Cdl caklum
(1882) 13, s-88
Q Longman Group UK Lid 1982
Microspectrofluorometry as a tool for investigation of non-calcium interactions of Indo-l F. BANCEL, J-M. SALMON, J. VIGO and P. WALLET Groupe de Microfloorim&rie Quantitative, UffA CNRS 1289, Universit6 de Pefpignan, Petpignan, France Abstract - Indo-l is a fluorescent calcium probe used to measure intracellular free calcium concentrations. These measurements are often performed by comparing the fluorescence intensities of Indo-l-treated cells at two selected wavelengths corresponding to the maxima of the fluorescence spectra of the calcium-bound and calcium-free forms. In this study, we used an optical multichannel anaiyser to numerise the fluorescence emitted by a single cell. A computerised resolution of numerlsed spectra was used on Intracellular Indo-i fluorescence. Calculation of numerical and graphic estimators allows us to evaluate the fit of the resolution. Dlfferent sets of characteristic spectra were compared using this method. It appeared that no linear combination of the two known forms of Indo-l and of the cell autofluorescence can fit with spectra of Indo-l-treated cells. In additlon, a study of the physico-chemical properties of Indo-l shows the existence of two other forms of the molecule: a protonated form (maximum emission at 455 nm) and a form in interaction with proteins (maximum emission at 438 nm). Taking into account the contribution of these two new forms leads to an improved spectral resolution of the fluorescence of Indo-l-treated living cells and, therefore, improves calcium measurements. Moreover, quantlficatlon of the amount of the protonated form of Indo-l allows a measurement of intracellular pH at the same time as calcium determination.
Abbreviations : Indo-1, lH-indole-6-carboxylic acid, 2-(4-bis-(carboxymethyl)amino-3-(2-(2-(bis~boxymethyl) amino-S-methylphenoxy)ethoxy)phenyl; Indo-l/AM, pentaacetoxymethyl ester of Indo-1; Fura-2, 1-[2-(karboxyoxazol-2-yl)-6-aminobenzofuran-S-oxy]-2-(%unino-5’ methylphenoxy)-et-N,N.N’,N’-tetmacetic acid; EGTA,
ethyleneglycol-bis-@-aminoethy1 ether)-N,N,N’,N’-tetraacetic acid; BAFTA, 1,2-bis(o-aminophenoxy)-ethaneN,N,N’.N’-tetraacetic acid; DMSO. diiethylsulfoxidt; DMEM, Dulbecco’s modified Eagle’s essential minii medium; BSA. bovine serum albumin, [Ca*+]i, inea~ellular calcium concentration; and pH,, intracelhrlsr pH
The use of fluorescent probes is often the most convenient technique for measuring intracellular ionic concentrations. Numerous probes have been designed with this aim, particularly in the case of the calcium ion, whose role as cellular messenger has been known for a long time [l]. Several generations of fluorescent calcium probes have been derived from the structure of the non-fluonzscent calcium chelator EGTA [2, 31. Indo-l belongs to the second generation of these tetracarboxylate calcium indicators. One of its most athractive properties is a 70 nm spectral shift of its 59
60
emission spectrum after calcium binding. This property has led to a widely used practice of calculating intracellular calcium concentration from the ratio of the fluorescence intensities obtained at two selected wavelengths using interference filters. The respective characteristic fluorescence spectra of the calcium-free and calcium-bound forms can be easily obtained, at pH > 7.5. in calcium-&e and high calcium solutions, respectively. The dissociation constant of the probe for the calcium ions is usually obtained using EGTA as a calcium buffer. The apparent Kd of EGTA for Ca2+depends on pH and the free calcium concentration of the solutions must be computed for each pH value [3]. Several authors have encountered problems with the analysis of fluorescence of the tetracarboxylate family of calcium probes in living cells [5-g]. It was reported that part of the fluorescence emitted by cells treated with Indo-l or Fura- could be attributed to neither the calcium-bound nor the calcium-free forms of the probes. The aim of this study was to connect such intracellular observations with physico-chemical properties of Indo-l and to provide a comprehensive basis for further utilization of non-calcium interactions of the probe. In order to analyse intracellular Indo-l fluorescence, we have used a technique based on computerised resolution of numerised fluorescence spectra [lo]. This technique allows us to calculate the contribution of several fluorescent compounds in a mixture. Investigation of intracellular fluorescence of Indwl was possible using a multichannel microspectrofluorometer for a numerical recording of fluorescence spectra emitted by single cells. After each resolution, calculation of fit estimators allows us to compare several sets of spectra in terms of accuracy and to determine whether or not experimental fluorescence spectra are correctly analysed. This study combines physico-chemical investigations of (a) acid-base properties [ll] and (b) interaction of Indo-l with proteins, and intracellular detection of the related fluorescence spectra in 3T3 fibroblasts. Results were obtained by spcctrofluorometry, for studies of the properties of the probe in solution, and multichannel microspectiofluorometry, for studies in single living 3T3 cells.
CELL CALCIUM
Materials and Methods Experimentsin solution
Indo-l, Indo-l/AM and BAPTA were purchased from Molecular Probes (Eugene, OR, USA). A stock solution of 1 mh4 was prepared in ultra-pure water (18.6 MQ 0.005 ppm Ca, Merck, used in all experiments in solution). Samples (10 ml) of 10% BSA (Sigma) were dialysed three times against 0.5 1 of water, in order to remove calcium ions. The actual concentration of albumin stock solution was assayed at 1 mM (7% w/v) using the Lowry method. Potentiomettic determination of the pH of all solutions was performed using a Taccussel PHN 81 digital pH-meter (+ 0.01 pH unit) with a Xerolyt electrode (Ingold). Molecular filtration experiments were performed using a Pharmacia PDlO (G25) column previously equilibrated with 10 ml H20/KOH, pH = 7.5. Poly-r.-lysine (MW = 300007OCOO) was purchased from Sigma. Fluorescence spectra were digitally recorded in a 1 cm width quartz cuvette with a Jobin-Yvon JY3D spectrofluorometcr interfaced with a Tandon AT 286 microcomputer. Excitation wavelength was 340 nm and spectra were recorded between 360 and 600 nrn, Fluorescence spectra were analysed using the characteristic spectra of each participating form of Indo-l and a resolution method developed in this laboratory [ 101as explained below. Cell culture and loading
3T3 mouse fibroblasts (Flow Laboratories) were routinely cultured in DMEM (Flow Laboratories) with 10% decomplemented fetal calf serum (FCS, Gibco) and 2 mM L-glutamine (Flow Laboratories) at 37°C in 25 cm2 flasks. For experiments, cells were seeded in Sykes-More chambers at 4000060000 cells per chamber at 37”C, 5% Co;?. Cells were incubated for 1 h in Dh4EM containing 5 p.M Indel/AM, 0.5% DMSO and washed three times with cold phosphate-saline buffer solution. The Sykes-Moore chambers were then placed on the thermostated stage (37°C) of the rnicrospcctrofluoromcter.
61
INVESTIGATION OF NON-CALCIUM INTERACTIONS OF INDO-
Fig. 1 Equation set generated for the determination of the composition of complex mixtures of fluorescent compounds. Diagonal terms are in bold characters and delimited by the solid line
Microspectrofluorometryon single cells The microspectrofluorometer is composed of an inverted microscope (Leitz) connected to an Optical Multichannel Analyser (OMA, Princeton Applied Research Corp.) equipped with a silicon intensified target (SIT) as detector. The excitation wavelength was selected at 337 MI from the emission of a xenon lamp with a monochromator. Spectra were recorded with a signal to noise ratio greater than 30 between 350 and 640 nm (400 channels, 0.72 nm per channel). Data were transferred to a PDP 11/73 microcomputer for storage and (Plessey ) calculations.
ITL is the intensity of experimental spectrum at the wavelength h, I$. is the intensity of the i* characteristic spectrum, and ai is a coefficient proportional to its contribution in the mixture. When a fluorescence speclrum emitted from a single cell is recorded, the amount of photons collected per frame scan (32 x 10m3s) for each channel is low. The signal to noise ratio can be increased by (a) increasing the number of accumulations of the signal and/or (b) integration of the fluorescence spectnun The first process is time consuming (3 s for the summation of 100 fluorescence spectra) while the second leads to the loss of spectral information. An integration of Equation 1 over the whole spectral domain gives:
Numerical resolutionoffluorescence spectra This method takes advantage of the fact that if a complex fluorescence spectrum results from the fluorescence of N chemicals, the intensity at each wavelength can be described as a linear combination of the intensities of each component at the same wavelength. This can be expnzssed as follows: N
ITA = C i=l
ai Ii,h
Eq. 1
c I’h
k=l.lJ
=
C ai C i=l
Ii,1
Eq. 2
h=b
A set of N independent equations is then generated from Equation 2 using N modulating functions depending on h in order to restore the spectral information. Theoretically. any non-random function cDj,h is suitable for this purpose. The originality of the method we have developed lies in (a) the use of the characteristic spectmm Ij,L of each
62
CELL CALCIUM
fluorescent entity as a modulating function and (b) the weighting of this function by the complex spectrum ITh, i.e.:
@j,h= ITk Ij,k
Eq. 3
l
The major advantages of generating tiquation sets in this way am (a) to maximise in each lquation the relative weight of only one component (this procedure generates a system more diagonai than the unweighted one and therefore increases the value of the determinant) and (b) to vary the weight of the data with respect to their intensity. The tinal equation set (Fig. 1) is then resolved using a classical routine. Once the results are obtained (i.e. values for al, az, ... aN), residuals are calculated as the difference between a reconstructed complex spectrum and the experimental one. Weighted residuals (WR) are then calculated as the ratio of the residuals versus the absolute value of the random noise superimposed on the signal (for more details, see [lo]). The fit between the result of the resolution and the experimental spectrum is represented by a graphic plot of the WR and a numeric estimator is calculated as the chi-square of the WR. For an optimally resolved spectrum WR should be randomly distributed about the zero value and chi-square should be close to 1. This technique allows us to calculate optimal contributions of several fluorescence spectra to the
400
500 Wavelength
Fig. 2 Changes in Indo-l fluomcence
600
(nm)
spectra in response to pH changes. Inset: titration curve accounting for the equilibrium between the L and LH forms
complex spectrum resulting from a mixtune of several chemicals. This implies that the characteristic fluorescence spectrum of each fluorescent compound is introduced in the resolution system The user must then record the characteristic spectrum of each form and store it in a library of spectra prior to analysis. R~ults i ud Discussion
Constructronof a library offluorescence spectra Calcium-@e and calcium-boundforms of Indo-l
The calch n-free (L) and calcium-bound (LM) spectra wa ; obtained by recording the fluorescence of solutions at (a) 50 pM Indwl, 2 r&I EGTA. pH = 8.0 and (b) 5 pM Indo-l with a saturating concentration of CaCh, pH 7.5 after elimination of a precipitate of Ca(OH)z by centrifugation, respectively. Indo-l interaction with calcium ions was described in detail in [ll]. Among the physico-chemical parameters which can act on Indo-l-fluorescence, pH has to be studied fast [12]. Acid-baseproperties of Indo-l
We have previously published a study of the acidbase properties of Indo-l in aqueous solution [ll]. In this study, a protonated form of Indo-l (LH) was detected, with an emission spectrum maximum at 455 nm. Figure 2 shows the changes in Indo-l fluorescence spectra in response to pH changes. A titration curve can be plotted from the spectroscopic data (inset), the logarithm of the acidity constant @Ka) was calculated at 6.18. The characteristic fluorescence spectrum of LH was obtained by subtracting the calculated contribution of the L spectrum from the spectrum of a solution containing 50 p.M Indo-l and 2 mM BARTA, pH = 6.0. This subtraction was performed according to the values of pKa (6.18) and the rate of quantum yields (R = YWYLH = 0.5) previously measured [ll]. interactionof Indo- withproteins
In order to complete the previous study [l 11, we have recorded the fluorescence spectrum of Indo-l in fetal calf serum and in DMEM, with 5 mM EGTA. We observed an unexpected fluorescence
INVESTIGATION
OF NON-CALCIUM
INTERACTIONS
63
OF INDO-
the fluorescence spectrum of the pentaacetoxymethyl ester of Indo-l fIndo-l/AM, 50 p&l in Hanks’ Balanced Salt Solution (Gibco), 5% DMSO] and (b) the intrinsic cellular fluorescence of untreated 3T3 fibroblasts (in our excitation conditions, i.e. 337 run, this fluorescence is mainly the emission of NAD(P)H). Obtaining a interactions
360
400
440
480
520
Wavelength Fig. 3 Changes varying Scatchard
in Indo-l
concentrations plot obtained
fluorescence
of
dialysed
580
800
(nm)
spectra in the presence of BSA,
from the resolution
pH
-
7.5.
Inset
of the experimental
spectra (v = [LP]/[BSA]total)
peak around 440 nm (data not shown), out of the pH range where LH is found (i.e pH > 7.5). We then studied the fluorescence of Indo-l solutions at pH = 7.5, in the presence of varying concentrations of dialysed BSA. We obtained an iso-emissive point (Fig. 3) with a spectrum attributed to a form of Indo-l interacting with BSA (maximum emission = 438 run, called LP in the following), and the L spectrum (maximum emission = 485 run), in proportions depending on BSA concentration. Each composite spectrum was then analysed using our resolution method. A Scatchard diagram plotted from these results (inset of Fig. 3), showed a high affinit site on BSA (dissociation constant: Kd c: Y 3.10- M) and several other sites of lower affinity. Interaction of Indo-l with proteins was not limited to BSA: we obtained the same spectral changes when Indo-l was in the presence of proteins as different as histones and trypsin. This could indicate that Indo-l is likely to interact with a great number of cellular proteins. This interaction could be mediated by a class of amino-acid rather than a specific binding site on BSA. In addition, interaction of Indo-l with intracellular proteins could be correlated with the observation of a slow-diffusing fraction of the probe by Blat&r et al. [13]. The library of fluorescence spectra was completed by recording in the same conditions (a)
comprehensive
pH-Dependence proteins
model
of
Indo-l
of the interaction of Indo-l
with
To study the pH-dependence of the interaction of Indo-l with proteins, the evolution of the molar fraction of LP in response to pH changes was followed in a solution containing 20 pM Indo-l and 5 pM dialysed BSA. Fluorescence spectra were recorded for pH values varying between 5.5 and 9.5 and the composition of the solutions at each pH value was determined using our resolution method. An example of such a resolution is shown in Figure 4. The molar fraction of the LP form was calculated from the contribution of LP spectrum (R = WYLP = 0.56). The variation of this molar fraction versus pH is plotted in the inset These results indicate an increase of the interaction of Indo-l with BSA as the pH decreases in the range 9-6.5. Further acidification leads to a decrease of the interaction. Two conclusions could be deduced from these observations: (a) according to the results reported by Konishi et al. on the interaction of Fura- with proteins [14], the type of interaction could be electrostatic and mediated by basic amino-acids, such as arginine and lysine, that are positivelycharged when protonated @Ka in the range 8-9); and (b) the lowering of the interaction of pH values below 6.5 occurs when the protonation of the basic amino acid residues is almost complete. This could be due to the absence of interaction between the basic amino-acid residues and the LH form of Indo-1. Chromatographic interaction
study
of
Indo-llproteints)
It would be interesting to know which form(s) of Indo-l is (are) likely to bind to proteins in order to determine a comprehensive model of Indo-l
64
CELL CALCIUM
7
0
PH
450
550 Wavelength
(nm)
Fig. 4 pH dependence of the interaction of Indo-l with BSA: example of resolu~on of the fluorescence speettum of a solution containing 20 @I Indo-1, 5 ph4 BSA, pH - 6.5 (filled circles) using the spectra of the L (open squares: 12.6%). Lh4 (filled squares: 17.4%). LH (open triangles: 18.3%) and LP (filled triangles: 61.1%) forms of Indo-1. x2 - 0.989. Weighted residues (WR) am represented (x 300). Inset: plot of the molar fraction of Indo-l bound to BSA veraus pH
interactions (and perhaps of the ones of the other tetracarboxylate calcium probes and chelators). To investigate this, we have carried out molecular filtration experiments on a G25 column (Fiig, 5). The column was loaded with 2.5 ml samples containing (a) 10 J.&I Indo-l alone or in the presence of (b) 0.07% BSA, pH = 7.5, (c) 0.07% BSA + 2 mM Ca2’ pH = 6.5, or (d) poly-L-lysine (O.Ol%), pH = 7.5. ‘For experiments (a), (b) and (d), the column was eluted with 20 ml H20, pH = 7.5. For experiment (c), the elution solution contained 5 mh4 Ca2+, pH = 6.5. Fractions (1 ml) were collected for fluorescence measurements. In a control experiment, eluted BSA tractions were detected by their absorbance at 280 run. In this case, an elution volume of 5.5 ml was obtained. This indicates that, due to its high molecular weight, the 70 kD protein is excluded from the inner volume of the column. When Indo- alone was used, an elution volume of 11.5 ml was obtained, corresponding to a free diffusion of the probe in the whole column. When Indo-l was in the presence of BSA or poly-L-lysine its fluorescence was found to be associated with the macromolecules. Similar
Elutlon
Volume
(ml)
Fig. 5 Elution volumes of: a solution of BSA at 0.07% (filled triangles); 10 pM Indo-l solutions containing (a) no proteins (open squares), (b) 0.07% BSA (filled squares), (c) 0.07% BSA + 2 mM Ca2’ (filled circles) or (d) 0.01% poly+lysine (open circles) on a G25 molecular filtration column
results were obtained for Fura(Bancel et al., submitted for publication). This similarity of the interaction of 1ndw-l with a protein and a synthetic polymer confums the non-specificity of the
INVESTIGATION
OF NON-CALCIUM
Wwolength Fig.
6
Spectral
analysis
of
w_aIONS
OF I~O_
(nm)
the
fluorescence
of
a single
Indo-l-treated
3T3 cell (filled circles) using the spectra of the L
(open squares:
28.6%)
Indo-l
and LM (filled circles:
and the cell autofluorescence
98.5 nM from the contributions
(+: 51.5%).
28.2%)
forms of
[Ca”]
found at
of L and LM fotms.
Weighted residues (WR) are represented
1
mobilization of the negative charges of the EGTA-like substructure of Indo-l in both calcium and proton binding that precludes an electrostatic interaction with the positive charges of the proteins. In addition, we have previously shown the mutual exclusion of proton and calcium binding by Indo-l We can then propose a general model [Ill. accounting for Indo-l behaviour in complex biological solutions (Scheme 1). According to this model, and using the acidity constant previously measured, the ratios of the quantum yields of L and LH forms, and the contribution of these two forms to the spectrum of Figure 4, the pH of this solution was calculated at 6.47 while the potentiometrically-measured value was 6.51.
L”x
x2 _ 1.279.
(x 100)
interaction and its possible electrostatic nature, as well as the role of the basic amino acids in the binding of Indo-1. Inhibition of Indo-l binding to BSA was almost total in the presence of 5 mM Ca2’, even at pH 6.5 which is the optimum pH for Indo-1-BSA interaction (inset of Fig. 4). We can deduce from this result that the LM form of Indo-l does not interact with the proteins. This does not agree with the model proposed by Konishi et al. 1141. In this paper, the proposed model is based on the absence of competition of EGTA for the binding of Fura- at pH = 7. Assuming that the EGTA-like substructure is not involved in the interaction with positive charges, it was concluded that a Furamolecule bound to a protein could still bind to Ca2+. In fact, the comparison of the binding properties of the fully deprotonated Fura- or Indo-l with the ones of EGTA at pH 7 may not be relevant. As the first two pKa’s of EGTA are measured at 9.46 and 8.85 141, the most represented form at pH = 7 will be HzEGTA2-. The structure of I-IaEGTA2- might be compared to that of the LH form of Indo-1, which is not likely to bind to proteins though it is still partially deprotonated. All these observations lead to the conclusion that only the L form of Indo-l is able to bind to
LPLH
tl +P
LP
Scheme
1
Application to the analysis of the fluorescence emitled by a single 3T3 fibroblast treated with Indo-IIAM
We present here results obtained on the same fluorescence spectrum of a single 3T3 fibroblast treated with Indo-l/AM. This allowed us to compare different sets of spectra in order to analyse the fluorescence emitted by a single cell, according to the physico-chemical properties observed in solution. Spectral evidence for the existence of more than two forms of Indo-l
whole fluorescence spectrum of an The Indo-l-treated single living cell was recorded using the microspectrofluorometer. An attempt to resolve
CELL CALCIUM
66
WR
observations am in accordance with the findings of Owen et al. who reported such spectral distortions in Indo-l-treated cell suspensions [8,9]. Introductionof LH spectrum in the equationset
Wavelength (nm)
Spectral analysis of the flwxescence of a single, Fig. 7 Indo-l-treated 3T3 cell (filled circles) using the spectra of the L (open squares: 18.2%). Lh4 (f&d squares: 28.7%) and LH (open triangles: 40.4%) forms of Indo-l and the cell autofluorescence (t: 14.7%). pH found at 6.25 from the contributions of L and L?l forms, [Ca*‘] at 158 r&l from the contributions of L and LM forms. x* - 1.061. Weighted residues (WR) an? represented (x100)
this fluorescence spectrum using the spectra of the calcium-bound Indo-1, the calcium-free Indo-l and the cell autofluorescence was unsuccessful, as illustrated in Figure 6. The results of the resolution has to be rejected since (a) both the graphic and numeric estimators of the fit between the resolution and the experimental spectrum indicated severe distortions in the determination and (b) the intensity found for the cell autofluorescence (Imax = 1000 counts) was overestimated, as compared to autofluorescence of control cells. As the resolution method we used calculates the best linear combination of the characteristic spectra [lo], the contribution of the cell autofluorescence was, therefore overvalued (furthermore, the representation of intrinsic fluorescence is very noisy: in this case, the contribution found for this form is 5-fold higher than the usual intensity of the cell autofluorescence and the noise is therefore increased by the same factor of 5). This overvaluation could be due to the fluorescence, around 450 run, of one or several fluorochromes
related to Indo-1, whose spectrum
have not been introduced in the equation set. These
A new attempt to resolve the fluorescence spectra of Indo-l-treated 3T3 cells using the characteristic spectra of L, LM, LH forms of Indo-l and the cell autofluomscence is shown in Figure 7. The intracellular pH values @H = 6.25, as calculated using the participation of L and LH spectra to the total fluorescence, and the rate of quantum yield and pKa as they were determined in aqueous solution), were far out of the biological rauge [15]. x2 was improved (1,061 vs 1.279), while the distrib$on of WR was not totally homogenous. Furthermore, the huge contribution to the total fluorescence calculated for LH indicated that another form of Indo-l was emitting within the same spectral region. Introductionof LP spectrum in the equationset
Using the fluorescence spectra of the L, LM, LH, and LP forms of Indo-l and the cell
4bO WavrlDngth
(nm)
Ng. 8 Spectral analysis of the fluo~~~~nce of a single, Indo-l-treated 3T3 cell (filled circles) using the spectra of the L (open squares: 23.4%), LM (filled squares: 17.0%). LH (open triangles: 7.7%) and LP (filled triangles: 39.4%) forms of Ind+l and the cell autofluorescence (+: 142%). pH determined at 7.08 from the contributions of L and LH forms, [Ca*‘] at 73 nM from the contributions of L and LM forms. 2 - 1.017. Weighted nxidues (WR) are represented (x 100)
INVESTIGATION
OF NON-CALCIUM
INTERACTIONS
OF INDO-
autofluorescence, it has been possible to resolve the fluorescence spectrnm of single 3T3 cells treated with Indo-l/AM. In Figure 8 we show the resolution of the same experimental spectrum as in Figures 6 and 7, but the characteristic spectrum LP was introduced in the system in addition to the previous set. Both graphic and numeric estimators wem improved. With this spectral analysis it has been possible to quantify each form of Indo-l as well as the cell autofluorescence. This allowed us to calculate for the single cell: pHi = 7.08, [Ca2+]i = 73 x lo-’ M, using the dissociation constants and the ratios of quantum yield obtained in aqueous solution [ill. Intracellular calibration of these parameters will be necessary but the pH value determined in this example is in agreement with the results of Lahmy et al. obtained in the same cell line using DCH (2,3dicyanohydroquinone) as a pH probe [16]. Test of the contributionof unhy&olysedIndo-l/AM Some authors [5-71 have suggested that unhydrolysed acetoxymethyl esters could be responsible for a calcium-insensitive pool of intracellular Indo-1. We have tested this hypothesis in 3T3 cells by including the characteristic spectrum of Indo-l/AM in the set of characteristic spectra presented in Figure 8. The contribution of this spectrum was so small (around 1% of total fluorescence) that it was below the sensitivity of the method. This may change from one cell line to another and 3T3 cells exhibit a high esterase activity towards other esteritied fluorescent probes [16]. However, due to the very low quantum yield of Indo-l/AM (YIJYAM = 55, in our conditions), a cell cannot be expected to exhibit simultaneously a high fluorescence and a large amount of unhydrolysed ester.
Conclusions Our observations of non-calcium interactions of Indo-l are consistent with numerous reports of anomalous intracellular spectra of this family of calcium probes [5-g, 11, 131. We point out the existence of two new forms of Indo-1: a protonated form (LH) and a form interacting with proteins (LP)
67
that should be taken into consideration for a interpretation of intracellular quantitative fluorescence of the probe. The fluorescence of these two forms peaks in the same spectral region (i.e. 455 and 438 run for LH and LP, respectively) and spectra of the four forms of Indo-l extensively overlap. A microspectrofluorometric approach is likely to quantify the contribution of each spectrum to the complex fluorescence spectrum emitted by a single cell. From these contributions, it is possible to calculate the molar fraction of each fhiorescent compound (using the ratio of their quantum yields). Taking into account the contribution of the fluorescence of LH and LP to the intracellular fluorescence, allows an accurate evaluation of intracellular calcium concentrations. Furthermore, intracellular pH can be simultaneously evaluated using the contributions of L and LH forms. This multiparametric use of Indo-l requires further intracellular assessment and calibration that will be presented in another paper. This study shows that four different forms of the fluorescent calcium probes are in equilibrium under physiological conditions. When the ratio of fluorescence intensities at two different wavelengths is used to monitor calcium changes within a cell, we have two equations to solve a system of four unknowns. Even so, intracellular calibration using ionophores is likely to provide an empirical relation between the ratio value and the actual calcium When the experiment itself is concentration. performed, one must then assume that calcium concentration is the only varying parameter. The main consequence is that the ‘ratio’ method can only provide semi-quantitative data on intracellular calcium concentrations when intracellular pH is modified. In many cases only qualitative information is required to elicit whether or not a given biological compound has an effect on cellular calcium levels. When experiments are performed with this aim, pH measurements could be necessary in order to check what parameter @Hi or [Ca2+]i) is modified when ratio changes am observed. This is particularly true if the changes observed in the ratio However, numerous are of low amplitude. biological events are known to be, or thought to be triggered by, or at least occur concomitantly with, changes in [Ca’+]i or pH1 [17, 181. It seems,
cBLLcALcIuM
68
therefore, that simultaneous evaluation of pH and calcium concentration will be of some interest in many biological fields. Thus, Indo-l appears to be a potential multiparametric fluorescent pH and calcium probe, but such a use requires that the fluorescence measurements are performed at more than two wavelengths.
Acknowledgements This work was supported by INSERM (Contract No. CRJ3399015) and by a fellowship of the ‘L,igue Nationale Fraqaisc contre le Cancer’. We thank Dr Gillian Hull for reading the manuscript.
References 1. Campbell AK. (1983) Intracellular Calcium: Its Universal Role as Regulator. Cbichester, Wiley. 2. Tsien RY. (1984) New calcium indicators and buffers with high selectivity against magnesium and protons: design, synthesis and properties of prototype structun~. Biochemistry, 19,2396-2404. 3. Grynkiewicz G. Poenie M. Tsien RY. (1985) A new generation of Ca’+ indicators with greatly improved fluomscence properties. J. Biol. C&m., 260,3440-3450. 4. Fabiato A. Fabiatc F. (1979) Calculator pmgtams for computing the composition of the solutions containing multiple metals and ligands used for experiments in skinned muscle cells. J. Physiol. (Paris), 75.463-505. 5. Lilckoff A. (1986) Measuring cytosolic free calcium concentration in endothelial cells with Indo-1: the pitfall of using the ratio of two fluorescence intensities recorded at different wavelengths. Cell Calcium, 7,233-248. 6. Highsmith S. Bloebaum P. Snowdone KW. $786) Sarcoplasmic teticuhtm interacts with the Ca indicator precursor Fura-2-AM. B&hem. Biophys. Res. Common., 138, 1153-1162. 7. Scanlon M. Williams DA. Fay FS. (1987) A Ca’+-insensitive form of Fum-2 associated with polymorphonuclear leukocytes. J. Biol. Chem., 262,
6308-6312. 8. Owen CS. Shuler RL. (1989) Spectral evidence for non-calcium interactions of intnue1lular Indo-1. Biochem. Bicphys. Res. Commun., 163,328-333. 9. Owen CS. Sykes NL. Shuler RI.,. Ost D. (1991) Non-calcium environmental sensitivity of inuacellular In&l. Anal. B&hem., 192,142-148. 10. Salmon JM. Vigo J. Viallet P. (1988) Resolution of complex fluorescence spectra recorded on single unpigmented living cells using a computerised method. Cytometry, 9.2532. Il. Bancel F. Vigo J. Salmon JM. Viallet P. (1990) Acid-base and calcium binding properties of the fluorescent calcium indicator Indo-1. J. Photo&m. Photobiol., 53,397-409. 12. Lattanxio FA (1990) The effects of pH and temperature on fluorescent calcium indicators as de&mined with cbelex-100 and EDTA buffer systems. B&hem. Biophys. Res. Commun., 171,102-108. 13. Blatter LA. Wier WG. (1990) Intracelhtlar diffusion, binding and compattmentalixation of the fluomscen t calcium indicatora Indo-l and Fura-2. Biophys. J., 58, 1491-1499. 14. Konishi M. Olson A. Holhngworth S. Baylor SM. (1988) Myoplasmic binding of Fura- investigated by steady-state fluomscence and absorbance meas umments. Biophys. J.. 54, 1089-l 104. 15. Roes A. Baton WF. (1981) IntnzelhdarpH. Physiol. Rev., 61.400-408. 16. Lahmy S. Salmon JM. Vigc J. Viallet P. (1989) pHl and DNA content modifications after ADR treatment in 3T3 fibroblasts. A micmfluommetric approach. Anti-r Res., 9,929-936. 17. Siskind MS. McCoy CH. Chobanian A. Schwartz JH. (1989) Regulation of intracellular calcium by cell pH in vascular smooth muscle cells. Am. J. Physiol., 256, C23CC240. 18. Dickens CJ. Gillespie JI. Greenwell JR Hutchinson P.(l990) Relationship between intracellular pH (pHi) and calcium (Ca2’l) in avian heart fibtoblasts. Hxp. Cell Res., 187,39-46.
Please send reprint requests to : Dr Fr&ric Bancel, Groupe de Microfluorim&rie Quantitative, URA CNR,S 1289, Universit6 de Perpignan, 52 Avenue de Villeneuve. F-66860 Perpignan cedex, France Received : 17 April 1991 Revised : 8 Cktober 1991 Accepted : 9 October 1991
(1882) 13, s-88
Q Longman Group UK Lid 1982
Microspectrofluorometry as a tool for investigation of non-calcium interactions of Indo-l F. BANCEL, J-M. SALMON, J. VIGO and P. WALLET Groupe de Microfloorim&rie Quantitative, UffA CNRS 1289, Universit6 de Pefpignan, Petpignan, France Abstract - Indo-l is a fluorescent calcium probe used to measure intracellular free calcium concentrations. These measurements are often performed by comparing the fluorescence intensities of Indo-l-treated cells at two selected wavelengths corresponding to the maxima of the fluorescence spectra of the calcium-bound and calcium-free forms. In this study, we used an optical multichannel anaiyser to numerise the fluorescence emitted by a single cell. A computerised resolution of numerlsed spectra was used on Intracellular Indo-i fluorescence. Calculation of numerical and graphic estimators allows us to evaluate the fit of the resolution. Dlfferent sets of characteristic spectra were compared using this method. It appeared that no linear combination of the two known forms of Indo-l and of the cell autofluorescence can fit with spectra of Indo-l-treated cells. In additlon, a study of the physico-chemical properties of Indo-l shows the existence of two other forms of the molecule: a protonated form (maximum emission at 455 nm) and a form in interaction with proteins (maximum emission at 438 nm). Taking into account the contribution of these two new forms leads to an improved spectral resolution of the fluorescence of Indo-l-treated living cells and, therefore, improves calcium measurements. Moreover, quantlficatlon of the amount of the protonated form of Indo-l allows a measurement of intracellular pH at the same time as calcium determination.
Abbreviations : Indo-1, lH-indole-6-carboxylic acid, 2-(4-bis-(carboxymethyl)amino-3-(2-(2-(bis~boxymethyl) amino-S-methylphenoxy)ethoxy)phenyl; Indo-l/AM, pentaacetoxymethyl ester of Indo-1; Fura-2, 1-[2-(karboxyoxazol-2-yl)-6-aminobenzofuran-S-oxy]-2-(%unino-5’ methylphenoxy)-et-N,N.N’,N’-tetmacetic acid; EGTA,
ethyleneglycol-bis-@-aminoethy1 ether)-N,N,N’,N’-tetraacetic acid; BAFTA, 1,2-bis(o-aminophenoxy)-ethaneN,N,N’.N’-tetraacetic acid; DMSO. diiethylsulfoxidt; DMEM, Dulbecco’s modified Eagle’s essential minii medium; BSA. bovine serum albumin, [Ca*+]i, inea~ellular calcium concentration; and pH,, intracelhrlsr pH
The use of fluorescent probes is often the most convenient technique for measuring intracellular ionic concentrations. Numerous probes have been designed with this aim, particularly in the case of the calcium ion, whose role as cellular messenger has been known for a long time [l]. Several generations of fluorescent calcium probes have been derived from the structure of the non-fluonzscent calcium chelator EGTA [2, 31. Indo-l belongs to the second generation of these tetracarboxylate calcium indicators. One of its most athractive properties is a 70 nm spectral shift of its 59
60
emission spectrum after calcium binding. This property has led to a widely used practice of calculating intracellular calcium concentration from the ratio of the fluorescence intensities obtained at two selected wavelengths using interference filters. The respective characteristic fluorescence spectra of the calcium-free and calcium-bound forms can be easily obtained, at pH > 7.5. in calcium-&e and high calcium solutions, respectively. The dissociation constant of the probe for the calcium ions is usually obtained using EGTA as a calcium buffer. The apparent Kd of EGTA for Ca2+depends on pH and the free calcium concentration of the solutions must be computed for each pH value [3]. Several authors have encountered problems with the analysis of fluorescence of the tetracarboxylate family of calcium probes in living cells [5-g]. It was reported that part of the fluorescence emitted by cells treated with Indo-l or Fura- could be attributed to neither the calcium-bound nor the calcium-free forms of the probes. The aim of this study was to connect such intracellular observations with physico-chemical properties of Indo-l and to provide a comprehensive basis for further utilization of non-calcium interactions of the probe. In order to analyse intracellular Indo-l fluorescence, we have used a technique based on computerised resolution of numerised fluorescence spectra [lo]. This technique allows us to calculate the contribution of several fluorescent compounds in a mixture. Investigation of intracellular fluorescence of Indwl was possible using a multichannel microspectrofluorometer for a numerical recording of fluorescence spectra emitted by single cells. After each resolution, calculation of fit estimators allows us to compare several sets of spectra in terms of accuracy and to determine whether or not experimental fluorescence spectra are correctly analysed. This study combines physico-chemical investigations of (a) acid-base properties [ll] and (b) interaction of Indo-l with proteins, and intracellular detection of the related fluorescence spectra in 3T3 fibroblasts. Results were obtained by spcctrofluorometry, for studies of the properties of the probe in solution, and multichannel microspectiofluorometry, for studies in single living 3T3 cells.
CELL CALCIUM
Materials and Methods Experimentsin solution
Indo-l, Indo-l/AM and BAPTA were purchased from Molecular Probes (Eugene, OR, USA). A stock solution of 1 mh4 was prepared in ultra-pure water (18.6 MQ 0.005 ppm Ca, Merck, used in all experiments in solution). Samples (10 ml) of 10% BSA (Sigma) were dialysed three times against 0.5 1 of water, in order to remove calcium ions. The actual concentration of albumin stock solution was assayed at 1 mM (7% w/v) using the Lowry method. Potentiomettic determination of the pH of all solutions was performed using a Taccussel PHN 81 digital pH-meter (+ 0.01 pH unit) with a Xerolyt electrode (Ingold). Molecular filtration experiments were performed using a Pharmacia PDlO (G25) column previously equilibrated with 10 ml H20/KOH, pH = 7.5. Poly-r.-lysine (MW = 300007OCOO) was purchased from Sigma. Fluorescence spectra were digitally recorded in a 1 cm width quartz cuvette with a Jobin-Yvon JY3D spectrofluorometcr interfaced with a Tandon AT 286 microcomputer. Excitation wavelength was 340 nm and spectra were recorded between 360 and 600 nrn, Fluorescence spectra were analysed using the characteristic spectra of each participating form of Indo-l and a resolution method developed in this laboratory [ 101as explained below. Cell culture and loading
3T3 mouse fibroblasts (Flow Laboratories) were routinely cultured in DMEM (Flow Laboratories) with 10% decomplemented fetal calf serum (FCS, Gibco) and 2 mM L-glutamine (Flow Laboratories) at 37°C in 25 cm2 flasks. For experiments, cells were seeded in Sykes-More chambers at 4000060000 cells per chamber at 37”C, 5% Co;?. Cells were incubated for 1 h in Dh4EM containing 5 p.M Indel/AM, 0.5% DMSO and washed three times with cold phosphate-saline buffer solution. The Sykes-Moore chambers were then placed on the thermostated stage (37°C) of the rnicrospcctrofluoromcter.
61
INVESTIGATION OF NON-CALCIUM INTERACTIONS OF INDO-
Fig. 1 Equation set generated for the determination of the composition of complex mixtures of fluorescent compounds. Diagonal terms are in bold characters and delimited by the solid line
Microspectrofluorometryon single cells The microspectrofluorometer is composed of an inverted microscope (Leitz) connected to an Optical Multichannel Analyser (OMA, Princeton Applied Research Corp.) equipped with a silicon intensified target (SIT) as detector. The excitation wavelength was selected at 337 MI from the emission of a xenon lamp with a monochromator. Spectra were recorded with a signal to noise ratio greater than 30 between 350 and 640 nm (400 channels, 0.72 nm per channel). Data were transferred to a PDP 11/73 microcomputer for storage and (Plessey ) calculations.
ITL is the intensity of experimental spectrum at the wavelength h, I$. is the intensity of the i* characteristic spectrum, and ai is a coefficient proportional to its contribution in the mixture. When a fluorescence speclrum emitted from a single cell is recorded, the amount of photons collected per frame scan (32 x 10m3s) for each channel is low. The signal to noise ratio can be increased by (a) increasing the number of accumulations of the signal and/or (b) integration of the fluorescence spectnun The first process is time consuming (3 s for the summation of 100 fluorescence spectra) while the second leads to the loss of spectral information. An integration of Equation 1 over the whole spectral domain gives:
Numerical resolutionoffluorescence spectra This method takes advantage of the fact that if a complex fluorescence spectrum results from the fluorescence of N chemicals, the intensity at each wavelength can be described as a linear combination of the intensities of each component at the same wavelength. This can be expnzssed as follows: N
ITA = C i=l
ai Ii,h
Eq. 1
c I’h
k=l.lJ
=
C ai C i=l
Ii,1
Eq. 2
h=b
A set of N independent equations is then generated from Equation 2 using N modulating functions depending on h in order to restore the spectral information. Theoretically. any non-random function cDj,h is suitable for this purpose. The originality of the method we have developed lies in (a) the use of the characteristic spectmm Ij,L of each
62
CELL CALCIUM
fluorescent entity as a modulating function and (b) the weighting of this function by the complex spectrum ITh, i.e.:
@j,h= ITk Ij,k
Eq. 3
l
The major advantages of generating tiquation sets in this way am (a) to maximise in each lquation the relative weight of only one component (this procedure generates a system more diagonai than the unweighted one and therefore increases the value of the determinant) and (b) to vary the weight of the data with respect to their intensity. The tinal equation set (Fig. 1) is then resolved using a classical routine. Once the results are obtained (i.e. values for al, az, ... aN), residuals are calculated as the difference between a reconstructed complex spectrum and the experimental one. Weighted residuals (WR) are then calculated as the ratio of the residuals versus the absolute value of the random noise superimposed on the signal (for more details, see [lo]). The fit between the result of the resolution and the experimental spectrum is represented by a graphic plot of the WR and a numeric estimator is calculated as the chi-square of the WR. For an optimally resolved spectrum WR should be randomly distributed about the zero value and chi-square should be close to 1. This technique allows us to calculate optimal contributions of several fluorescence spectra to the
400
500 Wavelength
Fig. 2 Changes in Indo-l fluomcence
600
(nm)
spectra in response to pH changes. Inset: titration curve accounting for the equilibrium between the L and LH forms
complex spectrum resulting from a mixtune of several chemicals. This implies that the characteristic fluorescence spectrum of each fluorescent compound is introduced in the resolution system The user must then record the characteristic spectrum of each form and store it in a library of spectra prior to analysis. R~ults i ud Discussion
Constructronof a library offluorescence spectra Calcium-@e and calcium-boundforms of Indo-l
The calch n-free (L) and calcium-bound (LM) spectra wa ; obtained by recording the fluorescence of solutions at (a) 50 pM Indwl, 2 r&I EGTA. pH = 8.0 and (b) 5 pM Indo-l with a saturating concentration of CaCh, pH 7.5 after elimination of a precipitate of Ca(OH)z by centrifugation, respectively. Indo-l interaction with calcium ions was described in detail in [ll]. Among the physico-chemical parameters which can act on Indo-l-fluorescence, pH has to be studied fast [12]. Acid-baseproperties of Indo-l
We have previously published a study of the acidbase properties of Indo-l in aqueous solution [ll]. In this study, a protonated form of Indo-l (LH) was detected, with an emission spectrum maximum at 455 nm. Figure 2 shows the changes in Indo-l fluorescence spectra in response to pH changes. A titration curve can be plotted from the spectroscopic data (inset), the logarithm of the acidity constant @Ka) was calculated at 6.18. The characteristic fluorescence spectrum of LH was obtained by subtracting the calculated contribution of the L spectrum from the spectrum of a solution containing 50 p.M Indo-l and 2 mM BARTA, pH = 6.0. This subtraction was performed according to the values of pKa (6.18) and the rate of quantum yields (R = YWYLH = 0.5) previously measured [ll]. interactionof Indo- withproteins
In order to complete the previous study [l 11, we have recorded the fluorescence spectrum of Indo-l in fetal calf serum and in DMEM, with 5 mM EGTA. We observed an unexpected fluorescence
INVESTIGATION
OF NON-CALCIUM
INTERACTIONS
63
OF INDO-
the fluorescence spectrum of the pentaacetoxymethyl ester of Indo-l fIndo-l/AM, 50 p&l in Hanks’ Balanced Salt Solution (Gibco), 5% DMSO] and (b) the intrinsic cellular fluorescence of untreated 3T3 fibroblasts (in our excitation conditions, i.e. 337 run, this fluorescence is mainly the emission of NAD(P)H). Obtaining a interactions
360
400
440
480
520
Wavelength Fig. 3 Changes varying Scatchard
in Indo-l
concentrations plot obtained
fluorescence
of
dialysed
580
800
(nm)
spectra in the presence of BSA,
from the resolution
pH
-
7.5.
Inset
of the experimental
spectra (v = [LP]/[BSA]total)
peak around 440 nm (data not shown), out of the pH range where LH is found (i.e pH > 7.5). We then studied the fluorescence of Indo-l solutions at pH = 7.5, in the presence of varying concentrations of dialysed BSA. We obtained an iso-emissive point (Fig. 3) with a spectrum attributed to a form of Indo-l interacting with BSA (maximum emission = 438 run, called LP in the following), and the L spectrum (maximum emission = 485 run), in proportions depending on BSA concentration. Each composite spectrum was then analysed using our resolution method. A Scatchard diagram plotted from these results (inset of Fig. 3), showed a high affinit site on BSA (dissociation constant: Kd c: Y 3.10- M) and several other sites of lower affinity. Interaction of Indo-l with proteins was not limited to BSA: we obtained the same spectral changes when Indo-l was in the presence of proteins as different as histones and trypsin. This could indicate that Indo-l is likely to interact with a great number of cellular proteins. This interaction could be mediated by a class of amino-acid rather than a specific binding site on BSA. In addition, interaction of Indo-l with intracellular proteins could be correlated with the observation of a slow-diffusing fraction of the probe by Blat&r et al. [13]. The library of fluorescence spectra was completed by recording in the same conditions (a)
comprehensive
pH-Dependence proteins
model
of
Indo-l
of the interaction of Indo-l
with
To study the pH-dependence of the interaction of Indo-l with proteins, the evolution of the molar fraction of LP in response to pH changes was followed in a solution containing 20 pM Indo-l and 5 pM dialysed BSA. Fluorescence spectra were recorded for pH values varying between 5.5 and 9.5 and the composition of the solutions at each pH value was determined using our resolution method. An example of such a resolution is shown in Figure 4. The molar fraction of the LP form was calculated from the contribution of LP spectrum (R = WYLP = 0.56). The variation of this molar fraction versus pH is plotted in the inset These results indicate an increase of the interaction of Indo-l with BSA as the pH decreases in the range 9-6.5. Further acidification leads to a decrease of the interaction. Two conclusions could be deduced from these observations: (a) according to the results reported by Konishi et al. on the interaction of Fura- with proteins [14], the type of interaction could be electrostatic and mediated by basic amino-acids, such as arginine and lysine, that are positivelycharged when protonated @Ka in the range 8-9); and (b) the lowering of the interaction of pH values below 6.5 occurs when the protonation of the basic amino acid residues is almost complete. This could be due to the absence of interaction between the basic amino-acid residues and the LH form of Indo-1. Chromatographic interaction
study
of
Indo-llproteints)
It would be interesting to know which form(s) of Indo-l is (are) likely to bind to proteins in order to determine a comprehensive model of Indo-l
64
CELL CALCIUM
7
0
PH
450
550 Wavelength
(nm)
Fig. 4 pH dependence of the interaction of Indo-l with BSA: example of resolu~on of the fluorescence speettum of a solution containing 20 @I Indo-1, 5 ph4 BSA, pH - 6.5 (filled circles) using the spectra of the L (open squares: 12.6%). Lh4 (filled squares: 17.4%). LH (open triangles: 18.3%) and LP (filled triangles: 61.1%) forms of Indo-1. x2 - 0.989. Weighted residues (WR) am represented (x 300). Inset: plot of the molar fraction of Indo-l bound to BSA veraus pH
interactions (and perhaps of the ones of the other tetracarboxylate calcium probes and chelators). To investigate this, we have carried out molecular filtration experiments on a G25 column (Fiig, 5). The column was loaded with 2.5 ml samples containing (a) 10 J.&I Indo-l alone or in the presence of (b) 0.07% BSA, pH = 7.5, (c) 0.07% BSA + 2 mM Ca2’ pH = 6.5, or (d) poly-L-lysine (O.Ol%), pH = 7.5. ‘For experiments (a), (b) and (d), the column was eluted with 20 ml H20, pH = 7.5. For experiment (c), the elution solution contained 5 mh4 Ca2+, pH = 6.5. Fractions (1 ml) were collected for fluorescence measurements. In a control experiment, eluted BSA tractions were detected by their absorbance at 280 run. In this case, an elution volume of 5.5 ml was obtained. This indicates that, due to its high molecular weight, the 70 kD protein is excluded from the inner volume of the column. When Indo- alone was used, an elution volume of 11.5 ml was obtained, corresponding to a free diffusion of the probe in the whole column. When Indo-l was in the presence of BSA or poly-L-lysine its fluorescence was found to be associated with the macromolecules. Similar
Elutlon
Volume
(ml)
Fig. 5 Elution volumes of: a solution of BSA at 0.07% (filled triangles); 10 pM Indo-l solutions containing (a) no proteins (open squares), (b) 0.07% BSA (filled squares), (c) 0.07% BSA + 2 mM Ca2’ (filled circles) or (d) 0.01% poly+lysine (open circles) on a G25 molecular filtration column
results were obtained for Fura(Bancel et al., submitted for publication). This similarity of the interaction of 1ndw-l with a protein and a synthetic polymer confums the non-specificity of the
INVESTIGATION
OF NON-CALCIUM
Wwolength Fig.
6
Spectral
analysis
of
w_aIONS
OF I~O_
(nm)
the
fluorescence
of
a single
Indo-l-treated
3T3 cell (filled circles) using the spectra of the L
(open squares:
28.6%)
Indo-l
and LM (filled circles:
and the cell autofluorescence
98.5 nM from the contributions
(+: 51.5%).
28.2%)
forms of
[Ca”]
found at
of L and LM fotms.
Weighted residues (WR) are represented
1
mobilization of the negative charges of the EGTA-like substructure of Indo-l in both calcium and proton binding that precludes an electrostatic interaction with the positive charges of the proteins. In addition, we have previously shown the mutual exclusion of proton and calcium binding by Indo-l We can then propose a general model [Ill. accounting for Indo-l behaviour in complex biological solutions (Scheme 1). According to this model, and using the acidity constant previously measured, the ratios of the quantum yields of L and LH forms, and the contribution of these two forms to the spectrum of Figure 4, the pH of this solution was calculated at 6.47 while the potentiometrically-measured value was 6.51.
L”x
x2 _ 1.279.
(x 100)
interaction and its possible electrostatic nature, as well as the role of the basic amino acids in the binding of Indo-1. Inhibition of Indo-l binding to BSA was almost total in the presence of 5 mM Ca2’, even at pH 6.5 which is the optimum pH for Indo-1-BSA interaction (inset of Fig. 4). We can deduce from this result that the LM form of Indo-l does not interact with the proteins. This does not agree with the model proposed by Konishi et al. 1141. In this paper, the proposed model is based on the absence of competition of EGTA for the binding of Fura- at pH = 7. Assuming that the EGTA-like substructure is not involved in the interaction with positive charges, it was concluded that a Furamolecule bound to a protein could still bind to Ca2+. In fact, the comparison of the binding properties of the fully deprotonated Fura- or Indo-l with the ones of EGTA at pH 7 may not be relevant. As the first two pKa’s of EGTA are measured at 9.46 and 8.85 141, the most represented form at pH = 7 will be HzEGTA2-. The structure of I-IaEGTA2- might be compared to that of the LH form of Indo-1, which is not likely to bind to proteins though it is still partially deprotonated. All these observations lead to the conclusion that only the L form of Indo-l is able to bind to
LPLH
tl +P
LP
Scheme
1
Application to the analysis of the fluorescence emitled by a single 3T3 fibroblast treated with Indo-IIAM
We present here results obtained on the same fluorescence spectrum of a single 3T3 fibroblast treated with Indo-l/AM. This allowed us to compare different sets of spectra in order to analyse the fluorescence emitted by a single cell, according to the physico-chemical properties observed in solution. Spectral evidence for the existence of more than two forms of Indo-l
whole fluorescence spectrum of an The Indo-l-treated single living cell was recorded using the microspectrofluorometer. An attempt to resolve
CELL CALCIUM
66
WR
observations am in accordance with the findings of Owen et al. who reported such spectral distortions in Indo-l-treated cell suspensions [8,9]. Introductionof LH spectrum in the equationset
Wavelength (nm)
Spectral analysis of the flwxescence of a single, Fig. 7 Indo-l-treated 3T3 cell (filled circles) using the spectra of the L (open squares: 18.2%). Lh4 (f&d squares: 28.7%) and LH (open triangles: 40.4%) forms of Indo-l and the cell autofluorescence (t: 14.7%). pH found at 6.25 from the contributions of L and L?l forms, [Ca*‘] at 158 r&l from the contributions of L and LM forms. x* - 1.061. Weighted residues (WR) an? represented (x100)
this fluorescence spectrum using the spectra of the calcium-bound Indo-1, the calcium-free Indo-l and the cell autofluorescence was unsuccessful, as illustrated in Figure 6. The results of the resolution has to be rejected since (a) both the graphic and numeric estimators of the fit between the resolution and the experimental spectrum indicated severe distortions in the determination and (b) the intensity found for the cell autofluorescence (Imax = 1000 counts) was overestimated, as compared to autofluorescence of control cells. As the resolution method we used calculates the best linear combination of the characteristic spectra [lo], the contribution of the cell autofluorescence was, therefore overvalued (furthermore, the representation of intrinsic fluorescence is very noisy: in this case, the contribution found for this form is 5-fold higher than the usual intensity of the cell autofluorescence and the noise is therefore increased by the same factor of 5). This overvaluation could be due to the fluorescence, around 450 run, of one or several fluorochromes
related to Indo-1, whose spectrum
have not been introduced in the equation set. These
A new attempt to resolve the fluorescence spectra of Indo-l-treated 3T3 cells using the characteristic spectra of L, LM, LH forms of Indo-l and the cell autofluomscence is shown in Figure 7. The intracellular pH values @H = 6.25, as calculated using the participation of L and LH spectra to the total fluorescence, and the rate of quantum yield and pKa as they were determined in aqueous solution), were far out of the biological rauge [15]. x2 was improved (1,061 vs 1.279), while the distrib$on of WR was not totally homogenous. Furthermore, the huge contribution to the total fluorescence calculated for LH indicated that another form of Indo-l was emitting within the same spectral region. Introductionof LP spectrum in the equationset
Using the fluorescence spectra of the L, LM, LH, and LP forms of Indo-l and the cell
4bO WavrlDngth
(nm)
Ng. 8 Spectral analysis of the fluo~~~~nce of a single, Indo-l-treated 3T3 cell (filled circles) using the spectra of the L (open squares: 23.4%), LM (filled squares: 17.0%). LH (open triangles: 7.7%) and LP (filled triangles: 39.4%) forms of Ind+l and the cell autofluorescence (+: 142%). pH determined at 7.08 from the contributions of L and LH forms, [Ca*‘] at 73 nM from the contributions of L and LM forms. 2 - 1.017. Weighted nxidues (WR) are represented (x 100)
INVESTIGATION
OF NON-CALCIUM
INTERACTIONS
OF INDO-
autofluorescence, it has been possible to resolve the fluorescence spectrnm of single 3T3 cells treated with Indo-l/AM. In Figure 8 we show the resolution of the same experimental spectrum as in Figures 6 and 7, but the characteristic spectrum LP was introduced in the system in addition to the previous set. Both graphic and numeric estimators wem improved. With this spectral analysis it has been possible to quantify each form of Indo-l as well as the cell autofluorescence. This allowed us to calculate for the single cell: pHi = 7.08, [Ca2+]i = 73 x lo-’ M, using the dissociation constants and the ratios of quantum yield obtained in aqueous solution [ill. Intracellular calibration of these parameters will be necessary but the pH value determined in this example is in agreement with the results of Lahmy et al. obtained in the same cell line using DCH (2,3dicyanohydroquinone) as a pH probe [16]. Test of the contributionof unhy&olysedIndo-l/AM Some authors [5-71 have suggested that unhydrolysed acetoxymethyl esters could be responsible for a calcium-insensitive pool of intracellular Indo-1. We have tested this hypothesis in 3T3 cells by including the characteristic spectrum of Indo-l/AM in the set of characteristic spectra presented in Figure 8. The contribution of this spectrum was so small (around 1% of total fluorescence) that it was below the sensitivity of the method. This may change from one cell line to another and 3T3 cells exhibit a high esterase activity towards other esteritied fluorescent probes [16]. However, due to the very low quantum yield of Indo-l/AM (YIJYAM = 55, in our conditions), a cell cannot be expected to exhibit simultaneously a high fluorescence and a large amount of unhydrolysed ester.
Conclusions Our observations of non-calcium interactions of Indo-l are consistent with numerous reports of anomalous intracellular spectra of this family of calcium probes [5-g, 11, 131. We point out the existence of two new forms of Indo-1: a protonated form (LH) and a form interacting with proteins (LP)
67
that should be taken into consideration for a interpretation of intracellular quantitative fluorescence of the probe. The fluorescence of these two forms peaks in the same spectral region (i.e. 455 and 438 run for LH and LP, respectively) and spectra of the four forms of Indo-l extensively overlap. A microspectrofluorometric approach is likely to quantify the contribution of each spectrum to the complex fluorescence spectrum emitted by a single cell. From these contributions, it is possible to calculate the molar fraction of each fhiorescent compound (using the ratio of their quantum yields). Taking into account the contribution of the fluorescence of LH and LP to the intracellular fluorescence, allows an accurate evaluation of intracellular calcium concentrations. Furthermore, intracellular pH can be simultaneously evaluated using the contributions of L and LH forms. This multiparametric use of Indo-l requires further intracellular assessment and calibration that will be presented in another paper. This study shows that four different forms of the fluorescent calcium probes are in equilibrium under physiological conditions. When the ratio of fluorescence intensities at two different wavelengths is used to monitor calcium changes within a cell, we have two equations to solve a system of four unknowns. Even so, intracellular calibration using ionophores is likely to provide an empirical relation between the ratio value and the actual calcium When the experiment itself is concentration. performed, one must then assume that calcium concentration is the only varying parameter. The main consequence is that the ‘ratio’ method can only provide semi-quantitative data on intracellular calcium concentrations when intracellular pH is modified. In many cases only qualitative information is required to elicit whether or not a given biological compound has an effect on cellular calcium levels. When experiments are performed with this aim, pH measurements could be necessary in order to check what parameter @Hi or [Ca2+]i) is modified when ratio changes am observed. This is particularly true if the changes observed in the ratio However, numerous are of low amplitude. biological events are known to be, or thought to be triggered by, or at least occur concomitantly with, changes in [Ca’+]i or pH1 [17, 181. It seems,
cBLLcALcIuM
68
therefore, that simultaneous evaluation of pH and calcium concentration will be of some interest in many biological fields. Thus, Indo-l appears to be a potential multiparametric fluorescent pH and calcium probe, but such a use requires that the fluorescence measurements are performed at more than two wavelengths.
Acknowledgements This work was supported by INSERM (Contract No. CRJ3399015) and by a fellowship of the ‘L,igue Nationale Fraqaisc contre le Cancer’. We thank Dr Gillian Hull for reading the manuscript.
References 1. Campbell AK. (1983) Intracellular Calcium: Its Universal Role as Regulator. Cbichester, Wiley. 2. Tsien RY. (1984) New calcium indicators and buffers with high selectivity against magnesium and protons: design, synthesis and properties of prototype structun~. Biochemistry, 19,2396-2404. 3. Grynkiewicz G. Poenie M. Tsien RY. (1985) A new generation of Ca’+ indicators with greatly improved fluomscence properties. J. Biol. C&m., 260,3440-3450. 4. Fabiato A. Fabiatc F. (1979) Calculator pmgtams for computing the composition of the solutions containing multiple metals and ligands used for experiments in skinned muscle cells. J. Physiol. (Paris), 75.463-505. 5. Lilckoff A. (1986) Measuring cytosolic free calcium concentration in endothelial cells with Indo-1: the pitfall of using the ratio of two fluorescence intensities recorded at different wavelengths. Cell Calcium, 7,233-248. 6. Highsmith S. Bloebaum P. Snowdone KW. $786) Sarcoplasmic teticuhtm interacts with the Ca indicator precursor Fura-2-AM. B&hem. Biophys. Res. Common., 138, 1153-1162. 7. Scanlon M. Williams DA. Fay FS. (1987) A Ca’+-insensitive form of Fum-2 associated with polymorphonuclear leukocytes. J. Biol. Chem., 262,
6308-6312. 8. Owen CS. Shuler RL. (1989) Spectral evidence for non-calcium interactions of intnue1lular Indo-1. Biochem. Bicphys. Res. Commun., 163,328-333. 9. Owen CS. Sykes NL. Shuler RI.,. Ost D. (1991) Non-calcium environmental sensitivity of inuacellular In&l. Anal. B&hem., 192,142-148. 10. Salmon JM. Vigo J. Viallet P. (1988) Resolution of complex fluorescence spectra recorded on single unpigmented living cells using a computerised method. Cytometry, 9.2532. Il. Bancel F. Vigo J. Salmon JM. Viallet P. (1990) Acid-base and calcium binding properties of the fluorescent calcium indicator Indo-1. J. Photo&m. Photobiol., 53,397-409. 12. Lattanxio FA (1990) The effects of pH and temperature on fluorescent calcium indicators as de&mined with cbelex-100 and EDTA buffer systems. B&hem. Biophys. Res. Commun., 171,102-108. 13. Blatter LA. Wier WG. (1990) Intracelhtlar diffusion, binding and compattmentalixation of the fluomscen t calcium indicatora Indo-l and Fura-2. Biophys. J., 58, 1491-1499. 14. Konishi M. Olson A. Holhngworth S. Baylor SM. (1988) Myoplasmic binding of Fura- investigated by steady-state fluomscence and absorbance meas umments. Biophys. J.. 54, 1089-l 104. 15. Roes A. Baton WF. (1981) IntnzelhdarpH. Physiol. Rev., 61.400-408. 16. Lahmy S. Salmon JM. Vigc J. Viallet P. (1989) pHl and DNA content modifications after ADR treatment in 3T3 fibroblasts. A micmfluommetric approach. Anti-r Res., 9,929-936. 17. Siskind MS. McCoy CH. Chobanian A. Schwartz JH. (1989) Regulation of intracellular calcium by cell pH in vascular smooth muscle cells. Am. J. Physiol., 256, C23CC240. 18. Dickens CJ. Gillespie JI. Greenwell JR Hutchinson P.(l990) Relationship between intracellular pH (pHi) and calcium (Ca2’l) in avian heart fibtoblasts. Hxp. Cell Res., 187,39-46.
Please send reprint requests to : Dr Fr&ric Bancel, Groupe de Microfluorim&rie Quantitative, URA CNR,S 1289, Universit6 de Perpignan, 52 Avenue de Villeneuve. F-66860 Perpignan cedex, France Received : 17 April 1991 Revised : 8 Cktober 1991 Accepted : 9 October 1991
