Fluorescent biosensor for quantitative real-time measurements of inositol 1,4,5-trisphosphate in single living cells.

The second messenger inositol 1,4,5-trisphosphate (IP(3)) plays a central role in the generation of a variety of spatiotemporally complex intracellular Ca(2+) signals involved in the regulation of many essential physiological processes. Here we describe the development of "LIBRA", a novel ratiometric fluorescent IP(3) biosensor that allows for the quantitative monitoring of intracellular IP(3) concentrations in single living cells in real time. LIBRA consists of the IP(3)-binding domain of the rat type 3 IP(3) receptor fused between the fluorescence resonance energy transfer pair cyan fluorescent protein and yellow fluorescent protein and preceded by a membrane-targeting signal. We show that the LIBRA fluorescent signal is highly selective for IP(3) and unaffected by concentrations of Ca(2+) and ATP in the physiological range. In addition, LIBRA can be calibrated in situ. We demonstrate the utility of LIBRA by monitoring the temporal relationship between the responses intracellular IP(3) and Ca(2+) concentrations in SH-SY5Y cells following acetylcholine stimulation.

From the ‡Department of Dental Pharmacology, School of Dentistry, Health Sciences University of Hokkaido, Ishikari-Tobetsu, Hokkaido 061-0293, Japan and the ¶Membrane Biology Section, Gene Therapy and Therapeutics Branch, NIDCR, National Institutes of Health, Bethesda, Maryland 20892 The second messenger inositol 1,4,5-trisphosphate (IP 3 ) plays a central role in the generation of a variety of spatiotemporally complex intracellular Ca 2؉ signals involved in the regulation of many essential physiological processes. Here we describe the development of "LI-BRA", a novel ratiometric fluorescent IP 3 biosensor that allows for the quantitative monitoring of intracellular IP 3

concentrations in single living cells in real time.
LIBRA consists of the IP 3 -binding domain of the rat type 3 IP 3 receptor fused between the fluorescence resonance energy transfer pair cyan fluorescent protein and yellow fluorescent protein and preceded by a membrane-targeting signal. We show that the LIBRA fluorescent signal is highly selective for IP 3 and unaffected by concentrations of Ca 2؉ and ATP in the physiological range. In addition, LIBRA can be calibrated in situ. We demonstrate the utility of LIBRA by monitoring the temporal relationship between the responses intracellular IP 3

and Ca 2؉ concentrations in SH-SY5Y cells following acetylcholine stimulation.
Many plasma membrane receptors act by stimulating the phospholipase C (PLC) 1 -dependent hydrolysis of phosphatidyl-inositol 4,5-bisphosphate (PIP 2 ) to produce the intracellular messenger inositol 1,4,5-trisphosphate (IP 3 ) (1). IP 3 in turn releases Ca 2ϩ from intracellular stores via IP 3 receptors (IP 3 Rs) resulting in the activation of a variety of Ca 2ϩ -dependent processes. IP 3 -dependent Ca 2ϩ responses can exhibit complex spatial and temporal patterns, and a number of competing models have been proposed to account for them (1)(2)(3). Because of its central role in these responses, the ability to quantitatively monitor temporal changes in [IP 3 ] i in single living cells is of vital importance to understanding these phenomena. Since IP 3 Rs are the natural physiological target for IP 3 , we reasoned that their IP 3 -binding domain (4, 5) might be used to construct an effective and specific intracellular IP 3 detector. Here we describe such a ratiometric fluorescent biosensor that in addition allows for in situ calibration. We demonstrate the utility of this biosensor by examining the relationship between the responses of [ (6).
Plasmid Construction-A CFP/YFP fusion construct was made by cutting EYFP out of pEYFP-N1 (Clontech) using BamHI and XbaI and ligating into pECFP-C1 (Clontech) cut with the same enzymes. The multiple cloning site of this vector was then removed by cutting with BspEI and BamHI and replaced with a linker generated from two synthetic oligonucleotides. The forward sequence of this linker was TCC GGA AAG CTC GAG GCA GTA AGA TCT GGC TCC GCC GAC GAT GAC GAT AAG GCC GGA TCT GTC GAC GCA GTC GGA TCC, where the reconstituted BspEI and BamHI sites have been included for clarity and the sequence has been parsed into codons. This fusion construct was referred to as pCY-N. The linker plus EYFP sequence of pCY-N was then cut out with BsrGI and ligated into pECFP-mem (Clontech) cut with the same enzyme. Finally the multiple cloning site originating from pECFP-mem was removed by cutting with Eco47III and SmaI and religating. The resultant construct, mCY, codes for ECFP preceded by the N-terminal 20 amino acids of neuromodulin (a membrane-targeting signal) and followed by the above linker and EYFP.
To construct LIBRA (luminous inositol trisphosphate-binding domain for ratiometric analysis) the IP 3 -binding domain of the rat type 3 IP 3 R (amino acids 1-604) was amplified by PCR using pCB-EGFP: IP 3 R3 (7) as the template and incorporating XhoI sites at either end. This sequence was then ligated into the XhoI site in the linker region of mCY using standard methods. LIBRA⌬N was constructed in the same way using amino acids 227-604 of the rat type 3 IP 3 R (8). The forward PCR primers used were ACG CAT ACT CGA GAT GAA TGA AAT GTC CAG C for LIBRA and AAG CAT ACT CGA GTT CCG GGA CCA TCT GGA G for LIBRA⌬N. The same reverse primer, AGC GTA TCT CGA GCT TCC GGT TGT TGT GCA G, was used for both PCR reactions. The correctness of all constructs was confirmed by restriction digestion and sequencing.
Cell Culture and Transfection-SH-SY5Y cells purchased from Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (Braunschweig, Germany) were cultured in Dulbecco's modified Eagle's medium (Invitrogen) with low glucose (100 mg/ml), supplemented with 10% newborn calf serum, 584 mg/ml L-glutamine, 110 mg/ml sodium pyruvate, 100 units/ml penicillin and 100 g/ml streptomycin. Cells were transfected with plasmids using LipofectAMINE 2000 (Invitro-* This work was supported in part by the Academic Sciences Frontier Project and by Grant-in-aid for Scientific Research 13877314 (to Y. T.) from the Ministry of Education, Science, Sports and Culture of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AB161231.
Measurement of Fluorescence-SH-SY5Y cells were grown on fibronectin-coated glass cover slips incorporated into an experimental chamber (6). Cells were washed with HBSS-H and rested for at least 15 min prior to experiments. In some experiments, cells were loaded with fura-2 by incubation for 5 min at room temperature in HBSS-H containing 0.25-0.5 M fura-2/AM (Dojin Chemicals, Kumamoto, Japan). Cells in HBSS-H or Ca 2ϩ -free HBSS-H (in which 1.3 mM CaCl 2 was replaced with 0.1 mM EGTA) were stimulated by exposure to various concentration of acetylcholine (Ach) or 1 M ionomycin as indicated.
Permeabilization was performed by exposing cells to ICM containing 100 g/ml (w/v) saponin (ICN, Cleveland, OH) for ϳ1 min. Permeabilized cells were washed with ICM and then exposed to ICM containing various concentrations of IP 3 and other reagents.
Fluorescence images were captured using a dual-wavelength ratio imaging system (Hamamatsu Photonics, Shizuoka, Japan) consisting of a cooled CCD camera (HiSCA) and W-View optics coupled to a Nikon Diaphot 300 inverted fluorescence microscope equipped with a Nikon Fluor 40 oil immersion objective (NA 1.3). Fluorescence of LIBRA, LIBRA⌬N, and mCY were monitored with excitation at 425 nm and dual-emission at 480 and 535 nm. Simultaneous monitoring of LIBRA and fura-2 was performed with sequential excitation at 380 nm (for fura-2) and 425 nm (for LIBRA) and dual-emission at 480 nm (for LIBRA) and 535 nm (for LIBRA and fura-2). Data were analyzed with AQUACOSMOS software (Hamamatsu Photonics). [Ca 2ϩ ] i was calculated from the fura-2 fluorescence intensity, F, using the formula where K d ϭ 135 nM, and F min and F max are the values of F at zero and limitingly high [Ca 2ϩ ], respectively (9). F min was determined by assuming that the resting [Ca 2ϩ ] i was 50 nM; F max was measured following the application of 1 M ionomycin.

RESULTS AND DISCUSSION
We designed two potential IP 3 biosensors (Fig. 1a and see "Experimental Procedures") consisting of the IP 3 -binding domain of the rat type 3 IP 3 R (8) fused between the well established FRET pair CFP and YFP (10,11) and preceded by the N-terminal 20 amino acids of neuromodulin (12), a plasma membrane-targeting signal. The first of these constructs, which we refer to as LIBRA contains residues 1-604 of the rat type 3 IP 3 R, while LIBRA⌬N contains residues 227-604. The control molecule mCY lacks the IP 3 -binding domain. Using confocal microscopy we found that these recombinant proteins localized to the plasma membrane and Golgi area when expressed in SH-SY5Y human neuroblastoma cells (data not shown); also ϳ90% of their fluorescence was retained by the cells after permeabilization with saponin (Fig. 1b).
We looked for FRET between CFP and YFP by exciting CFP at 425 nm and recording emitted fluorescence at 480 and 535 nm. When permeabilized SH-SY5Y cells transfected with LI-BRA were exposed to 10 M IP 3 an increase in the 480 nm signal (ϳ4.5%) and a parallel decrease in the 535 nm signal (ϳ3.6%) were observed within ϳ1 s (Fig. 1c). By contrast neither LIBRA⌬N nor mCY showed any detectable fluorescence changes in response to 10 M IP 3 (Fig. 1d). The CFP donor quenching of LIBRA that was relieved by YFP acceptor photobleaching (13) was ϳ3.9% in the presence of 10 M IP 3 and ϳ8.6% in its absence, confirming that the FRET efficiency of LIBRA is decreased in its IP 3 -bound form. It has been demonstrated that residues 226 -576 of the mouse type 1 IP 3 R constitute the essential "core" region required for high affinity IP 3 binding and that the N-terminal 225 residues act to suppress the IP 3 binding affinity of the core (14). Interestingly, LIBRA⌬N, which lacks this N-terminal suppressor region, shows no IP 3 -dependent changes in fluorescence. Accordingly we speculate that the conformational change underlying the IP 3 -dependent LIBRA FRET signal could be related to an effect of IP 3 on the interaction between the core and suppressor regions.
Changes in the LIBRA 480/535 nm emission ratio (⌬Ratio) were highly sensitive to [IP 3 ] and selective for 1,4,5-IP 3 versus other inositol phosphates (Fig. 2). ⌬Ratio had a monophasic dependence on [IP 3 ] with apparent dissociation constant (K d ) of 404 nM and Hill coefficient (n) of 1.118 (Fig. 2b). The K d for adenophostin A, a more potent agonist for the IP 3 R than IP 3 (15), was 65 nM and we estimate that inositol 1,3,4,5-tetrakisphosphate (IP 4 ), inositol 4,5-bisphosphate (IP 2 ), and inositol 1,3,4-trisphosphate (1,3,4-IP 3 ) bind to LIBRA with K d values of ϳ15, ϳ40, and Ͼ Ͼ10 M, respectively (Fig. 2b). It is known that the activity of the IP 3 R is modulated by Ca 2ϩ (16,17) and ATP (18), although their effects are not mediated by the IP 3 -binding domain (5,19). In control experiments we have confirmed that both Ca 2ϩ (0 -1 M) and ATP (0 -3 mM) have no effect on LIBRA fluorescence or on the ⌬Ratio seen in response to IP 3 (Fig. 2c). We also found that, while changes in pH affected the basal emission ratio as a result of the well known effects of pH on YFP fluorescence (13), the ⌬Ratio elicited by IP 3 was unaffected over the pH range 7.0 -7.6 ( Fig. 2d). Taken together the above results show that ⌬Ratio (at constant pH) is a sensitive and specific measure of [IP 3 ].
We next used the ⌬Ratio of LIBRA to follow the dynamics of [IP 3 ] i after muscarinic stimulation of intact SH-SY5Y cells (Fig.  3a). In the presence of extracellular Ca 2ϩ , application of 10 or 100 M Ach increased ⌬Ratio in Ͼ80% of LIBRA-expressing cells, and 1 M Ach elicited a response in ϳ60% of the cells tested. Increasing [Ca 2ϩ ] i with ionomycin had no effect on ⌬Ratio (Fig. 3, a and e)  cells or in cells loaded with the intracellular pH indicator BCECF (data not shown), confirming the specificity of the LIBRA response. The PLC inhibitor U73122 (20) completely blocked the Ach-induced increase in ⌬Ratio but had no effect on the resting emission ratio (Fig. 3b) suggesting that the resting [IP 3 ] i in these cells is below the detectable range of LIBRA.
The magnitude of the LIBRA ⌬Ratio response increased with [Ach] reaching a maximum 2-5 min after stimulation (Fig. 3, a  and c). As discussed in more detail below the time scale of these responses is much longer than the onset of Ach-induced Ca 2ϩ spikes in this cell type (5-20 s). Peak ⌬Ratio values were similar in the presence and absence of extracellular Ca 2ϩ (Fig.  3c); however, significant differences were seen in the time courses of these responses (Fig. 3a). In the presence of Ca 2ϩ , in 20 out of 32 responding cells, ⌬Ratio rose monotonically to a sustained maximum level, while in the remaining 12 cells ⌬Ratio rose to a peak then fell to a plateau level that was Ͼ50% of the maximal response. In contrast, in the absence of Ca 2ϩ , only 9 out of 43 responding cells showed a sustained rise in ⌬Ratio in response to Ach; in the remaining 34 cells ⌬Ratio rose to a peak then fell to a value close to resting levels. In both experimental conditions the application of Ach typically elicited a rapid spike in [Ca 2ϩ ] i (monitored using fura-2), which then fell to a lower sustained level above base line in the presence of extracellular Ca 2ϩ and to base-line levels in its absence (data not shown). The sustained rise in ⌬Ratio in the presence of extracellular Ca 2ϩ and the transient rise in its absence (Fig. 3a) suggest a role for a sustained rise in [Ca 2ϩ ] i in maintaining increased [IP 3 ] i .
To look for effects of LIBRA expression on Ach-induced Ca 2ϩ release, we examined the latency for the onset of Ca 2ϩ spikes after Ach stimulation. Following the application of 1 M Ach in the absence of extracellular Ca 2ϩ this latency was 16.7 Ϯ 4.2 s (n ϭ 11) in LIBRA-expressing cells and 18.9 Ϯ 2.0 s (n ϭ 24) in non-expressing cells. Since any buffering of intracellular [IP 3 ] by LIBRA would have been expected to increase this latency (i.e. to delay the Ca 2ϩ spike), these results are consistent with the hypothesis that LIBRA expression has little if any buffering effect To directly explore the relationship between the Ach-induced [IP 3 ] i response and Ca 2ϩ release from intracellular stores, we monitored the fluorescence of both LIBRA and the [Ca 2ϩ ] i indicator fura-2 simultaneously in the same cells in Ca 2ϩ -free medium. In experiments where cells were exposed to two successive applications of a low [Ach], the second application of Ach consistently elicited a significant but smaller increase in [IP 3 ] i than the first and was accompanied by little or no increase in [Ca 2ϩ ] i (Fig. 3d) (Fig. 3, d and e). Similar results were observed in the presence of extracellular Ca 2ϩ (data not shown). At low [Ach] a subsequent application of a higher [Ach] resulted in additional Ca 2ϩ release (Fig. 3d)  By assuming that the response of the LIBRA ⌬Ratio to IP 3 is the same in intact and permeabilized cells, one can convert the above LIBRA signals to [IP 3 ] i using the concentration-response curve shown in Fig. 2b. Consistent with this assumption we have confirmed that the increase in ⌬Ratio due to microinjection of a supramaximal concentration of IP 3 into intact LIBRAexpressing cells is not significantly different from the maximal increase in ⌬Ratio due to IP 3 in permeabilized cells (data not shown). Thus we estimate that the threshold [IP 3 ] i required to elicit a Ca 2ϩ spike in SH-SY5Y cells (⌬Ratio ϳ 0.008) is ϳ50 nM. This value is very close to the threshold concentration of photoreleased IP 3 (ϳ60 nM) previously found to be required for triggering [Ca 2ϩ ] i spikes in Xenopus oocytes (22). We also estimate that the peak [IP 3 ] i achieved by 1, 10, and 100 M Ach in SH-SY5Y cells are ϳ100, 210, and 360 nM, respectively.
Our results demonstrate the utility of LIBRA to directly follow physiologically relevant changes in [IP 3 ] i in intact mammalian cells in real time. LIBRA has several important advantages over a previous method for monitoring [IP 3 ] i that employs the intracellular redistribution of a chimeric protein consisting of the pleckstrin homology domain (PHD) of PLC␦1 fused to GFP (GFP-PHD) (23). This probe was originally designed to detect changes in membrane PIP 2 levels (24, 25) but subsequently was also shown to be sensitive to [IP 3 ] i (2, 23). More recent experiments have confirmed that GFP-PHD redistribu-tion can reflect physiologically relevant changes in both IP 3 and PIP 2 and that its selectivity varies with GFP-PHD expression level (26,27). In addition to these complexities, for technical reasons (26, 27) it has not been possible to quantitate the GFP-PHD signal.
The temporal dynamics of G protein coupled receptor-mediated Ca 2ϩ responses have been shown to depend on the effects of cytosolic and luminal [Ca 2ϩ ] on IP 3 Rs (1, 6, 17, 28) as well as on positive feedback loops to enhance IP 3 production (29). In addition, recent studies suggest the involvement of PKC and scaffolding proteins on IP 3 synthesis (30). We anticipate that the ability to quantitatively monitor [IP 3 ] i with LIBRA and related ratiometric biosensors will aid in the clarification of the roles of these and other effects on the generation of spatiotemporally complex Ca 2ϩ signals.