Functional Compartmentalization of Opioid Desensitization in Primary Sensory Neurons 1

Transcription

1 /00/ $03.00/0 THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS Vol. 294, No. 2 Copyright 2000 by The American Society for Pharmacology and Experimental Therapeutics Printed in U.S.A. JPET 294: , 2000 /2438/ Functional Compartmentalization of Opioid Desensitization in Primary Sensory Neurons 1 GARY M. SAMORISKI and ROBERT A. GROSS Departments of Neurology and Pharmacology & Physiology, University of Rochester School of Medicine & Dentistry, Rochester, New York Accepted for publication May 3, 2000 This paper is available online at Persistent activation of opioid receptors results in the gradual loss of the pharmacological action of opiate compounds (homologous desensitization) and, in some cases, other compounds selective for distinctly different receptor systems (heterologous desensitization). Because opioid receptors are coupled to a variety of G proteins, and thereby regulate multiple signaling pathways, the biochemical mechanisms resulting in desensitization are complex and not fully understood. Multiple opioid-activated pathways have been implicated in this process, including activation of the phospholipase C-protein kinase C (PKC) pathway (Spencer et al., 1997; Strassheim et al., 1998). Desensitization probably also involves a complex array of interacting signaling pathways, including G protein receptor kinases, which may phosphorylate agonist-occupied receptors, and arrestins, which functionally uncouple the phosphorylated receptor from the G Received for publication December 28, This study was supported by Grants DA07232 (to G.M.S.) and DA10514 (to R.A.G.) from the National Institutes of Health. Preliminary results of this study have been presented at the 27th (1997 Oct 25 30, New Orleans, LA) and 28th (1998 Nov 7 12, Los Angeles, CA) annual meetings of the Society for Neuroscience. ABSTRACT The cellular correlates of desensitization or tolerance are poorly understood. To address this, we studied acute and long-term -opioid desensitization, with respect to Ca 2 currents, in cultured rat dorsal root ganglion (DRG) neurons. Exposure of DRG neurons to the -agonist [D-Ala 2,N-MePhe 4,Gly-ol 5 ]-enkephalin (DAMGO; 3 M) reduced whole-cell currents 35%, but with continued agonist application, 52% of the response was lost over 10 to 12 min. In contrast, exposure of DRG neurons to DAMGO for 24 h resulted in a nearly complete loss of Ca 2 channel regulation after washing and re-exposure to DAMGO. Responses to the -aminobutyric acid B agonist baclofen were not affected in these neurons. Acute desensitization preferentially affected the voltage-sensitive component of -opioid and -aminobutyric acid B responses. Facilitation of both the DAMGO- and baclofen-inhibited current by a strong depolarizing prepulse was significantly attenuated in acutely desensitized neurons. Because G -subunits mediate neurotransmitter-induced changes in channel voltage-dependent properties, these data suggest an altered interaction of the G -subunit with the Ca 2 channel. Block of N-type Ca 2 channels with -conotoxin GVIA revealed a component of the opioid response that did not desensitize over 10 min. We conclude that acute and long-term -opioid desensitization in DRG neurons occurs by different mechanisms. Acute desensitization is heterologous and functionally compartmentalized: the pathway targeting non-n-type channels is relatively resistant to the early effects of continuous agonist exposure; the pathway targeting N-type channels in a largely voltage-insensitive manner is partially desensitized; and the pathway targeting N-type channels in a largely voltage-sensitive manner is completely desensitized. protein (Inglese et al., 1993; Chuang et al., 1996; Ferguson et al., 1996). To describe this phenomenon at the cellular level, the goal of this study was to examine the characteristics of desensitization resulting from short-term and prolonged opioid exposure in primary sensory neurons and to identify a potential target by which the loss of opioid response occurs. An important signaling pathway regulated by -opioids involves the rapid inhibition of high voltage-activated Ca 2 channels (Wilding et al., 1995). In sensory neurons, N- and P/Q-type Ca 2 channels are affected predominantly, with lesser effects on other subtypes (Schroeder et al., 1991; Seward et al., 1991; Moises et al., 1994a; Rusin and Moises, 1995). Regulation of voltage-gated Ca 2 channels can have significant consequences on neuronal activity because they play a central role in membrane excitability and neurotransmitter release (Lipscombe et al., 1989; Luebke et al., 1993). Inhibition of Ca 2 channels by -opioids occurs through the receptor-mediated activation of a pertussis toxin-sensitive G protein (G o ; Moises et al., 1994b), resulting in a shift in the voltage-dependence of ion channel activation (Hille, 1994). Qualitatively, the partial reversal of the current inhibition and restoration of fast activation kinetics reflect, in part, this ABBREVIATIONS: PKC, protein kinase C; DAMGO, [D-Ala 2,N-MePhe 4,Gly-ol 5 ]-enkephalin; DRG, dorsal root ganglion; V h, holding potential; CTOP, D-Phe-Cys-Tyr-D-Trp-Orn-Thr-Pen-Thr-NH 2 ; GABA, -aminobutyric acid; IR, inhibition ratio. 500

2 2000 Compartmentalization of Neuronal Opioid Desensitization 501 change in calcium channel gating when the cell is strongly depolarized. Voltage-dependent modulation of calcium channels has been shown to result from receptor-mediated dissociation of the G protein and subsequent binding of the subunit to the calcium channel (Herlitze et al., 1996; Ikeda, 1996; Delmas et al., 1998a; Zamponi and Snutch, 1998). However, this does not account for all of the channel inhibition in sensory neurons. This study was undertaken to investigate acute and longterm opioid densensitization with respect to Ca 2 channel regulation. These experiments confirm and extend the findings of Nomura et al. (1994). First, we examined whether the duration of exposure to the -opioid agonist [D-Ala 2,N- MePhe 4,Gly-ol 5 ]-enkephalin (DAMGO) differentially affected G o -mediated inhibition of high voltage-activated Ca 2 channels in neonatal rat dorsal root ganglion (DRG) neurons. Furthermore, it has been well documented that G o -linked receptors inhibit voltage-activated Ca 2 channels through distinct voltage-sensitive and voltage-insensitive mechanisms (for review, see Jones and Elmslie, 1997). Because G o activation inhibits different calcium channel types by different mechanisms, it was of interest to test whether the changes that occur with short-term exposure to DAMGO were selective for a specific channel type and whether this involved a voltage-dependent or -independent pathway. Materials and Methods Preparation of Neuronal Cultures DRG neurons were prepared from 7- to 9-day-old Sprague-Dawley rats (Harlan Sprague-Dawley, Inc., Indianapolis, IN). All animal use procedures were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the University of Rochester Committee on Animal Resources. The ganglia were dissected away from the lumbo-sacral region of the spinal cord and collected on ice in Dulbecco s PBS without Ca 2 or Mg 2 (Life Technologies, Grand Island, NY). After multiple rinses with the Ca 2 /Mg 2 -free medium, the ganglia were treated with trypsin (1.25 mg/ml; Life Technologies) and collagenase (3 mg/ml; Sigma Chemical Co., St. Louis, MO) for 30 min at 37 C. Enzymatic digestion was terminated by rinsing the cells in Hanks buffered saline solution (Life Technologies) containing equine serum (10%; Hyclone Laboratories, Logan, UT) and DNase I (0.05 mg/ml; Worthington Biochemical Corporation, Lakewood, NJ). The cells were suspended in plating medium [minimum essential medium (Life Technologies) supplemented with nerve growth factor (50 ng/ml; Life Technologies), equine serum (5%; Hyclone Laboratories), fetal bovine serum (5%; Hyclone Laboratories), glucose (5 mg/ml; Sigma Chemical Co.), L-glutamine (2 mm; Life Technologies), penicillin (100 U/ml; Life Technologies), streptomycin (100 g/ml; Life Technologies), and gentamicin (0.05 mg/ml; Life Technologies)] and mechanically dispersed by trituration. The suspension (100 l/dish 1 2 DRG/dish) was then plated on 35-mm culture dishes with laminin (Collaborative Biomedical Research, Bedford, MA) as the substrate. After a 2-h incubation at 37 C the volume was brought to 2 ml with 50% plating medium and 50% feeding medium [minimum essential medium containing equine serum (10%), nerve growth factor (50 ng/ml), glucose (5 mg/ml), and L-glutamine (2 mm)]. After 24 h, a 50% exchange with feeding medium was performed. The cultures were maintained at 37 C in a humidified atmosphere of 95% air and 5% CO 2 with a 50% exchange of feeding medium performed once weekly. To maximize the use of all DRG neurons from each preparation and to conduct all experiments in cells that had been maintained in culture for 4 to 14 days without the use of compounds needed to inhibit the growth of nonneuronal cells, some DRG neurons from each preparation (before plating) were placed in plating medium containing 10% dimethyl sulfoxide (Sigma Chemical Co.) and stored at 80 C. When needed, these cells were warmed at 37 C for 2 min, plated ( 1 2 DRG/dish) and, after a 1-h incubation at 37 C, rinsed with plating medium. The medium was replaced with feeding medium and the cultures maintained as described above. Only neurons without processes and with adequate space-clamp (rapid deactivation of tail currents) were accepted for analysis. Freshly prepared neurons and those previously frozen were identical in their electrophysiological and pharmacological properties. Electrophysiology Whole-cell voltage-clamp recordings were executed with the whole-cell variation of the patch-clamp technique (Hamill et al., 1981). Glass recording patch pipettes were shaped from microhematocrit tubes (Chase Instruments Corporation, Norcross, GA) with a Sutter Instruments Flaming/Brown P-87 micropipette puller. Pipettes had resistances of 1.2 to 1.8 M when filled with the following recording solution: 140 mm CsMeSO 3, 10 mm HEPES, 5 mm EGTA [or 1,2-bis(2-aminophenoxy)ethane-N,N,N,N -tetraacetic acid, tetracesium salt], 5 mm ATP-Mg 2, and 0.1 mm GTP-sodium [all reagents from Sigma Chemical Co., except 1,2-bis(2-aminophenoxy)ethane-N,N,N,N -tetraacetic acid; Molecular Probes, Eugene, OR]. The ph was adjusted to 7.35 with 1 N CsOH after the addition of ATP. The osmolarity was 5 to 10% less than the bath solution ( mosm). Aliquots of the internal recording solution were stored at 80 C and kept on ice after thawing. The cells were bathed in an external Ca 2 buffer (ph 7.35; mosm) containing 67 mm choline chloride, 5.3 mm KCl, 100 mm tetraethylammonium chloride, 5.6 mm glucose, 10 mm HEPES, 0.8 mm MgCl 2,and5mM CaCl 2 (all reagents from Sigma Chemical Co.). Recordings were made at room temperature with an Axopatch 1-B patch-clamp amplifier (Axon Instruments, Foster City, CA). Pipette and whole-cell capacitance and series resistance were corrected by compensation circuitry on the amplifier. Series resistance compensation of 80 to 90% was possible without significant noise or oscillation. Whole-cell Ca 2 currents were evoked every 30 s by 100-ms voltage steps to 10 mv (test pulse) from holding potential (V h ) 80 mv. In experiments in which the voltage-sensitive component of agonist regulation of Ca 2 channels was assessed, a 95-ms depolarizing prepulse to 100 mv was elicited 5 ms before the test pulse. Currents were filtered with a Bessel filter at 5 khz ( 3 db) and the records digitized at 5 khz. Data acquisition and analysis was performed with pclamp software (versions 6.0 and 7.0; Axon Instruments) installed on a microcomputer with Pentium processor. Leak current was determined by a P/P6 protocol. This current was digitally subtracted from the relevant inward current to obtain the calcium current. Solution Preparation and Application DAMGO, D-Phe-Cys-Tyr-D-Trp-Orn-Thr-Pen-Thr-NH 2 (CTOP; both from Research Biochemicals International, Natick, MA), and -conotoxin GVIA (Peptides International, Louisville, KY) were prepared as 5 mm, 625 M, and 100 M stock solutions in distilled water, respectively, partitioned into 20- l aliquots, lyophilized, and stored at 20 C. On the day of the experiment, the lyophilized compound was dissolved in the external Ca 2 buffer at the desired concentration. BSA (0.1%; Sigma Chemical Co.) was included in the CTOP solution to minimize peptide binding to the application system. A 10- l aliquot of DAMGO, reconstituted in distilled water, was added directly into the culture medium to achieve a final concentration of 5 M for experiments in which neurons were exposed to the -opioid agonist for 24 h. Baclofen and naloxone hydrochloride (both from Sigma Chemical Co.) were prepared on the day of the experiment at the desired concentration in external Ca 2 buffer. DAMGO, CTOP, baclofen, and naloxone were applied with a gravity-fed U-tube microperfusion system in which the microenviron-

3 502 Samoriski and Gross Vol. 294 ment of the cell was continuously perfused with either external Ca 2 buffer (control) or drug-containing solution. Whole-cell Ca 2 currents were evoked 20 s after switching solutions. -Conotoxin GVIA was applied to the cell under study from a blunt-tipped ( m tip i.d.) glass micropipette positioned 30 m from the cell with pressure ejection (6 10 kilopascals) for a duration of 20 s. The puffer pipette was removed from the bath when not in use. In all experiments, drug concentrations were minimized in the remainder of the culture dish by the continuous gravity-fed influx ( 0.3 ml/min) and vacuum-removed efflux of external bath solution. However, where long applications of drug were used (principally short-term desensitization experiments), only 1 cell/dish was used to eliminate prior exposure to agonist, antagonist, or toxin as a confounding influence on the measured response. Generally, experiments in which the effect of prolonged DAMGO exposure on Ca 2 channel regulation was tested, more than 1 cell/dish was studied. Pretreatment and Testing Protocols Prolonged DAMGO Exposure. DRG neurons were either left untreated or exposed to a 5 M concentration of DAMGO for 24 h. After the pretreatment period, the culture medium ( DAMGO) was removed by rinsing the cells two to three times with external Ca 2 buffer. Healthy neurons were identified, the whole-cell recording was established, and voltage-activated Ca 2 currents were elicited. DAMGO (3 M) or baclofen (50 M) responses were assessed with the U-tube microperfusion application system. Short-Term DAMGO Desensitization. After removal and exchange of the culture medium by rinsing two to three times and replacing with external Ca 2 buffer, whole-cell recordings in DRG neurons were initiated and voltage-activated Ca 2 currents elicited. Once the magnitude of the peak current had stabilized, DAMGO (3 M) was continuously applied with the U-tube microperfusion application system. Calcium currents were elicited every 30 s and desensitization assessed by monitoring the increase in the peak current magnitude with continuous exposure to the -opioid agonist. On completion of desensitization (increase in peak current magnitude had stopped), drug application was terminated by switching the perfusion solution back to external Ca 2 buffer. Calcium currents were then reassessed in the absence of drug to determine the extent of current rundown. -Aminobutyric Acid (GABA) B Modulation after Short- Term DAMGO Desensitization. Because evidence from pilot experiments suggested that a single, brief application of baclofen resulted in a reduced response on subsequent exposure, the protocol used to establish short-term DAMGO desensitization was modified from that described above to avoid the potential confounding influence of repeated baclofen exposure. After identification of a healthy DRG neuron and before establishing the whole-cell patch configuration, a 10-min DAMGO (3 M) application was accomplished with the U-tube microperfusion system. The whole-cell patch configuration was then established and Ca 2 currents were elicited every 30 s until the peak current magnitude was stable (at this point we assumed, based on previous results, that both DAMGO desensitization and current stabilization were complete). Application of the DAMGO-containing solution was then stopped, the DAMGO removed by vacuum-removed efflux, and the perfusion solution switched to one containing only the external Ca 2 buffer. Calcium currents were again elicited every 30 s until the peak current magnitude was stable (i.e., recovery from the DAMGO-mediated inhibition was complete). The effect of a brief application of baclofen (50 M conditioning prepulse) on peak current magnitude was then assessed. Data Analysis The magnitude and time of the whole-cell peak calcium current was determined with the peak detect feature of the pclamp software (Axon Instruments). With few exceptions, currents were elicited every 30 s. Cells exhibiting 7.5% rundown over the course of the experiment were excluded from the analysis. Analysis consisted of comparisons of current peak magnitude, time-to-peak, and percentage change in the peak current exhibited in the presence of drug. The magnitude of the change in peak whole-cell calcium current was defined as the difference between the current magnitude in the absence and presence of drug and expressed as a percentage of the control current magnitude (% change 100 [peak control current/ peak current in presence of drug)/peak control current]). The frequency of -opioid responding cells also was determined. Drug-sensitive cells were defined as those exhibiting an inhibition of peak current magnitude of 10% in the presence of the agonist compared with control. The magnitude of desensitization was defined as the proportion of the initial drug response that was lost on completion of the desensitization paradigm. The voltage-dependent component of agonist-mediated inhibition was assessed with a two-pulse protocol. A test pulse to 10 mv was preceded by a 95-ms depolarizing prepulse to 100 mv with an interpulse interval of 5 ms. The current generated by this test pulse (P2) was compared with the test pulse (P1) that immediately preceded the two-pulse protocol. Reversal of agonist-mediated inhibition was quantified by determining the degree of current facilitation (facilitation ratio) after the conditioning prepulse and expressed as the ratio of the peak current amplitude after prepulse to the peak current without prepulse (P2/P1). Current inhibitions in the absence and presence of a conditioning prepulse were compared and expressed as the inhibition ratio (IR). The voltage-dependent component of current inhibition (percentage) was then estimated to be /IR. Means between different samples were compared with Student s two-tailed t test. Changes within cells were statistically compared with the paired Student s t test. Proportions were compared with the 2 test. Except where indicated, data are presented as the mean S.E. Results -Opioid, but not GABA B, Regulation of Voltage- Gated Ca 2 Channels Is Lost after Prolonged Exposure to DAMGO. Neonatal rat DRG neurons exhibited rapidly activating inward Ca 2 currents when elicited by a 100-ms depolarizing test pulse to 10 mv from a V h of 80 mv. Of the 165 untreated DRG neurons tested with the -opioid selective agonist DAMGO (3 M), 66% exhibited a 10% reduction in peak Ca 2 current ( DAMGO-sensitive ). Brief application of DAMGO resulted in reversible inhibition of the peak whole-cell Ca 2 current, an action that was completely blocked by coapplication of the -opioid antagonist CTOP (3 M). The magnitude of peak current inhibition varied greatly in DAMGO-sensitive cells (10 74%; Fig. 1A) with a mean of % ( S.E.; n 108). Associated with the DAMGO-mediated decrease in peak Ca 2 current magnitude was a slowing of the activation kinetics, as indicated by an increase in the time-to-peak current magnitude (see current traces in Fig. 1B). Before the brief application of DAMGO, the peak of the Ca 2 current was achieved ms after the start of the voltage step to 10 mv. In contrast, the time-to-peak was significantly increased in the presence of the -opioid agonist ( ms; P.001; paired two-tailed t test). To determine the effect of prolonged exposure to DAMGO on -opioid regulation of calcium channels, DRG neurons from paired cultures (as the controls described above) were exposed for 24 h to 5 M DAMGO. Neurons pretreated with the -opioid agonist exhibited a definable DAMGO-mediated reduction in peak Ca 2 current in only 3% of neurons tested (n 73), a significantly lesser proportion compared with

4 2000 Compartmentalization of Neuronal Opioid Desensitization 503 untreated DAMGO-sensitive cells (P.001; ; Fig. 1A). In addition, the opioid-mediated slowing of the activation of targeted calcium channels was not observed in any pretreated cell tested (see current traces in Fig. 1B). Greater concentrations of DAMGO could not restore the loss of -opioid-mediated regulation of Ca 2 channels observed after prolonged opioid exposure. Neither 3 M nor greater concentrations of DAMGO (10 M, n 4; 30 M, n 8) effected a 10% inhibition in the peak current magnitude in any cell tested. Cotreatment of neurons with 5 M DAMGO and 10 M naloxone prevented the loss of DAMGO responses. Finally, the loss of the opioid response was specific and not related to a change in Ca 2 current magnitude (P.1). To determine whether the loss of the opioid response after prolonged exposure to DAMGO was due to an inability of G proteins to interact with targeted Ca 2 channels, the effect of baclofen, a specific GABA B receptor agonist, was tested. GABA B receptors, like -opioid receptors, regulate voltagegated Ca 2 channels through a membrane-delimited pathway mediated by G o (Dolphin and Scott, 1986; Campbell et al., 1993). Therefore, a subset of untreated and DAMGOpretreated neurons was tested with both DAMGO and baclofen to examine the effect of prolonged DAMGO exposure on GABA B regulation of Ca 2 channels. Of 74 untreated DRG neurons tested with both baclofen (50 M) and DAMGO (3 M), 55% were sensitive to both baclofen and DAMGO, 30% responded only to baclofen, 12% only to DAMGO, and 3% did not exhibit a definable response to either agonist. The magnitude of current inhibition mediated by each agonist was similar for the respective drug-sensitive cells. That is, Fig. 1. Effect of prolonged exposure to DAMGO on -opioid and GABA B regulation of high voltage-activated Ca 2 channels. A and C illustrate the distribution of untreated and DAMGOtreated neurons as a function of the magnitude of agonist-mediated inhibition of peak Ca 2 current. The magnitude of inhibition is expressed as the percentage of reduction in peak current after a brief application of the agonist compared with the control current just before the acute drug application. A, significantly greater proportion of untreated cells exhibited a definable DAMGO response ( 10% reduction of the peak current) compared with those that had been pretreated with DAMGO for 24 h ( ; P.001; n 165 untreated, 73 pretreated). C, in contrast, the proportion of DRG neurons that were baclofen sensitive was not significantly different between the two treatment groups ( ; P.55; n 74 untreated, 29 pretreated). B and D, whole-cell Ca 2 currents were evoked by 100-ms voltage steps to 10 mv from V h 80 mv in the absence (control) or presence of 3 M DAMGO (B) or 50 M baclofen (D) in neurons that were either untreated or exposed to 5 M DAMGO for 24 h. baclofen-sensitive neurons exhibited a % current reduction when tested with baclofen, whereas DAMGO effected a % inhibition of peak current magnitude in DAMGO-sensitive cells. In contrast to the effect of prolonged exposure to DAMGO on -opioid regulation of voltage-dependent Ca 2 channels in the above-described DRG neurons, GABA B -mediated inhibition of Ca 2 currents in these cells was unchanged after long-term DAMGO desensitization (Fig. 1C). Although a 24-h exposure to 5 M DAMGO virtually eliminated the DAMGO (3 M) response (7% were DAMGO sensitive), a definable baclofen (50 M) response was observed in 90% of the cells tested (n 29). This fraction was not different from the 85% that were baclofen sensitive in the untreated condition (P.55; ). Similarly, the magnitude of the baclofenmediated inhibition of Ca 2 currents in baclofen-sensitive cells was % and was not different compared with the magnitude of the baclofen response in untreated neurons ( %; n 63; P.06). Baclofen, like DAMGO, slowed activation of targeted Ca 2 channels when the cell was depolarized to 10 mv from a V h of 80 mv. Before the brief application of baclofen, the timeto-peak current amplitude in untreated baclofen-sensitive cells was ms and in the presence of the GABA B agonist activation was slowed to ms (P.001; paired two-tailed t test; n 63; Fig. 1D). Although this characteristic is lost in cells exposed to DAMGO for 24 h when tested with DAMGO, baclofen-mediated slowing of activation is unaltered in neurons pretreated for 24 h with the -opioid agonist. The time-to-peak current amplitude after

5 504 Samoriski and Gross Vol. 294 application of baclofen in baclofen-sensitive DAMGO pretreated cells was ms (n 26) and was not different compared with the time-to-peak in untreated neurons (P.5; two-tailed t test). -Opioid and GABA B Regulation of Voltage-Gated Ca 2 Channels Is Reduced after Short-Term Exposure to DAMGO. Continuous application of 3 M DAMGO resulted in a significant decrease in -opioid-mediated inhibition of voltage-sensitive Ca 2 currents (short-term desensitization). Figure 2, A and B illustrate the effect of short-term application of DAMGO on whole-cell Ca 2 currents compared with its initial response (compare traces 2 and 3) and the failure of the response to recover completely within 10 min (traces 1 and 2 ). Initial exposure to DAMGO reduced the magnitude of the peak Ca 2 current % of the control current amplitude. In comparison, the peak Ca 2 current was inhibited by only % (P.001; n 32) at the end of continuous DAMGO application. Once shortterm desensitization was complete, there was a % loss of DAMGO-mediated inhibition of the Ca 2 current (Fig. 2C). In addition, short-term exposure to DAMGO significantly affected the agonist-induced slowing of channel activation. The time-to-peak current amplitude with initial drug administration was ms and was significantly slower compared with that of the predrug current ( ms; P.001; paired two-tailed t test). After short-term desensitization, however, channel activation in the presence of the agonist was more rapid (time-to-peak ms). Although the time-to-peak did not return to the predrug value, short-term DAMGO exposure significantly affected the ability of the -opioid agonist to slow activation of the channel. So as to avoid confusion between current rundown and loss of response, any cell that exhibited 7.5% current rundown from its predrug current magnitude was excluded from the analysis. Desensitization of -opioid regulation of voltage-gated Ca 2 channels developed rapidly and was partially reversible over 10 min (Fig. 2D). A brief reapplication of 3 M DAMGO 10 min after complete desensitization resulted in an opioid response that was, in general, 60% of the initial DAMGO response (n 5), which remained significantly smaller than the initial DAMGO-mediated reduction (P.05). Two observations suggested that the reduction of DAMGOmediated regulation of voltage-activated Ca 2 channels was due to agonist-induced desensitization of the signaling pathway between the -opioid receptor and the Ca 2 channel and not due to a perfusion-mediated loss, or rundown, of modulation capability. First, reapplication of DAMGO, 10 min after short-term desensitization was complete and the drug removed, resulted in a partial recovery ( 60% of initial drug response) of the response (Fig. 2D). Second, in separate experiments, the magnitude of the drug-mediated inhibition of peak current amplitude was not altered with three brief ( 30 s) applications of DAMGO (3 M) over the course of 8 min (data not shown). To determine whether the desensitization induced by short-term exposure to DAMGO was similar to long-term desensitization and limited to -opioid regulation of Ca 2 channels, GABA B -mediated inhibition of voltage-sensitive Ca 2 currents was examined in cells briefly exposed to DAMGO. Evidence from pilot experiments suggested that a single, brief application of baclofen resulted in a slightly reduced response on subsequent exposure. Baclofen elicited an initial reduction in the peak current magnitude of % (n 6), whereas a second application (8 min after the first) caused a reduction of %. Although these Fig. 2. Effect of short-term exposure to DAMGO on -opioid regulation of high voltage-activated Ca 2 channels. A, whole-cell Ca 2 currents were evoked by 100-ms steps to 10 mv from V h 80 mv. The traces were recorded in a representative DRG neuron at the times indicated (after patch rupture) in B. C, mean ( S.E.) DAMGO-mediated inhibition of Ca 2 currents is significantly reduced after an 10-min exposure to the agonist (late DAMGO) compared with the initial response (initial DAMGO). ***P.001, twotailed paired t test, n 32. D, results (mean S.E.) from a subset of neurons illustrated in C in which recovery was tested 10 min after the removal of the agonist. *P.05; NS; n 5.

6 2000 Compartmentalization of Neuronal Opioid Desensitization 505 values are not significantly different (P.07; two-tailed t test), the possibility exists that repeated exposure to the agonist could alter, however subtly, the signaling pathway between the GABA B receptor and the Ca 2 channel. Therefore, to avoid any potential confounding influence of repeated baclofen exposure, the protocol for establishing short-term DAMGO desensitization was modified from the above-described protocol. Before establishing the recording, 3 M DAMGO was continuously applied to the cell under study for approximately 10 min with the U-tube application system. At this time the whole-cell patch configuration was established and Ca 2 currents were elicited every 30 s until the peak current magnitude was stable ( 5 min from patch rupture); at this point we assumed that both DAMGO desensitization and current stabilization were complete). Application of the DAMGO-containing solution was then stopped; the DAMGO was removed by vacuum-removed efflux of bath solution and the perfusion solution switched to one containing only the external Ca 2 buffer. Calcium currents were then elicited every 30 s until the peak current magnitude was again stable (the assumption was that recovery from the DAMGO-mediated inhibition was complete). The effect of a brief application of baclofen (50 M) on peak current magnitude was then assessed. Only those cells that exhibited a noticeable increase in current magnitude on removal of DAMGO were considered to be DAMGO sensitive and were included in the analysis. In contrast to the unchanged baclofen response after long-term DAMGO exposure, the baclofen response was significantly reduced after short-term DAMGO desensitization. Twelve of 13 (92%) untreated DRG neurons were baclofen sensitive and exhibited a % inhibition of peak current magnitude. Conversely, a 15-min pre-exposure to DAMGO significantly attenuated the baclofen-mediated inhibition of peak current amplitude ( %; P.001; n 7). Because there was ample time for the DAMGO to be removed before testing with baclofen, these results suggest that short-term DAMGO exposure results in cross-desensitization of GABA B -mediated regulation of voltage-activated Ca 2 channels and not occlusion of the baclofen response due to the continued presence of the -opioid agonist. -Opioid Regulation of Non-N-Type Ca 2 Channels Is Not Affected with Short-Term Exposure to DAMGO. Regulation of N-type Ca 2 channels has both voltage-dependent and voltage-independent components, whereas modulation of non-n-type channels is principally voltage-independent (Luebke and Dunlap, 1994; Bourinet et al., 1996; Delmas et al., 1998a; Sun and Dale, 1998). We therefore wished to test whether desensitization differed according to the targeted channel type and its mode of regulation with respect to voltage. If this were the case, DAMGO-mediated desensitization might not occur after eliminating the N-type channel component of the whole-cell current. To test this, the selective and essentially irreversible N- type channel blocker -conotoxin GVIA (10 M) was applied to DAMGO-sensitive neurons; this was followed by a 7-min continuous application of 3 M DAMGO (short-term desensitization is 89 2% complete within this time frame in cells not exposed to the toxin; Fig. 2). The current traces in Fig. 3A (left) illustrate the DAMGO-mediated inhibition of peak current magnitude in the absence of the -conotoxin GVIA (trace 2) and the remaining whole-cell current after a 20-s exposure to the N-type channel blocker (trace 3). The peak current Fig. 3. -Opioid regulation of non-n-type Ca 2 channels does not exhibit short-term DAMGO desensitization. A, superimposed whole-cell Ca 2 currents evoked by a 100-ms test pulse to 10 mv from V h 80 mv at the times (after patch rupture) indicated in B. Currents were evoked before (1, 2) and after (3, 4, 5) a 20-s application of 10 M -conotoxin GVIA. DAMGO-mediated inhibition was first determined for all targeted channels (trace 2). After elimination of the N-type component of the whole-cell current (trace 3), the effect of DAMGO (3 M) was assessed on the remaining current. The opioid response targets primarily N-type channels (compare [1 2] and [3 4] in A and B and and f in C) in these cells. Naloxone was used to terminate the DAMGO response after -conotoxin GVIA pretreatment to confirm that the small effect of DAMGO was due to an opioid action. C, magnitude of DAMGO-mediated inhibition of peak Ca 2 current on all targeted Ca 2 channels ( ) and on non-n-type Ca 2 channels (f) for both early DAMGO versus late DAMGO. Data are the mean S.E. (n 10; NS; P.20), two-tailed paired t test). magnitude was reduced by % (n 10) in response to the first application of DAMGO. Application of -conotoxin GVIA resulted in a % reduction in the whole-cell current in these neurons. After a period of time (2 4 min) to ensure stability of the baseline current after exposure to the N-type channel blocker, DAMGO was applied to the cell under study. The effect of DAMGO on the remaining wholecell current, both before and after a 7-min continuous exposure to the agonist, is depicted in Fig. 3, A (right), B, and C. The reduction of peak current magnitude effected by DAMGO was % of the post-gvia current (or % of the initial DAMGO response), which was not significantly different than the reduction observed at the end of DAMGO application ( %; P.20). Because of the relatively small proportion of non-n-type Ca 2 channels reg-

7 506 Samoriski and Gross Vol. 294 ulated by DAMGO, naloxone was used to terminate the DAMGO effect so as to eliminate the influence of any residual drug on assessing the degree of rundown during continuous DAMGO exposure. Neurons with 7.5% rundown were excluded from the analysis. Short-Term DAMGO Desensitization Significantly Attenuates the Voltage-Sensitive Component of -Opioid Regulation of Ca 2 Channels. Inhibition of highthreshold Ca 2 currents by G protein-coupled receptor agonists results from a shift in the voltage dependence of gating of the Ca 2 channels effected by specific membrane-delimited pathways (Bean, 1989; Hille, 1994). This change from a willing to reluctant channel state likely depends on the interaction of G protein -subunits with Ca 2 channel subunits (Herlitze et al., 1996; Ikeda, 1996; Ford et al., 1998; García et al., 1998). DAMGO-mediated inhibition of Ca 2 currents in neonatal rat DRG neurons exhibited many of the qualitative features of voltage-dependent inhibition. In addition to the DAMGO-mediated slowing of activation kinetics already described, a strong depolarizing prepulse both partially reversed DAMGO-mediated inhibition (facilitation) and restored the fast activation kinetics. Because one of the features of DAMGO desensitization was a shift in the activation kinetics back toward one of more rapid activation, the hypothesis that it was the voltage-dependent component of -opioid regulation of Ca 2 channels that was significantly attenuated during desensitization was tested. Currents, with (P2) and without (P1) a strong depolarizing prepulse to 100 mv, were compared at the beginning of DAMGO (3 M) exposure and again when desensitization was complete. One measure of voltage-dependent regulation of Ca 2 channels derived from this comparison is the facilitation ratio (P2/P1; see Materials and Methods; Ikeda, 1991). With a voltage protocol in which a 95-ms depolarizing prepulse ( 100 mv) is elicited 5 ms before the test pulse, the voltage-dependent component of -opioid and GABA B regulation of Ca 2 channels was determined. The facilitation ratio in DRG neurons in the absence of agonist is (n 6), suggesting little or no tonic inhibition of the Ca 2 channel under our recording conditions. The current traces in Fig. 4A 1 show that DAMGO exposure resulted in a % (n 15) reduction of the peak Ca 2 current and that a strongly depolarizing prepulse relieved 50% of the DAMGO-mediated inhibition (i.e., a DAMGOinduced reduction of %). DAMGO-mediated desensitization in this experiment (Fig. 3C) was % (P Fig. 4. Short-term DAMGO desensitization reduces the voltage-dependent relief of -opioid-mediated inhibition that is elicited by a strong depolarizing prepulse. A, superimposed whole-cell Ca 2 currents evoked by a 100-ms test pulse to 10 mv from V h 80 mv without (P1) or with (P2) a 95-ms depolarizing prepulse ( 100 mv) 5 ms preceding the test pulse. Responses after the initial application of 3 M DAMGO (A 1 ), with or without a conditioning prepulse, are compared with responses after short-term continuous exposure to DAMGO (A 2 ). The protocol used to establish short-term DAMGO desensitization was like that shown in Fig. 2. B, facilitation (mean S.E.) of peak current magnitude in the presence of the agonist (P2/P1) on initial application of the -opioid agonist (acute) and after desensitization (desensitized). C, comparison of DAMGO-mediated inhibition (mean S.E.) before ( ) and after (f) a strong depolarizing prepulse on initial agonist application and after desensitization. This is summarized as the IR (ratio of inhibition before the prepulse to that after the prepulse, mean S.E.). D, scatter plot demonstrating that the remaining DAMGO-mediated inhibition in desensitized neurons (F) lacks a significant voltage-dependent component (expressed as the IR) compared with initial agonist exposure (E). An IR 1.00 (dotted line) indicates that the inhibition is entirely voltage-independent. ***P.001; NS; n 15.

8 2000 Compartmentalization of Neuronal Opioid Desensitization compared with the initial response). Facilitation of the DAMGO inhibited current by a strong depolarizing prepulse also was significantly reduced in desensitized DRG neurons (Fig. 4A 2 ). Facilitation on initial exposure to DAMGO was compared with after desensitization. The apparent loss of voltage-dependent inhibition after desensitization, as detected by the decrease in the facilitation ratio, was highly significant (P.001; Fig. 4B). Another estimate of the voltage-dependent component of neurotransmitter action on Ca 2 channels is the IR (Delmas et al., 1999). The IR (see Materials and Methods) may be a better estimate of the voltage-dependent component of opioid action on Ca 2 channels because, unlike the facilitation ratio, the IR does not depend on the degree of inhibition. DAMGO-mediated inhibition, on initial exposure to the agonist, was found to have a large voltage-dependent component ( %) with an IR of (Fig. 4C). After desensitization, however, the IR was significantly reduced ( ; P.001; n 15). Furthermore, the voltagedependent component was lost entirely in 8 of 15 cells (IR 1.00; Fig. 4D) and reduced to % in the remaining seven DRG neurons. Together these data demonstrate that the voltage-dependent component of DAMGO-mediated regulation of Ca 2 channels is lost or significantly attenuated with only brief exposure to the agonist. Because the G subunit is thought to be responsible for the voltage-dependent component of neurotransmitter-mediated inhibition of Ca 2 channels (Herlitze et al., 1996; Ikeda, 1996), these data strongly suggest that it is the interaction of the G subunit with the Ca 2 channel that is altered with continuous agonist exposure, resulting in an attenuated drug response. In a separate experiment, neurons exposed to 10 M -conotoxin GVIA did not exhibit facilitation of the DAMGOinhibited current after a depolarizing prepulse (data not shown). This observation suggests that -opioid regulation of non-n-type Ca 2 channels in neonatal rat DRG neurons lacks a significant voltage-dependent G -component. Short-Term DAMGO Desensitization Significantly Attenuates the Voltage-Sensitive Component of GABA B Regulation of Ca 2 Channels. To test whether short-term DAMGO desensitization affected the voltage-dependent component of GABA B -mediated regulation of high voltage-activated Ca 2 channels without the potential confounding influence of repeated baclofen exposure, an alternate desensitization protocol was used (see above; see Materials and Methods). The same prepulse facilitation protocol, as that described above, was used to examine whether the voltage-dependent component of GABA B regulation of Ca 2 channels was reduced after short-term DAMGO desensitization. Although a conditioning prepulse to 100 mv partially reversed baclofen-mediated inhibition of peak Ca 2 current magnitude in control neurons, inhibition in the presence of baclofen, both before and after a strong depolarizing prepulse, was significantly attenuated in DRG neurons pre-exposed to 3 M DAMGO for 15 min (Fig. 5A). The prepulse to 100 mv reduced the inhibition of peak current magnitude in control baclofen-sensitive neurons from to % (P.001; n 12; Fig. 5B). In contrast, baclofen-mediated inhibition was not reversed after a conditioning prepulse in neurons desensitized with DAMGO (P.14; n 7). Although % of the current inhibition in baclofensensitive cells was voltage-dependent (IR ), this component was significantly reduced (two of seven cells, IR ) or eliminated (five of seven cells, IR 1) in desensitized cells [IR (combined) ; P.005]. Fig. 5. Short-term DAMGO desensitization reduces the voltage-dependent relief of GABA B -mediated inhibition that is elicited by a strong depolarizing prepulse. A, superimposed whole-cell Ca 2 currents evoked by a 100-ms test pulse to 10 mv from V h 80 mv without (P1) or with (P2) a 95-ms depolarizing prepulse ( 100 mv) 5 ms preceding the test pulse. Traces show baclofen responses with and without conditioning prepulses in control DRG neurons (A 1 ), and in separate cells pre-exposed for 15 min to DAMGO (A 2 ). In this experiment, a modified protocol from that shown in Fig. 2 was used to establish short-term DAMGO desensitization. Before establishing the whole-cell patch configuration, a 10-min DAMGO (3 M) application was accomplished with the U-tube microperfusion system. At this time the cell was patched and Ca 2 currents were elicited every 30 s until the peak current magnitude was stable ( 5 min from patch rupture). Application of the DAMGO-containing solution was then stopped, the perfusion solution switched to one containing only the external Ca 2 buffer, and the DAMGO removed by continuous gravity-fed influx and vacuum-removed efflux of bath solution. Calcium currents were again elicited every 30 s until the peak current magnitude was stable (the assumption was that recovery from the DAMGO-mediated inhibition was complete). The effect of a brief application of baclofen (50 M conditioning prepulse) on peak current magnitude was then assessed. B, comparison of baclofen-mediated inhibition (mean S.E.) before ( ) and after (f) a strong depolarizing prepulse in control neurons (n 12) and in cells pre-exposed to DAMGO for 15 min (n 7). o indicates the mean responses expressed as the IR (see text). **P.01 (two-sample t test), ***P.001 (paired t test).

9 508 Samoriski and Gross Vol. 294 These data suggest that short-term DAMGO desensitization affects the voltage-dependent component of neurotransmitter regulation of Ca 2 channels, probably at the level of G -interaction with the Ca 2 channel. Together, these data provide electrophysiological and pharmacological evidence that desensitization of DAMGO-mediated regulation of non- N-type Ca 2 channels does not occur. Discussion This report demonstrates that there are two apparent components of -opioid-induced desensitization in neonatal rat DRG neurons, depending on the duration of agonist exposure, and that each likely involves different mechanisms. Prolonged exposure (24 h) to DAMGO resulted in a near complete loss of Ca 2 channel regulation by the -opioid agonist but did not affect the baclofen-mediated inhibition of targeted Ca 2 channels. In contrast, the early phase of desensitization developed rapidly, was maximal within 10 to 15 min of continuous DAMGO exposure, and began to reverse within minutes after removal of the agonist. During this period, not only was inhibition of targeted Ca 2 channels by DAMGO reduced by approximately 50% but also regulation of high voltage-activated Ca 2 channels by the GABA B agonist baclofen was significantly attenuated. Moreover, the voltage-dependent component of agonist-mediated inhibition was significantly reduced or entirely lost as a consequence of short-term DAMGO-induced desensitization. The differing features of the two components of -opioid desensitization suggest separate mechanisms, as yet incompletely described, leading to the development of these states. Activation of -opioid receptors results in the inhibition of Ca 2 current amplitude in agonist-sensitive DRG neurons by way of a membrane-delimited pathway that involves coupling of the receptor to the channel through a G o -type G protein. However, current inhibition is never complete, indicating that only a subset of the total channel population is targeted by the activated receptor. Previous studies have shown (Rusin and Moises, 1995), and our data confirm, that the majority of channels affected by -agonists are N-type (85 90% of targeted current), with lesser effects on non-ntype channels. Roughly two-thirds of the whole-cell current in rat neonatal DRG neurons is N type; however, only half of that, constituting approximately one-third of the cell Ca 2 current, is inhibited by DAMGO. Regulation of Ca 2 channels by DAMGO in DRG neurons exhibits in part the characteristic properties associated with voltage-dependent modulation after G protein activation. First, the inhibition of current amplitude is accompanied by a slowing of activation kinetics. Second, a strong depolarizing prepulse before the test depolarization restores the fast activation kinetics and partially reverses the current inhibition. G protein -subunits have been shown to cause neurotransmitter-like changes in channel voltage-dependent properties (Herlitze et al., 1996; Ikeda, 1996), presumably through an interaction with channel subunits (Bourinet et al., 1996; Ford et al., 1998; Sun and Dale, 1998) that readily reverses after strong depolarizing stimuli. In contrast, prepulse protocols yielded no facilitation of control currents, or of those currents remaining after block of N-type currents with -conotoxin GVIA. Together, these findings support the idea that the voltage-dependent component of G protein-mediated inhibition of Ca 2 currents in sensory neurons is selective for -conotoxin GVIA-sensitive N-type channels (Bean, 1989; Hille, 1994; but also see Luebke and Dunlap, 1994; Sun and Dale, 1997; Dolphin, 1998). The channel inhibition that remains after a strong depolarizing prepulse is presumably either due to a voltage-independent component or one that is less voltage sensitive to our prepulse protocol. We thus consider it likely that DAMGO inhibits N-type channels primarily via -subunits. The voltage-insensitive component of N- and non-n-type (presumably P/Q) channel inhibition is either mediated by -subunits for which the interaction with Ca 2 channels is less sensitive to voltage, or, perhaps, by -subunits (Jones and Elmslie, 1997; Delmas et al., 1998b). A novel aspect of our results, one not described in other desensitization paradigms, is that desensitization within a single cell varied according to the -opioid-dependent signaling pathway. Specifically, short-term DAMGO exposure eliminated or significantly attenuated the voltage-dependent component of -opioid regulation of Ca 2 channels. In contrast, control currents or DAMGO-reduced non-n-type Ca 2 currents showed no prepulse facilitation. We thus conclude that little tonic inhibition of Ca 2 channels by G proteins exists in our preparation and that DAMGO-induced reduction of non-n-type Ca 2 currents is mediated by voltageinsensitive G protein subunits. The corollary is that the voltage-sensitive -dependent pathway mediating Ca 2 current reduction is most susceptible to desensitization. Our results also show, however, that there is a component of N-type channel inhibition by DAMGO that does not reverse by depolarizing prepulses and that does desensitize. It is possible that there is a voltage-sensitive component of -mediated channel inhibition that we cannot fully assay with our prepulse protocol. However, we favor the alternative explanation that there are pathways of Ca 2 channel inhibition that are either non- -dependent (or nonvoltage-sensitive -dependent), but which are still under the influence of the desensitization process. Additional experiments will be required to address this more fully. Our results therefore support the view that opioid desensitization in DRG neurons occurs by at least two time-dependent processes and that the acute phase of desensitization is remarkable by virtue of the apparent existence of different compartments of desensitization. We hypothesize that long-term desensitization is mediated primarily by receptortargeting desensitization processes, such as G protein receptor kinases (Inglese et al., 1993; Chuang et al., 1996; Ferguson et al., 1996). This idea is supported by the longer time course required for the development of desensitization, the completeness of the loss of opioid signaling, and that the desensitization is homologous with respect to GABA B receptors. We further hypothesize that acute desensitization is primarily a process that targets signaling components downstream of the receptor, perhaps by influencing the interaction of -subunits with Ca 2 channels. In support of this idea is the finding that the voltage-dependent -pathway is completely desensitized and that opioid-induced desensitization also reduces signaling within a convergent pathway (GABA B ) that uses the same G protein subtype and that targets similar Ca 2 channel subtypes. One mechanism by which this may occur is by PKC-induced phosphorylation of the targeted Ca 2 channel. PKC has been shown to phosphorylate the 1 subunit of the Ca 2 at or near the region to which the