Subtype-selective regulation of IP(3) receptors by thimerosal via cysteine residues within the IP(3)-binding core and suppressor domain.

IP(3)R (IP(3) [inositol 1,4,5-trisphosphate] receptors) and ryanodine receptors are the most widely expressed intracellular Ca(2+) channels and both are regulated by thiol reagents. In DT40 cells stably expressing single subtypes of mammalian IP(3)R, low concentrations of thimerosal (also known as thiomersal), which oxidizes thiols to form a thiomercurylethyl complex, increased the sensitivity of IP(3)-evoked Ca(2+) release via IP(3)R1 and IP(3)R2, but inhibited IP(3)R3. Activation of IP(3)R is initiated by IP(3) binding to the IBC (IP(3)-binding core; residues 224-604) and proceeds via re-arrangement of an interface between the IBC and SD (suppressor domain; residues 1-223). Thimerosal (100 μM) stimulated IP(3) binding to the isolated NT (N-terminal; residues 1-604) of IP(3)R1 and IP(3)R2, but not to that of IP(3)R3. Binding of a competitive antagonist (heparin) or partial agonist (dimeric-IP(3)) to NT1 was unaffected by thiomersal, suggesting that the effect of thimerosal is specifically related to IP(3)R activation. IP(3) binding to NT1 in which all cysteine residues were replaced by alanine was insensitive to thimerosal, so too were NT1 in which cysteine residues were replaced in either the SD or IBC. This demonstrates that thimerosal interacts directly with cysteine in both the SD and IBC. Chimaeric proteins in which the SD of the IP(3)R was replaced by the structurally related A domain of a ryanodine receptor were functional, but thimerosal inhibited both IP(3) binding to the chimaeric NT and IP(3)-evoked Ca(2+) release from the chimaeric IP(3)R. This is the first systematic analysis of the effects of a thiol reagent on each IP(3)R subtype. We conclude that thimerosal selectively sensitizes IP(3)R1 and IP(3)R2 to IP(3) by modifying cysteine residues within both the SD and IBC and thereby stabilizing an active conformation of the receptor.

INTRODUCTION

IP3Rs [IP3 (inositol 1,4,5-trisphosphate) receptors] and RyRs (ryanodine receptors) are related families of intracellular Ca2+ channels (see Supplementary Figure S1 at http://www.biochemj.org/bj/451/bj4510177add.htm). IP3Rs are expressed in most animal cells [1] and RyRs are expressed in many cells, but most abundantly in excitable tissues [2]. Most IP3Rs and RyRs are expressed within the membranes of the endoplasmic or sarcoplasmic reticulum, where they mediate the release of Ca2+ stored within them. Vertebrate genomes have three genes for IP3R subunits and most have three genes for RyR. Functional intracellular Ca2+ channels are tetrameric assemblies of these very large subunits, probably only homotetramers for RyR [2] (but see [3]), whereas IP3R form both homo- and hetero-tetrameric channels [1]. Activation of IP3R is initiated by binding of IP3 to the IBC (IP3-binding core; residues 224–604) of each IP3R subunit [4] and then proceeds via re-arrangement of an interface between the IBC and SD (suppressor domain; residues 1–223) [5,6] (see Supplementary Figure S1). The links between these conformational changes within the NT (N-terminal residues 1–604) domains of the IP3R and gating of the channel formed by transmembrane helices close to the C-terminal are poorly understood, but they are known to require the SD [7].

The two families of intracellular Ca2+ channels, IP3Rs and RyRs, share many characteristics including related structures [6], similar pores [8] and many aspects of their regulation. The latter includes regulation by cytosolic Ca2+ [9,10] and modulation by thiol-reactive reagents [11,12]. Indeed, modification of cysteinyl residues within ion channels is widespread [13-17], and contributes both to pathological responses to oxidative stress [18] and to physiological signals, including nitric oxide [19], hydrogen peroxide [20] and oxygen [21], which generate reactive oxygen or nitrogen species (see references in [22]).

Thimerosal (also known as thiomersal) was once widely used as an antiseptic and as a preservative, particularly in childhood vaccines, but after considerable controversy it is now rarely used [23]. However, as a membrane-permeable thiol-oxidizing agent, thimerosal remains a useful experimental tool that interacts with free thiol groups to form a thiomercuryethyl complex. Many studies have reported thimerosal-evoked increases in cytosolic Ca2+ concentration and suggested that they, at least partly, result from its effects, whether direct or via oxidation of GSH, on IP3Rs or RyRs (reviewed in [24]).

Many thiol reagents, including thimerosal [25], biphasically regulate RyR activity: low concentrations stimulate activity, whereas higher concentrations are inhibitory [12,22]. This pattern is reminiscent of the biphasic regulation of RyR by cytosolic Ca2+ [10], and consistent with evidence that oxidation of critical cysteine residues increases the sensitivity of RyRs to endogenous regulators [26] (reviewed in [12]). There are approximately 100 cysteine residues in each RyR subunit and within the reducing environment of the cytosol approximately half of these are in the reduced state [22]. A few of these residues are particularly susceptible to oxidation, and their susceptibility depends on whether the RyR is open or closed. But there is not yet any clearly defined relationship between oxidation of specific cysteine residues and channel activity (reviewed in [12]).

An abundant amount of evidence demonstrates that IP3R activity is also modulated by thiol-oxidizing agents [27] (reviewed in [11]). In many cells, for example, the effects of thimerosal on IP3-evoked Ca2+ release are rather similar to its biphasic effects on RyRs: potentiation of responses at low concentrations and inhibition of Ca2+ release at higher concentrations of thimerosal [11]. Although the responses to thimerosal in intact cells are undoubtedly made more complicated by its effects on additional Ca2+-handing proteins and perturbation of GSH/GSSG [28], there are clearly direct biphasic effects on IP3R behaviour [29] (see references in [11]). There are 60 cysteine residues in each IP3R1 subunit and most are likely to be in the reduced form in their native setting [11]. None of the 12 cysteine residues within the S1 splice variant of the IP3R1 NT are required for effective gating of IP3Rs by IP3 [6], but two conserved cysteine residues within the C-terminal tail (Cys2610 and Cys2613) are essential [7], and three reduced cysteine residues within the third luminal loop of IP3R1 (Cys2496, Cys2504 and Cys2527) are essential for its redox-sensitive interaction with a luminal thioredoxin-related protein, ERp44 [30]. The mechanisms whereby thimerosal or other thiol reagents modulate IP3R behaviour have not been resolved, but they are relevant to understanding redox regulation of IP3Rs under physiological and pathological conditions, and to defining further the mechanisms of IP3R activation. The mechanism by which thimerosal regulates the three subtypes of IP3Rs is addressed in the present study.

MATERIALS AND METHODS

Materials

Thimerosal and heparin (sodium salt from bovine intestinal mucosa) were from Sigma–Aldrich. IP3 was from Enzo Life Sciences (Exeter). [3H]IP3 (18 Ci/mmol) was from PerkinElmer Life and Analytical Sciences. Dimeric IP3 (structure shown in Figure 3A) was synthesized and characterized as reported previously [5]. Sources of other materials are specified in previous publications [31,32] or within the relevant methods sections.

Figure 3

Thimerosal does not affect binding of an antagonist or partial agonist

(A and B) Concentration-dependent effects of the IP3 dimer (A) or heparin (B) on specific [3H]IP3 binding (0.25 nM) to the NT from IP3R1 alone or after incubation with thimerosal (100 μM, 10 min). Results are means±S.E.M. for four independent experiments. The structure of the IP3 dimer is shown within (A). ○, control; ●, thimerosal.

Expression of NT fragments of IP3 receptors

NT fragments of rat IP3R1 (NT1 and IBC) (Supplementary Table S1 at http://www.biochemj.org/bj/451/bj4510177add.htm) were amplified by PCR from the full-length coding sequence of IP3R1 lacking the S1 splice site (GenBank accession number: GQ233032.1) and ligated into pGEX-6P-2 vectors (GE Healthcare). The open reading frame encoding NT2 (residues 1–604 from mouse IP3R2) (GenBank: GU980658.1) and NT3 (residues 1–604 from rat IP3R3) (GenBank: GQ233031.1) were amplified by PCR using the primers listed in Supplementary Table S2 (at http://www.biochemj.org/bj/451/bj4510177add.htm). PCR products were digested with SmaI and XhoI and ligated into pGEX-6P-2 vectors for expression in Escherichia coli. The methods used to express an NT1CL (cysteine-less NT1) in which all endogenous cysteine residues of the NT of IP3R1 were replaced by alanine, and a chimaeric NT (RyR2A-IBC) in which the A domain of the type 2 RyR (RyR2; residues 1–210) (GenBank: GI164831) [33] was fused to the IBC of rat IP3R1 were described recently [6]. The plasmid encoding NT1CL−IBC (NT1 where cysteine residues within the IBC were replaced by alanine) (Supplementary Table S1) was generated from the NT1 construct using QuikChange multi-site-directed mutagenesis (Agilent). The NT1CL construct was used as the template to prepare a plasmid encoding NT1CL−SD (NT1 where all endogenous cysteine residues within the SD were replaced by alanine) (Supplementary Table S1). The sequence encoding the cysteine-less SD was PCR-amplified from the NT1CL template using the primers listed in Supplementary Table S2. PCR products were then ligated into the pET41a(+) vector containing the open reading frame of the IBC of IP3R1 as SpeI/EcoRV fragments for expression in E. coli. The coding sequences of all expression constructs were confirmed. Supplementary Table S1 lists the proteins used and their abbreviations.

All N-terminal fragments of IP3R were expressed as fusion proteins linked to N-terminal GST (glutathione transferase) via a PreScission cleavage site in E. coli BL21 (DE3) exactly as described previously [32]. Bacteria were harvested (6000  for 15 min at 4°C), washed twice with phosphate-buffered saline and stored at −80°C before purification of the fragments. Bacterial pellets were suspended (1 g/10 ml) in Tris-buffered medium [50 mM Tris and 1 mM EDTA (pH 8.3)] and lysed by incubation with 100 μg/ml lysozyme (Sigma–Aldrich), 5 units/ml DNase (Sigma–Aldrich) and 10 μg/ml RNase (Sigma–Aldrich) for 1 h on ice, followed by sonication (30 s). After centrifugation (30000  for 60 min at 4°C), the supernatant was mixed with glutathione–Sepharose 4B beads [GE Healthcare; lysate/beads (v/v) 50:1) and incubated with gentle rotation for 1 h at 20°C. The beads were then washed with Tris-buffered medium supplemented with 1 mM dithiothreitol, centrifuged (500  for 5 min at 4°C) and incubated in the same medium (1 ml) supplemented with 160 units/ml GST-tagged PreScission protease (GE Healthcare) for 16 h at 4°C with gentle rotation. After centrifugation (500  for 5 min at 4°C), the NT fragments, cleaved from their GST tags, were recovered in the supernatant and rapidly frozen in liquid nitrogen before storage at −80°C. Immunoblots of the proteins used are shown in Supplementary Figure S3 (at http://www.biochemj.org/bj/451/bj4510177add.htm).

[3H]IP3 binding

Equilibrium competition binding assays were performed in Tris-buffered medium (500 μl) containing purified protein (1–5 μg), [3H]IP3 (0.25–0.75 nM) and appropriate concentrations of competing ligand. Where indicated, thimerosal was added 10 min before the addition of [3H]IP3. Reactions were terminated after 5 min by addition of poly(ethylene glycol)-8000 [Sigma–Aldrich; 30% (w/v) 500 μl] and γ-globulin (Sigma–Aldrich; 750 μg in 30 μl of Tris-buffered medium). Bound and free [3H]IP3 were then separated by centrifugation (20000  for 5 min). Results were fitted to Hill equations using GraphPad Prism (version 5.0, GraphPad) from which IC50, and thereby Kd (equilibrium dissociation constant) and pKd, values were calculated.

Ca2+ release from the intracellular stores of DT40 cells expressing IP3R subtypes

DT40 cells, in which the genes for all three endogenous IP3R subtypes had been disrupted [DT40-KO (knockout) cells], were used to generate stable cell lines expressing rat IP3R1, mouse IP3R2 or rat IP3R3 (DT40-IP3R1–3 cells) or a chimaeric IP3R (RyR1A-IP3R1) in which the SD of IP3R1 was replaced by the equivalent A domain of rabbit type 1 RyR (RyR1; residues 1—210; GenBank accession number: X15209) (DT40-RyR1A-IP3R1 cells) (Supplementary Table S1). The methods used to establish these stable cell lines have been described previously [6,34]. All DT40 cells were grown in suspension at 37°C in an atmosphere of 95% air and 5% CO2 in RPMI 1640 medium supplemented with 2 mM l-glutamine, 10% fetal bovine serum, 1% heat-inactivated chicken serum and 50 μM 2-mercaptoethanol. Cells were used or passaged when they reached a density of ~2×106 cells/ml. Expression levels of IP3Rs in the cell lines were compared by immunoblotting using antisera selective for each IP3R subtype or an anti-peptide serum (AbC) that interacts equally with all three IP3R subtypes. The immunoblotting methods were reported previously [34].

A low-affinity Ca2+ indicator, Mag-fluo-4, trapped within the endoplasmic reticulum, was used to measure luminal free Ca2+ concentration [31]. The endoplasmic reticulum of DT40 cells (~4×106 cells/ml) was loaded with indicator by incubating cells in the dark (for 1 h at 20°C) with 20 μM Mag-fluo-4/AM (acetoxymethyl ester) in Hepes-buffered saline [135 mM NaCl, 5.9 mM KCl, 11.6 mM Hepes, 1.5 mM CaCl2, 11.5 mM glucose and 1.2 mM MgCl2 (pH 7.3)], supplemented with Pluronic F127 (1 mg/ml) and BSA (0.4 mg/ml). After centrifugation (650  for 2 min), cells (~4×106 cells/ml) were suspended in Ca2+-free CLM [cytosol-like medium; 140 mM KCl, 20 mM NaCl, 2 mM MgCl2, 1 mM EGTA and 20 mM Pipes (pH 7.0)], supplemented with saponin (final concentration 20 μg/ml) to allow for selective permeabilization of the plasma membrane. After ~4 min at 37°C, when all cells had become permeable to 0.1% Trypan Blue, the cells (5×105 cells/ml) were recovered (650  for 2 min) and suspended at 20°C in Mg2+-free CLM, supplemented with CaCl2 (375 μM, to provide a final free Ca2+ concentration of 220 nM after addition of 1.5 mM MgATP) and 10 μM FCCP (carbonyl cyanide p-trifluoromethoxyphenylhydrazone). The cells were then distributed into poly-l-lysine-coated 96-well plates (50 μl/well), centrifuged (300  for 2 min) and used for fluorescence measurements. Fluorescence (excitation 485 nm, emission 520 nm, measured at 1.5 s intervals) was recorded at 20°C using a FlexStation III plate-reader (Molecular Devices) [31]. MgATP (1.5 mM) was added to allow intracellular stores to sequester Ca2+. After the stores had loaded to steady state (~150 s), IP3 was added with thapsigargin (1 μM to inhibit Ca2+ re-uptake). Ca2+ release was measured after a further 60 s. The Ca2+ release evoked by IP3 is expressed as a fraction of the ATP-dependent Ca2+ uptake [31]. Concentration–effect relationships were fitted to Hill equations using non-linear curve fitting (version 5.0, GraphPad Prism).

Statistical analyses of IP3R sensitivity to IP3 used pEC50 and pKd values (negative logarithm of the EC50 and Kd respectively). Results are presented as means±S.E.M. For clarity, mean EC50 and Kd values (derived from pEC50 or pKd) are shown alongside. In paired comparisons of the effects of thimerosal, ΔpEC50 (or ΔpKd) values are shown, where ΔpEC50=EC50thimerosal−pEC50control. Statistical analyses used ANOVA followed by paired Student's t test. P<0.05 was considered significant.

RESULTS AND DISCUSSION

Thimerosal potentiates IP3-evoked Ca2+ release via IP3R1 and IP3R2

In DT40 cells stably expressing only IP3R1 (DT40-IP3R1 cells) (Figure 1A), IP3 caused a concentration-dependent release of Ca2+ from intracellular stores (pEC50=7.95±0.04) (Figure 1B). Pre-incubation with thimerosal (10 μM for 120 s) increased the sensitivity to IP3 (pEC50=8.95±0.31, ΔpEC50=1.00±0.32) without affecting loading of the Ca2+ stores or the maximal response to IP3 (Table 1 and Figure 1B). Similar results were obtained with DT40 cells expressing only IP3R2 (DT40-IP3R2 cells) (Figure 1A), where 10 μM thimerosal also increased the sensitivity to IP3 (ΔpEC50=0.60±0.20) (Figure 1C). For both IP3R1 and IP3R2, higher concentrations of thimerosal inhibited responses to IP3 (Table 1). However, in DT40 cells expressing IP3R3 (DT40-IP3R3 cells) (Figure 1A), 10 μM thimerosal inhibited IP3-evoked Ca2+ release (ΔpEC50=−0.17±0.01) (Figure 1D), without significantly affecting loading of the intracellular Ca2+ stores (Table 1). The latter demonstrates that the diminished responses to IP3 were not due to thimerosal activating IP3R3 and thereby draining the Ca2+ stores before addition of IP3. Functional effects of thimerosal were reported previously for DT40 cells expressing IP3R1 or IP3R3 [29], but use of the same single sub-threshold concentration of IP3 for both IP3R subtypes when IP3R3 is much less sensitive to IP3 (Figure 1 and Table 1) [35] compromised the conclusion that thimerosal sensitizes only IP3R1. Our results, demonstrating that low concentrations of thimerosal sensitize IP3R1 and IP3R2 to IP3 while inhibiting IP3R3, are consistent with previous studies in which thimerosal [29,36-42] or other thiol reagents [40,43,44] potentiated IP3-evoked Ca2+ signals in cells expressing predominantly IP3R1 [29,37,42] or IP3R2 [38,40,43,44], but not in cells where IP3R3 predominates [29,39,41]. We conclude that within homotetrameric populations of IP3R, thimerosal potentiates responses from IP3R1 and IP3R2, but inhibits IP3R3.

Figure 1

IP3-evoked Ca2+ release via IP3R1 and IP3R2 is sensitized by thimerosal

(A) Immunoblot (4×104 cells/lane) using an antiserum (AbC) that recognizes a conserved NT sequence in all IP3R subtypes demonstrates that expression levels of the three IP3R subtypes and chimaeric IP3R are broadly similar in each of the four cell lines. There are no detectable IP3Rs in the DT40-KO cells. (B–D) Concentration-dependent effects of IP3 on Ca2+ release are shown for control cells and after treatment with thimerosal (10 μM, 2 min) for DT40 cells expressing each of the three IP3R subtypes. Results are means±S.E.M. for at least three independent experiments. The structure of thimerosal is shown within (B). ○, control; ●, thimerosal.

Table 1

Effects of thimerosal on IP3-evoked Ca2+ release by different IP3Rs

DT40 cells expressing each of the IP3R subtypes were incubated with the indicated concentrations of thimerosal during loading of the intracellular stores with Ca2+ (2 min) before addition of IP3. Results show Ca2+ uptake (as a percentage of its value in the absence of thimerosal), maximal IP3-evoked Ca2+ release (as a percentage of the Ca2+ content of the stores before addition of IP3) and the Hill coefficient (h), pEC50 and EC50 values for the IP3-evoked Ca2+ release. All results (except EC50, which shows only the mean, see the Materials and methods section) are shown as means±S.E.M. for three to four independent experiments. ND, not determined.

IP3R subtypes	Thimerosal (μM)	Ca2+ uptake (%)	EC50 (nM)	pEC50	h	Maximal Ca2+ release (%)
IP3R1	0	100	11	7.95±0.04	2.0±0.1	84±2
	1	98±5	7	8.16±0.13	1.2±0.2	89±1*
	10	97±7	1	8.95±0.31*	2.7±0.8	85±2
	100	93±6	77	7.11±0.18*	1.0±0.4	42±11*
IP3R2	0	100	23	7.64±0.01	1.6±0.1	74±2
	1	99±2	14	7.87±0.22	1.3±0.2	78±3
	10	96±7	6	8.24±0.26*	2.2±0.3	77±4
	100	93±3	35	7.46±0.28	0.9±0.2	67±2*
IP3R3	0	100	576	6.24±0.04	1.2±0.1	88±2
	1	99±3	687	6.16±0.08	1.0±0.1	90±5
	10	97±8	860	6.07±0.06*	1.3±0.1	78±8
	100	86±6	2185	5.66±0.02*	3.0±0.5	54±2*
RyR1A-IP3R1	0	100	65	7.19±0.01	2.4±0.1	81±4
	1	97±4	78	7.11±0.04	2.0±0.2	76±3
	10	89±4	238	6.62±0.18*	1.6±0.2	80±6
	100	66±8*	ND	ND	ND	12±1≠

*P<0.05 relative to its control. ≠Response to 1 μM IP3.

Thimerosal stimulates IP3 binding to the NTs of IP3R1 and IP3R2

Thimerosal stimulates IP3-evoked Ca2+ release via purified IP3R1 [37] and it stimulates [3H]IP3 binding to purified IP3R1 [37] and its isolated NT [29]. This confirms that thimerosal directly affects IP3R1. Activation of IP3R is initiated by IP3 binding to the IBC and proceeds via re-arrangement of interactions between the IBC and SD [6]. We therefore examined [3H]IP3 binding to the NTs of IP3R1–IP3R3, a region that comprises the critical SD and IBC (Supplementary Figure S1A).

Incubation (10 min) of purified NT from IP3R1 (NT1) with thimerosal caused a concentration-dependent stimulation of [3H]IP3 binding that was maximal with 100 μM thimerosal and then reversed at higher concentrations (Figure 2A). A biphasic effect of thimerosal on [3H]IP3 binding to full-length IP3R1 [45] and an NT region of IP3R1 (residues 1–581) was reported previously [46]. The maximally effective concentration of thimerosal (100 μM) caused a 12-fold increase in the affinity of NT1 for IP3 (ΔpKd=1.07±0.28) (Figure 2B and Table 2). The effect of maximally effective concentrations of thimerosal (10–100 μM) on the sensitivities of functional responses of IP3R1 to IP3 (ΔpEC50=1.00±0.32) and IP3 binding to NT1 (ΔpKd=1.07±0.28) are similar: both are increased ~10-fold. The results were similar with IP3R2: 100 μM thimerosal caused a ~10-fold increase in the affinity of NT2 for IP3 (ΔpKd=1.05±0.19) (Figure 2C and Table 2). There was no significant effect of thimerosal on IP3 binding to the NT of IP3R3 (Figure 2D and Table 2).

Figure 2

Thimerosal stimulates [3H]IP3 binding to the NT of IP3R1 and IP3R2

(A) Specific binding of [3H]IP3 (0.5 nM) to the NT of IP3R1 is shown after pre-incubation (10 min) with the indicated concentrations of thimerosal. Results are shown as the percentage of specific [3H]IP3 binding measured in the absence of thimerosal. (B–D) Concentration-dependent effects of IP3 on specific [3H]IP3 binding (0.5 nM) to the NT from each of the three IP3R subtypes alone or after incubation with thimerosal (100 μM, 10 min). (A–D) Results are means±S.E.M. for four independent experiments. ○, control; ●, thimerosal.

Table 2

Effects of thimerosal on [3H]IP3 binding to the NT of each IP3R subtype

Equilibrium-competition binding experiments using [3H]IP3 (0.5 nM) and NT1–NT3 show the effects of thimerosal (100 μM, 10 min) on the Kd (mean), pKd (mean±S.E.M.) and Hill coefficient (h, mean±S.E.M.) for IP3. ΔpKd=pKdthimerosal−pKdcontrol. Results are from at least four independent experiments.

	pKd (Kd; nM) h
NT subtype	Control	Thimerosal	ΔpKd
NT1	8.62±0.10	9.69±0.26*	1.07±0.28
	(2.39)	(0.20)
	1.2±0.2	1.2±0.3
NT2	9.12±0.07	10.17±0.15*	1.05±0.19
	(0.76)	(0.07)
	0.6±0.04	0.6±0.02
NT3	8.06±0.06	7.91±0.17	0.15±0.22
	(8.72)	(12.4)
	0.7±0.06	1.0±0.05

*P<0.05 relative to its control.

Previous work suggested that thimerosal stimulates [3H]IP3 binding to full-length IP3R1, but not to IP3R3, whereas it stimulated binding to the NT region (residues 1–581) of IP3R1 and IP3R3, but not to that of IP3R2 [29,46]. The previous conclusions are neither internally consistent nor consistent with our observations using NT1–NT3 (residues 1–604) (Figure 2). The residues missing in the shorter NT fragments are reasonably conserved between IP3R subtypes and they lack cysteine residues (Supplementary Figure S1B). However, from the high-resolution structure of the IBC [4], an NT region truncated at residue 581 is likely to disrupt an α-helix (α9 in [4]), the NT of which forms essential interactions with IP3. Furthermore, the published affinities for IP3 of these shorter fragments from IP3R1 and IP3R3 are indistinguishable, which is inconsistent with evidence that in both NT and full-length proteins, IP3R3 has lower affinity than IP3R1 [35] (Table 2). It seems likely that the shorter fragments used in the previous studies [29,46] may not retain the native structure of the IBC.

Our results with [3H]IP3 binding to the NT from all three IP3R subtypes (Figure 2), evidence from [3H]IP3 binding to full-length IP3R1 and IP3R3 expressed in Sf9 insect cells [29], our functional analyses of DT40 cells expressing homogenous populations of IP3R subtypes (Figure 1) and substantial evidence from native cells expressing different IP3R subtypes (discussed above) are consistent with the conclusion that low concentrations of thimerosal stimulate IP3 binding and IP3-evoked Ca2+ release via IP3R1 and IP3R2, but not via IP3R3. These observations support the hypothesis that cysteine residues within the NT of IP3R1 and IP3R2 mediate the sensitizing effects of thimerosal on IP3-evoked Ca2+ release. Subsequent experiments address this hypothesis.

Thimerosal stimulates agonist, but not antagonist, binding to the NT of IP3R1

Heparin is a competitive antagonist of IP3R [47], and dimeric IP3 (see Figure 3A for structure) is a high-affinity partial agonist: the IP3 dimer binds to the IP3R, but causes lesser activation than IP3 such that the channel opens less frequently [5]. Both heparin and IP3 dimer completely displaced specifically bound [3H]IP3 from NT1, but thimerosal (100 μM, 10 min) affected the affinity of neither: ΔpKd was 0.11±0.09 and −0.06±0.19 for heparin and IP3 dimer respectively (Figure 3 and Table 3). These results are important because they demonstrate that thimerosal selectively affects the binding of ligands that activate IP3R. It is, however, noteworthy that thimerosal alone does not activate IP3R. Concentrations of thimerosal (≤10 μM) that cause substantial sensitization of IP3R1 and IP3R2 do not affect the Ca2+ content of the stores in the absence of IP3 (Table 1). These results suggest that thimerosal might selectively stabilize an agonist-bound active conformation of IP3R1.

Table 3

Effects of thimerosal on binding of heparin and IP3 dimer to the NT from IP3R1

Equilibrium-competition binding experiments using [3H]IP3 (0.25 nM) and NT1 show the effects of thimerosal (100 μM, 10 min) on the Kd (mean), pKd (mean±S.E.M.) and Hill coefficient (h, mean±S.E.M.) for IP3 dimer and heparin. ΔpKd= pKdthimerosal−pKdcontrol. Results are from ≥four independent experiments.

	pKd (Kd) h
	Control	Thimerosal	ΔpKd
IP3 dimer	9.81±0.07	9.75±0.14	−0.06±0.19
	(0.15 nM)	(0.18 nM)
	1.2±0.1	1.7±0.4
Heparin	7.06±0.09	7.17±0.03	0.11±0.09
	(87 ng/ml)	(68 ng/ml)
	0.82±0.17	0.73±0.09

Cysteine residues within both the SD and IBC are required for thimerosal to stimulate IP3 binding

Previous work demonstrated that thimerosal stimulates [3H]IP3 binding to NT1, but not to the IBC [29]. Results shown in Figure 4 confirm that conclusion. We have shown that NT1CL is structurally and functionally indistinguishable from native NT1 [6]. The lack of effect of thimerosal on [3H]IP3 binding to NT1CL (Figure 4A) confirms that cysteine residues mediate its effects. We had anticipated, in keeping with earlier evidence [29], that cysteine residues within the SD were likely to mediate the effects of thimerosal. However, thimerosal had no effect on [3H]IP3 binding to NT lacking cysteine in either the SD (NT1CL−SD) or the IBC (NT1CL−IBC) (Figures 4B and 4C). These results demonstrate that stimulation of IP3 binding by thimerosal requires cysteine residues within both the SD and IBC.

Figure 4

Stimulation of IP3 binding by thimerosal requires cysteine residues within the SD and IBC

(A) Effects of thimerosal (100 μM, 10 min) on specific binding of [3H]IP3 (0.5 nM) to each of the indicated proteins. Results are expressed as a percentage of the binding measured in the absence of thimerosal. (B and C) Concentration-dependent effects of thimerosal on specific [3H]IP3 binding to NT1CL−SD (B) and NT1CL−IBC (C). All results are means±S.E.M. for four independent experiments.

Residues within the NT are required for thimerosal to potentiate IP3-evoked Ca2+ release

It would be instructive to examine the functional effects of thimerosal on full-length IP3R1 in which cysteine residues within the NT are replaced by alanine and so verify whether the NT mediates the effects of thimerosal on IP3-evoked Ca2+ release. We have demonstrated that IP3R1 with a cysteine-less NT is functional, but only in transiently transfected DT40-KO cells [6], in which it is impracticable to complete the concentration–effect relationships needed to define the effects of thimerosal. We have not yet succeeded in establishing stable DT40 cell lines expressing IP3R1 with a cysteine-less NT. We therefore adopted another approach and used a stable DT40 cell line expressing a chimaeric IP3R1 in which the native SD is replaced by the equivalent region (the A domain) from a RyR (RyR1A-IP3R1) [6] (Figure 5A, and Supplementary Table S1).

Figure 5

A chimaeric IP3R1 is not stimulated by thimerosal

(A) The proteins used. (B) Concentration-dependent effects of thimerosal on specific [3H]IP3 (0.5 nM) binding to RyR2A-IBC. Results are expressed as percentages of the binding measured in the absence of thimerosal. (C) Concentration-dependent effects of IP3 on Ca2+ release from DT40-RyR1A-IP3R1 cells are shown for control cells and after treatment with thimerosal (10 μM, 2 min). Results in (B and C) are means±S.E.M. from four independent experiments.

The A domain of RyR, which is structurally similar to the SD of IP3R [6,48-50] (Supplementary Figure S1C and S1D), can functionally replace the SD of IP3R1 to allow both IP3-evoked Ca2+ release from a chimaeric IP3R and appropriate regulation of IP3 binding to a chimaeric NT [6]. All effective concentrations of thimerosal (1 μM–1 mM) inhibited [3H]IP3 binding to the chimaeric NT (RyR2A-IBC) (Figures 5A and 5B), although the effect appeared biphasic. Thimerosal (10 μM, 2 min) also inhibited IP3-evoked Ca2+ release from the chimaeric IP3R (RyR1A-IP3R1) (Figure 5C and Table 1). These results suggest that these cysteine residues present within the SD of IP3R1, but absent from the A domain of RyR, are required for thimerosal to potentiate IP3-evoked Ca2+ release.

Conclusions

Thimerosal potentiates IP3-evoked Ca2+ release via IP3R1 and IP3R2, but not via IP3R3 (Figure 1). The parallel behaviour of NTs from each of the three IP3R subtypes (Figure 2), our demonstration that cysteine residues are required within both the IBC and SD for thimerosal to stimulate IP3 binding (Figure 4) and evidence that thimerosal does not potentiate responses from a chimaeric IP3R1 in which the SD is replaced by the equivalent domain of a RyR (Figure 5) provide persuasive evidence that cysteine residues within the SD and IBC mediate functional responses to thimerosal. Furthermore, thimerosal selectively stimulates agonist, but not antagonist, binding to NT1 (Figure 3), suggesting that its interactions with critical cysteine residues selectively stabilize an active conformation of the NT. Because IP3 binding re-arranges interactions between the SD and IBC [6], this conclusion aligns with a previous suggestion that thimerosal stabilizes an interaction between the IBC and SD of IP3R1 [29].

We do not know which specific cysteine residues are modified by thimerosal. Despite its subtype-selective actions (Figures 1 and 2) and evidence that sensitization by thimerosal is mediated by cysteine residues within the NT, we have been unable to identify candidate cysteine residues from sequence data alone (see legend to Supplementary Figure S2 at http://www.biochemj.org/bj/451/bj4510177add.htm). Preliminary MALDI-TOF (matrix-assisted laser desorption ionization–time-of-flight) analyses were uninstructive, and it is impracticable to assay NT in which every pair of cysteine residues are mutated. It is perhaps surprising that cysteine residues in both the SD and IBC are required for thimerosal to sensitize IP3R1. This requirement is unlikely to result from cross-linking of a pair of cysteine residues by thimerosal because available evidence suggests that thimerosal forms a mercuryethyl adduct with a single cysteine [24]. An alternative possibility is that thimerosal disrupts a disulfide bridge between cysteine residues in the SD and IBC that constrains IP3R activation. This too seems unlikely. There is no evidence from the high-resolution structure of the NT for such a disulfide bridge [6]. Furthermore, IP3R with cysteine-less NT are not constitutively active and respond normally to IP3 [6]. We also used algorithms that predict the accessibility [51] and reactivity [52] of cysteine residues together with modelled structures of NT2–NT3 [53] to identify cysteine residues that might account for the differential sensitivities of IP3R1–IP3R3 to thimerosal. The results suggest that within the SD, there are two accessible cysteine residues (Cys206 and Cys214 in IP3R1) and they are predicted to be equally accessible in all three subtypes, but the residue equivalent to Cys214 in IP3R1 is less reactive in IP3R3 (Supplementary Figure S2). Within the IBC, Cys556 is predicted to be accessible in IP3R1 and IP3R2, but it is buried in IP3R3. Cys214 in the SD and Cys556 in the IBC of IP3R1 therefore have the properties expected to explain the sensitization of IP3R1 and IP3R2, but not IP3R3, by thimerosal. Further experimental work would be needed to assess this directly. We have provided the first systematic analysis of the effects of thiol reagents on each IP3R subtype. We conclude that thimerosal sensitizes IP3R1 and IP3R2, but not IP3R3, to IP3. It does so by modifying cysteine residues within both the SD and IBC and thereby stabilizing an active conformation of the IP3R.