Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Long appreciated for its central role in fundamental cellular biochemistry, zinc is necessary for the proper function and regulation of a diverse array of biomolecules. Zinc serves in catalytic as well as structural roles for many proteins, is critical for lipid metabolism, and directly regulates gene expression and cell proliferation. Accordingly, human zinc deficiency is associated with a host of clinical abnormalities including growth retardation, suppressed immunity, and deteriorated mental capabilities (reviewed by Prasad, 1993).

Recently, the role of zinc in the central nervous system has received particular attention. Certain neurons accumulate large amounts of vesicularized zinc, and zinc modulates, at least in vitro, the activity of numerous channels, transporters, and receptors participatory in neural activity (Harrison and Gibbons, 1994). Perhaps most importantly, an expanding body of evidence shows zinc to be a potent neurotoxin. Excessive accumulation of intracellular zinc kills cultured neurons and glia (Choi and Koh, 1998). In animal models, zinc accumulates in neurons destined to die after ischemia, seizure, and blunt trauma (Frederickson et al., 1989; Koh et al., 1996; Suh et al., 2000). Although the precise source of zinc is currently debated, it is clear that these injury paradigms involve mobilization and redistribution of zinc already present in the brain, giving rise to its classification as an endogenous neurotoxin (Koh et al., 1996; Cuajungco and Lees, 1998; Lee et al., 2000).

The investigation of zinc-mediated cell death has prompted the development of methodologies for measuring intracellular free zinc ([Zn²⁺]_i). Historically, probes used to detect tissue zinc required sample fixation (Frederickson, 1989). In live neurons, initial measurements of [Zn²⁺]_i employed fluorescent, ion-sensitive indicators closely related to fura-2 (Sensi et al., 1997; Cheng and Reynolds, 1998). Though famous for measuring [Ca²⁺]_i and [Mg²⁺]_i, these indicators are in fact very sensitive to heavy metals, with affinities for Zn²⁺ ranging from 2 to 30 nM (Tsien, 1999). However, selective detection of [Zn²⁺]_i has typically necessitated exclusion of Ca²⁺ and/or Mg²⁺ from the extracellular buffer. The advent of the first Zn²⁺-selective live-cell fluorophore, Newport Green DCF,¹ allowed straightforward measurement of [Zn²⁺]_i while minimizing confounding signal from other biologically important divalent cations (Canzoniero et al., 1999; Sensi et al., 1999; Aizenman et al., 2000). However, Newport Green is a single-wavelength fluorophore and unfortunately lacks advantages that popularized dual-wavelength (i.e., ratiometric) indicators, such as fura-2.

A new family of Zn²⁺ dyes may provide superior alternatives. These probes are modifications of pre-existing single- or dual-wavelength Ca²⁺ dyes, are available in cell-permeant forms, and have affinities for zinc in the micromolar range. In the present study, we initially evaluated properties of two of these novel indicators, FuraZin-1 and FluoZin-2, and compared them with the more established dyes Newport Green and mag-fura-2. Commercial data regarding fluorescent indicators can be misleading, because testing is almost always performed in spectrofluorometer-based, cell-free assays. Our comparisons therefore used fluorescence microscopy recordings from intact, dye-loaded neurons, a setting more consistent with their intended practical use. Our characterization of these dyes revealed that the high-affinity indicator mag-fura-2 (K_D, Zn²⁺ ~20 nM) displayed a sensitivity for [Zn²⁺]_i that was surprisingly similar to other indicators of much lower affinities (~1-3 µM). A reconsideration of ion-dye binding stoichiometry predicted that apparent dye sensitivity would in theory be heavily dominated by intracellular dye concentration rather than the affinity constant. We tested this principle in neurons containing different concentrations of mag-fura-2. In agreement with modeling predictions, dye saturation in neurons varied according to the intracellular dye concentration.

These results establish the utility of novel Zn²⁺-sensitive indicators in the practical setting of fluorescence microscopy. Furthermore, we demonstrate that although dye-based estimations of [Zn²⁺]_i are misleading, calculations that consider intracellular dye concentration may allow accurate determination of the total Zn²⁺ load.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Culture. All procedures using animals were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh. Primary cultures of embryonic rat forebrain neurons were prepared as described previously (Brocard et al., 2001). Briefly, embryonic day 17 Sprague-Dawley rat fetuses were surgically removed from an anesthetized dam. Fetal forebrains were then excised, dissociated by trypsinization, and plated on poly-D-lysine-coated 31-mm glass coverslips in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, penicillin (100 U/ml), and streptomycin (100 µg/ml). Twenty-four hours after plating, the medium was replaced with Dulbecco's modified Eagle's medium containing 10% horse serum in place of fetal bovine serum, and coverslips were inverted to inhibit proliferation of neuroglia. Neurons were kept in a 5% CO₂, 37°C incubator, and all experiments were performed after 9 to 16 days in culture.

Reagents and Solutions. All ion-sensitive fluorophores and N,N,N',N'-tetrakis(2-pyridalmethyl)ethylenediamine (TPEN) were purchased from Molecular Probes (Eugene, OR); all other reagents were purchased from Sigma-Aldrich (St. Louis, MO) unless otherwise noted. During microfluorimetric measurements, coverslips were continuously superfused with HEPES-buffered salt solution (HBSS) containing 150 mM NaCl, 5 mM KCl, 0.9 mM MgSO₄, 1.4 mM CaCl₂, 20 mM HEPES, and 5.5 mM glucose, and adjusted to pH 7.4 with NaOH. When appropriate, Zn²⁺ was added to the buffer from a 1000× stock of ZnCl₂ in 25 mM HCl. The Zn²⁺-specific ionophore sodium pyrithione (1-hydroxypyridine-2-thione) was added at a concentration of 20 µM from a 20 mM stock in dimethyl sulfoxide. To reduce [Zn²⁺]_i, the membrane-permeant heavy metal chelator TPEN was included at a concentration of 25 or 50 µM from a 25 mM stock in dimethyl sulfoxide. Experiments addressing dye sensitivity to high [Mg²⁺]_i used HBSS containing high extracellular Mg²⁺ concentration (9 mM) as described previously (Stout et al., 1996).

Fluorescence Microscopy. Recordings were performed at room temperature (20-25°C). [Zn²⁺]_i was measured using the Zn²⁺-sensitive fluorophores mag-fura-2, FuraZin-1, FluoZin-2, and Newport Green DCF. Figure 1 shows the structure, K_D for Zn²⁺, and excitation and emission wavelengths for each dye. (The properties of FuraZin-1 and FluoZin-2 are described only in product information inserts, which can be found on the Molecular Probes website: http://www.molecularprobes.com.) To load neurons with cell-permeant dye derivatives, coverslips were incubated in 1 ml of HBSS containing 5 mg/ml of bovine serum albumin and 5 µM dye in the dark at 37°C. The incubation period was 20 min for mag-fura-2 and FuraZin-1 and 30 min for Newport Green and FluoZin-2. After loading, coverslips were rinsed thoroughly, mounted in a recording chamber, and superfused with HBSS at 10 ml/min at room temperature. For each coverslip, the responses of approximately 10 to 15 neurons were compiled to generate a single mean trace. The PC-based imaging system used in these experiments consisted of the following components: a Nikon Diaphot 300 microscope equipped with a 40× oil-immersion objective (Tokyo, Japan), a charge-coupled device camera (Hamamatsu Photonics, Hamamatsu City, Japan), a monochromator-driven xenon light source (ASI, Eugene, OR), and SimplePCI imaging software (Compix, Cranberry, PA). For simultaneous imaging of mag-fura-2 and Newport Green DCF, neurons were loaded with Newport Green (5 µM) for 30 min. Ten minutes later, 5 µM mag-fura-2 acetoxymethyl ester (AM) was added, and coincubation with both dyes continued for an additional 20 min.

View larger version (28K): [in this window] [in a new window]   Fig. 1.   Structures of Zn2+-senstive dyes. The chemical structures of mag-fura-2, Newport Green DCF, FuraZin-1, and FluoZin-2 are accompanied by respective excitation and emission wavelengths employed in this manuscript. Also included for each dye is the reported affinity for Zn2+ (in micromolar). Note that mag-fura-2 and FuraZin-1 are ratiometric dyes, using two distinct excitation wavelengths. Newport Green DCF and FluoZin-2 are single-wavelength indicators.

Intraneuronal Dye Concentration. To determine the amount of intracellular dye accumulation in response to a known extracellular concentration of cell-permeant AM, three to six coverslips were exposed to varying concentrations of mag-fura-2 AM (0.1, 0.5, 1, 5, and 20 µM) for 20 min at 37°C in HBSS supplemented with bovine serum albumin. Newport Green DCF (5 µM for 30 min) and fura-2 AM (5 µM for 40 min) were also assayed. Using 1 to 1.5 ml of low ionic strength buffer (containing 120 mM KCl, 10 mM HEPES, and 0.2 mM EGTA, pH 7.2), neurons were lysed and collected using a rubber policeman. To ensure recovery of all intracellular dye, recovered material was subjected to three freeze/thaw cycles. Cellular particulate was then separated by centrifugation (10 min at 10,000g), and the dye-containing supernatant was decanted and its volume measured.

Statistics. All experiments were repeated on at least three different coverslips taken from three separate culture preparations. All data plots, modeling, and appropriate statistical analyses were generated using GraphPad Prism, version 3.0 (GraphPad Software, San Diego, CA).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

We first evaluated the properties of the new dyes to establish excitation and emission optima in cells (Fig. 2). Although in vitro spectra are available for these dyes (Molecular Probes product literature), it cannot be assumed that dyes will show identical spectroscopic and ion-binding properties in the cellular environment. (Here and in all subsequent usage, "in vitro" refers to cuvette-based, cell-free assays performed in a spectrofluorometer.) Accordingly, we loaded neurons with mag-fura-2, Newport Green DCF, FuraZin-1, and FluoZin-2 and obtained excitation spectra under basal conditions where [Zn²⁺]_i is very low. Excitation spectra were again obtained from the same neurons after they were exposed to high extracellular Zn²⁺ in the presence of the Zn²⁺-selective ionophore pyrithione, a maneuver known to substantially elevate [Zn²⁺]_i. From these experiments, we selected optimal excitation wavelengths for each dye. For mag-fura-2 and FuraZin-1, ₁ was chosen at 335 and 340 nm, respectively. Fluorescence produced at these wavelengths was adequately higher than background under resting conditions, while still exhibiting a modest increase in response to high [Zn²⁺]_i. For ₂, 375 and 380 nm were chosen because they exhibited a large [Zn²⁺]_i-induced decrease in fluorescence that remained sufficiently higher than background intensity (Fig. 2, A and C). In later figures, fluorescence values for mag-fura-2 and FuraZin-1 are shown in ratio form (i.e., ₁/₂). For the single-wavelength indicators Newport Green and FluoZin-2, excitation at 490 nm gave modest fluorescence under resting conditions while demonstrating several fold increases at high [Zn²⁺]_i conditions (Fig. 2, B and D). Newport Green and FluoZin-2 fluorescence is normalized to starting values (F/F₀) in later figures.

View larger version (26K): [in this window] [in a new window]   Fig. 2.   Intracellular excitation spectra for Zn2+-sensitive dyes acquired by fluorescence microscopy. Coverslips of neurons incubated with 5 µM cell-permeant forms of mag-fura-2 (A), Newport Green (B), FuraZin-1 (C), or FluoZin-2 (D) were illuminated through a range of excitation wavelengths at 2-nm intervals to establish excitation spectra in conditions of low [Zn2+]i (). Neurons were then stimulated with pyrithione (20 µM) and Zn2+ (300 µM) to produce high [Zn2+]i. The illumination procedure was then repeated to determine excitation spectra under conditions of high [Zn2+]i (). The vertical dashed lines correspond to the excitation wavelengths used for each dye in subsequent fluorescence microscopy experiments. (See Materials and Methods for experimental details.)

The utility of these dyes depends heavily on their selectivity for Zn²⁺ over other biologically relevant cations. Because these agents are derived from Ca²⁺ indicators, we first investigated their sensitivity to elevated [Ca²⁺]_i (Fig. 3). Our previous studies in neurons showed that very high [Ca²⁺]_i can be achieved by persistent activation of glutamate receptors (Stout and Reynolds, 1999). Exposing neurons to glutamate and the coagonist glycine (100 and 10 µM, respectively) elevates mag-fura-2 ratio values by 2- to 3-fold (Stout et al., 1998), consistent with the ability of mag-fura-2 to detect high [Ca²⁺]_i. In contrast, activation of glutamate receptors did not increase Newport Green fluorescence (Fig. 3, A and D). Glutamate caused a slight increase in the FuraZin-1 ratio (~0.05 units; Fig. 3, B and E) and a 50% increase in normalized FluoZin-2 fluorescence. (Fig. 3, C and F). These changes might reflect a modest sensitivity to [Ca²⁺]_i or perhaps a [Ca²⁺]_i-induced increase in [Zn²⁺]_i. In any case, glutamate-induced changes in FuraZin-1 and FluoZin-2 were small relative to those elicited by Zn²⁺ and pyrithione (see Fig. 4). Therefore, it is reasonable to conclude that FuraZin-1 and FluoZin-2 are relatively insensitive to [Ca²⁺]_i. We also examined dye sensitivity to [Mg²⁺]_i using an approach described previously (Stout et al., 1996). This method elevates [Mg²⁺]_i by 2 to 3 mM, which can be detected with mag-fura-2. However, under the same conditions, we observed no signal increase from either Newport Green, FuraZin-1, or FluoZin-2 (data not shown), demonstrating that these dyes are insensitive to [Mg²⁺]_i.

View larger version (16K): [in this window] [in a new window]   Fig. 3.   [Zn2+]i-sensitive dyes are relatively insensitive to toxic levels of intracellular Ca2+. Neurons incubated with 5 µM cell-permeant Newport Green (A), FuraZin-1 (B), or FluoZin-2 (C) were stimulated by glutamate (100 µM) with glycine (10 µM) for a 10-min period, as indicated by the dashed lines, to generate high [Ca2+]i. Respective summary data are presented in D, E, and F. Dye signal was taken before (Pre) and immediately after (Post) the glutamate + glycine stimulus, and bars represent mean ± S.E. For Newport Green (A and D) and FluoZin-2 (C and F), y-axis depicts normalized F/F0 fluorescence values; for FuraZin-1 (B and E), y-axis depicts ratio values (1/2).

View larger version (24K): [in this window] [in a new window]   Fig. 4.   [Zn2+]i-sensitive dyes exhibit similar response profiles, despite varying affinities for Zn2+. Neurons incubated with 5 µM cell-permeant forms of mag-fura-2 (A), Newport Green (B), FuraZin-1 (C), or FluoZin-2 (D) were exposed to pyrithione (Pyr; 20 µM) in the presence of sequentially increased concentrations of added extracellular Zn2+ (0, 0.3, 3, 30, and 300 µM, each for 5 min) as indicated by the dashed and solid lines. Return of [Zn2+]i to resting levels was achieved by addition of the high-affinity, membrane-permeant chelator TPEN (50 µM; starting at arrow). For the dual-wavelength dyes mag-fura-2 and FuraZin-1, ratio values are displayed on the y-axis; for the single-wavelength indicators Newport Green and FluoZin-2, dye signal is presented as normalized (F/F0) fluorescence.

We next evaluated dye responsiveness to elevated [Zn²⁺]_i. Previously, we showed that pyrithione provides a useful way to deliver Zn²⁺ into neurons, elevating [Zn²⁺]_i in proportion to the extracellular Zn²⁺ concentration (Dineley et al., 2000). This approach allows characterization of dyes in terms of the minimum amount of added Zn²⁺ necessary to elicit a change in dye signal and also the amount of added Zn²⁺ that produces a maximal, or saturating, response. The minimum and maximum responses, as well as the general profile of the response, provide parameters that can be compared between dyes. One would predict that dyes of higher affinity would be more sensitive and saturate more quickly in response to lower concentrations of added Zn²⁺, whereas dyes of lower affinity would be insensitive to low concentrations of added Zn²⁺ and saturate more slowly. Figure 4 shows typical traces from these experiments. In the case of each dye, the introduction of pyrithione alone had relatively little effect on the fluorescence signal, indicating that ambient Zn²⁺ in the buffer was low. Small signal increases were obtained by adding 0.3 µM Zn²⁺ to the extracellular solution, progressively larger responses resulted from 3 and then 30 µM added Zn²⁺, and modest increments (if any) were encountered at 300 µM Zn²⁺. For each indicator, addition of the high-affinity, membrane-permeant Zn²⁺ chelator TPEN restored fluorescence signals to prestimulus values. The data from a number of these experiments are summarized in Fig. 5. It is surprising that all four of these dyes demonstrate very similar response profiles to the same series of Zn²⁺ stimuli, given that their in vitro affinities vary from 20 nM (mag-fura-2) to ~3 µM (FuraZin-1). (With respect to the y-axes, it should be noted that the fluorescence readout values for each dye will differ due to intrinsic differences in dye properties but that the general profiles of the responses are broadly similar between the dyes, regardless of y-axis scale.) Another set of experiments provided further confirmation. Neurons were coloaded with mag-fura-2 and Newport Green so that the responses of both dyes could be recorded simultaneously from the same neurons. Again, the response of the dyes was surprisingly similar given the ~100 fold difference between their in vitro affinities (Fig. 6).

View larger version (26K): [in this window] [in a new window]   Fig. 5.   [Zn2+]i summary of [Zn2+]i-induced changes in Zn2+-sensitive dyes. For each dye in Fig. 4, fluorescence values were taken during basal conditions and at the end of each 5-min stimulus (20 µM pyrithione alone and 20 µM pyrithione in the presence of 0.3, 3, 30, or 300 µM extracellular Zn2+). For the dual-wavelength dyes mag-fura-2 and FuraZin-1, ratio values are displayed on the y-axis; for the single-wavelength indicators Newport Green and FluoZin-2, dye signal is presented as normalized (F/F0) fluorescence. Bars represent mean ± S.E.

View larger version (21K): [in this window] [in a new window]   Fig. 6.   Simultaneous recordings of mag-fura-2 and Newport Green. Neurons coloaded with cell-permeant Newport Green and mag-fura-2 (5 µM each) were exposed to pyrithione (Pyr; 20 µM) in the presence of sequentially increased concentrations of added extracellular Zn2+ (0, 0.3, 3, 30, and 300 µM, each for 5 min), as indicated by dashed and solid lines. At the end of the zinc treatment, TPEN (25 µM) was applied, beginning at (). Mag-fura-2 ratio values are represented by  and displayed on the left y-axis, whereas normalized (F/F0) Newport Green values are represented by  and displayed on the right y-axis. In comparing Fig. 6 with Fig. 4, the reader will notice different y-axis scale for mag-fura-2. The optics used for dual dye detection were different from those used to detect mag-fura-2 alone. The spectral bias of dual dye optics tends to compress the absolute range of the mag-fura-2 ratio. It should be emphasized, however, that the appropriate comparison involves the response profile rather than specific y-axis values.

These unexpected observations prompted us to reconsider the interpretation of these experiments. Interactions between dye and ion are governed by the principles of receptor-ligand binding. The saturation of dye by ion can be described by a standard receptor-ligand binding equation, which we have superficially modified so that "dye-ion" terminology is used instead of classic receptor-ligand parlance:

(1)

where [ion] represents the concentration of free Zn²⁺, [dye_t] the concentration of free dye, [dye·ion] the concentration of dye·Zn²⁺ complex, and [K_D] the dissociation constant of the dye for Zn²⁺. Indeed, it is a modification of this approach that was first proposed in the landmark paper by Grynkiewicz and colleagues (1985) and is widely used for calibrating Ca²⁺-sensitive dyes. However, this approach assumes that 1) the quantity of ion bound to dye is insignificant compared with the total ion pool available and consequently 2) the formation of dye·ion complex does not appreciably deplete [ion]. However, high concentrations of intracellular dye are often achieved by the diffusion-limited dye-loading techniques employed in most live-cell fluorescence microscopy. In such experiments, sufficiently high dye concentrations will inevitably deplete [ion], thereby invalidating the above assumption. Fortunately, depletion of [ion] occurs in direct proportion to total intracellular dye concentration and can be accounted for with a modification to eq. 1 (Kenakin, 1993):

(2)

which is solved in eq. 3:

(3)

where [ion_t] represents the total ion concentration. It is apparent that the formation of dye·ion will be limited (saturable) as complex formation depletes available ion. When considering ion-induced changes in dye signal, it should be understood that the change in any dye signal will be proportional to the amount of dye·ion complex formed; that is, dye responses are a function of the fractional occupancy of the dye by ion. Using eq. 3, which accounts for ion depletion, we can predict fractional dye occupancy as a function of added ion under various dye concentrations. Figure 7 shows a series of such curves for a dye of fixed affinity for Zn²⁺ (20 nM; e.g., mag-fura-2) when [dye] is increased from 1 to 1000 µM. As is evident from this graph, dye concentration has a substantial impact on the sensitivity of the system, and each 10-fold increase in dye concentration causes a large rightward shift in the fractional occupancy curve. Figure 8 again uses eq. 3 in an alternative consideration: fractional occupancy curves are plotted for dyes with different Zn²⁺ affinities (20 nM-20 µM), whereas dye concentration remains fixed (which essentially models the comparisons made in this study). Figure 8A shows dyes of different affinity assuming 10 µM [dye], whereas Fig. 8B plots the same hypothetical dyes at a concentration of 1000 µM. The curves, barely discrete at 10 µM dye, are essentially superimposed at 1000 µM dye, regardless of dye K_D. These surprising results show that dye response to ion fluxes is heavily dominated by the concentration of the dye itself, whereas K_D is virtually meaningless.

View larger version (21K): [in this window] [in a new window]   Fig. 7.   Dye fractional occupancy corrected for intracellular dye concentration. Using eq. 3, and assuming a KD for Zn2+ of 20 nM, theoretical fractional occupancy curves were generated for various intracellular dye concentrations (1, 10, 100, and 1000 µM). These curves model the saturation response of mag-fura-2 as a function of Zn2+; increasing intracellular dye concentration results in a rightward shift.

View larger version (21K): [in this window] [in a new window]   Fig. 8.   High dye concentrations determine fractional occupancy of Zn2+-sensitive dyes. Using eq. 3, theoretical fractional occupancy curves were generated for dyes of varying affinity (0.02, 0.2, 2.0, and 20 µM), each at a concentration of 10 µM (A). B presents similar data, but at a concentration of 1000 µM. Increasing dye concentration to millimolar levels results in virtually identical fractional occupancy, regardless of dye affinity.

These results raise two important issues. First, intracellular dye concentration becomes a dominating parameter in evaluating the magnitude of the ion-induced dye responses. Second, the findings predict that the saturation characteristics of any dye should vary according to the intracellular dye concentration. We investigated both issues using mag-fura-2 because it is a well characterized dye that loads effectively into neurons, generating relatively bright fluorescence signals. However, there is little information available regarding the intracellular concentrations of mag-fura-2 following standard loading conditions. We incubated neurons with concentrations of mag-fura-2 AM between 0.1 and 20 µM, washed the cells carefully, lysed the neurons, and recovered the dye in the supernatant after centrifugation of cell debris. Using a standard curve generated with known concentrations of mag-fura-2 tetrapotassium salt, we estimated dye concentration in the recovered samples, which were then normalized to protein concentration (see Materials and Methods). Although we have been unable to directly measure the intracellular volume of neurons using radioisotope exclusion, volume in liver parenchymal cells has been estimated to be 2 µl/mg protein (Kletzien et al., 1975). Using this premise, calculated [mag-fura-2]_i achieved millimolar concentrations (Fig. 9A). Proving that high intracellular dye concentrations are not unique to mag-fura-2, similar measurements with fura-2 and Newport Green also approached or exceeded millimolar values (Fig. 9B). Using mag-fura-2 loading concentrations consistent with those of Fig. 9A, we repeated cell-based, fluorescence microscopy recordings where pyrithione was applied in the presence of sequentially increased extracellular Zn²⁺ concentration. The extent to which we were able to decrease intracellular dye concentration was essentially limited by the fluorescence-imaging hardware. Nevertheless, it was possible to test loading concentrations as low as 0.5 µM and up to 20 µM (Fig. 10A). Interestingly, neurons loaded with the least amount of dye showed the greatest sensitivity to the addition of chelator before the addition of Zn²⁺. This suggests the presence of a finite pool of free intraneuronal Zn²⁺ or perhaps Zn²⁺ that the dye has leeched from internal binding sites. In either case, any signal from this relatively small amount of zinc is presumably obscured by high concentrations of intracellular dye. Most importantly, and as predicted by Fig. 7, neurons loaded with lower dye concentrations displayed greater Zn²⁺ sensitivity and saturated with less added Zn²⁺ compared with those loaded with higher dye concentrations. The results of several such experiments are summarized in Fig. 10B.

View larger version (15K): [in this window] [in a new window]   Fig. 9.   Accumulation of intracellular dye in response to varied concentrations of extracellular AM. A presents intracellular mag-fura-2 concentration as a function of extracellular AM concentration. The soluble, dye-containing fraction was recovered from neurons incubated with varied mag-fura-2 AM concentrations (0.1, 0.5, 1, 5, and 20 µM) and analyzed for mag-fura-2 content using a spectrofluorometer (see Materials and Methods). Both axes plot molar dye concentration (M) on log10 scale. Data points are presented as mean ± S.E. B, comparison between different cell-permeant dyes. Similar to A, the soluble, dye-containing fraction was obtained from neurons incubated with 5 µM of either mag-fura-2, fura-2, or Newport Green and analyzed for dye content. Bars, depicting intracellular dye concentration in millimolar, are presented as mean ± S.E.

View larger version (44K): [in this window] [in a new window]   Fig. 10.   Neurons incubated with different concentrations of mag-fura-2 AM display different Zn2+ saturation profiles. Neurons incubated with 0.5 (solid trace), 1 (), 5 (dashes), or 20 µM () mag-fura-2 AM were treated with TPEN (50 µM) for 5 min, washed with normal HBSS for 10 min, then stimulated with pyrithione (Pyr; 20 µM) in the presence of sequentially increased concentrations of added extracellular Zn2+ (0, 0.3, 3, 30, and 300 µM, each for 5 min) as indicated by the dashed and solid lines (A). B, summary data from experiments shown in A. Mag-fura-2 ratio values obtained at the end of each 5-min stimulus are expressed as a fraction of the maximum ratio value observed in an individual experiment. Bars depict mean ± S.E.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we investigated the properties of several novel fluorescent dyes that may be suitable for measuring [Zn²⁺]_i in cells. We established that FuraZin-1 and FluoZin-2 respond to altered [Zn²⁺]_i and that their response is selective for [Zn²⁺]_i over [Ca²⁺]_i or [Mg²⁺]_i. More importantly, these studies also revealed a surprising similarity in [Zn²⁺]_i sensitivity between dyes that have considerably different in vitro affinities for the ion. The apparent similarity between different dyes is the consequence of a poorly appreciated stoichiometric relationship between dye concentration and ion measurement that has a profound influence on the quantitative interpretation of intracellular ion signals. This issue becomes particularly important when using dyes of differing affinities and also when monitoring dye-ion interactions in which the ion has a high affinity for the dye.

Mag-fura-2 and Newport Green showed essentially identical sensitivities and saturation characteristics in cells. This prompted us to reassess conventional interpretation of dye signals. Using salts of the two dyes, we confirmed that their in vitro affinities were indeed very different (approximately 20 nM and 1 µM, respectively; data not shown). Our initial suspicion was that the intracellular environment caused one or both dyes to display altered properties. However, our data argued against this. First, high protein (bovine serum albumin) or lipid (cell membrane extracts) concentrations did not affect dye characteristics in vitro. The addition of sucrose to increase viscosity also did not affect dye responsiveness. Finally, dye recovered from loaded neurons behaved identically to the purchased salt, arguing against some unexpected enzyme-mediated dye alteration.

Instead, different dyes respond similarly in cells because standard loading protocols achieve high concentrations of intracellular dye. The critical oversight involves the assumption that the total ion and free ion remain essentially equal in these experiments. This assumption is valid in most receptor-ligand applications, because the concentration of added ligand is in great excess compared with the number of available receptors in biological preparations. However, with [ion]_i measurements, it is clear that levels of free ion can be substantially depleted by high concentrations of dye. Thus, although [Zn²⁺]_i may rise as high as a few hundred nanomolar in neurotoxic paradigms, [dye]_i may approach or exceed millimolar values (Fig. 9). In these conditions, there will be substantial depletion of added Zn²⁺. Equations 2 and 3 account for ion depletion; however, this correction reveals that dye·ion binding stoichiometry is determined by dye concentration, regardless of dye affinity for the ion of interest. This principle is illustrated in Figs. 7 and 8 and is experimentally validated in Fig. 10.

It is generally accepted that ion-induced changes in dye signal can be used to calculate absolute [ion]_i; it should be appreciated, however, that equations used for this purpose do not consider depletion of ion by excessive [dye]_i. Intracellular dye concentration is known precisely when the indicator is injected by micropipette (Helmchen et al., 1996). The vast majority of applications, however, use the diffusion-mediated loading approach employed in this report, and estimating [dye]_i under these circumstances is much more difficult. We extracted soluble fractions from neuronal cultures and calculated cytoplasmic concentrations based on dye recovery. The dye concentrations we report might be compromised by a variety of factors. Most obvious is the parameter used to estimate cell volume (Kletzien et al., 1975). This estimate was not made in neurons, and it is likely that neurons with their many processes would yield a higher level of protein per unit volume. Such an error would lead to overestimated [dye]_i. Similarly, some dye may be sequestered in cellular compartments not accessible to ion influx. Our recovery methods would most likely liberate this pool of dye into the soluble fraction, again causing an overestimation of [dye]_i available to interact with [ion]_i. We would note, however, that visual inspection of our fluorescence images yields no evidence of dye compartmentalization. Furthermore, prior studies support our estimates. Under broadly similar loading protocols, one report suggested that ~40 µM dye accumulated after loading with 5 µM fura-2 AM (Fink et al., 1998). Others reported millimolar concentrations using related fluorophores (Tsien, 1999). Indeed, it is generally accepted that loading with cell-permeant esters routinely yields a final [dye]_i of 10 to 100 times the loading concentration (Haugland, 1996). Finally, we would emphasize that the modeling predictions presented in Figs. 7 and 8 use fairly conservative dye concentrations compared with actual [dye]_i measurements shown in Fig. 9. Thus, our main conclusions would remain unaffected by even a 10-fold overestimation of [dye]_i.

Depletion of ion by high [dye]_i is exacerbated when the indicator has a particularly high affinity for the ion of interest (e.g., when using mag-fura-2 to detect [Zn²⁺]_i). Obviously, the dye will bind the ion very effectively, resulting in profound depletion of the free ion. It is generally recognized that an ideal fluorescent indicator should have a K_D near the anticipated concentration of the free ion. It is far less appreciated that [dye]_i should also be in the same range as K_D. A useful guideline for illustrating potential ion depletion involves a quick ratio calculation of [dye]_i/K_D. Depletion must be accounted for once this value exceeds unity (i.e., the point at which [dye]_i exceeds K_D) (Kenakin, 1993). In the case of Zn²⁺ and mag-fura-2, this ratio is exceedingly high, >>10³ if one assumes [dye]_i of 1 mM and K_D of 20 nM. With an affinity of 3 µM, a similar concentration of FuraZin-1 would still produce a ratio of >10². The detection of biologically relevant [Zn²⁺]_i necessitates high affinity, but affinity in turn dictates a concentration ceiling at which the fluorophore can be present without depleting free ion. For example, an ideal [Zn²⁺]_i indicator might exhibit an affinity of ~500 nM. However, [dye]_i in excess of 500 nM will lead to considerable ion depletion. The issue now becomes one of practicality, as it is improbable that any [Zn²⁺]_i indicator could even be detected by fluorescence microscopy at [dye]_i low enough to also avoid depletion of free ion. As our results show, reducing [dye]_i is limited in practice by the detection capability of the recording system. In conclusion, it can now be stated that for any Zn²⁺-sensitive dye, conversion of fluorescence to absolute [Zn²⁺]_i must account for Zn²⁺ depletion. These quantitative constraints, however, do not invalidate the use of these dyes for qualitative comparisons within and between cells.

The above considerations are also relevant to measurements of [Ca²⁺]_i and [Mg²⁺]_i. Using fura-2 to detect [Ca²⁺]_i, for example, [dye]_i/K_D is still well in excess of 10³ (assuming [fura-2]_i = 250 µM and K_D = 140 nM). Thus, depletion still confounds calculations of free ion concentrations. The effect is much smaller with very low-affinity Ca²⁺ indicators, so that the use of mag-fura-2 to measure [Ca²⁺]_i (K_D, Ca²⁺ ~50 µM) may be valid (Stout et al., 1998; Brocard et al., 2001). The problem disappears entirely when [Mg²⁺]_i is measured with mag-fura-2 (Stout et al., 1996), because the K_D for Mg²⁺ (0.5 mM) is very near our measured [dye]_i. It is also possible that comparisons between indicators of relatively similar affinities will be impacted more by differences in [dye]_i than by slightly differing affinities. Conversely, the differences between high- and low-affinity Ca²⁺ indicators previously reported (Hyrc et al., 1997; Stout and Reynolds, 1999) might be erroneously exaggerated if the low-affinity indicator loads into cells more effectively (as seems to be the case for mag-fura-2).

The application of ion-sensitive indicators in biological systems is accompanied by strong motivation to convert fluorescence signals into precisely quantified intracellular ion concentrations. In this regard, ratiometric indicators, such as fura-2, are considered superior to their single-wavelength counterparts. In principle, variations in illumination intensity, specimen thickness, and dye concentration can be ignored when fluorescence intensities at two distinct wavelengths are ratioed. Because variations in these parameters affect both wavelengths proportionally, ratiometric tactics conveniently factor them out. It is now evident that dye concentration cannot be ignored, regardless of whether single-wavelength or ratiometric approach is used. It must be noted, however, that the pioneers of ratio imaging did not intend for dye concentration to be ignored altogether; rather, their argument pertained to local variations of fluorophore concentration within a sample (Grynkiewicz et al., 1985). Those wishing to extrapolate precise values for intracellular ion concentrations have erroneously generalized this particular virtue of ratiometric imaging.

The key conclusion here is that it is difficult to infer [Zn²⁺]_i, and probably [Ca²⁺]_i, from the standard approaches to live-cell fluorescence microscopy. The solution to this problem is less clear. It is evident, however, that a critical parameter in these experiments is the intracellular dye concentration. If it is possible to determine [dye]_i effectively, then the total ion flux in response to a stimulus could be estimated with some confidence. This parameter, together with the cytoplasmic volume, could then be used to approximate the intracellular ion concentration, provided that the impact of fixed intracellular buffers can be accounted for (Neher, 1995; Helmchen et al., 1996). Given that none of these parameters have yet been unequivocally established for Zn²⁺ in neurons, we feel that quantitative estimates of [Zn²⁺]_i are premature, even though this conclusion does not invalidate the approach for qualitative comparisons within or between cells.