Endosomal Accumulation of pH Indicator Dyes Delivered as Acetoxymethyl Esters

The predominant mechanism for concentrative uptake of most nutrients by plants, fungi and many bacteria is symport with protons, drawing upon the transmembrane electrochemical gradient of H⁺ to build finite countergradients of sugars, amino acids, inorganic ions, nucleotide precursors, etc. And indeed, over the past 6–8 years, molecular biological techniques have led to the cloning and sequencing of large numbers of these symport systems (Quagliariello and Palmieri, 1992; Friedlander and Mueckler, 1992; Reuss et al. 1993), to detailed structure–function studies (see, for example, Kaback, 1992) and to mapping of expression with respect to cell and tissue type and developmental stage. The large-celled fungus Neurospora crassa has been especially useful for functional studies of these transport systems, since it has permitted easy measurement of currents associated with the nutrient fluxes (Hansen and Slayman, 1978; Sanders et al. 1983; Rodriguez-Navarro et al. 1986) in a single-cell system, thereby simplifying interpretation of ion-flux data and permitting, in principle, detailed electrical-kinetic analysis of the type described for Na⁺-coupled amino acid transport in patch-clamped mouse pancreatic acinar cells (Jauch and Läuger, 1986; Jauch et al. 1986) and for Na⁺-coupled glucose transport expressed from rabbit into Xenopus oocytes (Parent et al. 1992a,b).

Although improved patch-clamp technology (Bertl et al. 1992) has now made these same experiments possible on a genetically more compliant and popular organism, the yeast Saccharomyces cerevisiae, some features are nevertheless particularly clearly observed in Neurospora. Measurements with H⁺-sensitive intracellular microelectrodes had established the normal value of cytoplasmic pH in Neurospora to be 7.2±0.1 (Sanders and Slayman, 1982) and cytoplasmic buffer capacity to be near its minimum (approximately 38 mequiv H⁺ per pH unit) at that pH, but had also revealed a number of regulatory and kinetic anomalies associated with the conditions evoking various H⁺-coupled symporters, including enhanced cytoplasmic buffer capacity (5-to 10-fold; Sanders and Slayman, 1982) and pH-dependent shifts of apparent H⁺:substrate stoichiometry for the high-affinity glucose uptake system (GLU II) in the organism (Sanders and Slayman, 1984).

Technical constraints of the pH-electrode technique, however, directed us to alternative procedures for measuring cytoplasmic pH and, specifically, to the increasingly popular technique of ratio-fluorimetry (Tsien et al. 1985; Bright et al. 1987), based on esterified, membrane-permeant, fluorescent dyes (Thomas et al. 1979; Tsien, 1981; Loew, 1988), which can be broken down by cytoplasmic hydrolases and thereby trapped as intracellular indicator ions. During the course of these experiments, we discovered, or rather rediscovered (see Discussion), that the intracellular distribution of at least certain fluorescein-and rhodafluor-derived dyes is very far from homogeneous; in other words, that the end target for them is not free cytosol, but endomembranal vesicles (EMVs) of all sizes: endosomes, lysosomes and vacuoles. The findings emphasize that for all eukaryotic cells – plant, fungal and animal – interpretation of dye-based ′measurements′ of cytoplasmic pH and proton flux depends critically upon quantitative knowledge of the dye distribution.

Neurospora crassa strain RL21a was grown both in liquid shaking cultures (see Fig. 1) and in stationary, cellophane-on-agar cultures (see Figs 2–5, 7), which were generally handled as previously described (Slayman, 1965). The growth medium contained N-minimal salts (Vogel, 1956) supplemented with 2% sucrose at 25°C, and the staining medium was either 0.29× growth medium or a standard phosphate buffer (SPB: 25 mmol l⁻¹ K⁺ and 22.7 mmol l⁻¹ phosphate, at pH 5.8, plus 1 mmol l⁻¹ NH₄Cl) with or without 0.1–1.0% (w/v) glucose, as indicated. Mycelium was harvested by floating off the culture surface, rinsing (usually in SPB), transferring to a coverslip and blotting off excess medium. Dyes were applied by pipetting 50 μl samples of 3–100 μmol l⁻¹ -AM derivatives onto the coverslip and incubating the preparation in the dark for varying periods (see figure legends). Staining was terminated by blotting off excess fluid, rinsing in medium + 1 mmol l⁻¹ CaCl₂, and mounting the coverslip onto a thin flow-through chamber on the microscope stage.

Full-field microscopy (see Figs 1, 2, 4A, 8) was carried out on a Zeiss Axiophot microscope at Yale, using mainly a 63× apochromat adapted for both phase contrast and interference contrast, plus the Zeiss fluorescein filter pack. Confocal microscopy (see Figs 3, 4B,C, 5, 7) was carried out in the Developmental Resource for Biophysical Imaging and Opto-electronics (DRBIO) at Cornell, using a Zeiss IM35 inverted microscope fitted with a 100× planachromat, a BioRad MRC600 confocal head with wide pinhole, the 488 nm beam from an ILT 5425 laser (Ion Laser Technology, Salt Lake City UT) and the BioRad GHS filter block. The primary beam was directed to an acousto-optic modulator and adjusted so that the secondary peak passed through the microscope. The modulator was then used as a blanking switch to limit exposure of the preparation to the image-accumulating sweeps. In addition to the primary confocal fluorescence images, simultaneous or sequential phase-contrast and interference-contrast images were obtained, via an optical fiber gathering transmitted laser light at the back of the microscope condenser.

All digital images were assembled from 1 s scans, collected and processed via the BioRad SOM software operating on an NEC Powermate 386/20 microcomputer. End-processing of the images was carried out on an IBM RS-6000 Workstation in DRBIO, using ′Pixer′ image-processing software (written by Dr Russel Loan). Final printouts were made on a Tektronix Phaser IISDX color printer.

Esterified derivatives of three pH indicator dyes were tested: BCECF-AM [2′,7′-bis-(2-carboxyethyl)-5(and 6)-carboxyfluorescein], C.SNARF-AM [5(and 6)-carboxy-seminaphthorhodafluor-1] and C.SNARF-calcein-AM [5(and 6)-carboxy-seminaphtho-rhodafluor-1 calcein]. The dyes (nos B1150, C1271 and S3056, respectively, from Molecular Probes, Inc., Eugene, OR) were dissloved at 10 mmol l⁻¹ in anhydrous dimethylsulfoxide (DMSO), stored in 3–5 μl samples, then thawed and diluted into the growth medium or buffer solution immediately prior to use.

The overwhelming impression obtained from full-field examination of Neurospora hyphae was of a very heterogeneous distribution of fluorescence, for all concentrations of BCECF-AM tested and for all exposure times greater than about 15 min. A typical result for 10 h germinated conidia of Neurospora is shown in Fig. 1. Fig. 1A, taken in interference contrast, shows several hyphae (a) containing scattered vacuoles 2–4 μm diameter separated by nearly homogeneous cytoplasm, together with one hypha (b) packed with smaller vacuoles, and several hyphae of intermediate structure (c). Fig. 1B shows BCECF fluorescence in the same field and reveals a punctate distribution corresponding precisely to visible vacuoles. In several hyphae, the cytoplasm fluoresced intensely and almost homogeneously, but showed a faint mottled appearance that did not correspond to identifiable vacuoles. Overt vacuoles in these hyphae were stained at varying intensities relative to the background. The vacuole-packed hypha fluoresced only faintly, and that almost entirely from the vacuoles themselves, especially the smaller ones (c). Also, for most intermediate hyphae, the vacuoles alone fluoresced, but generally very brightly.

Prolonged observation revealed the fluorescence pattern in these cells to evolve towards almost complete vacuolar localization, but that process will be better demonstrated by experiments on the larger, mature hyphae, as described below. It must be emphasized that fungal vacuoles are acidic organelles (Nicolay et al. 1982; Legerton et al. 1983; Anraku et al. 1992), like animal lysosomes and most endosomes (Ohkuma and Poole, 1978; Tycko et al. 1983; Mellman et al. 1986), which should not accumulate anionic dyes, such as de-esterified BCECF, by simple pH trapping.

A major advantage of using Neurospora for these dye-distribution studies is that the extent to which endomembranal aggregation occurs, and the sizes of the resultant vacuoles visible in the light microscope, can be controlled by specific metabolic shifts, of which the simplest is glucose withdrawal (C. L. Slayman, E. Bashi, U.-P. Hansen, K. E. Allen and D. Sanders, in preparation).

Full-field phase-contrast pictures of two extreme cases are shown in Fig. 2. Fig. 2A displays a large hypha from mature mycelium grown in the presence of 2% sucrose (thus, metabolizing glucose). Normally visible features were the (outer) cell wall, a single cross-wall (xw), a typical single large vacuole (lv), a cluster of small vacuoles (sv) and a generally speckled cytoplasm. Hyphae such as this are packed with smaller vacuoles, endosomes, mitochondria and nuclei (McClure et al. 1968; That and Turian, 1978), all of which are obscured in photographs of living cells by out-of-focus imaging through the hyphal thickness (here, 14 μm), as well as by streaming and Brownian motion of most cytoplasmic particles. Fig. 2B shows a pair of nearby hyphae after 110 min of carbon starvation (carried out in a perfusion chamber on the microscope stage; see Materials and methods). Despite evident image blurring, the cytoplasm of the larger hypha can be seen to have been packed with vacuoles, ranging from 2.5 to 6 μm in diameter, and the smaller hypha with somewhat smaller vacuoles.

Corresponding confocal fluorescence images of a single hypha stained with BCECF-AM are shown in Fig. 3: near the beginning of glucose starvation (Fig. 3A) and approximately 3 h later (Fig. 3B). The punctate distribution of fluorescence is evident in both frames, at least at moderate overall intensities, but in the first the bright spots are irregularly distributed and small (mostly <1 μm diameter), whereas in the second they appear uniformly packed and 2–4 μm in diameter. It seems, therefore, that the confocal fluorescence images resolved EMVs or small vacuoles which had defeated the full-field phase microscope during early glucose starvation and also sharpened images of the densely packed enlarged vacuoles. The most important inference from Figs 2 and 3 is that BCECF, presented in the esterified form, targets to vesicular structures but becomes more clearly visible there as the organelles coalesce or enlarge into overt vacuoles. Direct comparison of end-fluorescence patterns with simultaneous phase images has confirmed this inference.

The linear geometric path for dye transfer from medium to vacuoles would require passage into and through the cell wall, wide distribution in the cytosol, and – finally – end accumulation in the vacuoles. This progression does in fact occur, and the first two stages are demonstrated in the experiments of Fig. 4. Fig. 4A shows a full-field fluorescence image made within 5 min of exposing mycelium to BCECF-AM and while dye was still present in the extracellular medium. Comparison of cytoplasmic fluorescence intensity with background shows that, by this stage, very little fluorescence was present in the cytoplasm. However, all cells were outlined by fluorescence, which was particularly bright at the thickened junctures between cross-walls and the outer boundary wall. Since fungal walls contain hydrophobic polymers, including chitin, their staining with lipophilic dyes per se was not surprising, but the appearance of fluorescence implies that at least some extracellular hydrolysis of BCECF-AM had occurred.

A separate experiment, now on the confocal microscope, shows simultaneous interference-contrast (Fig. 4B) and fluorescence (Fig. 4C) images from a moderately vacuolated hypha which had been exposed to BCECF-AM for less than 20 min. Although plenty of vacuoles were evident, the principal fluorescence was homogeneously distributed, except in the immediate vicinity of the cross-wall (far left of Fig. 4B,C), and in scattered intense spots (e) which do not represent vacuoles, but hydrophobic, ergosterol-bearing, hexagonal plates (Tsuda and Tatum, 1961; Hoch and Maxwell, 1974). These structures can be as large as 2 μm across and generally lie flat against the plasma membrane. However, during violent streaming – as upon cell rupture – they are stripped off and can actually serve as plugs (Slayman, 1965). Other pH probes also distributed into the hexagonal plates (see below).

Obviously, BCECF cannot be used to monitor cytosolic pH, at least for experiments lasting more than a few minutes, and in searching for a truely cytosolic pH indicator, we tested C.SNARF, which has recently been used in the -AM acetate form on cardiac myocytes, fibroblasts, yeast cells and a variety of other preparations (Haworth et al. 1991; Seksek et al. 1991; Blank et al. 1992). Distribution of this dye in hyphae having different degrees of vacuolation is demonstrated in Fig. 5, for a preparation maintained in the presence of extracellular C.SNARF-AM. Comparison of the fluorescence image (Fig. 5B) with the phase-contrast image (Fig. 5A) shows clearly that the dye was excluded from the vacuoles and was essentially homogeneously distributed in non-vacuolated cytoplasm. This result should be contrasted specifically with that of Fig. 3. Also in contrast with BCECF, the dye appeared not to be concentrated in the cell wall and may actually have been slightly excluded from the cross-wall (visible in Fig. 5A at top center). Nevertheless, it still accumulated in the hexagonal plates (e; top right of Fig. 5B). Unfortunately, the utility of C.SNARF as a cytosolic pH indicator proved severely limited, since it washed out of Neurospora cytoplasm with a time constant of approximately 30 s, when extracellular dye was removed. That fact, however, did make possible visualization of residual dye in the vacuoles, at a few per cent of the average cytoplasmic level, after long exposures to C.SNARF-AM (i.e. 1–3 h; data not shown). But such fluorescence was smeared around well-defined spots in the confocal micrographs, suggesting a slow leak of residual dye out of the vacuoles, following loss of the major fraction from the cytoplasm.

A structural comparison of BCECF-AM and C.SNARF-AM is provided in Fig. 6. The outstanding structural difference between these molecules is that only two groups on C.SNARF are available for external esterification, whereas five groups can be esterified on BCECF. To assess whether this might be important to the different modes of distribution for the two dyes, we tested esterified C.SNARF-calcein, which contains a total of four externally esterified residues (Fig. 6C). Fluorescence from this compound, indeed, distributed like BCECF, as demonstrated in Fig. 7 for highly vacuolated hyphae.

Fungal and plant cells are in some respects metabolically quite distinct from animal cells, so these observations on Neurospora were at first viewed as being of specialized, rather than of general, interest. Subsequent experiments with murine fibroblasts, however, changed that view. Fig. 8 displays a comparison of phase-contrast (Fig. 8A) and fluorescence (Fig. 8B) images from a confluent culture of normal NIH 3T3 cells. Again the most impressive feature is heterogeneity. Cytoplasmic (or perinuclear) fluorescence in some cells (a) appeared almost homogeneous, i.e. indistinguishable from background graininess in the photographic film, while in other cells (b) a strongly mottled or punctate distribution dominated. Nuclei were mostly dark (c), but in a few cases were either themselves fluorescent or overlain by thick fluorescent cytoplasm. Fluorescence from peripheral regions of most cells was characteristically spotty. These results are qualitatively similar to ones previously reported for BCECF-AM-treated Swiss 3T3 cells stimulated by addition of serum to quiescent cultures (Bright et al. 1989), although in that case punctate fluorescence was attributed to accumulation of dye in mitochondria rather than in EMVs.

Many, but not all, cultures of NIH 3T3 cells displayed such heterogeneity upon staining with BCECF-AM. Staining at lower concentrations and shorter times reduced the appearance of punctate fluorescence. But so did very slight defocusing of the microscope, thus emphasizing in a practical sense the critical dependence of perceived results upon the quality of the optical path.

Resolvable punctate distribution of BCECF fluorescence in fibroblasts was greatly enhanced in the giant cells which develop in cultures transfected with the gene for Neurospora plasma-membrane H⁺-ATPase. This heterologous protein, whose presence has been demonstrated by both cytochemical and functional assays (V. V. Moussatos, R. Nakamoto, T. Claudio, C. W. Slayman and C. L. Slayman, in preparation) distributes widely among surface and cytoplasmic membranes and is associated with extensive enlargement of endomembranal vesicles (EMVs). Phase contrast and fluorescence images of one such cell from isolate B17 are shown in Fig. 8C,D. In this typical case, nuclear fluorescence was intense and nearly homogeneous, while cytoplasmic fluorescence, both perinuclear and peripheral, was conspicuously punctate against a fine-grained background.

Although systematic work to explore cytosolic pH in plants and fungi by means of fluorescent H⁺ probes has only just begun, an extensive literature for comparison already exists with respect to cytoplasmic Ca²⁺ in plants and fungi. No single method of loading cells with calcium indicators has proved generally satisfactory, but it is safe to say that the least satisfactory method tested has been that of the -AM derivatives (variants of Indo-and Fura-dyes). In some cases, these derivatives are either non-penetrant or not trapped (Gilroy et al. 1986; Read et al. 1992), while in others they target promptly to vacuoles and/or intracellular tubulovesicular networks (Rathore et al. 1991; Kiss et al. 1991; Read et al. 1992). Electroporation, low-pH loading (extracellular pH 4.5–5.5) and mild detergent washing, for potassium salts of the Ca²⁺ indicators, are much less prone to give punctate fluorescence than is -AM loading. In the few cases where plants and other fungi have actually been tested with esterified H⁺ probes, the results have been similar to those just reported in Neurospora, albeit with cell-or species-specific variations. The most extensive data exist for the yeast Saccharomyces cerevisiae. C.SNARF(-AM), for example, is obviously excluded from yeast vacuoles (Haworth et al. 1991), yielding images very much like those for the vacuolated hyphae in Fig. 5 . However, in clear contrast with the results on Neurospora, C-SNARF has a rather long half-life (>1.5 h) in the apparent free cytosol of yeast. And BCECF(-AM) is said to be almost totally excluded from yeast cells (Haworth et al. 1991), while another popular pH indicator, 6-carboxyfluorescein (not tested on Neurospora), is strongly accumulated in yeast vacuoles when delivered as the neutral diacetate (Preston et al. 1989).

Such results are not at all surprising in view of the fact that one of the many functions of plant and fungal vacuoles is that of lysosomes (Wiemken et al. 1979; Wagner, 1981). They are acidic (typical vacuolar pH values of approximately 6.0±0.5) and store a variety of lytic enzymes (Wiemken et al. 1979; Preston et al. 1989). They also accumulate weak-base indicators, such as quinacrine (Preston et al. 1989; Haworth et al. 1991), just as animal lysosomes do, but without such conspicuous osmotic swelling as in the ′lysosomotropism′ described by de Duve et al. (1974). Finally, they sequester endocytosed macromolecules, such as fluorescein-labelled dextran (Makarow, 1985; Makarow and Nevalainen, 1987; Preston et al. 1989), and even do so when the macromolecules are injected through intracellular pipettes (Read et al. 1992).

The problem of cytoplasmic compartmentation is more subtle in animal cells than in plant or fungal cells, because of the much smaller average size of the EMVs. Furthermore, vesicles easily visible in the light microscope, such as the osmotically swollen lysosomes seen in lysosomotropism, certainly are morphologically abnormal. Therefore, despite scattered reports on inhomogeneous distributions of pH probes (see, for example, Bright et al. 1989; Mozingo and Chandler, 1990; Rashid and Horobin, 1991; Miller et al. 1993), the major concern in the development of cytosolic probes for animal cells has been to find combinations of probes and experimental conditions which give no evidence for dye compartmentation.

Most importantly, low concentrations of dyes have been used in the loading medium (1–10 μmol l⁻¹, rarely less than 1 μmol l⁻¹) to avoid obvious pitfalls, such as saturation of cytosolic esterases and photodynamic damage, which have repeatedly been proposed as mechanisms producing dye accumulation in cytoplasmic vesicles. Other important physical measurements have also been carried out on cells presenting visually homogeneous distributions of pH indicator dyes: fluorescence photobleaching recovery and fluorescence anisotropy of BCECF, to look for anomalous diffusion or viscosity which might signal dye-binding to large molecules or dye-trapping in vesicles (Bright et al. 1987; Fushimi and Verkman, 1991; Kao et al. 1993); and concentration–relaxation following mild detergent treatment of loaded cells, to look for multiple-component efflux (see, for example, Paradiso et al. 1986; Grinstein et al. 1989). The overall conclusion from such experiments has been that cells that do not show microscopically visible evidence of dye compartmentation also do not show other physical evidence of compartmentation.

However, despite the elegance and sophistication of these various measurements, interpretation of many results is not straightforward. What, for example, does detergent-stimulated release of dye molecules with rapid single-component kinetics really mean when low extracellular concentrations of other membrane agents (e.g. nigericin; Maxfield, 1982) can be used to control ion concentrations inside bona fide endosomes? And although the interpretation of bleach–relaxation and anisotropy experiments seems clear with respect to buffers and viscous sugar solutions, how would they compare with measurements on dense cytoplasm-like suspensions of actual endosomes, lysosomes or vacuoles?

An essential point about the experiments reported here on Neurospora is that the distribution of vacuolar sizes was determined by growth conditions, not by loading conditions, as is clearly demonstrated by comparison of Fig. 2 (unstained cells) with Fig. 3. For extracellular concentrations of BCECF-AM between 3 μmol l⁻¹ and 100 μmol l⁻¹, the only significant difference was a rise of fluorescence intensity in parallel with concentration, not an increase of vesicle size. In other words, our usual dye assay concentration of 30 μmol l⁻¹ made the vacuoles brighter – and easier to observe – than did the ′safer′ 3 μmol l⁻¹, but it did not produce appreciable swelling. The same was also true, at least approximately, for the giant PMA1-transfected fibroblasts (Fig. 8C,D).

The principal advantage, then, of using fungi for dye-distribution studies, rather than either plant or animal cells, is that they contain an appreciable population of intermediate-sized EMVs, which are small enough not to obscure the free cytosol (as often happens in plant cells) but large enough to be easily visible under ordinary microscopic conditions. Furthermore, their size can be controlled. And the fact that even the smaller visible EMVs in normal (sugar-replete) Neurospora hyphae concentrate BCECF from the -AM derivative (cf. Fig. 2A and Fig. 3A) strongly suggests that still smaller, unresolvable vesicles also concentrate the dye.

The general problem of how to visualize putative dye accumulations in submicroscopic vesicles has not been solved. In principle, well-separated fluorescent particles can be distinguished individually – if they are bright enough. But dim signals, close packing and out-of-focus imaging would all work to convert ′lawns′ of fluorescent particles into observed quasihomogeneous fluorescent fields in the microscope. Indeed, the possibility exists that such is the origin of the mottling of fluorescence in the brightly stained ′homogeneous′ cytoplasm of Fig. 1 (Neurospora) or in the nearly homogeneous perinuclear regions of Fig. 8B (NIH-3T3 cells), as well as in numerous published photographs of cellular regions said to show cytosolic distributions of BCECF fluorescence (see, for example, Bright et al. 1987; Fushimi and Verkman, 1991). Clearly, the confocal microscope can help considerably with this problem, by providing 10-to 100-fold greater rejection of background than does conventional fluorescence microscopy (Webb et al. 1990; Sandison and Webb, 1994). But newer techniques for suboptical resolution may be required to give a fully convincing picture of intracellular dye distribution.

Elementary dye chemistry, per se, yields few clues to the reasons for accumulation of BCECF and C.SNARF-calcein fluorescence in acidic organelles. Unlike weak-base indicators and drugs, they should not accumulate by pH-trapping. Their actual entry into Neurospora vacuoles, and presumably into the fibroblast EMVs as well, is probably by electroneutral diffusion – the process supposed to underlie cytosolic entry – or perhaps by voltage-driven flux of partially de-esterified anions. (The vacuolar membrane voltage in Neurospora is about -40 mV, cytoplasm negative to vacuolar interior; C. L. Slayman, unpublished experiments.) An additional possibility is that some anionic forms of BCECF or C.SNARF-calcein might enter via organic anion porters known to exist in fungal vacuolar membranes (Sato et al. 1984; Kulakovskaya et al. 1991; Wada et al. 1992). But the most likely mechanism for dye accumulation, per se, in vacuoles is complete de-esterification in the vacuole by the action of resident hydrolases (Wiemken et al. 1979; Jones, 1983). This circumstance raises important unanswered questions about the nature and localization of esterases which have been presumed to act on -AM esters within free cytosol.

The finite extracellular (background) fluorescence, together with transient fluorescence in cell walls (Fig. 4), in hexagonal plates (Fig. 5) and in cytosol (Fig. 4), raises an additional problem. Since fully esterified forms of these dyes are nearly colorless and non-fluorescent (Martin and Lindqvist, 1975; Whitaker et al. 1991; Haugland, 1992), partial hydrolysis must have occurred when any fluorescence is observed: i.e. in the medium and/or the cell wall and in the cytoplasm. At a minimum, this means that one of the nuclear acyl groups must have been removed and the resultant -OH residue oxidized, to allow opening of the lactone ring (Martin and Lindqvist, 1975). Perhaps if only that happens, the bulkier molecules still remain sufficiently lipophilic to cross the endosomal membranes. The observed rapid washout of cytoplasmic fluorescence following C.SNARF-AM loading, then, would represent ejection of partially de-esterified dye across the plasma membrane, perhaps aided by the substantial negative membrane voltage (approximately 200 mV, cytoplasm negative; Slayman, 1965). That would still leave the disparate behavior of cytoplasmic fluorescence due to BCECF, compared with C.SNARF, to be explained (see text description of Fig. 5), as well as the fact that depolarizing agents, such as sodium azide, or membrane-stabilizing agents, such as extracellular calcium (>100 μmol l⁻¹), do not act to conserve cytoplasmic fluorescence due to C.SNARF (test observations made during these experiments).

In the general literature on dye measurement of cytoplasmic pH, possible organellar targeting of AM-derived indicators is often mentioned, but also often dismissed, particularly when conventional spectrofluorimetry on cell suspensions is being performed, rather than microscope-based single-cell analysis. Mistargeting, however, is a serious and non-trivial problem, whose quantitative effects can vary with many experimental parameters. The facts that many EMVs lie below the limit of resolution in the light microscope (Tanaka, 1987; Gruenberg et al. 1989), and that satisfactory luminence-and contrast-enhancement techniques for visualizing these particles remain to be developed, mean that direct assessment of possible non-uniform dye distribution will remain a practical problem for at least a few more years.

A second problem lies in the meaning of pH calibration maneuvers. The standard technique for calibrating cytoplasmic pH indicators (Thomas et al. 1979) is to bathe the cells/tissues in buffer containing 10–100 μmol l⁻¹ nigericin or monensin at several fixed values of extracellular pH, on the assumption that ionophore-mediated H⁺/K⁺ and H⁺/Na⁺ exchanges will allow cytosolic pH to equilibrate with extracellular pH. Apart from the fact that ionophores of these classes do not work well or reliably on most intact plant or fungal cells, even for animal cells a heterogeneous distribution of indicator dye could vitiate calibrations based on ionophore action solely at the plasma membrane. That complication is not reduced by the fact that the same ionophores can be used with intact cells to alter the pH of EMVs (Maxfield, 1982).

A third point is that, since the number and size distribution of EMVs, and concomitant cytoplasmic buffer capacity (Sanders and Slayman, 1982), can clearly be altered by simple physiological maneuvers in at least some species (certainly in Neurospora, probably also in yeast), it is at least worth asking to what extent such things might occur in other types of cells, becoming yet another source for errors in calculation of cytoplasmic pH changes and H⁺ fluxes.

The safest route to take from these experiments, then, is to avoid using esterified forms of BCECF, C.SNARF and related pH indicators on compartmented biological materials. When such dyes are to be used to judge cytosolic pH (or other ion concentrations), then methods should be found to load them as anions, rather than as neutral molecules or weak bases. For studies of individual cells, microinjection could be used; for studies on populations, transient disruption of plasma membranes via reversible chemical reagents (e.g. mercurials, elevated concentrations of extracellular nucleotides, detergents), mild mechanical disturbance (e.g. scrape-loading; McNeil et al. 1984, contact pressure from glass beads, mild sonication) or electroporation could be used.

C.L.S. is indebted to Yale University for sabbatical salary and to many students and fellows in the Developmental Resource for Biophysical Imaging and Opto-electronics (DRBIO) at Cornell University, Ithaca: especially to Mr. Kevin Hodgson and Dr David Sandison. We are also indebted to Dr Michael Caplan (Yale University), for use of the Zeiss Axiophot microscope and for advice on the manuscript, and to Drs Richard Haugland (Molecular Probes, Inc.) and John E. T. Corrie (National Institute for Medical Research, Mill Hill, London), for references and insights into dye chemistry. The work was supported by Research Grant DEFG02-85ER 13359 from the US Department of Energy (C.L.S.), by Fellowship DF92-040 from the Donaghue Medical Research Foundation (V.V.M.), and by Grants 08-P1RR04-224A from the National Institutes of Health and DIR-8800278 from the US National Science Foundation, to DRBIO at Cornell (W.W.W.).