Detection in Living Cells of Ca2+-dependent Changes in the Fluorescence Emission of an Indicator Composed of Two Green Fluorescent Protein Variants Linked by a Calmodulin-binding Sequence

We have designed a novel fluorescent indicator composed of two green fluorescent protein variants joined by the calmodulin-binding domain from smooth muscle myosin light chain kinase. When (Ca2+) 4-calmodulin is bound to the indicator (K d = 0. 4 nm), fluorescence resonance energy transfer between the two fluorophores is attenuated; the ratio of the fluorescence intensity measured at 505 nm to the intensity measured at 440 nm decreases 6-fold. Images of microinjected living cells demonstrate that emission ratios can be used to monitor spatio-temporal changes in the fluorescence of the indicator. Changes in indicator fluorescence in these cells are coupled with no discernible lag (<1 s) to changes in the cytosolic free Ca2+ ion concentration, ranging from below 50 nm to ∼1 μm. This observation suggests that the activity of a calmodulin target with a typical 1 nmaffinity for (Ca2+) 4-calmodulin is responsive to changes in the intracellular Ca2+ concentration over the physiological range. It is likely that the indicator we describe can be modified to detect the levels of ligands and proteins in the cell other than calmodulin. We have designed a novel fluorescent indicator composed of two green fluorescent protein variants joined by the calmodulin-binding domain from smooth muscle myosin light chain kinase. When (Ca2+) 4-calmodulin is bound to the indicator (K d = 0. 4 nm), fluorescence resonance energy transfer between the two fluorophores is attenuated; the ratio of the fluorescence intensity measured at 505 nm to the intensity measured at 440 nm decreases 6-fold. Images of microinjected living cells demonstrate that emission ratios can be used to monitor spatio-temporal changes in the fluorescence of the indicator. Changes in indicator fluorescence in these cells are coupled with no discernible lag (<1 s) to changes in the cytosolic free Ca2+ ion concentration, ranging from below 50 nm to ∼1 μm. This observation suggests that the activity of a calmodulin target with a typical 1 nmaffinity for (Ca2+) 4-calmodulin is responsive to changes in the intracellular Ca2+ concentration over the physiological range. It is likely that the indicator we describe can be modified to detect the levels of ligands and proteins in the cell other than calmodulin. The Ca2+-binding protein calmodulin (CaM) 1The abbreviations used are: CaM, calmodulin; BAPTA, 1, 2-bis (2-aminophenoxy) ethane-N, N, N′, N′-tetraacetic acid; TRH, thyrotropin-releasing hormone; FRET, fluorescence resonance energy transfer; GFP, green fluorescent protein; BGFP, blue-shifted GFP; RGFP, red-shifted GFP; FIP, fluorescent indicator protein; FIP-CBSM, a fluorescent indicator protein in which BGFP and RGFP domains are joined by the CaM-binding domain from smooth muscle myosin light chain kinase. is a key transducer of intracellular Ca2+ ion signals, largely through its Ca2+-dependent activation of many enzyme activities (1Means A. R. VanBerkum M. F. Bagchi I. Lu K. P. Rasmussen C. D. Pharmacol. Ther. 1991; 50: 255-270Crossref PubMed Scopus (198) Google Scholar, 2Manalan A. S. Klee C. B. Adv. Cyclic Nucleotide Protein Phosphorylation Res. 1984; 18: 227-278PubMed Google Scholar, 3Lu K. P. Means A. R. Endocr. Rev. 1993; 14: 40-58Crossref PubMed Scopus (272) Google Scholar, 4Vanaman T. C. Klee C. B. Adv. Protein Chem. 1982; 35: 213-321Crossref PubMed Scopus (738) Google Scholar). Yet little is known about the extent and kinetics of enzyme activation in vivo, mainly because of the difficulty of directly monitoring target activation in the cell. Mitra et al. (5Mitra R. D. Silva C. M. Youvan D. C. Gene (Amst. ). 1996; 173: 13-17Crossref PubMed Scopus (264) Google Scholar) have recently reported that changes in fluorescence resonance energy transfer (FRET) between variants of green fluorescent protein (GFP) can be used to monitor cleavage at a protease site within a linker amino acid sequence. We have designed a similar fluorescent indicator protein in which the GFP variants are linked by a CaM-binding sequence. This indicator exhibits a large CaM-dependent change in its fluorescence emission due to disruption of FRET when calmodulin is bound to the linker sequence. This response can be monitored in living cells, where it closely follows changes in the intracellular Ca2+ concentration. The vector for expression of FIP-CBSM is similar to the one described by Mitra et al. (5Mitra R. D. Silva C. M. Youvan D. C. Gene (Amst. ). 1996; 173: 13-17Crossref PubMed Scopus (264) Google Scholar). The coding sequences for the BGFP (6Heim R. Prasher D. C. Tsien R. Y. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 12501-12504Crossref PubMed Scopus (1506) Google Scholar) and RGFP (7Delegrave S. Hawtin R. E. Silva C. M. Yang M. M. Youvan D. C. Bio/Technology. 1995; 13: 151-154PubMed Google Scholar) domains were produced by amplifying the GFP-encoding sequences in the BioBlue™ and BioYellow™ vectors obtained from PharMingen, Inc. (San Diego, CA). The vector pETIC, encoding a fluorescent indicator protein (FIP) control consisting of RGFP and BGFP domains joined by the linker sequence, GTSSGSSTGA, was generated first. The RGFP domain in pETIC is fused to the His6-tag/thrombin/S-tag/enterokinase leader sequence derived from pET30a (Novagen, Inc. , Madison, WI). The C terminus of the BGFP domain is fused to an additional His6 sequence, also derived from pET30a. The vector pETIC-1 encodes FIP-CBSM, which is identical to the FIP control, except that the linker has the sequence: GTSSRRKWNKTGHAVRAIGRLSSTGA. Boldface type denotes the CaM-binding sequence from avian smooth muscle myosin light chain kinase (8Guerriero Jr. , V. Russo M. A. Olson N. J. Putkey J. A. Means A. R. Biochemistry. 1986; 25: 8372-8381Crossref PubMed Scopus (76) Google Scholar). For expression, pETIC and pETIC-1 were transformed intoEscherichia coli strain BL21 (DE3). Cells containing pETIC or pETIC-1 were grown at 23 °C to an A 600 of 0. 6–0. 8, and protein expression was induced by addition of isopropyl-1-thio-β-d-galactopyranoside to 0. 5 mm. After incubating at 23 °C for ∼40 h, cells were harvested. Control FIP and FIP-CBSM were purified using His6 affinity chromatography essentially as described by Mitra et al. (5Mitra R. D. Silva C. M. Youvan D. C. Gene (Amst. ). 1996; 173: 13-17Crossref PubMed Scopus (264) Google Scholar). Vertebrate CaM expressed in E. coli was purified as described previously (9Persechini A. Blumenthal D. K. Jarrett H. W. Klee C. B. Hardy D. O. Kretsinger R. H. J. Biol. Chem. 1989; 264: 8052-8058Abstract Full Text PDF PubMed Google Scholar). The concentrations of control FIP and FIP-CBSM were determined using an ε490 of 89 mm−1cm−1. Concentrations of FIP-CBSM stock solutions were verified by titration with a standard CaM solution. Fluorescence measurements were performed using a Photon Technology International (Monmouth Junction, NJ) QuantaMaster™ photon counting spectrofluorometer. Reaction volumes (3 ml) were incubated at 30 °C in a stirred cuvette. Excitation and emission slit widths were 5 nm. An excitation wavelength of 380 nm was used forin vitro measurements of FIP fluorescence. For most experiments, a buffer containing 25 mm Tris, 0. 1m NaCl, and 300 μm CaCl2, pH 7. 5, was used. For experiments in which the free Ca2+ ion concentration was varied, a buffer containing 50 mm Tris, 0. 1 m NaCl, 0. 5 mm MgCl2, and 3 mm1, 2-bis (2-amino-5, 5′-dibromophenoxy) ethane-N, N, N′, N′-tetraacetic acid, pH 7. 5, was used. Aliquots of standard CaCl2solutions were added to achieve various levels of free Ca2+ion, which were calculated using the MaxChelator program (10Bers D. Patton C. Nuccitelli R. Methods Cell Biol. 1994; 40: 3-29Crossref PubMed Scopus (497) Google Scholar). The FIP-CBSM fluorescence emission spectrum is essentially independent of the pH between 7. 0 and 8. 0, either in the presence or absence of bound (Ca2+) 4-CaM. Human embryonic kidney cells (HEK-293) stably transfected with an epitope-tagged thyrotropin-releasing hormone (TRH) receptor (11Yu, R. , and Hinkle, P. M. (1997) Mol. Pharmacol. , in press. Google Scholar) were grown on glass coverslips to 60–80% confluence, rinsed in Hank's balanced salt solution, and placed in a Sykes-Moore chamber maintained at 37 °C. Microinjections were performed on an Eppendorf Transjector 5246 equipped with a Micromanipulator 5171 using Femtotips from Eppendorf (Madison, WI). Microinjection solutions were centrifuged and filtered through 0. 2-μm nitrocellulose filters and injected at pressures of 50–100 hectopascals for 0. 1 s. Successful injections were visualized in brightfield and by observing at 530 nm the fluorescence of RGFP excited directly at 495 nm. After microinjection, cells were allowed to recover for at least 30 min. Dynamic measurements were performed using a Dage CCD72 camera and Geniisys image intensifier system (Michigan City, IN) and IMAGE-1/AT analysis software from Universal Imaging (Media, PA). Fura-2 340/380 fluorescence excitation ratios were obtained as described previously (12Nelson E. J. Hinkle P. M. Endocrinology. 1994; 135: 1084-1092Crossref PubMed Scopus (23) Google Scholar). The fluorescence emission of FIP-CBSM and control FIP excited at 380 nm was measured at 500-ms intervals using a 510-nm emission filter. Still photographs were obtained with a Cohu CCD camera (San Diego, CA) and a 440A integrator (Colorado Video, Inc. , Boulder, CO) and analyzed with Metamorph software from Universal Imaging. Cells were illuminated at 380 nm, and emitted light was collected for 10–20 s at 510 nm and then at 440 nm. We have designed a fluorescent indicator protein containing two green fluorescent protein variants, with reported fluorescence excitation and emission maxima of 382 and 448 nm (BGFP) (6Heim R. Prasher D. C. Tsien R. Y. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 12501-12504Crossref PubMed Scopus (1506) Google Scholar) and 495 and 509 nm (RGFP) (7Delegrave S. Hawtin R. E. Silva C. M. Yang M. M. Youvan D. C. Bio/Technology. 1995; 13: 151-154PubMed Google Scholar), joined by an amino acid linker containing the CaM-binding sequence from smooth muscle myosin light chain kinase (8Guerriero Jr. , V. Russo M. A. Olson N. J. Putkey J. A. Means A. R. Biochemistry. 1986; 25: 8372-8381Crossref PubMed Scopus (76) Google Scholar) (Fig. 1). We term this particular indicator FIP-CBSM. Excitation of the fluorophore in the BGFP domain at 380 nm results in fluorescence emission at 505 nm from the fluorophore in the RGFP domain due to FRET between the fluorophores. FRET is essentially eliminated when FIP-CBSM binds (Ca2+) 4-CaM, and theF 505/F 440 ratio decreases from a value of 1. 7 to a value of 0. 3 (Fig. 1). FRET between the fluorophores in a control indicator protein lacking a CaM-binding sequence is not significantly affected by (Ca2+) 4-CaM. FIP-CBSM binds (Ca2+) 4-CaM with a K d of 0. 4 nm, which is close to the 1 nm apparentK d value for the complex between (Ca2+) 4-CaM and smooth muscle myosin light chain kinase (Fig. 2, A and B) (13Adelstein R. S. Klee C. B. J. Biol. Chem. 1981; 256: 7501-7509Abstract Full Text PDF PubMed Google Scholar). Figure 2Characterization of CaM binding by FIP-CBSM. A, titration of a mixture of 8. 2 nm FIP-CBSM and 200 nm CaM with increasing concentrations of free Ca2+ ion. Data were fit to an equation of the form: F = α (L n) / (L n +K a), where F is the fractional saturation of FIP-CBSM with CaM, given in the figure as 1 − (F −F min) / (F max −F min), L is the free Ca2+ ligand concentration, K a is an apparent dissociation constant that depends upon the total amount of CaM, and n is the number of interacting sites, which equals 3. 9 for the curve shown. This indicates that (Ca2+) 4-CaM is the species bound by FIP-CBSM. F is the fluorescence measured at 505 nm, F max and F min are the values for F measured at maximal and minimal free Ca2+ ion concentrations, and α is a correction factor allowing adjustment of the maximal fraction of FIP-CBSMbound to give the best fit. α values are between 1 and 1. 1 for the curves shown in panels A and B. B, binding of (Ca2+) 4-CaM to FIP-CBSMat concentrations of 1. 4 (▪) and 1. 0 (▴) nm. The data measured at a 1 nm FIP-CBSM concentration were fit to a standard single-site kinetic model as described previously (20Persechini A. McMillan K. Leakey P. J. Biol. Chem. 1994; 269: 16148-16154Abstract Full Text PDF PubMed Google Scholar). The data measured at a 1. 4 nm FIP-CBSMconcentration were fit to an equation of the form: F = α (P t + L t +K d) − (P t +L t + K d) 2 − 4 (P t) (L t) 0. 5/2 (P t), where L t is the total CaM concentration andP t is the total FIP-CBSMconcentration. In both cases a K d value of 0. 4 nm was obtained. Emission spectra for titration of 1. 4 nm FIP-CBSM with (Ca2+) 4-CaM are shown in theinset. View Large Image Figure ViewerDownload Hi-res image Download (PPT) Purified GFP dimerizes in solution, and crystallographic data suggest that the two GFPs are in an antiparallel orientation with the chromophores ∼25 Å apart (14Ormo M. Cubitt A. B. Kallio K. Gross L. A. Tsien R. Y. Remington S. J. Science. 1996; 273: 1392-1395Crossref PubMed Scopus (1934) Google Scholar, 15Yang F. Moss L. G. Phillips G. N. Nat. Biotechnol. 1996; 14: 1246-1251Crossref PubMed Scopus (1301) Google Scholar). Molecular modeling indicates that the linker sequence between the GFP domains in FIP-CBSM is long enough to allow them to interact in a similar manner. In the structures of the complexes between (Ca2+) 4-CaM and the CaM-binding sequences in several targets, including smooth muscle myosin light chain kinase, CaM enfolds the CaM-binding sequence forming a globular structure ∼ 40 Å in diameter (16Crivici A. Ikura M. Annu. Rev. Biophys. Biomol. Struct. 1995; 24: 85-116Crossref PubMed Scopus (697) Google Scholar). Thus, when (Ca2+) 4-CaM binds to the linker in FIP-CBSM, the inter-fluorophore distance is likely to be increased from ∼25 Å to ∼65 Å (Fig. 1). Since the efficiency of FRET is proportional to 1/r 6, where ris the inter-fluorophore distance, this would explain the reduction in FRET observed when FIP-CBSM binds (Ca2+) -CaM (17Stryer L. Annu. Rev. Biochem. 1978; 47: 819-846Crossref PubMed Scopus (1978) Google Scholar). CaM-dependent changes in FRET between the fluorophores in FIP-CBSM can be monitored in living cells, providing a view of free (Ca2+) 4-CaM levels in the cell. We have microinjected FIP-CBSM, with or without equimolar CaM, into HEK-293 cells stably transfected with the Gq/11-coupled Ca2+-mobilizing receptor for TRH. We estimate an intracellular FIP-CBSM concentration in microinjected cells of 1–10 μm, similar to estimates for the intracellular concentrations of high-abundance CaM targets, such as smooth muscle myosin light chain kinase (18Tansey M. G. Luby-Phelps K. Kamm K. E. Stull J. T. J. Biol. Chem. 1994; 269: 9912-9920Abstract Full Text PDF PubMed Google Scholar). Injection of FIP-CBSMundoubtedly perturbs the balance between CaM and its targets. Coinjection of the indicator with equimolar CaM should help to restore it, but the exact mole ratio of CaM to its binding sites in the cell is unknown. The intracellular free Ca2+ ion concentration (Ca2+i) increases from below 50 nm, when cells are incubated in media containing BAPTA, to greater than 1 μm, when extracellular Ca2+ and ionomycin are added (data not shown). Corresponding fluorescence images show a clear Ca2+-dependent reduction in theF 510/F 440 ratios in cells injected either with FIP-CBSM alone or with FIP-CBSM and CaM. This indicates an increase in the fraction of FIP-CBSM bound to (Ca2+) 4-CaM (Fig. 3). A greater reduction in theF 510/F 440 ratio is evident in cells injected with both FIP-CBSM and CaM. The basal fluorescence intensity of the RGFP acceptor (Fig. 3, Cand F) is brighter in some regions of the cell than others due to the distribution of the probe within the cells. Some cells were injected in the nucleus, some into the cytoplasm, and some into both nuclear and cytoplasmic regions. FIP-CBSM remained localized to the nucleus or cytoplasm in cells, depending upon its initial site of injection. The indicator, with a molecular mass of 63. 4 kDa, would not be expected to move through nuclear pores, and such translocation was not observed. No systematic effort was made to compare the responses of the indicator in the nucleus and cytosol, although it is clear that similar responses occur in these compartments (Fig. 3). Ca2+-dependent changes in the 510/440 emission ratio measured in cells suggest microheterogeneity in the response that may be artifactual, perhaps reflecting small changes in cell shape occurring during data collection. Investigations of a possible physiological basis for the apparent microheterogeneity are clearly a high priority. To more precisely establish the kinetics and magnitude of the FIP-CBSM response to changes in Ca2+i, the F 510 of microinjected cells was measured at 500-ms intervals, while Ca2+i was manipulated (Fig. 4). Reductions in F 510 mirror increases in Ca2+i caused by addition of TRH, Ca2+, and ionomycin with no discernible lag (<1 s) (Fig. 4). Measurements using Fura-2 indicate that basal Ca2+i in these cells is below 50 nm; it increases to 200–400 nm with TRH and to greater than 1 μm with the combination of external Ca2+ and ionomycin. In cells injected with equimolar FIP-CBSM and CaM, the combination of Ca2+ and ionomycin causes a 30% reduction in F 510. This appears to represent the saturated response of the indicator, since it is reached and maintained when Ca2+i is still increasing (Fig. 4 A). Treatment of cells injected with FIP-CBSM alone with TRH causes a ∼6% reduction inF 510; the combination of Ca2+ and ionomycin causes a ∼10% reduction (Fig. 4 B). Thus, saturation of the FIP-CBSM response is not approached in these cells, indicating that the CaM concentration is limiting. The maximum fractional reduction in the F 510 of FIP-CBSM observed in cells was 30%, or about half of the maximal 65% CaM-dependent reduction inF 510 measured in vitro (Fig. 1). Given the very different experimental systems used for measurements of fluorescence in vitro and in vivo, this difference in the maximal signals appears acceptable. The amount of probe injected was kept low to minimize perturbation of Ca2+-CaM homeostasis. A consequence is that fluorescence filters with bandwidths of ∼40 nm were required to obtain an adequate fluorescence signal. In contrast, the monochromator bandwidths used forin vitro measurements were about 3 nm. Some of the small discrepancy between in vitro and in vivomeasurements of CaM-dependent changes in indicator fluorescence may therefore be attributable to the different optical systems used. Our results clearly indicate that changes in Ca2+i ranging from below 50 nm to ∼1 μm are coupled to changes in theF 510 of FIP-CBSM. This suggests that the activity of a calmodulin target with a typical 1 nmaffinity for (Ca2+) 4-calmodulin is responsive to changes in the intracellular Ca2+ concentration over the physiological range. It also suggests that physiological changes in Ca2+i are coupled to changes in the free (Ca2+) 4-CaM concentration in the low nanomolar range. The free concentrations of (Ca2+) 4-CaM occurring in the cell are therefore ∼1000-fold less than the total concentration of CaM (18Tansey M. G. Luby-Phelps K. Kamm K. E. Stull J. T. J. Biol. Chem. 1994; 269: 9912-9920Abstract Full Text PDF PubMed Google Scholar). Thus, the calmodulin concentration in the cell is limiting; essentially all the (Ca2+) 4-CaM present in the cell must be bound to targets, as has been proposed based on studies of the intracellular mobility of tagged CaM (19Gough A. H. Taylor D. L. J. Cell Biol. 1993; 121: 1095-1107Crossref PubMed Scopus (79) Google Scholar). The very low physiological levels of free (Ca2+) 4-CaM indicate that small changes in the affinity of a typical target should significantly affect its level of activity at a submaximal Ca2+i, as recently demonstrated for smooth muscle myosin light chain kinase activity (18Tansey M. G. Luby-Phelps K. Kamm K. E. Stull J. T. J. Biol. Chem. 1994; 269: 9912-9920Abstract Full Text PDF PubMed Google Scholar). The ability to monitor free (Ca2+) 4-CaM levels in living cells provides an exciting new approach for dissecting the processes that link variations in Ca2+i to diverse cellular responses. Indeed, FIP-CBSM appears to represent a new class of ligand-dependent indicators that have the potential of reporting the levels of a variety of proteins and other ligands in the cell, depending upon the nature of the linker sequence.