A Primary Cation Transport by a V-Type Atpase of Low Specificity

Across the epithelia constituting the Malpighian tubules, the gut, the salivary glands and the epidermal sensilla of insects, there is an electrochemical K⁺ gradient directed from the outside or the lumen of the organs to the haemolymph. The electrochemical potential difference (PD) of K⁺ energizes a range of physiological functions such as excretion, absorption and signal transmission (summarized by Harvey et al. 1983; see also Küppers and Thurm, 1982). At least some of the cells of these epithelia are characterized by an apical membrane that is studded with particles, and it is this region of the membrane that transfers K⁺ from the cytosol to the outside. In vitro, the K⁺ transport, both at the midgut and at epithelial sensilla, was shown to be a cation transport of low specificity, preferring Rb⁺ over K⁺ and other monovalent cations (Zerahn, 1971; Thurm, 1974). The function of Malpighian tubules reflects the low ion specificity in vivo (Maddrell, 1980). There is a fundamental similarity in K⁺ transport in the various organs, as indicated by a similarity in physiological performance, by the presence of characteristic ‘portasomes’ (Harvey, 1980; Harvey et al. 1981) and by the poor ion selectivity of the process. The transport at insect sensilla and that in the gut of Manduca sexta have been shown to possess further features in common: they are insensitive to the F₁F_o-ATPase inhibitors oligomycin and NaN₃ (Wieczorek et al. 1986; J. Küppers, unpublished observation), and antibodies to the M. sexta ATPase label the sites of K⁺ transport at epidermal sensilla of other insect species (Klein, 1992). Cloning and sequencing of the subunits of the transporting enzyme of the midgut and of Malpighian tubules revealed that it belongs to the family of vacuolar-type ATPases (Schweikl et al. 1989; Meagher et al. 1990; Novak et al. 1992; Dow, 1992; Pietrantonio and Gill, 1993; Nelson, 1992; Gogarten and Taiz, 1992).

Cioffi and Harvey (1981) analysed the transepithelial short-circuit current, I_sc, at the midgut of Manduca sexta. The I_sc was carried exclusively by K⁺, and they identified the transporting enzyme as an electrogenic cation-ATPase mediating the primary transport of K⁺.

Wieczorek and associates proposed a concept differing from that deduced from I_sc analysis. They observed changes in the fluorescence of voltage-and pH-sensitive dyes induced by harmaline and amiloride and by other experimental procedures in vesicle preparations of the apical membrane of the goblet cells of M. sexta. From the interpretation of the fluorescence quenching, they inferred that the K⁺ transport of insects is driven by a proton gradient, generated by a V-type H⁺-ATPase, via a distinct, electrophoretic nH⁺/K⁺ antiport (Wieczorek et al. 1989; Wieczorek, 1992).

This paper examines the K⁺ transport at the apical membrane of the accessory cells of mechanoreceptor sensilla of the cockroach Periplaneta americana in an in situ preparation that preserves an unequivocal orientation of the transporting membrane. Both the input and the output compartments of the transport are accessible to microelectrodes. This allows quantitative measurements of the voltage, the intra-and extracellular pH and the impedance of the transporting membrane. The dependence of these variables on energy metabolism and the effect of putative inhibitors of a H⁺/K⁺ exchanger are reported. Results are supplemented by previous measurements of the K⁺ activities in the same system (Thurm and Küppers, 1980). Some of the results have been published in a brief form (Küppers and Thurm, 1994).

This investigation was centred on the apical membrane of the accessory cells of epidermal thread hair sensilla located on the ventral side of the cerci of Periplaneta americana. The accessory cells, the tormogene, trichogene and thecogene cells, represent a specialized part of the functionally monolayered epidermis. They envelop the dendrite of the sensory cell and confine a compartment beneath the cuticle: the receptor lymph space. This cavity is separated from the haemolymph and from the subcuticular space of the surrounding integument (Keil, 1984). The isolation prevents any net flux of solutes and confines the electrical circuit to the epithelial cells of the sensillum.

The apical membrane of the accessory cells constitutes at least 95 % of the resistance measured between the receptor lymph and the haemolymph (J. Küppers, unpublished observation). Poisoning energy metabolism using CN^–or NaN₃ raises this resistance from approximately 20 to 40–60 Mil. The impedance of the apical membrane in the active state and during ATP depletion has been determined using current-clamp (Küppers, 1984). The conductance of this membrane is almost completely governed by an ATP-dependent pathway, as revealed by fitting a minimalist electrical circuit diagram to the vector plots of magnitudes and phase shifts recorded between 0.1 and 100 Hz. The ATP-dependent branch has a conductance of about 50 nS, while the sum of all conductances in parallel, including the intercellular clefts and the unstimulated dendrite, amounts to only approximately 5 nS.

The voltage across the apical membrane of the accessory cells is about 105 mV, outside positive, the K⁺ activity (aK⁺) within the receptor lymph is 140 mmol l^–1 (corresponding to an aqueous KCl solution of 180 mmol l^–1) and aK⁺ in the cytosol is about 105 mmol l^–1 (Thurm and Küppers, 1980). Thus, the electrochemical potential difference for K⁺, Δ η_K, adds up to 110 meV (unit charge times ion-motive force, the energy referred to a single ion) or approximately 11 kJ mol^–1.

The only known function of the transport at the accessory cells of insect sensilla is the generation or enhancement of Δ η_K, which drives the receptor current. Upon adequate mechanical stimulation, the receptor current enters the apex of the sensory dendrite and depolarises the spike-generating basal membrane of the sensory cell on the haemolymph side (Thurm and Küppers, 1980; Küppers and Thurm, 1979). The transport at sensilla is not concerned with excretion or resorption, unlike that of the gut, the rectum, the Malpighian tubules and the salivary glands.

The cerci of adult Periplaneta americana L. were isolated. Their distal end was cut and they were glued to a glass capillary to be perfused by gravity (4 kPa). The flow rate was 150–200 μl min^–1. The standard perfusate used throughout these experiments was a high-K⁺ saline which contained (in mmol l^–1): 140 KCl, 5 KHCO₃, 1 KH₂PO₄, 5 MgCl₂, 5 Tris, 5 EGTA, 30 glucose and 120 sucrose. pH was adjusted to 7.2 and pCa to approximately 9. 5 mmol l^–1 NaN₃, 0.2 mmol l^–1 harmaline or 1 mmol l^–1 amiloride was added to this solution. The high K⁺ concentration of this ‘cytosol substitute’ depolarizes the basolateral membrane to approximately 0 mV, indicating a dominating K⁺ conductance. The values of the physiological parameters observed remained constant in successful preparations for several hours.

A substitute for haemolymph differed from the high-K⁺ saline in the concentrations of the main cations, which were (in mmol l^–1): 125 Na⁺, 20 K⁺ and 5 Ca²⁺ (0 EGTA), according to an analysis of the Periplaneta americana haemolymph by Pichon (1970). This solution, used by Thurm and Küppers (1980), polarizes the basal membrane to approximately -30 mV without a significant effect on the ATP-dependent voltage drop across the apical membrane. The voltage between the receptor lymph and the haemolymph is about +70 mV in that case.

A double-barrelled, bevelled glass electrode (total tip diameter ⩽3 μm, see below) was guided by means of a motor-driven micromanipulator under stereomicroscopic control. The electrode was inserted through the thin cuticular cupola at the base of the hair into the receptor lymph space or was further advanced into a cell. One barrel of the microelectrode recorded the voltage, the other was used for either current injection or determination of pH (Fig. 1). The reference electrode made contact with the perfusate at the open end of the cercus. All three electrodes were filled with the standard perfusate. They were connected via AgCl-coated silver wires to a current-clamp device and two high-impedance amplifiers, respectively. The micromanipulator and electronic equipment were designed and constructed by our collaborators H. Meschede and J. Weil. Results were recorded by means of a two-channel chart recorder (SE 130, ABB Goerz, Wien) with the facility for pen synchronization.

The local transepithelial voltage (TEV) is a sensitive monitor of any artefactual shunt. A TEV of more than 100 mV (in the active state ) ensures that a recording originates from the receptor lymph of an intact, unstimulated sensillum. Until the acute dependence of the low cytoplasmic pH on energy metabolism had been established (see Figs 3 and 7), the reversible polarization of the basal membrane by the haemolymph substitute was used to establish the intracellular position of the electrode. The very large tip of the electrode, which was required to penetrate the cuticle, resulted in a low percentage of successful cell impalements. The intracellular recordings are believed to originate from the outermost accessory cell, the tormogene cell, because this cell is by far the largest of the whole complex.

The resistance between the receptor lymph space and the haemolymph space was determined using a sinusoidal current, I_cl, of 0.1 Hz and an amplitude of ±0.4 nA. The resulting alternating voltage reflects the resistance and is superimposed on the actual transepithelial voltage.

For the pH microelectrodes, we used the liquid ion exchangers ‘Hydrogen Ionophore I – Cocktail B’ and ‘Hydrogen Ionophore II – Cocktail A’ from Fluka. One tip of the double-barrelled electrode was siliconized and the tip was filled with ionophore to a length of 300–500 μm. The electrodes were calibrated using commercial buffers of pH 6.0, 7.0 and 8.0 (Fluka) before and after each experiment. The response time was shorter than the time resolution of the recordings. Only those electrodes that responded with a change of more than 55 mV per pH unit were used. Measurements were discarded if the drift of the electrode during an experiment exceeded 0.1 pH unit h^–1.

We found that Ionophore II (but not Ionophore I) responded with approximately 60 mV per decade to harmaline (Fig. 2). We used this sensitivity of Ionophore II to verify that harmaline actually reached both faces of the apical membrane when added to the perfusate. The concentration of this agent in the receptor lymph was 90 % of that in the perfusate within approximately 5 min.

We found that electrodes containing Ionophore I were not suitable for intracellular recordings: a few minutes after impalement of a cell, the reading started to drift by several hundred millivolts, indicating a pH of 2 or even less. The electrodes slowly recovered in calibration buffer but showed little recovery in the perfusate. We were, therefore, unable to monitor the effect of harmaline on the intracellular pH.

The voltage across the apical membrane approached zero when energy metabolism was inhibited (Fig. 3). The pH of the receptor lymph was 6.8±0.1 (mean ± S.D., N=10) and that of the cytoplasm was 7.0±0.15 (N=13). This difference in pH was statistically significant (P⩽0.01; Student’s t-test), indicating that there is an anionic matrix within the receptor lymph cavity. We observed no significant changes in these values for periods of up to an hour, and so we define this state as the equilibrium state.

When metabolic activity was restored, the receptor lymph cavity became positive with respect to the cell by 105 mV and the pH of the receptor lymph increased by 0.35 to 7.15±0.1 (N=17) (Fig. 3). The increase in this transport-dependent pH is significant (P<0.001). At the same time, the pH of the cytoplasm decreased by 0.70 to 6.3±0.2 (N=18), presumably as a result of the presence of acid metabolic intermediates or products.

In conclusion, the active transport reduced the H⁺ activity of the receptor lymph. The pH difference generated across the apical membrane amounted to 1.0 unit compared with the equilibrium distribution. Thus, the inwardly directed LT_H was approximately 45 meV (4.5 kJ mol^–1) at a voltage of 105 mV. Under these conditions, Δ η_K is 110 meV as previously determined (see above).

The influx of three protons would be required for the outward transport of one potassium ion via an antiport judging from the ratio Δ η_H/Δ η_K only. Inhibiting such an electrophoretic antiport should markedly increase the H⁺ activity of the receptor lymph and the resistance of the apical membrane. In addition, the membrane voltage will increase if the load to the primary transport is reduced (see Discussion). In contrast, inhibiting the ATP-dependent transport reduces the voltage and increases the resistance (Figs 4 and 5, see also Thurm, 1974).

To inhibit a putative ηH⁺/K⁺ exchanger we used the drugs harmaline and amiloride, which have been reported to be effective blockers in vesicle preparations of the goblet cell apical membrane of M. sexta (Wieczorek et al. 1991). We raised the concentration of harmaline to 0.2 mmol l^–1 and that of amiloride to 1 mmol l^–1 (that is twice and 1.5 times, respectively, the concentrations used by Wieczorek and collaborators) until nonspecific effects clearly impeded the primary transport (see Fig. 4).

The effects of harmaline and amiloride on membrane voltage and impedance are compared with those of NaN₃ in Fig. 4A (a representative recording taken from seven experiments) and Fig. 4B (showing a representative recording from four experiments). Depletion of ATP caused the membrane resistance to increase by about 160 % (from 20 to 53 MΩ in Fig. 4A and from 24 to 65 MΩ in Fig. 4B). Both inhibitors produced a slight increase in the resistance (30 % for harmaline and 10 % for amiloride) and a slight decrease in the voltage, which were only partly reversible. This combination suggests some inhibition of the primary transport rather than an effect on an antiport. This inference is corroborated by the finding that harmaline or amiloride had no effect on the membrane resistance in the ATP-depleted state (compare Fig. 5A, which is representative of 13 recordings, and Fig. 5B, which is representative of four recordings, with Fig. 4).

Harmaline and amiloride did not significantly affect the pH of the receptor lymph (Fig. 6). Some of the slight decrease in the pH during the application of harmaline may be due to the reduced voltage and a corresponding increase in H⁺ activity. The transient changes in pH at the beginning and end of amiloride application might reflect intracellular transients (Fig. 7).

Observations of voltage changes and of shifts in the activities of a very restricted number of ion species in response to different experimental conditions cannot yield a convincing, positive identification of the ion specificity of a transport ATPase. However, a quantitative comparison of electrochemical potential differences, supplemented by measurements of resistance, may enable likely, possible and impossible mechanisms to be identified.

The thread hair sensillum of P. americana represents a very simple system in this context. The morphology prevents a significant net flux of solutes, and the low conductance in parallel to the transport provides nearly open-circuit conditions for the transport itself. Simultaneous recordings of voltage and pH or of voltage and resistance ensure that recordings originate from an uninjured sensillum and provide information about the position of the electrode tip, thus enhancing the reliability of the data and of their interpretation.

The conclusions drawn rely on reversible and reproducible ATP-dependent changes (Fig. 8) which were observed at the K⁺ activity at which the transport actually works in vivo. They are inferred from two independent lines of evidence.

During ATP depletion, the following equilibrium state across the apical membrane was approached: the voltage was close to 0 mV, the pH of the receptor lymph was 6.8 and that of the cytosol 7.0. The transport generated a voltage of +105 mV and rendered the receptor lymph alkaline by 0.35 pH units (P<0.001). The pH of the cytosol was 6.3 in this active state, i.e. the cell became acidic by 0.7 pH units. The absolute, and even more the relative, increase of pH in the receptor lymph space by 1 unit was too large to be explained by experimental error.

The finding that the activity of the apical transport system renders the output compartment alkaline corresponds to observations on the midgut epithelium of lepidopteran larvae and on Malpighian tubules of Drosophila melanogaster (Chao et al. 1991; Wessing et al. 1993). The extremely high pH within the midgut lumen (>11) is consistently ascribed to the equilibration of protons with the high voltage across the open-circuited midgut epithelium (Dow, 1992).

To adapt the observations at the midgut to the H⁺-ATPase hypothesis, a concept has been developed by Harvey (1992), Martin (1992) and Martin and Harvey (1994) that joins a primary, electrogenic proton transport with the alkalization of its output side by means of an electrophoretic ηH⁺/K⁺ antiport which is driven by a proton gradient. This model of ‘ion circuit analysis’ has been accepted by others. On the basis of sophisticated assumptions, the concept tries to explain final states by their development. During this step-by-step analysis, the First Law of Thermodynamics is circumvented, because the system eventually arrives at a state that allows the driving energy of the proton potential difference to be ascribed only to the voltage component and the chemical potential difference to be disregarded. As a further explicit outcome, ‘ion circuit analysis’ allows the membrane voltage to exceed the reversal potential of the pump, which is limited by ΔG_ATP/m3z_jF, where ΔG_ATP is the free molar energy of ATP hydrolysis, m is a stoichiometric factor, z_j is the valence of the ion species j and F is the Faraday constant.

The relative maximum entropy production is characterized by the state: L’Vz_jF=Lμ_j because then d(Δ Ψz_jF)=-d(Δμ_j) and d(TΔS*)=0. If either Δ Ψz_jF or Lμ_j reaches the input energy of the pump, ΔS* declines to its minimum (equation 2). It follows from the Second Law of Thermodynamics that no term of Δ η_j will ever exceed the limit given by ΔG_ATP/m, and, thus, invert the sign of the other term. This would require the input of entropy from an additional, independent energy source. This energy cannot be derived from a secondary process that is driven by the reaction under consideration.

Thus, an exclusive proton pump, such as a conventional V-ATPase, will render the output side both more positive and more acidic. This phenomenological analysis presumes nothing but causality between pump, voltage and ion distribution. It does not rely on coupled transporters, on details of the ion fluxes induced or on the physicochemical properties of the surroundings. It applies both to rather simple systems, such as vesicles or the closed receptor lymph cavity which do not allow the net flux of any matter, and to functionally sophisticated epithelia.

Because the active transport reduced the H⁺ activity in the receptor lymph, we infer that the transport processes at the apical membrane are not governed by a primary proton transport.

The transport generated a Δ η_H of 45 meV and a Δ η_K of 110 meV. From the ratio LT_H/LT_K, it follows that a hypothetical nH⁺/K⁺ antiport might transfer K⁺ if the value of n is greater than 2. Inhibiting such an electrophoretic antiport should perceptibly increase the membrane resistance. In this study, there is direct evidence (see Materials and methods) that harmaline, a putative antiport inhibitor, reached the receptor lymph space. From the effects on the primary transport process (Figs 4, 6), we infer that amiloride also reached the apical membrane. However, the pH, the voltage across the apical membrane and the resistance of this membrane did not reflect a specific target for these drugs (see Figs 4, 5, 6). In the ATP-depleted state, the conductance proved to be completely insensitive to both harmaline and amiloride (Fig. 5). This lack of effect might be because these drugs are not appropriate tools with which to inhibit an electrophoretic antiport. The result, however, is consistent with that obtained from the analysis of the electrical impedance, which showed no significant conductance in parallel to the ATP-dependent pathway (Küppers, 1984). We therefore conclude that the apical membrane of cockroach sensilla lacks an electrophoretic proton/cation exchanger.

The fact that the proton distribution did not equilibrate with the transport-dependent membrane voltage may be most simply ascribed to an appreciable H⁺ affinity of the primary transport itself, as suggested by Wieczorek et al. (1986). Since the H⁺ activity is 5–6 orders of magnitude lower than aK⁺, it may be estimated that the proton specificity of the transport exceeds even that for K⁺ by some orders of magnitude, thus leading to acidification of the output side under K⁺-free experimental conditions, as observed by Wieczorek et al. (1991) and Wessing et al. (1993).

The electrochemical potential differences for K⁺ and H⁺ only add up to 155 meV. A lower limit for ΔG_ATP released by the pump may be estimated assuming the stoichiometric factor 2 (estimated by Harvey et al. 1981, for K⁺/ATP). We calculate that ΔG_ATP is greater than 300 meV or 30 kJ mol^–1, which is consistent with general estimates for ΔG_ATP, which may be 40–50 kJ mol^–1 in living animal cells.

In conclusion, both from the ATP-dependent shift in pH and from Δ η_H/Δ η_K and the missing electrophoretic antiporter, we infer that the K⁺ transport at the apical membrane of insect sensilla is not secondary to a proton transport. The transport phenomena observed match the performance of a primary, electrogenic cation pump of the kind that has been reviewed by Harvey et al. (1983).

This genuine concept of a primary K⁺ transport through insect epithelia relies on analyses of the short-circuit current at the midgut epithelium of Hyalophora cecropia (Harvey and Nedergaard, 1964) and the midgut of M. sexta. In the latter case, the I_sc constitutes 100 % K⁺ (Cioffi and Harvey, 1981; verified by Wolfersberger and Giangiacomo, 1983). Precise short-circuit conditions, defined by zero electrochemical gradients for the solvent and all solutes across the transporting membrane, are difficult to achieve in experimental situations. But even if short-circuit conditions are only approximated, we do not see how the flux of that ion species, which is driven by a metabolic reaction, can fail to contribute at least a detectable amount to the composition of the current. In these and related experiments, however, the transport of K⁺ always behaved as an indivisible functional entity.

In view of the heuristic value of short-circuit current analyses, and in view of the difficulties that arise in explaining established observations with the H⁺-ATPase/antiport hypothesis (an example is discussed above), it is surprising that the well-founded K⁺ transport concept has been so precipitately dismissed. A critical discussion of the antiport hypothesis has been avoided by the postulate that the hypothesized H⁺ and K⁺ fluxes are so intimately coupled to each other that the transport system as a whole behaves like an ‘electrogenic K⁺ pump’. The physicochemical nature of such tight coupling remains, however, in question. In considering these problems, Dow (1992) approached the outcome of our paper, when he suggested: ‘Perhaps the interaction is so tight that alkali metal ions can actually substitute for protons in the ATPase’.

The common structural and biochemical features outlined in the Introduction, the functional analogies and the labelling by antibodies against the ATPase of M. sexta (Klein, 1992) justify the assumption that the K⁺ pump at epidermal sensilla does not differ in essence from the catalytic element of the K⁺-transporting system in the midgut or Malpighian tubules. In these epithelia, the ATPase has been reliably identified as a V-type ATPase (see references in Introduction). F-ATPases and the related V-ATPases are characterized by their specific molecular structure, showing homologies from the quaternary down to the primary level. Most of them seem to exhibit the extremely high proton specificity that is a prerequisite for the operation of H⁺-translocating ATPases in the presence of abundant Na⁺ and K⁺.

The function ascribed to the V-type ATPase of Enterococcus hirae (Streptococcus faecalis) is essentially different from the electrogenic H⁺ transport of most V-ATPases. It has been suggested that it drives an electroneutral exchange of Na⁺ for K⁺ (Kakinuma and Igarashi, 1989, 1994). The F₁F_o-ATPase of the anaerobic bacterium Propionigenium modestum allows Na⁺ to substitute for protons (Laubinger and Dimroth, 1989; Dimroth, 1994), and it has been shown that Na⁺ and H₃O⁺ share the same binding site in the F_o moiety (Kluge and Dimroth, 1993). Acetobacterium woodii yields a further example of a Na⁺-translocating F-ATPase (Reidlinger and Müller, 1994; Reidlinger et al. 1994). Site-directed mutagenesis of subunit c changed the binding specificity of F_o from Escherichia coli H⁺-ATPase (Zhang and Fillingame, 1995). These examples show that minor modifications may affect the specificity of a proton-translocating ATPase.

From the structural evidence and the physiological evidence gained from the intact midgut epithelium of lepidopteran larvae, from Malpighian tubules and from epithelial sensilla, we infer that the K⁺ transport at insect epithelia relies on a modified vacuolar ATPase. A V-ATPase with a reduced proton specificity, due to some modification at any structural level, will operate just like the electrogenic cation pump at insect epithelia.

We are grateful to Professor Dr Ulrich Thurm and Dr Dorothea-Ch. Neugebauer for critically reading the manuscript and for stimulating discussions. We also thank Dr Elizabeth Howes for very helpful editorial work on the manuscript.