Neuroscience Letters, 143 (1992) 185 189 1992 Elsevier Scientific Publishers Ireland Ltd. All rights reserved 0304-3940/92/$ 05.00
185
NSL 08887
Hypoglycemia-activated K + channels in hippocampal neurons Cinzia T r o m b a a, A n t o n i n o Salvaggi¢, Giorgio R a c a g n ? 'b and Andrea Volterra ~'b ~'Center qf Neuropharmacology and hlnstitute of Pharmacological Sciences. UniversiO"of Milan. Milan (ltalv) (Received 2 March 1992; Revised version received 11 May 1992; Accepted 18 May 1992)
Key words." ATP-sensitive potassium channel; Hypoglycemia; Hippocampal neuron: Sulphonylurea; Potassium channel opener Channels linking the electrical and metabolic activities of cells (KAvp channels) have been described in various tissues, including some brain areas thypothalamus, cerebral cortex and substantia nigra). Here we report the existence in hippocampal neurons of K ~ permeant channels whose activity is regulated by extracellular glucose. They are open at the cell resting potential and respond to transient hypoglycemia with a reversible increase in activity. The one type so far characterized has a conductance of ~100 pS in isotonic K +, is inhibited by the sulphonylurea glibenclamide (1/~M), and is activated by the potassium channel opener lemakalim (0.1 1 ¢tM). These data provide a direct demonstration of the presence, in hippocampal neurons, of glucose-sensitive channels that could belong to the KAyp t;amily.
Originally described in heart myocytes, K + channels linking the metabolic state to the electrical activity of cells have now been found in a variety of different tissues, including pancreatic fl cells, skeletal and smooth muscle cells and central neurons [3]. They have been generally termed KAT p channels due to the inhibitory regulation by intracellular ATE a decrease in the nucleotide concentration leading to channel activation. In cardiac myo c y t e s KAT P channels have been proposed to function as a cell defense mechanism: they activate in situations of energy impairment like hypoxia and ischemia, and cause hyperpolarization of the cells, thus reducing their metabolic demand [3, 14]. In pancreatic fl cells or neurons of the ventromedial nucleus of the hypothalamus (VMHN), KATpchannels represent the molecular basis of the physiological glucose-sensing property of these cells which regulates insulin secretion and appetite [2~,]. From a pharmacological standpoint, KAT P channels of different tissues seem all susceptible to the inhibitory action of hypoglycemic sulphonylureas (tolbutamide, glibenclamide etc.), although with different sensitivities; in some tissues, they also respond to the stimulatory action of potassium channel openers (KCO) such as diazoxide or cromakalim [3]. Radioligand binding studies have demonstrated a high density of 3H-glibenclamide binding sites at many brain locations, including the substantia Correspondenee." A. Volterra, Center of Neuropharmacology and Institute of Pharmacological Sciences, University of Milan, Via Balzaretti 9, 20133 Milan, Italy. Fax: (39)(2)2940-4961.
nigra, motor cortex and some regions of the hippocampus [11, 12]. These results have been taken as an indication of the presence of channels of the KAT p type in the CNS. As a matter of fact, KAT p channels have been characterized in neurons of the hypothalamus [4], cerebral cortex [5] and substantia nigra [17]. Considerable research is now focused on the hippocampal formation, because of the differential susceptibility of its neuronal circuitry to anoxic- ischemic insult. Recent studies based on pharmacological evidence suggest a possible role for KAT P channels in the cell's defense against anoxic insults. In fact, sulphonylureas and KCO drugs seem to modulate the glutamate-induced depolarization during anoxia in CA3 pyramidal neurons [6, 7], and KCO drugs have been reported to display a protective action in conditions of in vitro-induced excitotoxicity [1] or in vivo seizures [10]. In the present study we investigate the possibility of recording single KAT P channels in hippocampal cells in culture. As a first step in our research, we have used an in vitro model of transient hypoglycemia and looked for K ÷ permeant channels whose activity could change in response to different extracellular glucose concentrations. Primary neuronal cultures were obtained from embryonic (ED 18) rat hippocampi. Following enzymatic digestion (Trypsin 0.1%, DNase 0.05%), mechanical dissociation, and centrifugation (1100 rpm/4 min), the cells were resuspended in a medium consisting basically of 50% MEM\50% F10 (Gibco) with the addition of MgCI2 (0.5 mM), ascorbic acid (0.14 mM) and 10% foetal calf serum (FCS, ICN Flow), and plated on poly-ornithine
186 GLUCOSE
A
20 mM
OA[
0 mM
I
I
0
II -1
-2
-3
0
-1
-2
-3
I (pA)
B
20 mM
cz o . o L
,
0
, -
-~
0 mM
,
,
-8
-12
0
-4
-8
-12
I (pA)
pA z,0 ms Fig. 1. Two types of hypoglycemia-activatedchannels. Representative traces from two patches showing openings of a 0.7 pA (A) and a 6 pA (B) channel, respectively.Recordingswere made at a pipette potential (Vp) of 0 mV, corresponding to the cell resting potential (F~). Fm=-52+6.5 mV, n=6, as seen by switching to whole-cellmode at the end of some experiments. Insets represent the frequency distribution histograms of current values in the same experiments,in either 20 or 0 mM glucose solutions. Notice that, for both channels, hypoglycemia induces a significantenhancementof the current peak correspondingto the open channel state, i.e. 0.7 pA (A) and 6 pA (B). Histograms were generated from I-rain-longstretches of data.
treated Petri dishes. After 6 h, FCS in the medium was diluted to 1/4 by substitution with N1 supplement. Cultures were then grown in humidified atmosphere of 5% CO2\95% 02 and used for experiments after 5-8 days in vitro. The cell-attached mode of the patch-clamp technique was employed in all experiments. Recording pipettes with a resistance of 3-5 Mr2 were filled with (in mM): 130 KC1, 1 CaCI> 1 MgCI> 10 HEPES, pH 7.3 with KOH. The bathing solution was continously superfused onto cells at room temperature and consisted o f (in mM): 140 NaC1, 5.4 KCI, 1.8 CaC12, 1 MgCI> 5 HEPES (pH 7.4 with N a O H ) + 20 o-glucose (high-glucose solution) or + 20 o-galactose (glucose-free solution). Pharmacological agents were directly applied onto cells via a fast microperfusion system. Single channel currents were recorded with an EPC-7 (List Electronic) patch-clamp amplifier and stored on videotape for later analysis. One min-long stretches of data, filtered at 0.5 kHz and digitized at 2.5 kHz, were stored as data files on a computer
using the pClamp program (Axon Instruments) and then exported for analysis with a custom-made, automated trace idealizer program based on the method of Dudei and Franke [9]. Channel activity, measured from idealized traces, was expressed as the product Npo (where ..V is the number of active channels in the patch and Po is the probability of opening) and monitored continously by sequential calculation of the Npo value on l-s-long blocks of data [8]. In our experimental protocols we generally exposed hippocampal neurons to high-glucose (20 raM) solutions while recording elementary membrane currents at the cell resting potential ( ~ = 0 ) . Once a stable recording was achieved, we then switched to glucose-free solutions, inducing a progressive dilution o f glucose concentration in the experimental bath (4-5 min to get virtual zero glucose). In most o f the patches channel openings of various amplitude were recorded at resL Often channel activity did not show an obvious change in response to hypoglycemia (42 patches out of 53); however, in 11 experiments we did identify at least two types of channels with different single channel conductance whose activity was clearly enhanced by transient hypoglycemia (Fig. 1). Traces in Fig. 2A are taken from a representative patch containing the higher conductance channels. A few openings of a 6 pA channel were detectable when the cells were bathed in high-glucose medium (a). After starting the superfusion of glucose-free medium, channel activity gradually increased, until it reached a maximum after about 5 min (b). This effect was fully reversed upon return to initial conditions (c). Similar results were obtained in a total of 7 patches, where Npo product was found to augment from 0.007 _4-0.01 in 20 mM glucose to 1.186 _+ 0.88 in glucose-free medium with an average delay of 293 + 20 s. In order to characterize the properties of the channel described above, we constructed its current-voltage relationship (Fig. 2B). Measurement of the conductance from the slope of the line between +20 mV and - 6 0 mV (Vv) gave a value of 94.8 + 8.6 pS (n=3). On depolarization the activity of the channel seemed to increase while its amplitude decreased to reach zero value at ~60 mV positive to rest, in the range expected for E K in isotonic K + (see legend to Fig. 1). When patches were further depolarized, however, we did not observe a clear reversal of the current, often due also to the appearance of other channels. We then tested the effectiveness of the putative selective KATe channel antagonist glibenclamide. One example of the results obtained is shown in Fig. 3. In this patch openings of at least 2 channels of 6 pA were detectable in glucose-free medium (a). After about 3 min of stable recording glibenclamide (1 J~M) was applied and
187
a.
A
0mM GLUCOSE
a. 20mM GLUCOSE -C
'
b. GLIBENCLAMIDE (II.IM)
rTFTb. 0 m M
r-rrr
GLUCOSE
C. 2 0 m M
GLUCOSE , ,
,,, ......
c.
, .......
F , , , 1
-
-
0
.
~
-C
T. . . . . .
20ram GLUCOSE
d. LEMAKALIM (100nM) -
--7
B 20
0
I
-2-
y'
Vp (mV) -20 -40 -60
10pA I 300ms
-80
30 s) despite that our perfusion system allows the compounds to reach the cell within I 2 s.
a total of 4 experiments, Npo decreased by a maximum of 87.7 + 5.6% following application of glibenclamide (1 /~M). In the same experiments we found that these glucose- and glibenclamide-sensitive channels could also be potently activated by the KCO compound lemakalim (BRL 38227), the ( - ) enantiomer of cromakalim. An example is shown in Fig. 3d, where lemakalim was applied on the same patch where glibenclamide had been previously tested. Following wash out of the sulphonylurea and recovery in a glucose-free medium, the cell was exposed to a solution containing 20 mM glucose. This caused a marked reduction in the open probability of the channel (c). Then, lemakalim (100 nM) was applied and within 45 s a 10-fold increase in channel activity was observed. In 3 experiments where the effect of lemakalim (0.1 1/JM) was studied, Np~,value always showed a max-
188 imal enhancement of at least 40 times. Such enhancement was only partially reversed upon wash out of the compound. As already mentioned, in our experiments we observed (n=4) another type of K + permeant channel which is active at the cell resting potential and is also susceptible to activation by hypoglycemic conditions (see Fig. 1). Preliminary characterization of the properties of this channel confirms that it is different from the one above described since: (a) it has a much lower conductance (~10 pS in symmetrical K+); and (b) it displays a different voltage-dependency. The data here reported demonstrate that glucose-sensitive K + channels do exist in cultured hippocampal neurons. Moreover, at least two distinct conductances (100 and 10 pS) respond to hypoglycemic conditions with increased activity. Although we have not conducted yet a detailed study o f the permeability properties of these channels, two factors suggest that they preferably permeate K + ions: (a) in our experimental conditions K ~ is the main cation present in the patch pipette and (b) the zero current value for both channels lies in the range of the expected EK. The 100 pS channel is a good candidate to belong to the large family of KAVpchannels, based also on its sensitivity to the inhibitory action of glibenclamide. However, to verify this suggestion we need to test directly for ATPsensitivity in excised inside-out patches. The same channel is potently activated by the KCO drug lemakalim (0.1-1/aM). Recently, a K ÷ channel which opens upon application o f either cromakalim (100 MM) or energy depleting agents and is inhibited by glibenclamide, has been reported in hippocampal neurons [15]. Such a channel, however, is different from the one here described: it has a conductance of 50 pS in symmetrical K + [16] and is normally closed in the absence of pharmacological treatments. Therefore, an heterogeneity among the channels which respond to KCO and sulphonylurea compounds may exist in the hippocampus. In our experiments, glucose-sensitive K ~ channels have been recorded only in =20% of the tested patches. Studies on 3H-glibenclamide binding in the hippocampal formation indicate several possibilities to explain this data: (a) at birth, the density of glibenclimide binding sites is quite low, due to ontogenetic immaturity [12], (b) their regional distribution is uneven [12], and (c) also their cellular location may differ depending on the cell type. For example, in area CA3, binding sites seem preferentially located at the pre-synaptic level on the mossy fiber nerve endings, whereas in area CA1 they are probably post-synaptic on the pyramidal neurons [13, 18]. In our experiments we monitor only the channels that are present in the somatic region of hippocampal neurons
already developed at ED 18. lnnnunohistochemicai characterization of these cells will lead to a better understanding of the significance of glucose-sensitive channels in the hippocampal circuitry. From a functional poinl o1" view, it has been proposed that pre-synaptic K x~p channels in mossy fiber endings regulate glutamate release [6, 18], but no information has been available until now about somatic hippocampal KAT~,channels. The only example in the CNS where a clear function has been attributed to such channels is represented by the case of V M H N neurons. Here, as in pancreatic fl cells, Kwt, channels contribute in a significant way to the setting of the membrane potential and the firing activity of the cell, thereby constituting the main mechanism through which V M H N neurons express their distinctive glucose-sensing property [4]. Besides the common somatic location and glucose-sensitivity, the 100 pS hippocampal channel here described and the KAjp channel in the hypothalamus show similarities in at least two other respects: (a) both display a high conductance in symmetric K ÷ and (b) both are active at the cell resting potential in normoglycemic conditions. However, the contribution of the hippocampal channel to the resting potential has not been yet addressed. In this light, it will be of interest to perform experiments where the effects of modified glucose concentrations are tested in parallel on the single channel function and on the whole cell current response. This will allow an evaluation of the role played by the glucose-sensitive channels here described in the modulation of hippocampal cell excitability. The authors are grateful to the pharmaceutical companies Hoechst Italia Sud and SmithKline Beecham tbr the gift of glibenclamide and lemakalim, respectively. We would also like to thank Drs. I.S. Cohen and S.A. Siegelbaum for critical reading of the manuscript.
1 Abele,A.E. and Miller, R.J., Potassium channel activators abolish excitotoxicityin cultured hippocampal pyramidal neurons, Neurosci. Lett., 115 (1990) 195-200. 2 Ashcroft, F.M., Harrison, D.E. and Ashcroll, S.J.H., Glucose induces closureof single potassium channels in isolated rat pancreatic fl cells, Nature, 312 (1984)446-448. 3 Ashcroft, S.J.H. and Ashcroft, F.M., Properties and functions of ATP-sensitiveK channels, Cell. Signal., 2 (1990) t 97-214. 4 Ashford, M.L.J., Boden, RR. and Treherne,J.M., Glucose-induced excitation of hypothalamicneurones is mediated by ATP-sensitive K+channels, PfliigersArch., 415 (1990)479 -483. 5 Ashford, M.LJ., Sturgess, N.C., Trout, N.J., Gardner, N.J. and Hales, C.N., Adenosine-5'-triphosphate-sensitiveion channels in neonatal rat cultured central neurones, Pfltigers Arch., 4t2 (1988) 297- 304. 6 Ben-Ari, Y., Galanin and glibenctamide modulate the anoxic release of glutamate in rat CA3 hippocampal neurons, Eur. J. Neurosci., 2 (1990) 62-68.
189 7 Ben-Ari, Y.. Krnjevic', K. and Crepel, V., Activators of ATP-sensilive K ~ channels reduce anoxic depolarization in CA3 hippocampal neurons, Neuroscience, 37 (1990) 55 60. 8 Buttner, N.. Siegelbaum, S.A. and Volterra, A., Direct modulation o1" Aplysia S-K + channels by a 12-1ipoxygenase metabolile of arachidonic acid, Nature, 342 (19891 553 555. 9 Dudel. J. and Franke, C.H., Single glutamate-gated synaptic channels at the crayfish neuromuscolar junction, Pfliigers Arch., 408 (1987) 3(17 314. 10 Gandolfo, G., Gottesmann. C., Bidard, J.N. and Lazdunski, M., K channels openers prevent epilepsy induced by the bee v e n o m pcptide MCD, Eur. J. Pharmacol., 159 (19891 329 330. I 1 Mourre, C.. Ben-Aft, Y., Bernardi, H., Fosset. M. and Lazdunski, M.. Anlidiabetic sulfonylureas: localization of binding sites in the brain and effects on the hyperpolarization induced by anoxia in hippocampal slices. Brain Res.. 486 (19891 159 164. 12 Mourre, C.. Widmann, C. and Lazdunski, M., Sulphonylurea binding sites associated with ATP-regulated K channels in the central nervous system: autoradiographic analysis of their distribution and ontogenesis, and their localization in mutant mice cerebellum, Brain Res., 519 (19901 197 214.
13 Mourre, C., Widmann, C. and Lazdunski, M.. Specifc hippocampal lesions indicate the presence of sulphonylurea binding sites associated to ATP-sensitive K ~ channels both post-synaptically and on mossy fibers, Brian Res., 540 (1991 ) 340 344. 14 Noma, A.. ATP-regulated K channels in cardiac muscle, Nature, 305(1983) 147 148. 15 Politi, D.M.T. and Rogawski, M.A., Glyburide-sensitive K + channels in cultured rat hippocampal neurons: activation by cromakalim and energy-depleting conditions, Mol, Pharmacol., 40 ( 1991 ) 308 315. 16 PolitL D,M.T. and Rogawski, M.A.. Glyburide-sensitivc K' channels in cultured rat hippocampal neurons: activation by cromakalim and energy-depleting conditions. Soc. Neurosci. Abstr~, 17 (19911 1474. 17 R6per, J., Hainsworth, A.H. and Ashcroft. V.M., ATP-scnsiti~e K + ill guinea-pig isolated substantia nigra neurones are modulated by cellular metabolism, J. Physiol.. 430 (19901 13011. 18 Tremblay. E., Zini, S. and Ben-Ari, Y., Autoradiographic study of the cellular localization of [~H]glibenclamide binding sites in the rat hippocampus, Neurosci. Lett., 127 (1991) 21 24.
185
NSL 08887
Hypoglycemia-activated K + channels in hippocampal neurons Cinzia T r o m b a a, A n t o n i n o Salvaggi¢, Giorgio R a c a g n ? 'b and Andrea Volterra ~'b ~'Center qf Neuropharmacology and hlnstitute of Pharmacological Sciences. UniversiO"of Milan. Milan (ltalv) (Received 2 March 1992; Revised version received 11 May 1992; Accepted 18 May 1992)
Key words." ATP-sensitive potassium channel; Hypoglycemia; Hippocampal neuron: Sulphonylurea; Potassium channel opener Channels linking the electrical and metabolic activities of cells (KAvp channels) have been described in various tissues, including some brain areas thypothalamus, cerebral cortex and substantia nigra). Here we report the existence in hippocampal neurons of K ~ permeant channels whose activity is regulated by extracellular glucose. They are open at the cell resting potential and respond to transient hypoglycemia with a reversible increase in activity. The one type so far characterized has a conductance of ~100 pS in isotonic K +, is inhibited by the sulphonylurea glibenclamide (1/~M), and is activated by the potassium channel opener lemakalim (0.1 1 ¢tM). These data provide a direct demonstration of the presence, in hippocampal neurons, of glucose-sensitive channels that could belong to the KAyp t;amily.
Originally described in heart myocytes, K + channels linking the metabolic state to the electrical activity of cells have now been found in a variety of different tissues, including pancreatic fl cells, skeletal and smooth muscle cells and central neurons [3]. They have been generally termed KAT p channels due to the inhibitory regulation by intracellular ATE a decrease in the nucleotide concentration leading to channel activation. In cardiac myo c y t e s KAT P channels have been proposed to function as a cell defense mechanism: they activate in situations of energy impairment like hypoxia and ischemia, and cause hyperpolarization of the cells, thus reducing their metabolic demand [3, 14]. In pancreatic fl cells or neurons of the ventromedial nucleus of the hypothalamus (VMHN), KATpchannels represent the molecular basis of the physiological glucose-sensing property of these cells which regulates insulin secretion and appetite [2~,]. From a pharmacological standpoint, KAT P channels of different tissues seem all susceptible to the inhibitory action of hypoglycemic sulphonylureas (tolbutamide, glibenclamide etc.), although with different sensitivities; in some tissues, they also respond to the stimulatory action of potassium channel openers (KCO) such as diazoxide or cromakalim [3]. Radioligand binding studies have demonstrated a high density of 3H-glibenclamide binding sites at many brain locations, including the substantia Correspondenee." A. Volterra, Center of Neuropharmacology and Institute of Pharmacological Sciences, University of Milan, Via Balzaretti 9, 20133 Milan, Italy. Fax: (39)(2)2940-4961.
nigra, motor cortex and some regions of the hippocampus [11, 12]. These results have been taken as an indication of the presence of channels of the KAT p type in the CNS. As a matter of fact, KAT p channels have been characterized in neurons of the hypothalamus [4], cerebral cortex [5] and substantia nigra [17]. Considerable research is now focused on the hippocampal formation, because of the differential susceptibility of its neuronal circuitry to anoxic- ischemic insult. Recent studies based on pharmacological evidence suggest a possible role for KAT P channels in the cell's defense against anoxic insults. In fact, sulphonylureas and KCO drugs seem to modulate the glutamate-induced depolarization during anoxia in CA3 pyramidal neurons [6, 7], and KCO drugs have been reported to display a protective action in conditions of in vitro-induced excitotoxicity [1] or in vivo seizures [10]. In the present study we investigate the possibility of recording single KAT P channels in hippocampal cells in culture. As a first step in our research, we have used an in vitro model of transient hypoglycemia and looked for K ÷ permeant channels whose activity could change in response to different extracellular glucose concentrations. Primary neuronal cultures were obtained from embryonic (ED 18) rat hippocampi. Following enzymatic digestion (Trypsin 0.1%, DNase 0.05%), mechanical dissociation, and centrifugation (1100 rpm/4 min), the cells were resuspended in a medium consisting basically of 50% MEM\50% F10 (Gibco) with the addition of MgCI2 (0.5 mM), ascorbic acid (0.14 mM) and 10% foetal calf serum (FCS, ICN Flow), and plated on poly-ornithine
186 GLUCOSE
A
20 mM
OA[
0 mM
I
I
0
II -1
-2
-3
0
-1
-2
-3
I (pA)
B
20 mM
cz o . o L
,
0
, -
-~
0 mM
,
,
-8
-12
0
-4
-8
-12
I (pA)
pA z,0 ms Fig. 1. Two types of hypoglycemia-activatedchannels. Representative traces from two patches showing openings of a 0.7 pA (A) and a 6 pA (B) channel, respectively.Recordingswere made at a pipette potential (Vp) of 0 mV, corresponding to the cell resting potential (F~). Fm=-52+6.5 mV, n=6, as seen by switching to whole-cellmode at the end of some experiments. Insets represent the frequency distribution histograms of current values in the same experiments,in either 20 or 0 mM glucose solutions. Notice that, for both channels, hypoglycemia induces a significantenhancementof the current peak correspondingto the open channel state, i.e. 0.7 pA (A) and 6 pA (B). Histograms were generated from I-rain-longstretches of data.
treated Petri dishes. After 6 h, FCS in the medium was diluted to 1/4 by substitution with N1 supplement. Cultures were then grown in humidified atmosphere of 5% CO2\95% 02 and used for experiments after 5-8 days in vitro. The cell-attached mode of the patch-clamp technique was employed in all experiments. Recording pipettes with a resistance of 3-5 Mr2 were filled with (in mM): 130 KC1, 1 CaCI> 1 MgCI> 10 HEPES, pH 7.3 with KOH. The bathing solution was continously superfused onto cells at room temperature and consisted o f (in mM): 140 NaC1, 5.4 KCI, 1.8 CaC12, 1 MgCI> 5 HEPES (pH 7.4 with N a O H ) + 20 o-glucose (high-glucose solution) or + 20 o-galactose (glucose-free solution). Pharmacological agents were directly applied onto cells via a fast microperfusion system. Single channel currents were recorded with an EPC-7 (List Electronic) patch-clamp amplifier and stored on videotape for later analysis. One min-long stretches of data, filtered at 0.5 kHz and digitized at 2.5 kHz, were stored as data files on a computer
using the pClamp program (Axon Instruments) and then exported for analysis with a custom-made, automated trace idealizer program based on the method of Dudei and Franke [9]. Channel activity, measured from idealized traces, was expressed as the product Npo (where ..V is the number of active channels in the patch and Po is the probability of opening) and monitored continously by sequential calculation of the Npo value on l-s-long blocks of data [8]. In our experimental protocols we generally exposed hippocampal neurons to high-glucose (20 raM) solutions while recording elementary membrane currents at the cell resting potential ( ~ = 0 ) . Once a stable recording was achieved, we then switched to glucose-free solutions, inducing a progressive dilution o f glucose concentration in the experimental bath (4-5 min to get virtual zero glucose). In most o f the patches channel openings of various amplitude were recorded at resL Often channel activity did not show an obvious change in response to hypoglycemia (42 patches out of 53); however, in 11 experiments we did identify at least two types of channels with different single channel conductance whose activity was clearly enhanced by transient hypoglycemia (Fig. 1). Traces in Fig. 2A are taken from a representative patch containing the higher conductance channels. A few openings of a 6 pA channel were detectable when the cells were bathed in high-glucose medium (a). After starting the superfusion of glucose-free medium, channel activity gradually increased, until it reached a maximum after about 5 min (b). This effect was fully reversed upon return to initial conditions (c). Similar results were obtained in a total of 7 patches, where Npo product was found to augment from 0.007 _4-0.01 in 20 mM glucose to 1.186 _+ 0.88 in glucose-free medium with an average delay of 293 + 20 s. In order to characterize the properties of the channel described above, we constructed its current-voltage relationship (Fig. 2B). Measurement of the conductance from the slope of the line between +20 mV and - 6 0 mV (Vv) gave a value of 94.8 + 8.6 pS (n=3). On depolarization the activity of the channel seemed to increase while its amplitude decreased to reach zero value at ~60 mV positive to rest, in the range expected for E K in isotonic K + (see legend to Fig. 1). When patches were further depolarized, however, we did not observe a clear reversal of the current, often due also to the appearance of other channels. We then tested the effectiveness of the putative selective KATe channel antagonist glibenclamide. One example of the results obtained is shown in Fig. 3. In this patch openings of at least 2 channels of 6 pA were detectable in glucose-free medium (a). After about 3 min of stable recording glibenclamide (1 J~M) was applied and
187
a.
A
0mM GLUCOSE
a. 20mM GLUCOSE -C
'
b. GLIBENCLAMIDE (II.IM)
rTFTb. 0 m M
r-rrr
GLUCOSE
C. 2 0 m M
GLUCOSE , ,
,,, ......
c.
, .......
F , , , 1
-
-
0
.
~
-C
T. . . . . .
20ram GLUCOSE
d. LEMAKALIM (100nM) -
--7
B 20
0
I
-2-
y'
Vp (mV) -20 -40 -60
10pA I 300ms
-80
30 s) despite that our perfusion system allows the compounds to reach the cell within I 2 s.
a total of 4 experiments, Npo decreased by a maximum of 87.7 + 5.6% following application of glibenclamide (1 /~M). In the same experiments we found that these glucose- and glibenclamide-sensitive channels could also be potently activated by the KCO compound lemakalim (BRL 38227), the ( - ) enantiomer of cromakalim. An example is shown in Fig. 3d, where lemakalim was applied on the same patch where glibenclamide had been previously tested. Following wash out of the sulphonylurea and recovery in a glucose-free medium, the cell was exposed to a solution containing 20 mM glucose. This caused a marked reduction in the open probability of the channel (c). Then, lemakalim (100 nM) was applied and within 45 s a 10-fold increase in channel activity was observed. In 3 experiments where the effect of lemakalim (0.1 1/JM) was studied, Np~,value always showed a max-
188 imal enhancement of at least 40 times. Such enhancement was only partially reversed upon wash out of the compound. As already mentioned, in our experiments we observed (n=4) another type of K + permeant channel which is active at the cell resting potential and is also susceptible to activation by hypoglycemic conditions (see Fig. 1). Preliminary characterization of the properties of this channel confirms that it is different from the one above described since: (a) it has a much lower conductance (~10 pS in symmetrical K+); and (b) it displays a different voltage-dependency. The data here reported demonstrate that glucose-sensitive K + channels do exist in cultured hippocampal neurons. Moreover, at least two distinct conductances (100 and 10 pS) respond to hypoglycemic conditions with increased activity. Although we have not conducted yet a detailed study o f the permeability properties of these channels, two factors suggest that they preferably permeate K + ions: (a) in our experimental conditions K ~ is the main cation present in the patch pipette and (b) the zero current value for both channels lies in the range of the expected EK. The 100 pS channel is a good candidate to belong to the large family of KAVpchannels, based also on its sensitivity to the inhibitory action of glibenclamide. However, to verify this suggestion we need to test directly for ATPsensitivity in excised inside-out patches. The same channel is potently activated by the KCO drug lemakalim (0.1-1/aM). Recently, a K ÷ channel which opens upon application o f either cromakalim (100 MM) or energy depleting agents and is inhibited by glibenclamide, has been reported in hippocampal neurons [15]. Such a channel, however, is different from the one here described: it has a conductance of 50 pS in symmetrical K + [16] and is normally closed in the absence of pharmacological treatments. Therefore, an heterogeneity among the channels which respond to KCO and sulphonylurea compounds may exist in the hippocampus. In our experiments, glucose-sensitive K ~ channels have been recorded only in =20% of the tested patches. Studies on 3H-glibenclamide binding in the hippocampal formation indicate several possibilities to explain this data: (a) at birth, the density of glibenclimide binding sites is quite low, due to ontogenetic immaturity [12], (b) their regional distribution is uneven [12], and (c) also their cellular location may differ depending on the cell type. For example, in area CA3, binding sites seem preferentially located at the pre-synaptic level on the mossy fiber nerve endings, whereas in area CA1 they are probably post-synaptic on the pyramidal neurons [13, 18]. In our experiments we monitor only the channels that are present in the somatic region of hippocampal neurons
already developed at ED 18. lnnnunohistochemicai characterization of these cells will lead to a better understanding of the significance of glucose-sensitive channels in the hippocampal circuitry. From a functional poinl o1" view, it has been proposed that pre-synaptic K x~p channels in mossy fiber endings regulate glutamate release [6, 18], but no information has been available until now about somatic hippocampal KAT~,channels. The only example in the CNS where a clear function has been attributed to such channels is represented by the case of V M H N neurons. Here, as in pancreatic fl cells, Kwt, channels contribute in a significant way to the setting of the membrane potential and the firing activity of the cell, thereby constituting the main mechanism through which V M H N neurons express their distinctive glucose-sensing property [4]. Besides the common somatic location and glucose-sensitivity, the 100 pS hippocampal channel here described and the KAjp channel in the hypothalamus show similarities in at least two other respects: (a) both display a high conductance in symmetric K ÷ and (b) both are active at the cell resting potential in normoglycemic conditions. However, the contribution of the hippocampal channel to the resting potential has not been yet addressed. In this light, it will be of interest to perform experiments where the effects of modified glucose concentrations are tested in parallel on the single channel function and on the whole cell current response. This will allow an evaluation of the role played by the glucose-sensitive channels here described in the modulation of hippocampal cell excitability. The authors are grateful to the pharmaceutical companies Hoechst Italia Sud and SmithKline Beecham tbr the gift of glibenclamide and lemakalim, respectively. We would also like to thank Drs. I.S. Cohen and S.A. Siegelbaum for critical reading of the manuscript.
1 Abele,A.E. and Miller, R.J., Potassium channel activators abolish excitotoxicityin cultured hippocampal pyramidal neurons, Neurosci. Lett., 115 (1990) 195-200. 2 Ashcroft, F.M., Harrison, D.E. and Ashcroll, S.J.H., Glucose induces closureof single potassium channels in isolated rat pancreatic fl cells, Nature, 312 (1984)446-448. 3 Ashcroft, S.J.H. and Ashcroft, F.M., Properties and functions of ATP-sensitiveK channels, Cell. Signal., 2 (1990) t 97-214. 4 Ashford, M.L.J., Boden, RR. and Treherne,J.M., Glucose-induced excitation of hypothalamicneurones is mediated by ATP-sensitive K+channels, PfliigersArch., 415 (1990)479 -483. 5 Ashford, M.LJ., Sturgess, N.C., Trout, N.J., Gardner, N.J. and Hales, C.N., Adenosine-5'-triphosphate-sensitiveion channels in neonatal rat cultured central neurones, Pfltigers Arch., 4t2 (1988) 297- 304. 6 Ben-Ari, Y., Galanin and glibenctamide modulate the anoxic release of glutamate in rat CA3 hippocampal neurons, Eur. J. Neurosci., 2 (1990) 62-68.
189 7 Ben-Ari, Y.. Krnjevic', K. and Crepel, V., Activators of ATP-sensilive K ~ channels reduce anoxic depolarization in CA3 hippocampal neurons, Neuroscience, 37 (1990) 55 60. 8 Buttner, N.. Siegelbaum, S.A. and Volterra, A., Direct modulation o1" Aplysia S-K + channels by a 12-1ipoxygenase metabolile of arachidonic acid, Nature, 342 (19891 553 555. 9 Dudel. J. and Franke, C.H., Single glutamate-gated synaptic channels at the crayfish neuromuscolar junction, Pfliigers Arch., 408 (1987) 3(17 314. 10 Gandolfo, G., Gottesmann. C., Bidard, J.N. and Lazdunski, M., K channels openers prevent epilepsy induced by the bee v e n o m pcptide MCD, Eur. J. Pharmacol., 159 (19891 329 330. I 1 Mourre, C.. Ben-Aft, Y., Bernardi, H., Fosset. M. and Lazdunski, M.. Anlidiabetic sulfonylureas: localization of binding sites in the brain and effects on the hyperpolarization induced by anoxia in hippocampal slices. Brain Res.. 486 (19891 159 164. 12 Mourre, C.. Widmann, C. and Lazdunski, M., Sulphonylurea binding sites associated with ATP-regulated K channels in the central nervous system: autoradiographic analysis of their distribution and ontogenesis, and their localization in mutant mice cerebellum, Brain Res., 519 (19901 197 214.
13 Mourre, C., Widmann, C. and Lazdunski, M.. Specifc hippocampal lesions indicate the presence of sulphonylurea binding sites associated to ATP-sensitive K ~ channels both post-synaptically and on mossy fibers, Brian Res., 540 (1991 ) 340 344. 14 Noma, A.. ATP-regulated K channels in cardiac muscle, Nature, 305(1983) 147 148. 15 Politi, D.M.T. and Rogawski, M.A., Glyburide-sensitive K + channels in cultured rat hippocampal neurons: activation by cromakalim and energy-depleting conditions, Mol, Pharmacol., 40 ( 1991 ) 308 315. 16 PolitL D,M.T. and Rogawski, M.A.. Glyburide-sensitivc K' channels in cultured rat hippocampal neurons: activation by cromakalim and energy-depleting conditions. Soc. Neurosci. Abstr~, 17 (19911 1474. 17 R6per, J., Hainsworth, A.H. and Ashcroft. V.M., ATP-scnsiti~e K + ill guinea-pig isolated substantia nigra neurones are modulated by cellular metabolism, J. Physiol.. 430 (19901 13011. 18 Tremblay. E., Zini, S. and Ben-Ari, Y., Autoradiographic study of the cellular localization of [~H]glibenclamide binding sites in the rat hippocampus, Neurosci. Lett., 127 (1991) 21 24.
