Author's present address J. A. Sim: Department of Neurobiology, Babraham Institute, Babraham Hall, Cambridge CB2 4AT, UK.
Corresponding authors A. A. Selyanko: Department of Pharmacology, University College London, Gower Street, London WC1E 6BT, UK. Email: email@example.com
1Whole-cell perforated-patch recording from cultured CA1-CA3 pyramidal neurones from neonatal rat hippocampus (20-22 °C; [K+]o= 2.5 mM) revealed two previously recorded non-inactivating (sustained) K+ outward currents: a voltage-independent ‘leak’ current (Ileak) operating at all negative potentials, and, at potentials ≥−60 mV, a time- and voltage-dependent ‘M-current’ (IK(M)). Both were inhibited by 1 mM Ba2+ or 10 μM oxotremorine-M (Oxo-M). In ruptured-patch recording using Ca2+-free pipette solution, Ileak was strongly enhanced, and was inhibited by 1 mM Ba2+ but unaffected by 0.5 mM 4-aminopyridine (4-AP), 1 mM tetraethylammonium (TEA) or 1-10 nM margatoxin.
2Single channels underlying these currents were sought in cell-attached patch recordings. A single class of channels of conductance ≈7 pS showing sustained activity at resting potential and above was identified. These normally had a very low open probability (Po < 0.1), which, however, showed a dramatic and reversible increase (to about 0.9 at ≈0 mV) following the removal of Ca2+ from the bath. Under these (Ca2+-free) conditions, single-channel Po showed both voltage-dependent and voltage-independent components on patch depolarization from resting potential. The mean activation curve was fitted by a modified Boltzmann equation. When tested, all channels were reversibly inhibited by addition of 10 μM Oxo-M to the bath solution.
3The channels maintained their high Po in patches excised in inside-out mode into a Ca2+-free internal solution and were strongly inhibited by application of Ca2+ to the inner face of the membrane (IC50= 122 nM); this inhibition was observed in the absence of MgATP, and therefore was direct and unrelated to channel phosphorylation/dephosphorylation.
4Channels in patches excised in outside-out mode were blocked by 1 mM Ba2+ but were unaffected by 4-AP or TEA.
5Channels in cell-attached patches were inhibited after single spikes, yielding inward ensemble currents lasting several hundred milliseconds. This was prevented in Ca2+-free solution, implying that it was due to Ca2+ entry.
6The properties of these channels (block by internal Ca2+ and external Oxo-M and Ba2+, and the presence of both voltage-dependent and voltage-independent components in their Po/V relationship) show points of resemblance to those expected for channels associated with both Ileak and IK(M) components of the sustained macroscopic currents. For this reason we designate them Ksust (‘sustained current’) channels. Inhibition of these channels by Ca2+ entry during an action potential may account for some forms of Ca2+-dependent after-depolarization. Their high sensitivity to internal Ca2+ may provide a new, positive feedback mechanism for cell excitation operating at low (near-resting) [Ca2+]i.
Using rat hippocampal pyramidal neurones maintained in culture, we have identified a single species of channel which exhibits several features common to those expected for channels associated with both of these macroscopic currents (including inhibition by muscarinic receptor stimulation). We therefore provisionally refer to them as Ksust (‘sustained current’) channels, and address their correspondence with channels expected for macroscopic currents in the Discussion. We provide evidence to show that these channels are tonically inhibited by resting intracellular Ca2+ and transiently inhibited by Ca2+ influx during the spike.
Experiments were performed on dispersed 8- to 25-day-old cultured pyramidal neurones from CA1-CA3 regions prepared as described in Alger, Sim & Brown (1994). Briefly, CA1-CA3 regions were dissected from 400 μm thick coronal slices of hippocampus from neonatal Sprague-Dawley rats (15-20 g) killed by decapitation. The tissues were incubated in trypsin (1.25 mg ml−1) and mechanically dissociated by trituration. Cell suspensions were centrifuged and the pellet resuspended in growth medium containing: Dulbecco's modified Eagle's medium-Leibowitz-15 medium (1 : 1), 10 % fetal calf serum, 2 mM L-glutamine, 30 mM D-glucose and 18 mM NaHCO3. Hippocampal neurones were plated on poly-L-lysine-coated plastic culture dishes (Nunc) and cultured at 37°C in a humidified atmosphere containing 5 % CO2. Glial cell growth was controlled by treatment with 10 μM cytosine β-d-arabinofuranoside (Ara-C) 2 days after plating. Pyramidal neurones were identified on the basis of their morphology.
Whole-cell and single-channel currents were recorded using different modifications of the patch-clamp method (Hamill, Marty, Neher, Sakmann & Sigworth, 1981). Cells were continuously perfused with the following standard bath solution (mM): 144 NaCl, 2.5 KCl, 2 CaCl2, 0.5 MgCl2, 5 Hepes, 10 glucose, pH 7.4 with Tris base. All experiments were conducted at room temperature (20-22°C).
Two methods of whole-cell recording were used: (i) perforated-patch technique (Rae, Cooper, Gates & Watsky, 1991) to maintain the cell's intracellular medium undisturbed, in particular its resting [Ca2+] level, and (ii) the conventional whole-cell recording technique for dialysis of the cell with an artificial ‘intracellular’ solution (Hamill et al. 1981). For whole-cell recording, pipettes were filled with internal solution of the following composition (mM): 175 KCl, 5 Hepes, 1 BAPTA, pH 7.2 with NaOH. For perforated-patch recording, this solution was supplemented with amphotericin B (5 μg ml−1).
K+ channel activity was recorded in cell-attached, inside-out and outside-out modes. In all these configurations, the external side of the membrane was exposed to the standard extracellular solution (see above) containing 2 mM Ca2+, 0.5 mM Mg2+and 2.5 mM K+. In cell-attached experiments, several modifications to the solution bathing the cell were used. Thus, the removal of Ca2+ (compensated by addition of 6 mM Mg2+) was tested either at low (2.5 mM) or high (25 mM) K+. The latter was used to ‘clamp’ the membrane at the depolarized level of the K+ equilibrium potential (EK) and no indication was found that the cell resting membrane potential changed significantly in response to variation in [Ca2+] in 2.5 mM K+ (see Results). Oxotremorine-M (Oxo-M) was tested in the solution containing 0 mM Ca2+-6 mM Mg2+-25 mM K+ (see Selyanko, Stansfeld & Brown, 1992).
In whole-cell, inside-out and outside-out experiments, the internal side of the membrane was exposed to the standard, nominally Ca2+-free ‘internal’ solution (mM): 175 KCl, 5 Hepes, 1 BAPTA, pH 7.2 with NaOH. In the solutions tested in inside-out configuration, free [Ca2+] was set at the desired levels using program React, version 2.01 (G. L. Smith, Department of Physiology, University of Glasgow, UK). In some specially noted experiments free [Ca2+] was set at 50 nM and the effect of raising [Ca2+] to 500 nM was tested in the presence of 3 mM ATP-Na+-0.3 mM GTP-Na+-0.5 mM free Mg2+.
Recording pipettes were manufactured from borosilicate glass capillaries (Clark Electromedical Instruments, Reading, UK), either standard-walled (1.5 mm o.d., 0.86 mm i.d., for single-channel recording) or thin-walled (1.5 mm o.d., 1.17 mm i.d., for whole-cell recording). They were coated with Sylgard and fire polished. Tip resistance was 10-20 and 3-5 MΩ for single-channel and whole-cell recordings, respectively.
Solution change and drug application
Ca2+ and Oxo-M were applied to the cell (in whole-cell or cell-attached recording modes) by bath perfusion (perfusion rate, 0.5 ml s−1; dead space, 2 ml). Solutions perfusing excised inside- or outside-out membrane patches were switched by a motor-driven stepper (solution change, < 10 ms; estimated by measuring current offset produced in a pipette by solutions with different saline concentrations).
Data acquisition and analysis
Data were acquired and analysed using pCLAMP software (version 6.0.3; Axon Instruments). Currents were recorded using an Axopatch 200A patch-clamp amplifier (Axon Instruments), filtered at 1 kHz and digitized at 4 kHz (further filtered, to 500 Hz, during the analysis and for illustrations). The analysis of single-channel records was identical to that previously described (Selyanko & Brown, 1996).
Drugs and chemicals
All culture media and solutions were purchased from Life Technologies, Paisley, UK; D-glucose from Merck, Poole, UK; Ara-C, poly-L-lysine, 4-aminopyridine (4-AP), tetrodotoxin (TTX) and barium chloride from Sigma; Oxo-M from Semat Technical, Herts, UK; tetraethylammonium chloride (TEA) from Lancaster Synthesis, Morecambe, UK; margatoxin (MgTX) from Peptide Institute, Japan; and (±)-1-aminocyclopentane-trans-1,3-dicarboxylic acid (t-ACPD) from Tocris Cookson, Bristol, UK.
Data are expressed as means ±s.e.m., unless indicated otherwise, and statistical significance was assessed using Student's two-tailed t test. P < 0.05 was considered significant.
Macroscopic (whole-cell) currents
Previous experiments have shown that hippocampal pyramidal neurones in culture can express both IK(M) (Gähwiler & Brown, 1985; Fiszman et al. 1991) and Ileak (Andrade & Nicoll, 1987), similar to those recorded from hippocampal slices (Halliwell & Adams, 1982; Andrade & Nicoll, 1987; Madison et al. 1987) or freshly dissociated hippocampal pyramidal neurones (Harata et al. 1996). Our initial experiments were performed to test for the presence of these two forms of sustained (non-inactivating) outward current in hippocampal pyramidal neurones cultured under our experimental conditions. We used two configurations of the patch-electrode method - amphotericin perforated-patch recording (Rae et al. 1991) and ruptured-patch recording with a Ca2+-free pipette solution. Apart from diffusive exchange of small ions, the former configuration leaves the cytoplasm relatively undisturbed, and so approximates the conditions we used for subsequent cell-attached pipette recording of single-channel currents; the ruptured-patch approach more closely resembles the conditions used when channels were recorded in membrane patches that were subsequently excised (in inside-out configuration) into a Ca2+-free solution (see below).
Two voltage protocols were used to identify the two components of sustained outward current. The time-dependent IK(M) component was detected in the form of inward current relaxations (signifying current deactivation) when hyperpolarizing step commands were applied to between -40 and -80 mV from a holding potential of -30 mV (Fig. 1A; cf. Halliwell & Adams, 1982; Gähwiler & Brown, 1985). On the other hand, the leak component could be observed (in isolation from IK(M)) when the membrane potential was held at a more negative level (-80 mV) where no time- or voltage-dependent K+ currents were present (Fig. 1B). This holding potential was close to the resting (zero-current) potential in these cells (-75.9 ± 2.7 mV; range, between -64 and -85 mV; n= 7).
Both time-dependent and time-independent components were reversibly inhibited by 10 μM Oxo-M (Fig. 1A and B). When quasi-steady-state I-V relationships were obtained from the records shown in Fig. 1Aa and Ab, Oxo-M clearly reduced the slope conductance, measured in both an outwardly rectifying IK(M) range (between -30 and -70 mV) and a linear Ileak range (at negative potentials) (Fig. 1Ca). The ramp currents obtained from the same cell in the presence of 1 mM Ba2+ showed that Ba2+ was capable of mimicking the effects of Oxo-M on the I-V relationship (Fig. 1C b).
The I-V relationships obtained in the presence of Oxo-M or Ba2+ crossed the control I-V curve at about -100 mV (Fig. 1Ca and Cb). This cross-over potential was close to the calculated EK (-107 mV) and was far away from the equilibrium potentials set for other physiological ions (0 mV or more positive potentials), which suggests that the sustained current inhibited by Oxo-M and Ba2+ is generated by K+ channels. Similar results to those shown in Fig. 1 were obtained in six other cells tested (inhibition by Ba2+ in three cells, inhibition by Oxo-M in five cells; small inhibition by 10 μM t-ACPD in four cells; no effect of 100 μM UTP in two cells). Thus, the channels involved in generating IK(M) and Ileak were present in the cells studied and were functional under our experimental conditions.
Ruptured-patch (Ca2+-free) recording
Under these conditions, the zero-current (resting) potential measured immediately after breaking through the membrane was -72.1 ± 0.9 mV (n= 8), and became progressively more negative with time of recording. During voltage-clamp recording, a corresponding time-independent outward current developed when the cell was held at -70 mV, near to its initial resting potential (Fig. 2Aa-Ac). This current was larger at depolarized (-50 mV) than at hyperpolarized (-90 mV) potentials and, therefore, resulted from an increase in the resting membrane conductance and a shift in zero-current potential towards a more negative level, close to EK. Both the initial (Fig. 2Aa) and the developed (Fig. 2Ac) currents were almost completely blocked by 1 mM Ba2+ (Fig. 2Ad).
When the increase in the resting current was fully developed, the steady-state I-V relationship became almost linear over a wide range of membrane potentials (-30 to -120 mV) with a superimposed inactivating outward current at potentials ≥ -30 mV (Fig. 2B). Under these conditions, the characteristic relaxations associated with IK(M) deactivation were not detected on stepping from -30 mV to more negative potentials (Fig. 2B). Ba2+ at 1 mM strongly suppressed the steady-state current (Fig. 2B); the cross-over potential for this response was close to the EK (-107 mV), suggesting it was mainly, if not exclusively, the K+ current which was inhibited by Ba2+. The inactivating current was insensitive to 1 mM Ba2+ (Fig. 2B).
Inhibition of the resting current by Ba2+ was both concentration- and voltage dependent (Fig. 2C a and Cb). A concentration of Ba2+ of 3 mM seemed to be capable of producing a complete block of the increased resting K+ current, with an IC50 value in the low micromolar range at all potentials tested. However, the inhibitory effect of Ba2+ was clearly enhanced by membrane hyperpolarization (see Fig. 2Cb).
On the other hand, 4-AP (0.5 mM; n= 2), TEA (1 mM; n= 2) and MgTX (1-10 nM; n= 2; a specific blocker of delayed rectifier K+ (Kv1.3) channels which presumably set the resting potential in T lymphocytes - see Defarias, Stevens & Leonard (1995) and references therein) had no effect (data not illustrated). The current was slightly inhibited by a Ca2+-free external solution, containing elevated (6 mM) Mg2+ and no added Ca2+. This inhibition was fast, fully reversible and was probably produced by the direct blocking effect of elevated Mg2+.
Single (Ksust) channel currents
Cell-attached patch recording
Non-inactivating (sustained) Ksust channels (see Introduction) were identified as K+ (outward-current) channels which could be detected at zero-patch potential (i.e. at the resting potential of the cell) and remained active, with no obvious inactivation, at depolarized patch potentials. They could be readily distinguished from small (∼3 pS) and large (∼70 pS) Ca2+-activated K+ channels (SKCa and BKCa channels, respectively), and also from high-threshold, inactivating (delayed rectifier) K+ channels (∼15 pS; see also Bossu & Gähwiler, 1996) recorded in these cells by their sustained activity and conductance (∼7 pS, see below).
The activity of these channels was normally very low: when measured in different patches at resting or depolarized potentials, the open probability (Po), although highly variable, was usually below 0.1. Under these conditions, it was not possible to reliably determine whether or not their activity was voltage dependent. However, the activity of Ksust channels was observed to be greatly increased when Ca2+ was removed from the extracellular (bath) solution (2 mM Ca2+ was present in the pipette throughout the experiment), and it was only then that the voltage dependence and other properties of Ksust channels could be precisely evaluated (see below).
Figure 3A shows the results of an experiment conducted at normal, 2.5 mM K+ when the patch was depolarized by 70 mV, to maximize K+ channel activity and to increase the single-channel current. Under these conditions a single Ksust channel was detected, which initially had a low Po (< 0.01) but whose activity clearly increased following a switch to a nominally Ca2+-free bath solution (mean Po increased to 0.26). The increase in activity was sustained and persisted as long as the recording, and was fully reversible on re-addition of Ca2+. Changes in Po occurred with long latencies of up to several minutes, i.e. they were much slower than the solution changes in the bath (see Methods). However, once started, the increase in Po was rapid (< 1 s; Fig. 3Ad).
Changes in channel parameters during the Ca2+ response are shown in Fig. 4A. Thus, inhibition by 2 mM Ca2+ in the bath was accompanied by the appearance of long gaps in the activity. At the same time, short bursts of openings persisted in the presence of Ca2+, and the open time was decreased and the frequency of opening was increased during these periods of activity (Fig. 4Ac and Ad), with no significant change in the current amplitude (Fig. 4Ae). Parameters of the activity increased by the removal of Ca2+ from the bath were not significantly affected by excision of the patch into a standard Ca2+-free internal solution (Fig. 4A).
Muscarinic sensitivity of Ksust channels was tested by the addition of 10 μM Oxo-M to the bath solution containing 25 mM K+ (to clamp the cell membrane potential) with no added Ca2+ (to avoid cell Ca2+ loading at depolarized potentials). Recordings under these conditions revealed that channel activity (increased due to the absence of Ca2+ in the bath) was completely and reversibly inhibited by Oxo-M (Fig. 3B). Similar results were obtained from two other patches tested.
Figure 4B illustrates the effect of Oxo-M on channel parameters. Interestingly, during the initial phase of Oxo-M-induced inhibition there was a transient increase in the frequency of opening (followed by a persistent reduction) and a clear reduction in the open time (dotted line, 1); the open time recovered following wash out of Oxo-M (dotted line, 2) (Fig. 4Bc and Bd).
Effect of Ca2+ on channel kinetics.
This was quantified by comparing shut and open time distributions in the presence and absence of Ca2+ in the bath (Fig. 5). Three shut times, short (τs1), medium (τs2) and long (τs3), and two open times, short (τo1) and long (τo2), were detected in the activity recorded in the presence of 2 mM Ca2+ in the bath (Fig. 5Aa and Ba and Table 1). The medium and long shut times were strongly reduced on removal of Ca2+ from the bath (Fig. 5Ab and Table 1) and the long shut times disappeared after excision of the patch into a Ca2+-free internal solution (Fig. 5Ac and Table 1). By contrast, the two open times remained unaffected by the removal of Ca2+ from the bath, whereas the long open time was somewhat prolonged after patch excision (Fig. 5B and Table 1). However, the contribution of the short open time component to the activity was considerably (about twofold) higher in the presence than in the absence of Ca2+ in the bath (Fig. 5B a and Bb and Table 1). Coupled with an increased frequency of opening during the bursts of activity in 2 mM Ca2+, this finding suggests that addition of Ca2+ induced a switch towards a greater prominence of short open times (though presumably this only contributes a minor component to change in Po).
Table 1. Kinetic components of the activity of Ca2+-inhibited K+ channels in cell-attached and excised inside-out patches
All the data were obtained from the same three patches. The data in cell-attached mode (at a bath concentration of 2 and 0 mM Ca2+) are means ± S.E.M. from three patches. The data in inside-out mode are individual measurements from two patches. Values in parentheses are the percentage contribution. Note that in inside-out mode, with the inner face of the patch exposed to the standard Ca2+-free internal solution, the longest shut times (τs3) were not detected.
2 mM Ca2+
0.6 ± 0.2
62.9 ± 28.7
1270 ± 758
1.5 ± 0.5
10.0 ± 4.0
(77.8 ± 12.0 %)
(18.2 ± 9.8 %)
(4.0 ± 2.3 %)
(50.9 ± 19.1 %)
(49.1 ± 19.1 %)
0.9 ± 0.3
5.0 ± 1.6
89.1 ± 50.9
1.5 ± 0.5
13.8 ± 4.0
(81.4 ± 7.5 %)
(12.1 ± 4.9 %)
(6.5 ± 2.7 %)
(28.1 ± 4.0 %)
(71.9 ± 4.0 %)
0.7 (93.7 %)
3.7 (6.3 %)
1.8 (2.9 %)
33.5 (97.1 %)
0.5 (93.6 %)
7.1 (6.4 %)
2.5 (1.1 %)
30.8 (98.9 %)
In eight cell-attached patches recorded from cells exposed to 0 and 2 mM Ca2+, at a pipette potential of -70 to -80 mV, the mean Po increased from 0.06 ± 0.03 to 0.63 ± 0.1 (P << 0.01) whereas single-channel current amplitude (i) did not change significantly (from 0.59 ± 0.07 to 0.63 ± 0.1 pA) in response to a switch from 2 to 0 mM Ca2+ bath solution. The lack of change in i indicates that there was no significant change in cell membrane potential under these conditions. When measured in all eighteen patches studied, the values of i and Po were 0.68 ± 0.05 pA and 0.07 ± 0.03, and 0.66 ± 0.04 pA and 0.67 ± 0.03, in 2 and 0 mM Ca2+, respectively.
Voltage dependence of channel activation.
In seven cell-attached patches single Ksust channels were held for a sufficient time for their voltage dependence and kinetics to be examined. Channel activity was increased by using Ca2+-free bath solution containing low, 2.5 mM K+. Figure 6A shows examples of single-channel activity recorded at different levels of patch depolarization from cell resting membrane potential which, on the basis of measurements of resting potential made using the perforated-patch technique (see above), was assumed to be ∼-80 mV. The channel was partially active at resting membrane potential and was further activated by patch depolarization; maximal activation was achieved at a patch depolarization of 80 mV, i.e. at a membrane potential of ∼0 mV. The current distribution had only one component (Fig. 6B), indicating the channel had only one conductance level.
Consistent with the Goldman-Hodgkin-Katz equation, the i-V relationship showed an outward rectification with permeability, P, equal to 4.5 × 10−14 cm3 s−1 in this patch or, on average, (4.3 ± 0.3) × 10−14 cm3 s−1 in seven patches studied (Fig. 6C). Assuming an EK in these cells of ∼-110 mV and taking into account single-channel current at a pipette potential of ∼-80 mV, the chord conductance at a membrane potential of 0 mV was calculated to be 8.6 pS in this patch and, on average, 6.7 ± 0.46 pS in sixteen patches studied. Stability plots for Po demonstrated the lack of channel inactivation and an increase in Po with membrane depolarization.
Figure 7A shows Po-Vpatch (Vpatch, patch depolarization) relationships for each of the seven Ksust channels studied including that shown in Fig. 6 (Fig. 7Af). All channels showed a voltage-dependent increase in Po, though with considerable variation in half-activation voltage (V½) which ranged between 26 and 77 mV patch depolarization (see Table 2). Because resting membrane potential (see above) and i in different patches showed a limited variation (see Table 2), it seems likely that the scatter in V½ can only be explained in part by the variation in the resting potential among different cells (though this cannot be excluded since we did not directly measure resting membrane potential in these seven cells).
Table 2. Parameters of voltage-dependent activation of single Ksust channels recorded from seven cell-attached patches
The activity was increased by using Ca2+-free bath solution. Parameters V½, Po,max, a and c were obtained by fitting Po-Vpatch relationships (see Fig. 7Aa-g) with the equation:
where c is the value of Po at potentials ≤ resting potential, Po,max is maximum voltage-dependent open probability, V½ is half-activation voltage, a is the slope factor, and Vpatch is patch depolarization (see text for details). i is the mean current measured at a patch depolarization of 80 mV.
Interestingly, in four out of seven patches, Po did not diminish to zero at resting potential but remained significantly above zero (0.1-0.57) as seen in Fig. 7. Because of the high Po at depolarized potentials, the apparent voltage-independent activity remaining at or near resting potential could be unequivocally attributed to the same (single) channel that exhibited an increase in activity with depolarization. Figure 7Ba shows superimposed Boltzmann plots of all seven patches (see Table 2 for the parameters of each fit) and Fig. 7Bb shows the mean plot fitted with a modified Boltzmann equation:
where Vpatch is patch depolarization, V½ is 46.3 mV, slope factor (a) is 12.0 mV, maximum voltage-dependent open probability (Po,max) is 0.61, and c (value of Po at potentials ≤ resting potential) is 0.21.
Single-channel activity contained multiple kinetic components: at least three shut (τs1, τs2, τs3) and two open (τo1 and τo2) components (see histograms in Fig. 8A and B). The shut times were reduced (Fig. 8Ca), whereas the open times were increased (Fig. 8C b), by patch depolarization. When the channels were fully activated (at ∼80 mV patch depolarization, see Fig. 7), the mean values of τs1, τs2, τs3, τo1 and τo2 (percentage contributions in parentheses) were found to be 0.7 ± 0.1 ms (80.5 ± 4.0 %), 3.7 ± 0.7 ms (15.1 ± 3.2 %), 42.1 ± 18.9 ms (4.4 ± 1.3 %), 2.2 ± 0.3 ms (23.7 ± 3.6 %) and 35.1 ± 8.3 ms (76.3 ± 3.6 %), respectively.
Po analysis. When Ksust channels were activated by moderate patch depolarization, they showed significant fluctuations in Po (data not shown). Because Ksust channels have two open times in their activity, these fluctuations might be due to a slow switching between two (high and low) modes of Po. We have therefore tested whether Ksust channels show this behaviour, and, if so, whether Ca2+ modulates modal gating. For this, runs analysis was conducted for Ksust channel activity recorded first at different membrane potentials (in the absence of Ca2+ in the bath) and, second, at the same potential in the presence and absence of Ca2+ in the bath.
When a single Ksust channel (recorded in cell-attached mode) was partially (Po, ∼0.5) activated by 30 mV patch depolarization, its Po value in 0.5 s segments varied greatly; however, at this potential the channel assumed no predominant level(s) of Po, i.e. it had no modal behaviour. Moreover, when the channel was fully activated (Po, ∼0.9) by strong (80 mV) patch depolarization, or deactivated at small (10 mV) patch depolarization, the channel assumed only one predominant Po level, close to 1 or to 0, respectively (data not shown). Similarly, when the channel was inhibited by 2 mM Ca2+ in the bath, it had only one Po level, close to 0, and this level shifted towards 1 following the removal of Ca2+ from the bath (data not shown). Also, after excision of the patch into a Ca2+-free solution the distribution of Po in 0.5 s segments retained only one peak, at Po close to 1 (not shown). Thus, Ksust channels did not exhibit two modes of Po in their activity and, therefore, their regulation by Ca2+ is not related to modal gating.
Inside-out patch recording: inhibition by intracellular Ca2+
In order to investigate the sensitivity of Ksust channels to internal Ca2+, cell-attached membrane patches containing single or multiple channels were excised into the standard (Ca2+-free) internal solution (see Methods) and a range of Ca2+ concentrations were applied to the patch, for several seconds, using a stepper device.
Figure 9A and B shows an example of the activity of four Ksust channels recorded from the same patch in cell-attached (A) and inside-out (B) modes. The activity had a low mean Po (< 0.05) in cell-attached mode (with 2 mM Ca2+ in the bath) but showed a sudden run-up (to nearly 1) several seconds after excision of the patch into a standard (Ca2+-free) internal solution. The activity was then completely and reversibly abolished by the addition of ≥ 175 nM intracellular Ca2+ (Ca2+1; Fig. 9B). Similar results (increase in activity of the channel and its inhibition by low concentrations of Ca2+) were obtained from all nine patches tested.
In three inside-out patches examined, Ksust channels were studied in a ‘physiological’ internal solution containing 50 nM free Ca2+, 0.5 mM free Mg2+, 3 mM ATP-Na+ and 0.3 mM GTP-Na+. Under these conditions, the activity of Ksust channels persisted and elevation of [Ca2+] to 500 nM was still capable of producing a complete and fully reversible block in all patches studied (Fig. 9C). Ksust channels remained active not only in Mg2+ and ATP but also in spermine (100 mM; n= 2; data not shown). These findings, coupled with the low channel conductance, clearly differentiated Ksust channels from inward rectifier K+ (KIR) channels (for a review see Doupnik, Lim, Kofuji, Davidson & Lester, 1995).
The concentration dependence of Ca2+-induced inhibition was studied in nine multichannel patches using a standard internal solution. In each patch, several different concentrations of Ca2+ were tested as exemplified in Fig. 9B and pooled data are shown in Fig. 9D. The mean concentration dependence of inhibition was fitted by the equation:
where A is the maximal inhibition of channel activity and nH is the Hill coefficient. Least-squares values for A, IC50 and nH were 100 %, 122 nM and 5.4, respectively. It should be noted that, in all patches tested, the activity was almost completely inhibited at [Ca2+]i≥ 175 nM.
The conductance of Ca2+-inhibited K+ channels obtained from excised inside-out membrane patches was similar to that obtained from cell-attached membrane patches. Thus, single-channel current (i) was 0.89 ± 0.02 pA at 0 mV (n= 12) and taking into account electrical driving force (membrane potential - EK= 107 mV) the chord conductance at this potential was 8.4 ± 0.2 pS (as measured from the i-V relationships obtained from four patches; the mean channel permeability, P, was 5.1 ± 0.2 cm3 s−1). These values were similar to those obtained from cell-attached patches at comparable potentials (see above).
Outside-out patch recording: K+ channel blockers
When Ksust channels were recorded from six excised outside-out membrane patches, their characteristics (conductance, lack of inactivation, increased activity in a reduced [Ca2+]i) were similar to those recorded in cell-attached and inside-out modes. Their activity was unaffected by 1 mM TEA (n= 2) or 0.5 mM 4-AP (n= 2) (not shown), but was effectively reduced by 1 mM Ba2+ (mean inhibition, 91.3 ± 1.1 %; n= 4; see Fig. 10a), which appeared to produce slow (non-flickering) block of Ksust channels (Fig. 10B).
When cell-attached recordings were made in control bath solution containing 2 mM Ca2+-0.5 mM Mg2+-2.5 mM K+ a small proportion of patches (11/55) showed spontaneous invasive spikes in the form of large biphasic transients. These appeared irregularly and could be blocked by addition of 0.1 μM TTX to the bath. In other cells, spikes were very infrequent or absent but cell firing could be transiently induced by bath perfusion with 25 mM K+. In three of eleven cell-attached patches showing spontaneous spiking activity in 2 mM Ca2+, Ksust channels had a high level of activity, which was clearly inhibited by firing.
Figure 11A-C demonstrates that a multichannel activity can clearly be reduced following bursts of spiking activity (Fig. 11A, B and Ca); this inhibition was completely and reversibly abolished by Ca2+-free-high-Mg2+ external solution, despite the increased firing rate recorded under these conditions (Fig. 11Cb). (This increase in firing in 0 mM Ca2+-6 mM Mg2+ was not itself a consistent observation. Thus, among 9 patches with spiking activity in control, only 3 showed an increase in firing, whereas 5 showed a decrease and 1 no change in response to 0 mM Ca2+-6 mM Mg2+. Similarly, among 44 ‘silent’ control patches, only 7 showed the spiking, whereas 28 remained silent on switching to 0 mM Ca2+-6 mM Mg2+.) These channels also became fully active following excision of the patch into a Ca2+-free internal solution and were completely blocked by ≥ 175 nM Ca2+1 (Fig. 11D). These findings suggest that the Ca2+ influx into the cell, resulting from cell firing, may be responsible for the inhibition of Ksust channels.
Therefore the question arising from these results is whether a single spike is capable of inhibiting Ksust channels, and if so, what is the time course of this inhibition. To answer this question, average records with spikes were obtained at a depolarized patch potential of 60 mV, when the channels were strongly activated by voltage and the channel current was maximized (Fig. 12A), and at 0 mV patch depolarization (i.e. cell resting potential), when the channel activity was low and the channel current was small (Fig. 12B). The difference current shows that a single spike was followed by a significant (40-50 %) inhibition of Ksust channels and that this response had a fast (∼40 ms) growth phase and a much slower (> 500 ms) decay (see Fig. 12C). Similar curves for the difference current were obtained at more depolarized (90-100 mV) patch potentials (data not shown).
The main result in the present study is the identification of non-inactivating K+ channels which have not been previously reported in hippocampal pyramidal neurones. The characteristic features of these channels are: (i) the presence of both voltage-independent and voltage-dependent components in their activity; (ii) lack of inactivation, even at positive membrane potentials; (iii) tonic inhibition at resting intracellular [Ca2+]; (iv) transient inhibition by Ca2+ influx during the spike, which probably does not require channel phosphorylation/dephosphorylation; (v) inhibition via activation of muscarinic acetylcholine receptors, mediated by a diffusible second messenger; and (vi) inhibition by external Ba2+ but not by TEA or 4-AP. We have thus far provisionally termed them Ksust (sustained current) channels: their possible relationship to channels expected for observed macroscopic currents in these cells is addressed below.
KM (M-) channels?
Some of the steady-state properties of Ksust channels clearly resemble those of channels previously identified as KM channels (M-channels) in rat sympathetic neurones (Selyanko et al. 1992; Selyanko & Brown, 1993, 1996). Thus, these latter KM channels have a comparable conductance in asymmetric [K+] (7-11 pS), gating kinetics (3 shut and 2 open times), steady-state voltage dependence, and sensitivity to [Ca2+]i and muscarine. Also, the enhancement of channel activity in cell-attached patches on removal of external Ca2+ resembles the increased macroscopic IK(M) amplitude previously observed in hippocampal pyramidal cells on the addition of Cd2+ ions (Gähwiler & Brown, 1985); both effects might reasonably have resulted from reduced Ca2+ entry and consequent fall in intracellular [Ca2+]. The mean half-maximal potential for the voltage-dependent component of Ksust channel activation (-46 mV: Fig. 7Bb) is ∼10 mV more negative than that for KM channel activation in sympathetic neurones. However, the activation curve for the macroscopic IK(M) in hippocampal neurones has never been accurately measured, but since the resting potential of hippocampal neurones is more negative than that of sympathetic neurones (about -76 mV in the present experiments using amphotericin-perforated patches versus−60 mV in perforated-patched sympathetic neurones; Selyanko et al. 1992), a corresponding negative shift of the IK(M) activation curve would not be unreasonable.
Nevertheless, two factors deter us from labelling these channels unequivocally as KM channels. First, many of the channels recorded in cell-attached mode were open at resting potential. This we observed (although at very low frequency) even in normal (2 mM) Ca2+ solution - indeed, this was how we first detected them - but it was more obvious in a Ca2+-free solution when the overall activity was enhanced (Fig. 7). While a low level of activity around resting potential (in high [K+] pipette solution; Stansfeld, Marsh, Gibb & Brown, 1993), or down to -70 mV in excised patches exposed to a Ca2+-free internal solution (Selyanko & Brown, 1996), has been noted in some previous recordings of sympathetic neurone KM channels, the amount of ‘resting’ activity observed in the present experiments seems greater than one might expect for conventional KM channels. Further, some at least of our Ksust channels showed what appeared to be a voltage-independent ‘pedestal’ of open probability at negative potentials (see Fig. 7Ab, Ae, Af and Ag), and a constant (voltage-independent) value of Po of 0.21 had to be added to obtain a reasonable Boltzmann fit to the pooled data (Fig. 7B b). (Since the mean curve might be distorted by the variability in the activation voltage range in different patches (possible reasons for which are discussed below), it would clearly be helpful to know whether this pedestal of activity persists at potentials negative to resting potential; this we have not yet determined, because of the small current amplitude at negative potentials.) Second, and most crucially, IK(M) is defined not solely through its voltage sensitivity but also through time-dependent activation and deactivation following voltage steps; we have not yet been able to determine activation and deactivation kinetics of these hippocampal cell channels (which requires many steps and hence prolonged stability), so we cannot verify this essential property of a KM channel.
Thus, while noting that these channels are distinctly ‘KM like’ in many respects, until this essential kinetic information is available, and their voltage dependence (or otherwise) at potentials negative to rest and in normal external [Ca2+] has been clarified, it would seem best to continue to assign these channels the neutral term Ksust channels rather than KM channels.
The fact that a proportion of channels recorded in cell-attached patches showed a finite open probability at resting potential raises the question of the extent of their possible contribution to the macroscopic Ileak. It seems very likely that the enhanced Ileak seen in cells dialysed with a Ca2+-free pipette solution results from the increased activity of Ksust channels since: (i) Ksust channels in excised patches also showed increased activity in Ca2+-free solution, and (ii) both the enhanced Ileak and Ksust channels (in outside-out patches) were blocked by Ba2+ ions, but not by TEA or 4-AP. Zhang, Pennefather, Velumian, Tymianski, Charlton & Carlen (1995) have previously noted the development of a K+ current in hippocampal CA1 neurones on using a Ca2+-free BAPTA-containing internal solution, which they interpreted to result from activation of the after-hyperpolarization (AHP) current (i.e. opening of SKCa channels). We have detected SKCa channels in excised patches from these cells on application of intracellular Ca2+ but these had a distinctly lower conductance (∼3 pS) than Ksust channels, and we could not see any evidence for SKCa channel activity in cell-attached patches in Ca2+-free solution. Hence, activation of Ksust channels seems the more likely explanation.
Thus, one possibility arising from our observations is that both IK(M) and the muscarine- and Ba2+-sensitive component of Ileak might be carried by the same channels, but operating in different (voltage-dependent and -independent) modes. Previously, these two currents have been regarded as being quite distinct, not only on obvious kinetic grounds, but also because of their differential sensitivity to some receptor agonists (e.g. Madison et al. 1987; Benson et al. 1988; Dutar & Nicoll, 1988). Clearly, it would be important to know whether these differences persisted under our experimental conditions, and how these matched the sensitivity of Ksust channels as recorded in cell-attached patch electrodes. This is far from trivial and, together with further information on the time-dependent kinetics of Ksust channels and their voltage dependence at negative potentials, will require much more experimentation.
Other leak channels?
Several channels have been suggested to contribute to the resting membrane conductance, including SKCa, KIR and ATP-sensitive K+ (KATP) channels (see Storm, 1990, 1993). Ksust channels could be distinguished from SKCa channels by their high activity in low [Ca2+] (and different single-channel conductance), from KATP channels by their lower conductance, and from KIR channels by their outward rectification; also, Ksust channels were not blocked by MgTX, thus excluding Kv1.3 channels (which contribute to resting K+ conductance in lymphocytes; Defarias et al. 1995). We did not detect opening of any of these channels, or any other channels, at resting potential in our experiments.
Recently, a family of voltage-independent (baseline or leak) K+ channels have been identified. The cloned members have two pore domains in tandem and some of them are expressed abundantly in hippocampus and cortex (e.g. TREK-1, a TWIK-1- (tandem of P domains in a weak inward rectifying K+ channel) related K+ channel; Fink et al. 1996). Although they have a similar pharmacology (sensitivity to Ba2+ and insensitivity to other K+ channel blockers), they are insensitive to intracellular Ca2+. Thus, they do not seem to correspond to Ksust channels. (It is not yet known whether they are affected by muscarinic receptor stimulation.)
Ksust channels in cell-attached patches were tonically suppressed at the resting level of 2 mM external Ca2+ and those in inside-out patches were directly blocked by low (IC50, ∼100 nM) concentrations of free Ca2+. Does this blocking range of [Ca2+] correspond to the resting [Ca2+]i in hippocampal pyramidal neurones? Although cytosolic [Ca2+] has not been measured in the present experiments, a number of previous reports provide this kind of estimate both at rest and under various experimental conditions. Thus, the mean resting [Ca2+], measured by means of fluorescent Ca2+ indicators at the resting potential and in the presence of normal extracellular [Ca2+], was found to be rather low, 30 nM (Knöpfel, Charpak, Brown & Gähwiler, 1990) or 20-80 nM (Iijima, Kudo, Ogura, Akita & Matsumoto, 1990). Membrane depolarization to -30 mV (Knöpfel et al. 1990) or muscarinic stimulation (Iijima et al. 1990; Wakamori, Hidaka & Akaike, 1993) increased this mean level considerably, up to 500 nM. Although muscarinic inhibition of Ksust channels might possibly be mediated by these large [Ca2+] rises, tonic channel inhibition by resting [Ca2+]i is inconsistent with the Ca2+ measurements. Indeed, as illustrated in Fig. 9D, in inside-out patches the channels were completely blocked by ≥ 200 nM Ca2+, whereas ≤ 75 nM Ca2+ (estimated mean [Ca2+]i; see above) produced very little if any effect and at least 175 nM Ca2+ was required to produce a degree of inhibition seen in cell-attached patches under resting conditions. This raises the possibility that the local, near-membrane level of [Ca2+]i might be several times higher than its average level in the cytosol - perhaps due to a sustained influx of Ca2+ through some Ca2+-permeable channels open at rest and an active removal of Ca2+ from the cytoplasm by homeostatic mechanisms (see Selyanko & Brown, 1996). Consistent with this hypothesis are the following observations. First, in our cell-attached patches, in the presence of Ca2+ in the bath, not only Ksust channels were inhibited, but also BKCa channels were activated (at depolarized potentials; A. A. Selyanko, unpublished observations) and both these effects reversed on wash out of Ca2+. Similar results were obtained from rat sympathetic neurones where KM and BKCa channels changed their activities in response to variations in extracellular [Ca2+]; interestingly, these effects were not accompanied by any significant changes in the mean cytoplasmic [Ca2+] (Selyanko & Brown, 1996). Second, in excised inside-out patches elevated [Ca2+] (> 200 nM) was required to activate BKCa channels and to inhibit Ksust channels to the degree seen in cell-attached patches under resting conditions. Third, recent experiments on chick sensory neurones showed that, even during stimulation with Ca2+ ionophores, the level of [Ca2+] in the cytoplasm may return to control, despite the continuous influx of Ca2+ and elevated [Ca2+] within 1 μm of the plasmalemma (Bolsover, Kater & Guthrie, 1996). The source of Ca2+ maintaining high resting [Ca2+]i near the membrane is not yet known. However, according to a recent report on dihydropyridine-sensitive voltage-dependent Ca2+ channels in hippocampal pyramidal neurones, some of them are open at rest (∼-70 mV) and contribute to the resting level of [Ca2+]i (Magee, Avery, Christie & Johnston, 1996).
Interestingly, the voltage dependence of individual Ksust channels was highly variable (see Fig. 7). Because the activation of Ksust channels depended on [Ca2+]i this might be due to variation in [Ca2+]i in different cells or in different loci of the cell. Thus, it would be interesting to check these possibilities by assessing the voltage dependence at different free [Ca2+] using excised inside-out patches. However, due to the exceptionally long periods of recording from excised patches required for these studies and the possibility that the channel properties may alter with patch excision, we have not attempted these experiments. An alternative approach to the same problem would be to try to correlate the channel voltage dependence in on-cell patches with [Ca2+]i measurements in the same cells. Whatever the approach, both experiments are difficult and so have yet to be done.
One cannot exclude the possibility that in cell-attached conditions some Ca2+-dependent cytoplasmic/membrane regulatory mechanisms might contribute to the inhibitory action of resting [Ca2+] on Ksust channels. Further, although inhibition of Ksust channels by [Ca2+]o is most likely to be mediated by the Ca2+ influx and subsequent elevation in [Ca2+]i, an alternative/additional mechanism of action of [Ca2+]o should be considered. Thus, the Ca2+-sensing receptor, originally cloned from bovine parathyroid cells (Brown et al. 1993) and homologous to the metabotropic glutamate receptor, has been detected in the nerve terminals in hippocampus (Ruat, Molliver, Snowman & Snyder, 1995). This receptor is strongly activated by physiological [Ca2+]o and, therefore, it might be involved in [Ca2+]o-induced inhibition of Ksust channels. However, until now only activation of non-selective cation channels (not seen in the present experiments) has been reported in hippocampal pyramidal neurones in response to elevation of [Ca2+]o (Ye, Kanazirska, Quinn, Brown & Vassilev, 1996).
Physiological role of Ksust channels
In hippocampal pyramidal neurones and, if present, in other brain neurones, Ksust channels could contribute to the following important processes.
Maintaining the resting membrane potential
In hippocampal and other brain neurones the resting potential is significantly more negative than, for example, in autonomic neurones. This may be required to prevent cell Ca2+ loading through voltage-dependent Ca2+ channels or NMDA channels, which, at negative potentials, are closed by voltage or are blocked by Mg2+, respectively. The contribution of Ksust channels to outward resting current would assist this.
Generation of slow (muscarinic) EPSP
Inhibition of Ileak and IK(M) is the main mechanism of muscarinic depolarization in these neurones (Gähwiler & Brown, 1985; Madison et al. 1987). In agreement with this, single Ksust channels were readily inhibited by a muscarinic stimulant, Oxo-M. Although another component, a Ca2+-dependent cation current, is also activated in response to muscarinic stimulation, it can be detected only when K+ currents are inhibited and is probably generated in dendrites (Benson et al. 1988; Colino & Halliwell, 1993).
Generation of after-depolarization (ADP)
In hippocampal pyramidal neurones cell firing is followed by AHP or, when AHP is inhibited by muscarinic stimulation, by a Ca2+-activated cation current which generates an ADP (Caeser, Brown, Gähwiler & Knöpfel, 1993). The present data obtained from cell-attached patches suggest that there is probably one more component of ADP, which is also induced by Ca2+ influx during the spike, but is associated with a decrease in resting K+ conductance. A macroscopic current with similar properties has been described in pyramidal olfactory cortex neurones where it is responsible for sustained neuronal excitation (Constanti, Bagetta & Libri, 1993).
Because muscarinic inhibition of Ksust channels is accompanied by membrane depolarization, which in turn augments Ca2+ influx (through voltage-dependent Ca2+ channels or NMDA channels), this property of Ksust channel inhibition may provide a mechanism of positive feedback for cell excitation.
This work was supported by the UK Medical Research Council. We thank Professor David Brown for support, advice and discussions.