ATP-Dependent Activation of the Atrial Acetylcholine-Induced K+ Channel Does Not Require Nucleoside Diphosphate Kinase Activity
Abstract—Prior reports by others have shown that cytoplasmically applied ATP can activate the acetylcholine-induced K+ channel in inside-out atrial membrane patches when no guanine nucleotides are present in the solution bathing the cytosolic face of the membrane. A nucleoside diphosphate kinase mechanism was proposed to explain the activation by ATP. We show in the present study that cytoplasmic adenylylimidodiphosphate mimics the activation by ATP. Unlike ATP, the activation by adenylylimidodiphosphate does not subside on washout. Although commercially available adenylylimidodiphosphate is contaminated by guanylylimidodiphosphate, the activation by adenylylimidodiphosphate still occurs after HPLC purification to remove guanine nucleotide contamination. Adenylylimidodiphosphate does not support phosphotransferase activity by nucleoside diphosphate kinase. Therefore, nucleoside diphosphate kinase activity cannot explain the activation of atrial acetylcholine-induced K+ current by ATP and adenylylimidodiphosphate. We hypothesize that the activation by millimolar concentrations of ATP is due to binding of adenine nucleotide to the guanine nucleotide binding site of the G protein(s) responsible for stimulating the acetylcholine-induced K+ current.
- acetylcholine-induced K+ channel
- atrial myocyte
- nucleoside diphosphate kinase
The acetylcholine-induced K+ current, IK,ACh, plays a major role in vagal control of the heart. Activation of IK,ACh contributes to vagally mediated slowing of spontaneous rate, slowing of atrioventricular nodal conduction, and decreasing the force of contraction of the atria. The mechanism of activation is well studied but still incompletely understood. Muscarinic agonists stimulate IK,ACh via a membrane-delimited pathway.1 A pertussis toxin–sensitive G protein couples the cardiac m2 muscarinic receptor to the channel.2 There is evidence to suggest that both the α-subunit from the Gi family of G proteins or G-protein βγ-subunits can activate IK,ACh.3 4 5 6 7 8 Physiological activation of IK,ACh requires an extracellular agonist (either acetylcholine or adenosine), a membrane-bound receptor, and intracellular GTP. The channel can also be activated by hydrolysis-resistant analogues of GTP at the intracellular side of the membrane.9 10
G proteins are a family of heterotrimeric proteins that serve as transducing elements for a large number of cell surface receptors. In the absence of agonist, G proteins exist predominantly in the inactive heterotrimeric (αβγ) GDP-liganded state. This is because the rate of hydrolysis of GTP by the G-protein α-subunit is much faster than the rate of dissociation of GDP (reviewed in Reference 1111 ). The binding of an agonist to a G-protein–coupled receptor increases the dissociation rate for GDP, allowing a significant fraction of the subset of G proteins that are coupled to that receptor to bind GTP and become activated. The activated G protein is thought to dissociate into α- and βγ-subunits.11 Both portions of the G protein can then interact with downstream effector molecules. In the case of IK,ACh, the channel protein is the effector molecule. The G protein deactivates by hydrolyzing GTP to GDP. The GDP-liganded state favors the reassociation of the α- and βγ-subunits and terminates the response if agonist is no longer present.11 In frog atrial cells, indirect measures of the basal GDP dissociation rate from GK resulted in an off rate of 0.3 min−1 in the absence of agonist.10 The rate of GTP hydrolysis by GK was estimated to be 135 min−1.10
In the absence of intracellular guanine nucleotides, IK,ACh can be activated by millimolar concentrations of ATP on the cytoplasmic face of inside-out membrane patches.12 13 The stimulation by ATP can occur in the absence of agonists12 13 but is not inhibited by the presence of agonists.14 The effects of ATP can be inhibited by nanomolar concentrations of GTP or GDP.12 13 Two groups concluded independently that an NDP kinase was responsible for the effect of ATP.12 13 This interpretation was based on the observations that (1) the effect of ATP required Mg2+, (2) the hydrolysis-resistant ATP analogue AppNHp did not activate IK,ACh in their hands, (3) ATPγS mimicked the effect of ATP, and (4) the effect of ATP was inhibited by cytoplasmic guanine nucleotides.12 13 Since these studies were conducted on inside-out membrane patches in the absence of cytoplasmic guanine nucleotide, the hypothetical NDP kinase mechanism required the kinase to phosphorylate GDP while it was bound to the G-protein α-subunit. A whole-cell patch-clamp study using intracellular dialysis with ATPγS also supported the notion that NDP kinase activity could affect IK,ACh.15 The data were consistent with the proposal that ATPγS served as a phosphate donor for the conversion of cellular GDP to GTPγS.15 A subsequent biochemical study demonstrated that NDP kinase was present in atrial membranes.16
One difficulty with the NDP kinase hypothesis is the requirement for phosphorylation of GDP while it is bound to the G-protein α-subunit. Attempts to demonstrate phosphorylation of G-protein–bound GDP were not successful.17 Experiments with two anti–NDP kinase antibodies, one that inhibited the phosphotransferase activity and one that did not, supported the idea that NDP kinase was involved in agonist-dependent activation of IK,ACh.18 Notably, it was concluded that the action of NDP kinase on agonist-dependent activation of IK,ACh did not involve the conversion of GDP to GTP.18 However, antibody experiments suggested that NDP kinase activity was not required for agonist-independent activation of IK,ACh by 4 mmol/L ATP.18
The present study resulted from a serendipitous observation. We were investigating IK,ACh activity in inside-out membrane patches. AppNHp was included in the bath solution to prevent the activation of IK,ATP after patch excision. Surprisingly, we observed an activation of IK,ACh that was similar to the ATP-dependent activation described by others.12 13 This initiated a reinvestigation of the effect of AppNHp on IK,ACh.
Materials and Methods
Primary cultures of canine atrial cells were prepared as previously described.19 Cells were used between 1 and 4 days in culture. The inside-out configuration of the patch-clamp technique was used to record single-channel activity.20 Sylgard-coated patch electrodes (Dow Corning Corporation) had resistances of 5 to 20 MΩ when filled with the pipette solution. An Axoclamp 1D amplifier was used at a gain of 100 mV/pA and a low-pass filter setting of 2 or 5 kHz. The amplifier voltage offset was adjusted to zero before forming a high-resistance seal.
Vesicles or restricted access patches were frequently formed after patch excision. We could successfully convert some of the restricted access patches to an inside-out membrane patch by briefly moving the pipette tip out of the bath solution into air.20 Patches that did not respond to bath application of 2 mmol/L ATP with an increase in IK,ACh channel activity that reached a peak by 3 minutes were rejected.
Current and voltage were recorded using a modified videocassette recorder (model 420 CR, A.R. Vetter) and analyzed off-line using PClamp 6.02 software (Axon Instruments). A Labmaster TL-1 board (Techmar) was used to digitize the data at 8, 10, or 20 kHz. Sampling frequency was chosen to be at least four times the corner frequency of the low-pass filter used before digitization.
NPo was determined either by dividing the time integral for channel activity by the product of the single-channel current amplitude and the time interval being analyzed or by event lists generated using the Fetchan module of PClamp 6.02. NPo was determined from event lists using the PStat module of PClamp 6.02. All patches used for this study contained multiple channels. Estimates of τo were made from portions of traces with infrequent overlapping channel events. Open-time histograms with a minimum of 750 openings were fit with a single exponential using a maximum likelihood estimator to determine τo.
Positive ions flowing from the extracellular to the cytoplasmic side of the membrane are represented as downward deflections (ie, the normal whole-cell current convention is used). The voltages reported correspond to the voltage on the cytoplasmic side of the membrane relative to the solution in the patch electrode on the extracellular side of the membrane. Patches used for this study either had no ATP-sensitive K+ channels or were used after the activity of ATP-sensitive channels had run down.
Symmetrical bath and pipette filling solutions were used. The solutions contained (mmol/L) potassium aspartate 150, MgCl2 0, 3, or 10, HEPES 10 (titrated to pH 7.25 with KOH), dextrose 5.5, and EGTA 1. In experiments with 0 or 3 mmol/L MgCl2, 10 mmol/L KCl was added to the solutions. When used, the disodium or magnesium salt of ATP, AppNHp, or GDP was added directly to the bath solution. Calculated free Mg2+ concentrations21 with 10 mmol/L total Mg2+ were 9.64 and 7.73 mmol/L in the absence and presence, respectively, of 2 mmol/L disodium ATP. Calculated free Mg2+ concentrations21 with 3 mmol/L total Mg2+ were 2.86 and 1.15 mmol/L in the absence and presence, respectively, of 2 mmol/L disodium ATP. When ATP was added, the pH was adjusted back to 7.25 with KOH. A conventional gravity-fed flow system was used for these studies. The flow rate was ≈4 mL/min. The bath volume was 0.8 mL. All studies were performed at room temperature (21°C to 23°C).
For one set of experiments (Figure 8⇓), a fast-flow system was used to change the solution superfusing the cytoplasmic face of inside-out membrane patches within 2 seconds. We have described this fast-flow system previously.22 A six-inlet single-outlet polytetrafluoroethylene manifold was used for the fast-flow system. The membrane patch was placed in the opening of the polyethylene outlet tube (1.14-mm internal diameter).
Whole-cell currents were measured as previously described.19 Holding potential was −10 mV in whole-cell experiments. Currents were measured in response to slow hyperpolarizing voltage ramps (−7 mV/s). The bath solution for this study contained (mmol/L) NaCl 144, KCl 25, HEPES-NaOH 10 (pH 7.4), CaCl2 1.8, MgCl2 1, glucose 5.5, and glibenclamide 0.01. When present, the concentration of carbachol was 10 μmol/L. The electrode filling solution contained (mmol/L) potassium aspartate 125, KCl 15, HEPES-KOH 10 (pH 7.2), EGTA 4, disodium phosphocreatine 5, MgATP 3, MgCl2 1, and GTP 0.2.
Commercially available AppNHp (Sigma Chemical Co) was used for analysis and further purification. AppNHp was stored at −20°C or colder and used within 3 months of purchase or purification. Analytical HPLC was performed using an HP 1050 Liquid Chromatograph (Hewlett Packard) supplied with diode array detector and Primesphere 5 C18-HC column (0.2×15 cm, Phenomenex). The mobile phase was 5% MeOH in 0.1 mol/L triethylammonium acetate (pH 7.0). The flow rate was 0.2 mL/min.
Purification of commercial AppNHp was carried out by preparative HPLC using a Dynamax system (Rainin Instrument Co Inc) supplied with a UV absorbance detector at 254 nm and using a Primesphere 10 C18-HC column (2.1×25 cm, Phenomenex). The mobile phase was the same as in analytical HPLC at a flow rate of 20 mL/min.
Activation of IK,ACh by Cytoplasmic ATP: Mg2+ Dependence
The agonist-independent activation of IK,ACh channels by ATP has been reported previously by Kaibara et al13 and Heidbüchel and colleagues.12 14 The response typically develops slowly and requires 2 to 3 minutes to reach steady state. A common technical difficulty when working with inside-out membrane patches is the formation of vesicles or patches with restricted access to the cytoplasmic face. ATP-dependent activation of IK,ACh was used as a positive control to ensure that bath-applied agents could reach the cytoplasmic surface of the membrane. All of the patches included in this study responded to bath-applied ATP (2 mmol/L) with a steady-state activation of IK,ACh within 3 minutes.
An example of the ATP-dependent activation and the Mg2+ dependence of this effect is shown in Figure 1⇓. Superfusion with 2 mmol/L ATP in the presence of 3 mmol/L Mg2+ induced an inward current at −80 mV. When 3 mmol/L Mg2+ was removed from the bath in the continued presence of 2 mmol/L ATP, channel activity decreased back to baseline levels. Channel activity resumed when Mg2+ was reintroduced. Similar Mg2+ dependence was seen in nine of nine patches.
Current-Voltage Relationship for the ATP-Induced Current
The current-voltage relationship for the ATP-activated single channel is shown in Figure 2⇓. The continuous trace at the top of Figure 2⇓ shows that the current reverses near 0 mV. The inwardly rectifying current-voltage relationship and slope conductance between −40 and −120 mV of 44 pS of are similar to what have been previously reported for IK,ACh. Mean open time for the ATP-activated channels was also consistent with the properties of IK,ACh (0.60±0.03 milliseconds at −80 mV).
If one extrapolates a linear fit to the data between −120 and −40 mV in Figure 2⇑ to estimate the reversal potential, the value is negative to 0. This is probably due to the inward rectification of the channel, since large uncompensated liquid junction potentials would be unlikely when working with symmetrical bath and pipette solutions. We have observed a similar phenomenon with whole-cell IK,ACh measured in 25 mmol/L extracellular K+ (Figure 3⇓). Theoretically, the nonlinearity in the whole-cell current-voltage relationship could be due to voltage-dependent changes in open probability. However, we did not detect a decrease in NPo for ATP-activated channels when the voltage was shifted positive from −120 to −40 mV. This suggests that the nonlinearity of the whole-cell current as the reversal potential is approached is due to rectification of the single-channel current amplitude.
At first glance, the nonlinear current-voltage relationship appears different from that normally reported for IK,ACh in freshly isolated myocytes. The apparent discrepancy could indicate that IK,ACh is not identical in cultured versus freshly isolated myocytes, but we believe that this is unlikely. Although never a point of emphasis, there are examples of IK,ACh single-channel current-voltage relationships recorded from freshly isolated cells with nonlinearity negative to the K+ equilibrium potential. If one extrapolates a straight line through the most negative three or four data points in some published figures, the voltage intercept is negative to 0. Two examples are seen in the cell-attached currents in Figure 2A of Yamada and Kurachi23 and Figure 2B of Nair et al.6 Nonlinear whole-cell currents negative to the K+ equilibrium potential have also been observed in native myocytes (Figure 2B of Kurachi et al9).
Inhibition of ATP-Activated IK,ACh by GDP
Figure 4⇓ shows an example of IK,ACh activation by ATP in the presence of 10 mmol/L total Mg2+. Early in the activation, discrete single-channel events could be resolved (Figure 4⇓, trace b). By 3 minutes, the effect of ATP had reached a steady state, and many overlapping channel openings were observed (Figure 4⇓, top and trace c). A similar activation by ATP was observed in 11 patches in the presence of 10 mmol/L total Mg2+. The inward rectification of the channel was documented during voltage-clamp steps to +80 mV (Figure 4⇓, trace g). In agreement with other reports,12 13 the effect of ATP was inhibited by the application of 10 μmol/L GDP (Figure 4⇓, top and trace d). The inhibitory effect of GDP subsided after several minutes of washout, and channels gradually reactivated in the continued presence of ATP. The effect of ATP subsided on washout. The inhibitory effect of GDP was observed in two of two patches.
Although we normally used the disodium salt of ATP to activate the channels, the activation was not Na+ dependent. The magnesium salt of ATP was as effective as disodium ATP (not shown). This observation differs from results obtained in cultured embryonic chick atrial cells,24 where ATP alone had little effect, but a priming by ATP was reportedly required to observe a Na+-dependent activation.
Activation by Commercially Available AppNHp
Figure 5⇓ is a continuation of the experiment shown in Figure 4⇑. After ATP was washed out, the patch was exposed to 2 mmol/L AppNHp and 10 μmol/L GDP. Channel activity did not increase during simultaneous exposure to AppNHp and GDP (Figure 5⇓, top and trace b). The guanine nucleotide was then washed out. After several minutes of superfusion with AppNHp alone, the inhibitory effect of GDP subsided, and IK,ACh channels were activated by AppNHp (Figure 5⇓, top and traces c and d). Unlike the effect of ATP, the effect of AppNHp did not subside on washout. Although the activation by AppNHp conflicts with published reports, the persistent activation by commercially available AppNHp was observed in six of six patches when the patches were first screened to determine if they responded to ATP. The inhibitory effect of 10 μmol/L GDP was observed in two of two patches. The inhibitory effect of GDP on activation by AppNHp could only be observed before activation. Application of 10 μmol/L GDP after persistent activation by AppNHp was ineffective (n=2, not shown). The results presented in the present study were obtained with AppNHp purchased from Sigma, but we have also seen activation of neonatal rat atrial IK,ACh by AppNHp purchased from Boehringer Mannheim (S. Sorota, unpublished data, 1990).
Contamination of Commercial AppNHp
At first glance, the effect of commercially available AppNHp could have been interpreted to indicate that NDP kinase was not required for the activation of IK,ACh by ATP. Further investigation revealed that interpretation of these experiments was problematic because of guanine nucleotide contamination in commercially available AppNHp. Commercial suppliers synthesize AppNHp from adenosine that has significant guanosine contamination. The specifications of most commercial suppliers of AppNHp indicate that guanine nucleotide contamination is less than or equal to one part in 10 000.
HPLC analysis in ion-pair reversed-phase mode was used to look for contamination of AppNHp by guanine nucleotides. Absorbance at 254 nm was monitored with full spectra being determined at maxima of the corresponding peaks (Figure 6⇓, top). We detected the presence of GppNHp impurity at the level of 0.1% to 0.2% when 10 μg of commercial AppNHp was injected. The peak at 4.8 minutes had a spectrum that was identical to a reference standard of GppNHp (not shown). The adenosine analogues ran as two peaks with retention times of 5.9 and 8.3 minutes in Figure 6⇓, top. The UV spectra of these two peaks were nearly identical, indicating that both peaks represent adenosine analogues (see Figure 6⇓, bottom, inset). The main peak at 8.3 minutes corresponds to AppNHp. The peak at 5.9 minutes is likely to be the diphosphate breakdown product of AppNHp, which is a common contaminant in AppNHp preparations.25
Activation by GppNHp
The presence of GppNHp in the AppNHp solutions could be problematic because hydrolysis-resistant GTP analogues are known to activate IK,ACh. The patch-clamp studies above were performed using 2 mmol/L of commercially available AppNHp. If GppNHp were present at 1:10 000, this would result in a final GppNHp concentration of 0.2 μmol/L. This concentration of GppNHp was found to effectively and persistently activate IK,ACh in two of two patches (not shown). This result rendered the experiments with commercially available AppNHp uninterpretable, since it could not be determined whether the activity of the preparation was due to AppNHp or the contaminating GppNHp.
Purification of AppNHp
To circumvent the difficulty with commercial AppNHp, preparative HPLC purification was used to separate the GppNHp impurity from the AppNHp sample. AppNHp solution (1 mL) in the mobile phase (25 mg/mL) was applied to the column (10 injections total), and fractions containing pure AppNHp (due to analytical HPLC data) were pooled, evaporated using a rotary evaporator at room temperature, kept in vacuo for ≈16 hours to remove the excess of triethylammonium acetate, dissolved in water, and finally were freeze-dried to yield the triethylammonium salt of AppNHp (264 mg). Purity was confirmed by applying 200 μg of AppNHp (20-fold more than in Figure 6⇑, top) to the HPLC system. After purification, there was no detectable GppNHp peak eluting from the column between 3 and 5 minutes (Figure 6⇑, bottom).
Effect of Purified AppNHp
After confirming that triethylammonium did not activate IK,ACh on its own or inhibit the activation by ATP (not shown), we repeated the inside-out patch-clamp experiments using the purified AppNHp. A representative experiment is shown in Figure 7⇓. Superfusion with ATP activated IK,ACh channels in the patch, and channel activity subsided on washout of ATP (Figure 7⇓, top and trace b). Next, 2 mmol/L purified AppNHp was applied in the presence of 10 μmol/L GDP. There was little change in channel activity over the next 3 minutes (Figure 7⇓, top and trace d). At that time, GDP was washed out, but the application of AppNHp was continued. A persistent activation of IK,ACh resulted (Figure 7⇓, top and trace e). As expected, the channel activated by AppNHp exhibited strong inward rectification (Figure 7⇓, trace f). The effect of purified AppNHp was seen in four of four patches. The inhibition of the AppNHp effect by GDP was observed in two of two patches.
At a concentration of 2 mmol/L, ATP increased channel activity more than AppNHp. The ratio of NPo in the presence of AppNHp to the NPo in the presence of ATP was 0.60±0.04. A portion of the smaller NPo could be attributed to a decreased τo value in the presence of AppNHp (ATP, τo=0.56±0.04 milliseconds; AppNHp, τo=0.41±0.02 milliseconds).
Qualitatively consistent with the data from Kim et al,26 the τo value is longer with ATP than with AppNHp, suggesting regulation of channel open time by a phosphorylation event. However, our τo values are much shorter than those reported by Kim et al, and the effect of ATP on open time is much less pronounced. Differences in experimental models (oocyte versus native cell), solutions, or protocol (Kim et al studied an effect of ATP on agonist and GTP-activated channels) could explain the quantitative discrepancies. The high chloride (150 mmol/L) used in the study by Kim et al may also be a factor. Elevated cytoplasmic chloride is known to inhibit the GTPase activity of G proteins.11 Our data are consistent with those of Kaibara et al13 in that the mean open time for ATP-activated channels (in the absence of GTP) is <1 millisecond.
Activation Time Course: Effect of Recent GDP Application
If the NDP kinase hypothesis is correct, ATP-dependent activation will require the presence of GDP on the G protein. However, guanine nucleotides will dissociate from G-protein α-subunits in the presence of millimolar concentrations of Mg2+.11 The dissociation of guanine nucleotides from G proteins is Mg2+ dependent.11 If the NDP kinase hypothesis is correct, then recent application of GDP should not inhibit and may in fact enhance the activation of IK,ACh by ATP. If, however, GDP dissociation from G-protein α-subunits is limiting activation by ATP, then recent application of GDP will slow the activation time course.
We examined the effect of recent GDP exposure on the activation time course using a fast-flow system that allowed rapid and complete solution changes. The protocol was to expose to ATP and wash out three different times. GDP was applied for 5 minutes as the first ATP washout was initiated. As GDP was washed out, the second exposure to ATP was initiated. GDP was not present during the second washout of ATP or third exposure to ATP.
The results of one experiment are shown in Figure 8⇓. In this patch, channel activity during the first exposure to ATP reached a steady state within several seconds. The first exposure to ATP occurred 12 minutes after patch excision to allow IK,ATP channels to run down. A 5-minute exposure to 10 μmol/L GDP was initiated as soon as ATP was washed out. The second exposure to ATP was initiated simultaneously with GDP washout. During the second application of ATP, channel activity was initiated with a lag and took much longer to achieve a steady state. The estimated half-time for activation during the second exposure to ATP was 100 seconds. A third exposure to ATP was initiated after a 5-minute washout period. Activation was rapid during the third exposure to ATP. This result is inconsistent with the NDP kinase hypothesis but is consistent with a model in which GDP dissociation limits the activation time course. Similar results were observed for a second patch (not shown), with half times for activation of 54 and 35 seconds during the first and third exposures to ATP, respectively, and 148 seconds during the second exposure to ATP just after exposure to GDP.
We have reproduced several aspects of the findings of Heidbüchel and colleagues12 14 and Kaibara et al13 with regard to the activation of IK,ACh by millimolar concentrations of cytoplasmically applied ATP. The time course for activation by ATP, Mg2+ dependence, current-voltage relationship, channel τo values, and the inhibition of the response by 10 μmol/L GDP indicate that we are studying the same phenomenon. The prior studies had further determined that the response to ATP was not blocked by pertussis toxin and could be mimicked by ATPγS. These data in combination with their failure to activate IK,ACh with AppNHp were consistent with the hypothesis that a phosphorylation event was involved. Both Heidbüchel and colleagues and Kaibara et al proposed that the phosphorylation of GDP that was bound to G-protein α-subunit by NDP kinase was responsible for the stimulatory effect of ATP. This action would convert GDP to GTP and could promote activation of the G protein. Heidbüchel et al16 subsequently used biochemical methods to demonstrate that NDP kinase activity is detectable in atrial membranes from frogs, guinea pigs, and humans.
Before the present study was undertaken, there were several aspects of the NDP kinase hypothesis that were problematic. Whole-cell IK,ACh is very low in the absence of agonist despite the presence of cytoplasmic GTP. Low basal activity is attributed to the ratio of the GTP hydrolysis rate and the rate for GDP dissociation (reviewed in Reference 1111 ). Since GDP dissociates from the G protein very slowly relative to the rate of GTP hydrolysis, the majority of the G protein is GDP-liganded. How then could NDP kinase activity result in sustained activation of IK,ACh in inside-out membrane patches during exposure to ATP? The phosphorylation of GDP that is bound to the G protein would have to be fast relative to the rate of GTP hydrolysis to cause most of the G protein to be GTP-liganded. Furthermore, the requirement that NDP kinase acts on bound GDP places significant constraints in terms of the spatial orientation of G protein, NDP kinase, GDP, and ATP that would allow for this proposed rapid rephosphorylation. A biochemical study to detect the phosphorylation of GDP while bound to G proteins has concluded that this does not occur.17
It is expected that GDP will dissociate from the G protein in inside-out membrane patches. Indirect estimates of the basal GDP dissociation rate from the G protein(s) that activates IK,ACh in native cells suggest that 30% of the GDP dissociates every minute.10 Heidbüchel et al12 have reported that the application of GTP to the cytoplasmic face of the membrane in the absence of extracellular agonist produces a transient activation of IK,ACh in the continued presence of cytoplasmic GTP. To account for this transient activation, one would have to propose that GTP had access to unliganded G protein (no guanine nucleotide bound), resulting in a short pulse in which the G protein exists primarily in the GTP-liganded state. This implies that there was no GDP on many of the G-protein molecules. GDP dissociation from the G protein in inside-out membrane patches clearly presents a problem for the NDP kinase hypothesis, since GDP-liganded G protein is required to explain the activation by ATP.12 13 If the NDP kinase hypothesis is correct, one would predict that ATP would become less effective over time after patch rupture as GDP is lost from the patch. To date, there are no reports of a decreased effect of ATP over time after patch excision or during repeated exposures to ATP.
Xu et al18 have recently challenged the NDP kinase hypothesis as an explanation for agonist-independent activation of IK,ACh by ATP. Antibodies against NDP kinase were used. No inhibition of the agonist-independent activation by ATP was observed. Xu et al concluded that NDP kinase played no role in the agonist-independent activation of IK,ACh by ATP. In contrast, when carbachol was present on the extracellular face of the membranes, antibodies against NDP kinase inhibited ATP-dependent IK,ACh activity,18 suggesting a role in agonist-dependent activation of IK,ACh. However, Xu et al concluded that the inhibition of agonist-dependent activity was not attributable to a decrease in phosphotransferase activity, because a second antibody that bound NDP kinase without blocking the ability of the enzyme to phosphorylate NDPs also inhibited the ATP-dependent activation of IK,ACh when carbachol was present.
Since AppNHp will not support phosphotransferase activity, the effect of AppNHp observed in the present study indicates that it is not necessary to invoke an NDP kinase mechanism to explain agonist-independent activation of IK,ACh by cytoplasmic ATP. We do not know why prior studies failed to detect the activation of IK,ACh by AppNHp12 13 but are confident that the effect occurs in every patch if the AppNHp is fresh and there is not restricted access to the cytoplasmic face of the patch. A possible reason for the discrepancy is the difference in concentrations of AppNHp used. Former studies used 1 mmol/L, whereas we used 2 mmol/L. The results presented in the present study add experimental support to the conclusion of Xu et al18 (ie, that NDP kinase activity is not involved in the agonist-independent activation of IK,ACh by cytoplasmic ATP) and provide mechanistic insight into the activation of IK,ACh by cytoplasmic ATP.
We propose a simple hypothetical mechanism to explain the effect of ATP. It is possible that in the absence of guanine nucleotides ATP binds directly to the GTP binding site and activates the G protein, essentially substituting for the normal ligand (Figure 9⇓). However, there is a difference in the effects of ATP and GTP: IK,ACh remains active as long as ATP is present, whereas GTP is reported to result in transient activation.12 To account for the steady-state activation by ATP, it is necessary to propose that many of the G-protein α-subunits are present in the ATP-liganded state as opposed to the ADP-liganded state. This could occur if either (1) the rate of ATP hydrolysis was much lower that the rate of GTP hydrolysis and/or (2) the rate of ADP dissociation was much faster than the rate of GDP dissociation. Further studies are required to discriminate between these possibilities.
The data in Figure 8⇑ suggest that the rate of GDP dissociation would be the limiting factor for activation shortly after patch excision or shortly after exposing the patch to guanine nucleotides. If a patch is exposed to guanine nucleotide–free solutions for long times, the activation rate is shorter. These data are consistent with the experimental observations of Heidbüchel et al,14 indicating that in the absence of guanine nucleotides the time course for channel activation is shorter during a second exposure to ATP.
In summary, we found that AppNHp can support the activation of IK,ACh in the absence of extracellular agonist or intracellular guanine nucleotide. These data are not consistent with the NDP kinase hypothesis for activation of IK,ACh by ATP. This observation does not rule out a role for NDP kinase under some yet-to-be-identified conditions. Our results do indicate that the activation of IK,ACh by ATP in the absence of guanine nucleotides can be explained without invoking a role for NDP kinase.
Selected Abbreviations and Acronyms
|τo||=||open time constant|
|GK||=||G-protein coupling muscarinic receptors to IK,ACh|
|I K,ACh||=||acetylcholine-induced K+ current|
|I K,ATP||=||ATP-sensitive K+ current|
|NPo||=||product of the number of channels (N) and single-channel open probability (Po)|
This study was supported by National Institutes of Health grants 1-R29-HL-48514 and 1-R01-HL-53972.
- Received August 21, 1997.
- Accepted February 26, 1998.
- © 1998 American Heart Association, Inc.
Sorota S, Tsuji Y, Tajima T, Pappano AJ. Pertussis toxin treatment blocks hyperpolarization by muscarinic agonists in chick atrium. Circ Res. 1985;57:748–758.
Yatani A, Codina J, Brown AM, Birnbaumer L. Direct activation of mammalian atrial muscarinic potassium channels by GTP regulatory protein Gk. Science. 1987;235:207–211.
Cerbai E, Klöckner U, Isenberg G. The α subunit of the GTP binding protein activates muscarinic potassium channels of the atrium. Science. 1988;240:1782–1783.
Nair LA, Inglese J, Stoffel R, Koch WJ, Lefkowitz RJ, Kwatra MM, Grant AO. Cardiac muscarinic potassium channel activity is attenuated by inhibitors of Gβγ. Circ Res. 1995;76:832–838.
Breitwieser GE, Szabo G. Mechanism of muscarinic receptor-induced K+ channel activation as revealed by hydrolysis-resistant GTP analogues. J Gen Physiol. 1988;91:469–493.
Otero AS, Breitwieser GE, Szabo G. Activation of muscarinic potassium currents by ATPγS in atrial cells. Science. 1988;242:443–445.
Heidbüchel H, Callewaert G, Vereecke J, Carmeliet E. Membrane-bound nucleoside diphosphate kinase activity in atrial cells of frog, guinea pig, and human. Circ Res. 1992;71:808–820.
Randazzo PA, Northup JK, Kahn RA. Regulatory GTP-binding proteins (ADP-ribosylation factor, Gt, and RAS) are not activated directly by nucleoside diphosphate kinase. J Biol Chem. 1992;267:18182–18189.
Xu L, Murphy J, Otero AS. Participation of nucleoside-diphosphate kinase in muscarinic K+ channel activation does not involve GTP formation. J Biol Chem. 1996;271:21120–21125.
Sorota S, Hoffman BF. Role of G-proteins in the ACh-induced potassium current of canine atrial cells. Am J Physiol. 1989;257:H1516–H1522.
Du X-Y, Sorota S. Cardiac swelling-induced chloride current depolarizes canine atrial myocytes. Am J Physiol. 1997;272:H1904–H1916.
Yamada M, Kurachi Y. Spermine gates inward-rectifying muscarinic but not ATP-sensitive K+ channels in rabbit atrial myocytes. J Biol Chem. 1995;270:9289–9294.
Sui JL, Chan KW, Logothetis D. Na+ activation of the muscarinic K+ channel by a G-protein-independent mechanism. J Gen Physiol. 1996;108:381–391.
Kim D, Watson M, Indyk V. ATP-dependent regulation of a G protein-coupled K+ channel (GIRK1/GIRK4) expressed in oocytes. Am J Physiol. 1997;272:H195–H206.