Highly Cooperative Dependence of Sarco/Endoplasmic Reticulum Calcium ATPase (SERCA) 2a Pump Activity on Cytosolic Calcium in Living Cells*

Sarco/endoplasmic reticulum (SR/ER) Ca2+-ATPase (SERCA) is an intracellular Ca2+ pump localized on the SR/ER membrane. The role of SERCA in refilling intracellular Ca2+ stores is pivotal for maintaining intracellular Ca2+ homeostasis, and disturbed SERCA activity causes many disease phenotypes, including heart failure, diabetes, cancer, and Alzheimer disease. Although SERCA activity has been described using a simple enzyme activity equation, the dynamics of SERCA activity in living cells is still unknown. To monitor SERCA activity in living cells, we constructed an enhanced CFP (ECFP)- and FlAsH-tagged SERCA2a, designated F-L577, which retains the ATP-dependent Ca2+ pump activity. The FRET efficiency between ECFP and FlAsH of F-L577 is dependent on the conformational state of the molecule. ER luminal Ca2+ imaging confirmed that the FRET signal changes directly reflect the Ca2+ pump activity. Dual imaging of cytosolic Ca2+ and the FRET signals of F-L577 in intact COS7 cells revealed that SERCA2a activity is coincident with the oscillatory cytosolic Ca2+ concentration changes evoked by ATP stimulation. The Ca2+ pump activity of SERCA2a in intact cells can be expressed by the Hill equation with an apparent affinity for Ca2+ of 0.41 ± 0.0095 μm and a Hill coefficient of 5.7 ± 0.73. These results indicate that in the cellular environment the Ca2+ dependence of ATPase activation is highly cooperative and that SERCA2a acts as a rapid switch to refill Ca2+ stores in living cells for shaping the intracellular Ca2+ dynamics. F-L577 will be useful for future studies on Ca2+ signaling involving SERCA2a activity.

Intracellular Ca 2ϩ plays a pivotal role in controlling numerous cellular processes such as exocytosis, gene transcription, cell proliferation, muscle contraction, and cell survival (1). The level of intracellular Ca 2ϩ is determined by the balance between the influx that introduces Ca 2ϩ into the cytoplasm and the efflux that removes it from the cytoplasm. The key molecules involved in the regulation of intracellular Ca 2ϩ such as channels, pumps, and exchangers have been identified (2). Channels in the plasma membrane and sarco/endoplasmic reticulum (SR/ER) 8 membrane are responsible for the Ca 2ϩ influx, whereas pumps and exchangers carry out the Ca 2ϩ efflux. SR/ER Ca 2ϩ -ATPase (SERCA) is a Ca 2ϩ pump that transfers Ca 2ϩ from the cytosol to the lumen of the SR/ER at the expense of ATP hydrolysis (3). SERCA functions to determine the resting level of intracellular Ca 2ϩ (4) and to control the spatiotemporal profile of Ca 2ϩ transients and the frequency of Ca 2ϩ oscillations (5). Impairment of SERCA causes Ca 2ϩ homeostatic dysfunction, resulting in several important disease states such as heart failure, hypertension, diabetes, and Alzheimer disease (6).
The SERCA family consists of three isoforms and their splicing variants (SERCA1a, -1b, -2a, -2b, -3a, -3b, and -3c). The molecular masses of SERCA isoforms range from 105 to 115 kDa. Each SERCA isoform consists of four distinct functional domains, namely the nucleotide-binding, phosphorylation, actuator, and transmembrane domains (7). The three-dimensional structures of different conformational states of SERCA1a have been defined by x-ray crystallography (8), and structural models for the catalytic cycle of SERCA are well established (9).
The sequential conformational changes of SERCA are accompanied by active transport of 2 mol of Ca 2ϩ per 1 mol of bound ATP, and 1 mol of ATP is hydrolyzed during one reaction cycle (10). Thus, the conformational changes are directly linked with SERCA activity.
Cytosolic Ca 2ϩ increases evoked by extracellular stimuli are often observed in the form of oscillating Ca 2ϩ spikes (11). In most cell types, Ca 2ϩ spikes are generated by inositol 1,4,5trisphosphate (IP 3 )-induced Ca 2ϩ release from intracellular Ca 2ϩ stores (12). The IP 3 receptor is an intracellular Ca 2ϩ release channel, and its activity is controlled by not only IP 3 but also cytosolic Ca 2ϩ (13). Because submicromolar Ca 2ϩ concentrations activate IP 3 receptor, the upstroke of Ca 2ϩ spikes may occur by regenerative Ca 2ϩ release through IP 3 receptor (14). On the other hand, the role of SERCA in the generation of Ca 2ϩ spikes is more passive. SERCA transports Ca 2ϩ back to the Ca 2ϩ stores at a rate that is solely dependent on the intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) (15). The pump activity of SERCA has been approximated by the Hill equation to have a Hill coefficient of 2 for Ca 2ϩ binding (16) in theoretical studies of Ca 2ϩ dynamics (15,17). However, it remains to be elucidated whether the enzymatic activity of SERCA in living cells is identical to that measured in vitro.
The FRET technique is commonly used to study the conformational changes of functional proteins (18), and labeling with fluorescent proteins has been widely used for this purpose (19). However, as ectopic expression and/or dysfunction of the target protein can sometimes result from the labeling with these relatively large (ϳ27 kDa) proteins (20), techniques using small fluorescent molecules with less steric bulk have been greatly advanced (21). As an example, FlAsH-EDT 2 is a small membrane-permeable synthetic ligand with high affinity and specificity for a tetracysteine tag (CCPGCC; TC tag) (22). The binding of FlAsH-EDT 2 to a TC tag induces chromophore formation, giving rise to fluorescence (23).
Real-time visualization of SERCA pump activity in living cells should facilitate our understanding of the mechanism for the generation of intracellular Ca 2ϩ dynamics. In this study, we successfully detected the pump activity of SERCA2a in living cells using the FRET technique with a combination of enhanced CFP (ECFP) and FlAsH. Dual imaging of intracellular Ca 2ϩ and the FRET signals of our ECFP-and FlAsH-tagged SERCA2a, designated F-L577, provides significant insights into the regulatory mechanism of SERCA pump activity in living cells.

EXPERIMENTAL PROCEDURES
Plasmid Constructions-The details of these procedures are given in the supplemental data.
Cell Culture and Transfection-COS7 cells were cultured in DMEM (Invitrogen) supplemented with 10% FBS, 100 units/ml penicillin and 0.1 mg/ml streptomycin, and maintained in a humidified incubator with 5% CO 2 at 37°C. The cells were grown on 35-mm poly-L-lysine-coated glass-bottomed dishes (Matsunami). F-L577 and mRFP-KDEL cDNAs were transfected into the cells using TransIT (Mirus). The cells were used for experiments at 2 days after transfection. For FRET imaging, F-L577-expressing COS7 cells were loaded with FlAsH-EDT 2 reagent (Invitrogen) for 60 -90 min at room temperature and washed twice with BAL wash buffer (Invitrogen).
Subcellular Localization of F-L577-Fluorescence images of F-L577 expressing COS7 cells were taken under a confocal scanning microscope (CSU-XI; Yokogawa) attached to an IX81 inverted microscope (Olympus) with an EMCCD camera (Cas-cadeII; Roper) and a 60ϫ (oil; numerical aperture, 1.42) objective lens (Olympus). A 440-nm excitation filter, a 520-nm emission filter, and 450-nm dichroic mirrors were inserted into the light path. Data were analyzed using MetaMorph software (Molecular Devices).
Acceptor Photobleaching-Acceptor (FlAsH) photobleaching was performed using a confocal microscope (A1; Nikon). Repeated scans (120 s at maximum rates) with an unattenuated 514-nm illumination from a multi-argon laser were applied (total, 30 milliwatt maximum output). ECFP and FlAsH emission signals were acquired with a 515-nm dichroic mirror and a pair of 460 -500-nm and 520 -620-nm band pass filters, respectively. Data were analyzed using NIS elements (Nikon) and MetaMorph software (Molecular Devices).
Generation of Recombinant Baculoviruses and Expression-Spodoptera frugiperda (Sf9) cells (Invitrogen) were cultured at 27°C with stirring in Sf-900II SFM medium (Invitrogen) containing 10% FBS. P1 viral stocks of recombinant baculoviruses for F-L577 and WT-SERCA2a were generated using a Bac-to-Bac Baculovirus Expression System (Invitrogen) according to the manufacturer's instructions. Amplification of P2 viral stocks and recombinant protein expression of F-L577 and WT-SERCA2a were performed using Sf9 cells as described previously (24,25).
In Vitro Kinetic Assay of ATP-induced Ca 2ϩ Uptake by SERCA Pump-Microsomal vesicles were prepared from Sf9 cells as described previously (26). The measurements and analysis of the ATP-induced Ca 2ϩ accumulation were conducted as described elsewhere. 9 Briefly, the volume of the microsomal vesicles used in each measurement was adjusted using the absorbance at 600 nm acquired by a DU 640 spectrophotometer (Beckman Coulter). The microsomal vesicles were suspended in 500 l of cytosol-like medium (110 mM KCl, 10 mM NaCl, 5 mM KH 2 PO 4 , 1 mM DTT, 50 mM HEPES-KOH, pH 7.2, at room temperature), supplemented with 1 g/ml oligomycin (Sigma), 2 M MgCl 2 , 25 g/ml creatine kinase (Roche Diagnostics), 10 mM phosphocreatine (Sigma), and 2 M fura-2 (Dojindo). The suspension was continuously stirred at 30°C during measurements. The fluorescent signals of fura-2 were monitored with a 9 M. Enomoto, T. Michikawa, and K. Mikoshiba, unpublished data. CAF-110 spectrofluorometer (Jasco) and a PowerLab data acquisition system (ADInstruments). Fluorescence signals were recorded every 0.01 s at an emission wavelength of 510 nm with alternate excitation at 340 and 380 nm. The 1 mM ATPinduced Ca 2ϩ uptake into microsomal vesicles by the SERCA pump was initiated and monitored for 300 s. Free Ca 2ϩ concentrations were calculated from acquired fluorescent signals as described previously (27). To estimate the rate of Ca 2ϩ uptake, curve fitting of three-exponential function was conducted using Igor Pro 6.02 software (WaveMetrics) and t1 ⁄ 2 (half-decay time) was calculated.
Cell Permeabilization and State Fixation of F-L577-FlAsH-EDT 2 -loaded F-L577-expressing COS7 cells were permeabilized with 60 M ␤-escin (Sigma) in an internal solution (19 mM NaCl, 125 mM KCl, 10 mM HEPES-KOH, pH 7.4) supplemented with 5 mM EGTA for 3-5 min. The permeabilized cells were gently washed with the internal solution containing 5 mM EGTA and then used for FRET measurements. Reagents for the fixation of F-L577 in the four major conformational states were prepared according to methods described previously in the crystallization experiments of SERCA1a (28 -31). Cells were perfused with an internal solution containing 30 M thapsigargin and 5 mM EGTA for the E2 state; 100 M CaCl 2 for the E1-Ca 2ϩ state; 1 mM ADP, 0.33 mM AlCl 3 , 5 mM NaF, 1 mM MgCl 2 , and 100 M CaCl 2 for the E1-ATP state; or 2 mM BeCl 2 , 8 mM NaF, 1 mM MgCl 2 , and 5 mM EGTA for the E2P state. The free Ca 2ϩ concentration in the internal solutions was adjusted with K 2 HEDTA and CaHEDTA at 37°C as described previously (32). Fluorescent signals were acquired at 0.25 Hz with an IX71 inverted fluorescence microscope (Olympus) (see below).
ER Ca 2ϩ Imaging-ER luminal Ca 2ϩ imaging was performed with 10 M Mag-Indo-1 AM (Molecular Probes), according to the method described previously (33,34) with some modifications. The free Ca 2ϩ concentrations in the internal solutions were carefully adjusted with a Ca 2ϩ buffer and confirmed using a Ca 2ϩ -sensitive electrode (Metrohm) prior to use (41). For FRET and Ca 2ϩ imaging, the permeabilized COS7 cells expressing F-L577 were perfused with the internal solution containing 1 M free Ca 2ϩ , and then 0.01, 0.1, or 1 mM MgATP (Sigma) was applied in the internal solution containing 1 M free Ca 2ϩ . Fluorescent signals were acquired at 0.25 Hz with an IX71 inverted microscope (Olympus) (see below).
Live Cell Imaging-For imaging of intact cells, F-L577-expressing COS7 cells were loaded with 10 M Indo-5F AM (Dojindo) and FlAsH-EDT 2 reagent. Images were obtained under a constant flow (2 ml/min) of a balanced salt solution containing 115 mM NaCl, 5.4 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , 10 mM glucose, and 20 mM HEPES-NaOH, pH 7.4. Fluorescent signals were acquired at 2 Hz with an IX71 inverted fluorescence microscope (see below). For [Ca 2ϩ ] i calibration, fluorescence imaging of F-L577 and Indo-5F was performed using the IX81 inverted microscope. [Ca 2ϩ ] i was calculated using the formula of Grynkiewicz et al. (27). After agonist stimulation (1 M ATP for 2 min), the minimum and maximum fluorescence ratio of Indo-5F was obtained by treatment of the cells with 2 M ionomycin (Br-A23187) and 5 mM EGTA followed by 2 M ionomycin and 10 mM CaCl 2 .

Construction and Characterization of Fluorescently Labeled
SERCA2a- Fig. 1A shows a schematic representation of the F-L577 protein constructed in this study to investigate the conformational changes linked with SERCA2a activity in living cells. Because the N-terminal fusion of ECFP to SERCA1a has no notable effects on its ATPase activity (35), ECFP was fused to the N terminus of the A-domain of SERCA2a (Fig. 1A). The site of the TC tag insertion was determined based on the following criteria: 1) the relative distance between the donor and acceptor should change dramatically during the reaction cycle of SERCA, and 2) the site should be located within a surface loop connecting secondary structures, such as an ␣-helix and a ␤-sheet. Fig. 1B shows a schematic representation of the conformational changes during the reaction cycle of SERCA1a (28 -31, 36). Because the relative orientation of the N-domain is markedly changed during the cycle, the penultimate loop between Leu-577 and Glu-578 in the N-domain was chosen as the TC tag insertion site (Fig. 1, A and B). Fig. 2A shows a Western blot analysis of F-L577 expressed in COS7 cells. The signal for F-L577 (lane 2) was detected with the expected size. The subcellular localization of F-L577 in the ER was confirmed by confocal microscopy (Fig. 2B). The ER was labeled by cotransfection with mRFP-KDEL (an ER-targeted monomeric red fluorescent protein) (34). The fluorescent signals of ECFP and FlAsH colocalized well with the mRFP-KDEL signals, indicating that the ECFP fusion and TC tag insertion did not disrupt the ER localization of F-L577.
To confirm that FRET occurred in living COS7 cells, we performed acceptor (FlAsH) photobleaching of F-L577. After the photobleaching of FlAsH, the ECFP (donor) fluorescence intensity was increased to 130.1 Ϯ 10.1% (mean Ϯ S.D., n ϭ 6) (Fig. 2C). This finding indicates that FRET occurred between the N-terminal ECFP and the FlAsH label in the N-domain in resting COS7 cells.
We measured the Ca 2ϩ pump activity of F-L577 in vitro (Fig.  2D). Recombinant WT-SERCA2a and F-L577 were expressed in Sf9 insect cells. ATP-induced Ca 2ϩ accumulation in the microsomal vesicles containing WT-SERCA2a or F-L577 was monitored by spectrofluorometry. A representative trace of Ca 2ϩ uptake is shown in supplemental Fig. S1. The half-decay time (t1 ⁄ 2 ) of the ATP-induced Ca 2ϩ accumulation in the microsomal vesicles prepared from cells expressing WT-SERCA2a was significantly smaller than the t1 ⁄ 2 of Ca 2ϩ accumulation in the microsomal vesicles prepared from non-transfected cells (Fig. 2D). These findings indicate that the activity of the recombinant WT-SERCA2a was successfully detected in this experimental system. Overexpression of F-L577 also significantly decreased the t1 ⁄ 2 (Fig. 2D). Based on these findings, we con-cluded that the ECFP fusion and TC tag insertion did not abolish the SERCA2 pump activity.
FRET Signals of F-L577 in Four Major Conformational States of SERCA-To examine the ability of F-L577 to reflect the various conformational changes linked to SERCA2 pump activity, we measured its FRET signals in the four major conformational states (Fig. 3A). To isolate these states, we performed state fixation of F-L577 expressed in COS7 cells. Briefly, permeabilized COS7 cells expressing F-L577 were initially perfused with an internal solution containing 5 mM EGTA to provide an arbitrary F-L577 reference state. The cells were then perfused with an internal solution containing 30 M thapsigargin and 5 mM EGTA (E2 state) (29); 100 M CaCl 2 (E1-Ca 2ϩ state) (28); 1 mM ADP, 0.33 mM AlCl 3 , 5 mM NaF, 1 mM MgCl 2 , and 100 M CaCl 2 (E1-ATP state) (30); or 2 mM BeCl 2 , 8 mM NaF, 1 mM MgCl 2 , and 5 mM EGTA (E2P state) (31). AlF 4 Ϫ and BeF 3 Ϫ are phosphate analogues that can be used to fix SERCA1a in the E1-ATP and E2P states, respectively (30,36). Supplemental Fig.  S2 shows the relationship between the FRET signals of F-L577 and the concentration of thapsigargin, as a SERCA-specific inhibitor, in permeabilized COS7 cells. The EC 50 value was ϳ20 M, which was higher than that measured in vitro (37). Therefore, a saturating concentration (30 M) of thapsigargin was used for the state fixation experiments. Fig. 3B shows representative traces of the FRET signal changes from the reference state to the four different states. The average values of the FRET signal changes (⌬R/R base ) for the four different states are shown in Fig. 3C. When the cells were fixed in the E2 state (thapsigargin-bound state), F-L577 showed a 7.3 Ϯ 1.6% (mean Ϯ S.D.)  decrease in the FRET signal relative to the baseline signal in the reference state (Fig. 3, B and C). The FRET signals in the E1-Ca 2ϩ state (Ca 2ϩ -bound state) and E1-ATP state (ATP-bound state) showed 11.9 Ϯ 3.2% and 6.2 Ϯ 1.9% decreases relative to the baseline signal, respectively (Fig. 3, B and C). In contrast, the FRET signal showed a 2.4 Ϯ 0.86% increase following the transition from the reference state to the E2P state (Fig. 3, B and C). The overall dynamic range of the FRET signal changes of F-L577 was ϳ15%.
The distances between the N terminus and Leu-577 of SERCA1a in the four major conformational states were calculated using the molecular modeling software PyMOL. The calculated distances were 39 (36). Therefore, the order of the distance between the N terminus and Leu-577 was E1-Ca 2ϩ Ͼ E2 Ͼ E1-ATP Ͼ E2P. This order was consistent with the order of the relative amounts of the FRET signal changes of SERCA2a (Fig. 3C), suggesting that the FRET signal changes contain information about the distance between the N terminus and Leu-577. The measured FRET signal changes were predictive of the defined structural changes during the cycling of SERCA, strongly suggesting that F-L577 is capable of detecting the conformational changes of SERCA2a.
FRET Signal Changes of F-L577 Directly Reflect Ca 2ϩ Pump Activity-To confirm that the FRET signal changes of F-L577 reflect the Ca 2ϩ pump activity in living cells, we performed time-lapse dual imaging of F-L577 and the ER luminal Ca 2ϩ concentration during Ca 2ϩ pumping in permeabilized cells (Fig. 4A). For ER luminal Ca 2ϩ imaging, COS7 cells were loaded with a low affinity fluorescent Ca 2ϩ indicator, Mag-Indo-1 AM (K d ϭ ϳ35 M) using a method described previously (33,34). For time-lapse imaging during Ca 2ϩ pumping, a baseline was initially established by perfusion with an internal solution containing 1 M Ca 2ϩ without MgATP. Under these ATP-free conditions, the E2 and E1-Ca 2ϩ states should dominate (Fig. 3A). To initiate Ca 2ϩ accumulation in the ER, 1 mM MgATP was added to the perfusate. Following the application of MgATP, the FRET signal of F-L577 increased by 3-6%, accompanied by an increase in the ER luminal Ca 2ϩ concentration (Fig. 4A). The increase in the FRET signal suggests that the fractions of the E1-ATP and E2P states, which showed higher FRET signals than the E2 and E1-Ca 2ϩ states (Fig.  3C), are increased when SERCA2a is activated.
To determine the timing of the FRET changes relative to Ca 2ϩ uptake, we compared the first derivatives of the two responses (Fig.  4B). The peak responses were detected at the same sampling points, indicating that the timing of the conformational changes of F-L577 was coincident with Ca 2ϩ uptake into the ER.
The Ca 2ϩ pump activity of SERCA2 is dependent on the ATP concentration (38). To measure the ATP dependence of the FRET signals of F-L577, we repeated the above experiments in the presence of 0.01, 0.1, or 1 mM ATP (Fig. 4C). The results demonstrated that the FRET signals of F-L577 showed ATP concentration-dependent changes (Fig. 4C, left) and that the ATP dependence of the FRET signals paralleled the accumulation of Ca 2ϩ into the ER lumen (Fig. 4C, right).
We further analyzed the correlation between the F-L577 FRET and Mag-Indo-1 responses (Fig. 4D). Interestingly, the F-L577 (⌬R/R base ) and Mag-Indo-1 (⌬F/F base ) responses were significantly correlated (r ϭ 0.767, n ϭ 61). We also explored the effect of the cytosolic Ca 2ϩ concentration on the F-L577 response (supplemental Fig. S3). The FRET signal changes of F-L577 and the Mag-Indo-1 signal changes were significantly correlated even when the cytosolic Ca 2ϩ concentration varied from 0.13-1.0 M (r ϭ 0.651, n ϭ 38). These results suggest that the positive FRET signal changes of F-L577 are directly correlated with the instantaneous Ca 2ϩ pump activity of SERCA2a.
We compared the Mag-Indo-1 signal changes of non-transfected cells and F-L577-expressing cells under three different conditions (Fig. 4E). When permeabilized COS7 cells were stimulated by 0.1 or 1 mM MgATP in an internal solution containing 1 M Ca 2ϩ , the Mag-Indo-1 signal change in F-L577expressing cells was significantly larger than that in non-transfected cells. These results indicate that F-L577 itself has Ca 2ϩ pump activity in living cells.
The data described thus far demonstrate that 1) F-L577 shows different FRET signals depending on the conformational state, (2) the conformational change of F-L577 is coincident with Ca 2ϩ uptake, (3) the ATP and Ca 2ϩ concentration dependences of the F-L577 FRET signal are comparable with those of Ca 2ϩ uptake, (4) the FRET signal change of F-L577 is correlated with Ca 2ϩ uptake in permeabilized cells, and (5) F-L577 itself has Ca 2ϩ pump activity in living cells. In summary, the FRET signal changes of F-L577 directly reflect the Ca 2ϩ pump activity linked with its conformational changes in the SERCA reaction cycle.
Visualization of Ca 2ϩ Pump Activities of SERCA During Ca 2ϩ Oscillations in COS7 Cells-To investigate the dynamics of SERCA activity in living cells during ATP-evoked Ca 2ϩ oscillations, we performed dual imaging of intracellular Ca 2ϩ and the FRET signal changes of F-L577 in COS7 cells. Fig. 5A shows representative FRET signal changes of F-L577 (upper panel) and cytoplasmic Ca 2ϩ concentration changes (lower panel) observed in COS7 cells stimulated with 1 M ATP. The FRET signals of F-L577 showed oscillatory dynamics during Ca 2ϩ oscillations. Although the oscillatory SERCA activity has been predictable, these findings comprise the first experimental evidence for the detection of oscillatory SERCA activity in living cells. The representative traces of the FRET signal changes of F-L577 (continuous line) and Ca 2ϩ changes (broken line) are also shown on an enlarged time scale (Fig. 5B). There was no detectable delay between the F-L577 signals and the Indo-5F signals during either the rising or falling phases of the Ca 2ϩ spikes. These findings indicate that the Ca 2ϩ pump activity of SERCA2a is synchronized with cytosolic Ca 2ϩ concentration changes without any detectable delay. Fig. 5C shows the relationship between the FRET signal changes of F-L577 and [Ca 2ϩ ] i during the period of Ca 2ϩ oscil- lations evoked with 1 M ATP. We found a good fit between the data and the Hill equation with an apparent affinity for Ca 2ϩ of 0.41 Ϯ 0.0095 M and a Hill coefficient of 5.7 Ϯ 0.73 (Fig. 5C). The apparent Ca 2ϩ sensitivity of F-L577 was consistent with previously reported values (0.34 ϳ 0.45 M) for native SERCA2a measured in vitro (15,39,40). In contrast, the high degree of cooperativity differed from that measured in vitro (16).
To determine whether the high degree of cooperativity shown in this in vivo experiment is due to overexpression of the SERCA pump or fluorescent labeling, we investigated the FRET signal change of F-L577 under the steady-state condition in [Ca 2ϩ ] i (Fig. 6). Fig. 6 shows the relationship between [Ca 2ϩ ] i in the solutions applied to permeabilized cells and the FRET signal change of F-L577. We found a good fit between the data and the Hill equation with an apparent affinity for Ca 2ϩ of 0.46 Ϯ 0.26 M and a Hill coefficient of 0.94 Ϯ 0.41 (Fig. 6). Both the apparent Ca 2ϩ sensitivity of F-L577 and the Hill coefficient were consistent with that measured in vitro (15,39,40). These results confirm that the highly cooperative dependence of the FRET signal changes of F-L577 on cytosolic Ca 2ϩ does not arise from artifacts caused by the overexpression and/or fluorescent labeling. These findings indicate that the Ca 2ϩ pump activity of SERCA2a shows a high degree of cooperativity in the physiological condition of living cells.
We assumed that the FRET signal of F-L577 was proportional to the instantaneous SERCA pump activity based on the linear relationship between the FRET signal of F-L577 and the Mag-Indo-1 signal reflecting the ER luminal Ca 2ϩ concentration ([Ca 2ϩ ] ER ) (Fig. 4, B and D). To investigate the possibility that the high degree of cooperativity is caused artificially by a nonlinear relationship between the Mag-Indo-1 signal and [Ca 2ϩ ] ER (Fig. 7A), we evaluated the effect of the nonlinearity on the Hill coefficient as described below. Because [Ca 2ϩ ] ER is difficult to calibrate, we could not estimate the range of Ca 2ϩ concentrations observed in the ER Ca 2ϩ imaging experiments precisely. Therefore, we assumed several possible ranges in which the Mag-Indo-1 signal maximally changed by 30% relative to the basal level ( Fig. 4D and supplemental Fig. S3). Fig. 7B shows the relationship between the Mag-Indo-1 signal change  and [Ca 2ϩ ] ER change when the basal Mag-Indo-1 signal was varied from 0.01 to 0.7 of the maximum. On the basis of the linear relationship between the Mag-Indo-1 signal and the FRET signal of F-L577 (Fig. 4D), we obtained the relationship between [Ca 2ϩ ] ER change and the FRET signal change of F-L577 (Fig. 7C). By using this relationship, we estimated the cytosolic Ca 2ϩ dependence of the SERCA pump activity, which was directly correlated with [Ca 2ϩ ] ER change (Fig. 7D). We found that the Hill coefficient of the cytosolic Ca 2ϩ dependence of the SERCA pump activity was almost constant regardless of the value of the basal Mag-Indo-1 signal (Fig. 7E). These results clearly demonstrate that the high cooperativity was not caused by a nonlinear relationship between the Mag-Indo-1 signal and [Ca 2ϩ ] ER .
The cooperative dependence of SERCA on cytosolic Ca 2ϩ is an important parameter for the generation of complex Ca 2ϩ signals such as Ca 2ϩ oscillations because small changes in this parameter had a large impact on the behavior of the Ca 2ϩ dynamics. Similar to the action potential, which is generated by the combination of opposite membrane potential changes pro-moted by fast voltage-gated sodium channels and slowly activated voltage-gated potassium channels (41), cytosolic Ca 2ϩ spikes may be generated by the ingenious balance of Ca 2ϩ influx and Ca 2ϩ efflux promoted by Ca 2ϩ release channels and Ca 2ϩ pumps, respectively. Therefore, our findings provide us with evidence to reconsider not only the SERCA pump activity but also the Ca 2ϩ release activity in living cells to gain a better understanding of the mechanism underlying the generation of Ca 2ϩ dynamics. In addition, Ca 2ϩ is one of the most important cytosolic signals in living cells, but a prolonged elevation of the cytosolic Ca 2ϩ concentration results in irreversible damage, as observed during cardiac or cerebral ischemia (42). The high cooperativity of SERCA should work to rapidly take up intracellular Ca 2ϩ to reduce its cytotoxicity.
Currently, we do not know the exact reason for the discrepancy between the Hill coefficients measured in intact cells (n H ϭ 5.7 Ϯ 0.73) (Fig. 5C) and in vitro assays (n H Ϸ 2) (16). To explore the mechanism for the highly cooperative dependence on cytosolic Ca 2ϩ , we carried out a simulation analysis based on  (Fig. 4D) was used to convert ⌬F/F base to ⌬R/R base . The shaded area shows the range of the FRET signals of F-L577 measured during Ca 2ϩ oscillations in living cells (Fig. 5C). D, the cytosolic Ca 2ϩ dependence on the SERCA2a pump activity was assessed by ⌬[Ca 2ϩ ] ER when F base ϭ 0.5. The measured values of ⌬R/R base of F-L577 shown in Fig. 5C were converted into ⌬[Ca 2ϩ ] ER by the relationship shown in Fig. 7C. The Hill coefficient of the best-fit equation is 5.7. E, the Hill coefficient estimated from the cytosolic Ca 2ϩ dependence of the SERCA2a pump activity assessed by ⌬[Ca 2ϩ ] ER was plotted against the assumed F base value. The dashed line represents the average value of 5.7. Similar results were obtained when the linear relationship between the Mag-Indo-1 signal and the FRET signal of F-L577 in the presence of different [Ca 2ϩ ] i (y ϭ 2.5996, x ϩ 0.0881) (supplemental Fig. S3) was used to convert ⌬F/F base to ⌬R/R base . The error bars represent the S.D.
the six-state SERCA model described by Yano et al. (43) (supplemental Fig. S4A). The model includes the regulation of SERCA pump activity by both [Ca 2ϩ ] i and [Ca 2ϩ ] ER . The ATPase activity of the SERCA pump was calculated for a cycle of cytosolic Ca 2ϩ oscillation with a peak value of 1 M and the baseline concentration of 150 nM with a sinusoidal decrease in [Ca 2ϩ ] ER (44). [Ca 2ϩ ] ER was set as 100 M at the resting state, with the minimal [Ca 2ϩ ] ER set as 50 M during the cycle. We found that nonlinear modulation of the rate of the transition from the E2 state to E1 state by cytosolic Ca 2ϩ can produce the highly cooperative dependence of SERCA activity on cytosolic Ca 2ϩ (supplemental Fig. S4B). We measured the intracellular distribution of F-L577 in COS7 cells before and after ATP stimulation (supplemental Fig. S5A) and found that the distribution of F-L577 was not largely changed after ATP stimulation, even though the FRET signal was changed significantly (supplemental Fig. S5C). These results suggest that the fluorescent signal changes of F-L577 are caused mainly by intramolecular FRET, rather than intermolecular FRET. Therefore, the high cooperativity obtained in this study is not derived from oligomeric interaction of SERCA pumps. These results suggest that factors such as Ca 2ϩ -dependent SERCA binding proteins, which can be excluded by the permeabilization treatment, are involved in the modulation of the transition rate from the E2 state to the E1 state for the generation of the high cooperativity in living cells. In fact, there is a report that an interacting modulator, such as phospholamban, alters the cooperativity of Ca 2ϩ binding to SERCA (45). The mechanism of the highly cooperative dependence of SERCA2a pump activity on cytosolic Ca 2ϩ is an interesting issue to be clarified in future studies.
In conclusion, we succeeded in visualizing the dynamics of SERCA2a in living cells by constructing F-L577, whose FRET signal changes reflect the instantaneous Ca 2ϩ pump activity. We found that the Ca 2ϩ pump activity of SERCA2a is synchronized with cytosolic Ca 2ϩ concentration changes without delay during Ca 2ϩ oscillations observed in COS7 cells stimulated with ATP. The Ca 2ϩ pump activity of SERCA2a in intact cells can be expressed by the Hill equation with an apparent affinity for Ca 2ϩ of 0.41 Ϯ 0.0095 M and a Hill coefficient of 5.7 Ϯ 0.73. The marked Ca 2ϩ dependence allows SERCA2a to act as a switch to refill the Ca 2ϩ stores efficiently. The F-L577 protein constructed in this study will be useful for future studies on Ca 2ϩ signaling in normal and abnormal cellular processes that involve SERCA pump activity.