Does diacylglycerol regulate KCNQ channels?
- 782 Downloads
Some ion channels are regulated by inositol phospholipids and by the products of cleavage by phospholipase C (PLC). KCNQ channels (Kv7) require membrane phosphatidylinositol 4,5-bisphosphate (PIP2) and are turned off when muscarinic receptors stimulate cleavage of PIP2 by PLC. We test whether diacylglycerols are also important in the regulation of KCNQ2/KCNQ3 channels using electrophysiology and fluorescent translocation probes as indicators for PIP2 and diacylglycerol in tsA cells. The cells are transfected with M1 muscarinic receptors, channel subunits, and translocation probes. Although they cause translocation of a fluorescent probe with a diacylglycerol-binding C1 domain, exogenously applied diacylglycerol (oleoyl-acetyl-glycerol and dioctanoyl glycerol) and phorbol ester do not mimic or occlude the suppression of KCNQ current by muscarinic agonist. Blocking the metabolism of endogenous diacylglycerol by inhibiting diacylglycerol kinase with R59022 or R59949 slows the decay of diacylglycerol twofold but does not mimic or occlude muscarinic regulation and recovery of current. Blocking diacylglycerol lipase with RHC-80267 also does not occlude muscarinic modulation of current. We conclude that the diacylglycerol produced during activation of PLC, any activation of protein kinase C that it may stimulate, and downstream products of its metabolism are not essential players in the acute muscarinic modulation of KCNQ channels.
KeywordsDiacylglycerol Phospholipase C Inositol 1,4,5-trisphosphate Phosphatidylinositol 4,5-bisphosphate Muscarinic receptor M-current Protein kinase C 59022 RHC-80267
The various organellar membranes and the plasma membrane each has a unique lipid composition, and their integral and peripheral membrane proteins are sensitive to these lipids . Thus, the plasma membrane contains much of the cellular phosphoinositide, phosphatidylinositol 4,5-bisphosphate (PIP2), as well as numerous PIP2-sensitive proteins [2, 3]. KCNQ ion channels (Kv7.2 and Kv7.3) underlie the classical M current of sympathetic neurons, a non-inactivating K+ current that can be suppressed by agonists that activate phospholipase C (PLC). Recently, we and others presented evidence that KCNQ channels require PIP2 for function and that they are suppressed by PLC because PIP2 becomes depleted from the plasma membrane [4, 5, 6, 7, 8, 9, 10]. The evidence included showing that PLC is essential for agonist-induced suppression of current, that PIP2 is dramatically depleted when PLC is activated, that other molecules that bind up PIP2 also suppress current, that ATP and a lipid kinase on the PIP2-synthesis pathway are essential for recovery of current, that recovery had a similar time course to PIP2 regeneration, and that exogenously applied short-chain PIP2 analogs can speed recovery and slow suppression. We have made a kinetic model that successfully describes the events of muscarinic activation of PLC, hydrolysis, and resynthesis of PIP2, and the resulting effects on KCNQ currents [5, 8].
This paper focuses on possible roles of DAG in the short time scale of acute modulation of current. Is it essential for current modulation? Do DAG or DAG metabolites have significant direct or indirect effects on the channels during acute channel modulation? There is precedent for sensitivity to DAG. Older studies reported that activation of protein kinase C (PKC) by DAGs and phorbol esters can reduce M current [11, 12, 13, 14], and newer papers suggest that KCNQ channels exist in a complex with bound PKC and can be phosphorylated . In addition, some other ion channels are thought to be directly responsive to DAG independent of PKC [16, 17, 18]. We express KCNQ channels from their constituent KCNQ2 and KCNQ3 subunits together with PLC-coupled M1 muscarinic receptors in a mammalian cell line. The experiments consist of studying effects of adding exogenous DAG analogs and of slowing the metabolism of endogenous DAG. We use two fluorescent translocation probes (1) for DAG to verify its appearance in the plasma membrane and (2) for PIP2 to verify its depletion during activation of receptors.
Materials and methods
Human embryonic kidney tsA-201 (tsA) cells were cultured and transiently transfected using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) with various cDNAs, as described in : M1 muscarinic receptor (1 μg, from Neil Nathanson, University of Washington, USA), PH-PLCδ1-EGFP (PH-EGFP, 0.25 μg, from Pietro De Camilli, HHMI, Yale University, USA), PKC-C1A-EGFP (C1-EGFP, 0.25 μg, from Tobias Meyer, Stanford University), the channel subunits human KCNQ2 and rat KCNQ3 (Kv7.2 and Kv7.3; 1 μg, from David McKinnon, State University of New York, Stony Brook, NY, USA), and green fluorescent protein (0.1 μg) as a marker for transfection if needed. The muscarinic receptor agonist oxotremorine-M was used at 10 μM. Chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA).
For translocation experiments, tsA cells were imaged 24–48 h after transfection . Images were taken every 5 s on a Leica TCS/MP confocal microscope at room temperature and analyzed, as described . In the legends, we use the symbol F to represent the mean pixel intensity from the fluorescent probe in a cytoplasmic (in one case nuclear) region of interest, normalized so that the minima and maxima are 0 and 1.0. For electrophysiology experiments, cells were locally perfused with flowing solutions as KCNQ currents were recorded by whole-cell patch clamp, as described in . Currents were studied by holding the cell at −20 mV and applying a 500 ms hyperpolarizing step to −60 mV every 4 s. The plotted amplitude of the current is the outward current at the −20 mV holding potential. The voltage dependence of channel activation was measured from the normalized amplitudes of outward currents at −70 mV (tail currents) after 500-ms depolarizations to various potentials. The plotted value is the mean current in the period 10–20 ms after return to −70 mV.
Kinetic simulations are made with the virtual cell environment of the National Resource for Cell Analysis and Modeling, University of Connecticut Health Center http://www.nrcam.uchc.edu. We started with the model described in Horowitz et al. 2005  and made small changes. The revised working model with control values of rate constants and initial conditions is available at that web page for public use and modification under Shared/hillelab/SuhHilleEJPFig1b.
Modeling of DAG and IP3 kinetics
We begin by showing the time course of receptor-evoked production and disappearance of DAG and IP3 predicted by our previous kinetic modeling (Fig. 1b). As in our previous presentations of this model [5, 8], application of a supramaximal concentration (10 μM) of the muscarinic agonist oxotremorine-M (Oxo-M) activates PLC and causes PIP2 to fall 100-fold within a few seconds. Recovery of PIP2 begins only after the agonist is removed. All of the PIP2 that is cleaved is converted into equimolar amounts of DAG and IP3. The time course of these two products would be identical if they had the same lifetime; however, it seems that they are further metabolized at different rates. The model assigns DAG molecules a mean lifetime of 80 s, chosen to match our experiments with translocation of the C1-EGFP probe for DAG  (see also below). This decay of DAG represents the parallel actions of the three enzymes shown in Fig. 1a. The model assigns IP3 molecules a mean lifetime of 15 s, representing rapid dephosphorylation by IP3 5-phosphatase.
The literature does not have accurate measures of IP3 lifetime, but our value would allow IP3-dependent calcium release to fall rapidly after agonist is removed, as in our published experiments . An interesting point to note is that because PIP2 is quickly depleted, the rate of production of IP3 and DAG must fall dramatically while the muscarinic agonist is still present. This is particularly evident in the time course of short-lived IP3. In functional experiments that study IP3-induced calcium release, the potential depletion of PIP2 is often overlooked as a possible explanation for transients in IP3 production and in calcium elevations. In some systems, the depletion of PIP2 by PLC is probably much less severe than in this model, as there is a concomitant activation of the lipid kinases that synthesize PIP2. The factors regulating these lipid kinases are not well-understood.
Application of DAGs
In the resting condition, the DAG probe appears throughout the cytoplasm and nucleus, as there is little membrane DAG for it to bind to (Fig. 2a, first image), and the PIP2/IP3 probe is bound to the plasma membrane where there is much PIP2 (Fig. 2b, first image). Then Oxo-M is applied to activate PLC. The DAG probe translocates from cytoplasm to plasma membrane as DAG is formed there, and the PIP2/IP3 probe migrates the other way as membrane PIP2 is cleaved and cytoplasmic IP3 is formed (second images of Fig. 2a,b). The effect is reversible after Oxo-M is removed (third images), showing that DAG is metabolized, PIP2 is resynthesized, and presumably IP3 is broken down, all within <200 s.
The full-time courses of the transient changes of cytoplasmic probe fluorescence for this experiment are drawn in Fig. 2c,d as open symbols. Fluorescence in the nucleus is shown as filled symbols. At about 250 s, 10 μM OAG is applied. It induces slow migration of the DAG probe to the plasma membrane (Fig. 2a,c), showing that enough applied OAG penetrates gradually to the inner leaflet of that membrane to attract the C1-domain of PKC. This prolonged recruitment lasts long enough to draw probe from the nucleus as well (see ). Nevertheless, OAG does no affect the localization of the PIP2/IP3 probe or its subsequent reversible translocation during application of Oxo-M (Fig. 2b,d), showing that PLC can be activated, the products metabolized, and new PIP2 can be synthesized in normal fashion in the presence of added OAG. Figure 2e,f shows similar experiments with 10 μM DOG instead of OAG. The results look virtually the same (n=5).
As some published works report effects of DAG at concentrations higher than 10 μM DAG [12, 14, 25], we also tested a tenfold-higher concentration using both OAG and DOG (data not shown). The effects with 100 μM were similar to those with 10 μM except that the KCNQ current was slightly depressed during DOG or OAG application (15.8±0.7%, n=6, and 6.2±1.0%, n=4, respectively), an effect that took only a few seconds to develop and reverse. As before, DOG speeded activation of current and shifted the voltage dependence of activation (by −8 mV). The higher concentration of DAG analogs also did not induce translocation of the PIP2 probe or change its translocation during Oxo-M application (data not shown), confirming the specificity of the probe and a lack of major change of PLC activation.
Pathways of DAG degradation
The preceding negative experiments could be criticized because they involved application of unnatural DAG analogs. Actions of these DAGs could be different, and certainly the downstream metabolites are not the same as those of native DAGs. The predominant endogenous DAG produced by PLC action on PIP2 would be 1-stearoyl-2-arachidonyl-sn-glycerol . One approach to testing the endogenous DAGs would be to slow their further metabolism so that they would persist longer and so downstream metabolites would not be produced. DAGs are metabolized in three ways by three major enzymes (Fig. 1a). DAG kinase phosphorylates DAG to give phosphatidic acid, which has potential signaling activity. DAG lipase and monoacylglycerol lipase remove the fatty chains, to give first monoacylglycerol and then glycerol plus the free fatty acids, stearic acid, and arachidonic acid, which have potential signaling roles. Finally, DAG acyltransferase can yield triglyceride, which is not expected to have signaling activity.
Despite doubling the lifetime of DAG and preventing formation of phosphatidic acid, these inhibitors do not reduce the depth of muscarinic suppression of current or slow the time course of recovery of channels from muscarinic suppression (Fig. 4c,f). The half-time for recovery of current was 155±8 s (n=5) in control cells and 149±5 s (n=7) and 103±11 s (n=5) in cells treated with R59022 or R59949, respectively. The speeding of recovery by R59949 was statistically significant (P<0.002) and frequently accompanied by over-recovery, as in Fig 4f. Hence, endogenous DAG and its metabolism to phosphatidic acid are not essential elements of acute muscarinic suppression or of subsequent recovery of KCNQ current.
Phorbol esters and PKC
We have looked for roles of DAG in the strong, acute modulation of KCNQ channels by activation of PLC, and we find none. Applications of exogenous OAG and DOG do not mimic or change this modulation. Inhibitors of endogenous DAG metabolism do not intensify or weaken muscarinic modulation and its recovery. Phorbol ester does not mimic or change the muscarinic suppression. Using a fluorescent translocation probe for DAG, we show that each of the reagents used is applied at a concentration that would be recognized by the C1 domain of PKC or that significantly prolongs the lifetime of endogenous DAG.
Our observations give two indices of DAG metabolism in tsA cells. We found that the mean lifetime of endogenous DAGs is 80 s at room temperature and that breakdown via DAG kinase to phosphatidic acid accounts for about 50% of the DAG metabolism. The 50% figure assumes that the two DAG kinase inhibitors tested have high selectivity for this enzyme and inhibit it relatively completely. These conclusions compare favorably with a study of DAG metabolism in platelets where the fate of added [3H]DOG (labeled in the glycerol) was followed by chemical analysis of products . The DOG was half metabolized in 60 s at 37°C, and production of DOG phosphate by DAG kinase accounted for 40% of the loss. Cleavage to [3H]glycerol by di- and monoacylglycerol lipase accounted for another 20% of the loss. Eventually, [3H]dioctanolyphosphatidylinositol was also formed, presumably from the DOG phosphate (phosphatidic acid) product of DAG kinase.
Interestingly, this conversion had a half time of roughly 20 min at 37°C, showing, as is usually stated, that the synthesis of PIP2 after PLC activation must depend much more on the existing large, unphosphorylated phosphatidylinositol pool than on slow regeneration all the way from phosphatidic acid. These experiments also remind us that during functional studies with exogenous DAG analogs, numerous metabolic steps are generating additional downstream products as well. Review articles (e.g., ) emphasize that much cellular DAG is formed from abundant lipids, such as phosphatidylcholine, by pathways that do not involve PIP2-specific PLC, as well as by dephosphorylation of phosphatidic acid. We have no chemical analyses but can say that the C1-probe, which recognizes a broad range of DAGs, does not report a lot of plasma membrane DAG accumulation in tsA cells until PLC is activated. Hence, the DAG pools in unstimulated cells seem small by this criterion.
It is not surprising that DAGs have no direct effect on KCNQ channel modulation, as there have been few prior suggestions that they might. But now it seems very clear. We can say this both for exogenous OAG and DAG, as well as for the endogenous DAG. We did find that DOG, but not OAG, seemed to have direct effects on channel gating, facilitating opening. The rate of activation upon depolarization to −20 mV was speeded and the voltage dependence of activation was shifted in such a way as to increase the probability of opening at any voltage. A similar facilitation of opening was seen with R59949, and not R59022. Further structure-activity studies would be needed to interpret these perturbations and to relate them to any physiological signals.
From our inhibitor work, we would argue that the downstream products of DAG metabolism are not needed for the acute muscarinic suppression of KCNQ current and its subsequent recovery. First, inhibiting the formation of phosphatidic acid does not bother suppression or recovery. A similar result with R59022 was obtained in bullfrog . Second, all products of DAG metabolism (such as arachidonic acid) must be produced slowly, as DAG is normally metabolized with a time constant of 80 s. This would mean that these products are not generated significantly during the onset of inhibition (<7 s) and start to accumulate only during the recovery phase for short agonist exposures. Our experiments do not test whether downstream products may have direct actions. For that, they should be tested individually. Indeed, there already is much literature saying that arachidonic acid application increases KCNQ currents (reviewed in ). Our point is that such actions would have to be slow to develop after application of a stimulus for PLC. Ford et al.  did find that if DAG kinase was inhibited with R59022, repeated PLC activation eventually slowed recovery of M current, presumably reflecting the stoppage of phosphatidylinositol regeneration from phosphatidic acid.
The possibility of roles for PKC remains a more difficult question. From work on frog neurons and frog smooth muscle, several investigators proposed that PKC might mediate muscarinic suppression of M current [12, 13, 14]. A strong reduction of current by OAG and PMA in C. elegans KCNQ homologs (KQT channels) expressed in Xenopus has also been attributed to activation of PKC, and, of the mammalian channels, KCNQ5 was shown to be strongly affected by PKC and the others not . Our early work on frog neurons showed that exogenous DAGs did reduce M current somewhat, an effect blocked by blockers of PKC . We also showed that the PKC blockers did not block the acute muscarinic suppression of M current (see also ).
Hence, we concluded then that activation of PKC was not an essential link in muscarinic suppression. In this paper, we saw smaller effects of DAGs on expressed mammalian KCNQ2/KCNQ3 channels than we saw earlier on M current in frog neurons, in accordance with . Recent molecular work with mammalian channels showed that mammalian KCNQ channels bind the A-kinase anchoring protein AKAP 150, which in turn can bind PKC . Furthermore, it was found that activation of PKC can increase the sensitivity of KCNQ channels to muscarinic modulation, whereas in frog neurons the agonist sensitivity was said to be reduced by PKC .
Significant sensitization to muscarinic agonists could well be happening in our experiments but would not be apparent, as we used only supra-maximal concentrations of Oxo-M to evoke the suppression of current. Another possible reason we did not see larger effects of OAG, DOG, or PMA is the suggestion that in some unstimulated mammalian cell lines, the relevant phosphorylation of the channels is already complete . In either case, we continue to hold that acute activation of PKC during muscarinic signaling is not necessary for the rapid muscarinic suppression of current that we are studying.
We thank Greg Martin for help in the Keck Imaging Center and Lea Miller for expert technical assistance. This work was supported by National Institutes of Health grant NS08174.
- 24.Ford CP, Stemkowski PL, Light PE, Smith PA (2004) Experiments to test the role of phosphatidylinositol 4,5-bisphosphate in neurotransmitter-induced M-channel closure in bullfrog sympathetic neurons. J Neurosci 23:4931–4941Google Scholar