Sorting of ion transport proteins in polarized cells

As described in many of the contributions to this volume, the plasma membranes of polarized epithelial cells are divided into structurally and biochemically distinct domains (Simons and Fuller, 1985; Caplan and Matlin, 1989; Rodriguez-Boulan and Nelson, 1989). These domains subserve different purposes and are consequently characterized by unique protein compositions. In order to generate and maintain the anisotropic protein distributions typical of the polarized state, cells must be able distinguish among newly synthesized membrane polypeptides and to concentrate them within the appropriate plasmalemmal subdivisions. The capacity of cells to segregate and target membrane proteins implies the existence of sorting signals. A sorting signal can be considered to encompass any information embedded within the primary, secondary or tertiary structure of a given protein which somehow specifies its appropriate localization (Caplan and Matlin, 1989).

Our efforts to examine the nature of sorting signals have taken advantage of the existence of the E1-E2 family of ion transporting ATPases, whose individual members are restricted to distinct subcellular compartments despite their close structural and functional inter-relationships (Pedersen and Carafoli, 1987). By creating chimeras from these highly homologous proteins, we are able to limit our search for sorting information to a relatively few regions of sequence dissimilarity. This approach has allowed us to begin to identify sequence domains which appear to play important roles in pump targetting (Gottardi and Caplan, 1993a). Furthermore, this system allows us to demonstrate quantitatively that our chimeric constructs assume folding patterns compatible with these ion pumps’ highly conformation-dependent enzymatic activities (Blostein et al., 1993).

The E₁-E₂ class of ion pumps includes the Na⁺,K⁺-ATPase, the H⁺,K⁺-ATPase, the ER and cell surface calcium ATPases and the H-ATPase of the yeast plasmalemma. Each of these transport proteins shares a distinctive reaction mechanism characterized by a liganddependent progression through the several conformational states which are referred to in the designation ‘E1-E2’ (Pedersen and Carafoli, 1987). In the course of their catalytic cycles each of these pumps becomes transiently phosphorylated on an aspartate residue and each is sensitivite to inhibition by vanadate. Within this family, the Na⁺,K⁺-ATPase and the H⁺,K⁺-ATPase probably constitute the most closely related pair (De Pont et al., 1988). Both of these pumps are composed of ∼100 kDa a-subunits and ∼55 kDa p-subunits (Jorgensen, 1982; Saccomani et al., 1977; Okamoto et al., 1990). The α-subunits are non-glycosy-lated, span the membrane several times and appear to carry all of the determinants involved in enzymatic catalysis. The p-subunits are heavily glycosylated, span the membrane once and appear to play an important role in post-synthetic maturation (Caplan, 1990).

The genes encoding both subunits of the H⁺.K⁺- and Na⁺,K⁺-ATPases have been cloned, revealing -65% amino acid sequence identity for the a-subunits (Shull and Lin-grel, 1986) and ∼40% identity for the P-subunits (Reuben et al., 1990; Shull, 1990). Regions of these molecules known to be involved in activities common to all of the E₁-E₂ ATPases manifest even more dramatic similarity. The sequence of the putative high-affinity ATP binding site, for example, is identical among the two pumps. Several other long stretches are similarly conserved. Hydropathy plots suggest that these proteins’ resemblance extends to their tertiary structure as well. The Na⁺,K⁺- and H⁺,K⁺-ATPases share the same number and distribution of presumed membrane spanning domains, and intervening hydrophilic segments are also closely related in length and topology (Shull and Lingrel, 1986). Thus, within the limits of resolution imposed by these rather crude techniques for predicting structure, these proteins appear to be practically superimposable.

Given this level of compositional and structural homology, it is extremely interesting to note that the Na⁺,K⁺-ATPase and the H⁺,K⁺-ATPase are concentrated in distinct subcellular locations. The Na⁺,K⁺-ATPase occupies the basolateral plasmalemma of almost every polarized epithelial cell (Caplan, 1990). In contrast, the H⁺,K⁺-ATPase resides in both the apical membrane and a pre-apical storage compartment of gastric parietal cells (Hirst and Forte, 1985; Smolka et al., 1983; Urushidani and Forte, 1987; Smolka and Weinstein, 1986) (see Fig. 1). Functionally, these two pumps are characterized by different stoichiometries, affinities for different cations and susceptibilities to different inhibitors. It would appear, therefore, that the structures of these two closely related ion pumps encode distinct sorting signals and define individual functional properties. We have prepared molecular chimeras derived from these pumps’ subunit polypeptides. By expressing these constructs in polarized epithelial cells we have begun to identify the sequence domains which determine each pump’s cell biological and functional properties.

In order to study the sorting of pump chimeras in polarized cells we first needed to determine the rules which govern subunit assembly and cell surface delivery. We chose to study this problem by examining the behavior of Na⁺,K⁺-ATPase, H⁺,K⁺-ATPase and chimeric a-subunits transiently expressed by transfection in COS cells. By determining which pairs of subunits were able to attain a cell surface distribution we were able to draw conclusions about the sequence domains which specify a- and P-subunit interactions (Gottardi and Caplan, 1993b).

The cDNAs encoding the rat H⁺,K⁺-ATPase α-(kindly provided by G. Shull, University of Cincinnati) and the rabbit P-subunit (kindly provided by M. Reuben and G. Sachs, UCLA) have been subcloned into the pCB6 mammalian expression vector (kindly provided by M. Roth, University of Texas at Dallas). We have transiently expressed these cDNAs in COS cells and documented their expression by western blot. Examination of the COS cells by immunofluorescence revealed that transfected cells expressed extremely high levels of the exogenous proteins. Cells transfected with H⁺,K⁺ a alone exhibited a cytoplasmic labeling pattern consistent with an ER localization. When co-expressed in concert with the H⁺,K⁺ p, a fraction of the H⁺,K⁺ a-subunit was apparently able to depart from the ER, as suggested by a superposition of cell surface staining on top of the predominant ER pattern. Surface staining was also observed when these cells were stained with the H⁺,K⁺ P antibody. A similar pattern is generated when influenza HA protein expressed in COS cells is immunolabeled in non-permeabilized cells, strongly suggesting that these localizations do indeed represent surface staining. It is interesting to note that a punctate pattern, consistent with an intracellular vesicular compartment, was also found in cells labeled for H⁺,K⁺ fk No H⁺,K⁺ or Na⁺,K⁺ α-subunit labeling was ever seen to co-localize with H⁺,K⁺ β in these structures. To our surprise, an almost identical distribution was observed in COS cells transfected with H⁺,K⁺ p alone. Both surface and intracellular vesicle labeling were detected. These studies demonstrate that the H⁺,K⁺ α requires assembly with the H⁺,K⁺ p in order to to be delivered to the cell surface. In contrast, the H⁺,K⁺ β can apparently exit the ER without benefit of H⁺,K⁺ α assembly and accumulates at the plasmalemma as well as in an intracellular compartment which contains no Na⁺,K⁺ α-subunit.

Two H⁺,K⁺/Na⁺,K⁺ α-subunit chimeras have been generated. The construct referred to as H519N encodes the N-terminal half of the H⁺,K⁺-ATPase coupled to the C-ter-minal half of the Na⁺,K⁺-ATPase α-subunits. N519H constitutes the complimentary chimera. Both constructs have as their point of ‘in register’ fusion a Narl site, which resides in the sequence encoding amino acid 519, a component of the FITC-binding regions of both pumps (Shull and Lingrel, 1986; Shull et al., 1986). A cDNA encoding the rat Na⁺,K⁺-ATPase a-subunit was kindly provided by E. Benz (Yale University). All chimera construction was carried out in the BlueScript plasmid and products were analyzed by restriction mapping and sequencing. For the purposes of expression studies both constructs were subcloned into pCB6.

Both constructs were expressed by transient transfection in COS cells. Immunofluorescence analysis revealed that a fraction of the expressed H519N polypeptide was able to reach the cell surface. This plasma membrane localization was achieved in the absence of any exogenous H⁺,K⁺ β-subunit. The situation observed for N519H was somewhat different. When this protein was expressed alone only ER labelling could be detected. Co-transfection with the H⁺,K⁺ P-subunit, however, enabled a fraction of the N519H protein to attain a cell surface distribution. The H⁺,K⁺ β in these cells displayed the previously described surface and vesicular pattern. No N519H was detected in association with the vesicles. We interpret these results to indicate that N519H assembles with the H⁺,K⁺ β-subunit and requires this interaction in order to depart from the ER. We further speculate that H519N assembles with the endogenous Na⁺,K⁺ p expressed by the COS cells, and thus does not require the addition of an exogenous p protein in order to achieve a surface localization.

Our results (summarized in Fig. 2) strongly suggest that p-subunits assemble with the C-terminal portions of the H⁺,K⁺- and Na⁺,K⁺-ATPase a-subunits. These findings are entirely consistent with the observations of Luckie et al. (1992) and Lemas et al. (1992), who studied chimeras generated from portions of the Na⁺,K⁺- and Ca²⁺-ATPase α-subunits. These investigators found that a-subunits composed of the C-terminal third of the sodium pump coupled to the N-terminal portions of the Ca²⁺-ATPase assembled with the Na⁺,K⁺-ATPase β-subunit and were translocated to the cell surface. The complementary constructs showed no affinity for the p polypeptide and were retained in the ER.

The observations presented above also demonstrate that the process of α/ β-subunit assembly is at least somewhat selective and that the molecular basis for this selectivity resides in the C-terminal portion of the α-polypeptide. The H⁺,K⁺-ATPase α-subunit and the N519H chimera are unable to complex with Na⁺,K⁺-ATPase β-subunit. In contrast, the H519N chimera, whose C terminus is derived from the sodium pump, readily participates in heterodimers with the Na⁺,K⁺-ATPase p-subunit. In the context of this apparent capacity for discrimination, it is interesting to note that several investigators have found that the H⁺,K⁺-ATPase β-subunit can form functional complexes with the Na⁺,K⁺-ATPase α subunit. These experiments were performed in yeast (Eakle et al., 1992) and Xenopus oocyte (Horisberger et al., 1991; Noguchi et al., 1992) expression systems, which lack endogenous production of Na⁺,K⁺-ATPase β-subunit. As will be discussed below, we have never been able to detect assembly between endogenous Na⁺,K⁺-ATPase a-subunit and H⁺,K⁺-ATPase β-subunit under conditions in which the Na⁺,K⁺-ATPase β -protein is also being synthesized. It would appear, therefore, that the C-terminal portion of the H⁺,K⁺-ATPase α-subunit manifests a high degree of selectivity in choosing a p-subunit partner. The sodium pump’s a-polypeptide may be somewhat less selective, although our results suggest that it prefers its own p-subunit to that of the H⁺,K⁺ pump.

Finally, it should be noted that studies performed in collaboration with the laboratory of Dr R. Blostein (McGill University) demonstrate that the H519N chimera is active as a transport protein (Blostein et al., 1993). Measurements of ⁸⁶Rb uptake reveal that this protein mediates an ion flux which is sensitive to inhibition by both ouabain and SCH 28080, inhibitors of the Na⁺,K⁺-ATPase and H⁺,K⁺-ATPase, respectively. It has yet to be determined whether the counter ion for Rb movement is Na or a proton, nor is it known whether the flux is electrogenic or electroneutral. Experiments designed to answer these questions are underway. It is clear, however, that H519N is able to assume a tertiary structure compatible with the extremely conformation-dependent transport mechanism of an E1-E2 ATPase. We feel confident, therefore, that this protein’s folding has not been adversly affected by chimera construction and that its sub-domain structure is intact.

We wished to examine the nature of the sorting information responsible for localizing the H⁺,K⁺-ATPase to the parietal cell apical membrane. For these purposes we stably transfected the polarized LLC-PK1 epithelial cell line with cDNAs encoding the H⁺,K⁺-ATPase subunits. The presence of H⁺,K⁺ subunit polypeptides was confirmed by western blotting. Indirect immunofluorescence performed on the H⁺,K⁺-ATPase-expressing cell line reveals that coexpression with its H⁺,K⁺ β-subunit is required in order for the H⁺,K⁺-ATPase a to reach the cell surface. Confocal analysis confirms that both the H⁺,K⁺ α- and β-subunits immunolocalize predominantly to the apical brush border. Neither a-nor P-subunits of the endogenous Na⁺,K⁺-ATPase appear to be mis-sorted in these cells. This observation strongly suggests that the pump subunits exhibit strict fidelity with respect to assembly when expressed in this system. The observed spatial segregation of H⁺,K⁺ from Na⁺,K⁺ subunit polypeptides argues against the formation of hybrid pump dimers. Clearly, this cell line is able to distinguish the H⁺,K⁺ α/ β complex from the Na⁺,K⁺ a/p complex and to target each to its appropriate membrane domain.

A stable cell line expressing only the H⁺,K⁺ P-subunit was generated in order to determine which of the two subunits encoded the apical sorting information. A western blot of this cell line shows that both mature and immature forms of the protein are present. Immunofluorescence analysis reveals that the H⁺,K⁺ β-subunit is localized to the apical brush border as well as to a population of large subapical vesicles. The endogenous Na⁺,K⁺ α-subunit does not appear to escort H⁺,K⁺ β-to the cell surface, as it is found only in its normal basolateral distribution.

The apical localization of the H⁺,K⁺ β-subunit would suggest that it is in fact the p that encodes the sorting information responsible for the polarized cell surface distribution of the H⁺,K⁺-ATPase α/β dimer. In order to test this hypothesis, we have prepared a stably transfected LLC-PKI cell line which expresses a high level of the H519N chimera. Western blot analysis reveals that this chimeric protein has the same mobility as the full-length H⁺,K⁺-ATPase α -subunit and is not detected with an antibody against the carboxy-terminal half of the H⁺,K⁺-ATPase.

We next used indirect immunofluorescence and confocal microscopy in order to determine the cell surface distribution of this chimera. We find that the H519N chimera is almost exclusively localized to the apical membrane. The endogenous Na⁺,K⁺-ATPase a-subunit maintains its steady-state basolateral distribution, demonstrating that expression of this chimera does not alter the sorting of the endogenous ATPase. Interestingly, the Na⁺,K⁺-ATPase β -subunit co-localizes with the chimera at the apical brush border as well as with the Na⁺,K⁺ α -subunit at the basolateral surface. Quantitative data on the distributions of the H⁺,K⁺-ATPase and H519N are presented in Table 1.

Both the chimera and the Na⁺,K⁺ P are actually components of the apical membrane and are exposed at the apical surface. This fact is demonstrated by incubating the monolayer with the Na⁺,K⁺ P antibody from the apical side before fixation and permeabilization, which results in strong apical staining that is seen only in our chimera transfected cells and not in the control cells. Taken together, these results suggest that cell surface expression of this chimeric a-subunit requires assembly with the Na⁺,K⁺-ATPase β -subunit. They further demonstrate that the Na⁺,K⁺ P-subunit does not encode dominant sorting information. It would appear that the N-terminal half of the H⁺,K⁺-ATPase must encode an apical signal which is able to function when expressed as part of a chimeric construct. In order to fully test this hypothesis, it will be necessary to evaluate the sorting of the complementary N519H chimera expressed in concert with H⁺,K⁺ p. We have recently succeeded in generating several LLC-PKi cells lines which express this combination of subunits and are in the process of establishing these proteins’ subcellular distributions.

In light of the preceding evidence that the β -subunit does not encode sorting information, it is surprising that the H⁺,K⁺ p expressed alone is sorted predominantly to the apical membrane. It is also interesting that in addition to its cell surface localization, the H⁺,K⁺ β was detected in a population of subapical vesicles. This distribution has never been observed for the Na⁺,K⁺ P-or a-subunit. In order to better understand the unique behavior of the H⁺,K⁺ β it is necessary to establish the identity of this vesicular compartment. We performed electron microscopy in which the localization of the H⁺,K⁺ β was revealed by the immunoperoxidase technique (see Fig. 3). Dense reaction product can be detected along the microvilli, within the endoplasmic reticulum and within coated pits and compartments that resemble tubular endosomes.

To further characterize these structures, we wished to determine whether this compartment could be loaded with a fluid-phase endocytosis marker such as horseradish peroxidase (HRP) (Stoorvogel et al., 1991). Stably transfected cells expressing solely the H⁺,K⁺ β -subunit were incubated in the presence of both HRP and the H⁺,K⁺ β antibody at 4°C to prebind the H+,K⁺ p-subunits at the cell surface. The cells were warmed to 37°C for various lengths of time and subsequently fixed and permeabilized for indirect immunofluorescence. Our results show that even after only 10 minutes both HRP and H⁺,K⁺ P colocalize within the same compartment. They continue to colocalize at 20 minutes and as long as 45 minutes after warming the cells. The structure they occupy is not labeled by antibodies to LGP 120, a lysosomal marker. It seems likely, therefore, that the vesicles seen in H⁺,K⁺ P-expressing cells are derived from the cell surface and represent an endosomal compartment.

These results also raise the interesting possibility that the H⁺,K⁺ β -subunit encodes an endocytosis or coated pit localization signal. Analysis of the H⁺,K⁺ β -subunit’s amino acid sequence reveals that its cytoplasmic tail contains a four amino acid motif, YXRF, which has been shown to function as the transferrin receptor’s coated pit localization sequence (Collawn et al., 1990; Girones et al., 1991). This motif is not present in the sequence of the Na⁺,K⁺-ATPase P-polypeptide (Gottardi and Caplan, 1993a). In attempting to suggest possible roles for this signal, it is useful to recall that in its native gastric parietal cell the H⁺,K⁺-ATPase α / β complex resides within a pre-apical compartment which rapidly fuses with the apical membrane in response to physiological stimulation. When the stimulatory signal is removed the H⁺,K⁺-ATPase is rapidly endocytosed and returned to its pre-apical compartment. In light of the endosomal localization of the H⁺,K⁺ β in transfected cells, it is tempting to speculate that it is the H⁺,K⁺ β -subunit that encodes the signal for the rapid retrievel of H⁺,K⁺ α / β enzyme from the apical membrane.

Finally, it is interesting to recall that coated pit localization signals have been shown to be sufficient to ensure basolateral targeting of membrane proteins in MDCK cells (Brewer and Roth, 1991; Hunziker et al., 1991; Hunziker and Mellman, 1989; Le Bivic et al., 1991). It is perhaps surprising, therefore, that the H⁺,K⁺ β, which appears to manifest such a signal, behaves as an apical protein in LLC-PKi cells. In our attempts to understand this behavior, it has been interesting to consider functional differences between renal cells derived from different portions of the nephron. The LLC-PK, cell line was derived from the cortex of the pig kidney and manifests the morphological and physiological properties of proximal tubule cells (Pfaller et al., 1990). The apical membrane of the proximal tubule participates in the recapture and resorption of peptides and proteins which have eluded the glomerular basement membrane and passed into the glomerular filtrate. Electron microscopic examination of this tubule segment reveals that it is well equiped for this task. Coated pits and coated vesicles are associated with the bases of microvilli throughout the proximal brush border (Rodman et al., 1984). In contrast, little or no receptor-mediated endocytosis has been shown to occur from the apical surfaces of the sort of distal tubule cells which gave rise to the MDCK cell line. This comparison has led us to speculate that coated pit localization sequences do not specify sorting to the basolateral surface per se but, rather, cause the proteins bearing them to accumulate in a given cell’s most endocytically active domain. According to this model, the relevant sorting signal would mediate targetting to a functionally defined instead of a topographically defined destination. Extrapolating from this model, one would predict that the H⁺,K⁺-ATPase p-subunit expressed in MDCK cells would behave as a basolateral polypeptide. Preliminary experiments suggest that this is, in fact, the case. Future studies with this and other proteins will hopefully determine how endo-cytosis signals are interpreted by the sorting machinery of diverse epithelial cell types.

Recently, several groups have begun to extend the insights gained through the study of sorting in epithelia to nonepithelial cell types. Elegant studies on cultured hippocampal neurons have led to the suggestion that neuronal sorting mechanisms may be closely related to those employed by epithelial cells. Infection of these cells with enveloped viruses revealed that influenza HA protein accumulates in axonal processes, while the VSV G protein is restricted to the dendrites (Dotti and Simons, 1990). Further experiments indicated that GPI-linked proteins are predominantly axonal in neurons (Dotti et al., 1991). These observations have prompted the speculation that the same signals and machinery involved in sorting to the basolateral and apical domains of epithelia function in neurons to mediate targetting to dendrites and axons, respectively. In order to examine this hypothesis further, we chose to determine the distribution in cultured hippocampal neurons of the Na⁺,K⁺-ATPase, which, as mentioned above, occupies the basolateral plasmalemmas of most polarized epithelia. The simple equation of basolateral and dendritic membranes would lead to the physiologically unsupportable conclusion that axonal membranes lack Na⁺,K⁺-ATPase. We wondered whether the sodium pump needs of the axon are met by a neuron-specific isoform of the Na⁺,K ⁺ -ATPase, which might be a substrate for apical sorting if expressed in epithelia.

Three isoforms of the Na⁺,K⁺-ATPase a-subunit have been identified (Shull et al., 1986). While the tissue distribution of α1 is fairly ubiquitous, previous studies of α3 expression suggested that it is found predominantly in neurons (McGrail et al., 1991). Using isoform-specific synthetic peptide antibodies we have shown that hippocampal neurons in culture express both the al and a3 subunits of the Na⁺,K⁺-ATPase (Pietrini et al., 1992). When the distributions of these proteins were compared with those of known dendrite and axon markers (MAP-2 and GAP43, respectively) (Caceres et al., 1986; Goslin et al., 1990), we found that both sodium pump isoforms are found in both types of neuronal process. There would appear to be no polarity present in the distribution of either protein. Since the α3 isoform is not endogenously produced in epithelia, we further wondered whether it would manifest a polarized distribution when expressed by transfection in LLC-PKi cells. We found that the distribution of the a3 protein in epithelial cells mirrored precisely that of al. Thus, while both isoforms are restricted to the basolateral surfaces of at least one polarized epithelial cell line, these two proteins are present in all of the subdomains of the neuronal plasmalemma. In light of these observations, it is interesting to note that the ankyrin-fodrin cytoskeleton underlies only the basolateral surfaces of most epithelial cells, whereas elements of this matrix are present in both the axons and dendrites of neurons (Nelson and Hammerton, 1989; Kordely and Bennett, 1991). The role that the cytoskeleton may play in determining the sodium pump’s distribution in neurons and epithelial cells remains to be established.

We have also carried out experiments designed to test the prediction that axonal proteins would be apically sorted when expressed by transfection in epithelial cells. Our initial studies have made use of the Na,Cl-dependent GABA transport system which populates the presynaptic plasma membranes of GABA-ergic nerve terminals (Radian and Kanner, 1983). This protein mediates the re-uptake of neurotransmitter from the synaptic space and is thus responsible for terminating GABA-ergic transmission. Similar transport systems exist for a number of neurotransmitters, including serotonin, norepinephrine and dopamine.

Immunoelectron microscopic examination of the distribution of the GABA transporter in situ reveal that this protein is limited in its distribution to axonal plasmalemma (Radian et al., 1990). We have localized this polypeptide in polarized hippocampal neurons grown in culture and find a similarly axonal localization. We have expressed a cDNA encoding the GABA transporter (gift from B. Kanner) in MDCK cells. Immunofluorescence, cell surface biotinylation and tracer uptake assays all indicate that this protein is predominantly present in the apical surfaces of the epithelial cells (Pietrini et al., 1993). This finding, in contrast to that presented above for the sodium pump, is in keeping with the hypothesis that the axonal and apical sorting pathways are functionally related. Perhaps of greater interest is the recent finding that MDCK cells endogenously express a close cousin of the GABA transporter. The betaine transport system is involved in the osmoregulation of renal epithelial cells and is ∼50% identical to the GABA transporter at the amino acid level (Yamauchi et al., 1992; Guastella et al., 1990). Tracer flux studies demonstrate that the betaine transporter is restricted to the MDCK cell’s basolateral plasma membrane (Yamauchi et al., 1991). We are in the process of developing chimeras between these transporters in order to identify polypeptide domains which participate in their epithelial as well as their neuronal localizations.

The problem of protein sorting is confronted by a large variety of cell types whose various membranous destinations are somewhat more subtly distinguishable than the dramatically differentiated cell surface domains of epithelial cells and neurons. Even in non-polarized cells, a sorting decision may be required in order to select the route that a secretory or membrane protein follows to the cell surface. Endocrine and exocrine cells, for example, possess both constitutive and regulated pathways to the plasmalemma (Burgess and Kelly, 1987). Proteins with potent biological activities, such as hormones and digestive enzymes, are stored within the cell in secretory granules, whose fusion with the plasmalemma is dependent upon stimulation by secretagogues. Polypeptides which serve non-regulated, house-keeping functions, such as components of the basement membrane, by-pass the storage step and are released from the cell as they are made. There is a large and fascinating of literature describing investigations into the mechanisms and signals which function to target proteins for regulated or constitutive delivery.

The H⁺,K⁺-ATPase is an example of one of a very few membrane proteins whose surface delivery is regulated by secretagogues. As mentioned above, the H⁺,K⁺-ATPase resides in a storage compartment composed of tubulovesic-ular elements in its native gastric parietal cell (Hirst and Forte, 1985; Smolka et al., 1983; Urushidani and Forte, 1987; Smolka and Weinstein, 1986). These vesicles also contain intrinsic factor, an exocrine secretory product involved in cobalamin absorption. Hormonal stimulation by gastrin, acetyl choline or histamine lead to increased gastric acid secretion by inducing the tubulovesicular elements to fuse with the parietal cell apical membrane. Thus, in addition to behaving as an apical plasma membrane protein, the H⁺,K⁺-ATPase can be thought of as a participant in a regulated pathway for exocytosis.

We wondered whether the information which directs the H⁺,K⁺-ATPase to its regulated compartment in intercalated cells would also mediate sorting to the regulated pathway in endocrine and exocrine cells. In order to examine this question, we have transfected RINm5AH-T2-B cells with cDNAs encoding subunits of the H⁺,K⁺ pump and our pump chimeras. This cell line is derived from a rat insulinoma and retains many of the differentiated characteristics of its pancreatic p cell progenitor (Nielsen, 1989). Most relevant among these is the presence of a well-developed regulated secretory pathway characterized by numerous insulin-containing intracellular storage vesicles whose secretion can be mobilized by membrane depolarization. Our preliminary studies suggest that the H⁺,K⁺-ATPase, as well as one of our pump chimeras, becomes concentrated in the insulin storage compartment. These data suggest that the H⁺,K⁺ pump carries the sorting information necessary to ensure its diversion to the regulated delivery pathway when it is heterologously expressed in endocrine cells. It also appears that the requisite information may be associated with the H⁺,K⁺-ATPase P-subunit, since this protein also accumulates in the insulin compartment when expressed individually.

Studies are underway to determine whether information and mechanisms involved in the pump’s sorting to the regulated secretory pathway are related to those which drive its apical localization in epithelia. We are especially interested in the role that the H⁺,K⁺ P-subunit’s putative endo-cytosis signal may play in mediating its diversion into the regulated compartment. If any theme can be detected in our observations on pump and co-transporter sorting, it is that these proteins are endowed with multiple sorting signals whose relative importance and functions appear to depend upon the cell types in which these polypeptides find themselves expressed. According to this formulation, sorting is a plastic phenomenon which can be adapted to the various needs of different tissues or to the various environmental conditions confronted by individual cells. The means by which cell types modify their sorting machinery in order to accommodate their physiological roles remain mysterious and will, in all likelihood, persist as a subject for investigation for quite some time to come.

The authors thank Drs G. Shull, M. Reuben, G. Sachs, J. G. Forte, D. C. Chow, E. Benz, G. Banker, J. Handler and K. Amsler for generously providing reagents, and the members of the Caplan lab for helpful discussions and valuable insights. This work was supported by NIH GM-42136 (M.J.C.) and a fellowship from the David and Lucille Packard Foundation (M.J.C.).