Methadone but not Morphine Inhibits Lubiprostone-Stimulated Cl− Currents in T84 Intestinal Cells and Recombinant Human ClC-2, but not CFTR Cl− Currents

In clinical trials, methadone, but not morphine, appeared to prevent beneficial effects of lubiprostone, a ClC-2 Cl− channel activator, on opioid-induced constipation. Effects of methadone and morphine on lubiprostone-stimulated Cl− currents were measured by short circuit current (Isc) across T84 cells. Whole cell patch clamp of human ClC-2 (hClC-2) stably expressed in HEK293 cells and in a high expression cell line (HEK293EBNA) as well as human CFTR (hCFTR) stably expressed in HEK293 cells was used to study methadone and morphine effects on recombinant hClC-2 and hCFTR Cl− currents. Methadone but not morphine inhibited lubiprostone-stimulated Isc in T84 cells with half-maximal inhibition at 100 nM. Naloxone did not affect lubiprostone stimulation or methadone inhibition of Isc. Lubiprostone-stimulated Cl− currents in hClC-2/HEK293 cells, but not forskolin/IBMX-stimulated Cl− currents in hCFTR/HEK293 cells, were inhibited by methadone, but not morphine. HEK293EBNA cells expressing hClC-2 showed time-dependent, voltage-activated, CdCl2-inhibited Cl− currents in the absence (control) and the presence of lubiprostone. Methadone, but not morphine, inhibited control and lubiprostone-stimulated hClC-2 Cl− currents with half-maximal inhibition at 100 and 200–230 nM, respectively. Forskolin/IBMX-stimulated hClC-2 Cl− currents were also inhibited by methadone. Myristoylated protein kinase inhibitor (a specific PKA inhibitor) inhibited forskolin/IBMX- but not lubiprostone-stimulated hClC-2 Cl− currents. Methadone caused greater inhibition of lubiprostone-stimulated currents added before patching (66.1 %) compared with after patching (28.7 %). Methadone caused inhibition of lubiprostone-stimulated Cl− currents in T84 cells and control; lubiprostone- and forskolin/IBMX-stimulated recombinant hClC-2 Cl− currents may be the basis for reduced efficacy of lubiprostone in methadone-treated patients.

Opioid receptors are present on the nerves of the gastrointestinal tract [11,12], but not all opioid effects on ion channels involve these receptors. Thus, L-type, but not T-type Ca 2? channels [13], several K ? channels [14,15] and the hERG K ? channel [16] have been shown to be differentially inhibited by methadone over morphine, in a manner independent of mu receptor occupation. Inhibition of hERG by methadone is problematic, as methadone treatment may lead to cardiotoxicity (long QT syndrome) [16,17]. Methadone inhibition of hERG has been suggested to be caused possibly by direct binding to the hERG channel, perhaps at the voltage sensor [16].
Although there have been no previous reports of inhibition of Cltransport in T84 cells as measured by Isc by any opioids, it appeared reasonable, in the face of the apparent failure of lubiprostone to ameliorate methadoneinduced constipation [3], and evidence for non-opioid receptor mechanisms of methadone inhibition of some ion channels [13][14][15][16], to test whether methadone and morphine had any effect on Isc in T84 cells.
ClC-2 is a time-dependent, voltage-activated Clchannel exhibiting inward rectification [8,[18][19][20][21] and is inhibited by CdCl 2 [18][19][20][21]. hClC-2 activation also occurs with forskolin/IBMX in a myristoylated PKI-sensitive manner [10] at two sites identified by site-directed mutagenesis [10]. The present studies of opioid effects on control hClC-2 Clcurrents and forskolin/IBMX activation of hClC-2 were also undertaken to determine whether opioid effects were limited to lubiprostone activation, or were rather a general effect on ClC-2. The present study of methadone and morphine effects on Clcurrents in T84 intestinal cells and on recombinant hClC-2 Clcurrents might explain the lack of effectiveness of lubiprostone in a clinical trial on opioid-induced constipation in patients on methadone, but not morphine therapy [3].

Materials and Methods
Human ClC-2-and hCFTR-transfected HEK293 cells, T84 cells, culture conditions, patch clamp, and Isc methods were as described in [4].

Short-circuit Current Measurements
The EasyMount Ussing chamber system (8 chambers) with VCCMC8 8-channel current-voltage (I-V) clamps from Physiologic Instruments (San Diego, CA) was used for Isc measurements across confluent T84 cell monolayers as previously described [4]. Transepithelial resistance of T84 cells was monitored with an EVOM epithelial volt ohm meter (World Precision Instruments). Cells were used when the transepithelial resistance of the monolayer was [1,200 X. Solutions were continuously gassed with 95 % O 2 -5 % CO 2 , also providing stirring, and the temperature was held constant at 37°C with a heating block. The clamps were connected to Acquire & Analyze software (Physiologic Instruments) for automatic data collection from all eight of the Ussing chambers. Ag/AgCl reference electrodes were used for measuring transepithelial voltage and passing current. The basolateral membrane bath solution contained (in mM) 120 NaCl, 25 NaHCO 3 , 3.3 KH 2 PO 4 , 0.8 K 2 HPO 4 , 1.2 MgCl 2 , 1.2 CaCl 2 (pH 7.4), and 10 mM glucose. The apical membrane bath solution was identical, except that the Clconcentration was reduced by substituting sodium gluconate for NaCl and CaCl 2 was increased to 4 mM as previously described [4,28] because of Ca 2? chelation by gluconate. Free [Ca 2? ] of the gluconate medium was calculated to be 1.2 mM using the Cabuf program (ftp:// ftp.cc.kuleuven.ac.be/pub/droogmans/cabuf.zip) as was used previously [29]. 10 mM mannitol was used instead of glucose to ensure the absence of any Na ? -glucose cotransport. To remove constraints on apical membrane Clcurrents, 300 lM 1-EBIO, a Ca 2? -activated K channel activator [30] was added to the basolateral bath solution and allowed to equilibrate.

Patch Clamp Measurement of Whole Cell Cl -Currents
Patch clamp and analytic methods were described previously [4]. Two voltage-clamp pulse protocols were used. For hClC-2-transfected and hCFTR-transfected HEK293 cells, currents were elicited by voltage-clamp pulses between -140 and ?40 in 20-mV increments from a holding potential of -30 mV, and 200 ms recordings were made. For hClC-2-transfected and mock-transfected HEK293EBNA cells, currents were elicited by voltageclamp pulses between -160 and ?40 mV in 20-mV increments from a holding potential of -30 mV, and 1,500 ms recordings were made. For both protocols, current values were taken at 200 ms. The bath (external) solution contained (in mM) 140 tetraethylammonium Cl, 1 MgCl 2 , 2 CaCl 2 , and 10 HEPES (pH 7.4). The pipette (internal) solution contained (in mM) 115 tetraethylammonium Cl, 2 MgCl 2 , 5 EGTA, and 10 HEPES (pH 7.4). Pipettes were prepared from borosilicate glass and pulled by a two-stage Narishige puller to produce 1-1.5-MX resistance. Data were acquired with an Axopatch CV-4 headstage, a Digidata 1200 digitizer, and an Axopatch 1D amplifier. Data were analyzed using pClamp 6.04 (Axon Instruments, Union City, CA), Microsoft Excel, and Origin software (OriginLab, Northampton, MA). Clcurrents were all measured at 200 ms and normalized to capacitance.

Statistics
Statistical significance between two means was calculated using the Student's t test. Significance was at P \ 0.05 or less. In Fig. 3b, using Origin 5.0 Professional, the data were fit using a modified Michaelis-Menten hyperbolic function as previously described [4]. The equation used was DI = DI max 9 [lubi]/(EC 50 ? [lubi]), where DI max is the maximum change in I, and EC 50 is [lubi] required for half-maximal response. As change in I was measured, the analysis was constrained to 0. All other graphs were plotted as mean ± SEMs joined by lines. Half-maximal inhibitory concentrations for methadone were estimated from the values at the control (no methadone) minus the values at the maximal concentration of methadone divided by two.

Effect of Selected Concentrations of Methadone and Morphine on Lubiprostone-stimulated Cl -Currents in T84 Cells and Effect of Naloxone
The effects of methadone and morphine on lubiprostonestimulated Isc in T84 cells were determined. The results are shown in Fig. 1. Prior addition of 5 lM morphine had no effect on 250 nM lubiprostone-stimulated Isc, but prior addition of 5 lM methadone caused major (83.1 %) inhibition of lubiprostone-stimulated Isc (Fig. 1a). The effect of selected concentrations of methadone on lubiprostonestimulated Isc is shown in Fig. 1b, and the methadone and morphine concentration response curves are shown in Fig. 1c. Morphine had no effect at any concentration tested. However, methadone inhibited the lubiprostone-stimulated Isc in a concentration-dependent manner with half-maximal inhibition of Isc at 100 nM.
No evidence for mu receptors on intestinal cells was found by several investigators [11,12,31,32], but in one study, evidence was found for mu receptors on human colonocytes [33]. Therefore, the effect of the nonspecific opioid receptor antagonist, naloxone, on methadone inhibition of Clcurrents (Isc) in T84 cells was examined. As shown in Fig. 1d, 250 nM lubiprostone-stimulated Isc across T84 cells. Prior addition of 10 lM naloxone alone had no effect, while prior addition of 1 lM methadone was inhibitory. Addition of 10 lM naloxone had no effect on methadone inhibition of lubiprostone-stimulated Isc.
Effects of Methadone and Morphine on Lubiprostonestimulated and Forskolin/IBMX-stimulated Cl -Currents in hClC-2-expressing HEK293 Cells The effects of methadone and morphine, together and separately, on lubiprostone-stimulated hClC-2 Clcurrents were next determined. Clcurrents in hClC-2-transfected HEK293 cells were measured in cells without lubiprostone (control), cells treated with 100 nM lubiprostone, and with either 1 lM methadone followed by 100 nM morphine or the reverse 100 nM morphine followed by 1 lM methadone. As shown in Fig. 2a, 100 nM lubiprostone-stimulated hClC-2 Clcurrents and 100 nM morphine had no effect on the currents. Subsequent addition of 1 lM methadone inhibited lubiprostone-stimulated Clcurrents. Also as shown in Fig. 2a, addition of 1 lM methadone inhibited lubiprostone-stimulated Clcurrents and subsequent addition of 100 nM morphine had no effect. The effects of 100 nM morphine and 1 lM methadone were also studied on hClC-2 Clcurrents activated by 5 lM forskolin/20 lM IBMX, an alternative means of activation of hClC-2 blocked by mPKI [4,10]. As shown in Fig. 2b, forskolin/IBMX activated hClC-2 Clcurrents, and these currents were not affected by morphine, but inhibited by methadone, whether added after or before morphine. Thus, methadone appears to inhibit hClC-2, regardless of the method of activation, whether by lubiprostone or forskolin/ IBMX. Methadone, but not morphine, whether added together or separately, inhibited both lubiprostone-and forskolin/IBMX-stimulated Clcurrents in hClC-2-transfected HEK293 cells. Methadone thus apparently inhibits the hClC-2 Clchannel by a mechanism independent of lubiprostone, per se.
Effects of Methadone and Morphine on Forskolin/ IBMX-stimulated hCFTR Cl -Currents CFTR is also found in the intestine, and in one study was not activated by lubiprostone [4]. In other studies, CFTR appeared to be activated as well [5,6,8]. Therefore, the c Effects of selected concentrations of methadone and morphine on lubiprostone-stimulated Isc are shown. Isc values plotted were taken at 16 min after start of the experiment, and 12 min after addition of lubiprostone. Mean ± SEM are plotted; n is indicated. d T84 cells were mounted in an Ussing chamber under short circuit conditions, and first treated with 300 lM 1-EBIO and then with nothing, followed by 10 lM naloxone, 1 lM methadone, or 10 lM naloxone plus 1 lM methadone. 250 nM lubiprostone was then added to all, and Isc was measured. Data are plotted as mean ± SEM and n is indicated. *P \ 0.0005 versus lubi ± naloxone effects of 100 nM morphine and 1 lM methadone were also studied on Clcurrents in HEK293 cells expressing recombinant hCFTR [4] after activation by 5 lM forskolin/ 20 lM IBMX. As shown in Fig. 2c, hCFTR Clchannel activity activated by forskolin/IBMX was not inhibited by methadone or morphine.
Effect of Lubiprostone on Cl -Currents in hClC-2transfected HEK293EBNA Cells As the previous published experiments with lubiprostone [4] were carried out with hClC-2 in HEK293 cells where hClC-2 expression is low, before examining methadone and morphine effects on lubiprostone-activated hClC-2 Clcurrents when expressed in HEK293EBNA cells, the effect of lubiprostone at selected concentrations was first examined, and the EC 50 was calculated. The cells were washed with three changes of medium over about 3-4 min, in between different concentrations of lubiprostone as described previously [4]. These washes were sufficient to completely wash out the lubiprostone and return the Clcurrent to control levels [4]. The results are shown in Fig. 3b, plotted as current at 200 ms and -140 mV for direct comparison with previously published experiments [4]. Lubiprostone-activated hClC-2 Clcurrents when expressed in HEK293EBNA cells in a concentrationdependent manner. The data were fit with a modified Michaelis-Menton equation, and the EC 50 was 28.2 ± 2.2 nM (4), not significantly different from the EC 50 measured for hClC-2 in HEK293 cells [4]. Also shown in the inset is the Hill plot of the data. The Hill coefficient was 0.91 ± 0.02 (4), R = 0.96 ± 0.09 (8), and P \ 0.0005. These data are not significantly different from those previously reported [4]. Control-and lubiprostonestimulated hClC-2 Clcurrents in HEK293EBNA cells were about -100 and -450 pA/pF, respectively, approximately fourfold higher than control and lubiprostone-  Fig. 4b, 20 nM lubiprostone increased hClC-2 Clcurrents, and they remained time dependent and voltage activated. Subsequent addition of methadone decreased the lubiprostone-stimulated hClC-2 Clcurrents. In contrast, Clcurrents measured in mock-transfected HEK293EBNA cells were very low (-40.3 ± 7.6 (3) pA/ pF), significantly different (P \ 0.02) from those measured in hClC-2-transfected HEK293EBNA cells, and 20 nM lubiprostone followed by 1 lM methadone had no effect. The corresponding I/V curves for hClC-2-transfected and mocktransfected HEK293EBNA cells are also shown. Both control (without lubiprostone)-and lubiprostone-stimulated hClC-2 Clcurrents in hClC-2-transfected HEK293EBNA cells showed inward rectification, while mock-transfected cells had linear I/V curves, exhibiting very small currents. Methadone caused significant decreases (P \ 0.05) in the hClC-2 Clcurrent resulting in 46.2 % inhibition at -140 mV.

Discussion
Lubiprostone is very effective in treating opioid-induced constipation induced by morphine and congeners [1][2][3]. The inhibitory effect of methadone on lubiprostone-mediated relief from constipation [3] may arise from methadone inhibition of ClC-2 Clcurrents. The present study was designed to test the hypothesis whether methadone, but not morphine, might inhibit Cltransport by epithelial cells. Lubiprostone-stimulated Clcurrents measured by Isc in T84 cells and control and lubiprostone-stimulated hClC-2 Clcurrents were inhibited by methadone, but not by morphine. The half-maximal concentration for methadone inhibition of Clcurrents measured by Isc was 100 nM, approximately 18 times the affinity of methadone for mu receptors [34]. Naloxone alone, or with methadone, had no effect. Morphine, even at 5 lM (2,500 times higher concentration than its affinity for mu receptors) [34], had no effect. Therefore, inhibition by methadone of T84 cell lubiprostone-stimulated Isc appears to be consistent with the reduced effect of lubiprostone on methadone-induced constipation [3]. Methadone and morphine both bind to mu receptors on target cells [34,35], although they belong to two different classes of opioids distinct in chemical structure (diphenylheptanes vs. phenanthrenes) and metabolic pathways. In the intestine, mu receptors have been localized largely to nerve terminals and synaptic elements in all intestinal layers [11,12]. Although human colonocytes have been suggested to have mu receptors [33], no evidence for mu receptors on intestinal epithelial cells was found by others [11,12,31,32]. Lubiprostone stimulates Cltransport across epithelial cells [4][5][6]. This raised questions regarding the mechanism whereby the beneficial lubiprostone effects could be affected by methadone, but not morphine in the clinical situation. The reported finding of methadone, but not morphine attenuating/preventing the beneficial effects of lubiprostone in the clinical trial [3] was difficult to relate to mu receptors, as both agents bind mu receptors with similar affinity (EC 50 ): 5.6 nM for methadone and 2.0 nM for morphine, tested with the cloned human mu receptor [34]. Thus, methadone might interfere with lubiprostone action by a mechanism independent of mu receptor binding.
The finding of methadone inhibition of T84 cell lubiprostone-stimulated Clcurrents measured by Isc was unexpected, as there are no reports of effects of methadone on Clcurrents in the literature. If mu receptors were not involved in the process of inhibition by methadone, as suggested by the lack of naloxone effect, then some other target required for Clcurrent activation might be affected. Direct action on ion channels or a process required for activation of ion channel function might be responsible for methadone inhibition. Methadone, but not morphine, inhibits L-type calcium channels [13], various potassium channels [14,15], and hERG [16,17]. Direct inhibition of hERG might underlie cardiotoxicity seen with methadone treatment. In the latter case, methadone binds largely to the inactivated/open channel, and the inhibitory effect occurs within 10 ms. Methadone has been suggested to act through binding to the voltage sensor of hERG [16,17]. These findings provided a basis for examination of whether methadone affected the function of either recombinant hClC-2 or hCFTR.
Both hClC-2 and hCFTR ion channels have been suggested to be involved in lubiprostone stimulation of T84 cell Isc [4][5][6], but have not been directly identified. A similar concentration dependence for lubiprostone stimulation of recombinant hClC-2 and T84 cell Isc was reported, while no stimulation of hCFTR was observed at concentrations as high as 1 lM lubiprostone [4]. In singlechannel studies, A6 cell ClC-2 as well as hClC-2 expressed in HEK293 cells were activated by lubiprostone at low concentrations (\100 nM), while CFTR was also activated by lubiprostone, but at concentrations 50 times higher than the concentration necessary to activate ClC-2 [8]. Knockdown of T84 cell ClC-2 ablated lubiprostone stimulation of T84 cell Clcurrents [7]. Based on T84 cell Isc studies [5,6] and mouse intestine Isc studies [5] in the presence and absence of CFTRinh172, ClC-2 has been suggested to not be involved in lubiprostone stimulation, but rather CFTR. In those studies, lubiprostone stimulation of Isc was inhibited by CFTRinh172, but it is unknown whether CFTRinh172 also inhibits ClC-2. CFTRinh172 did not inhibit Ca 2? -activated Clcurrents in human airway cells or volume-activated Clcurrents in Fischer rat thyroid cells [36] and these were the only Clchannels tested. Until CFTRinh172 has been shown to not inhibit ClC-2, it is difficult to evaluate whether ClC-2 or CFTR was being activated by lubiprostone in those studies.
Studies of effects of methadone on hClC-2 and hCFTR were undertaken. Previously established HEK293 cell lines stably transfected with hClC-2 or hCFTR [4], and a newly developed cell line, HEK293EBNA transfected with hClC-2 were used for studies of lubiprostone stimulation of hClC-2 and the effects of opioids. The newly developed cell line overexpressed hClC-2, leading to higher control Clcurrents (approximately -100 pA/pF compared to approximately -25 pA/pF), allowing studies of the effects of lubiprostone and opioids on the time-dependent, voltage-activated hClC-2 Clcurrents, infrequently exhibited by our previously used hClC-2 in HEK293 cell line [4,9,10]. The reason(s) for this infrequent time dependence and voltage activation is(are) not known. However, hClC-2 Clcurrents in parental human cystic fibrosis airway, IB3-1, cells were shown to lack time dependence and voltage activation, but subsequent overexpression of hClC-2 in IB3-1 cells resulted in time-dependent, voltage-activated Clcurrents [19].
Lubiprostone-stimulated hClC-2 Clcurrents were inhibited by methadone, but not morphine. Addition of methadone before or after patching altered the extent, but not the concentration of methadone giving half-maximal inhibition of lubiprostone-stimulated hClC-2 Clcurrents. Methadone inhibition was greater when added before patching, suggesting that methadone does not bind to activated or open hClC-2 Clchannels. Determination of the basis for this effect is beyond the scope of the present study, and this effect probably has no therapeutic implications as methadone is maintained throughout treatment. Methadone inhibition also occurred in the presence of morphine. Methadone was found to be effective in inhibiting lubiprostone and forskolin/IBMX-stimulated hClC-2 Clcurrents whether morphine was also present or not. The lack of effect of methadone on Clcurrents in hCFTR-transfected HEK293 cells suggested inhibitory methadone effects were related to hClC-2 or the processes involved in the activation of lubiprostone-stimulated hClC-2 Clcurrents.
Mu receptors do not appear responsible for methadone inhibition of recombinant hClC-2 or T84 Clcurrents based on the high concentrations of methadone required for inhibition compared with its affinity for mu receptors, and the lack of effect of morphine. HEK293 cells appear unlikely to have mu receptors as judged by lack of specific mu receptor binding and lack of mu receptor protein detected by immunoblot [37] and have been widely used to study recombinant mu opioid receptors [38,39]. HEK293 cells expressing recombinant ion channels are therefore suitable for studies of effects of methadone and morphine. Intestinal epithelia have been variously reported to lack opiate receptors [11,12,31,32], or have mu receptors on human colonocytes [33]. There are no reports on whether T84 cells themselves have mu receptors. However, there was no effect of naloxone [13] on methadone inhibition of lubiprostone-stimulated Clcurrents in T84 cells.
Using a newly developed cell line (HEK293EBNA transfected with hClC-2) exhibiting time-dependent, voltage-activated Clcurrents, methadone, but not morphine, inhibited these hClC-2 Clcurrents. Thus, methadone does not act by competition with lubiprostone, suggesting possible interaction between methadone and hClC-2 (or a closely related process required for activation of hClC-2 dependent Clcurrents). Further studies will be required to determine if hClC-2 interacts directly with methadone.
Inhibitory effects of methadone on forskolin/IBMX activated Clcurrents were also observed in hClC-2expressing HEK293EBNA cells. This effect was prevented by the PKA-specific inhibitor, mPKI. Thus, forskolin/ IBMX stimulation appeared to be through PKA activation of hClC-2, as previously demonstrated in functional and site-directed mutagenesis studies [4,10]. Lubiprostonestimulated hClC-2 Clcurrents were unaffected by mPKI. Thus, inhibition of hClC-2 by methadone does not appear to be through competition with lubiprostone, consistent with the possibility of direct binding of methadone to hClC-2. Alternatively, methadone might interfere with a process blocking Clcurrents in general (without directly affecting the ClC-2 channel protein). However, methadone was without effect on hCFTR Clcurrents. Therefore, methadone is apparently not affecting a process important to Cltransport, per se, but might interact with either ClC-2 or a process required for transport of Clby ClC-2.
Single-channel studies of hClC-2 have been published using the same HEK293 cells transfected with hClC-2 as used in Fig. 2 [8]. In Fig. 17 of this published article, the authors state referring to HEK293 cells expressing hClC-2: ''the anion channel activated by lubiprostone had channel kinetics and a current-voltage relationship that were essentially indistinguishable from the channels in A6 cells we had identified as ClC-2 channels.'' It is not clear whether single-channel studies of methadone effects on hClC-2 would lead to further understanding of the mechanism.