Polarized targeting of V-ATPase in kidney epithelial cells

A membrane-associated vacuolar H⁺-ATPase (V-ATPase) plays an important role in physiologically regulated proton transport in the urinary tubule. In the proximal tubule, all epithelial cells have an apically located V-ATPase that contributes to proton secretion and bicarbonate reabsorption, whereas in the collecting duct, transepithelial proton transport is restricted to a subpopulation of cells known as intercalated cells. These cells, as well as related cells in other transporting epithelia, shuttle V-ATPase between a specialized intracellular vesicular compartment and the plasma membrane in response to changes in the acid-base balance of animals in vivo, as well as in isolated perfused collecting ducts in vitro (Bastani et al. 1991; Brown et al. 1987; Cannon et al. 1985; Dorup, 1986; Gluck et al. 1982; Madsen and Tisher, 1983, 1984; Schwartz and Al-Awqati, 1985; Schwartz et al. 1985; Stetson and Steinmetz, 1983; Van Adelsburg and Al-Awqati, 1986). In this way, both the concentration of V-ATPase in the plasma membrane and net acid or base secretion can be varied, providing a sensitive mechanism by which the kidney can regulate systemic acid-base balance.

V-ATPase molecules can be inserted into either the apical or basolateral plasma membrane of subtypes of intercalated cells (Brown et al. 1988a; Schwartz et al. 1985; Stetson and Steinmetz, 1985; Verlander et al. 1988). Because most membrane proteins have a well-defined apical or basolateral distribution in polarized epithelia (Caplan and Matlin, 1989; Simons, 1987), this property of the V-ATPase makes it an unusual membrane protein. Therefore, in addition to the importance of the membrane shuttling process in physiological regulation, the recycling of V-ATPase molecules to and from opposite membrane domains in intercalated cells provides a unique model for the study of the cell biological mechanisms involved in the insertion and retrieval of membrane proteins.

Intercalated cells are specialized acid-secreting cells in the urinary tubule. The availability of antibodies against V-ATPase isolated from bovine kidney medulla (Gluck and Caldwell, 1987, 1988) has allowed the precise immunolocalization of this molecule in these cells, as well as in other locations along the nephron (Brown et al. 1988a,b). These results provided direct evidence for the presence of different subtypes of intercalated cells that have V-ATPase in opposite plasma membrane domains in the cortical collecting duct (Fig. 1). Affinity-purified, polyclonal antibodies against three subunits (31, 56 and 70×10³M_r) of the purified bovine medullary V-ATPase were used for immunocytochemical studies in rat kidney. Using light microscopy, all intercalated cells were shown to be heavily labeled by antibodies against three pump subunits. In the renal medulla, the labeling was concentrated at the apical pole of intercalated cells (Fig. 1A) and all three ATPase subunits were co-localized to the same membrane domains. Using electron microscopy, immunogold labeling was localized on the cytoplasmic side of all membrane domains that showed a stud-like coating material, which we have identified as the cytoplasmic domain of the V-ATPase (Brown et al. 1987). This finding is consistent with physiological data showing that the medullary collecting duct is involved in hydrogen ion secretion (Atkins and Burg, 1985; Stone et al. 1983a).

When the kidney cortex was examined, a more complex picture emerged, and the findings directly confirmed the existence of at least two subpopulations of intercalated cells in cortical collecting ducts (Fig. 2), as proposed in previous studies (Madsen and Tisher, 1986; Schwartz et al. 1985). The cells, known as A-cells (acid-secreting) and B-cells (bicarbonate-secreting) have opposite polarities of plasma-membrane-associated proton pumps (Brown et al. 1988a,b). Similar subpopulations of carbonic-anhydrase-rich cells are found in the turtle bladder (Stetson et al. 1985; Stetson and Steinmetz, 1985). The A-cells showed predominantly apical labeling, whereas the B-cells had a clearly defined basolateral membrane localization of the proton pump (Fig. 1B). This finding was the first example of adjacent cells in the same epithelium maintaining opposite polarities of a major plasma membrane protein (Brown et al. 1988a). However, in addition, other immunopositive cells had a less well-polarized distribution of proton pumps, with the immunostaining distributed diffusely in the cytoplasm, or even appearing to be concentrated at both poles of the cell, close to the respective plasma membranes.

Immunocytochemical characterization of intercalated cells revealed subtypes that either exhibit or lack immunoreactive band 3 protein, a chloride/bicarbonate anion exchanger (AE1) that is abundant in the erythrocyte plasma membrane, in their basolateral membrane (Drenckhahn et al. 1985; Holthofer et al. 1987; Schuster et al. 1986; Wagner et al. 1987). In the case of the A-type cell, this basolateral anion exchanger works in concert with apical V-ATPase to ensure the extrusion of protons at the apical pole and HCO₃⁻ at the basolateral pole of the cell. Double staining revealed that all of the intercalated cells with basolateral or diffuse/bipolar staining for proton pumps were AE1-negative (Fig. 3). However, the double staining procedure also revealed a small number of AE1-negative cells that had a predominantly apical proton pump localization. These studies demonstrated that cells with opposite polarities of a proton-pumping ATPase co-exist in the cortical collecting duct, as proposed on the basis of earlier studies. Similar results were obtained in the rabbit cortical collecting duct, except that well-polarized basolateral staining for V-ATPase was not observed in the B-cells (Schuster et al. 1991). The AE1-negative B-cells showed diffuse cytoplasmic staining for V-ATPase. Taken together, the immunocytochemical data are consistent with physiological data showing that the cortical collecting duct can secrete either net bicarbonate or net acid under different physiological conditions, but that the medullary collecting duct does not secrete bicarbonate (Koeppen and Helman, 1982; Star et al. 1985; Stone et al. 1983b).

It has been proposed that individual intercalated cells are interconvertible between the A and B configurations and can reverse the polarity of important transporting proteins in response to acid-base loads (Schwartz et al. 1985). This attractive hypothesis has great significance for cell biology, because it implies that the same protein(s) could modify its polarized insertion in an epithelial cell in response to some external stimulus. However, some quantitative studies based on cell morphology alone (Bastani et al. 1991; Dorup, 1986; Madsen and Tisher, 1983, 1984) and immunocytochemical staining with H⁺-ATPase antibodies (Bastani et al. 1991) have shown that although there is, indeed, a marked change in the cellular distribution of cytoplasmic vesicles and membrane domains containing proton pumps during adaptation to acidosis or alkalosis, this result does not necessarily involve an interconversion of A-and B-cells. It is more likely that A-cells can increase the number of proton pumps in their apical plasma membrane during acidosis by exocytosis, and that these pumps are removed from the membrane during alkalosis (Bastani et al. 1991). Moreover, we have recently shown that basolateral staining for AE1 is considerably reduced in cortical A-cells during alkalosis, which would further reduce the activity of A-cells under these conditions (Alper et al. 1991). While these changes are occurring in A-cells, B-cells are also responding to acidosis by removing proton pumps from their basolateral membrane (Bastani et al. 1991); in parallel, acidosis stimulates endocytosis of apical plasma membrane in these B-cells (Satlin and Schwartz, 1989). Indeed a vigorous apical endocytosis in B-cells (identified by double staining to reveal basolateral proton pumps) can be detected even in normal control rats, using fluid-phase markers of endocytosis (D. Brown and 1. Sabolic, unpublished observations). Thus, although A and B intercalated cells can both modulate their plasma membrane content of V-ATPase and anion exchanger, there is still no direct evidence for interconversion of these two cell types in vivo.

If the simple reversal of polarity model of A-and B-cells is correct, then apically located AE1 should be detectable in B-cells. However, efforts by several groups to reveal this exchanger by immunocytochemistry have failed. Nevertheless, physiological data do support the existence of an apical anion exchanger in B-cells (Satlin and Schwartz, 1989), but show that it has certain characteristics that are distinct from the AE1 exchanger in the basolateral membrane of A-cells. Thus, either the apical anion exchanger in B-cells is antigenically distinct from basolateral AE1 of A-cells (Alper et al. 1988), or important epitopes are in some way masked in the apical membrane of intact tissue. In the turtle bladder, the functional characteristics of the apical anion exchanger are also somewhat different from those of the basolateral exchanger (Kohn et al. 1990), implying that it may be a different (but perhaps related) protein.

The intercalated cell membrane domains implicated in proton secretion, as well as intracellular vesicles that shuttle V-ATPase to and from the plasma membrane, have a distinctive stud-like coating material on their cytoplasmic leaflet (Brown et al. 1987; Griffith et al. 1968; Madsen and Tisher, 1986; Stetson et al. 1980). Turning to the toad urinary bladder as a model epithelium, we used the rapid-freeze, deep-etch procedure to examine the structure of this coating material in pieces of membrane removed from intact proton-secreting cells by adherence to poly-L-lysine-coated coverslips (Brown et al. 1987). After rotary-shadowing these membrane fragments with platinum and carbon, a remarkable array of hexogonally packed 10 nm globular subunits was visualized on the underside of the plasma membrane from these cells (Fig. 4). In addition, a similar paracrystalline array of particles was found on the cytoplasmic surface of some intracellular vesicles. Affinity-purified V-ATPase reconstituted into phospholipid vesicles and examined by the same technique exhibited an identical hexagonally packed array of particles, confirming that the structures visualized in proton-secreting cells in vivo were indeed the cytoplasmic domains of the V-ATPase molecule. How the different subunits of the V-ATPase are arranged in these membrane-associated particles is unknown and is the subject of continuing studies.

It has been known for some time that acidification in the turtle bladder is reduced, but not completely eliminated, by microtubule disruption with colchicine (Stetson and Steinmetz, 1983). Studies on other polarized epithelial cells have suggested that microtubules are involved in the transport of membrane proteins to the apical pole of epithelial cells (Achler et al. 1989; Gutmann et al. 1989; Ojakian and Schwimmer, 1988), whereas basolateral targetting of membrane proteins is less, if at all, dependent on intact microtubules. Therefore, we postulated that colchicine treatment of intercalated cells might lead to the inhibition of apical proton pump insertion in type A cells, whereas basolateral insertion of V-ATPase in type B cells would be relatively unaffected. In fact, we found that membrane insertion of V-ATPase was equally affected in both A and B intercalated cells, so that, in the cortex of colchicine-treated rats, A-and B-cells could no longer be distinguished by their patterns of V-ATPase staining (Brown et al. 1990). All intercalated cells showed a diffuse cytoplasmic staining of the pump, because the V-ATPase-containing vesicles were scattered throughout the cytoplasm (Fig. 5). The acidosis-induced insertion of V-ATPase into the apical membrane of A-type cells is inhibited by colchicine (D. Brown and I. Sabolic, unpublished data), explaining previous results that this drug inhibits the stimulation of acidification in turtle bladder.

However, two other basolateral membrane proteins in the intercalated cells, AE1 and a facilitated glucose transporter Glut-1 (Thorens et al. 1990), retained a strong basolateral staining pattern (not shown), implying that colchicine was selective for certain proteins. In the medullary collecting duct, the A-cells also showed a diffuse staining pattern for V-ATPase - no cells showed basolateral staining after colchicine treatment, despite the proximity of many positive vesicles close to the basolateral plasma membrane of these cells.

A similar effect of colchicine treatment on the distribution of V-ATPase was seen in the proximal tubule, where the normally tight apical band of staining was replaced by a scattered distribution of positive vesicles throughout the cell cytoplasm (Brown et al. 1990). Because apical and basolateral membrane vesicles can be readily prepared from kidney cortex (Sabolic and Burckhardt, 1983, 1990), we examined the distribution of N- ethylmaleimide (NEM)-sensitive ATPase activity (a biochemical assay for V-ATPase) in these membrane fractions before and after colchicine treatment of rats. As expected, colchicine treatment greatly reduced the apical membrane NEM-sensitive ATPase activity (Fig. 6), confirming the reduction in apical V-ATPase content detected by immunocytochemistry as well as by Western blotting of isolated plasma membrane vesicles (data not shown). However, no increase in basolateral NEM-sensitive ATPase was found after colchicine treatment, showing that basolateral insertion of ‘misdirected’ vesicles did not occur.

Taken together with other studies on apical membrane proteins in the kidney (Gutmann et al. 1989), these results suggest (i) that microtubules can be involved in the targetting of intracellular vesicles to both the apical and basolateral pole of epithelial cells, (ii) that microtubules are not involved in the final fusion event, since ‘misdirected’ apical vesicles do not fuse with the basolateral plasma membrane, and (iii) that the effect of microtubule disruption of membrane protein distribution is protein-specific, rather than membrane-specific. Thus, membrane proteins that turn over or are recycled rapidly may be more susceptible to the effects of microtubule disruption than membrane proteins that have a longer time of residency in the membrane.

The relative distribution of V-ATPase among various intracellular compartments and the plasma membrane varies greatly in different cell types along the urinary tubule. cDNA cloning was used to search for isoforms of pump subunits in the kidney and other tissues that might provide clues concerning the mechanisms responsible for this differential intracellular targetting (Hirsch et al. 1988; Nelson et al. 1992; Wang and Gluck, 1990; Wang et al. 1989). Isoforms of the 56×10³ M_r subunit were detected that shared sequence similarity in the internal region of the molecule, but which showed divergence in the amino-and carboxy-terminal regions. mRNA hybridization for one form (the ‘kidney’ isoform) revealed that it was present only in kidney cortex and medulla, whereas the other isoform was also present in other tissues. Specific antibodies against a C-terminal peptide found only in the ‘kidney’ isoform showed intense localization in intercalated cells of the cortex, together with a weaker staining of principal cells and distal tubule cells. Immunostaining with this antibody was completely absent in the proximal tubule. In contrast, antibodies against shared epitopes from the internal sequence of the 56x10³M_r subunits stained proximal tubule cells in addition to the other cell types (Nelson et al. 1992).

Double staining with the ‘kidney’ isoform-specific antibody and a monoclonal antibody against the ×10³M_r subunit of the V-ATPase shows this remarkable cell specificity of labeling with the two antibodies. The 31×10₃M_r antibody labels both collecting duct epithelial cells and proximal tubule cells, whereas proximal tubules are unstained with the C-terminal anti-56×10Mr antibody (Fig. 7). Further examination of distinct isoforms of V-ATPase subunits may provide important information related to the intracellular targetting and functional distribution of this enzyme in kidney epithelial cells. Indeed, isoforms of the 31×10³Mr subunit appear to be present on brush-border microvilli and intracellular organelles in the proximal tubule (Wang et al. 1989).