Energization of sodium absorption by the H⁺-ATPase pump in mitochondria-rich cells of frog skin

The frog skin in vivo absorbs sodium and secretes hydrogen ions (Krogh, 1939) in an apparent 1:1 stoichiometery (Garcia-Romeu et al. 1969). The in vitro amphibian skin and urinary bladder mounted in an Ussing chamber can also actively secrete hydrogen ions under appropriate electrochemical gradients (Frazier and Vanatta, 1971; Ludens and Fanestil, 1972; Machen and Erlij, 1975; Ehrenfeld and Garcia-Romeu, 1977; Ramsay, 1982). The outermost living cell layer of frog skin epithelium (the first reactive cell layer) is composed of two cell types, cuboidal granular cells and mitochondria-rich cells (MR). The granular cells along with the deeper spinosum and germinativum cells form a functional sodium transport syncytium (Rick et al. 1978). The Na⁺ absorption function of the epithelium is carried out by these principal (P) cells (Koefoed-Johnsen and Ussing, 1958; Kristensen and Ussing, 1985), whereas active H⁺ secretion is restricted to the mitochondria-rich cells (Harvey and Ehrenfeld, 1988a; Ehrenfeld et al. 1989).

The specialized transport functions of MR cells and principal cells are found in a wide range of high-resistance epithelia. Na⁺ absorption and H⁺ secretion mechanisms are similar in amphibian skin and urinary bladder, gill of freshwater fish and crustaceans and in mammalian colon and renal collecting tubule (Kelly et al., 1980; Schwartz et al. 1982; Madsen and Tisher; 1985; Durham and Nagel; 1986; Koeppen, 1987; Steinmetz, 1988; Harvey and Ehrenfeld, 1988a; Sauer et al. 1990; Larsen, 1991; Kirschner, 1982, 1988; Krippeit-Drews et al. 1989; Avella and Bornancin, 1990; Perrone et al. 1990; Fromter, 1988). The proton secretion function and high carbonic anhydrase content of MR cells in amphibian skin (Ehrenfeld et al. 1985; Rosen and Friedley, 1973; Katz and Gabbay, 1988) indicate a close similarity with the alpha-type intercalated cell of collecting tubule and turtle urinary bladder (Steinmetz, 1974, 1986; Schwartz and Steinmetz, 1977; Dixon and Al-Awqati, 1979; Husted et al. 1981; Steinmetz and Andersen, 1982; Brown et al. 1988a). Proton secretion in frog skin is affected by agents which inhibit different types of H⁺-ATPases (for reviews on pharmacology of H⁺ pumps, see Senior and Wise, 1983; Solioz, 1984; Stone and Xie, 1988; Forgac, 1989; Al-Awqati, 1986). Inhibitors of V-ATPase, such as dicyclohexylcarbodiimide (DCCD) and N-ethylmaleimide (NEM), are effective blockers of J_H, as are diethylstilboestrol (DES) and oligomycin (Ehrenfeld et al. 1985, 1990; Sabolic and Burckhardt, 1986; Harvey and Ehrenfeld, 1988a), which interact with F₁F₀-class ATPases (Fanestil and Park, 1981; Strid et al. 1988). Inhibition of JH by vanadate (which is ineffective against F-ATPases and V-ATPases) may be direct (Arruda et al. 1981 ) or indirect via effects on metabolism or insertion of H⁺ pumps into the plasma membrane. In this paper I will focus mainly on the role of the vacuolar H⁺-ATPase (V-ATPase) pump in energizing Na⁺ absorption across frog skin. I will describe novel uses of ion-sensitive microelectrodes and patch-clamp techniques to study the cellular pathways involved in JH and JNa and the rapid regulation of H⁺ secretion in single MR cells by the exocytotic insertion of membrane containing active proton pumps.

The transport pathways for transepithelial H⁺ secretion and Na⁺ absorption through mitochondria-rich and principal cells are summarized in Fig. 1. Most in vitro studies of Na⁺ absorption in frog skin use high-NaCl Ringer’s solution bathing both sides of the tissue whereas in vivo Na⁺ uptake occurs from more dilute external solutions. For example, frogs can absorb Na⁺ in the absence of a permeant anion from fresh water containing less than 2 mmol l⁻¹ Na⁺. Under these conditions, J_Na and J_H are coupled in an apparent 1:1 stoichiometry and electroneutrality is conserved by equal movements of Na⁺ and H⁺ in opposite directions, but through different cell types (Harvey and Ehrenfeld, 1988a). The action of the V-ATPase pump in energizing Na⁺ uptake across apical membranes of P cells is illustrated by the experiment in Fig. 2. A double-barrelled Na⁺-sensitive microelectrode was used to measure intracellular Na⁺ activity in P cells during electrical and pharmacological modulation of J_Na and J_H.

When JH is inhibited by acetazolamide, the loss of transepithelial proton current produces a depolarized transepithelial potential (serosal side becomes more positively charged) and the apical membrane of principal cells also depolarizes. This results in a decrease in the driving force for Na⁺ uptake and a fall in intracellular Na⁺ activity AN_a (Fig. 2). Short-circuiting the tissue renders J_Na independent of J_H and restores Na⁺ uptake (Fig. 2). The dependence of J_Na on electrical gradients is critical under physiological ion gradients. In P cells exposed to dilute apical saline (2 mmol l⁻¹ Na⁺) under open-circuit conditions, intracellular Na⁺ activity is three times that in the external solution. A favourable electrical gradient exists, however, for passive uptake of Na⁺ even down to an external Na⁺ concentration of 0.1 mmol l⁻¹ (Harvey and Kernan, 1984). Under these conditions, Na⁺ uptake in P cells is energized by a transepithelial electrical gradient generated by rheogenic H⁺ secretion through MR cells. Sodium uptake through P cells can be calculated from the instantaneous rate of change in intracellular Na⁺ activity (A4Na) after closure of apical Na⁺ channels by amiloride (Harvey and Kernan, 1984). The equivalence between and J_Na (Fig. 2) supports the hypothesis that principal cells constitute the main pathway for Na⁺ absorption when J_Na is either coupled or uncoupled to J_H.

In the absence of an impermeant anion, J_Na is rate-limited by charge balance from rheogenic proton secretion. Hormones such as oxytocin or vasopressin, which increase apical Na⁺ permeability in P cells but which do not directly stimulate J_H, fail to activate Na⁺ absorption when J_Na and J_H are coupled (Ehrenfeld et al. 1990). Thus, the rate of acid secretion, or more precisely the acid–base status of the animal, will set an upper limit to Na⁺ absorption.

A transepithelial voltage-clamp produces opposite effects on J_Na and J_H (Fig. 3A). At negative serosal potentials, J_Na is stimulated whereas J_H decreases. Transepithelial potential gradients have been shown to act, via changes in apical membrane potential, on separate electrogenic pathways for Na⁺ uptake in principal cells and H⁺ secretion in mitochondria-rich cells (Ehrenfeld et al. 1985; Harvey and Ehrenfeld, 1986). The effects on Na⁺ uptake of acid-loading the epithelium are dependent on the relationship existing between J_Na and H⁺ secretion (Harvey and Ehrenfeld, 1988b). Under conditions where J_Na and J_H are coupled (open-circuited skins exposed to dilute external saline or impermeant anions), an acid load produces a reversible increase in net Na⁺ absorption (Fig. 3B,C). The basal and acid-stimulated J_Na are completely blocked by micromolar concentrations of amiloride and not affected by ethyl isopropyl amiloride, an inhibitor of electroneutral Na⁺/H⁺ exchange. These findings implicate amiloride-blockable Na⁺ channels as the pathway for the acid-stimulated Na⁺ uptake process. Under short-circuit conditions, when J_Na is independent of J_H, net Na⁺ absorption is inhibited by metabolic or respiratory acid loads (Fig. 3D).

The 1:1 electrical coupling between J_Na and J_H in the absence of a permeant anion would seem to indicate that the underlying transport mechanisms are localized to the same cell. Measurements of intracellular Na⁺ activity with double-barrelled ion-sensitive microelectrodes demonstrate that Na⁺ uptake occurs essentially through principal cells even under conditions of tight coupling between J_Na and J_H. Indirect evidence places the proton pump in MR cells. Correlation between JH and MR cell number and apical surface area (Ehrenfeld et al. 1990), the high carbonic anhydrase activity of these cells (Rosen and Friedley, 1973; Katz, 1986) and the inhibition of J_H by acetazolamide (Ehrenfeld and Garcia-Romeu, 1977) constitute the strongest proof. Adaptation to different environmental salinities and acid–base or hormonal challenges can change the number and morphology of MR cells, which correlate with alterations in J_H (Frazier, 1978; Ilic and Brown, 1980; Katz, 1986; Page and Frazier, 1987; Ehrenfeld et al. 1989), as found for intercalated cells of turtle urinary bladder (Husted et al. 1981; Schwartz et al. 1985; Steinmetz and Stetson, 1987) and in renal collecting duct (Madsen and Tisher, 1984).

I have sought direct evidence for the presence of V-ATPase pumps in MR cells by scanning pH gradients above MR cells. Double-barrelled pH-sensitive microelectrodes placed 0.2 μm above the apical membrane of MR cells detected a fall in extracellular pH (pHe) under conditions when active H⁺ secretion was abruptly stimulated (voltageclamping V_t from −100 mV to +100 mV) (Fig. 4). The voltage-activated extracellular acidification (ΔpHe) was observed solely above the pit of MR cells and never when the pH electrode was above principal cells. ApHe was blocked by the H⁺-ATPase probe (Solioz, 1984) dicyclohexylcarbodiimide (10 μmol l⁻¹) or by the carbonic anhydrase inhibitor ethoxzolamide (100 μmol l⁻¹) added to the apical solution (Fig. 4). The instantaneous rate of change in pHe was used to calculate H⁺ flux in single HR cells; it was 8×10⁻¹⁸equiv s⁻¹. At an imposed V_t of 100 mV, the DCCD-sensitive transepithelial current was 10 μA cm⁻², equivalent to a transepithelial H⁺ secretion rate of 373 equiv h⁻¹ cm⁻², which could be generated by 1.3×10⁵ fully activated MR cells per square centimetre of tissue (normally 1 cm² of frog skin contains approximately 10⁵ MR cells). If the pump number correlates with the density of studs (200000 per cell) associated with rod-shaped particles in MR cells of amphibian skin and urinary bladder (Brown et al. 1987b; Steinmetz and Stetson, 1987), then single-pump current is 3.9×10⁻¹⁷ A.

A Na⁺ transport role for MR cells seems necessary to explain the stimulated JNa during an acid load (in low external salinity, OCC) when Na⁺ uptake via principal cells is reduced by H⁺-induced closure of Na⁺ channels. Investigations into sodium absorption by MR cells have led to conflicting conclusions, although there is a consensus that MR cells and P cells have specialized functions. An amiloride-and ouabain-sensitive swelling of MR cells has been demonstrated by video-image analysis in toad skin (Spring and Ussing, 1986; Larsen et al. 1987), which is consistent with the presence of apical Na⁺ channels and basolateral Na⁺/K⁺-ATPase. Electron microprobe elemental analysis of MR cells in frog skin (Rick et al. 1978; Dorge et al. 1989) failed to show significant changes in cellular Na⁺, K⁺ or Rb⁺ levels in response to diuretics, although recent microprobe data revealed that MR cells, which showed changes in Na⁺ content following ouabain and amiloride treatment, accounted for 50% and 25%, respectively, of the total MR cell population (Rick, 1992). Microelectrode measurements of membrane potential in turtle urinary bladder and renal nephron intercalated cells do not support the hypothesis that these cells could constitute a major pathway for transepithelial Na⁺ uptake (Durham and Nagel, 1986; Koeppen, 1987; Bello-Reuss, 1991), but recent patch-clamp studies provide direct evidence for highly selective Na⁺ channels in apical membranes of frog skin MR cells (Harvey and Larsen, 1992).

Single MR cells maintain their shape and polarity when isolated from amphibian skin by collagenase treatment and membrane currents can be studied using the patch-clamp technique (Larsen and Harvey, 1992). Recordings of single-channel Na⁺-selective currents (/Na) in an excised inside-out MR cell apical membrane of frog skin are shown in Fig. 5. The single-channel current–voltage (I–V) relationship was described by the Goldman-Hodgkin-Katz (GHK) equation for Na⁺ flux. The channel is highly selective for Na⁺ relative to K⁺ and has a single-channel chord conductance of 9pS with 120 mmol l⁻¹ Na⁺ in the patch pipette and 12 mmol l⁻¹ Na⁺ in the bath perfusate. In symmetrical Na⁺ solutions (120 mmol 1⁻¹) the I–V relationship was linear and singlechannel slope conductance was 3.6 pS. The GHK equation fitted to the I–V data yielded a value for the permeability selectivity coefficient, J_Na/P_K, of 35. Using this value, it is possible to determine the Na⁺ concentration of MR cells when the membrane potential is clamped to 0 mV by high external K⁺ (120 mmol l⁻¹). Under these conditions, the /Na reversal potential is equal to the Nemst potential for Na⁺ and the GHK fit to singlechannel Na⁺ currents in intact cells gives ENa of 50 mV and [Na⁺]_i of 16 mmol l⁻¹.

Another approach for detecting Na⁺ channels in MR cells uses the whole-cell patchclamp configuration. In this case, MR cells are patch-clamped in situ from the basolateral side of an intact epithelium mounted in an Ussing chamber. This experimental design has the advantage of maintaining functional polarity of the cells and allowing asymmetrical application of transport inhibitors or activators. Whole-cell voltage-clamp of MR cells in situ is possible since these cells are chemically and electrically uncoupled from neighbouring principal cells (Farquhar and Palade, 1964; Rosen and Friedley, 1973; Rick et al. 1978; Ilic and Brown, 1980; Kristensen and Ussing, 1985; Spring and Ussing, 1986). Amiloride-sensitive apical membrane current-voltage relationships (I_a− V_a) can be recorded from MR cells (in which K⁺ and Cl⁻ currents were eliminated by Ba²⁺, Cs⁺ and ionic substitution) by subtraction of whole-cell I – V relationships recorded in the absence and the presence of 5 pmol I⁻¹ apical amiloride (B. J. Harvey and E. H. Larsen, unpublished results). The almost ideal fit of the GHK equation for Na⁺ flux to I_a−V_a relationships provides firm evidence that the MR cell apical membrane possesses Na⁺ channels. Fluctuation analysis of amiloride-sensitive whole-cell current yields estimates of single-channel conductance and the number of Na⁺ channels in the MR cell apical membrane (Fig. 6). Power density spectra were fitted by a single Lorentzian function which describes whole-cell noise generated by a homogeneous population of fluctuating ion channels. The voltage-dependence of amiloride-sensitive noise shows minima at ENU, as expected for current fluctuations generated by openings and closures of Na⁺ channels. The low conductance and high Na⁺/K⁺ selectivity of Na⁺ channels in MR cells resembles quite closely the biophysical characteristics of patch-clamped Na⁺ channels in principal cells (Palmer and Frindt, 1986). The Na⁺ channel density of 450 channels per MR cell or 45X10⁶ channels per cm² tissue (for 10⁵ MR cells per cm²) is in the range 17×10⁶–77×10⁶ channels per cm² estimated by noise analysis in amphibian skin (van Driessche and Zeiske, 1985). It is theoretically possible for MR cells to account for all of J_Na. However, MR whole-cell Na⁺ currents and channel density are five times smaller than in granular cells (Harvey and Larsen, 1992), which constitute a high-capacity and syncytial route for Na⁺ absorption. The sensitivity of both MR and P cell Na⁺ channels to cytosolic H⁺ (Harvey et al. 1988; B. J. Harvey and E. H. Larsen, unpublished results) and the CO₂ activation of J_H-coupled J_Na implies that MR cell Na⁺ channels are protected against an acid load. This protection could be achieved by the dynamic buffering power of carbonic anhydrase-catalysed hydration of CO2. The high rate of H⁺ secretion through the H⁺ pump and the CCh-induced exocytotic insertion of pumps may also contribute to a rapid regulation of intracellular (pHi) (Cohen and Steinmetz, 1980; van Adelsberg and Al-Awqati, 1986).

The H⁺-ATPase pump of epithelia is thought to be composed of a cytoplasmic domain catalytic subunit and an intramembranous proton channel linked by an antechamber of high buffering power (Andersen et al. 1985; Steinmetz, 1988). This structure is analogous to that envisaged for V-ATPases and F-ATPases (Sebald et al. 1982; Senior and Wise, 1983; Gluck and Caldwell, 1988; Forgac, 1989; Futai et al. 1989). The rheogenic nature of proton secretion and its passage through a proton channel in the apical membrane opens the possibility of applying whole-cell patch-clamp recording to detect H⁺ pump current fluctuations. Using DCCD as a probe for the proton channel (Solioz, 1984), power density spectra were constructed from CO₂-induced noise in single MR cells in situ, in which conductive movements of Na⁺, K⁺ and Cl⁻ were eliminated (Fig. 7). Electrical access to the whole-cell membrane was achieved by perforating the membrane patch in cell-attached mode with the ionophore amphotericin B in the patch pipette. This technique (Hom and Marty, 1988) provides a low electrical resistance access to the cell while, presumably, maintaining metabolism and the cytoskeleton intact. (Hom and Marty used nystatin to perforate rat lacrymal gland cells; lower access resistance is achieved with amphotericin B in MR cells). Current fluctuations induced in MR cells by increasing m the serosal bath could be fitted by a single Lorentzian function, indicative of the presence of a homogeneous population of spontaneously fluctuating channels. The concomitant disappearance of the Lorentzian noise component and transepithelial H⁺ current when DCCD was added to the apical bath strongly suggest that the whole-cell current fluctuations originate from spontaneously fluctuating proton channels in the apical membrane. Changing from CO₂-free to 5% CO₂ medium produced a threefold increase in the calculated number of H⁺ pumps in the membrane accompanied by a 27% increase in single-pump current. The voltage-dependence of the Lorentzian plateau showed minima at membrane potentials of approximately −170 mV, which is in reasonable agreement with measured transepithelial voltages (130–210 mV) required to halt or reverse J_H (Dixon and Al-Awqati, 1979, 1980; Andersen et al. 1985; Ehrenfeld et al. 1985).

Aldosterone stimulates H⁺ secretion in amphibian skin and urinary bladder and in rat colon (Ludens and Fanestil, 1974; Al-Awqati et al. 1976; Perrone et al. 1990), activates H⁺-ATPase activity in renal collecting tubule (Mujáis, 1987; Khadouri et al. 1989), selectively binds to MR cells (Sapirstein and Scott, 1975) and changes their morphology (Voûte et al. 1972). In the presence of 5% CO₂/24 mmol I⁻¹ HCO₃⁻, aldosterone produced an additional stimulation of DCCD-sensitive apical membrane current fluctuations in single MR cells (Fig. 7B) and a 150% increase in the calculated number of H⁺ pumps, without significantly affecting single-pump current. From electron micrographs, the MR cell apical area is approximately 12.5 μm² and the density of proton pumps in CO2 is approximately 600/ttm’². Since MR cells account for approximately 10% of tissue area, this calculation yields a proton current of 12 μAcm^{− 2}, which is equivalent to measured rates of transepithelial H⁺ secretion (350 nequiv h^{− 1} cm^{− 2}).

The single-pump current calculated from noise analysis is about 500 times greater than that estimated from titration and morphological data. This could be explained if H⁺ pumps functioned in clusters, and their distribution over the membrane was non-uniform.

Exocytosis of cytosolic vesicles into plasma membranes can be a rapid means of increasing the transport capacity of a cell (Aimers, 1990; Schaerer et al. 1991). CO₂ stimulates H⁺ secretion in turtle urinary bladder (Schwartz and Steinmetz, 1971) by rapid insertion of vesicles containing preformed V-ATPase holoenzymes into the luminal membrane (Gluck et al. 1982; Stetson and Steinmetz, 1983, 1986; Brown et al. 1987a; Steinmetz, 1988). This phenomenon has also been described as a mechanism controlling acid secretion in renal tubule (Schwartz and Al-Awqati, 1985). The exocytotic event is dependent on changes in cell pH and calcium concentration (Cannon et al. 1985; van Adelsberg and Al-Awqati, 1986; Arruda et al. 1988, 1990), although the role of intracellular acidification per se in CO₂-induced J_H has been convincingly challenged (Adrogué et al. 1987). A dynamic control of acid secretion may be achieved by exocytotic insertion and endocytotic retrieval of V-ATPase pumps (Al-Awqati et al. 1983; Reeves et al. 1983; Schwartz and Al-Awqati, 1986; Dixon et al. 1986, 1988; Stetson, 1989). Exocytosis increases cell membrane area, which can be recorded as an increase in cell membrane electrical capacitance (Clausen and Dixon, 1986; Rich et al. 1990). Measurements of single-cell membrane capacitance (C_m) and exocytotic rates can be reliably made in the whole-cell patch-clamp recording configuration (Hom and Marty, 1988; Lindau and Neher, 1988). The application of this technique to single MR cells in situ is shown in Fig. 8. Superfusion of an MR cell in 5% CCb-buffered Ringer’s solution induces a rapid and reversible increase in membrane capacitance (C_m) from 4.6±10.7 pF to 22±11.5pF with a half-time for maximum change in C_m of 37±16s (N=10). The changes in MR cell membrane area and apical membrane DCCD-sensitive conductance occurred simultaneously, demonstrating that the newly inserted membrane contained functional H⁺ pumps. This result correlates well with an enhanced acid-secreting capacity of the epithelium with DCCD-sensitive transepithelial current increasing from 4.6 μAcm⁻² (172 nequiv h⁻¹ cm⁻²) in CO₂-free medium to 15.7 τ Acm⁻² (584 nequiv h⁻¹ cm⁻²) in 5% CO₂.

A similar approach was used to investigate the mechanism of aldosterone-stimulated H⁺ secretion in frog skin (Fig. 9). Aldosterone produces a slow increase in membrane capacitance and conductance of single MR cells (half-time 12 min). The conductance, but not the capacitance, was reduced by DCCD applied to the apical side, which indicates that aldosterone induces the insertion of new membrane containing functional H⁺ pumps into the apical membrane. DCCD acts as a blocker of proton current without affecting exocytosis/endocytosis. Aldosterone and CO₂-induced exocytosis are additive but mutually independent. How the hormone increases the activity of the membrane shuttle is at present unknown.

Besides their proton secretion function, there is evidence that MR cells are the site of passive apical C1⁻/HCO₃⁻ exchange (Ehrenfeld and Garcia-Romeu, 1978). Cl⁻ absorption/base secretion may not involve the alpha-type proton-secreting MR cell but rather a base-secreting MR cell similar to the beta-type intercalated cell in turtle urinary bladder and renal collecting tubule (Stetson and Steinmetz, 1985; Schwartz et al. 1985; Brown et al. 1988b). A concentration-and voltage-dependent Cl⁻ conductive uptake pathway has also been attributed to amphibian MR cells (Larsen and Rasmussen, 1982; Willumsen and Larsen, 1986; Katz et al. 1985; Katz and Scheffey, 1986; Nagel and van Driessche, 1991). Although this claim is currently a matter of contention (Dörge et al. 1988), recent whole MR cell patch-clamp studies provide direct evidence for voltagegated Cl⁻ currents (Larsen and Harvey, 1992), which may involve a novel gamma-type MR cell (Larsen, 1991). If Cl⁻ channels are inactivated in gamma-type MR cells exposed to low external Cl⁻ concentration, the parallel operation of H⁺-ATPase pumps and Cl⁻/HCO₃⁻ exchange will produce an apparent rheogenic transepithelial Cl⁻ absorption (Larsen et al. 1992). The possibility that proton pumps in MR cells energize Na⁺ and Cl⁻ absorption from dilute solutions in vivo is given added support from the positive correlation between Na⁺ and Cl⁻ absorption rates and MR cell number (Ehrenfeld et al. 1990; Devuyst et al. 1991).

The V-ATPase in mitochondria-rich cells appears to be a major determinant of sodium absorption through principal cells of amphibian skin in vivo or in vitro in the absence of transepithelial anion flux. The apparent 1:1 coupling between JNU and JH results from the equilibrium of circular current flow across the epithelium and is a manifestation of Kirchhoffs current law. An acid load produces closure of apical Na⁺ channels in principal cells and the MR cells provide an alternative cellular pathway for acid-stimulated amiloride-sensitive Na⁺ uptake.

Patch-clamp studies, as described here, of membrane conductance and capacitance of MR cells in situ in a polarized and functional epithelium offer unique opportunities to discover the mechanisms of ion transport regulation modulated by acid–base and hormonal challenges. This tool could also be used to probe further the analogy between proton channels and antidiuretic hormone water channels (Harris et al. 1991; Harvey et al. 1991).