Stimulation of Inositol 1,4,5-Trisphosphate (IP₃) Receptor Subtypes by Analogues of IP₃

Abstract

Most animal cells express mixtures of the three subtypes of inositol 1,4,5-trisphosphate receptor (IP₃R) encoded by vertebrate genomes. Activation of each subtype by different agonists has not hitherto been examined in cells expressing defined homogenous populations of IP₃R. Here we measure Ca²⁺ release evoked by synthetic analogues of IP₃ using a Ca²⁺ indicator within the lumen of the endoplasmic reticulum of permeabilized DT40 cells stably expressing single subtypes of mammalian IP₃R. Phosphorylation of (1,4,5)IP₃ to (1,3,4,5)IP₄ reduced potency by ∼100-fold. Relative to (1,4,5)IP₃, the potencies of IP₃ analogues modified at the 1-position (malachite green (1,4,5)IP₃), 2-position (2-deoxy(1,4,5)IP₃) or 3-position (3-deoxy(1,4,5)IP₃, (1,3,4,5)IP₄) were similar for each IP₃R subtype. The potency of an analogue, (1,4,6)IP₃, in which the orientations of the 2- and 3-hydroxyl groups were inverted, was also reduced similarly for all three IP₃R subtypes. Most analogues of IP₃ interact similarly with the three IP₃R subtypes, but the decrease in potency accompanying removal of the 1-phosphate from (1,4,5)IP₃ was least for IP₃R3. Addition of a large chromophore (malachite green) to the 1-phosphate of (1,4,5)IP₃ only modestly reduced potency suggesting that similar analogues could be used to measure (1,4,5)IP₃ binding optically. These data provide the first structure-activity analyses of key IP₃ analogues using homogenous populations of each mammalian IP₃R subtype. They demonstrate broadly similar structure-activity relationships for all mammalian IP₃R subtypes and establish the potential utility of (1,4,5)IP₃ analogues with chromophores attached to the 1-position.

Introduction

Most animal cells express inositol 1,4,5-trisphosphate receptors (IP₃R), which fulfil an essential role in linking the many cell-surface receptors that stimulate IP₃ formation to release of Ca²⁺ from the endoplasmic reticulum [1]. Vertebrates have genes for three IP₃R subunits, while invertebrates have only a single IP₃R gene. All functional IP₃R are tetrameric assemblies of these subunits. The similar primary sequences of the IP₃R subunits suggest that all IP₃R are likely to share similar structures, although we presently have only a limited understanding of the structure of the entire IP₃R [1], [2]. Each subunit has an N-terminal region to which IP₃ binds. This region comprises the N-terminal suppressor domain (SD, residues 1–223) and the IP₃-binding core (IBC, residues 224–604 in IP₃R1, Figure 1A), which is alone sufficient to bind IP₃ with appropriate selectivity [3]. The SD both modulates the affinity of the IBC for agonists and provides an essential link between IP₃ binding and opening of the pore [4], [5], [6], [7]. A large cytoplasmic region separates the N-terminal from the six transmembrane domains. The last pair of these, together with the intervening luminal loop, form the Ca²⁺-permeable pore [8] (Figure 1A). Each subunit terminates in a short C-terminal tail, which has also been implicated in the regulation of gating [9]. The diversity provided by three genes is further increased by multiple splice variants of at least two of the three IP₃R subtypes (IP₃R1 and IP₃R2), by formation of homo- or hetero-tetrameric assemblies of IP₃R subunits, by association with an enormous diversity of modulatory proteins and by post-translational modifications [10]. At present, we have only a limited understanding of the functional significance of this complexity for IP₃-evoked Ca²⁺ signals in native tissues.

Figure 1. Structure of the N-terminal of the IP₃ receptor and structures of the ligands used.

(A) Key regions of a single IP₃R subunit (numbering for rat IP₃R1) are shown highlighting N-terminal domains and the six C-terminal transmembrane domains (TMD) that form the pore. A high-resolution structure of the N-terminal (NT, residues 1–604) with (1,4,5)IP₃ bound is shown (Protein Data Bank, 3UJO). The NT comprises the suppressor domain (SD) and IP₃-binding core (IBC). The essential 4- and 5-phosphate groups of (1,4,5)IP₃ interact with residues in the β-domain and α-domain of the IBC, respectively. (B) Structures of the ligands used.

The broadly similar structures of the three IP₃R subunits are matched by many shared functional properties, most notably co-regulation of all IP₃R by IP₃ and Ca²⁺ [10], [11]. Nevertheless, there are differences in the patterns of expression of IP₃R in different tissues [12], [13], in their subcellular distributions [14], [15], sensitivities to IP₃ [16], modulation by accessory proteins and additional signals [17], [18], [19], and in the functional consequences of IP₃R ablation [20], [21]. Heterogeneous populations of IP₃R in most cells make it difficult to establish clearly the characteristics of each IP₃R subtype and to define their functional roles. A better knowledge of the ligand recognition properties of the three IP₃R3 subtypes is needed if ligands selective for IP₃R subtypes are to be developed to help resolve these problems. All known high-affinity agonists of IP₃R retain structures equivalent to the 4,5-bisphosphate and 6-hydroxyl groups of (1,4,5)IP₃ (Figure 1B) [22]. The only exception is a low-affinity analogue of adenophostin A (3″-dephospho-adenophostin A) in which interactions between the adenine moiety and IP₃R appear partially to compensate for loss of a phosphate (equivalent to the 5-phosphate of (1,4,5)IP₃) within the critical bisphosphate moiety [23]. Here we use a selection of synthetic analogues of IP₃ that preserve the key structures of the high-affinity agonists to assess their activity at each IP₃R subtype.

Materials and Methods

Materials

Thapsigargin was from Alomone Laboratories (Jerusalem, Israel). The structures of the ligands used and their abbreviations are shown in Figure 1B. (1,4,5)IP₃ was from Alexis Biochemicals (Nottingham, U.K.). 3-deoxy(1,4,5)IP₃, (1,3,4)IP₃ and (1,3,4,5)IP₄ were from Calbiochem (Nottingham, U.K.). (1,4,6)IP₃ from both Alexis Biochemicals and synthesized as reported previously [24] was used. Malachite green IP₃ (MG(1,4,5)IP₃) was synthesized using the methods described by Inoue et al. [25]. (4,5)IP₂ [26], 2-deoxy(1,4,5)IP₃ [27] and synthetic (1,3,4,5)IP₄ [28] were synthesized as previously described. All synthesized ligands were purified by ion-exchange chromatography, fully characterized by the usual spectroscopic methods and accurately quantified by total phosphate assay. ³H-IP₃ (185 Bq/mmol) was from PerkinElmer (Bucks, U.K.).

Anti-peptide antisera to peptides conserved in all mammalian IP₃R subtypes (AbC, residues 62–75 of rat IP₃R1) or unique to IP₃R1 (Ab1, residues 2724–2739) or IP₃R2 (Ab2, residues 2685–2701) were described previously [29]. A monoclonal antibody that recognizes N-terminal residues (22–230) of IP₃R3 (Ab3) was from BD Transduction Laboratories (Oxford, U.K.). Anti-β-actin antibody was from AbCam (Cambridge, U.K.). Donkey secondary antibodies (anti-rabbit or anti-mouse) conjugated to horseradish peroxidase were from Santa Cruz Biotechnology (Santa Cruz, CA, U.S.A.). Sources of other materials are provided in earlier publications [30], [31].

Cell Culture

DT40 cells lacking genes for all three IP₃R subtypes (DT40-KO cells) [32] and the same cells stably expressing rat IP₃R1 (GenBank accession number GQ233032.1) [33], mouse IP₃R2 (GU980658.1) [31] or rat IP₃R3 (GQ233031.1) [34] were cultured in RPMI 1640 medium supplemented with 10% foetal bovine serum, 1% heat-inactivated chicken serum, 2 mM glutamine and 50 µM 2-mercaptoethanol at 37°C in 95% air and 5% CO₂. Cells were passaged every 2 days when they reached a density of ∼1.5×10⁶ cells/mL.

Immunoblotting

Cells (∼7×10⁷) were centrifuged (650 g, 5 min), resuspended in Hepes-buffered saline (HBS), re-centrifuged, and the pellet was solubilized in 1 mL of medium containing 140 mM NaCl, 5 mM NaF, 10 mM Tris, 1 mM Na₄P₂O₇, 0.4 mM Na₃VO₄, 1% Triton X-100 and a protease inhibitor tablet (1 tablet/10 mL, Roche Diagnostics, Mannheim, Germany). 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. The solubilized cells were sonicated on ice (Trans Scientific 1420 sonicator, 50–60 Hz, 3×10 s), incubated with gentle rotation for 1 h at 2°C and then centrifuged (6000 g, 10 min). A sample of the supernatant (13 µL) was mixed with dl-dithiothreitol (2 µL, 100 mM) and NuPAGE LDS sample buffer (5 µL, Invitrogen, Paisley, U.K.). After heating (70°C, 10 min), 20-µL samples were loaded onto NuPAGE 3–8% Tris acetate gels for SDS-PAGE using Novex Tris acetate buffer (Invitrogen). Broad range spectrum marker and MagicMark molecular weight markers (Invitrogen) were used to monitor protein migration during SDS-PAGE and for calibration, respectively. After transfer to a PVDF membrane using an iBlot dry-blotting system (Invitrogen), the membrane was blocked by incubation (1 h) with Tris-buffered saline (TBS) containing 5% non-fat dry milk. It was then incubated with primary antiserum (1∶1000 and 1∶10,000 for IP₃R and β-actin antibodies, respectively) in TBS containing 2.5% w/v BSA for 12 h. TBS had the following composition: 140 mM NaCl, 20 mM Tris, 0.1% v/v Tween, pH 7.6. After incubation with primary antibodies, the blots were washed in TBS (3×5 min), incubated with horseradish peroxidase-conjugated secondary antibodies (1∶1000, donkey anti-rabbit or donkey anti-mouse) for 1 h in TBS supplemented with 2.5% BSA, and washed again (3×5 min). Bands were detected using Supersignal West Pico chemiluminescent substrate (ThermoScientific, Rockford, IL, U.S.A.) and quantified using GeneTools software (Syngene, Frederick, MD, U.S.A.).

Measurement of IP₃-evoked Ca²⁺ Release

A low-affinity Ca²⁺ indicator (Mag fluo-4) trapped within the ER lumen was used to measure IP₃-evoked Ca²⁺ release from permeabilized DT40 cells stably expressing mammalian IP₃R [35]. Cells (50 mL, 10⁶ cells/mL) were loaded with Mag fluo-4 AM (20 µM) in HBS supplemented with pluronic F127 (0.02% w/v) for 1 h at 20°C in the dark with gentle shaking. The cells were centrifuged (650 g, 5 min), resuspended in Ca²⁺-free cytosol like medium (CLM) and permeabilized by addition of saponin (10 µg/mL, ∼4 min, 37°C). Ca²⁺-free CLM had the following composition: 140 mM KCl, 2 mM NaCl, 1 mM EGTA, 2 mM MgCl₂, 20 mM Pipes, pH 7. After washing (650 g, 2 min), permeabilized cells were resuspended in CLM without Mg²⁺, but supplemented with CaCl₂ (375 µM) to give a final free [Ca²⁺] of ∼220 nM (after addition of 1.5 mM MgATP) and with carbonyl cyanide 4-trifluoromethoxy-phenyl hydrazone (FCCP, 10 µM) to inhibit mitochondrial Ca²⁺ uptake. Cells were distributed (5×10⁵ cells/well) into half area 96-well, black-walled, poly-lysine-coated plates and centrifuged (300 g, 2 min). Fluorescence (excitation and emission wavelengths of 485 nm and 520 nm, respectively) was recorded at 1-s intervals using a FlexStation™ 3 fluorescence plate reader (Molecular Devices, Berkshire, U.K.) at 20°C. Addition of MgATP (1.5 mM) allowed Ca²⁺ uptake into the intracellular stores, and after 150 s, IP₃ (or its analogues) was added together with thapsigargin (1 µM) to prevent further Ca²⁺ uptake. IP₃-evoked Ca²⁺ release is expressed as a fraction of that released by addition of ionomycin (1 µM) [35].

³H-IP₃ Binding to Native Type 1 IP₃ Receptors

Mouse cerebellar membranes (5 mg protein) in a final volume of 500 µL of CLM with a free [Ca²⁺] of 220 nM were incubated with ³H-IP₃ (1.5 nM) and appropriate concentrations of competing ligand at 4°C [4]. After 5 min, during which equilibrium was attained, the reactions were terminated by centrifugation (2000 g). The supernatant was removed and the pellet was washed (500 µL CLM) and then solubilized (200 µL CLM with 1% v/v Triton-X-100) before determining its radioactivity using Ecoscint A scintillation cocktail (National Diagnostics, Atlanta, GA, USA, 1 mL). Total ³H-IP₃ binding was ∼3000 disintegrations/min (d.p.m.) and non-specific binding was ∼300 d.p.m. (determined by addition of 1 µM IP₃ or by extrapolation of IP₃ competition curves to infinite IP₃ concentration). Results were fitted to Hill equations (GraphPad Prism, version 5, GraphPad Software Inc., CA, U.S.A.) from which half-maximal inhibitory concentration (IC₅₀) and thereby K_D values were calculated.

Statistical Analysis

Concentration-effect relationships were fitted to Hill equations using Graph Pad Prism, from which Hill coefficients (h), the fraction of the intracellular Ca²⁺ stores released by maximally effective concentrations of agonist, and pEC₅₀ values were calculated. For clarity, some results are reported as EC₅₀ values, but all statistical comparisons used pEC₅₀ values. Each experiment included an analysis of the effects of (1,4,5)IP₃ to allow paired comparisons of pEC₅₀ values for IP₃ and each analogue (ΔpEC₅₀). Results are expressed as means ± SEM for n independent experiments, with each experiment performed in triplicate. Statistical comparisons used paired Student’s t test or ANOVA followed by Bonferroni test, with P<0.05 is considered significant.

Homology Modelling and Ligand Docking

Sequence alignments were performed using MUSCLE (http://www.ebi.ac.uk/Tools/msa/muscle/). Structural homology models were built in Modeller 9.10 [36] using templates of crystal structures of the IBC (Protein Data Bank: 1N4K) and SD (Protein Data Bank: 1XZZ). The geometric qualities of the models were evaluated with the Molprobity server [37]. Finally, the N-terminal regions of mouse IP₃R2 and rat IP₃R3 comprising the modelled SD and IBC structures of each were reconstructed by aligning the individual domains against cognate regions of the NT of rat IP₃R1 (Protein Data Bank: 3UJ4) [7] using UCSF Chimera 1.6.1 [38].

We used docking of IP₃ analogues into a rigid IBC and subsequent superposition of the structure onto a model of the entire IP₃R solely to assess whether full-length IP₃R was likely to bind the analogues without steric clashes. Docking of IP₃ analogues was performed using the IBC (Protein Data Bank: 1N4K) [3] and AutoDock Vina 1.1.2 [39]. Prior to docking, bound (1,4,5)IP₃ and all water molecules (except those within 5 Å of the IP₃-binding site) were removed. The search space comprised a grid of 20×20×20 points, each separated by 0.375 Å, around the IP₃-binding site. Ligands (except (1,4,5)IP₃) were drawn and energy-minimized with MM2 force field using ChemBioOffice 2008 (http://www.cambridgesoft.com). Polar hydrogens and the Gasteiger partial atomic charges were then added to the protein and ligands using AutoDockTools (http://autodock.scripps.edu/resources/adt), and the prepared structures were used as input files for docking. Only the best pose is considered for each ligand. For MG(1,4,5)IP₃, the ligand was superposed onto (1,4,5)IP₃ within the NT monomer (Protein Data Bank: 3UJ0A) and the complex (with (1,4,5)IP₃-removed) was energy-minimized using MMFF94 Forcefield [40]. PyMol was used to present docked structures (http://pymol.sourceforge.net/).

Results

Expression of Mammalian IP₃ Receptor Subtypes in DT40 Cells

Immunoblotting with IP₃R subtype-selective antisera (Ab1-3) established that each of the stable DT40 cell lines specifically expressed only a single IP₃R subtype (Figure 2A). An antiserum that recognizes a peptide sequence present in all mammalian IP₃R subtypes (AbC) [29] was used to quantify relative levels of IP₃R expression in the three cell lines (Figure 2A). The results demonstrate that, relative to IP₃R3 (100%), levels of expression of IP₃R1 and IP₃R2 were 71±8% and 48±5%, respectively (Figure 2A). Because the density of IP₃R may affect the sensitivity of intracellular Ca²⁺ stores to IP₃ [6] and it is impracticable to generate cell lines expressing identical levels of each IP₃R subtype, comparisons of the relative potencies of IP₃ analogues for each IP₃R subtype are expressed relative to the potency of (1,4,5)IP₃ in the same cell line (see Methods).

Figure 2. Functional expression of recombinant IP₃ receptor subtypes in DT40 cells.

(A) Western blots from DT40 cells stably expressing each of the IP₃R subtypes. Each lane was loaded with lysate derived from ∼1.4×10⁶ cells stably expressing IP₃R1, IP₃R2 or IP₃R3, or from DT40-KO cells (KO). The antisera used are selective for IP₃R1-3 (Ab1-3) or they interact equally with all three IP₃R subtypes (AbC) [29] (see Materials). An antiserum to β-actin was used to confirm the equivalent loading of lanes. Quantitative analysis of AbC immunoreactivity from 3 similar gels was used to establish expression levels of the three IP₃R subtypes relative to IP₃R3 (means ± SEM, n = 3) (lower panel). (B) A luminal Ca²⁺ indicator was used to record Ca²⁺ uptake into the intracellular stores of permeabilized DT40 cells expressing IP₃R1 after addition of MgATP. Subsequent addition of (1,4,5)IP₃ (at the concentrations shown) with thapsigargin (1 µM) allowed (1,4,5)IP₃-evoked Ca²⁺ release to be quantified. The typical experiment shows traces averaged from 3 wells in a single plate. For clarity only a selection of the (1,4,5)IP₃ concentrations are shown. The same methods were used to quantify the effects of all analogues. (C) Concentration-dependent effects of (1,4,5)IP₃ on Ca²⁺ release from the intracellular stores of cells expressing IP₃R1 (black), IP₃R2 (blue) or IP₃R3 (red). The same colour-codes are used in all subsequent figures. Results are means ± SEM from the number of independent experiments given in Table 1. Here, and in many subsequent figures, some error bars are smaller than the symbols.

IP₃-evoked Ca²⁺ Release by Subtypes of IP₃ Receptor

A low-affinity luminal Ca²⁺ indicator was used to report the free Ca²⁺ concentration within the endoplasmic reticulum of permeabilized DT40 cells (Figure 2B). IP₃ failed to evoke Ca²⁺ release in DT40-KO cells, which lack IP₃R (not shown) [34], but it was effective in DT40 cells stably expressing each of the three IP₃R subtypes (Figure 2B and C) [31]. The results demonstrate that in each cell line, IP₃ caused a concentration-dependent release of 60–75% of the intracellular Ca²⁺ stores, with Hill coefficients (h) of about 1, and pEC₅₀ values of 7.06±0.05, 6.84±0.06 and 6.38±0.05 for IP₃R1, IP₃R2 and IP₃R3, respectively (Table 1 and Figure 2C). Notwithstanding the moderate differences in IP₃R expression in the different DT40 cell lines (Figure 2A), the relative sensitivities of IP₃R1-3 to IP₃ in these assays (IP₃R1∼IP₃R2>IP₃R3) are consistent with a general consensus that IP₃R3 is the least sensitive of the IP₃R subtypes [16].

Table 1. Effects of IP₃ analogues on Ca²⁺ release by subtypes of IP₃ receptor.

	IP3R1					IP3R2					IP3R3
	EC50 (nM)	pEC50	h	Ca2+release (%)	n	EC50 (nM)	pEC50	h	Ca2+release (%)	n	EC50 (nM)	pEC50	h	Ca2+release (%)	n
(1,4,5)IP3	87	7.06±0.05	0.99±0.05	75±1	31	145	6.84±0.06	1.26±0.09	61±2	34	417	6.38±0.05	1.26±0.07	64±2	30
(4,5)IP2	8128	5.09±0.12	1.11±0.15	69±4	4	6310	5.20±0.20	1.67±0.27	56±4	5	11482	4.94±0.07	1.28±0.03*	83±2	3
(1,3,4,5)IP4	11749	4.93±0.05	1.09±0.16	66±6	3	5495	5.26±0.09	2.26±0.18	61±5	3	114815b	3.94±0.03	1.08±0.22	36±1a	3
2-deoxy(1,4,5)IP3	123	6.91±0.15	0.84±0.07	82±3	4	115	6.94±0.11	1.71±0.37	65±2	6	324	6.49±0.02	1.36±0.10	74±3	3
(1,4,6)IP3	4365	5.36±0.19	0.98±0.10	81±4	6	12589	4.90±0.16	1.06±0.17	68±2	6	15849	4.80±0.12	2.58±0.61*	46±1	6
3-deoxy(1,4,5)IP3	3311	5.48±0.09	1.75±0.24*	61±2	6	5888	5.23±0.07	1.78±0.34	57±6	6	15849	4.80±0.13	2.59±0.68	53±7	6
MG(1,4,5)IP3	447	6.35±0.12	0.95±0.10	73±4	9	708	6.15±0.09	1.04±0.08	57±3	9	1738	5.76±0.11	1.18±0.11	64±4	8
(1,3,4)IP3	ND	ND	ND	39±8a	3	ND	ND	ND	18±6a	3	ND	ND	ND	2±1a	3

The EC₅₀, pEC₅₀, Hill coefficient (h) and fraction of the intracellular Ca²⁺ stores released by a maximally effective concentration of each analogue are shown for each IP₃R subtype. All results (except EC₅₀) are shown as means ± SEM from n independent experiments. The (1,3,4,5)IP₄ was provided by Calbiochem. ND, not determined.

^aCa²⁺ release evoked with 100 µM of the analogue (where the highest attainable concentrations of ligand failed to saturate the response).

^bEC₅₀ estimated by assuming that maximally effective concentrations of (1,3,4,5)IP₄ and (1,4,5)IP₃ release the same fraction of the Ca²⁺ stores.

^*Denotes Hill slopes that are significantly different (P<0.05) from 1. Statistical comparisons of maximal Ca²⁺ release and pEC₅₀ values used paired comparisons (Δmax or ΔpEC₅₀), which are presented in Table 2.

Interactions of IP₃ Metabolites with IP₃ Receptor Subtypes

Inositol 1,3,4,5-tetrakisphosphate ((1,3,4,5)IP₄) is the immediate product of (1,4,5)IP₃ phosphorylation by IP₃ 3-kinase [41] (Figure 3A). Although (1,3,4,5)IP₄ stimulated Ca²⁺ release via each of the three IP₃R subtypes (Figure 3B), it was ∼100-fold less potent than IP₃. Indeed in DT40-IP₃R3 cells, which are the least sensitive to IP₃ (Figure 2C), even 100 µM (1,3,4,5)IP₄ failed to release the entire IP₃-sensitive Ca²⁺ store (Figures 3B and C, and Table 1). The purity of (1,3,4,5)IP₄ supplied by Calbiochem is only ∼98%. We were therefore concerned that its effects on Ca²⁺ release might be mediated by contaminating (1,4,5)IP₃. However, similar experiments using DT40-IP₃R1 cells and synthetic (1,3,4,5)IP₄, where the synthesis provides no opportunity for contamination by (1,4,5)IP₃ [28], established that the two sources of (1,3,4,5)IP₄ were equipotent. Relative to (1,4,5)IP₃, ΔpEC₅₀ values were 2.09±0.07 and 1.97±0.04 (n = 3) for commercial and synthetic (1,3,4,5)IP₄, respectively. Synthetic (1,3,4,5)IP₄ (100 µM) had no significant effect on the Ca²⁺ content of the intracellular stores of permeabilized DT40 cells lacking IP₃R (DT40-KO cells) (not shown). We conclude that (1,3,4,5)IP₄ itself stimulates Ca²⁺ release via all IP₃R, but only at concentrations ∼100-fold higher than with (1,4,5)IP₃ (Figures 3B and C). The much reduced potency of (1,3,4,5)IP₄ is consistent with loss of the 3-hydroxyl group reducing potency (Table 1 and see below) and with docking analyses, which suggest that although the 3-phosphate can be accommodated, the orientations of the phosphate groups are slightly distorted relative to (1,4,5)IP₃ bound to the IBC (Figure 3F).

Figure 3. (1,3,4,5)IP₄ is a low-affinity full agonist of IP₃ receptors.

(A) (1,4,5)IP₃ can be dephosphorylated to (1,4)IP₂ or phosphorylated to (1,3,4,5)IP₄, which can then be dephosphorylated to (1,3,4)IP₃. (B) Effects of (1,3,4,5)IP₄ (Calbiochem) on Ca²⁺ release via each of the three IP₃R subtypes. (C) Paired comparisons of the effects of (1,4,5)IP₃ and (1,3,4,5)IP₄ are shown for each IP₃R subtype. (D) Typical results from DT40-IP₃R1 cells stimulated with (1,4,5)IP₃ alone (black trace) or in the presence of (1,3,4,5)IP₄ (5 µM, added 35 s before IP₃; red trace). (E) Summary results show the effect of (1,3,4,5)IP₄ on the concentration-dependent effects of (1,4,5)IP₃ on Ca²⁺ release. Results (B, C and E) are means ± SEM from the number of independent experiments given in Table 1. (F) (1,3,4,5)IP₄ docked into the IBC of IP₃R1 adopts a conformation in which the underlying (1,4,5)IP₃ scaffold is only slightly distorted from that of (1,4,5)IP₃ within the crystal structure of the IBC (right panel) [3]. Docking of the NT into a 3D reconstruction of the entire IP₃R1 [2] has suggested a likely arrangement of the NT within the tetrameric IP₃R [see 7] (left panel). Within this proposed arrangement, the IBC can still bind (1,3,4,5)IP₄ without obvious steric clashes.

It has been suggested that (1,3,4,5)IP₄ potentiates the Ca²⁺ release evoked by (2,4,5)IP₃ at IP₃R1 [42], although in other cells there was no evidence for this interaction [43]. We re-examined this phenomenon by first adding (1,3,4,5)IP₄ to permeabilized DT40-IP₃R1 cells at a concentration (5 µM) near its threshold for evoking Ca²⁺ release, and then assessed subsequent responses to (1,4,5)IP₃ (Figure 3D). The results, which show that (1,3,4,5)IP₄ has no effect on the response to any concentration of (1,4,5)IP₃ (Figure 3E), provide no support for the suggestion that (1,3,4,5)IP₄ potentiates IP₃-evoked Ca²⁺ release [42]. Furthermore, the lack of inhibition of IP₃-evoked Ca²⁺ release by (1,3,4,5)IP₄ demonstrates that it is not a partial agonist of IP₃R. We conclude that (1,3,4,5)IP₄ is a low-affinity full agonist of all three IP₃R subtypes.

Dephosphorylation of (1,3,4,5)IP₄ by IP₃ 5-phosphatase produces (1,3,4)IP₃ (Figure 3A), which can accumulate to concentrations considerably exceeding that of (1,4,5)IP₃ during sustained stimulation [44]. (1,3,4)IP₃, even at a concentration of 100 µM, failed to stimulate release of the entire (1,4,5)IP₃-sensitive Ca²⁺ store in any of the three cell lines (Table 1). (1,3,4)IP₃ is, therefore, at least 1000-fold less potent than (1,4,5)IP₃. Comparing the fraction of the Ca²⁺ stores released by 100 µM (1,3,4)IP₃ in cells expressing each of the three IP₃R subtypes indicates that the response was greatest in DT40-IP₃R1 cells and smallest in DT40-IP₃R3 cells (Table 1). The effects of (1,3,4)IP₃ on the three IP₃R subtypes (IP₃R1> IP₃R2>> IP₃R3) therefore match the rank order of potency of (1,4,5)IP₃ (Table 1), consistent perhaps with minor contamination (<0.1%) of the (1,3,4)IP₃ with (1,4,5)IP₃ or (1,3,6)IP₃ accounting for the activity. These results suggest that (1,3,4)IP₃ is itself unlikely to bind to IP₃R. This is consistent with a requirement for a vicinal 4,5-bisphosphate structure in all known inositol phosphate ligands of IP₃R [45]. It is impossible for us to verify this independently by equilibrium competition ³H-IP₃ binding because DT40 cells are the only homogenous source of IP₃R subtypes presently available to us, but the low density of IP₃Rs would require impracticably large numbers of cells for binding studies.

Type 3 IP₃ Receptors are Slightly Less Sensitive to Loss of the 1-phosphate from (1,4,5)IP₃

Removal of the 1-phosphate from (1,4,5)IP₃ to give (4,5)IP₂ caused a similar ∼83-fold decrease in potency for IP₃R1 and IP₃R2 (Figures 4A and B, Tables 1 and 2). Contamination of (4,5)IP₂ with (1,4,5)IP₃ cannot explain this activity because (4,5)IP₂ was prepared by total synthesis and purified by ion-exchange chromatography [26]. This is consistent with previous functional analyses of native IP₃R [45] and with the ability of (4,5)IP₂ to compete with ³H-(1,4,5)IP₃ for binding to these IP₃R subtypes heterologously expressed in Sf9 cells [46].

Figure 4. IP₃ receptor subtypes differ slightly in their requirements for a 1-phosphate group in (1,4,5)IP₃.

(A) Effects of (4,5)IP₂ on Ca²⁺ release via each IP₃R subtype. (B) Paired comparisons of the effects of (1,4,5)IP₃ and (4,5)IP₂ are shown for each IP₃R subtype. (C) The 1-phosphate group of (1,4,5)IP₃ interacts directly with R568 and via water (blue sphere) with K569 within the α-domain of the IBC [3]. The primary sequences of this region (residues 561-580) are similar for each IP₃R subtype. (D) Structural homology models of the NT of mouse IP₃R2 and rat IP₃R3 were built using structures of the SD (Protein Data Bank: 1XZZ) and IBC (Protein Data Bank: 1N4K) from IP₃R1 and these were then aligned using the published structure of the NT of rat IP₃R1 (Protein Data Bank: 3UJ0).

Table 2. Relative potencies of IP₃ analogues at different IP₃ receptor subtypes.

	IP3R1			IP3R2			IP3R3
	ΔpEC50	Δmax (%)	(1,4,5)IP3max (%)	ΔpEC50	Δmax (%)	(1,4,5)IP3max (%)	ΔpEC50	Δmax (%)	(1,4,5)IP3max (%)
(4,5)IP2	1.92±0.15	13±5	81±3	1.92±0.05	8±3	65±4	1.38±0.06*	−1±2	77±4
2-deoxy(1,4,5)IP3	0.09±0.10	−1±2	81±3	0.19±0.14	−3±3	63±4	−0.17±0.08	0±2	77±4
(1,4,6)IP3	1.57±0.09	−2±5	76±5	1.67±0.15	−3±6	64±5	1.45±0.09	8±6	55±7
3-deoxy(1,4,5)IP3	1.61±0.12	21±2*	80±4	1.64±0.21	12±4*	68±3	1.66±0.17	20±5*	66±3
(1,3,4,5)IP4 a	2.09±0.07	17±5a	77±1	2.18±0.04	7±0*	67±6	2.38±0.07	NDa	77±4
MG(1,4,5)IP3	0.69±0.10	4±5	77±3	0.7±0.14	11±3*	64±3	0.64±0.15	4±4	61±3

From paired comparisons with (1,4,5)IP₃, the relative potency and differences in maximal Ca²⁺ release are shown for each analogue at each IP₃R subtype. Results show means ± SEM, with n provided in Table 1. To allow direct comparison with responses evoked by (1,4,5)IP₃, maximal Ca²⁺ release evoked by (1,4,5)IP₃ in experiments paired with the analogues are also shown ((1,4,5)IP₃ max).

^aWith (1,3,4,5)IP₄ it was impossible to attain concentrations (>100 µM) that evoked a maximal response for all IP₃R subtypes; in these cases the maximal responses were not subject to statistical analysis. The relative potency of (1,3,4,5)IP₄ at IP₃R subtypes was estimated by assuming it was a full agonist that released the same fraction of the intracellular Ca²⁺ stores as a maximally effective concentration of (1,4,5)IP₃ in parallel analyses. ND, not determined.

^*Denotes values significantly different (P<0.05) from IP₃R1 (for ΔpEC₅₀ values) or from (1,4,5)IP₃ in paired comparisons with the analogue (for Δmax values).

However, the difference in potency of (1,4,5)IP₃ and (4,5)IP₂ at IP₃R3 was only ∼24-fold (ΔpEC₅₀ = 1.38±0.06) (Figure 4B, Table 2). Our results suggest that although removal of the 1-phosphate from (1,4,5)IP₃ reduces potency at all IP₃R subtypes, IP₃R3 is least affected. Within the IBC of IP₃R1, the 1-phosphate of (1,4,5)IP₃ interacts directly with R568 and, via water, with K569 [3] (Figure 4C). These residues and their immediate neighbours are conserved in IP₃R2 and IP₃R3 (Figure 4C). The only residues known to interact directly with (1,4,5)IP₃ are within the IBC (Figure 1A) and primary sequences of the IBC are highly conserved between IP₃R subtypes. It is therefore unsurprising that homology models based on the IBC of IP₃R1 [3] suggest almost indistinguishable structures for the IBCs from IP₃R2 and IP₃R3 (Figure 4D). Furthermore, the IBCs from the three IP₃R subtypes have the same affinity for (1,4,5)IP₃ [16]. However, both full-length IP₃R and the NT from different IP₃R subtypes differ in their affinities for (1,4,5)IP₃. These observations demonstrate that interactions between the IBC and other parts of the IP₃R, notably the suppressor domain (SD, residues 1–223), influence ligand binding [4], [16]. The complexity of these interactions between the IBC and other domains together with the need for only very modest conformational differences between subtypes to account for the small differences in ligand potency make it difficult to define the residues responsible for modulating the interactions between the IBC and the 1-phosphate group of (1,4,5)IP₃ in IP₃R3.

Removal of the 2- and 3-hydroxyl Groups from (1,4,5)IP₃ or Inverting their Orientations Similarly Affect Interactions with all IP₃ Receptor Subtypes

The structure of (1,4,5)IP₃ bound to the IBC of IP₃R1 shows the 2-hydroxyl group of (1,4,5)IP₃ exposed to solvent (Figure 1A) [3] and previous structure-activity studies of native IP₃R suggested that removal of the 2-hydroxyl moiety from (1,4,5)IP₃ to give 2-deoxy(1,4,5)IP₃ minimally affects activity [4], [46]. Our results confirm that observation for all three IP₃R subtypes: 2-deoxy(1,4,5)IP₃ and (1,4,5)IP₃ are equipotent at all three IP₃R subtypes (Figures 5A and B, Tables 1 and 2).

Figure 5. The 2- and 3-hydroxy groups of (1,4,5)IP₃ interact similarly with each IP₃ receptor subtype.

(A) Effects of 2-deoxy-(1,4,5)IP₃ on Ca²⁺ release via each IP₃R subtype. (B) Paired comparisons of the effects of (1,4,5)IP₃ and 2-deoxy-(1,4,5)IP₃ are shown for each IP₃R subtype. (C and D) Similar analyses of 3-deoxy-(1,4,5)IP₃. Results (A-D) are means ± SEM from the number of independent experiments given in Table 1.

Removal of the 3-hydroxyl of (1,4,5)IP₃ (3-deoxy(1,4,5)IP₃) caused a ∼40-fold decrease in potency (ΔpEC₅₀ ∼1.6) that was similar for all three IP₃R subtypes (Figures 5C and D, Tables 1 and 2). This decrease is larger than that observed for either Ca²⁺ release from cells expressing largely IP₃R1 (∼3-fold) [47] or for binding to heterologously expressed IP₃R1-3 (7- 20-fold) [46].

(1,4,6)IP₃ is an analogue of (1,4,5)IP₃ in which the orientations of the 2- and 3-hydroxyl groups are inverted (Figures 1B and 6A). (1,4,6)IP₃ was less potent than (1,4,5)IP₃ and the loss of potency was similar for all three IP₃R subtypes (Figures 6B and C, Tables 1 and 2). The loss of potency probably results from inverting the 3-hydroxyl group from its equatorial position in (1,4,5)IP₃ to an axial position in (1,4,6)IP₃ (Figure 6A) because l-scyllo(1,2,4)IP₃, which differs from (1,4,5)IP₃ only in its inverted orientation of the 2-hydroxyl group, was reported to have the same affinity as (1,4,5)IP₃ for all three IP₃R subtypes [46].

Figure 6. Interactions of (1,4,6)IP₃ with IP₃ receptor subtypes.

(A) Structures of (1,4,5)IP₃ and (1,4,6)IP₃ drawn to show how they differ only in the relative orientations of the 2- and 3-hydroxyl groups of (1,4,5)IP₃. (B) Effects of (1,4,6)IP₃ on Ca²⁺ release via each IP₃R subtype. (C) Paired comparisons of the effects of (1,4,5)IP₃ and (1,4,6)IP₃ are shown for each IP₃R subtype. Results (B and C) are means ± SEM from the number of independent experiments given in Table 1.

MG-(1,4,5)IP₃ is a Full Agonist of IP₃ Receptors with Slightly Lesser Affinity than (1,4,5)IP₃

MG(1,4,5)IP₃ in which 4-carboxy-malachite green is attached via an aminopropyl linkage to the 1-phosphate of (1,4,5)IP₃ (Figure 1B) was originally synthesized to explore its potential as a ligand of IP₃R that might allow chromophore-assisted laser inactivation (CALI) of IP₃R [25], [48]. The first study, using surface plasmon resonance to assess binding to an N-terminal fragment of IP₃R1 (residues 1–885), surprisingly suggested that MG(1,4,5)IP₃ had ∼170-fold greater affinity than (1,4,5)IP₃, whereas fluorescein similarly attached to the 1-position of (1,4,5)IP₃ had no significant effect on affinity [25]. The results were important because they suggested that MG(1,4,5)IP₃ might be the ligand with the highest known affinity for IP₃R, and they were unexpected because disrupting interaction of the 1-phosphate group of (1,4,5)IP₃ with the IP₃R would be expected to reduce affinity (Figure 4 and Table 1) [45]. A subsequent study of Ca²⁺ release from permeabilized smooth muscle cells concluded that the EC₅₀ for MG(1,4,5)IP₃ was ∼7-fold higher than that for (1,4,5)IP₃ [25]. The disparity between the reported very high affinity of MG(1,4,5)IP₃ for the N-terminal of IP₃R1 and its modest potency in functional assays of smooth muscle has not been explained. We considered two possibilities. MG(1,4,5)IP₃ may be a high-affinity partial agonist or it may differ massively in its affinity for IP₃R1 and the endogenous IP₃R of smooth muscle.

Our results indicate that MG(1,4,5)IP₃ is ∼5-fold less potent than (1,4,5)IP₃ at each IP₃R subtype (Figures 7A and B, Tables 1 and 2). This is consistent with functional assays of smooth muscle [25] and with our results from native IP₃R (largely IP₃R2) in rat hepatocytes, where ΔpEC₅₀ was 0.57±0.03 for (1,4,5)IP₃ (pEC₅₀ = 6.81±0.02) and MG(1,4,5)IP₃ (pEC₅₀ = 6.24±0.01) (Taylor CW, unpublished data). These results establish that MG(1,4,5)IP₃ interacts similarly with all three IP₃R subtypes and that it is less potent than (1,4,5)IP₃. In aqueous solution, the triphenylmethane component of MG(1,4,5)IP₃ exists as a mixture of several inter-converting species, whose relative proportions are sensitive to pH [49]. At pH 7.4, a colourless triphenylmethanol form with a tetrahedral structure is likely to co-exist with the coloured propeller-shaped form shown in Figures 1B and 7F. Because the earlier surface plasma resonance experiments [25], [48] and our analyses were performed at similar pH, we assume that the proportions of the two forms were similar in each analysis.

Figure 7. MG(1,4,5)-IP₃ is a full agonist of all IP₃ receptor subtypes.

(A) Effects of MG(1,4,5)IP₃ on Ca²⁺ release via each IP₃R subtype. (B) Paired comparisons of the effects of (1,4,5)IP₃ and MG(1,4,5)IP₃ are shown for each IP₃R subtype. Results (A and B) are means ± S.E.M. from the number of independent experiments given in Table 1. (C) Equilibrium-competition binding of ³H-IP₃ (1.5 nM) to cerebellar membranes in CLM containing 220 nM free [Ca²⁺] in the presence of the indicated concentrations of (1,4,5)IP₃ or MG(1,4,5)IP₃. (D) Typical results showing effects of MG(1,4,5)IP₃ (30 nM, added 35 s before (1,4,5)IP₃) on Ca²⁺ release from DT40-IP₃R cells. (E) Summary results (means ± SEM from 3 independent experiments) show concentration-dependent effects of (1,4,5)IP₃ alone or in the presence of MG(1,4,5)IP₃ (300 nM). (F) MG(1,4,5)IP₃ docked into the IBC of IP₃R1 (top; (1,4,5)IP₃ shown in gray) shows that it can be accommodated within the likely structure of the tetrameric IP₃R without steric clashes (bottom).

In equilibrium-competition binding assays in CLM using cerebellar membranes, which predominantly express IP₃R1 [14], (1,4,5)IP₃ bound with an affinity that was ∼7-fold greater than that of MG(1,4,5)IP₃ (ΔpK_D = 0.86±0.08, n = 3): pK_D = 8.29±0.03 and 7.43±0.08, for (1,4,5)IP₃ and MG(1,4,5)IP₃, respectively (Figure 7C and Table 3). The ∼7-fold lesser affinity of MG(1,4,5)IP₃ for IP₃R1 relative to (1,4,5)IP₃ and its similarly reduced potency (∼5-fold, Table 2) are consistent with the large malachite green structure perturbing interaction of the 1-phosphate of (1,4,5)IP₃ with the IBC, and they suggest that MG(1,4,5)IP₃ is a full agonist of IP₃R. The latter conclusion is further substantiated by the results shown in Figures 7D and E. These show that the Ca²⁺ release evoked by (1,4,5)IP₃ is unaffected by the presence of 300 nM MG(1,4,5)IP₃, which itself released 34±4% of the Ca²⁺ stores. The pEC₅₀ was 7.41±0.09 and 7.01±0.37 (n = 3) for (1,4,5)IP₃ alone and in the presence of MG(1,4,5)IP₃, respectively (Figure 7E). A weak partial agonist, by contrast, would be expected, at concentrations sufficient to evoke a response, to occupy enough IP₃R to shift the concentration-dependence of the response to (1,4,5)IP₃ to higher concentrations [4]. These results establish that MG(1,4,5)IP₃ is not a partial agonist of IP₃R1. We conclude that MG(1,4,5)IP₃ is a full agonist of IP₃R with an affinity that is ∼7-fold less than that of (1,4,5)IP₃.

Table 3. Interactions of MG(1,4,5)IP₃ with type 1 IP₃ receptors.

	DT40-IP3R1 cells			Cerebellum
	EC50 (nM)	pEC50 (/M)	Maximal Ca2+ release (%)	KD (nM)	pKD (/M)	pKD-pEC50
(1,4,5)IP3	91	7.04±0.1	77±2	5.13	8.29±0.03	1.25±0.29
MG(1,4,5)IP3	447	6.35±0.12	73±4	37.2	7.43±0.08	1.08±0.35

Results from functional assays of Ca²⁺ release from DT40-IP₃R1 cells (Figure 7B) and of equilibrium competition binding to cerebellar membranes (Figure 7C) compare the pEC₅₀ and pK_D values for (1,4,5)IP₃ and MG(1,4,5)IP₃. Results (except EC₅₀ and K_D) show means ± SEM from 3–9 independent experiments. The final column shows pK_D-pEC₅₀ for each ligand.

Discussion

The primary sequence of the IBC, which is responsible for recognition of (1,4,5)IP₃ by IP₃R, is well conserved between IP₃R subtypes, and in isolation the IBC from each IP₃R subtype binds IP₃ with similar affinity [16]. Although there are high-resolution structures of this region for only IP₃R1 [3], [7], [50], [51], homology modelling suggests that the IBC structures of the three IP₃R subtypes are very similar (Figure 4D). It is, therefore, unsurprising that structure-activity relationships for the three IP₃R subtypes, most derived from comparisons of cells expressing mixtures of IP₃R subtypes, are similar [45], [52], [53]. Nevertheless, residues outside the IBC, notably the SD (Figure 1A), reduce the affinity of the IBC to differing extents for different IP₃R subtypes [16]. Such differences and the minor differences in the primary sequence of the IBC between IP₃R subtypes leave open the possibility that it may be possible to develop subtype-selective ligands of IP₃R. Hitherto, the only systematic comparison of the ligand recognition properties of homogenous populations of IP₃R subtypes examined only ligand binding to mammalian IP₃R expressed in insect Sf9 cells [46]. We have now extended the analysis to functional assays of mammalian IP₃R expressed in a null background, namely DT40 cells in which the genes for all endogenous IP₃R have been disrupted [32] (Figure 2).

All high-affinity inositol phosphate agonists of IP₃R have structures equivalent to the 4,5-bisphosphate and 6-hydroxyl groups of (1,4,5)IP₃ [22], [45], [46], and with the exception of (1,3,4)IP₃ (which was effectively inactive), all ligands examined herein retain these groups (Figure 1B). Most of the ligands examined were either inactive ((1,3,4)IP₃) or their potency relative to (1,4,5)IP₃ suggested a lack of selectivity for IP₃R subtypes (MG(1,4,5)IP₃, (1,4,6)IP₃, (1,3,4,5)IP₄, 2-deoxy(1,4,5)IP₃ and 3-deoxy(1,4,5)IP₃) (Figures 3, 5, 6 and 7, Tables 1 and 2). The potencies of each of these agonists for each IP₃R subtype are consistent with known structure-activity relationships [45], [46] and with the structure of (1,4,5)IP₃ bound to the IBC [3] (Figure 1A). Loss of the 2-hydroxyl minimally affects activity, loss of either the 3-hydroxyl or 1-phosphate or inverting the orientation of the 3-hydroxyl causes the potency to decrease substantially, and addition of malachite green to the 1-phosphate of (1,4,5)IP₃ causes a modest (∼7-fold) decrease in potency (Tables 1 and 2). We conclude that the structure-activity relationships for all three mammalian IP₃R are broadly similar.

MG(1,4,5)IP₃ was reported to bind with unexpectedly high affinity to an N-terminal fragment of IP₃R1 [25], but our results, in line with subsequent analyses of smooth muscle cells from the same group [25], suggest that MG(1,4,5)IP₃ interacts similarly with all IP₃R subtypes and that it is a full agonist with an affinity that is ∼7-fold lower than that of (1,4,5)IP₃ (Figure 7 and Tables 2 and 3). This is consistent with evidence suggesting that even substantial additions to the 1-phosphate of (1,4,5)IP₃ (eg, fluorescein and 3-aminopropyl) are well-tolerated [25], [48], [54]. It is also consistent with our docking studies, which suggest that even these very large additions to the 1-phosphate moiety of (1,4,5)IP₃ do not create steric clashes within the tetrameric IP₃R (Figure 7F).

The necessity of the 4,5-bisphosphate and 6-hydroxyl moieties [55], and the considerable loss of affinity associated with modifying the 3-hydroxyl group of (1,4,5)IP₃ (Figures 3C, D and 5C, D) restrict opportunities to tag (1,4,5)IP₃ to modifications of the 1- and 2-positions. Our previous work has shown that modification of the 2-position is compatible with high-affinity binding to IP₃R, but it reduces efficacy [4], [56]. Our demonstration that MG(1,4,5)IP₃ is a high-affinity full agonist of IP₃R (Figure 7) therefore identifies an opportunity to develop fluorescent, or otherwise modified, ligands of IP₃R that might be expected to come close to mimicking (1,4,5)IP₃ in their interactions with IP₃R.

Only one synthetic ligand discriminated modestly between IP₃R subtypes. (4,5)IP₂ lacks the 1-phosphate group of the endogenous ligand, (1,4,5)IP₃ (Figure 1B). In keeping with considerable published evidence [45], [46], this causes a substantial decrease in potency at all IP₃R subtypes (Figure 4, Tables 1 and 2). The loss of potency is, however, significantly less pronounced (∼24-fold) for IP₃R3 than for the other IP₃R subtypes (∼80-fold) (Table 2). The difference is unlikely to be due to residues with the IBC itself because the IBCs of all three IP₃R subtypes bind (1,4,5)IP₃ with indistinguishable affinity [16] and the residues that interact with the 1-phosphate group are conserved between IP₃R (Figure 4C). We instead suggest that subtype-selective interactions of the IBC with other domains, perhaps the SD, subtly modify interaction of the 1-phosphate group of (1,4,5)IP₃ with two basic residues within the IBC (Figure 4C).

Using homogenous populations of mammalian IP₃R and a variety of (1,4,5)IP₃ analogues, we have shown that the three IP₃R subtypes have very similar ligand recognition properties. However, the decrease in affinity caused by loss of the 1-phosphate group is less for IP₃R3. Finally, we have shown that MG(1,4,5)IP₃ is a full agonist of IP₃R with only modestly reduced affinity, suggesting that attachment of fluorescent tags to the 1-phosphate of (1,4,5)IP₃ [57] is a feasible strategy for producing modified analogues of (1,4,5)IP₃ that closely mimic the native messenger.

Funding Statement

This work was supported by grants from the Wellcome Trust to CWT [085295], and BVLP and AMR [082837], and from Biotechnology and Biological Sciences Research Council to CWT [BB/H009736/1]. HS is supported by a research studentship from the Jameel Family Trust. TR was a research fellow of Pembroke College, Cambridge. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.