A. G. Clark: Organon Laboratories Ltd, Newhouse, Lanarkshire ML1 5SH, UK.
Corresponding author M. J. Shipston: Membrane Biology Group, Department of Biomedical Sciences, University of Edinburgh, Medical School, Teviot Place, Edinburgh EH8 9AG, UK. Email: email@example.com
1We investigated the effect of ATP in the regulation of two closely related cloned mouse brain large conductance calcium- and voltage-activated potassium (BK) channel α-subunit variants, expressed in human embryonic kidney (HEK 293) cells, using the excised inside-out configuration of the patch-clamp technique.
2The mB2 BK channel α-subunit variant expressed alone was potently inhibited by application of ATP to the intracellular surface of the patch with an IC50 of 30 μM. The effect of ATP was largely independent of protein phosphorylation events as the effect of ATP was mimicked by the non-hydrolysable analogue 5′-adenylylimidodiphosphate (AMP-PNP) and the inhibitory effect of ATPγS was reversible.
3In contrast, under identical conditions, direct nucleotide inhibition was not observed in the closely related mouse brain BK channel α-subunit variant mbr5. Furthermore, direct nucleotide regulation was not observed when mB2 was functionally coupled to regulatory β-subunits.
4These data suggest that the mB2 α-subunit splice variant could provide a dynamic link between cellular metabolism and cell excitability.
Substantial evidence suggests that BK channels are potently modulated by reversible protein phosphorylation through the action of multiple protein kinases and phosphatases that require ATP hydrolysis and are associated with the channel complex (White et al. 1991; Levitan, 1994; Reinhart & Levitan, 1995).
Here we demonstrate a novel mode of mammalian BK channel regulation through a direct action of ATP, independent of hydrolysis or protein phosphorylation events, in the mouse brain mslo α-subunit homologue mB2 expressed in HEK 293 cells. Our data suggest that the mB2 α-subunit splice variant could act as a metabolic sensor in mammalian brain and provide a dynamic link between cellular metabolism and cellular excitability. This novel mode of BK channel action, in concert with other ATP-dependent potassium channels (Tucker & Ashcroft, 1998), may constitute an important cellular defence mechanism, e.g. during ischaemic and hypoxic insults (LeBlond & Krnjevic, 1989; Martin et al. 1994; Faraci & Heistad, 1998).
Channel expression in HEK 293 cells
The mouse brain BK channel mslo variant (mB2) cDNA in pBluescript was kindly provided by Dr Leo Pallanck, University of Wisconsin, USA (Pallanck & Ganetzky, 1994). The BamHI-XbaI fragment of mB2 in pBluescript containing the entire coding region and > 1 kb of the 5′-untranslated sequence was subcloned into the mammalian expression vector pcDNA3.1+ (Invitrogen BV, Leek, The Netherlands).
The cDNA of a closely related mslo variant (mbr5) was kindly provided by Dr Lawrence Salkoff, Washington University, St Louis, MO, USA (Butler et al. 1993). The KpnI-XbaI fragment of mbr5 in the plasmid vector BSmxt was also subcloned into the mammalian expression vector pcDNA3.1+ (Invitrogen BV).
The bovine tracheal smooth muscle cell BK channel β-subunit (BKβ) cDNA was kindly provided by Dr Reid J. Leonard, Merck Research Laboratories, Rahway, NJ, USA (McManus et al. 1995). The EcoRI-NotI fragment was subcloned into the mammalian expression vector pcDNA3.1+ zeo (Invitrogen BV).
HEK 293 cells, a generous gift from David N. Sheppard (University of Edinburgh, UK) were maintained in Dulbecco's modified Eagle's medium containing 10 % fetal calf serum (Harlan Seralab, Crawley Down, UK) in a humidified atmosphere of 95 % air and 5 % CO2 at 37°C. Reverse transcriptase-polymerase chain reaction (RT-PCR) analysis (not shown) revealed no endogenous BK channel expression in this cell line. Cells were routinely passaged every 3-4 days using 0.25 % trypsin in Hanks’ buffered salt solution (HBSS) containing 0.1 % EDTA after reaching 80 % confluency. For transient transfections, cells were seeded onto glass coverslips 24 h before transfection at 40 % confluency and transfected with 1 μg of the expression plasmid using lipofectamine (Gibco BRL, Paisley, UK) essentially as described by the manufacturer. For co-transfections, β-subunit DNA was transfected at a 5-fold excess over the α-subunit. Cells were used 24-72 h after the start of transfection at between 40 and 80 % confluency. Stable cell lines were created as above by seeding in 24-well cluster dishes (Costar-Corning, Cambridge, MA, USA) and stable transformants selected, and maintained, for neomycin or zeocin resistance using 0.8 mg ml−1 Geneticin (Gibco BRL) and/or 0.2 mg ml−1 zeocin (Invitrogen), respectively. No qualitative difference in response to ATP analogues was observed between HEK 293 cells transiently or stably expressing the mB2 construct.
BK channels were analysed under voltage clamp in the excised inside-out configuration of the patch-clamp technique at room temperature (20-24°C) using physiological potassium gradients. The pipette solution contained (mM): 140 NaCl, 5 KCl, 20 glucose and 10 Hepes; pH 7.4. The bath solution contained (mM): 140 KCl, 2 MgCl2, 1 or 5 BAPTA, 30 glucose and 10 Hepes; pH 7.35 with free calcium ([Ca2+]i) buffered to the concentration indicated in the respective figure legend. For [Ca2+]i greater than, or equal to, 1 μM, dibromo-BAPTA was used as the calcium buffer and all analysis was performed in the presence of 2 mM MgCl2 to further exclude effects of weak Ca2+ chelation by free ATP (Klockner & Isenberg, 1992). Similar inhibition was observed under two different calcium buffering capacities using 1 or 5 mM BAPTA with the corresponding differences in total added calcium to achieve the same buffered level of free calcium calculated as described by Tsien (1980). To verify that ATP, or analogue addition, had no significant effect on [Ca2+]i in bath solutions, [Ca2+]i was verified by fura-2 (Calbiochem) fluorescence by determining the 340 nm/380 nm ratio in a cuvette-based Fluoromax 230 fluorimeter (SPEX Industries Inc., Edison, NJ, USA) at room temperature. With 2 mM MgCl2 in the bath solution at any given [Ca2+]i, determined using 1 or 5 mM BAPTA as buffer, subsequent addition of 1 mM of the ATP or GTP analogues had no significant effect on [Ca2+]i.
Data acquisition and voltage protocols were controlled by an Axopatch-200A amplifier and pCLAMP6 software (Axon Instruments). All recordings were sampled at 10 kHz and filtered at 2 kHz. For patches in which unitary currents could be resolved, channels were voltage clamped at the potential indicated in the respective figure legend. Mean steady-state single-channel open probability (Po) was determined from at least 30 s of continuous recording under each experimental condition. For macropatch recordings, outward BK currents were evoked by 200-500 ms step depolarizations (-50 to +40 mV), and the steady-state current amplitude, averaged from five consecutive depolarizations, between 200 and 500 ms was determined at each potential. For macropatch recordings with seal resistances > 10 GΩ, leak subtraction was not routinely applied. Pipettes were manufactured from Garner no. 7052 glass coated with Sylgard, and had typical resistances of between 2 and 10 MΩ in bath solution after fire polishing. Patches containing a single BK channel (verified at +80 mV and 10 μM [Ca2+]i) were extremely rare (< 0.5 % of total patches), even with pipette resistances > 10 MΩ. For quantitative analysis of the effects of ATP or ATP analogues on mean single-channel open probability or macropatch current amplitude, data were only included from patches that had received no prior experimental manipulation.
Chemicals and materials
ATP magnesium or potassium salt, ADP potassium salt, guanosine 5′-triphosphate (GTP) lithium salt, GTPγS tetralithium salt, 5′-adenylylimidodiphosphate (AMP-PNP) lithium salt and adenosine 5′-O-(3-thiotriphosphate) (ATPγS) tetralithium salt were all purchased from Sigma/Aldrich (USA) and stored as buffered 1 M stock solutions at -20°C prior to use. ATP and analogues buffered in bath solution to pH 7.35 were applied to the intracellular patch by gravity perfusion with 10 volumes of bath solution at a flow rate of 1-2 ml min−1. BAPTA, dibromo-BAPTA and fura-2 were obtained from Calbiochem (Nottingham, UK). Dehydrosoyasaponin-I (DHS-I) was a generous gift from Dr Owen McManus, Merck Research Laboratories, Rahway, NJ, USA.
Data are expressed as means ±s.e.m. unless otherwise stated. Statistical differences were compared using Student's t test as appropriate. A P value of less than 0.05 was considered to be significant.
mB2 is directly regulated by ATP
Excised inside-out patch-clamp recordings were made from HEK 293 cells expressing the large conductance calcium- and voltage-activated potassium (BK) channel variant mB2. Macropatch outward currents resulting from mB2 expression were characterized by their voltage dependence (V½,max was ∼+15 mV in 10 μM intracellular free calcium) and sensitivity to [Ca2+]i (Fig. 1A and B). Unitary mB2 currents were characterized by their large single-channel conductance (162.3 ± 3.05 pS, n= 4,Fig. 2). RT-PCR analysis revealed that mock transfected HEK 293 cells do not express endogenous BK channels, and this was confirmed by patch-clamp recording (Fig. 1A).
Application of 1 mM ATP (as Mg2+ or K+ salts) to the intracellular face of inside-out macropatches resulted in a rapid, reversible reduction in the steady-state outward macropatch current (Fig. 1A). This inhibition results from a reduction in mean open probability (NPo) of unitary currents rather than a reduction in unitary current amplitude (Fig. 2A and B). ATP inhibition was concentration dependent with an IC50 determined at +40 mV and 0.5 μM [Ca2+]i of approximately 30 μM (similar to that observed for other ATP-sensitive channels (Tucker & Ashcroft, 1998)) and a Hill coefficient (nH) of 0.5 (Fig. 1A).
In all patches tested, 1 mM ATP significantly reduced mB2 channel activity; there was no significant difference in the effect of ATP measured from single-channel records and from macropatch records containing multiple channels. Mean inhibition pooled from both macropatch and unitary current recordings was 66.9 ± 7.4 % (n= 19,P < 0.01) with a range of 30-97 % (Fig. 2A). mB2 channel activity was stable over 30 min of control recordings in the absence of ATP. Importantly, the inhibitory action of ATP was not simply a non-specific action of nucleotide triphosphates, as 1 mM GTP had no significant effect on channel activity (n= 3,Fig. 2A).
However, application of the non-hydrolysable ATP analogue AMP-PNP to the intracellular face of patches resulted in inhibition of channel activity by 72.7 ± 6.3 % of pretreatment control (n= 9,P < 0.01), which was fully reversible on washout (Fig. 2A and D). These results were similar to those observed with ATP. As AMP-PNP is very poorly hydrolysed by protein kinases, this strongly suggests that the mechanism of ATP action was independent of protein phosphorylation (Fig. 2A and D). To test this hypothesis further, we used the thiophosphate ATP analogue ATPγS, which can be hydrolysed by protein kinases, but the resultant thiophosphorylated target protein is largely resistant to dephosphorylation. If the effect of ATP was mediated solely via phosphorylation-dependent mechanisms, the effect of ATPγS should thus be essentially irreversible. Application of ATPγS to the intracellular face of patches resulted in an inhibition of 70.4 ± 9.2 % (n= 5,P < 0.01), which rapidly reversed to 80 % of control upon washout (Fig. 2A). Once again, these results are similar to those obtained using ATP, and support a mechanism of action that is independent of protein phosphorylation. The lack of complete reversal with ATPγS or ATP, compared with AMP-PNP, suggests that in addition to a direct effect of ATP, ATP-dependent hydrolysis/phosphorylation events also regulate mB2 activity in this system.
Under our experimental conditions, the potential effects of ATP analogues on calcium and pH buffering are eliminated (see Methods). However, in order to confirm that the inhibitory effects of ATP are not secondary to any effects of calcium chelation, pH changes or introduction of endogenous contaminants in ATP preparations (Bowlby & Levitan, 1996), we examined the effect of ATP and its analogues on the closely related mouse brain BK channel variant mbr5, which has previously been reported to be regulated by ATP only through phosphorylation-dependent mechanisms (Muller et al. 1996). Expression of mbr5 in HEK 293 cells resulted in macropatch currents (Fig. 3A) whose voltage- and calcium-sensitivity were similar to that of mB2 and with unitary currents of large conductance (142.3 ± 2.1 pS, n= 6,Fig. 3A). Under recording conditions identical to those used for mB2, addition of 1 mM ATP had no significant effect at the macropatch (Fig. 3A) or unitary current level (Fig. 3A). Channel activity expressed as a percentage of pretreatment control (100 %) was 104.3 ± 6.8 % (n= 6,Fig. 3A). Furthermore, no significant inhibition was observed after application of AMP-PNP to the intracellular face of the patch (110.1 ± 4.2 % of control activity, n= 6,Fig. 3A). In addition, 10 μM barium, a possible contaminant in ATP that can induce long blocked states of maximally activated BK channels, had no significant effect on the activity of unitary mB2 currents (101.5 ± 9.0 % of control activity, n= 3). These results confirm that the inhibition of mB2 channels is not an indirect, artifactual effect of adding ATP salts to the buffer solution.
Physiological ADP levels partially antagonize the direct effect of ATP
It has been proposed that certain ATP-sensitive potassium channels can act as metabolic sensors, by monitoring and responding to the [ATP]/[ADP] ratio (Tucker & Ashcroft, 1998). To investigate whether the ATP-sensitive mB2 α-subunits expressed alone act in this way, we examined whether physiological levels of ADP found in brain (50 μM (Veech et al. 1979)) could antagonize the effects of ATP. Application of 50 μM ADP alone to the intracellular face of patches had no significant effect on channel activity: activity was 92.3 ± 4.9 % (n= 3) of the pretreatment control value. However, in patches exposed to 50 μM ADP, currents were inhibited in the presence of ATP (1 mM) by only 41.3 ± 2.0 % (n= 4), which is significantly less (P < 0.05) than the inhibition observed in the presence of ATP alone.
β-Subunit co-expression prevents direct ATP inhibition
In vascular smooth muscle cells and cochlear hair cells, BK channel α-subunits are functionally coupled to regulatory β-subunits; this results in a leftward shift in calcium sensitivity and confers sensitivity to activation by dehydrosoyasaponin-I (DHS-I) (McManus et al. 1995; Vergara et al. 1998; Ramanathan et al. 1999). In addition, co-expression of the pore-forming α-subunits with β-subunits has been reported to modify the regulation of the channel by intracellular signalling pathways (Dworetzky et al. 1996). However, BK channel β-subunits are expressed at low levels in brain; furthermore, their functional role, distribution and association with brain α-subunits are poorly characterized (Tseng-Crank et al. 1994; Knaus et al. 1996). We examined whether the presence of the β-subunit modifies the direct ATP regulation of the mB2 channel complex. Co-expression of α- and β-subunits resulted in a leftward shift in calcium sensitivity of ∼65 mV and conferred sensitivity to activation by DHS-I (Fig. 4A). In patches in which functional interaction between α- and β-subunits was confirmed by reversible DHS-I activation, the effect of ATP was dramatically reduced (P < 0.01) compared with the effect of ATP on α-subunits alone. The activity of α+β channels was inhibited by only 24.0 ± 4.7 % (n= 5), whereas the activity of α-subunits alone was inhibited by 66.9 ± 7.4 % compared with control (n= 19) (Fig. 4A and C, cf. Fig. 1). Importantly, the effect of AMP-PNP was completely abolished: in the presence of AMP-PNP, channel activity was 102.8 ± 5.9 % of control (n= 6,Fig. 4A). The effect of ATPγS was also significantly attenuated, showing inhibition of only 22.5 ± 0.8 % (n= 3). The residual effect of ATP or ATPγS, in contrast to AMP-PNP, in co-expression studies probably reflects phosphorylation-mediated effects of these hydrolysable ATP analogues.
Here we show that the mslo BK channel α-subunit variant mB2 is directly regulated by intracellular ATP through a mechanism that is independent of protein phosphorylation. This direct inhibition of mB2 channels by ATP is significantly attenuated upon co-expression with β-subunits and, furthermore, the closely related mslo homologue mbr5 does not show direct inhibition by ATP. These data indicate that differential expression of splice variants and/or targeting of α- and β-subunits to different tissues or cellular compartments may allow different ATP-dependent signalling pathways to regulate BK channel function. Previous reports of direct ATP actions have been hampered by possible effects on calcium chelation or pH buffering when using EGTA as the calcium buffer (Klockner & Isenberg, 1992); we have eliminated such artifacts in our experiments.
The mechanism of ATP inhibition reported here is likely to result from direct binding of ATP to the α-subunit or an associated protein that interacts with the channel subunit in vivo. Although dslo variants contain a defined ATP-binding domain (Atkinson et al. 1991; Bowlby & Levitan, 1996) analysis of the amino acid sequence of mB2 (Pallanck & Ganetzky, 1994) does not reveal structural elements thought to be responsible for nucleotide binding in other proteins (Saraste et al. 1990). However, many ATP-binding proteins do not contain consensus sequences for nucleotide binding (e.g. KATP (Tucker & Ashcroft, 1998)), but rather amino acid residues required for ATP-binding domains are distributed across the primary sequence and are brought together during tertiary folding of the protein. Indeed, although mB2 is highly homologous to mbr5, it differs at multiple splice sites and at the C-terminus (Pallanck & Ganetzky, 1994), which could affect the assembly of any potential ATP-binding domain. Whether these sites confer ATP binding or allow association with endogenous ATP-binding proteins in vivo remains to be determined.
Although the inhibitory effect of AMP-PNP on mB2 α-subunits was fully reversible, the small residual inhibition conferred by ATP or ATPγS upon washout suggests that an ATP-dependent hydrolysis/phosphorylation component also regulates mB2 activity in this system. In further support of this, the inhibitory effect of AMP-PNP is completely abolished when α-subunits are co-expressed with β-subunits, whereas the hydrolysable analogues (ATP and ATPγS) elicit a partial inhibition of the α+β-channel complex. Indeed, BK channel activity is potently regulated by multiple protein phosphorylation cascades (Levitan, 1994; Reinhart & Levitan, 1995; Vergara et al. 1998).
Our data show that the direct inhibitory effect of ATP on mB2 channel activity was partially antagonized by physiological concentrations of ADP (Veech et al. 1979), an effect similar to that observed on several inwardly rectifying KATP channels (Tucker & Ashcroft, 1998). This suggests that BK channels, in conjunction with KATP channels, may act as metabolic sensors monitoring the intracellular ratio of [ATP]/[ADP]. Such regulation may play an important role in the cellular response to metabolic insult, e.g. during cerebral hypoxia, which produces profound effects on neuronal excitability and vascular tone observed during cerebral ischaemia (Martin et al. 1994). Although the molecular events immediately following ischaemic insults are poorly understood at the cellular level, many central neurones characteristically undergo rapid hyperpolarization due to the activation of potassium conductances, including BK channels, as a result of the elevation of [Ca 2+]i and the decrease in [ATP] during oxygen deprivation (Martin et al. 1994). Indeed, the predominant outward potassium current under experimental ischaemia in rat CA1 pyramidal neurones and locus coeruleus neurones is calcium dependent and charybdotoxin sensitive (LeBlond & Krnjevic, 1989; Harata et al. 1997; Murai et al. 1997), suggesting an involvement of BK channels in this response. Activation of BK channels during hypoxia in the cerebral vasculature may also be involved in the regulation of cerebral vasodilatation and blood flow (Faraci & Heistad, 1998). Direct modulation of BK channels by intracellular ATP would allow the channel to operate as a metabolic sensor under conditions in which the [ATP]/[ADP] ratio changes significantly, and allow them to elicit an immediate response if the metabolic state of the cell was compromised. Importantly, as ATP only inhibits mB2 α-subunits that are not functionally coupled to β-subunits (and hence much less sensitive to intracellular calcium), a fall in ATP levels would result in cell hyperpolarization in the absence of a significant rise in intracellular calcium normally required to activate the channels. Other pathophysiological changes evident during progressive ischaemia and anoxia may contribute further to the regulation of BK channel behaviour. In addition to falling ATP levels and elevated intracellular free [Ca2+], the channels may also be activated by multiple metabolic factors, such as changes in intracellular pH and redox potential, as well as intracellular signalling pathways including protein kinase/phosphatase cascades. KATP channels, which are activated during ischaemia and anoxia, have a similar IC50 range for ATP inhibition in isolated patches, as observed in this study with the mB2 variant (Tucker & Ashcroft, 1998). It is therefore possible that KATP and BK channels act in concert depending on the severity of the insult (Shinoda et al. 1997) to provide a mechanism of damage limitation by eliciting K+ efflux and allowing cell membrane repolarization.
This work was supported by The Wellcome Trust (grant ref. no. 048693). We are grateful to David N. Sheppard and other colleagues in the Membrane Biology Group for helpful discussions on the manuscript.