Chemicals and solutions

For in vitro measurements, the Calcium Calibration Buffer Kit #1 (Life Technologies, Darmstadt, Germany) (pH 7.2, ϑ = 22°C, I = 0.1) was used to prepare 11 solutions of varying Ca²⁺ concentrations [Ca²⁺]_free (in nM: 0, 20, 40, 70, 100, 150, 230, 360, 620, 1390, 39660). For investigating protein-dye interaction, Ca²⁺-free and Ca²⁺-saturated buffer solutions were prepared containing 18 mM NaCl, 2 mM MgCl₂·6 H₂O, 40 mM KCl, 80 mM K gluconate, 5 mM HEPES and 1% (m/v) bovine serum albumin (BSA) (Sigma Aldrich, Deisenhofen, Germany). Additionally, the Ca²⁺-free buffer solution contained 2 mM EGTA and the Ca²⁺-saturated buffer solution 2 mM Ca(NO₃)₂. The pH was adjusted to pH 7.3 with 1 N HCl.

The K⁺-salts of the Ca²⁺-sensitive dyes Asante Calcium Red (ACR 25 µg, Teflabs Inc., Austin, USA) and Asante Calcium Green (ACG 25 µg, Teflabs Inc., Austin, USA) were dissolved in 50 µL double-distilled water to obtain 0.5 mM and 0.4 mM stock solutions, respectively. These stocks were then diluted in the Ca²⁺ buffer solutions, reaching final concentrations of [ACR]  = 2.5 µM and [ACG]  = 0.9 µM, respectively.

Cockroach physiological saline contained 160 mM NaCl, 10 mM KCl, 2 mM MgCl₂·6 H₂O, 2 mM CaCl₂·2 H₂O, 10 mM glucose and 10 mM Tris, pH 7.4. According to [4], a Ca²⁺-free stock solution of 0.5 M K₂H₂EGTA and a Ca²⁺-saturated stock solution of 1 M K₂CaEGTA were prepared by potentiometric back titration for in situ calibration experiments. These stocks were then diluted in a buffer solution (160 mM NaCl and 10 mM Tris) to the final concentration of 10 mM K₂H₂EGTA and 10 mM K₂CaEGTA calibration buffer solutions (pH 7.4, adjusted with HCl). By mixing K₂H₂EGTA and K₂CaEGTA, various free Ca²⁺ concentrations [Ca²⁺]_free could be obtained according to Equation (1).(1)

The given EGTA dissociation constant depends on temperature, ionic strength and pH, and under recent conditions (pH 7.4, ϑ = 20°C, I = 0.16) it could be determined through a Beers fitting procedure at  = 68.1 nM [19].

For in situ calibration experiments, the non-ionic surfactant Triton X-100 (Sigma Aldrich, Deisenhofen, Germany) was used to equilibrate defined extracellular and intracellular Ca²⁺ concentrations [20], [21]. Hence, the salivary glands were continuously perfused with calibration buffer solutions containing a defined [Ca²⁺]_free (in nM: 0, 20, 80, 460, 790, 2360, 33960, 680450) and 0.1% Triton X-100 (v/v). A 10 mM stock solution of dopamine (Sigma Aldrich, Deisenhofen, Germany) in double-distilled water was prepared daily and dissolved in physiological saline immediately before an experiment to a final dopamine concentration of 1 µM. The acetoxymethyl (AM)-ester of ACR (50 µg, Teflabs Inc., Austin, USA) was diluted in 27 µL Pluronic F-127 (20%-solution in DMSO, Sigma Aldrich, Deisenhofen, Germany), divided into 1 µL aliquots and stored at −20°C. The aliquots were dissolved in hypotonic physiological saline (75% physiological saline +25% water) immediately before an experiment to the final dye concentration of 5 µM.

Absorption and fluorescence measurements

Absorption measurements were performed with a Lambda 750 UV/VIS spectrometer (Perkin Elmer, Waltham, USA). To determine the absorption coefficients, absorption spectra in Ca²⁺-free and Ca²⁺-saturated buffer solutions were recorded. The dye concentration varied from 1.7 µM to 12.5 µM for ACR and from 0.9 µM to 1.8 µM for ACG. Fluorescence quantum yields of the Ca²⁺-saturated dye forms were determined absolutely with the C 9929 integration sphere system (Hamamatsu, Hamamatsu City, Japan). Since the fluorescence quantum yields of the Ca²⁺-free dye forms were below the detection limit of this system (Φ_F <0.01), fluorescence quantum yields of these dye forms were determined relative to the respective Ca²⁺-saturated form as a fluorescent reference [25], [26]. Steady-state fluorescence spectra were recorded with FluoroMax 4 (Horiba, Kyoto, Japan). For time-resolved fluorescence measurements in the BSA-buffer, ACR was excited by a supercontinuum source (SC-400-PP, Fianium, Southhampton, UK) operating at λ_ex = 550 nm with a repetition rate of 20 MHz and a pulse width of ∼30 ps. The laser beam was fiber-guided towards the fluorescence lifetime spectrometer FL920 (Edinburgh Instruments, Edinburgh, UK), where the emitted fluorescence was detected by a multichannel plate (ELDY EM1-123/300, EuroPhoton, Berlin, Germany) in the time-correlated single photon counting (TCSPC) mode.

2P fluorescence excitation spectra

2P fluorescence excitation action cross-sections Φ_Fσ₂ were determined from relative measurements using the well-characterized 2P-reference rhodamine B in methanol [22], [23]. Rhodamine B concentrations were adjusted for the respective samples and controlled by absorption spectra if possible. Thus, for 2.5 µM ACR, the rhodamine B concentration was adjusted to 3 nM and 0.1 µM in Ca²⁺-free and Ca²⁺-saturated conditions, respectively. For 0.9 µM ACG, the rhodamine B concentration was adjusted to 10 nM and 5.0 µM in Ca²⁺-free and Ca²⁺-saturated conditions, respectively. The Φ_Fσ₂ values with 10⁻⁵⁰ cm⁴ s/photon  = 1 GM were calculated according to Equation (2)

(2)where c is the dye concentration, Φ_F the fluorescence quantum yield, and the integral of the 2P-fluorescence emission spectra [22], [23]. The subscript r indicates the spectroscopic parameters of the 2P-reference rhodamine B. 2P-excitaiton was carried out by a Ti:Sa laser system (Tsunami 3960; Spectra Physics, Mountain View, USA) tuning the wavelength between λ_ex,2P = 720 nm–900 nm, with a 82 MHz repetition rate and pulse width of ∼80 fs. The laser beam was coupled to the fluorescence lifetime spectrometer FL920 (Edinburgh Instruments, Edinburgh, UK) and was focused by a lens on the quartz cuvette. The fluorescence emission was detected with a photomultiplier (S300 blue-sensitive R1527, Hamamatsu, Hamamatsu City, Japan) in a spectral range of 400 nm–700 nm in 1 nm steps. The obtained emission spectra were integrated, unraveling . The stability of the average laser power was controlled between the measurements with a power meter (Fieldmaster LM10 HTD with a detection range of 10 mW to 10 W). The error of the determined σ₂ values is estimated at 10%.

Tissue preparation

A colony of the American cockroach P. americana (L.) was reared at 27°C under a light/dark cycle of 12 h: 12 h at the Department of Animal Physiology (University of Potsdam). The animals had free access to food and water. Only male adults aged between 4–6 weeks were used for experiments. Salivary glands were dissected in physiological saline as described previously [24]. Small lobes consisting of several acini with their corresponding branched duct system were examined.

2P-FLIM recordings

Salivary gland lobes were incubated for 60 min in hypotonic saline containing 5.1 µM ACR/AM. After dye loading, the gland lobes were transferred to physiological saline for at least 10 min, then fixed on a coverslip coated with the tissue adhesive Vectabond Reagent (Enzo Life Science, Lörrach, Germany) and mounted on a microscope stage. During the experiment, the gland lobes were continuously perfused with physiological saline at a flow rate of 2 ml/min.

MicroTime200 (PicoQuant, Berlin, Germany) was used for time-resolved fluorescence measurements of in vitro (Calcium Calibration Buffer Kit #1) and in situ imaging. The system includes an inverted microscope (IX71, Olympus, Hamburg, Germany) equipped with an Olympus PlanApo ×100/NA1.4 oil-immersion objective. A mode-locked fs-fiber laser (C-Fiber A 780; MenloSystems, Martinsried, Germany) was used as a 2P-excitation source, operating at λ_ex,2P = 780 nm with a repetition rate of 50 MHz and a ∼90 fs pulse width. The laser beam was guided towards the objective via the microscope side port using a dichroic mirror (2P-dichroic 725; Chroma, Fürstenfeldbruck, Germany). Fluorescence was guided through a 100 µm pinhole and was detected by a single-photon avalanche diode (SPCM-AQR-13; Perkin Elmer, Waltham, USA). For rejection of the excitation light, two short pass filters were used (1×SP420-680 OD2, Baader, Mammendorf, Germany; 1×SP400-680 OD4, Edmund Optics, Karlsruhe, Germany). Time-resolved fluorescence was performed in the TCSPC-mode by using a PicoHarp300 (PicoQuant, Berlin, Germany) with a time resolution of 8 ps. Laser power was adjusted to achieve average photon counting rates of <10⁵ photon/s and peak rates close to 10⁶ photons/s, thus below the maximum counting rate allowed by the TCSPC electronics to avoid pulse pile-up. The image acquisition occurred by raster scanning the objective using a xy-piezo-positioner (Physik Instrumente, Karlsruhe, Germany). Thereby, full-frame images of 80 µm×80 µm (150 pixel ×150 pixel) were acquired in around 30 s with a pixel dwell time of 0.6 ms. Data acquisition and analysis were performed by the SymPhoTime software version 5.3.2.2 (PicoQuant, Berlin, Germany). All photons collected in a distinct region of interest (ROI) were used to form a global histogram. The obtained decay curves were analyzed by a deconvolution fitting routine, whereby the quality of the fit was estimated by reduced values and randomly distributed residuals. The required instrument response function (IRF) was measured every day by recording the backscattered excitation light [21]. Its full width at half-maximum (FWHM) was calculated at 220±2 ps (N = 21). Global multiexponential fluorescence lifetime analyses of several fluorescence decay curves with linked decay time components were carried out using FAST Software 2.13 (Edinburgh Instruments Ltd., Livingston, UK). In this case, two types of values were calculated. An individual value defines the quality of fit to an individual decay curve within the global analysis. A global value characterizes the quality of fit for the entire set of data and was calculated as the root mean square of all individual values. Based on the fluorescence lifetime analysis, every pixel in the image was then treated in SymPhoTime in the same way, but this time with a maximum likelihood estimator resulting in false color-coded FLIM images. In addition, fluorescence intensity images could be calculated by integrating all detected photons in every pixel, thereby ignoring the temporal information.

Data analyses

Statistically analyzed sets of data were distributed normally (D'Agostino-Pearson normality test). Thus, the data sets were further analyzed by repeated-measures ANOVA and Holm-Sidak's multiple comparison post-hoc tests. Differences were considered statistically significant if P<0.05. Statistical analyses were performed using GraphPad Prism 4.01 (GraphPad Software, San Diego, USA). Graphical presentation was carried out using Origin 8 (OriginLab, Northampton, USA). Data are presented as the mean ± the standard error of the mean (SEM). The dissociation constants K_D were calculated by fitting the data points to Equation (3)

or rather

(3)where v corresponds to fluorescence intensity and fluorescence decay time, respectively [6]. The subscript min refers to Ca²⁺-free buffer conditions and the subscript max to Ca²⁺-saturated buffer conditions.

In vitro steady-state absorption and fluorescence measurements

ACR is composed of a Ca²⁺ chelating BAPTA ligand and seminaphtofluorescein as the fluorescent backbone, as published recently (Figure 1A) [16]. According to a rise in [Ca²⁺]_free, the absorption maximum of ACR shifted from 537 nm (Ca²⁺-free) to 542 nm (Ca²⁺-saturated) (Figure 1C). Concurrently, the molar absorption coefficient decreased from (55.0±1.3) 10³ M ⁻¹ cm ⁻¹ (N = 3, λ_abs = 540 nm) to (47.8±0.8) 10³ M ⁻¹ cm ⁻¹ (N = 3, λ_abs = 540 nm). These absorption coefficients fit well with the published data [16]. ACR exhibited a broad emission band around 654 nm, whose fluorescence intensity was elevated due to increasing [Ca²⁺]_free (Figure 1C). The fluorescence quantum yield of Ca²⁺-bound ACR at λ_ex = 515 nm was determined absolutely at Φ_F = 0.06±0.01 (N = 6), whereas it was estimated for the Ca²⁺-free ACR relatively at Φ_F = 0.0017 as its fluorescence quantum yield was below the limit of detection to be determined by absolute measurements. Here, ACR in a Ca²⁺-saturated buffer was used as the fluorescent reference [25], [26]. Tuning the excitation wavelength to shorter wavelengths (e.g. λ_ex = 488 nm), a further fluorescence emission band around 525 nm appeared (Figure S1). However, the fluorescence intensity of this emission band was almost independent of [Ca²⁺]_free changes. Thus, ACR can be used as an emission-ratiometric dye by recording fluorescence at a Ca²⁺-dependent and a Ca²⁺-independent emission wavelength, which is particularly useful in terms of reliable Ca²⁺ quantification [7].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Steady-state absorption and fluorescence measurements for ACR and ACG.

Chemical structures of (A) ACR and (B) ACG [16], [27]. Absorption (black) and relative fluorescence (red) spectra in Ca²⁺-free (solid line, [Ca²⁺]_free  = 0 nM) and Ca²⁺-saturated (dashed line, [Ca²⁺]_free  = 40 µM) buffers. (C) ACR (c = 2.5 µM, λ_ex = 540 nm). (D) ACG (c = 0.9 µM, λ_ex = 517 nm). (E) Fluorescence enhancement factor (FEF) of ACR (dashed line) and ACG (solid line) as a function of excitation wavelength. FEF is the ratio of the fluorescence intensities of the excitation spectra under Ca²⁺-saturated conditions and Ca²⁺-free conditions. The curves are the smoothed means of two measurements.

https://doi.org/10.1371/journal.pone.0105334.g001

Analog to ACR, ACG is composed of a Ca²⁺ chelating BAPTA ligand and a fluorescein derivative as the fluorescent backbone (Figure 1B) [27]. The absorption maximum of ACG shifted from 515 nm (Ca²⁺-free) to 519 nm (Ca²⁺-saturated) (Figure 1D), whereas the molar absorption coefficient also increased from (130±2.0) 10³ M ⁻¹ cm ⁻¹ (N = 3, λ_abs = 520 nm) to (144±1.6) 10³ M ⁻¹ cm ⁻¹ (N = 3, λ_abs = 520 nm). Increasing [Ca²⁺]_free caused a rise in the fluorescence intensity with an emission maximum at 542 nm. The fluorescence quantum yield of Ca²⁺-bound ACG at λ_ex = 515 nm was determined absolutely at Φ_F = 0.42±0.01 (N = 6), whereas it was estimated for the Ca²⁺-free ACG-relatively at Φ_F = 0.0020 using ACG in a Ca²⁺-saturated buffer as the fluorescent reference. Presumably, for ACG as well as ACR, a photoinduced electron transfer is supposed to be the mechanism for changing fluorescence characteristics. Thereby, the Ca²⁺ complexation by BAPTA moiety blocks the electron transfer from the electron donor BAPTA to the fluorophore, which acts as an electron acceptor. Here, the fluorescence intensity also increases [28]. ACG and ACR absorb and emit in the visible spectral range. This fact is quite advantageous for biological applications, because side effects such as cell toxicity and autofluorescence occurring with excitation in the UV spectral range are minimized. A measure for the dynamic sensor range is the fluorescence enhancement factor (FEF) as a ratio of the fluorescence intensities in Ca²⁺-saturated and Ca²⁺-free buffers. Thereby, both dyes displayed a dependency of the FEF and the chosen excitation wavelength (Figure 1E). A maximum FEF was achieved for ACR (41-fold) at λ_ex = 548 nm and for ACG (231-fold) at λ_ex = 525 nm. The excitation-dependent fluorescence enhancement for ACR has been described recently [16]. For both dyes, the obtained FEF values correspond well with literature data [16], [29].

From steady-state measurements in commercial Calcium Calibration Buffer Kit #1, the dissociation constants K_D,i were calculated using Equation (3). In the case of ACG, the received K_D,i of 265 nM±33 nM (N = 4) is nearly twice as high as that of the manufacturer (135 nM) [29]. However, due to missing data on the experimental conditions, a comparison of the data is not feasible, as varying ionic strength, viscosity and pH might have an impact on the determined K_D,i. Depending on the excitation wavelength, ACR can be used in either single wavelength (λ_ex = 540 nm, λ_em = 655 nm) or dual wavelength (λ_ex = 488 nm, λ_em = 655 nm/525 nm) mode [16]. The fluorescence intensity of the emission band at 525 nm is almost independent of [Ca²⁺]_free. In this case, the ratio of the fluorescence intensities at 655 nm and 525 nm were used to calculate the corresponding K_D,i = 415 nM±43 nM (N = 4). In the single wavelength mode of ACR, a K_D,i = 463 nM±59 nM (N = 4) was calculated. In both cases, experimental data agree well with the manufacturer's data (K_D,i = 400 nM) [29]. Concluded from the determined K_D,i values, ACG and ACR are both high-affinity Ca²⁺ indicators [7].

2P fluorescence excitation spectra

With regard to biological applications, 2P-excitation offers several advantages in comparison to 1P-excitation, such as deeper tissue penetration, less light scattering, lower cell toxicity and autofluorescence [30]. To choose optimal excitation properties, the knowledge of 2P fluorescence excitation spectra is necessary. The efficiency of the 2P-excitation process can be quantified by the fluorescence excitation action cross-section, which is the product of fluorescence quantum yield Φ_F and 2P-absorption cross-section σ₂. The latter was determined by relative measurements, whereby rhodamine B in methanol as well as the characterized 2P-reference was used [22], [23]. The fluorescence quantum yield of rhodamine B (Φ_F = 0.50±0.003; N = 5) was measured absolutely and is in good agreement with the published data [31]. By means of absorption spectra, the dye concentrations were determined more accurately [32]. In the range of λ_ex,2P = 720 nm–860 nm the obtained 2P-absorption cross-sections σ₂ of ACR were rather constant, and only increased slightly at longer wavelengths σ₂ (Figure 2A). Moreover, the obtained σ₂ values for Ca²⁺-free and Ca²⁺-bound ACR were relatively small compared to other long excitation wavelength Ca²⁺-sensitive dyes and 2P-references rhodamine B or fluorescein [22], [23], [33]. However, even dyes with small 2P-action cross-sections can be successfully applied in 2P-experiments, as we have previously shown for the Cl^- sensor MQAE and the Na⁺ sensor SBFI, exhibiting relatively small Φ_Fσ₂ values <1 GM [17], [34]. For ACR, the calculated Φ_Fσ₂ values for the Ca²⁺-bound species were always around one order of magnitude higher than that of the Ca²⁺-free species at λ_ex,2P = 780 nm, the wavelength used for 2P-FLIM in situ recordings; Φ_Fσ₂ values of 0.04 GM (Ca²⁺-free) and 1.38 GM (Ca²⁺-bound) were determined (Figure 2A). So, an increase in 2P-excited fluorescence intensity due to increasing [Ca²⁺]_free is expected. Interestingly, recently published relative 2P-excited fluorescence spectra recorded in neuronal cells showed a different trend, with a maximum shifted to below 800 nm [16].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. 2P fluorescence excitation spectra for ACR and ACG.

Logarithmic plot of 2P fluorescence excitation action cross-sections Φ_Fσ₂ as a function of excitation wavelength. (A) ACR (means ± SEM, N = 6). (B) ACG (means ± SEM, N = 6). The black circles correspond to the 2P-reference rhodamine B in methanol and data were taken from the literature [22]. Ca²⁺-free ([Ca²⁺]_free  = 0 nM) and Ca²⁺-saturated ([Ca²⁺]_free  = 40 µM) conditions are depicted by blue and red squares, respectively. (C) Double logarithmic plot of the measured fluorescence intensity I_F as a function of 2P-excitation power P for rhodamine B (black circles) in methanol, ACR (red squares) and ACG (green triangles) in a Ca²⁺-saturated buffer (λ_ex,2P  = 780 nm; N = 3). The data points were fitted by a linear function (r²≥0.98).

https://doi.org/10.1371/journal.pone.0105334.g002

ACG showed a local maximum of Φ_Fσ₂ at λ_ex,2P = 820 nm for the Ca²⁺-free as well as Ca²⁺-bound species and continues to increase at longer wavelengths (Figure 2B). For the Ca²⁺-bound ACG species, higher Φ_Fσ₂ values were determined in comparison to ACR. Thus, the Φ_Fσ₂ values of the Ca²⁺-bound ACG species were in the same range as those of rhodamine B and thus slightly higher than those of other commercially available Ca²⁺ fluorophors like OGB-1 [21]. Similar to ACR, the calculated Φ_Fσ₂ values for the Ca²⁺-bound species were always higher than those of the Ca²⁺-free species, and at λ_ex,2P = 780 nm, Φ_Fσ₂ values of 0.15 GM and 44.55 GM were determined (Figure 2B). Thus, a rise in the 2P-excited fluorescence intensity due to increasing [Ca²⁺]_free is expected.

2P-excitation is a non-linear process. In this case, the measured fluorescence intensity I_F is a quadratic function of the excitation power, so in terms of ideal conditions the slope in a double-logarithmic plot is 2 [35]. This coincided for ACR (slope  = 2.0±0.1) and ACG (slope  = 2.0±0.1), and a slightly lower slope was calculated for rhodamine B (slope  = 1.6±0.1). Nevertheless, in all cases, a 2P-absorption process could be assumed (Figure 2C).

In vitro time-resolved fluorescence measurements

Concerning quantitative [Ca²⁺]_i recordings in living cells, time-resolved fluorescence measurements are an alternative approach to well-established fluorescence intensity measurements as has already been shown [5], [6], [13], [21]. In contrast to the latter, the fluorescence decay time is mostly independent of dye concentration, so that problems like dye leakage and dye bleaching are circumvented [36]. The suitability of ACR and ACG for time-resolved fluorescence measurements was tested in vitro by using the commercial Calcium Calibration Buffer Kit #1. Therefore, 11 solutions of varying [Ca²⁺]_free were prepared, and subsequently time-resolved fluorescence measurements were carried out. Both dyes were 2P-excited at 780 nm and the emitted fluorescence was recorded using the TCSPC technique. Theoretically, the easiest assumption would be the formation of 1∶1 complexes between Ca²⁺ ions and the Ca²⁺-sensitive dye. Hence, two distinct fluorescence decay time components would be expected, corresponding to the Ca²⁺-free and Ca²⁺-bound dye forms, respectively. Only their fractions relative to the entire fluorescence decay curve will change with varying [Ca²⁺]_free.

The obtained fluorescence decay curves of the Ca²⁺ concentration series for ACR and ACG showed a complex decay behavior and were globally fitted by a biexponential deconvolution fitting routine, in which the two decay time components were linked. The fits resulted in reasonable values and uniformly alternating residuals. Only around the fluorescence peak maximum were larger deviations observed (Figure 3A–3B). This could hint at an additional decay time component. However, a triexponential fitting model yielded slightly improved values and residuals (ACR: 3.09 vs. 2.20 and ACG: 2.03 vs. 1.60). The third decay time component for both dyes was calculated at approximately 20 ps–30 ps. These short decay times were considered uncertain, since such short decay times cannot be reliably resolved with the present setup displaying an IRF of 220 ps (FWHM). Thus, ACR exhibited two distinct fluorescence decay time components most probably attributed to the Ca²⁺-free (τ_free = 0.12 ns±0.006 ns) and the Ca²⁺-bound species (τ_bound = 0.57 ns±0.003 ns), respectively. For ACG, the decay time component of the Ca²⁺-free form is in the same range as that for ACR (τ_free = 0.25 ns±0.04 ns). On the other hand, the decay time component of the Ca²⁺-bound form is considerably higher (τ_bound = 2.38 ns±0.02 ns). The stated errors of the calculated fluorescence decay times were fitting errors derived from the FAST fitting routine.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Time-resolved fluorescence recordings of ACR and ACG after 2P-excitation at 780 nm.

Fluorescence decay curves of (A) ACR and (B) ACG in Ca²⁺-free (blue, [Ca²⁺]_free  = 0 nM) and Ca²⁺-saturated (red, [Ca²⁺]_free  = 40 µM) buffer solutions and the corresponding global biexponential deconvolution fits with values; IRF: instrument response function (black). The black arrow indicates increasing [Ca²⁺]_free. The residuals and individual values of the global biexponential fits are shown below. Normalized amplitudes α_i (squares, solid lines) and amplitude-weighted average fluorescence decay time τ_av,amp (triangles, dashed line) as a function of [Ca²⁺]_free for (C) ACR and (D) ACG. The blue squares correspond to the normalized amplitudes of the short decay time component (ACR: 0.12 ns±0.006 ns and ACG: 0.25 ns±0.04 ns; Ca²⁺-free species), whereas the normalized amplitudes of the long decay time component (ACR: 0.57 ns±0.003 ns and ACG: 2.38 ns±0.02 ns; Ca²⁺-bound) are depicted by red squares, (means ± SEM, N = 5).

https://doi.org/10.1371/journal.pone.0105334.g003

For both dyes, the normalized amplitudes α_i showed an opposite trend due to changes in [Ca²⁺]_free. Thereby, the normalized amplitudes α_f of the short decay time component, corresponding to the Ca²⁺-free dye form, decreased when [Ca²⁺]_free increased, whereas the normalized amplitudes α_b of the long decay time component, corresponding to the Ca²⁺-bound dye form, rose due to increasing [Ca²⁺]_free (Figure 3C–3D). It is somehow surprising that the normalized amplitudes α_i did not reach the initial fractions one and zero in Ca²⁺-saturated and Ca²⁺-free buffers, as we expected to find only the ion-free dye form in the Ca²⁺-free buffer and the ion-bound dye form of ACR and ACG in the Ca²⁺-saturated buffer. This phenomenon has been already discussed for other Ca²⁺-sensitive dyes and was either attributed to dye-aggregates or dye-impurities [13], [14], [21]. Nevertheless, a final explanation has not yet been confirmed.

Moreover, the determined fitting parameters allowed the calculation of the amplitude-weighted average decay time τ_av,amp. Hence, in both cases τ_av,amp increased with rising [Ca²⁺]_free. Fitting the obtained data by means of Equation (3) resulted in corresponding K_D,t values (Figure 3C–3D). So, in Ca²⁺ calibration buffer solutions, ACR and ACG exhibited K_D,t values of 114 nM±11 nM and 15 nM±3 nM, respectively. These dissociation constants obtained from time-resolved fluorescence measurements K_D,t deviate considerably from the dissociation constants determined from steady-state fluorescence measurements K_D,i. Similar characteristics have been described previously and a correlation between K_D,i and K_D,t was given by Equation (4) [6], [37], [38], [39].(4)

The subscripts f and s correspond to Ca²⁺-free and Ca²⁺-saturated conditions, respectively. In the case of 2P-excitation, the molar absorption coefficient ε needed to be replaced by its 2P analog, the 2P-absorption cross-section σ₂, resulting in Equation (5).(5)

Thus, it is possible to estimate K_D,t for ACR and ACG from steady-state fluorescence measurements (see above) to approximately 60 nM±5 nM and 9 nM±3 nM, respectively. The calculated K_D,t values for both dyes are slightly lower than the measured one, but they are within the same order of magnitude. In general, the dissociation constant K_D indicates the measurable analyte concentration range from approximately 0.1 to 10 times the K_D. However, the apparent value of K_D is determined by the particular measured parameter as it has been shown, for instance, for other Ca²⁺ indicator dyes recording fluorescence intensities or fluorescence decay times [6], [37]. Thus, when altering the measured parameter of an indicator dye, its dynamic range could probably be extended.

In vitro steady-state and time-resolved fluorescence measurements of ACR in the presence of BSA

Concerning intracellular ion recordings, the spectroscopic parameters of ion-sensitive fluorescent dyes can be changed due to factors such as viscosity, pH and ionic strength demanding fluorescence measurements under cellular conditions. Due to manifold factors, which are additionally often incalculable, the comprehensive analysis of interactions between a dye and cellular compounds would require extensive studies. Since particularly the presence of proteins significantly influences dyes, the protein-dye interaction can be initially investigated by using BSA [40], [41], [42]. The presence of BSA (1% (m/v)) in the in vitro buffer solutions resulted in a bathochromic shift in the Ca²⁺-free ACR absorption maximum from 537 nm to 555 nm, and increasing [Ca²⁺]_free led to a further bathochromic shift from 555 nm to 563 nm (Figure 4A). Moreover, the molar absorption coefficients were also decreased by about 20% compared to those measured in the BSA-free in vitro calibration buffer.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Spectroscopic properties of ACR in 1% BSA in vitro buffer solutions.

(A) Absorption (black) and relative fluorescence (red) spectra of ACR (c = 2.5 µM, λ_ex  = 550 nm) under Ca²⁺-free (solid lines) and Ca²⁺-saturated (dashed lines) conditions. (B) Fluorescence enhancement factor (FEF) of ACR as a function of excitation wavelength (means ± SEM, N = 3). The FEF is the ratio of the fluorescence intensities of the excitation spectra under Ca²⁺-saturated and Ca²⁺-free conditions. (C) Fluorescence decay curves of ACR in Ca²⁺-free (blue) and Ca²⁺-saturated (red) buffers (λ_ex = 550 nm, 1P-excitation, λ_em = 640 nm) and the corresponding global biexponential deconvolution fits with the value; IRF: instrument response function (black). The black arrow indicates increasing [Ca²⁺]_free. The residuals and individual values of the global biexponential fits are shown below.

https://doi.org/10.1371/journal.pone.0105334.g004

The fluorescence intensity of ACR under Ca²⁺-free conditions was increased in the presence of BSA and rose upon [Ca²⁺]_free elevation. However, the fluorescence spectrum exhibited a hypsochromic shift from 654 nm to 641 nm (Figure 4A). In the presence of BSA, the maximum FEF was achieved at λ_ex = 538 nm with a FEF ∼8, which was about 4 times lower than that in the absence of BSA (Figure 4B vs. Figure 1E). This behavior could be also revealed by the absolutely determined fluorescence quantum yields Φ_F = 0.01 (N = 3) for the Ca²⁺-free (λ_ex = 543 nm) and Φ_F = 0.08 (N = 3) for the Ca²⁺-bound (λ_ex = 550 nm) ACR species in 1% BSA in vitro buffer solutions. Obviously, the influence of BSA seemed to be stronger on the Ca²⁺-free than on the Ca²⁺-bound ACR species. In fact, the bathochromic shift in the absorption spectrum and the hypsochromic shift in the emission spectrum as well as enhanced intensity have also been reported for other Ca²⁺-sensitive dyes such as Fura-2, Indo-1 and Quin-2 [41], [42], [43]. Since the fluorescent backbones of these dyes are different, it was suggested that the carboxyl groups of the dye's Ca²⁺-chelating side interact with proteins. Due to their higher affinity to Ca²⁺ ions the carboxyl group would be rather occupied by Ca²⁺ than by proteins. Thus, in the presence of Ca²⁺ the formation of a protein-dye complex is interfered.

The obtained fluorescence decay curves could be adequately fitted by a deconvolution biexponential fitting routine yielding reasonable residuals and values (Figure 4C). The determined fluorescence decay time components for the Ca²⁺-free and the Ca²⁺-bound species were about 5 times longer than those obtained in BSA-free environments and amounted to τ_free = 0.77±0.002 ns (N = 3) and τ_bound = 2.86±0.04 ns (N = 3), respectively.

The 2P-absorption cross-sections of ACR in the presence of BSA were determined at λ_ex,2P = 780 nm as described previously, and yielded σ₂ = 40 GM±1 GM (N = 3) for the Ca²⁺-free and σ₂ = 24 GM ±1 GM (N = 3) for the Ca²⁺-bound dye species. Nevertheless, the 2P-action cross-section Φ_Fσ₂ of the Ca²⁺-free and Ca²⁺-bound dye species were calculated to 0.4 GM and 1.9 GM, still indicating an increase in the ACR fluorescence intensity due to increasing [Ca²⁺]_free in the presence of BSA. All spectroscopic parameters determined in BSA-containing buffer solutions deviated from those perceived in BSA-free calibration buffer solutions, which can be attributed to dye-protein interactions. This fact underlines the need for determining the calibration parameters used for intracellular ion concentration quantification individually in each biological system.

Time-resolved fluorescence measurements of ACR in situ

Although the in vitro measurements of ACG have unraveled larger absorption coefficients, higher fluorescence quantum yields, and larger FEFs compared to those of ACR, at this time ACG is still only commercially available in K⁺-salt form, but not in the ester form required for experiments in living tissues. Indeed, the ACG salt form could be microinjected into a selected living cell for in situ recordings. However, this would be an invasive approach and the simultaneous analysis of several cells would be virtually impossible. On the other hand, ACR is commercially available as an AM-ester. So, subsequently the feasibility of ACR for 2P-FLIM recordings in living cells was tested. The investigated salivary gland ducts showed a relatively low cellular autofluorescence after 2P-excitation at λ_ex,2P  = 780 nm, but with a distinct signal induced by the luminal cuticle (Figure 5A). In contrast, the fluorescence intensity of the duct cells was increased nearly 6 times after ACR loading (Figure 5B). Moreover, an inhomogeneous dye loading in terms of significantly weaker nuclei staining was observed. The same behavior has recently been described for the Ca²⁺ sensor OGB-1 [21].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Behavior of ACR in cockroach salivary glands.

2P-excited (780 nm) fluorescence intensity images of (A) unloaded (1 luminal cuticule, 2 ductal lumen, 3 apically located, point-shaped structures) and (B) ACR-loaded salivary gland ducts (median optical sections). The graphs below the images display the fluorescence intensity traces along the white lines in the images.

https://doi.org/10.1371/journal.pone.0105334.g005

Furthermore, the influences of dye leakage and dye bleaching of ACR after 2P-excitation were determined, as both effects are limiting factors for long-term measurements. Therefore, fluorescence intensity images were recorded for 60 min at a low acquisition rate of 0.067 min⁻¹ (P = 3.3 mW, measured at objective aperture) and for 30 min at an increased acquisition rate of 1.3 min⁻¹ (P = 3.7 mW). In the first case, the initial fluorescence intensity is reduced by one-third after 60 min, which can presumably be attributed to dye leakage (Figure S2A). This is in a comparable range to other Ca²⁺-sensitive dyes [21]. In the second case, the recorded fluorescence intensity showed a gradual decrease and was fitted by a monoexponential decay function, which yielded a half-time t_½ of fluorescence dye bleaching of about t_½ = 14 min (Figure S2B). Compared to other Ca²⁺-sensitive dyes such as OGB-1, the t_½ of ACR is almost two times shorter [21]. This fact has to be taken into account for long-term measurements, especially for fluorescence intensity-based ones.

As already shown for BSA, the spectroscopic properties of ACR are influenced by their environment. Additionally, several studies have shown that in vitro dissociation constants K_D differ from in situ K_D [21], [44]. Due to this, the K_D value has to be determined individually for each experimental setup and biological system. Thus, in situ calibration experiments were performed using the non-ionic surfactant Triton X-100 as an equilibrating agent [21]. Hence, the salivary glands were continuously perfused with a calibration buffer solution containing defined [Ca²⁺]_free and 0.1% Triton X-100. This Triton X-100 concentration enabled a [Ca²⁺]_free equilibration without significant intracellular dye loss and tissue deterioration within the experimental time period in the cockroach salivary glands, as reported recently [21]. Every minute a 2P-FLIM image was recorded, and for data analysis images were taken with the maximum Ca²⁺ effect, but without visible tissue deformation. Thereby, 126 fluorescence decay curves were obtained from ROIs, which were fitted globally by the biexponential deconvolution function, yielding reasonable residuals and values (Figure 6A). Thus, two distinct decay time components were identified at τ = 0.56 ns and τ = 2.44 ns, corresponding most probably to Ca²⁺-free and Ca²⁺-bound ACR, respectively. These time components were in the same range as those measured in the BSA-buffer, indicating a protein-dye interaction causing longer fluorescence decay times. The obtained fitting parameters α_i and τ_av,amp were plotted against corresponding [Ca²⁺]_free and fitted by Equation (3), unraveling the in situ dissociation constant of ACR to K_D,t = 180 nM±80 nM (Figure 6B). This is about 1.7 times larger than the measured in vitro K_D,t value. The only available in situ K_D values for ACR so far were published by Ljubojević et al. [44]. In that study, K_D,i values were determined in rat ventricular cardiomyocytes at 2183 nM in nucleoplasm and 1336 nM in cytoplasm, respectively. However, both K_D values are not directly comparable to the values presented here, since they are based on fluorescence intensity-based measurements which seem to result in larger K_D values. The measurable analyte concentration range is thought to be between 0.1 and 10 times that of the K_D [7]. Thus, the determined in situ K_D,t of ACR fits well with the expected [Ca²⁺]_i changes in cockroach salivary gland duct cells [21], [45]. On the other hand, the changes in τ_av,amp were unfortunately very small, so quantification of low [Ca²⁺]_i changes seems to be challenging (Figure 6B). Conversion of the effective amplitude-weighted average fluorescence decay time τ_av,amp into [Ca²⁺]_i has been suggested previously according to Equation (6)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Determination of in situ K_D,t for ACR in salivary duct cells.

(A) Fluorescence decay curves extracted from 2P-FLIM images of ACR-loaded salivary duct cells under Ca²⁺-free (blue, [Ca²⁺]_free  = 0 nM) and Ca²⁺-saturated (red, [Ca²⁺]_free  = 68 µM) conditions and the corresponding global biexponential deconvolution fits with the value. The residuals and individual values of the global biexponential fits are shown below. (B) Normalized amplitudes α_i (squares, solid lines) and amplitude-weighted average fluorescence decay time τ_av,amp (triangles, dashed line). The blue squares correspond to the normalized amplitudes of the short decay time component (0.56 ns, Ca²⁺-free species), whereas the normalized amplitudes of the long decay time component (2.44 ns, Ca²⁺-bound species) are depicted by red squares (N = 11–23). The dotted line marks the determined K_D,t.

https://doi.org/10.1371/journal.pone.0105334.g006

(6)with the extreme τ_av,amp values at Ca²⁺-free (τ_av,amp,min) and Ca²⁺-saturated (τ_av,amp,max) conditions, respectively [38], [37].

The feasibility of ACR for in situ measurements was tested in duct cells of cockroach salivary glands, representing a well-established model system for investigating aminergic controlled transepithelial ion transport processes [18]. The biogenic amine dopamine induces a saliva secretion accompanied by a reversible, slow, almost tonic, dose-dependent increase in [Ca²⁺]_i in the duct cells from 48 nM±4 nM to 311 nM±43 nM as demonstrated by steady-state fluorescence recordings using Fura-2 [45], [46]. A second, subsequent dopamine stimulus can yield an even four-times larger [Ca²⁺]_i elevation [45]. Recently, a dopamine-induced [Ca²⁺]_i increase from 144 nM±15 nM to 372 nM±160 nM was confirmed using the Ca²⁺ sensor dye OGB-1 for 2P-FLIM recordings [21]. Thus, this Ca²⁺ signaling response is a valuable tool for testing new Ca²⁺ sensor dyes in situ. When [Ca²⁺]_i increases, the corresponding ACR τ_av,amp should increase as well. To verify this assumption, ACR-loaded salivary duct cells were stimulated twice with 1 µM dopamine, and 2P-FLIM images were recorded continually (image acquisition time ∼30 s, acquisition rate 1 image min⁻¹). The appropriate fluorescence decay curves were fitted by a biexpontential decay function with fixed decay time components from the in situ calibration. Based on the fitting results, the normalized amplitudes α_i as well as τ_av,amp values were calculated (Figure 7A). As ACR bleached quickly under the chosen conditions (see also above), the quality of the fluorescence decay curves deteriorated during the experiment. Nevertheless, a sufficient fluorescence signal was obtained until the end of a measurement, indicated by reasonable residuals and values. After a bath application of dopamine, the normalized amplitudes showed an opposite trend as expected and τ_av,amp rose statistically significantly and recovered to resting state level after dopamine washout (Figure 7A–7B). The second dopamine stimulus proved the reversibility of this process, as well as the viability of the tissue under these 2P-FLIM experimental conditions. By means of the in situ calibration, it was possible to correlate τ_av,amp values with quantitative [Ca²⁺]_i changes. Thereby, resting [Ca²⁺]_i was calculated to 21 nM±18 nM (N = 15). In the presence of dopamine, [Ca²⁺]_i increased to 590 nM±300 nM (N = 15). The second dopamine stimulus resulted in a similar rise to 500 nM±260 nM (N = 15) (Figure 7C). The described behavior could also be identified by false color-coded 2P-FLIM images, in which warmer colors correspond to higher [Ca²⁺]_i (Figure 7D). The obtained results fit qualitatively well to the previously published data using Fura-2 and OGB-1 [21], [45]. However, large variations of τ_av,amp values could be recognized between the individual experiments, causing broader calcium concentration distributions (Figure 7B–7C). A reason for this could be found in the spatio-temporal behavior of the dopamine-induced [Ca²⁺]_i rise. At first, in addition to a tonic [Ca²⁺]_i increase during dopamine application, some salivary gland preparations exhibit a slow decrease after reaching a [Ca²⁺]_i peak value. It is also known that the [Ca²⁺]_i rise starts at several cells and spreads slowly over the duct epithelium, but not all regions of the duct seem to respond [45]. Thus, depending on the relatively small duct region investigated (80 µm×80 µm), variable responses to dopamine could be expected. In addition, the image acquisition rate was limited to 1 min⁻¹, making recordings of [Ca²⁺]_i rises of different magnitude also more likely. So, faster Ca²⁺ imaging could be realized by recording smaller regions of interest or directly at one single point. But then, a higher impact of photobleaching and tissue damage has to be taken into consideration. Another source of errors could be the weak fluorescence of the Ca²⁺-free ACR species. So, contribution of cellular autofluorescence to the overall fluorescence decay is increasingly strengthened by decreasing [Ca²⁺]_i and consequently might no longer be negligible. Evidence can therefore be seen in the resting [Ca²⁺]_i, just alternating around zero. Another aspect which must be considered is the possible impact of varying intracellular pH. For ACR, it has been reported that protonation due to acidification leads to a fluorescence intensity increase in the Ca²⁺-free species without effecting the Ca²⁺-bound species [16]. Thus, a [Ca²⁺]_i rise accompanied by intracellular acidification could probably be detected with ACR at a higher [Ca²⁺]_i level than actually occurs. From the literature it is known that cockroach salivary gland duct cells acidify after dopamine stimulation, from about pH 7.3 to pH 6.9 [47]. This acidification would lead to an increase in the signal of the Ca²⁺-free ACR species of about 10%, which might cause a slight overestimation of the [Ca²⁺]_i rise.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Analysis of dopamine-induced [Ca²⁺]_i changes in ACR-loaded salivary duct cells by 2P-FLIM recordings.

(A) Normalized amplitudes α_i (squares, solid lines) and amplitude-weighted average fluorescence decay time τ_av,amp (triangles, dashed line). The blue squares correspond to the normalized amplitudes of the short decay time component (0.56 ns, Ca²⁺-free species), whereas the normalized amplitudes of the long decay time component (2.44 ns, Ca²⁺-bound species) are depicted by red squares (means, N = 15). Black bars indicate the periods of 1 µM dopamine presence. (B) Statistical analysis of the dopamine effect; repeated-measures ANOVA and Holm-Sidak's multiple comparison tests (* P<0.05, *** P<0.001). (C) Data converted to [Ca²⁺]_i. The error bars shown were calculated by error propagation of τ_av,amp. Black bars indicate the periods of dopamine presence and (D) corresponding false color-coded 2P-FLIM images at distinct time points.

https://doi.org/10.1371/journal.pone.0105334.g007