Portal de Periódicos da CAPES

Genetically Encoded Indicators of Cellular Calcium Dynamics Based on Troponin C and Green Fluorescent Protein

2004; Elsevier BV; Volume: 279; Issue: 14 Linguagem: Inglês

10.1074/jbc.m312751200

1083-351X

Nicola Heim, Oliver Griesbeck,

RNA Research and Splicing

Genetic calcium probes offer tremendous potential in the fields of neuroscience, cell biology, and pharmaceutical screening. Previously, ratiometric and non-ratiometric indicators of cellular calcium dynamics have been described that consist of mutants of the green fluorescent protein (GFP) as fluorophores and calmodulin as calcium-binding moiety in several configurations. However, these calmodulin-based types of probes have a series of deficiencies, such as reduced dynamic ranges, when expressed within transgenic organisms and lack of calcium sensitivity in certain targetings. We developed novel types of calcium probes based on troponin C variants from skeletal and cardiac muscle. These indicators have ratio changes up to 140%, Kds ranging from 470 nm to 29 μm, and improved subcellular targeting properties. We targeted the indicators to the plasma membrane of HEK293 cells and primary hippocampal neurons. Upon long lasting depolarization, submembrane calcium levels in hippocampal neurons were found to be in equilibrium with bulk cytosolic calcium levels, suggesting no standing gradient persists from the membrane toward the cytosol. We expect that such novel indicators using specialized calcium sensing proteins will be minimally interacting with the cellular biochemical machinery. Genetic calcium probes offer tremendous potential in the fields of neuroscience, cell biology, and pharmaceutical screening. Previously, ratiometric and non-ratiometric indicators of cellular calcium dynamics have been described that consist of mutants of the green fluorescent protein (GFP) as fluorophores and calmodulin as calcium-binding moiety in several configurations. However, these calmodulin-based types of probes have a series of deficiencies, such as reduced dynamic ranges, when expressed within transgenic organisms and lack of calcium sensitivity in certain targetings. We developed novel types of calcium probes based on troponin C variants from skeletal and cardiac muscle. These indicators have ratio changes up to 140%, Kds ranging from 470 nm to 29 μm, and improved subcellular targeting properties. We targeted the indicators to the plasma membrane of HEK293 cells and primary hippocampal neurons. Upon long lasting depolarization, submembrane calcium levels in hippocampal neurons were found to be in equilibrium with bulk cytosolic calcium levels, suggesting no standing gradient persists from the membrane toward the cytosol. We expect that such novel indicators using specialized calcium sensing proteins will be minimally interacting with the cellular biochemical machinery. Genetically encoded fluorescent indicators used to visualize cellular calcium levels have many advantages over other fluorescent dyes that have to be applied externally. They are generated in situ inside cells after transfection, do not require cofactors, can be specifically targeted to cell organelles and cellular microenvironments, and do not leak out of cells during longer recording sessions. Furthermore, they can be expressed within intact tissues of transgenic organisms and thus solve the problem of loading an indicator dye into tissue, while allowing the labeling of specific subsets of cells of interest (for review, see Ref. 1Zhang J. Campbell R.E. Ting A.Y. Tsien R.Y. Nat. Rev. Mol. Biol. 2002; 3: 906-918Google Scholar). Two classes of green fluorescent protein (GFP) 1The abbreviations used are: GFP, green fluorescent protein; BAPTA, tetrapotassium salt; FRET, fluorescence resonance energy transfer; CFP, cyan fluorescent protein; HEK, human embryonic kidney; MOPS, 4-morpholinepropanesulfonic acid; csTnC, chicken skeletal muscle troponin C. -based calcium indicators have been described so far: (i) ratiometric indicators (termed "cameleons") consisting of a pair of fluorescent proteins engineered for fluorescence resonance energy transfer (FRET) which carry the calcium-binding protein calmodulin as well as a calmodulin target peptide sandwiched between the GFPs (2Miyawaki A. Llopis J. Heim R. McCaffery J.M. Adams J.A. Ikura M. Tsien R.Y. Nature. 1997; 388: 882-887Google Scholar, 3Miyawaki A. Griesbeck O. Heim R. Tsien R.Y. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 2135-2140Google Scholar, 4Truong K. Sawano A. Mizuno H. Hama H. Tong K.I. Mal T.K. Miyawaki A. Ikura M. Nat. Struct. Biol. 2001; 8: 1069-1073Google Scholar); and (ii) various non-ratiometric indicators with calmodulin directly inserted into a single fluorescent protein (5Baird G.S. Zacharias D.A. Tsien R.Y. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 11241-11246Google Scholar, 6Nagai T. Sawano A. Park E.S. Miyawaki A. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 3197-3202Google Scholar, 7Nakai J. Ohkura M. Imoto K. Nat. Biotechnol. 2001; 19: 137-141Google Scholar, 8Griesbeck O. Baird G.S. Campbell R.E. Zacharias D.A. Tsien R.Y. J. Biol. Chem. 2001; 276: 29188-29194Google Scholar). However, calmodulin-based indicators show deficiencies in certain applications, e.g. they display only a reduced dynamic range in transgenic invertebrates compared with in vitro data of the purified indicator proteins and acute transfections (9Reiff D.F. Thiel P.R. Schuster C.M. J. Neurosci. 2002; 22: 9399-9409Google Scholar, 10Kerr R. Lev-Ram V. Baird G.S. Vincent P. Tsien R.Y. Schafer W.R. Neuron. 2000; 26: 583-594Google Scholar, 11Fiala A. Spall T. Diegelmann S. Eisermann B. Sachse S. Devaud J.M. Buchner E. Galizia C.G. Curr. Biol. 2002; 12: 1877-1884Google Scholar) and fail to show calcium responses when targeted to certain sites within cells. No successful transgenic expression in mammals has been reported yet. Calmodulin is a ubiquitous signal protein in cell metabolism and thus under stringent regulation involving a plethora of calmodulin-binding proteins (12Jurado L.A. Chockalingam P.S. Jarrett H.W. Physiol. Rev. 1999; 79: 661-682Google Scholar). It activates numerous kinases and phosphatases, modulates ion channels (13Saimi Y. Kung C. Ann. Rev. Physiol. 2002; 64: 289-311Google Scholar), and is itself extensively phosphorylated by multiple protein serine/threonine kinases and protein tyrosine kinases (14Benaim G. Villalobo A. Eur. J. Biochem. 2002; 269: 3619-3725Google Scholar). Therefore, we explored ways of constructing new types of calcium probes with more specialized calcium-binding proteins that should be minimally influenced by the cellular regulatory protein network. Instead of calmodulin, we chose to try variants of troponin C (TnC), the calcium sensor in skeletal and cardiac muscle (15Gordon A.M. Homsher E. Regnier M. Physiol. Rev. 2000; 80: 853-924Google Scholar), and combinations of TnC and parts of its physiological binding partner troponin I (16Mercier P. Li M.X. Sykes B.D. Biochemistry. 2000; 39: 2902-2911Google Scholar) as linker proteins in ratiometric indicator types. Troponin C is a well characterized dumbbell-shaped calcium-binding protein with two globular domains connected by a central linker resembling calmodulin in its overall conformation (17Vassylyev D.G. Takeda S. Wakatsuki S. Maeda K. Maeda Y. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 4847-4852Google Scholar). Here, we describe novel types of calcium probes that are based on troponin C and mutants of the green fluorescent protein and demonstrate their usefulness for dynamic imaging within live cells. Gene Construction—Full-length and truncated troponin C domains were obtained by PCR from the wild-type gene sequence of chicken skeletal muscle troponin C and the human cardiac muscle gene sequence in which the intrinsic SacI site was deleted by a silent mutation of the Glu135 codon (GAG → GAA). By using primers containing an SphI site at the 5′ end and a SacI site at the 3′ end, the troponin C DNA fragments were inserted between cyan fluorescent protein (CFP) and citrine in the bacterial expression vector pRSETB (Invitrogen). TnI fusions with TnC were created by inserting TnI DNA fragments carrying either an SphI site or an SacI site at both ends at the existing SphI or SacI sites in the indicator fusion constructs. To alter calcium affinities of single EF-hands of TN-L15, point mutations were introduced into the troponin C sequence by site-directed mutagenesis using the primer extension method (QuikChange, Stratagene). For protein expression in mammalian cells, an optimized Kozak consensus sequence (GCC GCC ACC ATG G) was introduced by PCR at the 5′ end of CFP; the entire indicator fragments obtained by BamHI/EcoRI restriction of the pRSETB constructs were then subcloned into the mammalian expression vector pcDNA3 (Invitrogen). Membrane targeting of TN-L15 was achieved by fusing the 20 amino acid sequence KLNPPDESGPGCMSCKCVLS of the c-Ha-Ras membrane-anchoring signal to the C terminus of citrine, resulting in the construct TN-L15-Ras. Alternatively, the N-terminal 20 amino acids of GAP-43 (the sequence MLCCMRRTKQVEKNDEDQKI) were used for membrane-targeting of indicators. Protein Expression, in Vitro Spectroscopy, and Titrations—Proteins were expressed in Escherichia coli BL21 and purified as described previously (5Baird G.S. Zacharias D.A. Tsien R.Y. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 11241-11246Google Scholar). In vitro fluorescence measurements were performed in a Cary Eclipse fluorometer (Varian) equipped with a stopped flow RX2000 rapid kinetics accessory unit for kinetic measurements (Applied Photophysics). Calcium titrations were done with premixed calcium buffers (calcium calibration kit with magnesium #1, calcium calibration kit #3 to which magnesium was added when necessary; Molecular Probes). Magnesium titrations were done in 10 mm MOPS, pH 7.0, 100 mm KCl, and varying amounts of MgCl2. For measurements of dissociation kinetics, 6 μm purified protein in 10 mm MOPS, pH 7, 200 mm KCl, 1 mm tetrapotassium salt (BAPTA), 1 mm free Mg2+, and 1 or 50 μm free Ca2+ (TN-L15 D107A: 50 μm or 300 μm free Ca2+) were mixed with 20 mm BAPTA (TN-L15 D107A: 35 mm BAPTA) in 10 mm MOPS, pH 7, 200 mm KCl, and 1 mm free Mg2+; mixing dead time was 8 ms. Identical dissociation constants were obtained independently of calcium concentration (data not shown). Samples were excited at 432 nm and emission was monitored at 528 nm. Data sets from at least five experiments were averaged, and rate constants were derived from mono-exponential curve fittings. Cell Culture and Imaging—HEK293 cells were transfected with Lipofectin reagent (Invitrogen) and imaged on a Zeiss Axiovert 35M microscope with a charge-coupled device camera (CoolSnap, Roper Scientific). Hippocampal neurons were prepared from 17-day-old rat embryos, transfected by calcium phosphate precipitation 1-2 weeks after preparation, and imaged 1-3 days after transfection. The imaging setup was controlled by Metafluor version 4.6 software (Universal Imaging). For ratio imaging, a 440/20 excitation filter, a 455 dichroic long-pass mirror, and two emission filters (485/35 for CFP, 535/25 for citrine) operated in a filter wheel (Sutter Instruments) were used. Ratiometric Indicators with New Calcium-binding Moieties—It was not straightforward to generate new indicators using troponin C variants. First, fragments tested resulted in minimal calcium sensitivity. Therefore, we had to go through a series of optimizations in which individual amino acids at the linking sequences close to the GFPs had to be exchanged or deleted. Overall, more than 70 different constructs were made, and the proteins were purified and tested individually for their calcium sensitivity. All of our troponin fragments were sandwiched between CFP as donor and citrine (8Griesbeck O. Baird G.S. Campbell R.E. Zacharias D.A. Tsien R.Y. J. Biol. Chem. 2001; 276: 29188-29194Google Scholar) as acceptor fluorophore (Fig. 1). Calcium binding to troponin C leads to a conformational change in the protein, thereby enhancing the FRET from CFP to citrine. The ratio of acceptor to donor emission was used as a measure of the free ligand concentration. We first constructed a set of new indicators using the fast chicken skeletal muscle troponin C (csTnC) (18Reinach C.F. Karlsson R. J. Biol. Chem. 1988; 263: 2371-2376Google Scholar). As first constructs with simple insertion of the full-length gene yielded an indicator of modest performance, we tested a series of mutations and deletions at the linker regions. After consecutively deleting 14 amino acids from the N terminus of csTnC, we arrived at TN-L15, an indicator that showed a 140% ratio change in cuvettes with no magnesium present. Insertion of the human cardiac troponin C between CFP and citrine resulted in an indicator of a maximally 120% ratio change (termed TN-humTnC). In contrast to TN-L15, further attempts to improve the performance of TN-humTnC by engineering only led to indicators of slightly worse performance. In addition to the full-length sequences of humcTnC and csTnC and versions thereof with truncations and modified linkers, shorter csTnC domains were engineered in which only specific structural elements of the protein were used individually, such as the N-terminal regulatory lobe (amino acids 1-90, termed TnC-N90) of TnC alone. In our efforts to construct potentially minimally perturbing indicators, we also tested whether the conformational change within individual EF-hand motifs of TnC would alone be sufficient to allow the construction of indicators with satisfying performance. Unfortunately, such constructs (termed TnC-EF1-4 for the individual EF hands of TnC) never exceeded ratio changes of 5-20%. Because TnI is known to form a complex with TnC in vivo, and because some of these interactions are modified by calcium (16Mercier P. Li M.X. Sykes B.D. Biochemistry. 2000; 39: 2902-2911Google Scholar), peptide sequences of csTnI considered to be responsible for binding to the N- and C-terminal csTnC domains were selected according to the literature (17Vassylyev D.G. Takeda S. Wakatsuki S. Maeda K. Maeda Y. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 4847-4852Google Scholar). We hoped that such additional binding peptides could enhance FRET efficiency and the dynamic range of the indicators. This way, we created constructs in which either region 1-48 of TnI was fused to the C terminus of TnC, or regions 95-133 and 96-116 of TnI were fused to the N terminus of TnC. The fusions were connected either directly or were separated by the different spacer sequences GG, GSG, or GGSGG. Such constructs, however, never exceeded the dynamic range of the indicators without TnI sequences. A summary of basic constructs with modest to good performance as well as a basic evaluation of their function can be seen in Fig. 1. Among these constructs, the two with remarkable changes in their emission spectrum after addition of calcium, TN-L15 and TN-humTnC, were selected for further characterization, together with a series of mutants of TN-L15 in which individual calcium-binding sites were inactivated by exchange of single amino acids within the EF-hand motifs. Analogously to the related "cameleons," more convenient names for laboratory use of these indicators could be "S-troponeon" (S for skeletal) for TN-L15 and "C-troponeon" (C for cardiac) for TN-humTnC. In Vitro Properties of TN-L15 and TN-humTnC—Next, we set out to investigate further the calcium-binding properties of the chosen constructs. Examples of the emission spectrum of TN-L15 and TN-humTnC at zero and saturating calcium levels are shown in Fig. 2, A and B, respectively. The C-terminal domain of TnC is known to have two high-affinity calcium-binding sites that also bind magnesium. The N-terminal lobe binds calcium specifically with a somewhat lower affinity (19Cheung H.C. Dong W.J. Lakowicz J.R. Topics in Fluorescence Spectroscopy - Volume 6: Protein Fluorescence. Kluwer Academic/Plenum Publishers, New York2000: 257-283Google Scholar). In agreement with this, addition of 1 mm magnesium reduced the maximal dynamic range of TN-L15 and TN-humTnC obtainable by the addition of calcium from 140 to 100% and 120 to 70%, respectively. Determination of Kd values for calcium was performed by Ca2+ titrations in the presence of 1 mm-free Mg2+. Calcium titrations resulted in response curves with apparent dissociation constants of 470 nm for TN-humcTnC and 1.2 μm for TN-L15 (Fig. 2C). Kds for magnesium binding were 2.2 mm and 0.5 mm for TN-L15 and TN-humTnC, respectively. Site-directed mutagenesis has been used extensively to study ligand-binding properties and conformational change within troponin C. Therefore, we inactivated individual EF-hands systematically by exchanging crucial aspartate or glutamate residues within the binding loops with either alanine or glutamine (20Sorensen M.M. da Silva A.C.R. Gouveia C.S. Sousa V.P. Oshima W. Ferro J.A. Reinach F.C. J. Biol. Chem. 1995; 270: 9770-9777Google Scholar, 21Pearlstone J.R. Chandra M. Sorensen M.M. Smillie L.B. J. Biol. Chem. 2000; 275: 35106-35115Google Scholar). The mutation D107A (20Sorensen M.M. da Silva A.C.R. Gouveia C.S. Sousa V.P. Oshima W. Ferro J.A. Reinach F.C. J. Biol. Chem. 1995; 270: 9770-9777Google Scholar), by which the third, C-terminal EF-hand was inactivated within TN-L15, resulted in an indicator with reduced calcium affinity, in agreement with published results. The apparent calcium Kd of this construct was determined to be 29 μm. As a consequence, the response curve in calcium titrations was significantly shifted to the right (Fig. 2C). Therefore, this mutant seems to be more suitable for measuring larger changes in calcium that can be encountered, for example, when targeting indicators to synaptic sites or in close vicinity to channels. Magnesium binding to the D107A mutant was also significantly reduced to a Kd of 5 mm. Inactivation of both C-terminal binding sites of the TN-L15 construct by the double mutation D107A/D143A did not yield a functional protein in our hands. For comparison, however, inactivating both N-terminal sites by the double mutation E42Q/E78Q yielded a protein that left only the C-terminal high-affinity components intact, resulting in a Kd for calcium of 300 nm (Fig. 2C). As calcium binding to EF-hands is known to be pH-dependent, we also investigated to what extent pH changes affected the ratios obtained at zero calcium or calcium saturation. As expected, ratios were dependent upon pH (Fig. 2D). Ratios started to drop beginning below pH 6.8, thus reflecting the pH-properties of citrine and CFP (8Griesbeck O. Baird G.S. Campbell R.E. Zacharias D.A. Tsien R.Y. J. Biol. Chem. 2001; 276: 29188-29194Google Scholar). In the physiological range of cytosolic pH fluctuations between pH 6.8-7.3, the ratios were, however, remarkably stable. pH-dependent differences in calcium binding to TnC versus calmodulin were thus not observed. pH-resistance of our probes is a clear advantage over recent non-ratiometric probes based on calmodulin and a single GFP as fluorophore (5Baird G.S. Zacharias D.A. Tsien R.Y. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 11241-11246Google Scholar, 6Nagai T. Sawano A. Park E.S. Miyawaki A. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 3197-3202Google Scholar, 7Nakai J. Ohkura M. Imoto K. Nat. Biotechnol. 2001; 19: 137-141Google Scholar, 8Griesbeck O. Baird G.S. Campbell R.E. Zacharias D.A. Tsien R.Y. J. Biol. Chem. 2001; 276: 29188-29194Google Scholar), as these probes are intrinsically sensitive to pH changes and, therefore, are artifact-prone even when expressed in the cytosol. We were also interested in the binding kinetics of the indicators. Binding of calcium to TnC, especially to the N-terminal sites, is considered to be diffusion limited (∼108m-1 s-1), with the corresponding conformational change occurring concomitantly or with a slight delay in cardiac troponin C (15Gordon A.M. Homsher E. Regnier M. Physiol. Rev. 2000; 80: 853-924Google Scholar). In our experience, on-rates of genetically encoded calcium probes seemed not to be a problem in experiments. However, slow off-rates are the main obstacle to follow fast changing signals. Therefore, we focused on measuring the dissociation rates of calcium bound to our indicator proteins. As expected, for first-order reaction kinetics, these rates were independent of the chosen calcium concentration. The τ values obtained from the three constructs were 860 ms for TN-L15, 580 ms for TN-L15 D107A, and 560 ms for TN-humTnC (Fig. 2E). In comparison with our proteins, yellow cameleon 2.3 displayed a τ value of 870 ms (Fig. 2E). Ca2+ Imaging in HEK293 Cells and Rat Hippocampal Neurons—Constructs of the indicators with optimized Kozak consensus sequences for initiation of translation were expressed in HEK293 cells. Troponin C is a part of the troponin complex and usually not expressed as an isolated protein within the cytosol. Therefore, it was satisfying to see that our indicators showed good cytosolic expression. Fluorescence was distributed evenly and homogenously within the cytosol with no signs of aggregation (Fig. 3, A and G). The nucleus was excluded as expected for proteins with molecular masses of 69.7 and 72.5 kDa, respectively, for TN-L15 and TN-humTnC. To examine the function of the indicators inside cells, we used the carbachol response of HEK293 cells that can be stimulated by means of muscarinic receptors. Responses of HEK293 cells expressing TN-L15 after stimulations with 100 μm carbachol can be seen in Fig. 3. Ratios (Fig. 3B) and intensity changes of the individual wavelengths (Fig. 3C) are depicted for two cells expressing different levels of the probe. In good agreement with the in vitro properties of the indicator, carbachol-induced oscillations of cellular-free calcium were readily imaged, with repeated cycles of reciprocal intensity changes of CFP and citrine. Imaging turned out to be dynamic and reproducible, and it was no problem to obtain Rmax and Rmin. TN-L15 was also functional in primary cultures of rat hippocampal neurons. Spontaneous activity of hippocampal neurons was imaged using TN-L15, demonstrating its use in detecting endogenous activity patterns (Fig. 3, D and E). Responses to high potassium or glutamate stimulation were also readily detected (data not shown). A response of HEK293 cells expressing TN-humTnC is shown in Fig. 3F. Maximal ratio changes within cells were 100% for TN-L15 and 70% for TN-humTnC, in accordance with the in vitro values of the indicators. For comparison, the maximal ratio change obtainable with yellow chameleon 2.1 on our set-up was 70% (data not shown). Subcellular Targeting Properties of TN-L15—We next set out to evaluate the targeting properties of our new indicators within cells. In principle, one great potential of genetic probes is that they can be targeted to cellular organelles and microenvironments with the precision of molecular biology. Subsequently, we generated a series of subcellular targetings using TN-L15 and compared them to the otherwise analogous calmodulin-based cameleons. Constructs were made in the identical manner and differed only in the calcium-binding moiety. In all cases tested so far, TN-L15 retained activity whereas the corresponding calmodulin-based probes were significantly modified in dynamic range (data not shown and Fig. 4). One very attractive cellular site to us seemed to be the plasma membrane, as submembrane calcium domains play a crucial role in a series of biological phenomena ranging from secretion, ion channel modulation, to signal transduction, yet are beyond the resolution of conventional light microscopy. Apart from these immediate applications, membrane-targeting pathways are also often involved in other labelings such as localizing the indicator to pre- or postsynaptic structures or fusions to the pores of calcium channels. Although membrane targetings of the calmodulin-based indicators ratiometric perikams and yellow chameleons have been described before (22Pinton P. Tsuboi T. Ainscow E.K. Pozzan T. Rizzuto R. Rutter G.A. J. Biol. Chem. 2002; 277: 37702-37710Google Scholar, 23Issiki M. Ying Y.S. Fujita T. Anderson R.G.W. J. Biol. Chem. 2002; 277: 43389-43398Google Scholar), we were not able to put them to use successfully, as in our hands, cameleons had lost calcium sensitivity when targeted to the membrane (Fig. 4), and perikams were not considered preferential for primary neurons, as these cells show substantial cytosolic pH-fluctuations after stimulation (24Chesler M. Kaila K. Trends Neurosci. 1992; 15: 396-402Google Scholar). Our TnC-based indicators turned out to be superior under these conditions. Fig. 4 shows a comparison of membrane targeting of TN-L15 and yellow cameleons using several membrane-targeting sequences. All constructs were made in the identical manner and differed only in their calcium-binding moiety. In Fig. 4, A and B, the membrane targeting sequence of GAP43 was used. The 20 N-terminal amino acid residues of GAP43 were added in the identical manner to the N terminus of TN-L15 or YC2.1. The functionality of these constructs was tested in HEK293 cells. Imaging with GAP43-TN-L15 revealed long lasting calcium oscillations after stimulation with carbachol. As expected, the submembrane calcium oscillations in these cells were very similar to those seen with a cytosolic indicator. Finally, calibration with ionomycin/10 mm CaCl2 and ionomycin/20 μm EGTA to obtain Rmax and Rmin verified that the indicator had its full dynamic range and full functionality when targeted to the plasma membrane (Fig. 4A). In contrast, GAP43-YC2.1 performed poorly under identical conditions (Fig. 4B). No oscillations were detectable, and also, calibration with ionomycin indicated a reduced dynamic range, suggesting that the indicator had lost significant features of its calcium-binding properties on the pathway to membrane insertion. The trace shown is representative of seven experiments, none of which gave a response. Similar results were obtained by fusing the indicators to the presynaptic protein synaptobrevin (Fig. 4, C and D) and the C-terminal membrane targeting sequence of Ha-Ras (25Aronheim A. Engelberg D. Li N. al-Alawi N. Schlessinger J. Karin M. Cell. 1994; 78: 949-961Google Scholar) (Fig. 4, E-G). In the case of TN-L15-synaptobrevin, we saw a decrease of the dynamic range of the indicator, which we also found when expressing and purifying the protein from bacteria. Therefore, we attribute this fact mostly to sterical problems within the indicator protein. Interestingly, YC3.3Ras was fully calcium sensitive when purified from bacteria and tested in vitro, suggesting that the addition of the targeting sequence did not result in steric hindrance within this construct. YC3.3Ras also appeared to fold and target correctly as it showed similar membrane staining as TN-L15Ras in HEK293 cells, and both CFP and YFP were expressed and folded properly (data not shown). For evenly labeled membranes of hippocampal neurons, we used the membrane anchor sequence of c-Ha-Ras (25Aronheim A. Engelberg D. Li N. al-Alawi N. Schlessinger J. Karin M. Cell. 1994; 78: 949-961Google Scholar). Both in HEK293 cells and hippocampal neurons, clear and homogenous membrane labeling was evident (Figs. 4E and 5C). For imaging, we defined small regions following the contours of the membrane. We were interested in determining whether long lasting submembrane calcium domains could exist in primary neurons that are not in equilibrium with bulk cytosolic transients. Such domains were detected in the rat smooth muscle-derived cell line A7r5, in which global long lasting submembrane calcium domains were detected that reached peak values as high as 100 μm (26Marsault R. Murgia M. Pozzan T. Rizzuto R. EMBO J. 2001; 16: 1575-1581Google Scholar). Interestingly, also under resting conditions, submembrane calcium seemed to be elevated in these cells (26Marsault R. Murgia M. Pozzan T. Rizzuto R. EMBO J. 2001; 16: 1575-1581Google Scholar). In primary neurons, resting submembrane calcium levels were not found to be elevated (Fig. 5B). However, TN-L15-Ras came close to saturation after stimulation with high potassium (Fig. 5B). Therefore, we turned to using the low-affinity version TN-L15D107ARas which, with a Kd of 29 μm, seemed more suitable to quantify submembrane calcium rises in these cells and compare them to bulk cytosolic transients. Using this probe, glutamate or high potassium stimulation elicited average ratio changes of about 19% (n = 34) (Fig. 5D). The probe was not saturated, as calibration with ionomycin/10 mm calcium resulted in a further increase in ratio to the maximum obtainable with this probe. Average cytosolic calcium transients, however, were not significantly lower on this time scale (data not shown), suggesting that no permanent gradient in calcium concentration from the membrane toward the cytosol is maintained in these cells under these conditions. It will be interesting to address these questions further with improved temporal resolution. In this report, we characterize a novel type of ratiometric calcium indicator and demonstrate its use for dynamic imaging within live cells. These indicators work in targetings in which chameleons failed and should be more compatible with transgenic expression within whole organisms. Instead of calmodulin, we incorporated a specialized calcium-binding protein that has no other known function than regulating muscle contraction and thus should minimally interfere with cellular biochemistry. It was also fortuitous to obtain functional indicators without the need of additional binding peptides such as the calmodulin-binding peptide M13 from myosin light-chain kinase that is present in chameleons and other non-ratiometric indicators and that could, in principle, be bound by endogenous calmodulin when expressed within cells. Genetically encoded indicators have the great advantage that they can be targeted to sites of interest with the means of molecular biology. The success rate of obtaining functional targetings was, however, relatively low with existing calcium indicators. Successful functional targetings have, for example, been achieved to the endoplasmic reticulum (2Miyawaki A. Llopis J. Heim R. McCaffery J.M. Adams J.A. Ikura M. Tsien R.Y. Nature. 1997; 388: 882-887Google Scholar), the nucleus (2Miyawaki A. Llopis J. Heim R. McCaffery J.M. Adams J.A. Ikura M. Tsien R.Y. Nature. 1997; 388: 882-887Google Scholar), the Golgi (8Griesbeck O. Baird G.S. Campbell R.E. Zacharias D.A. Tsien R.Y. J. Biol. Chem. 2001; 276: 29188-29194Google Scholar), or mitochondria (6Nagai T. Sawano A. Park E.S. Miyawaki A. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 3197-3202Google Scholar). Many potentially very interesting targetings have so far not been reported, but certainly not for the lack of trying: for example, fusing indicators to the diverse array of calcium channels to monitor free calcium directly at the channel pore, targeting probes to the active zone of presynaptic terminals, or simply using genetically encoded calcium probes as activity sensors to monitor activity at the level of individual synapses. We do not know at present what type of modification restricts the calcium sensitivity of the calmodulin moiety in many targetings. It could be post-translational modification such as phosphorylation on the pathway to membrane insertion, interaction with calmodulin-binding proteins that have distinct localization patterns within cells, or simply binding by endogenous calmodulin, which may be present at extremely high concentrations under the membrane and especially in the vicinity of channels and specialized areas such as synapses. TnC-based indicators thus may complement the existing calmodulin-based indicators, as they promise to have a higher success rate in obtaining functional targetings when calmodulin-based indicators fail. It remains to be seen whether transgenic expression of TnC-based probes, especially in mammals, also solves some of the problems that may arise from the use of the highly regulated calmodulin within indicator proteins. Human cardiac troponin C is also a direct drug target for calcium sensitizing agents that modulate the calcium-binding behavior of cardiac TnC and strengthen the contraction of the human heart (27Pollessello P. Ovaska M. Kaivola J. Tilgmann C. Lundström K. Kalkinnen N. Ulmanen I. Nissinen E. Taskinen J. J. Biol. Chem. 1994; 269: 28584-28590Google Scholar). TN-humTnC may thus also become useful to design simple in vitro assays to identify small chemical compounds or polypeptides of clinical interest that act as calcium-sensitizing agents, in particular, for certain clinical settings, such as congenital heart failure. Genetically encoded indicators have tremendous potential in several areas of the life sciences, ranging from high throughput screening to cell biology to neurophysiological applications. They offer a means to implant an indicator deeply inside mature living tissue and thus complement new microscopy techniques such as multiphoton microscopy that aim at noninvasive imaging of physiological processes. Transgenic expression of our TnC-based indicators may be used in such diverse areas as elucidating sites of drug action within whole tissues to monitoring neuronal activity within the intact brain. Such functional genetically encoded optical probes of neuronal activity are urgently needed in cellular as well as systems neuroscience. We thank Drs. L. B. Smillie, R. J. Solaro, and K. Hastings for providing cDNA coding for troponin subunits. N. Hillen and O. Zapata Hommer provided technical assistance.