On the physiological roles of PIP2 at cardiac Na+–Ca2+ exchangers and KATP channels: a long journey from membrane biophysics into cell biology
- 1Department of Physiology, University of Texas Southwestern Medical Center at Dallas, Dallas, TX 75390, USA
- Corresponding author D. W. Hilgemann: Department of Physiology, UT Southwestern Medical Center, Dallas, TX 75390-9040, USA. Email: donald.hilgemann{at}utsouthwestern.edu
Abstract
Over the last 10 years we have tried to understand the roles of PIP2 in regulating cardiac Na+–Ca2+ exchangers and KATP K+ channels, both of which are directly activated by PIP2. Up to now, the idea that hormones might physiologically regulate these mechanisms by causing changes of PIP2 concentrations in the cardiac sarcolemma, either locally or globally, is not well supported. In intact myocardium, but not excised patches, phosphatidylinositol 4-phosphate 5-kinase (PIP5K) activity appears to be Ca2+ activated and dependent on cardiac activity. Potentially therefore the primary second messenger of the heart, cytoplasmic Ca2+, may regulate PIP2 and therewith numerous cardiac membrane processes. In general, however, PIP2 may simply serve to strongly activate various cardiac channels and transporters when they are inserted in the sarcolemma, while a lack of PIP2 on internal membranes maintains transporters and channels inactive during trafficking and processing. As in most, if not all, strong regulatory systems of cells, the activating effects of PIP2 can apparently be countered by strong inactivation mechanisms. In this context, our recent work suggests that internalization of cardiac Na+–Ca2+ exchangers is promoted by increased PIP2 synthesis, especially in combination with other cell signals. Assuming that multiple adapter–PIP2 interactions are necessary to initiate the budding of individual membrane vesicles, the dependence of endocytosis on PIP2 in the surface membrane can potentially be a very steep function. Thus, a better understanding of the regulation of cardiac lipid kinases may be key to understanding when and how cardiac ion transporters and channels are internalized.
It is more than 10 years ago that we noticed PIP2 in our efforts to understand how ion transporters and channels are activated by cytoplasmic ATP (Hilgemann & Ball, 1996). At first, we were surprised that an anionic phospholipid seemed to be important (Hilgemann & Collins, 1992), and then we were surprised that the lipid seemed to be PIP2. Given the central position of PIP2 in phosphoinositide signalling, we fully expected to define an important coupling mechanism of receptors, in particular G-protein-coupled receptors (GPCRs), to ion transporters and channels in the heart. At least for the cardiac Na+–Ca2+ exchanger (NCX1) and for cardiac KATP channels, this has not clearly been the case up to now. Rather, our work to understand PIP2 in the heart has led us to a substantially different cell biological interpretation of the role of PIP2. While not discounting other roles and possiblities, it seems well established at present that PIP2 is a surface membrane marker, which activates a large variety of channels and transporters when they are inserted into the surface membrane. The relative lack of PIP2 on internal membranes will silence these same transporters and channels during their trafficking, when their activity might be detremental to the ion homeostasis of internal membrane compartments. Two articles accompanying this report outline evidence that in addition PIP2 metabolism may regulate NCX1 activity via its functional roles in the remodelling of cytoskeleton and endocytic processes (Yaradanakul et al. 2007; Shen et al. 2007). This report will subsequently summarize progress in addressing several questions that must be answered in order to understand the roles of PIP2 for Na+–Ca2+ exchangers and KATP channels in cardiac muscle.
Does regulation of KATP channels and NCX1 play a role in normal cardiac physiology?
To a rather surprising extent we are unsure about the detailed physiological roles of both KATP channels and Na+–Ca2+ exchange in cardiac myocytes, and this raises difficulties for understanding what role their regulation by PIP2 might actually play in cardiac physiology. Most authors would agree that, while the KATP conductance is among the largest of the heart, when activated, it plays a role only in circumstances of cardiac metabolic stress. The major problems for knock-out mice come in periods of cardiac stress, both acute and prolonged (Gumina et al. 2006; Tong et al. 2006). Similarly, it would seem that large changes of cardiac NCX1 expression do not fundamentally affect cardiac excitation–contraction coupling (ECC) (Henderson et al. 2004; Reuter et al. 2004; Pott et al. 2005). Changes in the number of transporters have little effect presumably because the exchanger is working across its thermodynamic equilibrium during ECC (i.e. is both a Ca2+ influx and extrusion mechanism). To summarize, up-regulation of KATP channels might help the heart to respond more sensitively to stress situations by shorting the action potential, bringing to bear an anti-arrhythmogenic influence, and decreasing cardiac frequency. Up-regulation of NCX1 would potentially decrease the number of contraction cycles needed to turn over the Ca2+ pool used in ECC and eventually help to unload extra Ca2+ that enters the myocytes at high frequencies and with adrenergic input.
While we have not been able to identify hormonal influences in the heart that strongly modify PIP2 or Na+–Ca2+ exchanger function (Nasuhoglu et al. 2002b; Yaradanakul et al. 2007), there is a potential influence of cytoplasmic Ca2+ that could be physiologically important. It is described by Berberian and colleagues (Berberian et al. 1998) that the generation of PIP2 on cardiac vesicles incubated with ATP is strongly dependent on Ca2+, being activated severalfold by 1 μm Ca2+. In contrast, the generation of PIP is not Ca2+ dependent. That this mechanism may be physiologically relevant is supported by our findings that PIP2 in heart, but not PIP, increases with increase of contraction frequency (Nasuhoglu et al. 2002b). We have verified the results of Berberian using crude cardiac membranes, and found that the stimulatory effect of Ca2+ on PIP2 levels was lost when vesicles were resuspended and recentrifuged. Furthermore, we have never seen evidence that stimulatory effects of ATP on cardiac K+ conductances were enhanced by cytoplasmic Ca2+. Thus, the mechanism is labile and not maintained in ‘washed' surface membranes. As shown in Fig. 1, we have found additionally that calmodulin can strongly promote the generation of PIP2, but not PIP, in crude cardiac membranes in a Ca2+-dependent manner. At this time, we know little about the underlying mechanism. The potential physiological role would be to promote KATP channel opening at high contraction frequencies, and with high Ca2+ loads, and to promote Ca2+ turnover across the cell membrane when the Ca2+ load is high, thereby supporting the already strong direct activating effect of Ca2+ on NCX1 (Hilgemann et al. 1992). In this same connection, it seems important that PIP2 levels in intact myocardium are severalfold higher than in isolated cardiac myocytes (Nasuhoglu et al. 2002a). Thus, KATP channel openings might occur more often than one might otherwise expect in the absence of metabolic stress. In our hands, high PIP2 can largely suppress inhibitory effects of high ATP concentrations (Yaradanakul et al. 2007).
Why does the heart have fast phosphoinositide turnover?
In cardiac muscle, as in most eukaryotic cells, PIP2 appears to be turning over rapidly to phosphatidylinositol via its dephosphorylation and rephosphorylation to PIP2. Our support for this impression comes from experiments in which PIP2 metabolism was perturbed and allowed to return to steady state (Nasuhoglu et al. 2002b). One perturbation that works in many cell types is shrinkage by hypertonic solution, followed by reswelling in isotonic solution. In many different cell types, we have found that both PIP and PIP2 rise in proportion to shrinkage within as little as 2 min, and return in a similar time frame to baseline upon reswelling. Changes of phosphatidylinositol mirror the changes of PIP and PIP2 quantitatively in most cells. In mouse hearts, for example, the sum of PIP and PIP2 increases by 0.8% of total anionic phospholipid with 250 mm sucrose, while phosphatidylinositol decreases by 0.9% of anionic phospholipid. In BHK cells a doubling of total PIP and PIP2 corresponds to an increase by 1.2% of total anionic phospholipid, and phosphatidylinositol decreases by 1.4% of total (P < 0.01; D.W. Hilgemann & P. Dong, unpublished data). In cell lines, we have also used PH domains to follow changes of phosphoinositide metabolism, and at 37°C we find the PIP2 replenishes with time constants of 20–40 s after activating Phospholipase C's (PLC's) by either a few seconds of M1 receptor activation or Ca2+ influx by Na+–Ca2+ exchanger. From all these results, it seems certain that the average PIP2 turnover time is physiologically less than 2 min.
Why is the basal turnover rate so high? First, fast basal PIP2 turnover will allow faster signalling when PLCs are activated and then inactivated. Second, fast dephosphorylation of PIP2 will tend to localize PIP2 signals at the sites of lipid synthesis. Third, the rapid turnover of PIP2 to phosphatidylinositol may simply reflect that PIP2 metabolism is part-and-parcel of the continuous remodelling of membrane cytoskeleton and the turnover of surface membrane via trafficking mechanisms (Padron et al. 2003; Yin & Janmey, 2003). Our analysis of the effects of PIP2 depletion on NCX1 activity suggest that the exchanger is affected by both cytoskeleton remodelling and membrane trafficking with changes of PIP2 (Yaradanakul et al. 2007).
Is PIP2 a physiological ligand at cardiac ion channels and transporters?
PIP2 interactions with ion channels and transporters are in general not highly specific, and other anionic phospholipids can substitute for PIP2 (Hilgemann et al. 2001). Even more impressively, acyl CoAs can evidently substitute to a substantial extent for PIP2 at both KATP channels (Schulze et al. 2003) and NCX1 (Riedel et al. 2006). Thus, we need evidence that PIP2 is really the critical anionic lipid physiologically within intact cells. For the neuronal M-current, an elegant molecular biological approach has resulted in definitive evidence that PIP2 is indeed the key physiological activator of the channels (Suh et al. 2006). We have used siRNA to knock-down the kinases that generate PIP2 (Wei et al. 2002; Wang et al. 2003, 2004) and test whether function is altered in excised patches, in particular whether the activation of channels and transporters by cytoplasmic ATP is affected. As shown in Fig. 2, we have found that knock-down of one specific PIP5K, the human PIP5K-Iβ, specifically abolishes the ability of ATP to support KATP channel activity in HEK293 cells.
These experiments use an HEK293 cell line that expresses both the SUR2A and Kir6.2 subunits that make up the cardiac KATP channels (Giblin et al. 2002a,b). Using siRNA (Wang et al. 2004), we reduced the expression of four different lipid kinases and analysed the effects on KATP currents in excised patches as illustrated in Fig. 2. Figure 2A shows the usual behaviour of this conductance in a giant excised patch. With an outwardly directed K+ gradient, substitution of Cs+ on the cytoplasmic side for K+ activates only a very small current. After application of 1 mm ATP, the current at first decreases and then increases (rebounds) over about 1 min. After ATP removal, the direct inhibitory effect of ATP is released, and the full current is observed. In the Mg2+-containing and Ca2+-free solutions employed, the current runs down back toward baseline within a few minutes. Application of PIP2 (30 μm) from the cytoplasmic side can fully reactivate the current, and afterward the current is largely insensitive to ATP. As shown in Fig. 2B, when the hPIP5K-Iβ is knocked down by siRNA, ATP has almost no activating effect on the current. However, PIP2 is fully effective in activating the current, thereby showing that the number of channels in the cells is not changed, nor is their sensitivity to activation by PIP2. These experiments verify that the stimulatory effects of ATP on Kir channels in excised patches are indeed due to the generation of PIP2 via lipid kinases, and they identify the major PIP5K, at least for HEK293 cells. However, we still need to apply new experimental means to test whether PIP2 is a major activating ligand at KATP channels and Na+–Ca2+ exchangers in intact heart, rather than some other ligand that may be lost or metabolized in excised patches.
Given the multiple functions of PIP2, it is not surprising that knock-down and over-expression of lipid kinases over the course of one or more days (i.e. several cell generations) give rise to complex phenotypes, and the complexities to be described next underscore the importance of new approaches that permit fast acute changes of PIP2 metabolism in cells (Suh et al. 2006; Varnai et al. 2006). On the other hand, the specific nature of some phenotypes suggests that individual lipid kinase isoforms indeed play specific roles (Wei et al. 2002; Yin & Janmey, 2003; Wang et al. 2004). Our results for over-expression of lipid kinases are illustrative.
While we could ‘knock out’ the stimulatory effect of ATP on KATP channels, as just described, we could not enhance the stimulatory effects of ATP in excised patches by over-expressing either PI4- or PIP5-kinases. In some cases, as described in Fig. 3, a ‘dominant negative’ effect was obtained as if a lipid kinase with low activity has supplanted one with high activity. As shown in Fig. 3, the stimulatory effect of ATP on NCX1 current, when assessed in excised patches from BHK cells (Linck et al. 1998), was routinely rather small. It was established, however, that the current could be powerfully activated by exogenous PIP2. Thus, expression of more lipid kinase activity in the surface membrane was expected to enhance the stimulation of current by ATP. That was not the case, however, for any lipid kinase tested, and over-expression of the PIP5K-Iα actually reduced the stimulatory effect of ATP.
In another example, the 10-fold overexpression of a PI4-kinase in a transgenic mouse did not enhance the stimulatory effect of ATP in excised patches, and the exchange current densities in patches were actually decreased (Shen et al. 2007). A general explanation of these results is that enhanced phosphoinositide generation at the cell surface leads to the removal of lipid kinases, as well as transporters, and/or their mislocation to internal membranes. A general explanation may be that endocytosis is PIP2 dependent. Also, specific partners needed for lipid kinase targeting and activation (Doughman et al. 2003; Perez-Mansilla et al. 2006) may be in short supply.
Are there local PIP2 concentrations and/or local PIP2 signals in the sarcolemma of cardiac cells?
This question states one of the most important issues that must be resolved to define the significance of PIP2 metabolism in the heart. According to Cho and colleagues (Cho et al. 2005b), individual cardiac GPCRs can control individual cardiac ion channels in a directed manner because PIP2 diffusion in the cardiac membrane is several hundred times slower than expected for simple phospholipid bilayers (Cho et al. 2005a). If correct, the results define fundamentally new possibilities for membrane signalling. Also, according to Morris and colleagues (Morris et al. 2006) PIP2 metabolism in myoctes is occurring in so-called raft domains which show an enhanced level of PIP2 and larger changes of PIP2 during receptor activation than the bulk of surface membrane.
In response to these results, we attempted similar experiments in multiple cell types using ‘giant’ patch pipettes to perfuse phospholipid liposomes into cells. We found that pipette perfusion of fluorescent phospholipids into cells results in diffuse labelling of the entire cytoplasm with little preferential labelling of the surface membrane. Therefore, in our experience, it would be problematic to define a diffusion coefficient with exogenous phospholipids in intact cells. Instead, we used PH-domains to characterize PIP2 gradients in cells across a giant patch pipette boundary in whole-cell configuration. Using BHK and CHO cells expressing muscarinic M1 receptors, PIP2 depletion transmits rapidly across the pipette wall when receptors are activated within the pipette. And vice versa, activation of receptors outside of the pipette results in rapid depletion of PIP2 within the pipette (A. Yaradanakul & D.W. Hilgemann, unpublished observations). While these experiments give no evidence for restricted PIP2 diffusion, they are admittedly limited in important respects. For example, membrane cytoskeleton is strongly dependent on PIP2, and PIP2 diffusion is proposed to be dependent on intact cytoskeleton (Cho et al. 2005a). If so, depletion of PIP2 will disrupt the diffusion barrier.
Conclusions
Every critical reader of the PIP2 literature asks how a single molecule can be involved in so many cellular processes and still be used as a second messenger in signalling. One simple mechanism to allow selectivity would be that the head group of PIP2 can be metabolized, cleaved or dephosphorylated, even when the phospholipid side chains are bound by membrane proteins. and for cytoskeletal proteins this tentatively appears to be possible (Fukami et al. 1994). But we can also be wrong about some of the roles suggested for PIP2. As outlined in this review, PIP2 may possibly play only a constitutive role at cardiac ion channels and transporters. Some of the dramatic results obtained for membrane transporters and channels in excised patches (e.g. Yaradanakul et al. 2007) may actually not apply to intact cells because other lipids can assume the same activating function in intact cells. The internalization of membrane transporters, such as NCX1, is another point at which we suggest that PIP2 may play a critical role (Shen et al. 2007). It is proposed that a local generation of PIP2 on budding vesicles is part of the endocytic mechanism (Boucrot et al. 2006; Massol et al. 2006). This model requires that PIP2 diffusion is restricted across the neck of a 50 nm membrane bud. If the model is correct, the PIP2 that activates Na+–Ca2+ exchangers in the surface membrane and the PIP2 that supports their internalization may be from different sources and may not mix with one another significantly in the surface membrane. As 50 nm is still beyond the resolution of standard optical methods, it is clear that the application and improvement of new imaging technologies will be required for real progress with these issues. In conclusion, the molecular and biophysical details of PIP2 compartmentation promise to be a worthy challenge for many years to come.
Acknowledgements
I express gratitude to all members of our laboratory and collaborators in these projects who are coauthors of the accompanying articles on PIP2 and NCX1 (Yaradanakul et al. 2007; Shen et al. 2007). The HEK293 cell line expressing Kir6.2 with SUR2A was kindly provided by Dr Andrew Tinker (University College, London). The work described was supported by NIH-HL0679420 and HL051323.
Footnotes
-
(Received 19 March 2007; accepted after revision 20 April 2007; first published online 26 April 2007)
-
This report was presented at The Journal of Physiology Symposium on Regulation of ion channels and transporters by phosphatidylinositol 4,5-bisphosphate (PIP2), Baltimore, MD, USA, 2 March 2007. It was commissioned by the Editorial Board and reflects the views of the author.
- 2007 The Author. Journal compilation © 2007 The Physiological Society
References
Figure 1. Calmodulin-dependence of PIP2 in crude cardiac membranes Relative concentrations of PIP and PIP2 in crude cardiac membranes incubated with 0.5 μm free Ca2+ (1 EGTA and 0.5 mm CaCl2 at pH 7.0), 0.5 mm ATP, and the given concentrations of calmodulin (CAM). PIP2, but not PIP, is markedly enhanced by micromolar free Ca2+ and calmodulin in cardiac membranes.
Figure 2. KATP potassium currents in giant excised patches from HEK293 cells expressing cardiac KATP channels (Kir6.2 with SUR2A) The solutions (Hilgemann & Ball, 1996) are chosen so that KATP channels generate a large outward K+ current upon opening. A, control patch. Currents are defined by substituting 120 mm Cs+ for 120 mm K+ on the cytoplasmic side. Initially the K+ current is very small. Application of ATP (1 mm) immediately inhibits current and then causes current rebound. The full magnitude of the current is then revealed upon removing ATP, which directly inhibits the channels. In the Mg2+ (0.5 mm)-containing and Ca2+-free (1 mm EGTA) solution, the current runs down with a time constant of about 2 min. Thereafter, application of PIP2 (30 μm) can fully reactivate the current, and thereafter the direct inhibitory effect of ATP is largely suppressed. B, patch from an HEK293 cell transfected with siRNA against hPIP5K-Iβ. The ability of ATP to activate KATP channels is largely abolished, while PIP2 remains fully effective.
Figure 3. Typical outward NCX1 currents in giant excised patches from BHK cells with stable NCX1 expression Under the conditions of these experiments (Hilgemann & Ball, 1996), maximal outward exchange currents are activated by substituting 40 mm Na+ for 40 mm Cs+ on the cytoplasmic side. The currents typically decay (inactivate) by about 75%. Addition of Mg-ATP to the cytoplasmic side (2 mm) activates currents by reducing inactivation, and exogenous PIP2 would activate the current further (not shown). Overexpression of the hPIP5K-Iα results in substantial suppression, not enhancement, of the stimulatory effect of ATP, which reflects endogenous PIP5K activity (Hilgemann & Ball, 1996).
