Monitoring changes in membrane phosphatidylinositol 4,5-bisphosphate in living cells using a domain from the transcription factor tubby

Authors


Corresponding author A. Tinker: BHF Laboratories and Department of Medicine, University College London, 5 University Street, London WC1E 6JJ, UK. Email: a.tinker@ucl.ac.uk

Abstract

Phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P2) is a key component in signal transduction, being a precursor to other signalling molecules and itself associated with roles in signal transduction and cell biology. Tubby is a membrane bound transcription factor whose dysfunction results in obesity in mice. It contains a domain that selectively binds PtdIns(4,5)P2. We have investigated the use of a fluorescently tagged version of this domain to monitor changes in PtdIns(4,5)P2 concentration in living cells and compared it to the pleckstrin homology domain of PLCδ1. Our results show that selected mutants of this domain report receptor-mediated changes in cellular PtdIns(4,5)P2. In contrast to the pleckstrin homology domain of PLCδ1 it does not have a significant affinity for inositol 1,4,5-trisphosphate (IP3). Using a selected mutant, we examine the regulation of ATP-sensitive K+ channels via a Gq/11-coupled receptor. These experiments reveal a correlation between reporter translocation and the onset of current inhibition whilst the recovery of current after agonist removal is delayed when compared to the reporter. Furthermore our studies reveal the importance of Ca2+ in determining the overall activity of phospholipase C in living cells. This probe may be valuable in examining changes in PtdIns(4,5)P2 distinct from those of IP3 in intact cells in a variety of physiological settings.

Membrane phosphoinositides form a small total proportion of all membrane lipids and yet they have important roles in signalling. Specifically they make up ∼7% of all plasma membrane lipid, and PtdIns(4,5)P2 is a small component of this (∼5%) (Cullis et al. 1996; Voelker, 1996). PtdIns(4,5)P2 is the substrate of phospholipase C and PI-3 kinase and leads to the generation of important second messengers in signalling. However, it has become increasingly clear that PtdIns(4,5)P2 can act as a signalling molecule in its own right. An area of recent physiological interest is the role of this anionic phospholipid in the regulation of ion channels and transporters (Hilgemann et al. 2001; Suh & Hille, 2005; Gamper & Shapiro, 2007). For example, it is clear that PtdIns(4,5)P2 is an important cofactor for the activity of a number of K+ channels (Fan & Makielski, 1997; Huang et al. 1998; Zhang et al. 1999; Quinn et al. 2003; Delmas & Brown, 2005; Suh et al. 2006). However, it is not so clear whether physiological fluctuations in the levels of PtdIns(4,5)P2, as may occur with the activation of cell signalling pathways, are sufficient by themselves to regulate ion channels. For example, agonists at Gq/11-coupled muscarinic receptors have a profound influence on the excitability of peripheral and central neurones through the inhibition of the M-type K+ current (Brown & Adams, 1980; Marrion, 1997; Delmas & Brown, 2005). The second messenger has remained elusive and the hydrolysis of PtdIns(4,5)P2 by phospholipase C is now a leading potential mechanism (Suh & Hille, 2002; Zhang et al. 2003; Suh et al. 2006).

The ability to monitor changes in the levels of these anionic phospholipids in living cells is obviously critical. To some extent this can be realized using biochemical techniques (Willars et al. 1998); however, there has been considerable interest in the use of fluorescent biosensors that report changes in vivo. The pleckstrin homology (PH) domain of phospholipase Cδ fused to a fluorescent protein (PH fused to the cyan fluorescent protein in our case, PH-CFP) is one such tool (Stauffer et al. 1998; Meyer & Oancea, 2000; Varnai & Balla, 2006). This chimaeric protein binds PtdIns(4,5)P2 with high affinity (Kd∼2 μm) but it also binds inositol 1,4,5-trisphosphate (IP3, Kd∼0.1 μm) (Hirose et al. 1999). Thus whilst the reporter clearly translocates from the plasma membrane to the cytosol after agonist activation of a relevant receptor it is not clear whether this is because of PtdIns(4,5)P2 depletion or IP3 generation or a combination of the two (Hirose et al. 1999; Nash et al. 2001; Nahorski et al. 2003; Xu et al. 2003).

Defects in the transcriptional regulator tubby lead to maturity onset obesity and neurosensory problems in mice. The tubby mouse has a loss of function mutation in the Tub gene. In addition there are three additional tubby like proteins in the mouse (Carroll et al. 2004). Interestingly this protein is membrane-associated through a PtdIns(4,5)P2 binding domain (tubby domain). It has been proposed that activation of Gq/11 coupled receptors and PtdIns(4,5)P2 depletion lead to the membrane release of the full-length protein and nuclear translocation (Santagata et al. 2001). Specifically, it was shown using biochemical techniques that the domain bound specifically to bidentate-phosphorylated phosphoinositide lipid head groups in particular PtdIns(4,5)P2. However the behaviour of the tubby C-terminal domain in living cells and its potential to acts as a reporter for changes in PtdIns(4,5)P2 levels have not been examined in detail. We have previously demonstrated that the translocation of yellow fluorescent protein (YFP) tagged tubby mutant correlates with aspects of M-type K+ channel regulation in native sympathetic neurons (Hughes et al. 2007). In this study we delineate the fundamental properties of this reporter and compare the tubby domain, and a number of selected mutants, with PH-CFP as potential biosensors for membrane PtdIns(4,5)P2. One mutant seems to be a useful probe and this should be valuable in examining changes in PtdIns(4,5)P2 distinct from those of IP3 in intact cells in a variety of physiological settings.

Methods

Molecular biology

The tubby domain (amino acids 243–505) was amplified using PCR from full length tubby cDNA (tubby in pcDNA3, kindly provided by Professor L. Shapiro) and fused in frame using a KpnI and XhoI digest to yellow fluorescent protein (eYFP-C1 and eYFP-N1 vectors, Clontech) at the N- (tubby-nYFP) and C-terminus (tubby-cYFP). PH-CFP, CFP-Inp (a truncated version of the yeast inositol polyphosphate 5-phosphatase that specifically cleaves the phosphate at the 5 position of PI(4,5)P2 fused to CFP in frame with a domain from the FK506 binding protein and cyan fluorescent protein) and Lyn11-FRB (a membrane anchored domain from mTOR) cDNAs were kindly provided by Professor T. Meyer (Stauffer et al. 1998; Suh et al. 2006). IP3 5-phosphatase was kindly provided by Professor M. Iino (Hirose et al. 1999). Point mutations were introduced using the QuickChange kit (Stratagene). All constructs were sequenced to confirm their identity.

Cell culture

HEK293 cells were transiently transfected with each of these fluorescent constructs along with the M3 receptor using lipid-based methods as previously described (Quinn et al. 2003, 2004). Cells were visualized 24–48 h after transfection using a confocal microscope, and superfused with bath solution (pH 7.4) containing (mm): 130 NaCl, 5 KCl, 10 Hepes, 8 glucose, 2.6 CaCl2 and 1.2 MgCl2. The Ca2+ free solution omitted the Ca2+. Incubations with wortmannin and BAPTA-AM were relatively short (∼5 min). Values are given from at least two transfections.

Cell imaging

Cells for imaging were subcultured onto 35 mm culture dishes with integral no. 0 glass coverslip bottoms (Mattek). Cells were imaged using a Bio-Rad Radiance 2100 confocal microscope using a 60× Nikon Plan Apo oil objective (1.40 NA). CFP was excited with a 457 nm laser line and images were obtained using a 470–500 nm band pass filter. YFP was excited with a 514 nm laser line and emission measured between 530 and 570 nm. Dual imaging of CFP and YFP tagged constructs was performed simultaneously with minimal cross-talk between the two imaging channels. Intensities in the CFP and YFP channels were determined from cytoplasmic regions of interest drawn by hand at high magnification using the LaserPix software. Attempts to compare membrane-to-cytoplasmic fluorescence ratios were often confounded by changes in cell morphology during the experiment. Thus in general statistics were compiled from changes in cytoplasmic fluorescence derived from a region of interest placed within cytoplasm that was not affected by cell movement. Fluorescence was normalized to the initial baseline measurement of fluorescence intensity within the ROI (Fluorescence/FluorescenceControl). Cells were perfused with a gravity driven perfusion system. However we did perform a limited subset of experiments where we did analyse the membrane signal (see below).

Intracellular perfusion of cells with IP3

Standard patch clamping techniques were combined with confocal microscopy. Whole-cell patch pipettes were manufactured from borosilicate glass (OD 1.5 mm, ID 1.2 mm) using a PP-830 puller and fire-polished using a MF-830 microforge (both Narishige, Japan) to give pipette resistances of 2–3.5 MΩ. Membrane currents were studied with the whole-cell patch-clamp technique using an Axopatch 200B amplifier (Axon Instruments, Union City, CA, USA). Currents were filtered at 1 kHz and displayed on an oscilloscope. High resistance gigaohm seals were obtained and then a standard whole-cell configuration obtained after membrane rupture to allow dialysis of the intracellular contents. Pipette solutions (pH 7.2) contained (in mm): 107 KCl, 1.2 MgCl2, 1 CaCl2, 10 EGTA, 5 Hepes, 1 (Mg)ATP and 100 μm IP3 (adjusted to pH 7.2 with KOH, giving 140 K+).

Combined electrophysiology and imaging

Cells stably expressing Kir6.2 (Giblin et al. 2002) were grown in 300 μl chambers made by casting rings of Silastic MDX4-4210 (a gift of Dow-Corning) on 25 mm coverslips. At about 60% confluency they were transfected with plasmid DNAs coding for tubby-cYFP-R332H and for the M3 receptor. They were patched in the perforated whole-cell mode 24 to72 h later, on the stage of a Bio-Rad radiance 2100 confocal microscope using an Axopatch 200A amplifier (Molecular Devices). Clampex 9 (Molecular Devices), running on a portable computer equipped with an Adaptec 1460 SlimSCSI card, was used to drive the experiment via a Digidata 1322 digitizer (Molecular Devices). It generated voltage protocols and TTL pulses to command an automated gravity fed perfusion system (1.4 ml min−1) and to trigger frame acquisition by the confocal microscope (excitation at 514 nm/emission in a 530–570 nm band).

Images were obtained at 4 s intervals during application of carbachol and at 10 s interval during wash out. The complete imaging sequence lasted 4 min 20 s and counted 42 images. Each image was the Kallman average of three frames (512 × 512 pixels) scanned at 600 lines s−1. The confocal software sent tags that confirmed the completion of each frame. These were recorded by the acquisition board along with the membrane current.

The bath contained (in mm): 130 NaCl, 5 KCl, 10 Hepes, 1.8 CaCl2, 1 MgCl2 (pH adjusted to 7.35 with NaOH). The pipette solution contained (in mm): 33 KCl, 75 potassium gluconate, 10 NaCl, 1.2 MgCl2, 10 Hepes, 2 EGTA (pH adjusted to 7.3 with KOH; total K+ 120 mm). It was supplemented with 200 μg ml−1 amphotericin B (Sigma).

Experiments were carried out at 27–28°C. Clampfit 9 was used for analysis of the membrane currents. Images were analysed with the free ImageJ program (http://rsb.info.nih.gov/ij/).

From the perspective of correlating fluorescence with channel activity, assessing the loss of membrane fluorescence on carbachol application seems more relevant than measuring an increase in cytoplasmic fluorescence. Since the membrane was only clearly distinguishable at both end of our sequence, i.e. before tubby had left the membrane and after it has returned there, we used the first and last image to define, on each, the membrane compartment as a band of about 1 μm width. To minimize the influence that cell movement may have had on the membrane location during the sequence, the fluorescence within the two regions of interest so defined was measured on each image of the sequence, and the measures are weighted according to the proximity, in time, of the reference frames. The membrane fluorescence level attributed to each image is the so weighted mean of the measures in the initial and final bands. There was little bleaching of the reporter during sequential image acquisition.

Statistical analysis

Data are expressed as means ±s.e.m. Student's paired t test or one-way ANOVA with Bonferroni's or Dunnett's post hoc test was used as appropriate to calculate statistical significance using Origin 6.0 (OriginLab Corp., Northampton, MA, USA) or Prism v4.0 (GraphPad, San Diego, CA, USA). NS: not significant at the 5% level; asterisks: *P < 0.05, **P < 0.01 and ***P < 0.001. (Tocris Bioscience, Bristol, UK).

Results

Using standard molecular cloning techniques we fused the yellow fluorescent protein to the tubby domain at the N (tubby-nYFP) and C-terminus (tubby-cYFP). We transiently transfected these constructs into HEK293 cells and imaged the cells using confocal microscopy. It was clear that both constructs were present at the plasma membrane in agreement with Santagata et al. (2001)(Fig. 1A). We next examined whether these constructs translocated when coexpressed with the Gq/11 coupled muscarinic M3 receptor and subsequently stimulated with an appropriate agonist. Neither of these constructs translocated prominently from the plasma membrane under these conditions (Fig. 1A). There was a small degree of translocation with the tubby-nYFP domain and none with tubby-cYFP (Fig. 1A). This was a surprising result given the observation that activation of muscarinic receptors leads to substantial depletion of PtdIns(4,5)P2 in transiently transfected mammalian cells (Horowitz et al. 2005). We reasoned the affinity of these constructs for membrane PtdIns(4,5)P2 was high and that cellular fluctuations in membrane PtdIns(4,5)P2 after receptor activation were insufficient to lead to translocation of either of these reporters to the cytoplasm. Thus we incubated the cells with wortmannin at high concentrations (50 μm) that broadly inhibit phosphatidylinositol kinases (Fruman et al. 1998) and would thus affect PtdIns(4,5)P2 synthesis and further accentuate depletion on agonist application. The application of carbachol now led to an irreversible translocation of both domains though this was much more pronounced with the tubby-cYFP domain (Fig. 1B). We compared this behaviour to that of the PH domain fused in frame with CFP in similar conditions. When cotransfected with the M3 receptor, PH-CFP reversibly translocated after agonist application and removal. Incubating the cells with wortmannin at high concentrations (50 μm) had little effect on the behaviour (Fig. 2A). The mean increase in normalized cytoplasmic fluorescence in the various conditions is summarized in Fig. 1C for tubby-nYFP and tubby-cYFP and Fig. 2B for PH-CFP.

Figure 1.

Membrane localization of tubby-nYFP and tubby-cYFP, and translocation of these constructs from the membrane into the cytoplasm in response to M3 receptor stimulation
A, example of membrane localization of tubby-nYFP and tubby-cYFP, with effect of carbachol in cells contransfected with M3 and tubby-nYFP or M3 and tubby-cYFP. Left panels shows cells perfused with bath solution, centre panel shows cells in the presence of 10 μm carbachol and right panel shows cells after washout of carbachol. An example of a cross-section of a cell (red line) is inset, and example changes in fluorescence within a region of interest (bounded by a red box) are plotted to the right of each set of panels. B, effect of carbachol after preincubation for 10 min with 50 μm wortmannin. C, summary of data for tubby-cYFP and tubby-nYFP, before and after treatment with 50 μm wortmannin. n numbers are indicated in brackets and are from at least 2 transfections.

Figure 2.

Membrane localization of PH-CFP and translocation of this construct into the cytoplasm in response to M3 stimulation
A, top panels show examples of membrane localization of PH-CFP, and reversible translocation in response to carbachol in M3 cotransfected cells. Bottom panels show examples of translocation after preincubation with 50 μm wortmannin using and imaging the same field of cells. An example of a cross-section of a cell (red line) is inset, and example changes in fluorescence within a region of interest (bounded by a red box) are plotted to the right of each set of panels. B, summary of data for PH-CFP, before and after treatment with 50 μm wortmannin.

We noted that tubby-nYFP behaved less predictably and also led to a change in cell morphology: quite often the cells would become rounded. The high level expression of these domains may significantly decrease membrane PtdIns(4,5)P2 and this may result in changes in cytoskeletal interactions (Raucher et al. 2000). In contrast tubby-cYFP did not possess these properties and thus we chose to develop this construct further. Our results are compatible with the tubby domain having a high affinity for membrane PtdIns(4,5)P2. We hypothesized that weakening the interaction of the protein for the anionic phospholipid might result in a more useful reporter. From the crystal structure it is apparent that there are a number of key residues responsible for contacting the PtdIns(4,5)P2 head group (Santagata et al. 2001). We made five conservative mutations (equivalent to N310Q, R332H, R363K, K330R and K330H in the full length tubby protein and referred to as such here) in tubby-cYFP and expressed the constructs in HEK293 cells. All these constructs were membrane localized and now translocated reversibly in response to carbachol. Figure 3A and B shows the behaviour of three of these mutants (K330H, R363K, R332H). We further examined the effect that the phospholipase C inhibitor (U73122 at 10 μm) might have on the translocation of the tubby-cYFP-R332H mutant. We found that application of this agent severely abrogated the carbachol-induced translocation (Fig. 3B). Figure 3C summarizes the mean increase in normalized cytoplasmic fluorescence for each tubby mutant and Fig. 3D the effect of U73122 on the translocation of tubby-cYFP-R332H.

Figure 3.

Membrane localization of tubby-cYFP point mutants, and translocation of these constructs into the cytoplasm in response to M3 stimulation
A, two examples of tubby-cYFP mutants (K330H and R363K) where panels show translocation from the membrane to the cytoplasm in response to carbachol in M3 cotransfected cells. An example of a cross-section of a cell (red line) is inset, and example changes in fluorescence within a region of interest (bounded by a red box) are plotted to the right of each set of panels. B, example of tubby-cYFP-R332H where top panels show translocation in response to carbachol in M3 cotransfected cells, and bottom panels show responses in the same cells following 10 min incubation with 10 μm of the phospholipase C inhibitor U73122. C, summary of mean increase in normalized cytoplasmic fluorescence in response to carbachol for all tubby-CYFP mutants (N310Q, R332H, R363K, K330R, K330H).

We chose to focus on the behaviour of tubby-cYFP-R332H as a potential reporter of membrane PtdIns(4,5)P2 changes. We first used a recently developed tool to alter membrane PtdIns(4,5)P2 levels (Suh et al. 2006). We cotransfected cDNAs encoding CFP-Inp (a truncated version of the yeast inositol polyphosphate 5-phosphatase that specifically cleaves the phosphate at the 5 position of PI(4,5)P2 fused to CFP and a domain form the FK506 binding protein) and Lyn11-FRB (a membrane anchored domain from mTOR together with tubby-cYFP-R332H. The addition of iRAP promotes herterodimerization of these two domains and thus the translocation of the phosphatase to the membrane. In practice, we found some cells showed cytoplasmic localization of tubby-cYFP-R332H even before the addition of iRAP (not shown). However titration of cDNA levels so that the expression of CFP-Inp was reduced led to membrane localization of tubby-cYFP-R332H and translocation to the cytoplasm on the addition of iRAP with the concomitant movement of CFP-Inp to the membrane (Fig. 4). We observed this behaviour in two other separate transfections in a number of cells.

Figure 4.

The dependence of tubby subcellular localization on membrane PtdIns(4,5)P2
In cells coexpressing tubby-cYFP-R332H, CFP-Inp and Lyn11-FRB, the addition of 5 μm iRAP leads to the translocation of tubby-cYFP-R332H to the cytosol (left panels), line profile of tubby-cYFP-R332H at the point indicated (middle panels) and the translocation of CFP-Inp to the membrane (right panels). The latter signal is weak (see text) but nonetheless some redistribution is visible.

We next tested the IP3 dependence of translocation by directly introducing IP3 into the cells at a fixed high concentration (100 μm) using the whole-cell configuration of the patch-clamp combined with cell imaging. Importantly this allows us to control the IP3 concentration but also to fix the intracellular ion concentrations. In particular we are able to dialyse the cell with a high concentration of EGTA (calculated free Ca2+ in these solutions is ∼20 nm) that will act as a significant calcium buffer. Thus this technique allows us to examine the effect of IP3 whilst controlling the Ca2+ concentration. Dialysis of IP3 into the cell over a period of 3–4 min led to the translocation of the PH-CFP domain from the membrane to the cytoplasm (Fig. 5A). In contrast, tubby-cYFP and tubby-cYFP-R332H did not translocate under these conditions (Fig. 5A). Access to the intracellular contents on breakthrough was confirmed by the measurement of a capacitance transient consistent with our previous extensive characterization of these cells using electrophysiological techniques (Quinn et al. 2003, 2004). Figure 5B summarizes the mean increase in normalized cytoplasmic fluorescence. Thus tubby-cYFP and tubby-cYFP-R332H do not seem to have a substantial affinity for IP3in vivo in contrast to PH-CFP.

Figure 5.

Injection of IP3 into cells using patch-clamp pipette
A, examples of PH-CFP, tubby-cYFP and tubby-cYFP R332H before patching (left panel), after sealing (left centre panel) and after 2–5 min breakthrough (right hand panels). B, summary of normalized fluorescence in a cytoplasmic region of interest during dialysis of the cell with 100 μm IP3 with a patch pipette for PH-CFP, tubby-cYFP and tubby-cYFP-R332H.

We addressed the issue of whether these domains were responding to IP3 using another tool. The IP3 phosphatase was expressed together with the relevant domain and the M3 muscarinic receptor. The expression of the IP3 phosphatase abolished the translocation of PH-CFP, tubby-cYFP-R332H and tubby-cYFP (and in the presence of wortmannin) (Fig. 6). How is this result consistent with the one above? One difference is that in the experiments in Fig. 5, Ca2+ is likely to be better buffered and the Ca2+ rise smaller as the pipette solution contains 10 mm EGTA. Furthermore it is clear that in biochemical studies the activity of phospholipase C is Ca2+ dependent and the same is true in intact cells (Allen et al. 1997; Rebecchi & Pentyala, 2000; Young et al. 2003; Horowitz et al. 2005). Finally, we have previously shown pronounced increases in Ca2+ in intact HEK293 cells transfected with the M3 receptor (Leaney et al. 2004). Thus, we examined the possibility that the overexpression of IP3 phosphatase eliminates the carbahol-induced Ca2+ transient needed for the feed-back activation of PLC and this accounts for the lack of translocation seen in Fig. 6. The issue is to compare the relative Ca2+ dependence of the two constructs. The hypothesis is that IP3 will be produced even with low, buffered Ca2+ on the application of carbachol; however, the relative activity of PLC will be lower because there will not be a contribution from for example the Ca2+ sensitive PLCδ. Under these conditions, the translocation of tubby-R332H-YFP should be more Ca2+ dependent as it is not IP3 sensitive whilst the translocation of PH-CFP should be less Ca2+ dependent as IP3 is still produced (though less IP3 is probably produced under these conditions). As a result we examined the translocation of these domains in Ca2+ free bath solution and in Ca2+ free bath solution together with BAPTA-AM. BAPTA-AM is a cell-permeant ester of the Ca2+ chelator BAPTA. Under both these conditions the relative translocation to the cytoplasm of tubby-cYFP-R332H was reduced compared to that of PH-CFP (Fig. 7). In addition, we can also make comparisons within a probe. Thus comparing relative translocation for PH-CFP we find that there is no significant difference (P > 0.05) between control, no external Ca2+ and no external Ca2+/BAPTA-AM (one way ANOVA with Dunnett's post hoc test). In contrast for tubby-R332H-YFP there is a significant difference (control versus no external Ca2+, P < 0.05; control versus no external Ca2+/BAPTA-AM, P < 0.001) and thus we conclude that the translocation of this probe is more dependent on Ca2+ mobilization. Finally to further demonstrate this difference within the same cell, we cotransfected PH-CFP, tubby-cYFP-R332H and the M3 receptor into cells and simultaneously imaged CFP and YFP. In Ca2+ free bath solution after the application of agonist the two constructs behave differently with tubby-cYFP-R332H translocating less prominently than PH-CFP (Fig. 8A and B). A similar positive feedback regulation of PLC has been observed in heterologous expression systems by other investigators (Horowitz et al. 2005).

Figure 6.

Translocation of PH-CFP, tubby-cYFP and tubby-cYFP-R332H with overexpression of IP3 5-phosphatase
AD, examples of translocation responses to carbachol in M3-contransfected cells transfected as indicated. A and B are the same field of cells. E, summary of mean increase in normalized cytoplasmic fluorescence in response to carbachol for PH-CFP, tubby-cYFP, tubby-cYFP with wortmannin and tubby-cYFP-R332H, all with overexpressed IP3 phosphatase.

Figure 7.

Dependence on Ca2+
A, translocation of tubby-cYFP-R332H in absence of bath Ca2+ (top panels), and after incubation with 20 μm BAPTA-AM (second row of panels). B, translocation of PH-CFP in Ca2+ free bath solution (top row of panels) and after preincubation with 20 μm BAPTA-AM (2nd row of panels). An example of a cross-section of a cell (red line) is inset, and example changes in fluorescence within a region of interest (bounded by a red box) are plotted to the right of each set of panels. C, summary of mean increase in normalized cytoplasmic fluorescence in response to carbachol for PH-CFP and tubby-cYFP-R332H in the presence and absence of Ca2+, and in the absence of Ca2+ after pretreatment with 20 μm BAPTA-AM.

Figure 8.

Dual imaging of cotransfected PH-CFP and tubby-cYFP-R332H in the same cell
A, example of translocation of tubby-cYFP-R332H from the membrane to the cytoplasm in response to carbachol in M3 and PH-CFP cotransfected cells. Example line scan profiles (red line) are inset, and example changes in fluorescence within a region of interest (bounded by a red box) are plotted to the right of each set of panels. B, example of translocation of PH-CFP from the membrane to the cytoplasm in response to carbachol in M3 and tubby-cYFP-R332H cotransfected cells in the same field of cells as in A. C, summary of mean increase in normalized cytoplasmic fluorescence in response to carbachol for PH-CFP and tubby-cYFP-R332H in the same cell.

To test the potential of the tubby-cYFP-R332H reporter in signalling studies, we coexpressed it with the M3 muscarinic receptor in a HEK293 cell line stably expressing the clones (SUR2A and Kir6.2) constituting the cardiac ATP-sensitive K+ channel. We have previously established that Kir6.2 containing channels can be regulated via Gq/11 coupled receptors and membrane PtdIns(4,5)P2 (Quinn et al. 2003). After application of the channel opener pinacidil, simultaneous perforated whole cell recording and confocal imaging on these cells demonstrate that inhibition of the channel on carbachol application is concomitant with translocation of the tubby reporter (Fig. 9). Fitting the time course of current inhibition and loss of reporter fluorescence from the membrane with single exponential time constants gave an approximately 3-fold difference between the two (respectively 6.0 and 18.7 s, n= 4). However, recovery of the current was markedly slower than that of the membrane associated fluorescence level: on average the latter had returned to its initial level (with a time constant of 51.9 s) when only 43% of the current loss had been reversed. We also observed a clear over-recovery of the membrane bound fluorescence in 3 out of 4 experiments.

Figure 9.

The correlation between KATP channel regulation and tubby-cYFP-R332H translocation
A, upper section, perforated whole-cell current response to a sequence of channel opener, carbachol and channel blocker application. The cell was held at –80 mV, near the K+ reversal potential; the dotted line shows the zero current level. Depolarizing pulses to +20 mv at 1 or 5 s intervals were used to monitor the outward current level, largely dominated by a glibenclamide sensitive K+ conductance after application of pinacidil. Complete I–V relations were obtained at both ends of the sequence as well as before, during and after carbachol application. The vertical bars (green before, red during and blue after carbachol application) mark the time points at which an image was obtained. Lower section, three images of the patched cell acquired along the sequence at the time points marked by the arrows. Insets show the simultaneously recorded current responses to the depolarizing pulse. Note that the fluorescence returns to the membrane well before full recovery of the current. B, current–voltage relationships of the cell at various points in the sequence. Carbachol fully abolishes the pinacidil induced current. C, time course of the changes in fluorescence and current on application of carbachol, with origin at the onset of carbachol application.

Discussion

The use of fluorescent probes to monitor the activity of important biochemical pathways in living cells is a rapidly advancing field. Here we develop a novel probe for membrane PtdIns(4,5)P2 based on a domain from the protein tubby. Membrane PtdIns(4,5)P2 is a substrate for phospholipase C and PI-3 kinase, but in addition it has been proposed to act as a direct regulator of ion channel activity (Hilgemann et al. 2001). Depletion is generally associated with channel inhibition though in the case of TRP channels activation may occur (Voets & Nilius, 2007). However it is not clear whether significant depletion occurs after activation of phopholipase C and PI-3 kinase via different receptor mediated pathways and if so what is the dynamics of the response. Phosphatidylinositol lipids are a relatively minor component of membrane phospholipids and PtdIns(4,5)P2 is a small component of this (Cullis et al. 1996; Voelker, 1996). What consequences does activation of phospholipase C have on these levels? Biochemical work has revealed that PtdIns(4,5)P2 levels can fall by up to 60–70% (Creba et al. 1983; Willars et al. 1998). However, the remaining fraction is protected from depletion and once agonist is removed the levels of PtdIns(4,5)P2 recover rapidly (generally within a minute). PtdIns(4)P does not recover as quickly and in general it seems the cell seeks to maintain and replenish PtdIns(4,5)P2 at the expense of other phosphoinositides (Creba et al. 1983; Willars et al. 1998).

To address these questions in living cells it is important to have a tool that responds largely to PtdIns(4,5)P2. Meyer and colleagues pioneered the use of the pleckstrin homology domain from PLCδ but it also binds to IP3 and translocation is as much due the production of the latter as it is to depletion of PtdIns(4,5)P2, complicating the analysis (Stauffer et al. 1998; Hirose et al. 1999; Xu et al. 2003; Nahorski et al. 2003). Shapiro and colleagues studied the G-protein regulation of the transcription factor tubby and found in a number of biochemical assays that it only bound to PtdIns(4,5)P2, PtdIns(3,4)P2 and PtdIns(3,4,5)P3 in a selective fashion through a C-terminal tubby domain. It did not bind other inositol phospholipids (Santagata et al. 2001). The biochemical data and our results are consistent with the hypothesis that this domain largely responds to PtdIns(4,5)P2 in the cellular context. In our studies we used the domain alone fused to a fluorescent protein and found that in the absence of a mutation it did not translocate even after heterologous expression of a receptor and application of an appropriate agonist. Treatment with high concentrations of wortmannin to deplete cellular PtdIns(4,5)P2 levels, however, did lead to translocation suggesting that the affinity of the tubby domain for cellular PtdIns(4,5)P2 was high. The difference in the behaviour of full-length protein (Santagata et al. 2001) and that of the domain suggests a couple of possibilities. Other regions of the tubby protein might modulate the affinity of the PtdIns(4,5)P2 binding domain decreasing it in response to receptor activation. Alternatively specific G-protein coupled receptors might cluster with tubby and phospholipase C resulting in localized domains of PtdIns(4,5)P2 depletion. The translocation to and from the membrane was reversible in our experimental conditions.

We directly tested for response to IP3 using the whole-cell configuration of the patch-clamp to dialyse the cell with a supraphysiological concentration. Tellingly, PH-CFP translocated whilst tubby and a tubby point mutant (R332H) did not. However, further studies did reveal an indirect dependence of tubby translocation on IP3 generation. IP3 phosphatase abolished the translocation of all domains; however, in the case of the tubby domains this was related to the intracellular Ca2+ signal as Ca2+ free bath solution and BAPTA-AM attenuated the translocation of tubby and mutants but not PH-CFP. This is consistent with a Ca2+ dependence of one or more isoforms of phospholipase C. Thus Ca2+ generated after IP3-induced Ca2+ release leads to a positive feedback activation of phospholipase C. Such phenomena are important and could potentially influence the nature of Ca2+ oscillations (Young et al. 2003). The enzyme responsible could be one of the phospholipase Cβ isoforms but is probably more likely to be a phospholipase Cδ (Allen et al. 1997). Thus if intracellular Ca2+ is buffered, phospholipase C has a lower activity on activation but still leads to IP3 generation. The cell is able to largely maintain PtdIns(4,5)P2 levels via synthetic pathways and/or there is only modest relative PtdIns(4,5)P2 depletion. This scenario is reflected in the translocation of the PH-CFP domain but not the tubby-cYFP-R332H. If Ca2+ is allowed to rise, phospholipase C activity is potentiated and there is a rise in IP3 accompanied by a fall in PtdIns(4,5)P2 levels. Both PH and mutant tubby domains translocate. This hypothesis is consistent with our previous observations on the regulation of ATP-sensitive K+ channels by anionic phospholipids (Quinn et al. 2003). We could only demonstrate regulation of Kir6.2, which appears to be regulated largely through anionic phospholipids, in the perforated patch configuration. In contrast in the whole-cell configuration we could not see current inhibition on M3 receptor activation. In the latter recording mode, cytoplasmic Ca2+ was ∼20 nm and buffered with 10 mm EGTA, whilst in the former it is not and will increase on receptor activation.

We investigated this further and directly correlated SUR2A/Kir6.2 channel regulation via a Gq/11 coupled receptor with tubby-cYFP-R332H translocation. Such modulation of KATP currents is physiologically important in cardiac myocytes (Haruna et al. 2002). The initial inhibition of current, studied using perforated patch techniques, was correlated reasonably with the disappearance of membrane fluorescence consistent with our previous studies showing regulation of KATP channels containing Kir6.2 by anionic phospholipids and not PKC. The difference in the kinetics of initial inhibition of tubby-R332H-YFP translocation is not totally unexpected. Tubby-R332H-YFP contains a single binding site and its partitioning between the cytosol and membrane should be determined in principle by a simple binding isotherm. In contrast, KATP channels are tetrameric and contain four PtdIns(4,5)P2 binding sites. The relationship between channel activity and PtdIns(4,5)P2 density in the membrane may show significant cooperativity and a Hill coefficient greater than 1. The recovery from inhibition on agonist removal was, however, significantly slower than the recovery of the membrane tubby-cYFP-R332H fluorescence. This suggests that other factors may contribute. In our previous work we have seen little evidence for the involvement of protein kinase C in direct regulation of KATP channels containing Kir6.2; however, it cannot be fully excluded (Quinn et al. 2003). Furthermore, we have demonstrated strong biochemical binding of the C-terminal domains of inwardly rectifying K+ channels to PtdIns(4)P and this recovers more slowly than PtdIns(4,5)P2 levels (Willars et al. 1998; Thomas et al. 2006). Thus the initial inhibition on agonist application is closely related to PtdIns(4,5)P2 whilst recovery after agonist removal may be governed by additional processes.

We have also recently used tubby-cYFP-R332H in living sympathetic neurons and there was a good correlation between M-type K+ current suppression and tubby translocation on muscarinic receptor activation (Hughes et al. 2007). In addition, when bradykinin receptors were activated the translocation of this probe was much reduced in comparison. In this case the inhibition of channel activity is largely thought to be due to an increase in Ca2+ with relative preservation of membrane PtdIns(4,5)P2 levels (Winks et al. 2005; Brown et al. 2007). The behaviour of tubby-cYFP-R332H with of a variety of interventions supported this idea.

In summary we have developed a genetically encoded probe (specifically tubby-cYFP-R332H) that reports cellular PtdIns(4,5)P2 changes independently of IP3 generation. In contrast to reporters based on the PH domain, it translocates largely because of a fall in cellular PtdIns(4,5)P2 levels and its behaviour is relatively independent of IP3 generation per se. It should be useful to examine the pharmacology and dynamics of phospholipase C signalling in more native systems in physiological conditions.

Appendix

Acknowledgements

This work was supported by the British Heart Foundation and Wellcome Trust.

Ancillary