ABSTRACT
The midgut of the tobacco hornworm secretes K+ across the apical plasma membrane of its goblet cells. This secondary K+ transport results from K+/H+ antiport energized by the proton-motive force generated by a primary, H+-transporting plasma membrane V-ATPase. Thus, the lepidopteran midgut constitutes a well-established example of the emerging concept that the proton-motive force is an alternative to the classical sodium-motive force for the energization of animal plasma membranes.
K+/H+ antiport in the tobacco hornworm midgut is electrophoretic, exchanging 2H+ for 1K+. Under physiological conditions, it is energized by the voltage component of the proton-motive force. The strong coupling of electrophoretic K+/2H+ antiport with the electrogenic V-ATPase provides, in principle, the minimal device for the alkalization of the midgut lumen to pH values higher than 11.
K+/H+ antiport is insensitive to bafilomycin A1, but is inhibited by amiloride or Concanavalin A. Lectin staining of blots after SDS–PAGE revealed several glycosylated polypeptides in the goblet cell apical membrane which are not part of the V-ATPase and thus are candidates for the antiporter protein. Current efforts are focused on the isolation of the K+/H+ antiporter.
THE K+ PUMP IN THE LEPIDOPTERAN MIDGUT: AN EXCEPTION TO A GENERAL RULE
It is widely accepted that animal plasma membranes are energized by the Na+/K+-ATPase which generates a sodium-motive force driving sodium-dependent secondary transport processes (Harold, 1986). The larval lepidopteran midgut is a well-documented exception to this dogma since it lacks a Na+/K+-ATPase (Jungreis and Vaughan, 1977). Instead, all secondary transport processes across the midgut epithelium, including the absorption of amino acids and the regulation of the high pH in the midgut lumen, appear to be energized by a unique alkali metal ion pump (Harvey et al. 1983a). As K+ is the ion transported in vivo, this pump has been referred to simply as the K+ pump (pump = transport mechanism). It was discovered in the isolated midgut of the Cecropia silkworm 30 years ago (Harvey and Nedergard, 1964), shortly after the discovery of the Na+/K+-ATPase by Skou (1957), and was characterized as Na+-independent active transport of K+. The K+ pump resides in the apical plasma membrane (Wood et al. 1969) of goblet cells (Dow et al. 1984; Moffett and Koch, 1988), one of the two types of differentiated cells in the single-layered epithelium. It transports K+ electrogenically into the gut lumen, thereby creating a transmembrane voltage in excess of 250 mV, lumen positive (Dow and Peacock, 1989).
ACTIVE K+ TRANSPORT IS ENERGIZED BY A PLASMA MEMBRANE V-ATPASE
Active K+ transport and intracellular ATP concentration decrease in parallel during oxygen deprivation (Mandel et al. 1980). Moreover, the cytoplasmic side of the goblet cell apical membrane is studded by F1-like particles (Anderson and Harvey, 1966). Therefore, it was tempting to suggest that, like F-ATPases, the K+ pump is an ion-transporting ATPase, pumping K+ instead of H+ (Harvey et al. 1981). Indeed, partially purified membranes from larval Manduca sexta (Lepidoptera, Sphingidae) midgut contained a K+-modulated ATPase activity (Wolfersberger, 1979; Wolfersberger et al. 1982). Using ingenious methods for the isolation of highly purified midgut plasma membranes (Harvey et al. 1981; Cioffi and Wolfersberger, 1983), studies with purified goblet cell apical membranes proved the presence of K+-stimulated ATPase activity (Wieczorek et al. 1986). Characterization of this ATPase after solubilization and purification demonstrated that it belongs to the family of V-ATPases (Schweikl et al. 1989; Wieczorek et al. 1990, 1991; Wieczorek, 1992). A typical representative of these heteromultimeric proteins, the insect plasma membrane V-ATPase is composed of at least seven different subunits (Wieczorek, 1992). Amino acid sequences deduced from cDNAs encoding four of these subunits show substantial similarities to subunits of V-ATPases from other sources (Dow et al. 1992; Gräf et al. 1992, 1994; Novak et al. 1992).
Since the V-ATPase appears to be the only ATP-hydrolyzing enzyme in the goblet cell apical membrane, it necessarily has to energize active K+ transport. However, the insect plasma membrane V-ATPase is, like all V-ATPases, an electrogenic proton pump (Wieczorek et al. 1989). In contrast, goblet cell apical membranes do not transport H+ but K+.
K+/H+ ANTIPORT IS THE SOURCE OF K+ TRANSPORT
Using fluorescent Acridine Orange as a pH indicator in vesicle studies, the ATP-dependent proton transport was shown to have the same substrate and inhibitor specificity as enzyme-mediated ATP hydrolysis (Schweikl et al. 1989; Wieczorek et al. 1989). Nevertheless, there was a marked difference between the two activities. K+ and other alkali metal ions transported by the K+ pump stimulated ATPase activity (Wieczorek et al. 1986), whereas these same ions in vesicle experiments inhibited the development of a pH gradient (Wieczorek et al. 1989). Moreover, the addition of K+ and other cations to the extravesicular medium after the generation of an ATP-dependent pH gradient caused its dissipation (Wieczorek et al. 1989). Two explanations could account for this seemingly contradictory behaviour. On the one hand, the insect enzyme could be a K+-ATPase that pumps H+ in the absence of K+ or other alkali metal ions, but pumps cations in their presence; the dissipation of a pH gradient after addition of the ions would, in this case, be due to unspecific proton leaks in the vesicle membrane. On the other hand, the insect enzyme could be a proton pump energizing alkali metal ion/proton antiport in the same membrane; the apparent inhibition of the generation of a pH gradient in the presence of alkali metal ions and its dissipation after their addition would, in that case, be due not only to unspecific proton leaks but also to the presence of a K+/H+ antiport as a ‘specific’ proton leak. Two lines of evidence favour the latter hypothesis that the V-ATPase is a proton pump.
First, vesicles derived from the highly purified goblet cell apical membrane exhibit ATP-independent K+/H+ antiport (Wieczorek et al. 1991). Vesicles which had been preloaded with 20 mmol l−1 K+ developed a pH gradient, inside acidic, in a medium containing less than 1 mmol l−1 K+ (Fig. 1A). The addition of extravesicular K+ caused the dissipation of the K+-generated pH gradient in the same way as it dissipated an artificially generated pH gradient (Fig. 1B). These experiments alone could not exclude the possibility that the driving force for the distribution of protons was a cation diffusion potential established by the influx of cations through a cation channel. However, the existence of a cation channel would be inconsistent with the cation effects on the pH gradient generated by the electrogenic proton pump, since the ATP-dependent membrane potential had been shown to decrease by 15% in the presence of extravesicular K+ (Wieczorek et al. 1989).
Second, the insect V-ATPase was shown to be an exclusive proton pump (Wieczorek et al. 1991). Amiloride inhibited K+/H+ antiport with an I50 value of about 0.2 mmol l−1, whereas the proton pump was less sensitive, with an I50 value of about 0.7 mmol l−1. In the presence of 0.7 mmol l−1 amiloride, which completely inhibited K+/H+ antiport but only partially inhibited the proton pump, an ATP-dependent pH gradient developed even when extravesicular K+ was present (Fig. 2). Since the fluorescence quench was similar to that obtained in the presence of 0.7 mmol l−1 amiloride, but without extravesicular K+, ATP-dependent proton transport in the presence of amiloride could be considered not to be dependent on the extravesicular K+ concentration. Therefore, the V-ATPase in the goblet cell apical membrane appears to be a proton pump, working in parallel with a K+/H+ antiporter.
V-ATPASE AND K+/H+ ANTIPORTER ARE DIFFERENT PROTEINS
K+/H+ antiport is not inhibited by bafilomycin A1, in contrast to the V-ATPase in the same membrane (Wieczorek et al. 1991). From this result and from the differential sensitivities of both processes to amiloride, one could already deduce that ATP-dependent H+ transport and K+/H+ antiport are functionally different entities. It was not clear, however, whether both functions were mediated by the multisubunit V-ATPase or whether K+/H+ antiport was carried out by a separate antiporter protein. The latter possibility was supported by the finding that polyclonal antibodies to the V-ATPase holoenzyme, at a concentration of 0.6 mg ml−1 IgG, inhibited ATP-dependent proton transport and ATP hydrolysis completely, but left K+/H+ antiport unaffected (Wieczorek et al. 1991). Even after preincubation of vesicles with up to 6 mg ml−1 IgG for up to 5 h at room temperature, no substantial inhibition of K+/H+ antiport could be observed (A. Lepier and H. Wieczorek, unpublished results). Obviously ATPase and antiporter are two different molecules. Alternatively, failure to inhibit the antiport could be due to minor antigenicity of antiport epitopes, resulting in an amount of antibody too small to affect K+/H+ antiport. However, since all major subunits of the purified V-ATPase seen in protein-stained SDS gel blots could also be detected by immunostaining with the polyclonal antibodies, this possibility may be excluded. We cannot exclude completely, however, the possibility that most epitopes were not available in the vesicle experiments, since most of the antiporter moiety may be embedded in the membrane.
Independent evidence that K+/H+ antiport is not part of the V-ATPase was provided by solubilization of partially purified goblet cell apical membranes with cholate (A. Lepier and H. Wieczorek, unpublished results). After mixing cholate with the membrane suspension and subsequent centrifugation at 100 000 g for 1 h, the pellets were resuspended in a solution free of cholate and washed twice. Compared with control vesicles, cholate-treated vesicles exhibited a more than twofold higher specific fluorescence quench due to ATP-dependent H+ transport (Fig. 3). This result may be explained by the reorientation of the vesicles during cholate treatment (see Simon and Burckhardt, 1990). However, in contrast to untreated vesicles, the addition of K+ to the extravesicular solution no longer dissipated the established pH gradient. Thus, after cholate treatment, K+/H+ antiport evidently disappeared while ATP-dependent proton transport remained intact. SDS–PAGE (SDS–polyacrylamide gel electrophoresis) revealed that cholate-treated membranes contain fewer polypeptides than control membranes, but all polypeptides known to be part of the purified V-ATPase are still present (not shown). Evidently the K+/H+ antiporter was solubilized by cholate and should be found in the supernatant. These results support the hypothesis that the K+/H+ antiporter is not part of the V-ATPase. Conversely, they provide additional evidence that the V-ATPase is exclusively a proton pump.
STOICHIOMETRY OF K+/H+ ANTIPORT MAY EXPLAIN THE STRONG ALKALIZATION OF THE LUMINAL FLUID
Vesicle studies, using oxonol V as an indicator of membrane potential, had revealed that K+/H+ antiport in larval M. sexta midgut is electrogenic, exchanging one K+ for more than one H+ (Wieczorek et al. 1991). To analyze the antiport stoichiometry in highly purified vesicles from goblet cell apical membranes, we applied the static head method (Turner and Moran, 1982) using fluorescent Acridine Orange to estimate K+-dependent vesicle acidification (M. Azuma, W. R. Harvey and H. Wieczorek, in preparation). Vesicle membranes were clamped at zero voltage by 150 mmol l−1 chloride, which activates anion channels that are otherwise closed (Wieczorek et al. 1989). Clamped vesicles, preloaded with 150 mmol l−1 KCl in a pH 6.5 buffer, were acidified by K+/H+ antiport when they were injected into a K+-free solution of pH 7.0. After reaching a steady state of the fluorescence quench, different amounts of K+ were added to the reaction mixtures. The addition of sufficient K+ to account for a stoichiometry of 1K+ exchanged for 2H+ led to a dissipation of the fluorescence quench to a value comparable in size to that when K+-free control vesicles of pH 6.5 were injected into a K+-free solution of pH 7.0; in the latter case, the fluorescence quench was due only to the addition of acidic vesicles, but not to K+/H+ antiport. Thus, dissipation of the fluorescence quench by K+ under K+/2H+ conditions met static head conditions in which the driving forces of K+ and of H+ were balanced. Therefore, we concluded that the transport stoichiometry of potassium/proton antiport in the goblet cell apical membrane is 1K+/2H+.
V-ATPase and K+/2H+ antiport are tightly coupled since extravesicular K+ prevents ATP-dependent vesicle acidification (see above) and since, in transepithelial transport studies, the midgut pumps K+ rather than H+ (Harvey et al. 1983b; Dow, 1984). One important consequence of the tight coupling is that, in principle, V-ATPase and K+/2H+ antiporter provide the minimal device for the strong alkalization of the midgut lumen (Fig. 4); the luminal pH values in lepidopteran larvae are the most alkaline that have been measured in a biological system and can exceed 12 in certain species (Dow, 1984). The stoichiometry of 1K+/2H+ may explain how antiport results in electrophoretic exchange of lumen H+ for cellular K+. Since H+, is replaced by a K+, the midgut lumen must necessarily become alkaline. Thus, an antiport of more than one proton for one strong cation that is electrically coupled to a proton pump effectively reverses the direction of pumping so that protons are reabsorbed and the extracellular milieu becomes alkaline. This mechanism, in principle, can apply to any alkalization.
ATTEMPTS TO ISOLATE THE K+/H+ ANTIPORTER: A ROCKY ROAD
Lectins
Since the antiporter is a plasma membrane protein, it is likely to be glycosylated. Therefore, a number of lectins with different sugar specificities were tested for inhibition of K+/H+ antiport in vesicles obtained from highly purified goblet cell apical membranes (A. Lepier and H. Wieczorek, unpublished results). Vesicles were preincubated with lectin for 30 min at room temperature before recording K+/H+ antiport. Concanavalin A caused a distinct inhibition of K+/H+ antiport (Fig. 5), whereas wheat germ agglutinin, soybean agglutinin and peanut agglutinin had little or no effect (not shown). Except for peanut agglutinin, the lectins caused visible vesicle agglutination. Since soybean agglutinin also caused agglutination but did not inhibit K+/H+ antiport, mere agglutination of vesicles cannot be the reason for reduced antiport activity. Therefore, it seems probable that Concanavalin A binds to the K+/H+ antiporter.
Since cholate appears to have solubilized the K+/H+ antiporter (see above), we compared Concanavalin A staining in Western blots after SDS–PAGE of pellets and supernatants from cholate-treated membranes and of untreated membranes (Fig. 6). Several Concanavalin-A-binding polypeptides were almost completely solubilized by cholate, the most prominent being a band of approximately 95 kDa. Although it is tempting to suggest that this 95 kDa protein is the antiporter, the result should not be overemphasized. The turnover number of the antiporter may be high and thus it may be present in the goblet cell apical membrane only in small amounts. The antiporter could also be one of those weakly stained polypeptides in the supernatant. Current efforts are focused on the clarification of this point.
Antibodies
To obtain information about the antiporter at the DNA level, we raised antibodies to start an approach based on immuno-shotgun screening of a midgut lambda ZAP expression library (Gräf et al. 1994). The polyclonal antibodies were produced against a fraction obtained by solubilization of goblet cell apical membranes with C12E10 followed by density gradient centrifugation (A. Lepier and H. Wieczorek, unpublished results). The 10% sucrose fraction (Schweikl et al. 1989) contained virtually all the proteins from the goblet cell apical membrane except for the V-ATPase, but included those polypeptides which reacted with Concanavalin A. Unfortunately, the purified antibodies failed to inhibit K+/H+ antiport. This failure may have the same explanation as the failure of polyclonal antibodies against the V-ATPase to inhibit antiport (see above).
We also tested an inhibiting monoclonal antibody to the luminal 2Na+/H+ antiporter of the lobster hepatopancreas (De Couet et al. 1993; kindly provided by Dr G. Ahearn) on immunoblots of SDS gels from goblet cell apical membranes. Although this antibody showed cross reactions in a variety of crustacean tissues that exhibited 2Na+/H+ exchange (Kimura et al. 1994), no bands were recognized in the M. sexta preparation (A. Lepier and H. Wieczorek, unpublished results). The antibody was also tested for binding to the native goblet cell apical membranes in a dot-blot experiment. In contrast to the antiserum against the V-ATPase, which was used as a control, no signal could be detected with the monoclonal antibody, even at a concentration that inhibits 2Na+/H+ antiport in hepatopancreatic brush-border membrane vesicles.
A molecular approach by homology to vertebrate Na+/H+ antiporters
A further approach was based on the putative homology of the insect K+/H+ antiporter to the group of electroneutral Na+/H+ exchange proteins. Since no invertebrate amino acid sequence data were available, we took advantage of sequences derived from cDNAs encoding mammalian Na+/H+ antiporters (A. Lepier and H. Wieczorek, unpublished results). All mammalian Na+/H+ antiporters so far sequenced belong to one of four isoforms (NHE-1 to NHE-4) with remarkable similarities of primary structure, but differing with respect to their membrane localization and their sensitivity to amiloride (Tse et al. 1993; Bianchini and Pouyssegur, 1994). A number of degenerate oligonucleotide primers (17–20mers) were selected from suitable amino acid sequence stretches of the ‘housekeeping’ mammalian Na+/H+ antiporter isoform NHE-1 (Tse et al. 1993) for use in a reverse transcription/polymerase chain reaction (RT–PCR) experiment (Fig. 7). These primers were selected for minimal degeneracy and sufficient spanning fragment length. Sequences of amino acid residues that are conserved in all isoforms but would only allow the construction of highly degenerate primers were not considered. In a first experiment, reverse transcription of poly(A)+ mRNA from M. sexta midgut, from Malpighian tubules (which also exhibit K+/H+ antiport; Grinstein and Wieczorek, 1994) and, as a control, from human blood was primed with primer 5 (Fig. 7), which is specific for NHE-1. PCR amplifications using the cDNA as template were performed at low stringency (annealing temperature 40°C). Bands within the expected size range were excised from low-melting-point agarose gels and used for reamplification with one or two nested primers. Finally, fragments of about 240 bp (between primers 3 and 4, Fig. 7) from all three sources were cloned and sequenced completely. The sequences from M. sexta midgut and Malpighian tubules were identical; they turned out to be new sequences and were not related to any antiporters or other DNA sequences in the GenBank, whereas the sequence amplified from human blood was identical to the corresponding section of the known human Na+/H+ antiporter NHE-1 isoform (Sardet et al. 1989). This result suggests that no protein with sufficient homology to the mammalian NHE-1 isoform to be detected by this method is being expressed in these M. sexta tissues. Recently the putative apical localization of NHE-2 and NHE-3 in various mammalian epithelia has been confirmed (Tse et al. 1993; Bianchini and Pouyssegur, 1994). RT–PCR studies on mRNA from M. sexta midgut and Malpighian tubules using primers specific for NHE-2 are currently under way.
CONCLUSION AND PERSPECTIVES
Compelling evidence suggests that a K+/H+ antiporter is the K+-transporting component of the apical electrogenic K+ pump in larval M. sexta midgut. The antiporter is electrophoretic and exchanges 2H+ for 1K+, driven by the voltage component of the proton-motive force that is generated by the proton-pumping V-ATPase situated in the same plasma membrane. The lepidopteran midgut is the first example from animal tissues in which the proton-motive force generated by a primary proton pump has been established to be the energy source of secondary transport across the plasma membrane. The K+ pump of M. sexta midgut belongs to a group of insect alkali metal ion pumps involved in various physiological functions, such as amino acid absorption, water and ion homeostasis and sensory transduction (Harvey et al. 1983a). These cation pumps appear to constitute an evolutionarily conserved family of cation pumps consisting of a proton-pumping V-ATPase and a cation/proton antiporter (Grinstein and Wieczorek, 1994). Thus, insect cation/proton antiporters now are a well-established example of the emerging concept that the proton-motive force is an alternative to the sodium-motive force for the energization of animal plasma membranes.
Whereas K+/H+ antiport is electrophoretic in M. sexta midgut, this property seems not to be a general rule in insects; for example, K+/H+ antiport may be electroneutral in Malpighian tubules (see Grinstein and Wieczorek, 1994). Regardless of its hypothesized involvement in the production of a high luminal pH, the midgut K+/2H+ antiporter faces an alkaline lumen throughout the whole length of the gut (Dow, 1984). Even in the posterior midgut where the pH decreases to values lower than 9 or, more extreme, even at a neutral pH, the main driving force for active K+ transport has to be the voltage. Thus, the electrophoretic properties of the midgut K+/2H+ antiporter may be an evolutionary adaptation to the extracellular pH. Similarly, all prokaryotic Na+/H+ antiporters tested thus far appear to be electrophoretic, exchanging more than one H+ for one Na+ (Padan and Schuldiner, 1993); this stoichiometry has been interpreted to be an evolutionary adaptation to the necessity of acidifying the cytoplasm and of Na+ excretion in an alkaline environment where only the voltage, but not a pH gradient, can drive the exchange. It is tempting to speculate that insect and bacterial cation/proton antiporters may be somewhat homologous and that their similar functions may be based on more than simple convergence due to the same selective pressure.
ACKNOWLEDGEMENTS
This work was supported by German Research Foundation Grant Wi 698, European Economic Community Grant SC1*-CT90-0480, National Institutes of Health Grant AI22444 and the Ministry of Education, Science and Culture of Japan (International Research Fellowship, 1992-Y-146). We thank Ralph Gräf for critically reading the manuscript.