Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Since 1956, it has been reported that insulin pretreatment enhanced morphine-induced antinociceptive response (Davis et al., 1956). Singh et al. (1983) further suggested that this enhancement obtained in hypoglycemic animals is due to a lower level of intracellular ATP. After ATP-sensitive K⁺ (K_ATP) channels were explored in the brain (Bernardi et al., 1988), several in vivo studies suggested that K_ATP channels are involved in morphine-induced supraspinal analgesia: glibenclamide, a K_ATP channel blocker, when administered i.c.v., dose dependently decreased the antinociceptive response of morphine administered either intravenously (Ocaña et al., 1990; Roane and Boyd, 1993) or i.c.v. (Narita et al., 1992). Other K_ATP channel blockers but not the nonselective K⁺ channel blocker tetraethylammonium or 4-aminopyridine (Ocaña et al., 1993, 1995) also decreased the antinociceptive response of morphine. The potencies of K_ATP channel blockers in reversing morphine analgesia were correlated with their blocking activities of K_ATP channels (Ocaña et al., 1993). On the other hand, K_ATP channel openers potentiated the antinociceptive response of morphine (Vergoni et al., 1992; Narita et al., 1993; Lohmann and Welch, 1999). In addition, glibenclamide (i.c.v.) also decreased antinociceptive responses induced by other opiates, including levorphanol, methadone, and buprenorphine (Ocaña et al., 1995; Raffa and Martinez, 1995).

Nevertheless, the correlation between activation of K_ATP channels and morphine cellular actions in pain-related brain regions has not been studied in vitro. Given that the ventrolateral periaqueductal gray (PAG) is a crucial site for morphine-induced supraspinal analgesia (Yaksh et al., 1976), the present study was designed to test the hypothesis that effects of morphine in the ventrolateral PAG are mediated by activation of K_ATP channels. The cellular mechanism in the ventrolateral PAG for morphine-induced supraspinal analgesia is attributed mostly to inhibition of inhibitory synaptic transmission and partly to membrane hyperpolarization resulting from activation of inwardly rectifying K⁺ channels (Osborne et al., 1996; Vaughan et al., 1997; Chiou and Huang, 1999). Therefore, effects of morphine in the ventrolateral PAG were assessed by these two actions. The possible functional role(s) of K_ATP channels in the ventrolateral PAG were also explored by investigating effects of K_ATP ligands on synaptic transmissions and membrane currents.

It has been shown that the density of K_ATP channels is higher in adult rats compared with neonates (Mourre et al., 1990; Xia and Haddad, 1991). In addition to neonatal slices that are better preparations for blind patch-clamp recordings (Plant et al., 1995), experiments were also conducted in adult slices to see whether greater effects on morphine actions are observed when K_ATP channels are blocked. Given that the antinociceptive response induced by morphine is age-dependent (Auguy-Valette et al., 1978; Windh and Kuhn, 1995), the present study also examined whether the cellular actions of morphine in the ventrolateral PAG are dependent on age.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Brain Slice Preparations. Coronal brain slices (400 µm) containing the PAG were dissected as described previously (Chiou and Chou, 2000) from Wistar rats at the age of 12 to 16 days (neonates) or 8 to 12 weeks (adults). In some experiments, transverse hippocampal slices of 400 µm from adult rats were used. After equilibrated in the artificial cerebral spinal fluid (ACSF) for at least 1 h, slices were transferred to a submerged chamber and perfused with the ACSF at 2 to 3 ml min¹. The ACSF contained 117 mM NaCl, 4.5 mM KCl, 2.5 mM CaCl₂, 1.2 mM MgCl₂, 1.2 mM NaH₂PO₄, 25 mM NaHCO₃, and 11.4 mM dextrose and was oxygenated with 95% O₂ plus 5% CO₂ (pH 7.4).

Electrophysiological Recordings. Blind patch-clamp recording was conducted in ventrolateral neurons of PAG slices or in CA1 pyramidal neurons of hippocampal slices at 30°C. The membrane input resistance was monitored with the MEMBRANE TEST function of Clampex 7.0 (Axon Instruments, Foster City, CA) by applying small hyperpolarization pulses (3 mV). Membrane currents were recorded with an Axopatch 200B amplifier (Axon Instruments), digitized, and acquired with an A/D converter (DigiData 1200A; Axon Instruments) in a Pentium III PC computer running pClamp 7 (Axon Instruments).

Data Analysis and Chemicals. The effect of drugs was measured at the steady state of treatments as reported previously (Chiou, 2001). The shifting of holding current (I_hold) was expressed as positive for an outward shifting and negative for an inward one. The membrane current at 140 mV (I₁₄₀) was normalized by reference to the control current at 140 mV for each neuron. Data were expressed as means ± S.E.M. with n indicating the number of neurons recorded, each of which was taken from one slice from different rats. From each rat, usually two to three slices at caudal segment of the PAG were used. For a comparison, sequential treatments were applied in the same neuron, if possible. Student's t test was used for statistical analysis.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

PAG Neurons from Neonatal Rats

First, we performed experiments in PAG slices isolated from neonatal rats, which are better preparations for a successful whole cell recording than those from adult rats (Plant et al., 1995).

Glibenclamide Had No Effect on Evoked Synaptic Currents in Neonatal Neurons. Glibenclamide, a K_ATP channel blocker, at concentrations of 1 to 10 µM, affected neither EPSCs nor IPSCs (Fig. 1). The amplitude of EPSCs in the presence of 10 µM glibenclamide was 97 ± 4% of the control (n = 8) and that of IPSCs was 99 ± 3% of the control (n = 7). The holding currents (dotted lines in Fig. 1) recorded with microelectrodes containing ATP-free internal solution were also not affected by glibenclamide. They were 60.2 ± 6.6 versus 55.9 ± 5.6 pA (n = 8) and 2.7 ± 1.9 versus 1.9 ± 1.4 pA (n = 7), respectively, when EPSCs and IPSCs were recorded.

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Lack of effect of glibenclamide on evoked synaptic currents in neonatal ventrolateral PAG neurons. IPSCs (A) and EPSCs (B) were evoked at 0.03 Hz by focal stimulation in the absence (left) or presence of 10 µM glibenclamide (right). IPSCs were recorded at 63 mV in the presence of 1 mM kynurenic acid. EPSCs were recorded at 73 mV. Dotted line indicates the holding current of the control. ATP-free internal solution was used.

Glibenclamide Did Not Affect Morphine-Induced Inhibition of IPSCs. Morphine concentration dependently decreased the amplitude of IPSCs that were recorded in the presence of 1 mM kynurenic acid (Fig. 2), an ionotropic glutamate receptor blocker. The magnitude of inhibition by 50 µM morphine was 44 ± 6% (n = 5). Further addition of 10 µM glibenclamide did not affect the inhibition induced by morphine (Fig. 2). Increasing the concentration of glibenclamide or decreasing that of morphine also did not display any antagonism (Fig. 2).

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Lack of effect of glibenclamide on morphine-induced inhibition of IPSCs in neonatal PAG neurons. IPSCs were recorded at 15 mV in the presence of 1 mM kynurenic acid. A, IPSCs recorded from a neuron before and after treatment with 50 µM morphine and further with 10 µM glibenclamide. B, normalized IPSCs in the presence of morphine, alone (), or in combination with glibenclamide ( and ). Figures in the parentheses indicate the number of neurons. Cs+-gluconate internal solution was used. *p < 0.05 versus control.

Glibenclamide Did Not Affect Morphine Activation of Inwardly Rectifying K⁺ Channels. In 7 of 27 neurons tested, morphine increased the membrane current elicited by a hyperpolarization ramp from 60 to 140 mV and shifted the holding current outwardly (Fig. 3). The current increased by morphine showed inward rectification and had a reversal potential at 86 mV, which is close to the equilibrium potential of K⁺ ions, being 91 mV, according to the Nernst equation. Therefore, the current induced by morphine is an inwardly rectifying K⁺ current. The current at 140 mV was increased by 50 µM morphine to 135 ± 12% of the control (n = 7) (Fig. 4). In morphine-sensitive neurons, further treatment with 10 µM glibenclamide did not show any further change on the membrane current (Figs. 3 and 4). However, naloxone (1 µM) significantly antagonized the effect of morphine (Fig. 3).

View larger version (54K): [in this window] [in a new window]   Fig. 3.   Antagonism by naloxone but not glibenclamide of the morphine activation of inwardly rectifying K+ channels in neonatal PAG neurons. Membrane currents were elicited by hyperpolarization ramps (B, inset) from 60 to 140 mV for 400 ms every 30 s. Holding potential was 70 mV. A, chart recording of membrane currents elicited by voltage ramps from a neuron treated first with 50 µM morphine then with 10 µM glibenclamide and further with 1 µM naloxone. B, I-V curves of membrane currents in the absence or presence of morphine or morphine plus glibenclamide (left) and of morphine-induced currents in the absence or presence of glibenclamide (right). Note that the morphine-induced current, obtained by subtracting the control current from that in the presence of morphine, has a reversal potential near the equilibrium potential of K+ ion and shows an inward rectification. Normal K+-gluconate internal solution was used.

View larger version (29K): [in this window] [in a new window]   Fig. 4.   Statistic analysis of interactions of glibenclamide with morphine, lemakalim, and diazoxide on voltage ramp-elicited membrane currents in neonatal PAG neurons. Membrane currents were elicited by the same protocol as in Fig. 3. Shown are effects of morphine (50 µM), lemakalim (10-30 µM), and diazoxide (100-500 µM) in the absence () or presence of 10 µM glibenclamide (). A, changes of holding currents (Ihold) were denoted as positive for an outward shifting and as negative for an inward shifting. B, membrane currents at 140 mV (I140) were expressed as the percentage of control current at 140 mV. Insets, I-V curves of membrane currents under control conditions or during exposure to 30 µM lemakalim (left) or 500 µM diazoxide (right), alone or in combination with 10 µM glibenclamide. Figures in parentheses are the number of neurons recorded. For each group of columns, sequential treatments were applied in the same neuron, if possible. *p < 0.05 versus control.

K_ATP Channel Openers Had No Effect on Membrane Current. The negative result of glibenclamide on the membrane current might be due to that the K_ATP channels, if any, have been blocked by intracellular ATP in our recording condition. The effects of K_ATP channel openers were, therefore, further investigated. Neither lemakalim (10-30 µM) nor diazoxide (100-500 µM) affected the membrane current elicited by a hyperpolarization ramp or the holding current (Fig. 4, open columns). Further addition of glibenclamide, 10 µM, also did not produce any effect (Fig. 4, filled columns).

Effect of ATP-Free Internal Solution on Holding Current. Further experiments were conducted with microelectrodes containing ATP-free internal solution to see whether K_ATP channels could be disclosed by washing out the intracellular ATP. Figure 5 shows that, after whole cell configuration was formed, the holding current recorded with ATP-free internal solution was shifted inwardly but not outwardly.

View larger version (22K): [in this window] [in a new window]   Fig. 5.   Changes of holding currents after dialyzed with normal or ATP-free internal solution in neonatal PAG neurons. Changes of holding current (Ihold) after formation of whole cell configuration were recorded with microelectrodes filled with K+-gluconate internal solution with 5 mM ATP (, n = 5) or without ATP (, n = 5). The abscissa is the time after whole cell configuration was obtained.

PAG Neurons from Adult Rats

Immunohistochemical studies have shown that the density of K_ATP channels in neonates is lower compared with adults (Mourre et al., 1990; Xia and Haddad, 1991). To examine whether the ineffectiveness of K_ATP ligands is attributed to the low density of K_ATP channels in neonatal PAG neurons, further experiments were performed in slices isolated from adult rats.

Neither K_ATP Channel Ligands nor Morphine Affected Membrane Currents of Adult PAG Neurons. In adult ventrolateral PAG neurons, diazoxide at the concentration up to 500 µM, alone or in combination with glibenclamide, affected neither the membrane current elicited by a hyperpolarization ramp nor the holding current (Fig. 6). Interestingly, morphine, at the concentration of 50 µM, which induced membrane hyperpolarization in one-third of neonatal PAG neurons tested, failed to affect membrane currents in all recorded neurons (n = 9). When the concentration of morphine was increased to 100 µM in seven neurons or to 300 µM in nine neurons, membrane currents were still not affected (Fig. 6). No effect was observed by further addition of glibenclamide. However, in these morphine-resistant adult neurons, baclofen, a -aminobutyric acid_B receptor agonist, increased the membrane current reversibly (Figs. 6 and 7). The current increased by baclofen reversed at the potential close to the equilibrium potential of K⁺ ions and was characterized with inward rectification (Fig. 7).

View larger version (23K): [in this window] [in a new window]   Fig. 6.   Statistic analysis of the effect of morphine, diazoxide, or baclofen, alone or in combination with glibenclamide, on membrane currents of adult neurons. Membrane currents were elicited and expressed as in Fig. 4. Doublet bars represent neurons that were treated with morphine (50-300 µM) or 500 µM diazoxide in the absence () or presence of 10 µM glibenclamide (). Triplet bars represent neurons that were treated with 50 µM morphine (), 3 µM baclofen (), or washout (). *p < 0.05 versus control.

View larger version (45K): [in this window] [in a new window]   Fig. 7.   Activation of inwardly rectifying K+ channels by baclofen but not morphine in adult PAG neurons. Membrane currents were elicited by the same protocol as in Fig. 3. A, chart recording of the membrane current from a neuron that is insensitive to 300 µM morphine but responsive to 3 µM baclofen. B, I-V curves of membrane currents (left) before and after treatment with 300 µM morphine or 3 µM baclofen, and of the baclofen-induced current (right), which was obtained by subtracting the control current from that in the presence of baclofen.

Glibenclamide Did Not Affect Morphine Inhibition of IPSCs in Adult Neurons. To see whether the sensitivity to morphine is generally lower in adult PAG neurons, the effect of morphine on IPSCs of adult PAG neurons was examined. In all of the five neurons tested, IPSCs recorded in the presence of 1 mM kynurenic acid were depressed by 50 µM morphine (Fig. 8). The magnitude of depression is 41 ± 5% (n = 5), which is comparable with the magnitude of 44 ± 6% (n = 5) obtained in neonatal slices (Fig. 2). Further addition of 10 µM glibenclamide did not gain further change in IPSCs (Fig. 8).

View larger version (15K): [in this window] [in a new window]   Fig. 8.   Lack of effect of glibenclamide on morphine-induced inhibition of IPSCs in adult PAG neurons. IPSCs were recorded at 50 mV in the presence of 1 mM kynurenic acid. A, IPSCs were recorded from an adult neuron before and after treatment with 50 µM morphine and further with 10 µM glibenclamide. B, normalized IPSCs in the presence of morphine () or morphine plus glibenclamide (). Normal K+-gluconate internal solution was used. *p < 0.05 versus control.

Hippocampal Neurons from Adult Rats

K_ATP Channel Ligands Did Affect Membrane Current of Hippocampal CA1 Neurons. To see whether the K_ATP channel ligands used in our recording system are valid, we examined their effects on hippocampal CA1 neurons where K_ATP channel openers can induce membrane hyperpolarization (Fujimura et al., 1997). In contrast to the negative finding in PAG neurons, diazoxide (500 µM) induced an outward current, which was reversed by glibenclamide (10 µM) in hippocampal CA1 neurons (Fig. 9).

View larger version (7K): [in this window] [in a new window]   Fig. 9.   Antagonism by glibenclamide of diazoxide-induced outward current in a hippocampal CA1 neuron. The holding current was recorded at 70 mV in a hippocampal neuron treated with 500 µM diazoxide followed by further treatment with 10 µM glibenclamide. Note that diazoxide elicited an outward current that was antagonized by glibenclamide.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In ventrolateral PAG neurons, we demonstrated the following: 1) In neonates, morphine depressed the inhibitory synaptic transmission and activated inwardly rectifying K⁺ channels. 2) In adults, morphine failed to activate K⁺ channels in all recorded neurons but produced an inhibition of IPSC comparable with that in neonates. 3) Glibenclamide, a K_ATP channel blocker, did not alter morphine effects in both neonates and adults. 4) Neither openers nor blockers of K_ATP channels affected the synaptic transmission or membrane current in both neonates and adults.

Little Functional Role of K_ATP Channels in Ventrolateral PAG. K_ATP channels are inhibited by intracellular ATP at the concentration as low as 100 µM (Ashcroft and Kakei, 1989), which is far lower than the intracellular concentration of ATP in our recording condition. While this might explain why glibenclamide did not affect the membrane current per se, the openers of K_ATP channels, either lemakalim or diazoxide, also have no effect on the membrane current of ventrolateral PAG neurons. Although lemakalim was less effective in neuronal K_ATP channels (Stanford and Lacey, 1996; Schwanstecher and Bassen, 1997), diazoxide caused membrane hyperpolarization in both substantial nigra (Häusser et al., 1991) and hippocampal (Ben-Ari et al., 1990; Fujimura et al., 1997) neurons. In pancreatic  cells, the opening of K_ATP channels by diazoxide was hindered if the concentration of intracellular ATP was higher than 3 mM (Trube et al., 1986) and the intracellular MgADP is essential for the activation of K_ATP channels (Larsson et al., 1992). It is not clear whether these requirements are also applied to neuronal K_ATP channels. Nevertheless, using the same recording condition, we demonstrated that diazoxide induced a glibenclamide-sensitive outward current in hippocampal CA1 neurons, as reported by Fujimura et al. (1997). Therefore, the negative result of K_ATP channel ligands in the PAG is not due to a limitation from whole cell recording technique such as a dialysis of intracellular MgADP, or a deterioration of K_ATP channel ligands used. The latter inference is further supported by two assays (B-H. Liang and C. Y. Cheng, personal communication). First, the structure of used glibenclamide has been confirmed by NMR analysis. Second, glibenclamide, from the same lot used in the present study, antagonized lemakalim-induced inhibition of spontaneous contraction of rat portal veins.

K_ATP Channels Are Not Involved in Effects of Morphine in Ventrolateral PAG. Glibenclamide, of which the stability has been confirmed by NMR assay or bioassays in other preparations, did not affect morphine actions, either the presynaptic inhibition of IPSCs or postsynaptic activation of K⁺ channels, in the ventrolateral PAG. These findings suggest that K_ATP channels are not involved in morphine actions in the ventrolateral PAG. The mechanism of opioid inhibition of IPSCs in the PAG was proposed to be activation of a 4-aminopyridine- and dendrotoxin-sensitive K⁺ conductance through 12-lipoxygenase products and was suggested to be the mechanism of the supraspinal analgesic action of opioids (Vaughan et al., 1997). Nevertheless, it disagrees with the finding that 4-aminopyridine (i.c.v.) fails to affect the antinociceptive response of morphine (Ocaña et al., 1995).

Different Proportions of Neurons Hyperpolarized by Morphine and µ-Opioid Peptides. The findings that morphine inhibits the inhibitory synaptic transmission and activates inwardly rectifying K⁺ channels in the ventrolateral PAG agree with previous experiments using µ-opioid peptides (Chieng and Christie, 1994a,b; Vaughan et al., 1997; Chiou and Huang, 1999; Han et al., 1999; Chiou, 2001). However, the proportions of neurons that were hyperpolarized by morphine and µ-opioid peptides are different (Table 1). It seems that less neurons are sensitive to morphine than to µ-opioid peptides in both neonates and adults (Table 1). It is unclear whether this lower incidence of morphine response is due to the partial agonist property of morphine (Osborne et al., 2000). Differences have been reported in the phosphorylation, internalization, or desensitization of µ-opioid receptors activated by µ-opioid peptides or alkaloids (Law et al., 2000). It remains to be elucidated whether those differences between morphine and [D-Ala²,N-MePhe⁴,Gly-ol⁵]-enkephalin can explain the different susceptibility found in the ventrolateral PAG.

Age-Dependent Postsynaptic Action of Morphine in Ventrolateral PAG. The finding that none of adult neurons is hyperpolarized by morphine is interesting. In morphine-resistant adult neurons, baclofen activated inwardly rectifying K⁺ channels, which share the same G protein-coupled cascade with µ-opioids (Christie and North, 1988). Therefore, the insensitivity to morphine in K⁺ channel activation in adult neurons is not attributed to a defect at signaling(s) downstream from receptor activation. Interestingly, to µ-opioid peptides, there are also more sensitive neurons in neonates compared with adults (Table 1). The reason for this age-dependence is unclear. Since morphine induced a comparable inhibition of synaptic transmission in neonatal and adult PAG neurons, it is unlikely that there is a global difference between adult and neonatal neurons in opioid receptor density or receptor coupling efficacy. The possibility remains to be elucidated that a developmental shift occurs at the pre- or postsynaptic site in the distribution of µ-opioid receptors, G protein activation efficacy, or G protein-coupled receptor kinase activity.