Spontaneous Channel Openings Research Articles

Penicillin blocks human α 1β 1 and α 1β 1γ 2S GABA A channels that open spontaneously

We used the open-channel blocker, penicillin (10 mM), as a tool to investigate if the human α 1β 1 or α 1β 1γ 2S γ-aminobutyric acid type A (GABA A) receptor channels opened in the absence of GABA. Application of penicillin to cells expressing the receptors resulted in a transient inward whole-cell current, the off-current, upon penicillin removal. The amplitude of the off-current was dependent on the duration of the penicillin application, it reversed in polarity at depolarized potentials and exhibited “run-down” similar to the GABA-activated currents. Bicuculline (100 μM) blocked the off-current response. Pentobarbital (50 μM) enhanced the peak off-current amplitude by 2.8 and 3.4 in α 1β 1 and α 1β 1γ 2S receptors, respectively. Diazepam (1 μM) only enhanced the off-current peak response in α 1β 1γ 2S receptors (1.6) and induced the development of an inward current when applied alone. The results are consistent with that the α 1β 1 or α 1β 1γ 2S GABA A receptors can open in the absence of GABA and raise the question of what role spontaneous channel openings have in the function of GABA A receptors.

Molecular basis for modulation of recombinant alpha1beta2gamma2 GABAA receptors by protons.

We have previously shown that extracellular protons inhibit recombinant and native GABA(A) receptors. In this report, we studied the site(s) and mechanism by which protons modulate the GABA(A) receptor. Whole cell GABA-activated currents were recorded from human embryonic kidney (HEK) 293 cells expressing recombinant alpha1beta2gamma2 GABA(A) receptors. Protons competitively inhibited the response to GABA and bicuculline. In contrast, change in pH did not influence direct gating of the channel by pentobarbital, and it did not influence spontaneous channel openings in alpha1(L264T)beta2gamma2 receptors, suggesting pH does not modulate channel activity by affecting the channel gating process directly. To test the hypothesis that protons modulate GABA(A) receptors at the ligand binding site, we systemically mutated N-terminal residues known to be involved in GABA binding and assessed effects of pH on these mutant receptors. Site-specific mutation of beta2 Y205 to F or alpha1 F64 to A, both of which are known to influence GABA binding, significantly reduced pH sensitivity of the GABA response. These mutations did not affect Zn(2+) sensitivity, suggesting that H(+) and Zn(2+) do not share a common site of action. Additional experiments further tested this possibility. Treatment with the histidine-modifying reagent diethylpyrocarbonate (DEPC) reduced Zn(2+)-mediated inhibition of GABA(A) receptors but had no effect on proton-induced inhibition of GABA currents. In addition, mutation of residues known to be involved in Zn(2+) modulation had no effect on pH modulation of GABA(A) receptors. Our results support the hypothesis that protons inhibit GABA(A) receptor function by direct or allosteric interaction with the GABA binding site. In addition, the sites of action of H(+) and Zn(2+) in GABA(A) receptors are distinct.

Cellular remodeling in heart failure disrupts K(ATP) channel-dependent stress tolerance.

ATP-sensitive potassium (K(ATP)) channels are required for maintenance of homeostasis during the metabolically demanding adaptive response to stress. However, in disease, the effect of cellular remodeling on K(ATP) channel behavior and associated tolerance to metabolic insult is unknown. Here, transgenic expression of tumor necrosis factor alpha induced heart failure with typical cardiac structural and energetic alterations. In this paradigm of disease remodeling, K(ATP) channels responded aberrantly to metabolic signals despite intact intrinsic channel properties, implicating defects proximal to the channel. Indeed, cardiomyocytes from failing hearts exhibited mitochondrial and creatine kinase deficits, and thus a reduced potential for metabolic signal generation and transmission. Consequently, K(ATP) channels failed to properly translate cellular distress under metabolic challenge into a protective membrane response. Failing hearts were excessively vulnerable to metabolic insult, demonstrating cardiomyocyte calcium loading and myofibrillar contraction banding, with tolerance improved by K(ATP) channel openers. Thus, disease-induced K(ATP) channel metabolic dysregulation is a contributor to the pathobiology of heart failure, illustrating a mechanism for acquired channelopathy.

Distinct Molecular Basis for Differential Sensitivity of the Serotonin Type 3A Receptor to Ethanol in the Absence and Presence of Agonist

Ethanol can potentiate serotonin type 3 (5-HT(3)) receptor-mediated responses in various neurons and in cells expressing 5-HT(3A) receptors. However, the molecular basis for alcohol modulation of 5-HT(3) receptor function has not been determined. Here we report that point mutations of the arginine at amino acid 222 in the N-terminal domain of the 5-HT(3A) receptor can alter the EC(50) value of the 5-HT concentration-response curve. Some point mutations at amino acid 222 resulted in spontaneous opening of the 5-HT(3A) receptor channel and an inward current activated by ethanol in the absence of agonist. Among these mutant receptors, the amplitude of the current activated by ethanol in the absence of agonist was correlated with the amplitude of the current resulting from spontaneous channel openings, suggesting that the sensitivity of the receptor to ethanol in the absence of agonist is, at least in part, dependent on the preexisting conformational equilibrium of the receptor protein. On the other hand, point mutations that conferred greater sensitivity to ethanol potentiation of agonist-activated responses were less sensitive or insensitive to ethanol in the absence of agonist. For these receptors, the magnitude of the potentiation of agonist-activated responses by ethanol was inversely correlated with the EC(50) values of the 5-HT concentration-response curves, suggesting that these mutations may modulate ethanol sensitivity of the receptor by altering the EC(50) value of the receptor. Thus, distinct molecular processes may determine the sensitivity of 5-HT(3A) receptors to ethanol in the absence and presence of agonist.

Serotonin antagonizes the human neuronal α7 nicotinic acetylcholine receptor and becomes an agonist after L248T α7 mutation

The effects of serotonin (5-hydroxytryptamine or 5HT) on chick α7 nicotinic receptors have already been described. However similar studies on human α7 receptors have been lacking. To begin to fill this deficiency, studies were made on wild-type and mutant human α7 (hα7) receptors expressed in Xenopus oocytes or human BOSC 23 cells. In oocytes wild-type hα7 receptors were blocked by 5HT, and this block was voltage-dependent. In contrast, 5HT acted as an agonist on hα7-mutant receptors (L248T). Outside-out membrane-patches from BOSC 23 cells expressing hα7-mutant receptors exhibited spontaneous channel openings of two conductance levels (59 pS and 76 pS) and short mean open time (0.9 ms). hα7-Mutant channels activated by nicotine or 5HT displayed similar conductances and high Ca 2+ permeability; but longer duration (2.7 ms) than the spontaneous openings. Mutations at Cys190 and Cys191, in the extracellular N-terminus of the human α7 gene, did not prevent receptor expression and incorporation in the oocyte membrane (determined by α-bungarotoxin binding). However, both 5HT and nicotine were incapable of gating the channels, indicating that the mutated Cys residues are in, or near, the 5HT- and nicotine-binding site. This is the first report that α7 receptors have spontaneous openings; and that 5HT is an agonist of hα7-mutant receptors, and an antagonist of hα7-wild-type receptors, through interactions at, or near the acetylcholine-binding sites.

Mutation at the putative GABA(A) ion-channel gate reveals changes in allosteric modulation.

We have mutated a conserved leucine in the putative membrane-spanning domain to serine in human GABA(A) beta2 and investigated the actions of a number of GABA(A) agonists, antagonists and modulators on human alpha1beta2deltaL259Sgamma2s compared to wild type alpha1beta2gamma2s GABA(A) receptors, expressed in Xenopus oocytes. The mutation resulted in smaller maximum currents to gamma-aminobutyric acid (GABA) compared to alpha1beta2gamma2s receptors, and large leak currents resulting from spontaneous channel opening. As reported, this mutation significantly decreased the GABA EC50 (110 fold), and reduced desensitization. Muscimol and the partial agonists 4,5,6,7-tetrahydroisoxazolo[5,4-c]pyridin-3-ol (THIP) and piperidine-4-sulphonic acid (P4S) also displayed a decrease in EC50. In addition to competitively shifting GABA concentration response curves, the antagonists bicuculline and SR95531 both inhibited the spontaneous channel activity on alpha1beta2deltaL259Sgamma2s receptors, with different degrees of maximum inhibition. The effects of a range of allosteric modulators, including benzodiazepines and anaesthetics were examined on a submaximal GABA concentration (EC20). Compared to wild type, none of these modulators potentiated the EC20 response of alpha1beta2deltaL259Sgamma2s receptors, however they all directly activated the receptor in the absence of GABA. To conclude, the above mutation resulted in receptors which exhibit a degree of spontaneous activity, and are more sensitive to agonists. Benzodiazepines and other agents modulate constitutive activity, but positive modulation of GABA is lost. The competitive antagonists bicuculline and SR95531 can also act as allosteric channel modulators through the same GABA binding site.

Operative condition-dependent response of cardiac ATP-sensitive K+ channels toward sulfonylureas.

A defining property of ATP-sensitive K+ (K[ATP]) channels is inhibition by sulfonylurea drugs, yet the response of cardiac K[ATP] channels toward sulfonylureas during myocardial ischemia is not consistent. Altered channel sensitivity toward sulfonylureas has, in part, been ascribed to antagonism by cytosolic nucleotide diphosphates, although the mechanism of interaction remains unclear. Herein, in inside-out patches excised from cardiomyocytes, we observed a dual response of K[ATP] channels toward the sulfonylurea drug, glyburide, in the presence of cytosolic UDP. Specifically, glyburide failed to inhibit spontaneous K[ATP] channel activity in the presence of UDP but inhibited UDP-induced channel activity after rundown of spontaneous channel openings. Such behavior of K[ATP] channels cannot be explained by differences in the level of channel activity or by UDP-induced displacement of glyburide. Rather, the dual response toward the sulfonylurea could be attributed to a property of K[ATP] channels to switch between operative conditions (spontaneous versus UDP-induced) each associated with a distinct responsiveness toward ligands. Conversion of post-rundown K[ATP] channels to the spontaneously operative channel condition, by Mg-ATP, restored the ability of UDP to antagonize the inhibitory action of glyburide lost after rundown, suggesting that the response of the channel to glyburide is phosphorylation dependent. The existence of distinct operative conditions of cardiac K[ATP] channels could be the basis for the inconsistent response of the channel toward sulfonylurea drugs and should be considered when sulfonylureas are used to implicate the opening of K[ATP] channels in the myocardium.

Agonist-induced closure of constitutively open gamma-aminobutyric acid channels with mutated M2 domains.

Ligand-gated ion channels display a fundamental property-channels remain virtually closed at rest and open upon agonist binding. Here we show that substituting alanines for either of two amino acid residues (T314 or L317) in the M2 region of the gamma-aminobutyric acid (GABA) rho1 subunit results in spontaneous channel opening in the absence of ligand. Surprisingly, for two single point mutants (T314A or L317A), application of very low concentrations of agonist partially suppressed this spontaneous current, while higher concentrations re-activated the receptors. When both of these sites were mutated (T314A/L317A), GABA nearly completely suppressed the constitutive current and did not re-activate the current even at very high concentrations. This study provides important new insights into the structure-function relationship of ligand-gated ion channels, where modification of the structure of the channel pore region not only alters the allosteric transition of the receptor protein but also reverses the polarity of agonist regulation of channel gating. Our results suggest that the sites where these two residues are located are structurally critical for channel gating.

Congenital myasthenic syndromes caused by mutations in acetylcholine receptor genes

Article abstract-Congenital myasthenic syndromes (CMS) are heterogeneous disorders in which the safety margin of neuromuscular transmission is compromised by one or more specific mechanisms. According to the site of the primary defect, the CMS can be classified as presynaptic, synaptic, or postsynaptic. The postsynaptic CMS identified to date are characterized by a kinetic abnormality of the acetylcholine receptor (AChR) with or without AChR deficiency, or by AChR deficiency without a primary kinetic abnormality. We hypothesized and subsequently confirmed that a kinetic abnormality of the AChR at the single-channel level predicts a mutation in an AChR subunit gene, and that a severe deficiency of AChR can stem from nonsense mutations in AChR subunit genes. Seven dominant slow-channel mutations in different subunits and different domains of the subunits have been identified thus far. Mutations in the M2 and M1 transmembrane domains act predominantly by slowing channel closure, and mutations in M2 also allow spontaneous channel opening. Another slow-channel mutation in the extracellular domain of the alpha-subunit acts predominantly by enhancing the affinity of AChR for ACh, causing repeated channel reopenings during a prolonged ACh occupancy. The recently discovered low-affinity fast-channel syndrome is the physiologic opposite of the slow-channel syndrome. Here, a mutation in the extracellular domain of the epsilon-subunit reduces the affinity of AChR for ACh; this decreases the rate of channel opening, and results in briefer than normal bursts of openings. Finally, a number of homozygous or heterozygous nonsense mutations in an AChR subunit cause severe AChR deficiency. NEUROLOGY 1997;48(Suppl 5): S28-S35

Chloride channels in excised membrane patches from human platelets: effect of intracellular calcium

Human platelets were studied by patch clamp recordings from inside-out membranes; these were formed by briefly dipping the platelet, in cell-attached mode, into silicone grease. At 20°C, in symmetrical 150 mM NaCl, spontaneous channel openings were rarely observed at negative potentials, whereas depolarised potentials (+60 to + 100 mV) elicited sustained channel activity in 38% of patches. The single channel conductance was 53 ± 1 pS at + 80 mV (outward current), decreasing to 20 ± 2 pS at −80 mV (inward current). Io substitution experiments indicated that this channel conducts Cl − and not Na +. Furthermore, 5-nitro-2-(3-phenylpropylamino)benzoate (100 μM), a recognized inhibitor of anion channels, induced a reversible ‘flickery‘ channel block. We estimate that each platelet possesses ≥30 such channels. Kinetic analysis suggested at least two open channel states ( τ = 0.8 ± 0.2 ms, τ= 22 ± 14 ms, n = 4) and two closed states ( τ = 0.8 ± 0.2 ms, τ = 12 ± 0.6 ms, n = 4). Increasing [Ca 2+] i to 10 μM, following channel activation by depolarisation, had no significant effect on channel kinetics or open probability, however, elevated [Ca 2+] i (300 nM–10 μM) increased the number of anion channels activated by subsequent depolarisation. This study represents the first recordings of ionic currents in excised, inside-out membrane patches from human platelets, and provides further evidence for the existence of chloride channels in these cells.

Channel gating in the absence of agonist by a homooligomeric molluscan GABA receptor expressed inXenopus oocytes from a cloned cDNA

We have previously described the isolation of a complementary DNA (cDNA) from the freshwater mollusc Lymnaea stagnalis encoding a polypeptide that exhibits approximately 50% identity to the beta-subunits of vertebrate gamma-aminobutyric acid (GABA) type A (GABAA) receptor. When expressed in Xenopus laevis oocytes from in vitro-transcribed RNA, the snail subunit forms functional homo-oligomeric receptors possessing chloride-selective ion channels. In recordings from voltage-clamped oocytes held at -60 mV, GABA induced an inward current, whereas application of the chloride-channel blocker picrotoxin (in the absence of agonist) elicited an apparent outward current. Single channel recordings obtained from cell-attached patches have revealed a single population of approximately 20 pS channels, with an open probability greater than 90% (at a pipette potential of -100 mV) in the absence of GABA. The relationship between single channel current and pipette potential was linear over the studied range (-100 mV to +60 mV), but the open probability was less for hyperpolarizations than for depolarizations. The spontaneous channel openings were blocked by micromolar concentrations of picrotoxin. Functional hetero-oligomeric receptors were formed when the molluscan subunit was co-expressed in oocytes with the bovine GABAA receptor alpha 1-subunit, but the channels gated by these receptors did not open spontaneously.

Quantal and non-quantal ACh release at developing Xenopus neuromuscular junctions in culture.

1. Single acetylcholine receptor (AChR) channel openings, detected by the whole-cell patch clamp technique, were used to monitor quantal and non-quantal ACh release at synapses in 1- and 2-day-old co-cultures of Xenopus embryonic motoneurons and muscle cells. motoneuron growth cones in ways that presumably reflect muscle-nerve inductive influences and the development of neurotransmitter release mechanisms. 2. Miniature endplate currents (MEPCs) occurred at a mean frequency of approximately 0.6 s-1 with a skewed distribution and mean amplitude of about twenty channel openings. In addition, occasional brief episodes of rapid deviations in the baseline were observed in some cells, with mean amplitudes of 4-8 pA and durations of a few hundred milliseconds. However, these episodes did not closely resemble summated openings of AChR channels. Moreover, where tested, these episodes were not blocked by curare; and comparable episodes were seen in an uninnervated myocyte. Thus they appear not to reflect ACh release from the nerve terminal. 3. Single-channel openings that might have been responses to non-quantal release of ACh were observed at rates of 0.9-12.3 min-1 (mean 3.0 min-1), only 1-5 times the rate of spontaneous AChR channel openings in uninnervated myocytes (mean 1.4 min-1). 4. We conclude that there is no significant non-quantal ACh leak from the presynaptic contacts in these immature synapses under these culture conditions. This is in disagreement with other, less direct, experimental reports, but consistent with findings in mature frog motor nerve terminals.

Adenosine 5'‐triphosphate increases acetylcholine channel opening frequency in rat skeletal muscle.

1. The effects of extracellular adenosine 5'-triphosphate (ATP) on the acetylcholine (ACh) channel opening activity was studied in dissociated rat skeletal muscle cells using patch-clamp recording techniques in cell-attached configuration. 2. With 10 microM-ATP in the recording pipette, the spontaneous opening frequency on the the alpha-bungarotoxin-sensitive ACh channel increased significantly from 0.3 to 4.7 s-1, although the opening frequency was not as high as that activated by very low (0.4 microM) ACh concentrations (64 s-1). 3. Spontaneous ACh channel openings, and ATP-associated and ACh-activated channel openings had similar single-channel conductances, 55, 59 and 56 pS, respectively. 4. ATP-associated events and spontaneous ACh channel opening events had similar mean channel open durations (0.6 ms); however, these values were considerably shorter than the duration of ACh-activated events (2 ms). 5. Pre-treatment with alpha-bungarotoxin (100 nM) blocked spontaneous ACh channel openings, ATP-associated openings and ACh-activated openings. 6. When delivered through a separate drug pipette after the formation of a gigaseal, ATP increased ACh-activated single-channel open probability in a dose-dependent fashion. 7. The increase in channel open probability was due primarily to the increase in channel opening frequency. ATP did not significantly alter the mean channel open duration or the single-channel conductance. 8. The ATP analogue adenosine-5'-O-(3-thiotriphosphate) (ATP-gamma-S) also enhanced ACh-activated channel open probability with relatively less potency. ADP, AMP and adenosine (up to 1 mM) did not significantly increase ACh channel open probability. 9. It is concluded that ATP in the micromolar range facilitates both spontaneous and agonist-activated ACh channel opening. The facilitation is due to ATP itself and not to products of ATP hydrolysis. The facilitatory actions of ATP on ACh channels are manifested by the increase in the channel opening frequency, and they may be mediated by an intracellular second messenger.

G-proteins mediate intestinal chloride channel activation

The localization of several GTP-binding regulatory proteins in teh apical membrane of intestinal epithelial cells has prompted us to investigate a possible role for G-proteins as modulators of apical Cl- channels. In membrane vesicles isolated from rat small intestine or human HT29-cl.19A colon carcinoma cells, the entrapment of guanosine 5'-O-(3-thiophosphate (GTP gamma S) led to a large increase in Cl- conductance, as evidenced by an increased 125I- uptake and faster SPQ quenching. The enhancement was observed in the presence, but not in the absence of the K+ ionophore valinomycin, indicating that the increased Cl- permeability is not secondary to the opening of K+ channels. The effect of GTP gamma S was counteracted by guanosine 5'-O-(2-thiophosphate (GDP beta S) and appeared to be independent of cytosolic messengers, including ATP, cAMP, and Ca2+, suggesting that protein phosphorylation and/or phospholipase C activation is not involved. Patch clamp analysis of apical membrane patches of HT29-cl.19A colonocytes revealed a GTP gamma S-activated, inwardly rectifying, anion-selective channel with a unitary conductance of 20 +/- 4 pS. No spontaneous channel openings were observed in the absence of GTP gamma S, while the open time probability (Po) increases dramatically to 0.81 +/- 0.09 upon addition with GTP gamma S. Since the electrophysiological characteristics and regulatory properties of this channel are markedly different from those of the more widely studied cAMP/protein kinase A-operated channel, we propose the existence of a separate Cl(-)-selective ion channel in the apical border of intestinal epithelial cells. Our results suggest an alternative regulatory pathway in transepithelial salt transport and a possible site for anomalous channel regulation as observed in cystic fibrosis patients.

Axolemmal Abnormalities in Myelin Mutantsa

Evidence is reviewed that the paranodal axoglial junction plays important roles in the differentiation and function of myelinated axons. In myelin-deficient axons, ion flux across the axolemma is greater than that in myelinated fibers because a larger proportion of the axolemma is active during continuous, as opposed to saltatory, conduction. In addition, older myelin-deficient rats that have developed spontaneous seizures display small foci of node-like E-face particle accumulations in CNS axons as well as more diffuse regions of increased particle density and number. Assuming that the E-face particles represent sodium channels, such regions could underlie high sodium current density during activity, low threshold for excitation, and increased extracellular potassium accumulation. Depending on the degree of spontaneous channel opening, they could also represent sites of spontaneous generation of activity. The appearance of seizures and their gradual increase in frequency and severity could represent an increase in the number of such regions. In addition, diminution in the dimensions of the extracellular space during maturation would result in increased extracellular resistance, which, together with increasing axonal diameter, would tend to increase the likelihood of ephaptic interaction among neighboring axons as well as the likelihood of extracellular potassium rises to levels that could cause spontaneous activity.

Single calcium-activated potassium channel in cultured mammary epithelial cells

The properties of Ca2+-activated K+ channels in mouse mammary epithelial cells in primary culture were studied by the patch-clamp technique. In cell-attached patches, spontaneous channel openings were sometimes observed; the slope conductance of the currents was about 12 pS at negative membrane potentials with a physiological solution (152 mM Na+, 5.4 mM K+) in the pipette. External application of A23187, a calcium ionophore, activated this channel. In excised inside-out patches, the channel was activated by increasing the internal Ca2+ concentration (10(-7) to 10(-6) M). No voltage dependence of the channel was activated was observed. Internal Na+ blocked the outward K+ current in a voltage dependent manner and this block led to the non-linear I-V relationship at positive membrane potentials. The channel was blocked by internal Ba2+ (0.1 mM) and tetraethylammonium (TEA+, 20-50 mM). Ba2+ reduced the open probability but not the single channel conductance, whereas TEA+ reduced the single channel conductance. The single channel conductance of this channel, measured from the inward current with a high-K+ solution (150 mM K+) in the pipette, was large (about 40 pS), and showed inward rectification. These results suggest that this channel is different from the usual small conductance Ca2+-activated K+ channels observed in many other cells.

Single- and multi-channel currents recorded with patch electrodes in mouse eggs

Zona-free mouse eggs are amenable to patch recording as well as to whole-cell recording (Bland et al., 1984; Peres, 1986a). Although many patches are silent, others show spontaneous channel opening; the single channel current is 1.5 pA and inwardly directed at resting potential (RP). Current amplitudes distribute normally as a single population at RP, however at RP — 50 mV the amplitude histogram indicates two channel populations. Open time distribution is exponential with a mean open time between 4 and 7 msec at RP. In three cases out of thirty-eight, in which depolarizations were given from a holding potential of RP — 50 mV, microscopic currents very similar in kinetics and voltage-dependence to the whole-cell Ca 2+ current were observed. No single-channel currents could be resolved in these traces. The results reported here indicate that the voltage-dependent Ca 2+ current of the mouse egg goes through low-conductance channels located in high density spots on the egg surface.

Sidedness of reconstituted calcium channels from muscle transverse tubules as determined by D600 and D890 blockade

The verapamil-type calcium antagonist, D600, and its charged quaternary derivative, D890, were used to assess the sidedness of blockade in single calcium channels reconstituted from purified transverse tubules of skeletal muscle. Spontaneous single channel openings were induced with the agonist Bay-K8644 and recordings were made in a two-chamber planar bilayer setup so that drugs could be delivered to either side of the channel. Micromolar drug addition resulted in a greater than 10-fold decrease in probability of open channel events (po) without a significant change in single channel currents. Changes in po occurred in parallel with changes in mean open time and both parameters could be titrated with a similar IC50. At pH 7.2, cis or trans D600 blocked with an IC50 of 5 microM but for D890 the IC50 was cis 3 microM and trans greater than 75 microM (cis is the intracellular-equivalent side as defined by the voltage-dependent activation). The asymmetry of D890 blockade indicates that the drug can readily gain access to the blocking site from the aqueous phase adjacent to the inner but not extracellular end of the channel.