRNAs have also been documented in benzodiazepine-treated rodents (Heninger et al., 1990; Kang and Miller, 1991; O'Donovan et al., 1992; Zhao et al., 1995), it has been suggested that these changes may occur after the onset of tolerance and loss of receptor binding (Kang and Miller, 1991). Similarly, chronic administration of GABA agonists to chick embryos or to cortical neurons in vitro produces a down-regulation of GABA_A receptor binding and receptor polypeptides that precedes detectable changes in receptor subunit mRNAs (Baumgartner et al., 1994; Calkin et al., 1994; Calkin and Barnes, 1994a, 1994b) or translational rates of alpha-1 subunits (Miranda and Barnes, 1997).

Acute treatment of cortical neurons in culture with GABA or clonazepam induces the internalization of GABA_A receptor binding sites (Tehrani and Barnes, 1991), and comparable amounts of GABA_A receptor polypeptides are also sequestered from the cell surface (Calkin and Barnes, 1994a, 1994b). A large fraction of the internalized GABA_A receptor polypeptides appear to be degraded, providing a potential mechanism for use-dependent down-regulation (Calkin and Barnes, 1994a, 1994b; Miranda and Barnes, 1997). The identification of GABA_A receptor ligand binding on highly purified CCVs from rat brain suggests that receptor sequestion also occurs in vivo (Tehrani and Barnes, 1993). To further examine this pathway of GABA_A receptor regulation, lorazepam was chronically administered to mice using procedures that produce tolerance (Miller et al., 1988). Here we report that this benzodiazepine treatment leads to accumulation of GABA_A receptors on coated vesicles, whereas receptors on synaptic membranes declined. Some of these data have been presented in a preliminary form (Tehrani and Barnes, 1994).

	Methods

Top Abstract Introduction Methods Results Discussion References

Materials. [³H]Flunitrazepam (84.3 Ci/mmol) was obtained from New England Nuclear Research Products (Boston, MA), and [-³²P]ATP was from ICN (Costa Mesa, CA). Lorazepam was provided by Wyeth-Ayerst Research (Princeton, NJ), and clonazepam was provided by Hoffmann-LaRoche (Nutley, NJ). Osmotic pumps were purchased from Alza (Palo Alto, CA). Polyethylene glycol 400 and poly-L-lysine (64 kDa) were from Sigma Chemical (St. Louis, MO).

Lorazepam administration. Adult (10-12 weeks old) C57BL/6J mice were obtained from a breeding colony at Baylor College of Medicine. The animals were maintained in a light- and temperature-controlled environment and fed laboratory chow and water ad libitum. Lorazepam was administered as described by Miller et al. (1988). In brief, lorazepam was dissolved in polyethylene glycol 400 and placed in Alzet 2001 osmotic pumps to deliver 2 mg/kg/day. The mice were lightly anesthetized with methoxyflurane, and the pumps were implanted subcutaneously. Control animals were implanted with pumps containing only polyethylene glycol. The mice were treated for 7 days and then killed.

Tissue preparations. Trunk blood samples and brains were collected rapidly and chilled to 4°C. Coated vesicle fractions were isolated as described by Tehrani and Barnes (1993). The brains were homogenized individually in 3 volumes of isolation buffer [10 mM 2-(N-morpholino)ethanesulfonic acid, pH 6.5, 100 mM KCl, 1 mM EGTA, 0.5 mM MgCl₂, 0.3 mM phenylmethylsulfonyl fluoride, 50 µM DDT, 2 mM benzamidine, 0.1 mg/ml bacitracin and 0.1 mg/ml soybean trypsin inhibitor]. The homogenate was centrifuged at 20,000 × g for 30 min. The supernatant was collected, and the pellet was resuspended and recentrifuged as above. The pellets were retained, and the two supernatants were combined. To have sufficient amounts of material, the 20,000 × g supernatants derived from five brains were pooled and centrifuged at 100,000 × g for 1 hr to obtain a supernatant (cytoplasmic fraction) and a microsomal pellet. This pellet was resuspended in 2 ml of isolation buffer, mixed with 2 ml of a solution containing 12.5% Ficoll 400 (w/v) and 12.5% sucrose (w/v) in isolation buffer and then centrifuged at 42,000 g for 40 min. The resulting supernatant was collected and mixed with 3 volumes of isolation buffer containing Triton X-100 (where indicated) to give a final concentration of 0.1%. The mixture was incubated for 30 min on ice and then centrifuged at 100,000 × g for 90 min. The resulting pellet (TCV) was resuspended in isolation buffer.

[³H]Flunitrazepam binding assay. Crude synaptic membrane (60-100 µg of protein) or coated vesicle fractions (30-60 µg of protein) were added to an assay mixture containing 10 mM sodium phosphate, pH 7.4, 100 mM KCl, 1 mM EGTA, 0.5 mM MgCl₂ and 10 nM [³H]flunitrazepam in a final volume of 400 µl. For Scatchard analysis, concentrations of [³H]flunitrazepam from 0.25 to 40 nM were used. Nonspecific binding was defined with 1 µM clonazepam. Incubations were carried out for 30 min at 4°C and then terminated by the addition of 3.5 ml of ice-cold phosphate buffer and filtration through glass-fiber filters (no. 32; Schleicher & Schuell, Keene, NH) that had been treated with 0.3% polyethylenimine. The filters were washed twice with 3.5 ml of ice-cold assay buffer, dried and counted by liquid scintillation.

Clathrin light-chain kinase assay. Clathrin kinase activity in crude microsomal and CCV fractions was determined as previously described (Tehrani and Barnes, 1993). In brief, 35 µg of protein from tissue fractions was added to an assay mixture containing 25 mM HEPES, pH 7.4, 10 mM MgCl₂, 1 mM EGTA, 1 mM EDTA, 1 mM DTT and, where indicated, 5 µg of poly-L-lysine in a final volume of 50 µl. The mixtures were incubated for 5 min at 30°C, followed by the addition of [-³²P]ATP (20 µM final concentration, 5 Ci/mmol). After 3 additional min, the assay was terminated by the addition of 25 µl of stop solution (1% SDS, 20% glycerol, 0.2% phenol red, 0.1 M DTT). Aliquots were electrophoresed on 10% polyacrylamide-SDS gels (Garfin, 1990) that were dried and exposed to Kodak X-OMAT AR film.

Quantitative immunoblotting. GABA_A receptor alpha-1 subunit immunoreactivity was determined by Western blotting using an antibody (RP4) against an alpha-1(331-381) fusion protein described by Miranda et al. (1997). Clathrin heavy chains were detected on Western blots using monoclonal antibody F21-35C (provided by Dr. E. M. Lafer) and ECL enhanced chemiluminescent detection kits (Amersham, Arlington Heights, IL). Signal intensities on x-ray film were quantified using a laser scanner and QuantiScan software (Biosoft, Ferguson, MO). Western blots with a series of membrane protein concentrations were used to determine the linear range of analysis for GABA_A receptor alpha-1 subunit immunoreactivity (Miranda and Barnes, 1997). All of the alpha-1 subunit quantification was carried out within this linear range.

	Results

Top Abstract Introduction Methods Results Discussion References

CCV fractions from mouse brain were obtained essentially as previously described (Tehrani and Barnes, 1993) using modifications of the procedures of Bar-Zvi and Branton (1986). Microsomal fractions were centrifuged in 6.25% Ficoll 400/6.25% sucrose to produce a light vesicle supernatant greatly enriched in CCVs (crude CCVs). Further purification of coated vesicles was obtained by extracting smooth vesicle contaminants in 0.1% Triton to produce TCVs as suggested by Pearse (1982). As shown in fig. 1A, these latter fractions were markedly enriched in the 170-kDa clathrin heavy chain. A progressive increase in clathrin heavy chain was observed from homogenate through microsomal, crude CCV and TCV fractions. Purification of coated vesicles was also evaluated by measuring clathrin light-chain kinase, a poly-L-lysine-dependent activity found exclusively on these particles (Schook and Puszkin, 1985). Compared with the other fractions, TCVs showed a large increase in phosphorylation of the 34-kDa clathrin light chain (fig. 1B). The enrichment in clathrin heavy chain and clathrin light-chain kinase found in these experiments was similar to that obtained previously through gel filtration of the crude CCV fraction (Tehrani and Barnes, 1993).

View larger version (47K): [in this window] [in a new window]   Fig. 1.   Top, SDS-polyacrylamide gel analysis of brain fractions. Tissue fractions were prepared as described in Methods. Proteins (10 µg) were analyzed by electrophoresis on 6.5% polyacrylamide-SDS gels that were stained with Coomassie blue. Shown are homogenate (lane 1), cytoplasmic fraction (lane 2), crude synaptic membranes (lane 3), microsomal fraction (lane 4), crude CCV (lane 5) and TCV (lane 6). Right column, position of molecular mass markers (in kDa). Bottom, endogenous phosphorylation in tissue fractions. Aliquots of fractions (35 µg of protein) were incubated with [-32P]ATP as described in Methods in the absence () or presence (+) of 5 µg of poly-L-lysine. The products were analyzed on 10% polyacrylamide-SDS gels that were stained, dried and autoradiographed as described in Methods.

The levels of [³H]flunitrazepam binding and protein recovery in these fractions are summarized in table 1. For TCVs, the clonazepam-displaceable binding obtained in the presence of 10 nM [³H]flunitrazepam (74 fmol/mg of protein) represented 8.2% of the level found in crude synaptic membranes. A comparable value (5.9%) was obtained for the relative amount of [³H]flunitrazepam binding previously found for column-purified CCVs (Tehrani and Barnes, 1993). Triton extraction of the crude CCV fraction (table 1) increased the level of flunitrazepam binding by 120 ± 11%, whereas the amount of protein declined by 41%. This is consistent with removal of smooth vesicles by the detergent treatment.

Lorazepam was administered subcutaneously to adult mice via osmotic minipumps. To determine the serum concentration of lorazepam, a competitive binding assay was used. Unlabeled lorazepam displaced the specific binding of [³H]flunitrazepam in a dose-dependent manner with an IC₅₀ value of 25 nM (fig. 2A). The displacement plot was used as a standard curve to estimate the lorazepam content of serum. Specific [³H]flunitrazepam binding to synaptic membranes was displaced by sera from lorazepam-treated animals in a volume-dependent manner (fig. 2B). Serum from vehicle-treated animals was ineffective. The amount of lorazepam estimated from the standard curve was proportional to the amount of serum up to 20 µl (fig. 2C). At the end of the 7-day treatment with lorazepam (2 mg/kg/day), the average serum level in the experimental group was 103.3 ± 8.9 ng/ml (fig. 2D). Using the same treatment protocol, Miller et al. (1988) found a lorazepam level of 80 ng/ml through chromatographic analysis. Thus, the regimen used in our study appears to be sufficient to produce tolerance to lorazepam as defined by Miller et al. (1988).

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Assay of serum lorazepam by competitive receptor binding. A, Competitive displacement of [3H]flunitrazepam binding by lorazepam. Binding assays with synaptic membranes were carried out as described in Methods with 1 nM [3H]flunitrazepam and the indicated concentration of lorazepam. The results are expressed as a percentage of controls without lorazepam and represent values from a typical assay carried out in triplicate. B, Displacement of [3H]flunitrazepam binding by serum from lorazepam-treated animals. Binding assays as in A were carried out in the presence of sera from vehicle (Con)- or lorazepam (LZ)-treated animals. The results are expressed as a percentage of controls that lacked serum. C, Relationship of lorazepam content and serum volume. From the data of B, the amount of lorazepam was estimated using the plot in A as a calibration curve. D, Serum concentration of lorazepam in chronically treated mice. Lorazepam (2 mg/kg/day) was administered as described in Methods, and blood samples were obtained after 7 days of treatment. Assays were carried out as in C using 10 µl of serum. Results are mean ± S.E.M. from 15 animals.

Through Scatchard analysis, the parameters for [³H]flunitrazepam binding to synaptic membranes were determined for the vehicle- and lorazepam-treated animals. K_d values for the control and drug-treated groups, 1.55 ± 0.04 nM and 1.86 ± 0.52 nM (mean ± S.E.M., n = 6 preparations), respectively, did not differ significantly, whereas the B_max value for the mice receiving lorazepam showed a 33% reduction relative to controls (fig. 3, top). Because of the limited amounts of coated vesicles obtainable from mouse brain (~100 µg/animal; table 1), saturation studies of [³H]flunitrazepam binding were carried out only on tissue from untreated animals. For the coated vesicle fraction, a K_d value of 2.08 ± 0.14 nM and a B_max value of 220 ± 7 fmol/mg were obtained. A 10 nM concentration of [³H]flunitrazepam was used to estimate the level of receptors in TCVs from treated mice. As shown in fig. 3 (bottom), the level of binding to coated vesicles from the group exposed to lorazepam was increased by 82.9 ± 12.8% compared with controls. Assays of microsomal fractions did not reveal any significant differences in binding for the vehicle- vs. drug-treated mice (not shown).

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Effect of chronic administration of lorazepam on [3H]flunitrazepam binding to synaptic membranes and CCVs. Lorazepam (LZ) was administered for 7 days as in fig. 2. Controls (Con) were treated simultaneously with vehicle only. Crude synaptic membrane and TCV fractions were isolated, and [3H]flunitrazepam binding assays were carried out in triplicate as in table 1. For synaptic membranes (top), Scatchard analysis was carried out over a [3H]flunitrazepam concentration range of 0.25 to 40 nM. The values represent the mean ± S.E.M. Bmax value from six membrane preparations. Each preparation, derived from one brain, was from an independent experiment. For TCVs (bottom), the values are from binding assays containing 10 nM [3H]flunitrazepam and represent the mean ± S.E.M. from three preparations. Each preparation was from an independent experiment using five mice.

Quantitative Western blotting was used to determine the amounts of clathrin heavy chain and of GABA_A receptor alpha-1 subunits in the membrane fractions. As indicated in fig. 4 (left), immunoreactivity of the 170-kDa clathrin heavy chain in TCVs showed little difference in the lorazepam-treated animals compared with controls. Quantification of similar immunoblots of three independent TCV preparations from each treatment group (not shown) revealed that the clathrin level from the drug-treated mice represented 98.7 ± 3.9% (mean ± S.E.M.) of that of the vehicle-treated controls. This demonstrates that lorazepam administration had no gross effects on the yield of coated vesicles. On the other hand, the level of the GABA_A receptor 51-kDa alpha-1 subunit in TCVs from treated mice showed an increase compared with controls (fig. 4, right). The smaller amount of immunoreactivity at ~100 kDa appears to be nonspecific. Preabsorption of the antibody with the GABA_A receptor alpha-1(331-381) fusion protein used as the antigen (Miranda et al., 1997) eliminated the immunoreactive 51-kDa band but had little effect on the 100-kDa band (not shown). The relative amounts of the 51-kDa alpha-1 subunit on similar Western blots of three TCV preparations from each experimental group were determined with a previously validated method (Miranda and Barnes, 1997); this revealed that the content of GABA_A receptor alpha-1 subunits in TCVs from lorazepam-treated animals was increased by 58.5 ± 13.5% (mean ± S.E.M.) relative to the controls. When clathrin heavy-chain immunoreactivity was used as an internal control on each Western blot to normalize the amounts of GABA_A receptor alpha-1 subunits, similar results were obtained. This analysis showed that TCVs from the mice exposed to lorazepam had a 60.0 ± 10.3% (n = 3) elevation of alpha-1 subunits compared with those from the vehicle-treated animals. Consistent with fig. 4, quantification of immunoblots of crude synaptic membranes showed only an 11.5 ± 60.1% decline in alpha-1 subunits from the drug-treated mice compared with controls.

View larger version (64K): [in this window] [in a new window]   Fig. 4.   Western blotting of membrane fractions from control and lorazepam-treated mice. Membrane fractions from treated (LZ) and control (C) animals were prepared as in fig. 3. Left, clathrin immunoreactivity. Proteins (25 µg) from TCVs were resolved on 6.5% polyacrylamide-SDS gels, transferred to nitrocellulose and blotted with undiluted F21-35C hybridoma supernatant (30 min at 22°C). The membranes were washed and incubated for 30 min with a 1:3000 dilution of goat anti-mouse IgM antibody conjugated with horseradish peroxidase (Boehringer-Mannheim Biochemicals, Indianapolis, IN). Detection was by chemiluminescence (ECL kit, Amersham). Positions of molecular mass markers are shown (in kDa) (left margin). Right, GABAA receptor alpha-1 subunit immunoreactivity. Proteins (25 µg) from crude synaptic membranes (SM) or TCVs were resolved on 10% gels and then transferred and blotted with alpha-1(331-381) antibody as described by Miranda et al. (1997). 125I-Protein A was used for detection. Preimmune serum gave no detectable reactivity with either SM or TCV samples (not shown).

	Discussion

Top Abstract Introduction Methods Results Discussion References

The number of neurotransmitter receptors on the synaptic plasma membrane is regulated by the traffic of intracellular vesicles. Golgi-derived vesicles provide newly synthesized receptors to the cell surface, whereas CCVs are the initial vehicles for sequestration of surface receptors, which are ultimately degraded or recycled. Muscarinic acetylcholine receptors and beta-adrenergic receptors, which are subject to agonist-induced sequestration and down-regulation (Benovic et al., 1988; Thompson and Fisher, 1990), are found on CCVs (Chuang et al., 1986; Silva et al., 1986).

Acute exposure of cortical neurons in vitro to GABA or clonazepam induces sequestration of GABA_A ligand binding sites (Calkin and Barnes, 1994a, 1994b; Tehrani and Barnes, 1991). Consistent with operation of such a mechanism in vivo, we previously identified central benzodiazepine receptors on highly purified CCV fractions from rat brain (Tehrani and Barnes, 1993). However, in vivo studies have been hampered by difficulties in purifying CCV fractions from limited amounts of tissue. To facilitate such experiments, we changed the final purification step, eliminating gel filtration of crude CCVs and substituting an extraction with Triton X-100. Contamination by smooth vesicles, which account for ~50% of the protein in the crude coated vesicle fraction, was reduced through Triton extraction, a treatment that spares CCVs (Pearse, 1982; Weidenmann et al., 1985). We found earlier that >70% of the [³H]flunitrazepam binding sites on column-purified CCVs could be recovered after Triton extraction (Tehrani and Barnes, 1993). In the present study, the purification of coated vesicles was monitored by enrichment for the clathrin heavy chain (Bar-Zvi and Branton, 1986) and clathrin light-chain kinase (Schook and Puszkin, 1985). By these measures, the degree of enrichment of coated vesicles obtained here was similar to that reported earlier (Tehrani and Barnes, 1993). The level of [³H]flunitrazepam binding obtained under standard conditions (10 nM radioligand) was also similar: 59 fmol/mg for column-purified CCVs (Tehrani and Barnes, 1993) and 74 fmol/mg for TCVs (table 1).

To examine the possibility that agonist-dependent sequestration of GABA_A receptors occurs in vivo, we used procedures described by Miller et al. (1988) for the administration of lorazepam to mice. In that study, lorazepam treatment (2 mg/kg/day) for a 7-day period produced tolerance, a serum lorazepam level of 60 to 80 ng/ml and a 25% reduction in the B_max value for [³H]flunitrazepam binding to crude synaptic membranes (P₂ fraction) from the cerebral cortex. Using an identical protocol for drug administration, we found a serum lorazepam level of 103 ± 8.9 ng/ml and a 33 ± 4% reduction in the B_max value for [³H]flunitrazepam binding to synaptic membranes from whole brain. Thus, agreement between these two series of experiments seems to be excellent. As an alternative explanation of our results, the possibility that residual lorazepam could produce an artifactual displacement of [³H]flunitrazepam binding was considered. However, this appears unlikely because the washing procedures used during isolation of TCVs or treatment of crude synaptic membranes (see Methods) removes contaminating benzodiazepines (Miller et al., 1988; Wu et al., 1994).

We found that the level of [³H]flunitrazepam binding to coated vesicles (TCVs) from the lorazepam-treated animals was nearly twice that from the vehicle-treated controls. This result is highly significant (P < 0.025) and indicates that use-dependent sequestration of GABA_A receptors can occur in vivo. We considered three other explanations for our results. First, synthesis of some receptor subunits could have undergone an up-regulation as a result of the drug treatment (Galpern et al., 1990). Newly synthesized receptors are presumably present on smooth vesicles derived from the endoplasmic reticulum or Golgi complex, which could contaminate the coated vesicle fraction. Although such smooth vesicles should be present in even higher numbers in the microsomal fraction, we found that lorazepam administration had no significant effect on the microsomal binding of [³H]flunitrazepam. Second, the drug treatment could have produced an altered expression of GABA_A receptor subunits (Wu et al., 1994), which increased the affinity of [³H]flunitrazepam in TCVs. However, nearly saturating levels of [³H]flunitrazepam were used in the TCV assays. Furthermore, lorazepam administration did not significantly alter the K_d values for synaptic membranes. Finally, we considered the possibility that [³H]flunitrazepam binding to TCVs from the treated mice could have been stimulated allosterically by endogenous GABA. To remain after such extensive washing of membranes, any residual GABA would probably be trapped within vesicles, but intravesicular GABA should have been released by the Triton extraction. Furthermore, the GABA antagonist SR95531 has no effect on [³H]flunitrazepam binding to coated vesicles, showing that endogenous GABA does not compromise the assays (Tehrani and Barnes, 1995). Thus, we suggest that a lorazepam-dependent increase in receptor number on coated vesicles is the best explanation for the data. However, limitations in the amount of coated vesicles (table 1) that can be obtained from drug-treated mice prevented us from carrying out a complete saturation analysis of equilibrium [³H]flunitrazepam binding.

To show more directly that the drug treatment produced an increase in GABA_A receptor number in coated vesicles, we used quantitative immunoblotting of the alpha-1 subunit. This polypeptide is the major site for [³H]flunitrazepam photolabeling in rat brain (Macdonald and Olsen, 1994). The TCV fractions from treated mice showed nearly a 60% rise in alpha-1 subunit immunoreactivity relative to controls, whereas the synaptic membrane fraction experienced a very small decline (12%). This redistribution of GABA_A receptor alpha-1 subunits in the lorazepam-treated mice is similar to that observed for [³H]flunitrazepam binding. Thus, our data provide good support for the occurrence in vivo of lorazepam-dependent GABA_A receptor sequestration.

The mechanism of benzodiazepine tolerance is not understood. There is evidence that GABA_A receptor functions are compromised in some, but not all, brain regions of tolerant rodents (Marley and Gallager, 1989; Miller et al., 1988; Ramsey et al., 1991; Wilson and Gallager, 1988; Xie and Tietz, 1992). However, no clear picture of underlying alterations in receptor molecules has emerged. Some investigators have reported a down-regulation of GABA_A receptor ligand binding sites (Crawley et al., 1982; Miller et al., 1988; Tietz et al., 1986; Wu et al., 1994), which is in accord with our current study of [³H]flunitrazepam binding to synaptic membranes. However, others found no change in ligand binding to tissues from tolerant rats (Gallager et al., 1984; Impagnatiello et al., 1996). For GABA_A receptor subunit mRNAs, a complex pattern of drug-induced regional up and down variations was reported, but this pattern differs from that for the corresponding subunit polypeptides (O'Donovan et al., 1992; Impagnatiello et al., 1996). Our data suggest that some of these inconsistencies may be due to a subcellular redistribution of GABA_A receptors in the tolerant animals. For example, receptor sequestration on coated vesicles could reduce GABA-gated currents without producing changes detectable by ligand binding autoradiography or subunit immunoreactivity at a multicellular level. At a gross tissue level, the amount of GABA_A receptors on coated vesicles represents only a small fraction of the total. However, sequestration, like down-regulation (Tietz et al., 1986; Impagnatiello et al., 1996; Wu et al., 1994), may be anatomically regionalized.

The significance of benzodiazepine-induced sequestration of GABA_A receptors, either in vitro (Tehrani and Barnes, 1991) or in vivo, has not been clearly established. It has been shown that a fraction of sequestered GABA_A receptors are rapidly degraded, providing a pathway for use-dependent down-regulation (Calkin and Barnes, 1994a, 1994b). Miranda and Barnes (1997) also suggested that sequestration could provide intracellular signals for the regulation of GABA_A receptor expression. Alternatively, internalized receptors could be recycled to the neuronal surface (Barnes, 1996). Under appropriate conditions, such as drug withdrawal, this recycling could replenish the surface with GABA_A receptors, returning membrane currents to the normal state. Based on our current data, we advance the hypothesis that GABA_A receptor sequestration may contribute to the establishment of tolerance to benzodiazepines. Additional studies of coated vesicle components and their trafficking during benzodiazepine administration will be necessary to more definitively examine this hypothesis.