Cyclic AMP directs inositol (1,4,5)-trisphosphate-evoked Ca²⁺ signalling to different intracellular Ca²⁺ stores

Many cells respond to extracellular stimuli with an increase in intracellular free Ca²⁺ concentration, [Ca²⁺]_i. The complex spatio-temporal organization of these Ca²⁺ signals provides the versatility that allows them selectively to regulate many cellular processes (Berridge et al., 2000). Inositol (1,4,5)-trisphosphate (IP₃), by stimulating Ca²⁺ release from the endoplasmic reticulum (ER), provides a common link between receptors at the plasma membrane and cytoplasmic Ca²⁺ signals. Diverse stimuli, via their receptors, stimulate phospholipase C (PLC) and so formation of IP₃, which binds to IP₃ receptors (IP₃R) causing Ca²⁺ to leak into the cytosol (Foskett et al., 2007; Taylor and Tovey, 2010).

Parathyroid hormone (PTH) stimulates adenylyl cyclase (AC) via G protein-coupled receptors (GPCR). The type 1 PTH receptor (PTHR1) has attracted most attention because it has essential roles in bone remodelling and plasma Ca²⁺ homeostasis, and it is a target of drugs used to treat osteoporosis (Mannstadt et al., 1999). PTHR1 can stimulate both AC and an increase in [Ca²⁺]_i, but the relationships between these events are complex (reviewed in Taylor and Tovey, 2012). PTH invariably stimulates AC, but it can also stimulate PLC directly when the signalling proteins are expressed at high levels (Offermanns et al., 1996; Schwindinger et al., 1998; Taylor and Tovey, 2012) or when PTHR1 associates with the scaffold proteins, Na⁺-H⁺-exchanger regulatory factors (NHERF) (Wang et al., 2010). Cyclic AMP (cAMP), via exchange protein activated by cAMP (epac) and the monomeric G protein Rap, can also stimulate PLCε (Schmidt et al., 2001). We (Short and Taylor, 2000; Tovey et al., 2008; Tovey et al., 2003) and others (Schwindinger et al., 1998) find that in HEK 293 cells stably expressing physiological levels of PTHR1 (HEK-PR1 cells), PTH stimulates AC, but it does not alone evoke an increase in [Ca²⁺]_i. However, PTH (or endogenous β₂-adrenoceptors) potentiates the Ca²⁺ signals evoked by receptors that stimulate PLC (Kurian et al., 2009; Tovey et al., 2008). This effect of PTH is entirely mediated by cAMP, it requires activation of neither cAMP-dependent protein kinase (PKA) nor epacs, and it results from cAMP binding to IP₃R directly or to a protein that tightly associates with IP₃R (Tovey et al., 2008; Tovey et al., 2010). Although cAMP increases the sensitivity of all three IP₃R subtypes to IP₃ (Tovey et al., 2010), in intact HEK-PR1 cells a specific association of IP₃R2 and AC6, an ‘AC-IP₃R junction’, allows AC to deliver cAMP directly and at high concentration to IP₃R (Fig. 1A) (Tovey et al., 2008).

Lipid rafts are ordered regions of the plasma membrane in which the outer leaflet is enriched in cholesterol and sphingolipids (Simons and Toomre, 2000). Many signalling proteins, including GPCRs, G proteins, ACs, IP₃Rs and PLCs, associate with lipid rafts and are thereby organized into signalling complexes (Isshiki and Anderson, 1999; Ostrom and Insel, 2004; Simons and Toomre, 2000). Disruption of lipid rafts by cholesterol depletion has been reported to inhibit IP₃-evoked Ca²⁺ release (Nagata et al., 2007; Singleton and Bourguignon, 2004; Weerth et al., 2007) and to disrupt selective regulation of AC6 and AC8 by store-operated Ca²⁺ entry (Willoughby and Cooper, 2007). Evidence that delivery of cAMP from AC6 to IP₃R2 mediates the effects of PTH on IP₃-evoked Ca²⁺ signals in HEK-PR1 cells (Tovey et al., 2008) alongside reports that AC5/6 (Head et al., 2005; Ostrom et al., 2002; Willoughby and Cooper, 2007) and IP₃R2 (Weerth et al., 2007) can associate with cholesterol-rich lipid rafts prompted us to examine the role of lipid rafts in signalling by PTH.

Carbachol (CCh), a stable analogue of acetylcholine that stimulates IP₃ formation via endogenous M₃ muscarinic receptors (M₃R) (Luo et al., 2008), evoked Ca²⁺ release from the intracellular stores of HEK 293 cells (Short et al., 2000). This IP₃-evoked Ca²⁺ release was potentiated by PTH in HEK-PR1 cells (Fig. 1B), by isoproterenol, which stimulates endogenous β₂-adrenoceptors (supplementary material Fig. S1A) (Kurian et al., 2009), or by high concentrations of a membrane-permeant analogue of cAMP, 8-bromo-cAMP (8-Br-cAMP) (supplementary material Fig. S1B). Neither PTH nor isoproterenol alone stimulated Ca²⁺ release (Fig. 1B and supplementary material Fig. S1A). These results are consistent with earlier work suggesting that AC6-IP₃R2 junctions allow cAMP to be delivered directly to IP₃R at high concentrations to cause an increase in their sensitivity to IP₃ (Tovey et al., 2008; Tovey et al., 2010). Subsequent experiments assess whether cholesterol, an essential component of lipid rafts (Simons and Ikonen, 1997), is required for this selective communication (Fig. 1A).

Filipin-staining established that most cholesterol was in the plasma membrane and that exposure to β-methylcyclodextrin (βMCD, 2% w/v, 2 hours at 20°C) caused loss of plasma membrane cholesterol without perturbing cell morphology (Fig. 1C,D). Similar results were obtained with brief exposure (10 minutes) to βMCD (2% w/v) at 37°C (supplementary material Fig. S2A–C). In most experiments, the more prolonged (30 minutes to 2 hours) treatment at 20°C was used because it was then easier to adjust incubation times to achieve effective cholesterol depletion without causing cellular damage. Pre-treatment of cells with βMCD at 20°C massively attenuated the Ca²⁺ release evoked by CCh in populations of HEK-PR1 cells (Fig. 1E). The peak increase in [Ca²⁺]_i evoked by a maximal concentration of CCh (1 mM) was reduced by 80±11% from 234±22 nM to 51±26 nM (Fig. 1F). The sensitivity to CCh was not significantly affected: the pEC₅₀ (−logEC₅₀, where EC₅₀ is the half-maximally effective concentration) was 4.38±0.14 and 3.88±1.07 for control and βMCD-treated cells, respectively (Fig. 1F). CCh-evoked Ca²⁺ release was similarly inhibited by brief treatment with βMCD (2% w/v, 10 minutes) at 37°C (supplementary material Fig. S2D). The effects of cholesterol depletion on CCh-evoked Ca²⁺-release were reversed by cholesterol repletion (Fig. 1G).

Although loss of cholesterol almost abolished responses to CCh, it had no significant effect on the Ca²⁺ signals evoked by subsequent addition of PTH (Fig. 2A–C). Neither the amplitude of the Ca²⁺ signal evoked by addition of a maximal concentration of PTH after CCh (229±28 nM and 187±30 nM, in control and βMCD-treated cells, respectively) nor the sensitivity to PTH (pEC₅₀ = 7.25±0.21 and 7.11±0.17) was affected by loss of cholesterol (Fig. 2D). It is worth noting that in these and related experiments (Figs 2, 3; supplementary material Figs S1, S3B and S7), the Ca²⁺ signal evoked by addition of PTH directly reports the potentiated response because although CCh is present throughout, [Ca²⁺]_i has returned to its basal level before PTH is added. The results with cell populations were confirmed in single cells, where a maximal concentration of CCh (1 mM) evoked a transient release of Ca²⁺ in 92±3% of cells, and subsequent addition of PTH (100 nM) caused further Ca²⁺ release in 97±2% of cells (Fig. 2E–H). Pre-treatment with βMCD significantly reduced both the number of cells responding to CCh alone (to 63±7%, Fig. 2G) and the amplitude of the peak Ca²⁺ signal (Fig. 2H). Even in those cells that responded to CCh, the amplitude of the Ca²⁺ signal was reduced from 350±52 nM to 191±7 nM. However, βMCD had no effect on the number of cells in which PTH potentiated CCh-evoked Ca²⁺ signals (98±1% of cells, Fig. 2G) or the amplitude of the peak response to PTH (229±8 nM and 278±27 nM in control and βMCD-treated cells, respectively). It is important to recall that PTH evokes Ca²⁺ signals only when cells are co-stimulated with CCh (Fig. 1B), yet after βMCD-treatment many cells (37%) failed to respond to CCh alone, but almost all cells (98%) responded to CCh with PTH (Fig. 2G). Many βMCD-treated cells therefore responded to PTH with CCh despite not responding to CCh alone.

Addition of ionomycin, a Ca²⁺ ionophore, to cells in Ca²⁺-free HBS was used to define the amount of Ca²⁺ within intracellular stores (Fig. 2E,F). The results demonstrate that loss of cholesterol had no effect on the initial Ca²⁺ content of the stores (Fig. 2H, red), but their residual Ca²⁺ content after stimulation with CCh and PTH was significantly greater in βMCD-treated cells (Fig. 2H). We conclude that cholesterol depletion has no direct effect on the Ca²⁺ content of the intracellular stores, but it attenuates the Ca²⁺ signals evoked by CCh alone, without affecting those evoked by CCh with PTH.

Endogenous β₂-adrenoceptors in HEK 293 cells stimulate AC and thereby potentiate CCh-evoked Ca²⁺ release (supplementary material Fig. S1A). The response to β₂-adrenoceptors is larger in the parental HEK 293 cells; we therefore used these cells to examine responses to isoproterenol, a selective agonist of β-adrenoceptors. In populations of these cells, βMCD almost abolished the Ca²⁺ signals evoked by CCh (supplementary material Fig. S3A) without affecting the amplitude or sensitivity of the subsequent responses to isoproterenol (supplementary material Fig. S3B). In measurements of single cells, CCh (1 mM) caused a transient increase in [Ca²⁺]_i in 94±3% of cells, and subsequent addition of isoproterenol (1 µM) caused further Ca²⁺ release from 95±1% of cells (Fig. 3A–C). As with HEK-PR1 cells, pre-treatment with βMCD reduced the number of cells responding to CCh (Fig. 3C), and in those cells that responded the amplitude of the increase in [Ca²⁺]_i was significantly reduced (Fig. 3D). However, the Ca²⁺ signals evoked by addition of isoproterenol after CCh were unaffected by loss of cholesterol: 92±3% of cells responded and the amplitude of the increase in [Ca²⁺]_i was similar with (209±38 nM) or without βMCD-treatment (227±9 nM) (Fig. 3C,D). Just as with PTH, therefore, many cells in which βMCD-treatment abolished responses to CCh alone responded normally to stimulation with CCh and isoproterenol. Furthermore, the residual Ca²⁺ content of the intracellular stores after stimulation with CCh and isoproterenol was significantly larger in cells that had been depleted of cholesterol (Fig. 3D).

Potentiation of CCh-evoked Ca²⁺ signals by PTH and β₂-adrenoceptors depends on their ability to deliver cAMP locally at high concentrations to associated IP₃R (Fig. 1A) (Tovey et al., 2008), but the effects are mimicked by a uniformly high concentration of 8-Br-cAMP (supplementary material Fig. S1B). As with PTH and isoproterenol, under conditions where βMCD almost abolished responses to CCh alone, 8-Br-cAMP still potentiated CCh-evoked increases in [Ca²⁺]_i (Fig. 3E). These results demonstrate that whether cAMP is delivered focally or globally, it effectively allows CCh to evoke Ca²⁺ release under conditions where CCh alone is ineffective.

We conclude that loss of cholesterol massively attenuates Ca²⁺ signals evoked by CCh without affecting the potentiated signals evoked by CCh in combination with 8-Br-cAMP or stimulation of AC by endogenous β₂-adrenoceptors or heterologously expressed PTH receptors (Fig. 3F). There is, therefore, a surprising independence of the Ca²⁺ release evoked by IP₃ in response to CCh alone or CCh with cAMP that does not require spatially organized cAMP signals.

Maximal activation of endogenous M₃R in HEK cells generates insufficient IP₃ to directly release all IP₃-sensitive Ca²⁺ stores (Tovey et al., 2008) (Figs 2, 3). This is consistent with the EC₅₀ for CCh-evoked Ca²⁺-release and the K_D of CCh binding to M₃R (Burford et al., 1995) being similar. The lack of ‘spare receptors’ is an experimental advantage in that it allows use of saturating concentrations of CCh to evoke submaximal Ca²⁺ signals similar to those evoked by physiological stimuli, but it requires sensitive methods to resolve CCh-stimulated IP₃ formation. We used a FRET-based IP₃-biosensor to allow real-time recording of IP₃ levels in CCh-stimulated HEK-PR1 cells (see Materials and Methods and supplementary material Fig. S4). Results shown in Fig. 4A show that CCh (20 µM) evoked a rapid and sustained increase in cytosolic IP₃ concentration that was unaffected by subsequent addition of PTH (100 nM), but increased further when the CCh concentration was increased (1 mM). This confirms that PTH does not affect the intracellular concentration of IP₃ (Short and Taylor, 2000). Treatment with βMCD had no effect on the IP₃ signals evoked by CCh alone or CCh with PTH (Fig. 4A,B).

We used photolysis of ciIP₃ (caged iIP₃) non-disruptively loaded into cells as ciIP₃/PM [cell-permeant caged iIP₃, D-2,3-O-isopropylidene-6-O-(2-nitro-4,5-dimethoxy)benzyl-myo-inositol 1,4,5-trisphosphate-hexakis(propionoxymethyl)ester] (Dakin and Li, 2007) to assess the effects of βMCD on the Ca²⁺ signals evoked by activation of IP₃R directly. Photolysis of ciIP₃ in HEK-PR1 cells caused transient increases in [Ca²⁺]_i (Fig. 4C). The amplitudes of these Ca²⁺ signals varied between cells (supplementary material Fig. S5), perhaps reflecting differences in loading and de-esterification of ciIP₃/PM. However, in a large sample (n>400 cells), the response to iIP₃ was unaffected by treatment with βMCD (Fig. 4C–E). Neither the fraction of cells responding to photolysis of ciIP₃ (75±7% and 77±7% for control and βMCD-treated cells, respectively) nor the average amplitude of the increase in fluo-4 fluorescence (ΔF/F₀, 3.52±0.40 and 3.22±0.26) was affected by βMCD.

Treatment with βMCD caused HEK-PR1 cells to become slightly more rounded (Fig. 1C; Fig. 4F), but there was no obvious redistribution of the ER and it remained associated with the plasma membrane (Fig. 4F). Immunostaining for each of the three IP₃R subtypes confirmed that their distributions, whether assessed by widefield microscopy or total internal reflection fluorescence microscopy (TIRFM), were also unaffected by cholesterol depletion (Fig. 4G and supplementary material Fig. S6).

We conclude that selective inhibition of CCh-evoked Ca²⁺ release by cholesterol depletion occurs under conditions where CCh-evoked increases in cytosolic IP₃ are unaffected (Fig. 4B), intracellular Ca²⁺ stores are intact (Fig. 2H), IP₃R respond normally to IP₃ (Fig. 4C–E) and there are no obvious changes in the distribution of the ER (Fig. 4F) or IP₃Rs (Fig. 4G; supplementary material Fig. S6). This suggests that for cells stimulated with CCh alone cholesterol depletion disrupts effective delivery of IP₃ to IP₃R, yet it has no effect on IP₃ delivery to IP₃R that respond in the presence of either cAMP or 8-Br-cAMP.

The ability of CCh with PTH to evoke Ca²⁺ release when responses to CCh alone are almost abolished by cholesterol depletion is unexpected because PTH increases [Ca²⁺]_i only when paired with a stimulus, like CCh, that stimulates IP₃ production (Fig. 1B) (Tovey et al., 2008). How might loss of cholesterol disrupt delivery of IP₃ to IP₃R when CCh alone is the stimulus without affecting IP₃ delivery when CCh is combined with activation of receptors that stimulate AC? The effect cannot simply result from sensitization of IP₃R by cAMP counteracting inhibition of M₃R coupling to PLC because there was no detectable inhibition of IP₃ production (Fig. 4A), and inhibition of PLC with U73122 similarly inhibited responses to CCh alone or CCh with PTH (supplementary material Fig. S7). In most of our experiments, where cells were first stimulated with CCh before addition of PTH, there was more Ca²⁺ in the intracellular stores at the time of PTH addition for cells treated with βMCD (Fig. 2H; Fig. 3D). We had therefore to consider whether the lack of effect of βMCD on responses to CCh with PTH might result from a fortuitous combination of a larger intracellular Ca²⁺ pool together with PTH, via cAMP, increasing the sensitivity of IP₃R (Tovey et al., 2008). That explanation is unlikely because responses to every concentration of PTH were insensitive to βMCD (Fig. 2D). We nevertheless assessed the proposal directly using a protocol that allowed the CCh-sensitive Ca²⁺ stores to be depleted before treatment with βMCD and then stimulation with CCh and PTH.

Cells were first stimulated with CCh in nominally Ca²⁺-free HBS to deplete the stores that respond to CCh alone. They were then rapidly depleted of cholesterol by incubation with βMCD (2% w/v, 10 minutes, 37°C, supplementary material Fig. S2) and then the effects of PTH with CCh on [Ca²⁺]_i were assessed. The results demonstrate that prior stimulation with CCh or βMCD massively attenuated the Ca²⁺ signals evoked by subsequent addition of CCh, and combining the treatments almost abolished the response to CCh (Fig. 5A, black). However, the increase in [Ca²⁺]_i evoked by CCh with PTH was entirely unaffected by any of the treatments. Prior depletion of the CCh-sensitive Ca²⁺ stores, cholesterol depletion, or both together had no effect on the ability of PTH to potentiate the increase in [Ca²⁺]_i evoked by CCh (Fig. 5A, blue). These results demonstrate that the Ca²⁺ stores released by CCh alone and those released by CCh combined with PTH are functionally independent (Fig. 5B). We conclude that M₃R communicate with different IP₃R in different Ca²⁺ stores when cells are activated by CCh alone or by CCh with stimuli that increase cAMP. Cholesterol depletion distinguishes these two pathways, and although cAMP is delivered focally to IP₃R when its production is stimulated by PTH (Tovey et al., 2008) globally applied 8-Br-cAMP has the same effect (Fig. 3E and supplementary material Fig. S1B). The two pathways from M₃R to IP₃R are not, therefore, dependent on local delivery of cAMP, they must differ in how they deliver IP₃ to IP₃R.

Functionally distinct IP₃-sensitive Ca²⁺ stores respond independently to CCh alone or to CCh delivered with stimuli, like PTH, that activate AC (Fig. 5B). Cholesterol depletion almost abolishes responses to CCh (Fig. 1) without affecting Ca²⁺ signals evoked by CCh with PTH (Fig. 2), isoproterenol or 8-Br-cAMP (Fig. 3). Responses to PTH with CCh are unchanged despite there being more Ca²⁺ within the intracellular stores after cholesterol depletion (Fig. 2), and they are unaffected by depletion of the Ca²⁺ stores that respond to CCh alone (Fig. 5A). These results establish that even when M₃R are maximally activated, a population of M₃R stimulates Ca²⁺ release via IP₃R from a store that is unaffected by cAMP or 8-Br-cAMP. Because cAMP increases the sensitivity of all IP₃R subtypes to IP₃ (Tovey et al., 2010), we conclude that the IP₃ concentration to which these CCh-regulated IP₃Rs are locally exposed must be sufficient to cause their maximal activation. Evidence that βMCD massively attenuates CCh-evoked Ca²⁺ signals (Figs 1–f02,3, Fig. 5A; supplementary material Figs S2, S3 and S7) without affecting IP₃ formation (Fig. 4) is consistent with such a spatially organized delivery of IP₃ to IP₃R that is disrupted by cholesterol depletion (Fig. 5B, left). Responses to CCh alone are attenuated by loss of IP₃R1, but not of IP₃R2 (Tovey et al., 2008), suggesting that these locally activated IP₃R are likely to be IP₃R1 (or IP₃R3). Depleting cells, probably the plasma membrane (Fig. 1C), of cholesterol, disrupts communication between these M₃R and IP₃R (Fig. 1). Lipid rafts are enriched in M₃R, Gαq and PLCβ, they associate with IP₃R, and in many cells disrupting lipid rafts inhibits M₃R-evoked Ca²⁺ signals (Gosens et al., 2007; Lockwich et al., 2000). We propose that M₃R within lipid rafts locally deliver saturating concentrations of IP₃ to closely associated IP₃R (Fig. 5B, left). The concentration-dependent effects of CCh must then come largely from all-or-nothing recruitment of these IP₃R junctions rather than from graded recruitment of IP₃R within a junction. The graded CCh signal is transduced into a digital recruitment of IP₃R junctions. Cholesterol depletion disrupts lipid rafts and thereby the interactions between M₃R and IP₃R that facilitate focal delivery of IP₃, so that diffusion and degradation of IP₃ rapidly reduce its concentration to below the threshold for IP₃R activation, thereby abolishing the CCh-evoked Ca²⁺ signals (Fig. 5B, left).

A second population of M₃R is unaffected by cholesterol depletion and less intimately associated with IP₃R. Activation of these M₃R provides insufficient IP₃ to activate IP₃R directly, but these IP₃R respond when they are co-stimulated with cAMP (Fig. 5B, right). Previous work showed that loss of IP₃R2 attenuated Ca²⁺ signals evoked by CCh and PTH without affecting responses to CCh alone (Tovey et al., 2008). This suggests that IP₃R2 populate the Ca²⁺ store from which CCh evokes Ca²⁺ release only in the presence of cAMP (Fig. 5B, right).

We showed earlier that activation of M₃R and P2Y receptors in single HEK cells evokes Ca²⁺ release from functionally distinct IP₃-sensitive stores (Short et al., 2000), suggesting that IP₃ can be locally delivered to IP₃R. HEK cells are well-suited to addressing this feature because even maximally activated M₃R generate insufficient IP₃ to release Ca²⁺ from all IP₃-sensitive stores. Signalling from M₃R to IP₃R therefore remains focal even when M₃R are fully activated. In most cells with larger numbers of receptors, maximal activation is likely to generate high concentrations of IP₃ that flood the cytosol and obscure spatial organization. HEK cells provide the experimental convenience of working with all M₃R activated while retaining physiologically relevant spatial organization. This experimental opportunity allowed us to demonstrate that M₃R in lipid rafts deliver IP₃ at high local concentration to associated IP₃R (Fig. 5B, left). Such digital communication is robust and analogous to the digital signalling we proposed for cAMP delivery from AC6 to IP₃R2 (Fig. 1A) (Tovey et al., 2008). M₃R outside lipid rafts are less intimately associated with IP₃R and their activation fails to deliver enough IP₃ to activate IP₃R directly. Cyclic AMP, by sensitizing IP₃R (Tovey et al., 2008; Tovey et al., 2010), extends the range over which these M₃R can signal, allowing M₃R that are otherwise ineffective to recruit additional IP₃-sensitive Ca²⁺ stores. We conclude that M₃R can signal directly to IP₃R via digital junctions or more loosely via IP₃ diffusing over longer distances. The latter is ineffective unless cAMP tunes IP₃R sensitivity, effectively extending the range of action of IP₃ produced in response to these otherwise ineffective M₃R.

Cell culture media, Lipofectamine 2000, Alexa-488 conjugated secondary antibodies, fluo-4/AM, fura-2/AM and Ca²⁺ standard solutions for calibration of fura-2 fluorescence signals to [Ca²⁺]_i were from Life Technologies (Paisley, UK). Ionomycin and 8-bromo-cAMP (8-Br-cAMP) were from Merck Bioscience (Nottingham, UK). 1,2-bis(o-aminophenoxy)ethane- N,N,N′,N′-tetraacetic acid (BAPTA) was from Molekula (Gillingham, UK). U73122 was from Tocris (Bristol, UK). Foetal bovine serum, carbamylcholine chloride (CCh), poly-l-lysine, β-methylcyclodextrin (βMCD), filipin and isoproterenol were from Sigma–Aldrich (Poole, UK). G418 was from ForMedium (Hunstanton, UK). Cholesterol (plant-derived) was from Avanti Polar Lipids Inc. (Alabaster, USA). A peptide comprising residues 1–34 of human PTH (hereafter described as PTH) was from Bachem (St. Helens, UK). Cell-permeant caged iIP₃ (ciIP₃/PM) was from Sichem (Bremen, Germany). Inositol 1,4,5-trisphosphate (IP₃) was from Alexis Biochemicals (Nottingham, UK) and ³H-IP₃ (681Gbq/mmol) was from Perkin Elmer (Cambridge, UK). Cell culture plastics and 96-well assay plates were from Greiner (Stonehouse, UK). Imaging dishes (35-mm diameter with a 7-mm glass insert) were from MatTek Corporation (Ashland, USA) or PAA laboratories (Yeovil, UK). VECTASHIELD mounting reagent was from Vector Laboratories (Peterborough, UK). A plasmid encoding a Ca²⁺-indicator protein targeted to the ER lumen, D1ER, was a gift from R. Tsien (San Diego, USA) (Palmer et al., 2004). It was used to identify ER rather than to measure free [Ca²⁺]. An affinity-purified polyclonal antibody against the C-terminal of IP₃R1 has been described (Cardy et al., 1997). D. Yule (University of Rochester, USA) provided affinity-purified antibodies to the N- (residues 320–338) and C-terminals (residues 2686–2702) of IP₃R2 (Betzenhauser et al., 2009; Betzenhauser et al., 2008). A monoclonal antibody against IP₃R3 was from BD Transduction Laboratories (Oxford, UK). A polyclonal anti-GFP antibody was from Abcam (Cambridge, UK).

HEK 293 and HEK 293 cells stably transfected with the human type 1 PTH receptor (HEK-PR1 cells) (Short and Taylor, 2000) were cultured at 37°C in Dulbecco’s modified Eagle’s medium/Ham’s F12 with GlutaMAX™-1, supplemented with foetal bovine serum (10%) and 800 µg/ml G418 (HEK-PR1 cells only) in a humidified atmosphere containing 95% air and 5% CO₂. Medium was replaced every third day, and cells were passaged when they reached about 80% confluence. For experiments with cell populations, cells were seeded into 96-well plates (∼2×10⁴ cells/well). For single-cell analyses, cells were seeded (1.2×10⁵ cells/well) onto either 35-mm imaging dishes with a 7-mm glass insert or 22-mm round glass coverslips, each pre-coated with 0.01% poly-l-lysine. Cells were grown for a further 2–3 days before experiments or transfection. The latter used Lipofectamine 2000 according to the manufacturer’s instructions with 1 µg DNA/well for cells in 35-mm imaging dishes.

Almost confluent cultures of HEK 293 or HEK-PR1 cells in 96-well plates were washed, loaded with fluo-4/AM (2 µM), and [Ca²⁺]_i was measured in populations of cells using a FlexStation fluorescence plate-reader (MDS Analytical Devices, Wokingham, UK) (Tovey et al., 2008). All experiments were performed at 20°C in HBS, nominally Ca²⁺-free HBS or Ca²⁺-free HBS. HBS had the following composition: 135 mM NaCl, 5.9 mM KCl, 1.2 mM MgCl₂, 1.5 mM CaCl₂, 11.6 mM HEPES, 11.5 mM glucose, pH 7.3; CaCl₂ was omitted from nominally Ca²⁺-free HBS; Ca²⁺-free HBS was supplemented with 10 mM BAPTA. Single-cell analyses of [Ca²⁺]_i in HEK 293/HEK-PR1 cells loaded with fura-2/AM (2 µM) were performed as previously reported (Tovey et al., 2003). Fluorescence ratios were calibrated, after correction for background fluorescence, to [Ca²⁺]_i by reference to Ca²⁺-calibration solutions.

To allow photolysis of ciIP₃, HEK-PR1 cells were first incubated at 20°C for 45 minutes in HBS containing ciIP₃/PM (1 µM) (Dakin and Li, 2007; Smith and Parker, 2009) and then for 45 minutes with HBS containing fluo-4/AM (2 µM) and ciIP₃/PM (1 µM). After at least 60 minutes (during which any treatments used to deplete cholesterol were applied), cells were used for single-cell imaging. Imaging dishes were mounted on the stage of an Olympus IX81 inverted fluorescence microscope (60× TIRF objective, numerical aperture 1.45). Excitation light was provided by a 488-nm diode-pumped solid-state laser (Olympus Digital Laser Systems, Milton Keynes, UK) and emitted fluorescence (500–550 nm) was captured with an eMCCD camera (Andor ixon 897, Andor, Belfast, UK). Two high-intensity flashes (∼1 millisecond, 3000 µF, 300 V, ∼170 J) from a JML-C2 xenon flash-lamp (Rapp OptoElectronic, Hamburg, Germany) equipped with a 395-nm barrier filter were used to photolyse ciIP₃. Photolysis of ciIP₃ releases a form of IP₃ (iIP₃) in which the 2- and 3-hydroxyl groups are protected by an isopropylidene group. This does not prevent it from activating IP₃R, but iIP₃ is more resistant than IP₃ to degradation by endogenous enzymes (Dakin and Li, 2007). For these measurements of [Ca²⁺]_i using a non-ratiometric indicator (fluo-4), responses are reported as F/F₀, where F₀ is the average fluorescence intensity recorded from a defined region of interest (ROI) in the 5 seconds immediately before flash photolysis, and F is the fluorescence intensity from the same region after the flash.

Total internal reflection fluorescence microscopy (TIRFM) using an Olympus IX81 inverted microscope with a triple-line TIRF combiner and a 150× TIRF objective (numerical aperture 1.45) was used to visualize the ER in HEK-PR1 cells expressing D1ER (Palmer et al., 2004). Fluorescence of EYFP was excited using a 488-nm diode-pumped solid-state laser, with emitted fluorescence (500–550 nm) captured using an eMCCD camera (Andor ixon 897). Widefield images of D1ER were obtained by excitation at 504 nm, with emission collected at 530–570 nm.

Endogenous IP₃R were immunostained as described (Tovey et al., 2001), with all steps performed at 20°C in phosphate-buffered saline (PBS) unless stated otherwise. Cells on 35-mm imaging dishes were washed, fixed with paraformaldehyde (4% w/v in PBS) for 30 minutes, washed again and permeabilized with Triton X-100 (0.2% v/v in PBS, 10 minutes). Cells were then incubated in blocking medium (3% w/v BSA in PBS) for 1 hour, then with primary antibody (1∶100 in blocking medium) overnight at 4°C, washed (3×10 minutes), incubated with Alexa-Fluor-488 conjugated secondary antibody for 1 hour (1∶1000 in blocking medium) and washed (3×10 minutes). Immunofluorescence was imaged as described for D1ER. PBS had the following composition: 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄, pH 7.4.

Intracellular IP₃ was measured using a FRET-based IP₃-sensor derived from the IP₃-binding core (IBC) of IP₃R1 tagged at its N- and C-termini with ECFP and EYFP, respectively, each attached by a short linker (supplementary material Fig. S4A). The sequence encoding residues 224–604 of IP₃R1 was PCR-amplified from a full-length rat IP₃R1(S1⁺) sequence (GenBank: GQ233032.1) using primers 1 and 2. The sequences of all primers are provided in supplementary material Table S1. The fragment was cloned as a BamHI/XhoI fragment into the pENTR1 vector (Gateway) to generate the construct pENTR1-IBC. The open reading frame (ORF) of ECFP was PCR-amplified from the pECFP-ER vector (Clontech) using primers 3 and 4 and cloned as a BamHI/MluI fragment into pENTR1-IBC to generate the construct pENTR1-ECFP-IBC. The ORF of EYFP was PCR-amplified from the pC1-EYFP vector (Clontech) using primers 3 and 5, and cloned as an EcoRI/XhoI fragment into pENTR1-ECFP-IBC to generate the construct pENTR1-ECFP-IBC-EYFP. The latter was inserted into the expression vector pcDNA3.2 (Gateway) to generate the IP₃-sensor expression plasmid. Properties of the IP₃-sensor are shown in supplementary material Fig. S4.

Cells on 35-mm imaging dishes were transfected with the plasmid encoding the IP₃-sensor (1 µg/dish) using Lipofectamine 2000, and used 2 days later. An Olympus IX81 inverted microscope equipped with a 60× TIRF objective (numerical aperture 1.45) and a 440/520 nm dual band-pass dichroic mirror was used to record fluorescence using widefield excitation at 427 nm and simultaneous collection of CFP (455–485 nm) and YFP (520–550 nm) emissions using an Olympus U-SIP split imaging TV port fitted with a 505-nm dichroic mirror (supplementary material Fig. S4E). Split images were obtained at 1-second intervals using an eMCDD camera (Andor ixon 897). CFP and YFP emissions were background corrected and a normalized CFP/YFP ratio was calculated for each cell. IP₃ binding causes an increase in CFP/YFP ratio, indicative of decreased FRET (supplementary material Fig. S4D,E).

HEK cells transfected with the IP₃-sensor in 6-well plates were washed, scraped into PBS containing protease inhibitors (1 tablet/10 ml, Roche Diagnostics, Mannheim, Germany) and centrifuged (650 g, 2 minutes). The pellet was lysed by two freeze–thaw cycles in liquid nitrogen, sonication (1 minute) and passage through a syringe needle. After centrifugation (12,000 g, 5 minutes), the supernatant was used for western blotting or ³H-IP₃ binding. For blotting, proteins (50 µg) were loaded onto NuPAGE 4–12% Bis Tris gels (Life Technologies) and blotted with an anti-GFP antibody (1∶1000). Equilibrium-competition binding assays were performed as described (Rossi et al., 2009). Briefly, incubations (500 µl) at 4°C were performed in 50 mM Tris, 1 mM EDTA, pH 8.3 with ³H-IP₃ (0.75 nM) and cell supernatant (150 µg protein). Reactions were terminated after 5 minutes by addition of poly(ethylene glycol) 8000 (500 µl, 30% w/v) and γ-globulin (30 µl, 25 mg/ml) and centrifugation (20,000 g, 5 minutes). Radioactivity was determined by liquid scintillation counting. Non-specific binding was determined by addition of 1 µM IP₃.

After loading with a Ca²⁺ indicator (and, where appropriate, ciIP₃), cells were depleted of intracellular cholesterol by incubation with βMCD (2% w/v, ∼15 mM) for either 10 minutes at 37°C or for up to 2 hours at 20°C (Rodal et al., 1999; Sampson et al., 2004). After washing, cells were used for experiments. Cholesterol was restored as a βMCD∶cholesterol complex (10∶1, 0.26% w/v (∼2 mM) βMCD:200 µM cholesterol) added to cells for 1 hour at 37°C. Briefly, a 100 mM solution of cholesterol was prepared in methanol∶chloroform (2∶1) and complexes of βMCD∶cholesterol were formed by drop-wise addition of cholesterol to a continuously stirred (∼2 hours) solution of 0.26% w/v βMCD in HBS maintained at 80°C.

Free cholesterol was measured using filipin, a fluorescent antibiotic that binds to the free 3β-hydroxyl of cholesterol (McCabe and Berthiaume, 2001). Cells were fixed with paraformaldehyde at 20°C (4% w/v, 30 minutes), washed three times with PBS, and then with PBS containing glycine (1.5 mg/ml, 10 minutes) to terminate fixation. Cells were stained with filipin (50 µg/ml) in PBS containing foetal bovine serum (10%) for 2 hours at 20°C, washed with PBS (3×10 minutes), and mounted (VECTASHIELD). Filipin staining was visualized using an Olympus IX81 inverted fluorescence microscope with excitation at 380 nm and emission at 460–550 nm. Identical microscope and eMCCD camera settings were used to capture each image. The method used to quantify filipin staining is described in the legend to Fig. 1C.

Concentration–effect relationships for each experiment were individually fitted to Hill equations using non-linear curve-fitting (GraphPad Prism, version 5) and the results obtained from each (pEC₅₀, Hill coefficient h, maximal response) were pooled for statistical analysis and presentation. Student’s t-test was used for statistical analyses.

We thank Dr Ana Rossi for production and initial characterization of the IP₃-sensor, and Roger Tsien and David Yule for gifts of the D1ER expression plasmid and antibodies to IP₃R2, respectively.