An immunocytochemical analysis of the vacuolar proton pump in Dictyostelium discoideum

V-H⁺-ATPases are the proton pumps that acidify various organelles in members of all four eukaryotic kingdoms (Brown et al., 1987; Forgac, 1989; Padh et al., 1989b; Stone et al., 1990; Nelson, 1992). In both primary sequence and quaternary protein structure, they appear to be distant homologues of the F1-ATPases (Bowman, 1983; Nelson and Taiz, 1989; Bowman et al., 1992). It is thus possible that, in some primordial eukaryote, the proton pump was moved from plasma membranes to endomembranes and its function reversed so as to use ATP hydrolysis to form H⁺ gradients (Nelson, 1988; Forgac, 1989; Gogarten et al., 1989; Nelson, 1992; Kibak et al., 1992). The acidification of endocytic compartments might have benefitted early heterotrophs in countering the hazards of ingesting live prey (Nelson and Taiz, 1989). Proton gradients also drive solute co-transport and play an informational role in guiding proteins to the correct endomembrane compartment (Mellman et al., 1986; Forgac, 1989).

We now describe the application of polyclonal antisera directed against the V-H⁺-ATPase of the amoeba, Dic -tyostelium discoideum, as a cytochemical probe.

Dictyostelium discoideum, strain AX-3, was grown axenically to late log phase (5×10⁶ to 10×10⁶ cells/ml) as described (Padh et al., 1989a). Homogenization, subcellular fractionation and isolation of lysosomes and the proton pump-rich organelles, called acidosomes, have been described (Nolta et al., 1991). As a crude fraction of large organelles we used the pellet generated by centrifugation for 3 min at 4,000 r.p.m. in a Sorvall SS-34 rotor (Nolta et al., 1991). Contractile vacuole complexes were isolated from the remaining supernatant by two cycles of sucrose density gradient centrifugation (Nolta and Steck, unpublished data).

The V-H⁺-ATPase antigen was solubilized from purified acidosomes in dodecyl maltoside by sequential rate-zonal centrifugation and gel filtration (Nolta et al., 1991). V-H⁺-ATPaseactivity was assayed colorimetrically (Padh et al., 1989b; Nolta et al., 1991); we added 0.1 mM sodium vanadate here to inhibit extraneous P-type ATPase activity.

After drawing preimmune serum, three New Zealand white rabbits were immunized subcutaneously with 50-100 μg of native purified V-H⁺-ATPase every other week. Samples (50 ml) of blood were collected at those times and heat-treated, and aliquoted serum was stored at −70°C.

An 8 mg sample of purified native V-H⁺-ATPase was conjugated to 4 ml Affigel 15 beads as recommended by Bio-Rad (Richmond, CA). Antibodies were eluted with 0.1 M glycine-HCl (pH 2.2) plus 0.25 M MgCl₂; 2 ml aliquots were collected into 0.2 ml 1 M Tris-base. Fractions were concentrated in Amicon miniconcentrators with a 100 kDa cutoff (see Harlow and Lane, 1988, pp. 552-554).

We used 8% polyacrylamide slab gels in sodium dodecyl sulfate without reduction in Fig. 1 and 10% gels with reduction in Figs 2 and 4; in both cases the stacker was a 5% gel (Laemmli, 1970). Blotting by electrotransfer to nitrocellulose membranes (0.45 μm pore, Schleicher and Schuell, Keene, NH) and blocking with powdered milk were as described (Harlow and Lane, 1988, pp. 490-492). The blots were gyrated for 30 min in buffer T containing dilutions of antiserum and 1% powdered milk, washed in buffer T and gyrated with ¹²⁵I-Protein A at 1×10⁶ c.p.m./ml for 30 min (Harlow and Lane, 1988, pp. 497-503). The blots were washed, dried, and autoradiographed at −70°C for 1 to 3 days.

Homogenates were centrifuged for 30 min at 18,000 r.p.m. The pellets were mixed with PBS and 1% Triton X-100 and incubated for 30 min on ice with antibodies. An excess of fixed, prewashed Staphylococcus A particles (Sigma Chemicals, St. Louis, MO) was added and the mixture incubated for another 30 min (Harlow and Lane, 1988, pp. 465-467). The particles were washed extensively with PBS and incubated at 80°C in 50 mM Tris-HCl, 2% sodium dodecyl sulfate, 0.1% bromophenol blue, pH 7.4. The particles were pelleted and the unreduced supernatants electrophoresed as described above.

Suspensions were spread on glass coverslips for 30 min and the excess buffer removed with filter paper. Thin agar slabs were placed over the cells and then covered with a solution of 1% formaldehyde and 0.05% Triton X-100 in methanol for 10 min at −20°C (Yumura et al., 1984; Zhu and Clarke, 1992). The coverslips were rinsed well with buffer T and swirled at room temperature for 30 min in small Petri dishes containing antiserum diluted in buffer T. The coverslips were rinsed twice in buffer T and incubated for 30 min with a 1:100 dilution of TRITC-conjugated goat anti-rabbit antibody (Vector Labs, Burlingame, CA). Coverslips were rinsed, mounted with glycerol, and photographed (Padh et al., 1989a).

Suspensions of cells and subcellular fractions were mixed with equal volumes of fixation buffer: 2.5% glutaraldehyde in buffer E. The cells were rocked at 0-5°C for 2-4 h, washed with fresh fixation buffer, rocked overnight, washed twice with buffer E and pelleted. Pellets were post-fixed with 2% OsO₄ in 50 mM sodium cacodylate (pH 6.8) until the pellets were completely dark (2-6 h). Pellets were then dehydrated and embedded in Eponate 12 (Ted Pella, Inc., Redding, CA). Silver sections (=50-100 nm) were cut and counterstained with uranyl acetate and lead citrate.

Alternatively, cells were fixed with formaldehyde in methanol and processed as described for fluorescence microscopy. Fixed cells were pelleted, osmicated, embedded in Epon, and sectioned as described above.

The electron microscope was a Siemens 101 operated at 80 keV. Measurements were performed on primary micrographs taken on Kodak electron image film (SO-163).

Dehydrated pellets of fixed cells and isolated membranes were embedded in Lowicryl K4M at room temperature according to the manufacturer’s directions. Sections were placed on nickel grids and rinsed for 30 min in PBS containing 3% bovine serum albumin. Grids were floated for 1-2 h at room temperature on drops of primary antiserum diluted in this buffer, then rinsed dropwise with the buffer. The grids were then treated in the same way with goat anti-rabbit IgG conjugated to 20 nm gold particles (Electron Microscopy Sciences, Fort Washington, PA) diluted 50-fold in the aforementioned buffer. After a water rinse, the grids were counterstained for 15 min with 2-3% uranyl acetate.

For contractile vacuole complexes, the isolates were washed, suspended in PBS, reacted with primary antibody (1:1,000 dilution), washed again and incubated with the immunogold reagent described above (50-fold dilution). The mixture was centrifuged on a 25% to 45% sucrose gradient in a Beckman SW-41 rotor for 1 h at 40,000 r.p.m. The buoyant membrane fraction bearing the peak of alkaline phosphatase activity was washed, fixed, osmicated and embedded in Epon as described above.

While not shown, every immunocytochemical image shown here was accompanied by a control using preimmune serum; these were all essentially negative.

Polyclonal antisera were prepared against the purified V-H⁺-ATPase of Dictyostelium. Lane 1 of Fig. 1 shows the electrophoretic profile of the antigen in this gel system; the five largest of the eight polypeptide subunits of the V-H⁺-ATPase were resolved. Immunoprecipitation of washed, unfractionated organelles in the absence of detergent by antiserum of increasing concentration is shown in lanes 2-4 of Fig. 1. The major, low-mobility band trailing the V-H⁺-ATPase subunits was unreduced IgG (Kane et al., 1989). All of the proton pump polypeptides present, and no other Dictyostelium polypeptides, were precipitated by the antibody (lane 4); this reaction cleared less than 5% of the acid phosphatase (lysosomes), alkaline phosphatase (contractile vacuoles), F1-ATPase (mitochondria) and RNA (rough endoplasmic reticulum) from unfractionated organelle preparations. Thus, the organelles that contain most of the proton pumps did not contain appreciable amounts of other proteins.

The immunoprecipitation shown in lanes 1-4 would not have detected luminal antigen in organelles with unreactive surfaces (possibly, mitochondria). However, immunoprecipitates of unfractionated organelles, dissolved in 1% Triton X-100 to expose such latent antigens, brought down only the polypeptides of the V-H⁺-ATPase with dramatically lower yield (lanes 5-7). Controls demonstrated that none of the bands was pelleted when antibodies or Staphy -lococci were omitted (lanes 8 and 9) or preimmune serum was substituted. We also analyzed immunoprecipitates of unfractionated organelles on 8% polyacrylamide tube gels to resolve the lowest molecular mass subunits of the V-H⁺-ATPase; all eight polypeptide components of the V-H⁺-ATPase and no other bands were seen (not shown, but see Nolta et al., 1991; Nolta, 1993).

The polypeptides in whole cells reactive with antiserum at 50,000-fold dilution were precisely those of the purified acidosomes (lanes 1 and 2 of Fig. 2). The reactivity of these V-H⁺-ATPase polypeptides was band 2 ≃ 5 > 1 ≃ 3 > 4; it was uncertain whether bands 6, 7 and 8 were immunoreactive. The faint band at the dye front could have been band 7 and/or band 8 of the V-H⁺-ATPase or degradation products of any of the pump polypeptides (see below). The unidentified band between bands 1 and 2 (Mr ∼78,000) did not correspond to any pump polypeptide detected with Coomassie blue. The soluble cytosol did not react with our antisera (Fig. 2, lane 3). On the other hand, the growth medium (which contained yeast extract) contained immunoreactive material migrating at the dye front (not shown).

Affinity-purified antibodies reacted with all the same polypeptides at a titer about an order of magnitude lower than whole antiserum (Fig. 2, lane 4). We therefore routinely used unfractionated antisera. The reaction of preimmune sera with homogenate proteins was always negative (Fig. 2, lane 5).

Whole antisera and affinity purified antibodies but not preimmune serum were effective inhibitors of V-H⁺-ATPase activity (Fig. 3). The inhibitory activity of affinity-purified antibodies was about 10% of the whole antiserum. Neither oligomycin-sensitive mitochondrial nor vanadate-sensitive plasma membrane ATPase activity was affected by the antisera.

Fig. 4B shows the immunoblot of alternate fractions of an equilibrium sucrose density gradient of a Dictyostelium homogenate. For reference, gel A shows the Coomassie blue-stained polypeptide profile of the acidosomes. The peak of immunoreactivity in gradient fraction 9 coincided with the peak of V-H⁺-ATPase activity, which defines the acidosomes (Fig. 5). The immunologically reactive bands corresponded to those of the V-H⁺-ATPase (Fig. 2); the principal reactive polypeptides were, once again, bands 2 > 5 > 1 > 3 > 4.

Other organelles were not conspicuously reactive at this level of resolution. In particular, lysosomes and mitochondria were not visibly associated with proton pump antigens (Fig. 4, gel B, lane 1). The fact that the mitochondrial F1-ATPase is not cross-reactive with any of the V-H⁺-ATPase subunits has been previously reported for other organisms (Bowman, 1983; Konishi et al., 1990).

The disposition of proton pumps in intact Dictyostelium was surveyed by immunofluorescence microscopy (Fig. 6). The plasma membranes and cytosol were negative. The fluorescence was most intense over numerous bodies in the central cytoplasm, which ranged in size from small granules to one or two circular profiles of up to ≃2 μm diameter. The identity of these abundant membranous structures is uncertain at this level of resolution and was therefore pursued by electron microscopy.

In thin sections, the cytoplasm of vegetative D. discoideum bore a multitude of large and small vacuoles (Fig. 7; see also de Chastellier et al., 1978). Of particular interest were clusters of vesicles, vacuoles and slit-like tubules or saccules with strikingly clear lumina, such as those bracketed in Fig. 7A. Fig. 7B shows a high magnification view of a similar cell from which the background of soluble proteins was reduced by fixation in formaldehyde plus methanol. A cluster of clear vacuoles extended diagonally upwards from the lower right. As argued below, we take these to be the acidosomes (ac).

Upon isolation, acidosomes also appeared to be congregations of vacuoles and tubules with empty lumina and diameters of 0.1 to 1.0 μm (Fig. 8A). The persistence of clustering after homogenization could explain the ready sedimentability of acidosomes at low speeds (Nolta et al., 1991). Many of these vacuoles took the form of concentric rings in thin section (opposed arrows); we attribute these figures to deep invaginations rather than double membranes (Nolta, 1993). This invagination was attributed to the osmotic shrinkage of the luminal spaces of the acidosomes by the hypertonic sucrose density gradient and fixation media. The small profiles within the large double-membrane profiles might reflect additional finger-like invaginations. Numerous ≃12 nm diameter particles projected ≃12 nm from the outermost and innermost faces of the double membrane profile; we take these to be the V-H⁺-ATPase at the cytoplasmic surface of the acidosomes.

In negatively stained spreads, intact acidosomes had a pleiomorphic sacculotubular morphology (see also Heuser et al., 1990). Sometimes, large saccules tapered into sleeves with multiple irregular fingers (Fig. 8B); other acidosomes were elongated chains of broad sacs and narrow tubules (Fig. 8C). The constrictions in the latter complexes suggested that they were pinching apart. We imagine that thin sections across such complex figures would yield the kinds of profiles seen in Figs 7 and 8A. There was a profusion of ≃12 nm surface projections, the V-H⁺-ATPases (Nolta et al., 1991).

Virtually all the membranes in acidosome isolates were strongly reactive to immunogold decoration with our antisera, even at a dilution of 15,000-fold (Fig. 9A). The homogeneity of such images confirmed the purity of these isolates (Nolta et al., 1991). A total of 97% of the colloidal gold lay within 15 nm of membrane profiles; membranefree probe was negligible. Gold was occasionally seen on luminal surfaces. This may have reflected the reactivity of integral subunits extending to that face (Arai et al., 1988; Nelson, 1988; Moriyama and Nelson, 1989a,b; Parry et al., 1989). In addition, deep membrane invaginations (Fig. 8A) might have created the false impression of a luminal antigen. Furthermore, sidedness artefacts have been noted in the staining of membrane-associated proteins with immunogold (Baines and Korn, 1990; Zhu and Clarke, 1992).

We also examined the mixture of membranes in lowspeed pellets of homogenates (Fig. 9B). Most of the gold particles were concentrated on membranes resembling acidosome clusters (compare the upper left corner of Fig. 9B with Fig. 9A). Neither the nuclei (n) nor the mitochondria (m) were immunoreactive, confirming the specificity of our antisera and their lack of cross-reactivity with the F1-ATPase.

These organelles were identified by the moderate electron density of their lumina and their frequent membranous inclusions (arrows in Figs 7A and 9B; see also de Chastellier et al., 1978). Isolated acidic vacuoles, rich in acid hydrolases and endocytic markers, had the same morphology (Fig. 10; see also Rodriguez-Paris et al., 1993). Sometimes, lysosomes appeared within acidosome clusters (Fig. 7A).

Our antibodies to V-H⁺-ATPase stained the lysosomes both at their periphery and, in some instances, over their lumina. The antigen on the membrane was sparse (Fig. 10) and presumably corresponded to the ≃10% of cellular V-H⁺-ATPase activity associated with isolated lysosomes (Padh et al., 1991a,b; Nolta et al., 1991; Rodriguez-Paris et al., 1993). The immunodecoration of lysosomes was often more intense over their lumina than on their envelopes (Figs 9B and 10A). Generally, the luminal antigen was associated with internalized membranes. We could not distinguish whether the luminal membranes and their associated antigen derived from the simple invagination of the lysosome envelope or from the autophagy of acidosomes. The high pump density on the internal membranes supports the latter possibility.

The intensity of the immunocytochemical reaction of the lysosomes (Figs 9B and 10) was out of proportion to their small contribution to the cell’s total pump antigen in immunoblots (Figs 2 and 4) and their V-H⁺-ATPase activity on gradients (Padh et al., 1989b). Immunogold labelling of Lowicryl sections may favor proteolytically degraded pumps and thus not be quantitative.

Fig. 10 illustrates the complex contour of many lysosomes. Often, they appeared to be multilobulated, as if the product of multiple fusions. Sometimes, they had tubular extensions (asterisks in Fig. 10A).

Occasionally, crude membrane pellets contained what appeared to be microbes ingested into tight-fitting envelopes. The envelope was highly decorated by immunogold using antibodies to the V-H⁺-ATPase; membranes within the microbe were also reactive. The image in Fig. 11A was reminiscent of phagocytosed yeast particles (Hellio and Ryter, 1980). Because the putative primary phagosome in Fig. 11A appears to have not yet fused with other vacuoles, it may have acquired its V-H⁺-ATPases by some other means; for example, through interactions with acidosomes (Lukacs et al., 1990; Padh et al., 1991b).

Phagosomes containing latex particles were similarly decorated with immunogold in intact cells (Fig. 11B). We note that Fig. 11A contains several open membrane sheets that bear no gold (asterisks); these structures appear to be fragments of plasma membranes (Luna et al., 1984).

Heuser (1991) has suggested that the cytoplasmic surface projections previously seen on the accessory tubules (spongiomes) of contractile vacuoles (Patterson, 1980) might be vacuolar proton pumps. We have now obtained direct support for this hypothesis in Dictyostelium, Acanthamoeba and Paramecium (Nolta et al., 1992; Nolta and Steck, unpublished data; A. Fok, K. V. Nolta, T. L. Steck and R. Allen, unpublished data).

The one or two contractile vacuoles per vegetative Dic -tyostelium cell were recognized as large and bladder-like (cv in Fig. 7B). The spongiome ordinarily associated with contractile vacuoles has not been identified in intact Dic -tyostelium (see de Chastellier et al., 1978). However, spongiomes could be clearly identified in isolates of contractile vacuole complexes with the help of antibodies to the V-H⁺-ATPase (Fig. 12). They appeared as conglomerates of tiny tubules and/or vesicles strongly decorated by immunogold and associated with large lucent vacuoles. The occassional large, thick ring of V-H⁺-ATPase seen by immunofluorescence (Fig. 6) could thus reflect a rim of spongiome surrounding a contractile vacuole bladder. All of these data support the hypothesis that the pegs projecting from the surface of the contractile vacuole spongiome in Dictyostelium are proton pumps.

Having defined morphological landmarks on organelles in isolation, we examined thin sections of intact cells embedded in Lowicryl and decorated with immunogold. A variety of clusters consisting of vacuoles, tubules and/or saccules with clear lumina bore heavy stain on their cytoplasmic surfaces (Fig. 13A-C). The tubular forms were typically subjacent to the plasma membrane, as seen in (A) and (B). We took all of these structures to be acidosomes. The other major richly decorated organelles appeared to be lysosomes (Fig. 13D). As in isolates, these bodies were round or elongated, had electron-dense lumina, and often contained antigen-bearing membrane material inside (multivesicular bodies).

We examined the reactivity against Dictyostelium of affinity-purified rabbit antibodies from three other sources (Gillespie et al., 1991). The reagent prepared against intact chromaffin granule V-H⁺-ATPase was highly reactive with polypeptide band 1 of the Dictyostelium proton pump, slightly reactive with band 2, and not notably reactive with the other subunits. Two antibody preparations monospecific for the alpha and beta subunits of tonoplast V-H⁺-ATPase from Kalanchoe daigremontiana reacted exclusively with the corresponding polypeptides of Dictyostelium (bands 2 and 3; not shown). The patterns of cytochemical staining with these three heteroantibodies of Lowicryl sections of intact Dictyostelium and isolated fractions thereof were indistinguishable from those shown in Figs 9 and 13. We infer that there is a broad antigenic similarity between the V-H⁺-ATPases of the protozoan, Dictyostelium, and plants and animals. These experiments also corroborate the finding that our antibodies were specific for V-H⁺-ATPase.

Four types of animal cells were examined (Fig. 14A-D). In all cases, preimmune sera gave no reaction. Immune sera did not stain the cytosol or plasma membranes but reacted with a multitude of 0.2-1 μm diameter vesicles and vacuoles scattered throughout the cytoplasm. These patterns were similar to those elicited by antibodies generated against mammalian V-H⁺-ATPases (Yurko and Gluck, 1987; Rodman et al., 1991; Marquez-Sterling et al., 1991). They also resembled the pattern of acridine orange staining of acidic vacuoles in these cells. Noticeably fewer vacuoles reacted with our antisera than accumulated acridine orange. Similarly, the FITC-dextran ingested in parallel by macrophages assumed a pattern similar to that seen in Fig. 14D, except that the vacuoles bearing the endocytic dye were significantly more abundant than those stained by our antibodies. Such effects have also been reported with antibodies homologous to V-H⁺-ATPases in animal cells (Yurko and Gluck, 1987; Rodman et al., 1991; Marquez-Sterling et al., 1991).

The antisera prepared against the V-H⁺-ATPase of Dic -tyostelium discoideum were of high titer, stringent specificity and broad cross-reactivity with every eukaryotic species tested. They inhibited V-H⁺-ATPase catalytic activity; they selectively precipitated both the acidosomes and the proton pumps solubilized therefrom; and they labelled the five pump subunits of highest molecular mass on immunoblots. The subcellular distribution of the V-H⁺-ATPase antigen activity was similar to that of its catalytic activity; thus, there was no evidence for a substantial population of inactive proton pumps.

In other eukaryotes, the V-H⁺-ATPase has been identified principally in lysosomes, endosomes, coated vesicles, the Golgi apparatus, secretory vacuoles and phagosomes (Mellman et al., 1986; Yurko and Gluck, 1987; Anderson and Orci, 1988; Alper et al., 1989; Blair et al., 1989; Lukacs et al., 1990; Nelson, 1992). Fungi and plants have large, multi-functional acidic vacuoles (Nelson, 1992). In addition, acid-secreting urinary and intestinal epithelia have specialized apical proton pump-rich vesicles that reversibly fuse with the brush-border plasma membrane and secrete acid into the extracellular fluid (Brown et al., 1987, 1988; Schweikl et al., 1989).

Dictyostelium is the eukaryote of earliest evolutionary divergence in which vacuolar proton pumps have been described (Sogin et al., 1986; Padh et al., 1989b; Nolta et al., 1991). As in higher eukaryotes, Dictyostelium lysosomes and phagosomes bear vacuolar proton pumps. The acidification of the Golgi apparatus (see Lacoste et al., 1989) and secretory vacuoles (see Lenhard et al., 1989) in Dictyostelium remains untested.

We have now added two protozoan organelles to the list of endomembrane compartments with cytochemically demonstrable vacuolar proton pumps: the proton pump-rich organelle (acidosome) and the decorated spongiome of the contractile vacuole complex. It has been suggested that acidosomes and contractile vacuoles may be related to one another (Heuser, 1991), and this possibility is being investigated. It can be concluded, in any case, that acidification of organelles of endomembrane origin is a universal and archetypal feature of eukaryotic cell biology.

This work was supported by NIGMS grant R01-GM47282 (to TLS) and NSF grant MCB-9113366 (to HP). We are grateful to Dr David Apps at the University of Edinburgh for providing three antisera to the V-H⁺-ATPase of two higher eukaryotes. ¹²⁵I-Protein A was a generous gift from Dr Geoffrey Greene, University of Chicago. Canine hepatocytes were the generous gift of Dr Howard S. Tager; human foreskin fibroblasts were obtained from Dr Mitchel Villereal; and rat intestinal epithelial cells and mouse macrophages were provided by Dr Eugene Chang, all of the The University of Chicago. We also thank Dr Hewson H. Swift, Michael Shwen and Maya Moody of the EM Core Facility of the Digestive Diseases Center (supported by NIH grant DK 42086), where the electron microscopy was performed. We are grateful to Dr John E. Heuser of Washington University for his stimulating input early in this investigation.

An important study of the contractile vacuole in Dic -tyostelium discoideum is now in press (Heuser, J., Zhu, Q. and Clarke, M. Proton pumps populate the contractile vacuoles of Dictyostelium amoebae. J. Cell Biol.).