Mechanisms of K⁺ Uptake Across the Basal Membrane of Malpighian Tubules of Formica Polyctena: The Effect of Ions and Inhibitors

Malpighian tubules of forest ants (Formica polyctena) secrete primary urine. As in many insect species, they preferentially transport K⁺ into the lumen in control conditions, Cl⁻and water following passively (see Phillips, 1981, for a review). The fluid secretion rate is strongly dependent on the bath K⁺concentration (Van Kerkhove et al. 1989).

Since K⁺is always transported across the epithelium against its electrochemical gradient, K⁺transport must occur transcellularly. The active step is primarily situated in the apical membrane, where the extrusion of K⁺into the lumen is believed to be realized through the combination of an electrogenic H⁺pump and an electroneutral K⁺/H⁺antiporter acting in parallel (Weltens et al. 1992; Zhang et al. 1994; Dijkstra, 1993; Leyssens et al. 1993a,b). In Malpighian tubules of Formica polyctena, it was found that the bath K⁺concentration determines the intracellular K⁺content available for apical K⁺transport. This may be an intrinsic regulatory mechanism of fluid secretion (Leyssens et al. 1992, 1993a).

K⁺transport systems must exist across the basal membrane to ensure a sufficiently large K⁺uptake to maintain the rate of fluid secretion observed in the different bathing K⁺concentrations and to account for the change in intracellular K⁺concentration in response to the surrounding bath K⁺level. K⁺entry may occur passively through K⁺channels. Ion substitution and K⁺-selective measurements have demonstrated the appreciable K⁺permeability of the basal membrane (Leyssens et al. 1992, 1993a). The hyperpolarization of the basal membrane potential difference (V_bl) observed at all bath K⁺concentrations when Ba²⁺was used to block the K⁺channels (Weltens et al. 1992) suggested the existence of an electrochemical K⁺gradient favourable for K⁺uptake. However, the calculated electrochemical gradient for K⁺across the basal membrane was very small when the bath K⁺concentration was decreased and could even reverse (Leyssens et al. 1993a). The fluid secretion was also drastically, but not completely, blocked after 30 min in the presence of Ba²⁺(Weltens et al. 1992). This suggested that, as in other K⁺-transporting Malpighian tubules (for a review, see Nicolson, 1993), alternative pathways for K⁺uptake may be present in the basal membrane and that the relative rates of these hypothetical K⁺transport systems may vary under different transport conditions.

The present study focuses on these alternative routes for K⁺transport across the basal membrane at different bath K⁺concentrations. The effects of ion substitutions and blockers of primary and secondary active transport systems on fluid secretion, on intracellular and luminal K⁺concentration and on electrical potential differences were investigated.

It is shown that the rise in intracellular K⁺ concentration when the bath [K⁺] is increased still occurs after the basal K⁺ channels have been blocked. The existence of a basal coupled entry mechanism for Cl⁻ and K⁺ is demonstrated and its contribution to K⁺ uptake under different transport conditions is discussed. The role of Na⁺in transcellular ion transport is also investigated. The presence of a primary active K⁺ uptake system is critically evaluated. On the basis of these and previous findings, a transport model for transepithelial ion transport in the Malpighian tubule of Formica polyctena is proposed.

Worker ants of the species Formica polyctena Förster were collected from natural nests at the periphery of woods and kept in an artificial nest at a constant temperature of 20°C. The animals were fed with sugar and water.

After dissection, a single Malpighian tubule (1–2 mm) was transferred to the stage of an inverted microscope into a bathing droplet (50 μl). The bathing droplet was covered with paraffin oil to avoid evaporation (see Van Kerkhove et al. 1989).

The technique used for measuring fluid secretion rate has been described in detail by Van Kerkhove et al. (1989). A fluid secretion experiment consisted of a control (three times 10 min), a test and a wash-out period (3–4 times 10 min each). The bathing fluid could be renewed using a perfusion and suction pipette. When a change of bathing solution was needed, the new solution was added via the perfusion pipette at a rate of 150 μl min⁻¹over a 3 min period. Perfusion was then stopped until the next change in bath solution.

The effects of ions and inhibitors on fluid secretion rate were tested in 143 mmol l⁻¹Cl⁻Ringer’s solutions, containing varying K⁺concentrations (see Artificial salines and Results). As described previously (Weltens et al. 1992), in each experiment, secretion rate was expressed as a relative rate, i.e. as a percentage of the rate during the last control period before the change to that experimental solution. Furthermore, to account for a decrease in fluid secretion rate with time, for each collection period the ratio of the mean value of the experimental series over the mean value of the same period of fluid secretion experiments was calculated, where the tubules were bathed continuously in the control solution.

In order to test the statistical significance of the effect of a drug, the experimental relative rate was compared with the control relative rate at the same moment (unpaired Student’s t-test). The drug effect was considered to be fully reversible when no statistically significant difference (unpaired Student’s t-test) was found between the relative secretion rate during a wash-out period and the corresponding control period. Although we realize that this method was not ideal (the change in fluid secretion with time in the absence of a test solution had to be evaluated on a different set of tubules), it seemed to be the most suitable way to distinguish drug effects from time effects. Corrected results, taking into account the spontaneous decline of the fluid secretion rate with time are shown in the figures and tables (see Results).

The method has been described in detail previously (Leyssens et al. 1992). Measurements were performed with microelectrodes filled with 0.1 mol l⁻¹KCl and connected to a high-impedance electrometer (Duo 773, WPI) via a Ag/AgCl wire. The reference electrode was a low-resistance (1 MΩ) 3 mol l⁻¹KCl electrode earthed via a Ag/AgCl wire. In order to facilitate impalements, the Malpighian tubule was immobilized in the bathing droplet by two holding pipettes. The secreted fluid remained in the tubule. The measurement of the basal membrane potential difference (V_bl) was accepted if a sudden negative potential deflection occurred and was stable for at least a few minutes and if the electrode potential differed maximally by 3 mV from the baseline after electrode withdrawal. The transepithelial potential difference (V_te) was measured by advancing the microelectrode through the cell layer into the lumen of the tubule; V_bl and V_te were expressed with reference to the bath side. The apical potential difference (V_ap) was calculated as the difference between the transepithelial and transbasal potential differences and expressed with reference to the lumen [V_ap=-(V_te-V_bl)].

The bathing droplet was continuously perfused at a constant rate of 150 μl min⁻¹.

The technique used for the construction of double-barrelled ion-sensitive electrodes has been described previously (Leyssens et al. 1993a). The tip of the ion-selective barrel was filled with a short column of the ion-selective liquid K⁺exchanger (Corning 477317), and the rest of the barrel was backfilled with 1 mol l⁻¹KCl. The reference barrel was backfilled with 1 mol l⁻¹sodium acetate + 10 mmol l⁻¹KCl. The filled electrodes were bevelled in a rotating polishing alumina solution. After bevelling, the ion-sensitive and reference barrels had tip resistances ranging from 50 to 100 GΩ and from 0.5 to 1 GΩ, respectively. Three tracings were recorded on a pen recorder (Sefram or Linseis): the potential of the reference barrel, the potential of the ion-selective barrel and the difference between them, representing the K⁺-sensitive signal.

The K⁺-sensitive signal was calibrated in a 143 mmol l⁻¹Cl⁻Ringer’s solution, in which the K⁺concentration was varied from 5 to 113 mmol l⁻¹, using Na⁺as a substitute. Interference by Na⁺can be ignored (see Leyssens et al. 1993a). The measured intracellular ([K⁺]_i) or luminal ([K⁺]_l) K⁺‘signal’ was always expressed as the K⁺concentration that would be needed in a 143 mmol l⁻¹Cl⁻Ringer’s solution to result in the same reading of the K⁺-selective barrel. It should be kept in mind that the value for [K⁺]_i should, in fact, be corrected if the activity coefficient in the cell was different from that in the bath. Once the electrode has impaled the cell, it should, however, be possible to observe changes in [K⁺]_i without too much additional error. Other possible errors arising from tip and liquid junction potentials of the reference electrode have been discussed by Leyssens et al. (1993a). Since the calibration curve has a logarithmic scale, in the high concentration range (>100 mmol l⁻¹), small changes in intracellular and luminal concentrations (i.e. less than 10 mmol l⁻¹) correspond to minor changes in the K⁺-sensitive signal (i.e. 1–2 mV). On evaluating the results, one has to take into account that this approaches the limit of detection.

The ion-sensitive electrodes were accepted for use if a stable potential signal was obtained for both the ion-sensitive and the reference barrel, if the K⁺-sensitivity of the reference barrel was less than 1 mV and if the calibration curve could be fitted by a straight line with a slope of at least 48 mV per decade. Except when the effect of a certain test solution on the fluid secretion rate or on the electrical potentials or concentrations was not clearly reversible, the value of a specific electrophysiological variable, registered before and after ion substitution or drug application, was averaged and taken as the control value. In this way, possible time-dependent changes in potentials or concentrations were minimized.

The different control solutions containing varying K⁺and Na⁺concentrations (see Results) were obtained by mixing a K⁺-containing, Na⁺-free and a Na⁺-containing, K⁺-free standard Ringer’s solution (2 mmol l⁻¹CaCl₂, 13 mmol l⁻¹MgCl₂, 12.1 mmol l⁻¹Hepes, 2.8 mmol l⁻¹alanine, 10.6 mmol l⁻¹trehalose, 11.7 mmol l⁻¹maltose, 139 mmol l⁻¹glucose and 113 mmol l⁻¹KCl or NaCl, respectively; pH was adjusted to 7.20 with KOH or NaOH, respectively; the osmolality of the solution was 375 mosmol kg⁻¹H₂O).

Solutions were freshly prepared each week, filtered through 0.45 μm Millipore filters and kept at 2°C until use.

In some experiments, the following alterations in ion composition were made: 6 mmol l⁻¹BaCl₂ was added to the bath solution; Cl⁻was completely replaced by Br⁻; Na⁺was completely replaced by N-methyl-D-glucamine (NMDG⁺, Sigma). On replacing NaCl by NMDG⁺, the alkaline solution was titrated with an appropriate amount of HCl for pH adjustment; as a result, the final Cl⁻concentration remained virtually unchanged (i.e. 143 mmol l⁻¹).

The following pharmacological substances were tested: dihydro-ouabain (Sigma), sodium orthovanadate (Janssen Chimica), bumetanide (Leo Pharmaceutical Products), Schering 28080 (Schering-Plough; solvent, 0.05% methanol and 0.05% dimethylsulphoxide without any effect on fluid secretion; unpublished results).

Results are presented as mean values ± S.E.M. Statistical computations were made using Statview II (Abacus Concepts Inc., Berkeley, CA, USA, 1987). The fluid secretion measurements and electrophysiological measurements were evaluated using unpaired and paired two-tailed Student’s t-test, respectively, except when indicated differently. A value of P<0.05 was accepted as indicating statistical significance.

Ba²⁺is an effective blocker of the K⁺channels in the basal membrane of the Malpighian tubule cells of Formica polyctena. It probably reduces an inward K⁺current and, because of the activity of the electrogenic apical H⁺pump, both apical and basal membranes hyperpolarize (Weltens et al. 1992). In the present study, we compared the effects of a change in the bath K⁺concentration ([K⁺]_bl) on [K⁺]_i in the absence and presence of 6 mmol l⁻¹Ba²⁺. An example is shown in Fig. 1. In the absence of Ba²⁺, an increase in bath K⁺concentration ([K⁺]_bl) from 5 to 113 mmol l⁻¹resulted in a prompt depolarization of V_bl and a concomitant rise in [K⁺]_i (see also Leyssens et al. 1993a).

When [K⁺]_bl was decreased to 51 mmol l⁻¹[K⁺]_bl, V_bl hyperpolarized but [K⁺]_i remained at the same value. On returning to 5 mmol l⁻¹, V_bl and [K⁺]_i changed: [K⁺]_i decreased but to a slightly higher value than the control and V_bl was also slightly more negative than at the start of the experiment. In the presence of 6 mmol l⁻¹Ba²⁺, the response of V_bl to an increase in [K⁺]_bl was greatly reduced: V_bl depolarized by only 7 mV for an increase in [K⁺]_bl from 5 to 113 mmol l⁻¹. In contrast, the rise in [K⁺]_i on increasing [K⁺]_bl was comparable to the response of [K⁺]_i to a change in [K⁺]_bl in the absence of Ba²⁺.

In the experiment shown, the addition of 6 mmol l⁻¹Ba²⁺itself caused a considerable decrease of [K⁺]_i in both 5 and 113 mmol l⁻¹[K⁺]_bl. This type of behaviour was observed in two out of the four experiments. In the other two, [K⁺]_i increased. On average, in both 5 and 113 mmol l⁻¹K⁺, [K⁺]_i was not altered significantly on adding Ba²⁺to the bath solution (Table 1). Furthermore, the summarized results demonstrate that, although the K⁺-sensitivity of V_bl was lost under Ba²⁺treatment, suggesting a blockage of basal K⁺channels, the rise in [K⁺]_i on increasing [K⁺]_bl was not affected.

In order to detect any Na⁺-dependent transport system (the Na⁺site of the Na⁺/K⁺/2Cl⁻cotransporter, for instance, cannot transport large cations such as NMDG⁺, see Hedge and Palfrey, 1992), Na⁺was completely replaced by NMDG⁺. In 51 mmol l⁻¹[K⁺]_bl, there was no significant effect (Table 2A). In a low-K⁺solution (10 mmol l⁻¹[K⁺]_bl), the substitution of Na⁺by NMDG⁺caused a small but reversible inhibition of fluid secretion (Table 2A).

It is known that the Na⁺/K⁺/2Cl⁻and K⁺/Cl⁻cotransport systems have a different anion-dependence (Palfrey and Greengard, 1981; Ellory and Hall, 1988). When Cl⁻was omitted from the bath solution and replaced by Br⁻, fluid secretion rate was significantly stimulated in 51 mmol l⁻¹[K⁺]_bl but inhibited in 10 mmol l⁻¹[K⁺]_bl containing a high Na⁺concentration (Table 2A).

As a high K⁺concentration is competitive with ouabain (see Baker and Willis, 1970), the effect of dihydro-ouabain (DHO) was tested in a low [K⁺]_bl (5 mmol l⁻¹). No significant effect of a high dose of dihydro-ouabain (10⁻³ mol l⁻¹) was observed during the 30 min application of the drug (Table 2B). However, in this low [K⁺]_bl, fluid secretion rate was very slow before DHO was applied (i.e. 76±19 pl min⁻¹, N=8) and the time course was highly variable. Consequently, it was difficult to distinguish a small drug effect from the time-dependent changes.

The effect of dihydro-ouabain was also tested on V_bl (Fig. 2A): if dihydro-ouabain were to reduce [K⁺]_i, a depolarization of V_bl would be expected. However, in 5 mmol l⁻¹[K⁺]_bl, ouabain treatment (10⁻³ mol l⁻¹) for at least 10 min did not depolarize V_bl. If the putative pump were electrogenic, its effect on V_bl might be masked by the high K⁺permeability of the basal membrane. Therefore, the potential measurements were also performed in the presence of 6 mmol l⁻¹Ba²⁺(Fig. 2B) (for theoretical considerations, see also Weltens et al. 1992). The addition of 6 mmol l⁻¹Ba²⁺hyperpolarized V_bl. In the presence of Ba²⁺and following exposure to 10⁻³ mol l⁻¹dihydro-ouabain, V_bl decreased slightly; but after wash-out, V_bl declined a further 1 mV, suggesting a small time-dependent decay of V_bl. After wash-out of Ba²⁺, V_bl depolarized to approximately its initial control value. The potential measurements are summarized in Table 3A. Significant effects of 10⁻³ mol l⁻¹dihydro-ouabain on V_bl were detected neither in the absence nor in the presence of 6 mmol l⁻¹Ba²⁺.

At 10⁻⁴ mol l⁻¹, Sch28080 would only affect a K⁺/H⁺-ATPase, if present, or a V-type H⁺-ATPase and not the activity of a Na⁺/K⁺-ATPase (see Froissart et al. 1992). Its inhibitory action increases at lower bath K⁺concentrations (see Scott et al. 1987). In our hands, Sch28080 (10⁻⁴ mol l⁻¹) had no effect on fluid secretion rate either in a high (51 mmol l⁻¹) or in a low (10 mmol l⁻¹) [K⁺]_bl (Table 2B).

In 51 mmol l⁻¹[K⁺]_bl, 10⁻⁴ mol l⁻¹vanadate, known to be an inhibitor of E₁E₂ (P-type) ATPases (i.e. the Na⁺/K⁺- and K⁺/H⁺-ATPase) (see Nechay et al. 1986), did not affect the fluid secretion rate; but at 10⁻³ mol l⁻¹, fluid secretion was almost completely abolished within 40 min (Fig. 3A, Table 2B). In a high-K⁺, Na⁺-free solution, vanadate (10⁻³ mol l⁻¹) caused a similar inhibitory effect on fluid secretion (Fig. 3B, Table 2B). There was no statistically significant difference between the inhibitory effects in 51 or 113 mmol l⁻¹[K⁺]_bl. In both cases, the effect of vanadate was reversible.

The effect of vanadate was tested on intracellular and luminal potentials in a different series of tubules (Table 3A). In 5 and 51 mmol l⁻¹[K⁺]_bl, at a concentration of 10⁻³ mol l⁻¹, vanadate did not affect V_bl but significantly depolarized V_te; within 5 min, a steady state was reached. 10⁻⁴ mol l⁻¹vanadate had no significant effect on V_te. The depolarizing effect of 10⁻³ mol l⁻¹vanadate on V_te suggested the inhibition of an electrogenic transport system (for theoretical considerations, see also Weltens et al. 1992). Therefore, it was of interest to examine the effects of vanadate on the Ba²⁺-induced hyperpolarization of V_bl. A typical example is shown in Fig. 3C. Vanadate caused a depolarization (i.e. a reduction of the Ba²⁺-induced hyperpolarization) of V_bl within 5 min. The effect was statistically significant (Table 3A). The effects of vanadate on electrical potentials were not easily reversible within the experimental period (i.e. approximately 15 min wash-out).

Since 10⁻³ mol l⁻¹vanadate did not affect V_bl in the absence of Ba²⁺within 5 min, a drastic effect on [K⁺]_i was not expected. This was verified in 51 mmol l⁻¹[K⁺]_bl. Examples of intracellular and luminal measurements are given in Fig. 3D,E. The addition of 10⁻³ mol l⁻¹vanadate did not cause significant changes in V_bl or [K⁺]_i. In the lumen of another tubule, the reversible response of V_te and [K⁺]_l to a decrease in [K⁺]_bl in control conditions was first confirmed. Upon administation of 10⁻³ mol l⁻¹vanadate, V_te began to decrease within the first minute; after 6 min, V_te had decreased by 18 mV. Within the same period, [K⁺]_l was not affected. The experimental data are summarized in Table 3B. There was no significant effect on [K⁺]_i and [K⁺]_l. A similar lack of effect on [K⁺]_i (N=2) and [K⁺]_l (N=3) was observed (not shown) in 5 mmol l⁻¹[K⁺]_bl.

The loop diuretic bumetanide inhibits K⁺/Cl⁻and Na⁺/K⁺/2Cl⁻cotransport systems (see Ellory and Hall, 1988). A concentration of 10⁻⁵ mol l⁻¹is usually sufficient for maximal inhibition of the Na⁺/K⁺/2Cl⁻cotransporter, but much higher concentrations are generally needed to affect the K⁺/Cl⁻cotransporter (see Ellory and Hall, 1988; Palfrey and O’Donnell, 1992). Bumetanide was tested in 5, 51 and 113 mmol l⁻¹[K⁺]_bl. In 51 mmol l⁻¹[K⁺]_bl, 10⁻⁴ mol l⁻¹bumetanide significantly diminished fluid secretion rate; the application of 10⁻³ mol l⁻¹almost completely abolished it (Fig. 4A, Table 2B). Only at 10⁻⁴ mol l⁻¹was the effect reversible. In the lower [K⁺]_bl (5 mmol l⁻¹), bumetanide reversibly inhibited fluid secretion when applied at 10⁻⁵ mol l⁻¹. The inhibitory effect of 10⁻⁴ mol l⁻¹bumetanide was more pronounced and the fluid secretion did not recover within the 30 min wash-out period (Fig. 4B, Table 2B). In the high-K⁺, Na⁺-free solution, in contrast, bumetanide (10⁻⁴ mol l⁻¹) had no significant effect (Fig. 4C, Table 2B).

Since bumetanide (10⁻⁴ mol l⁻¹) diminished the fluid secretion rate in 51 and 5 mmol l⁻¹[K⁺]_bl, electrophysiological effects were tested in these bath solutions. As is shown in Fig. 4D, in 51 mmol l⁻¹[K⁺]_bl, V_te remained constant after the addition of 10⁻⁴ mol l⁻¹bumetanide; in 5 mmol l⁻¹[K⁺]_bl, V_te was 4 mV higher than the control value. On average (see Table 3A), neither in 5 nor 51 mmol l⁻¹[K⁺]_bl could a significant effect on V_te be detected. In 2 out of the 11 experiments performed in 51 mmol l⁻¹[K⁺]_bl, the drug was applied for more than 20 min without any clear change in potential difference being observed.

The effect of 10⁻⁴ mol l⁻¹bumetanide was tested on V_bl and [K⁺]_i in 5 and 51 mmol l⁻¹[K⁺]_bl on the same tubule (Fig. 4E). In control conditions, an increase in [K⁺]_bl from 5 to 51 mmol l⁻¹caused a depolarization of V_bl and an increase in [K⁺]_i. On introduction of the bumetanide-containing bath solution in 51 mmol l⁻¹[K⁺]_bl, V_bl increased slightly and [K⁺]_i remained virtually unaffected. When [K⁺]_bl was lowered to 5 mmol l⁻¹in the presence of the drug, V_bl hyperpolarized and [K⁺]_i decreased. On returning to 51 mmol l⁻¹[K⁺]_bl in the presence of bumetanide, V_bl depolarized again, while [K⁺]_i approached its previous value in 51 mmol l⁻¹[K⁺]_bl. Thus, bumetanide did not inhibit the rise of [K⁺]_i on increasing [K⁺]_bl. Washing out bumetanide in 5 mmol l⁻¹[K⁺]_bl did not change the V_bl and [K⁺]_i values measured in the presence of bumetanide. The small (irreversible) changes in V_bl and [K⁺]_i occurring in this experiment during administration of bumetanide were not systematically observed in other experiments. On average, there was no statistically significant effect of 10⁻⁴ mol l⁻¹bumetanide on V_bl or [K⁺]_i (Table 3B). The lack of an effect on V_bl was confirmed in eight experiments performed in 51 mmol l⁻¹[K⁺]_bl with 57 mmol l⁻¹Cl⁻in the solution (Cl⁻substituted by citrate): even after 20 min of application of 10⁻³ mol l⁻¹bumetanide, a concentration which effectively inhibited fluid secretion (Verhulst et al. 1988), V_bl did not change significantly (R. Weltens, personal communication).

Both Ba²⁺and bumetanide are able to impair fluid secretion substantially, probably by affecting different mechanisms. Neither Ba²⁺nor bumetanide seemed to block a change in intracellular [K⁺] when applied separately. It was of interest, therefore, to observe their combined action on fluid secretion and K⁺-selective measurements.

After three periods of 10 min in control Ringer containing 51 mmol l⁻¹[K⁺]_bl and 143 mmol l⁻¹Cl⁻, 6 mmol l⁻¹Ba²⁺was added during the following three 10 min periods. This caused a decrease in fluid secretion to 18±10% (N=3) of that in the previous control period. Over the next 30 min, 10⁻⁴ mol l⁻¹bumetanide was applied in the presence of Ba²⁺. Fluid secretion was completely inhibited within 20 min. The effect was only poorly reversible within the wash-out period (23±12% recovery after 30 min). In a control series, 6 mmol l⁻¹Ba²⁺alone was added for 60 min (N=3). This treatment also resulted in complete inhibition of fluid secretion.

In another series of experiments, the effects of a change in [K⁺]_bl on V_bl and [K⁺]_i were examined in the same cell of an isolated tubule, first in the presence of 6 mmol l⁻¹Ba²⁺and then in the presence of Ba²⁺together with 10⁻⁴ mol l⁻¹bumetanide. The change in [K⁺]_bl from 113 to 5 mmol l⁻¹was carried out in both directions. The results are summarized in Table 4. Surprisingly a change in [K⁺]_i was still observed when both bumetanide and Ba²⁺were present in the medium, although it was statistically smaller by 6±1.5 mmol l⁻¹(paired t-test, N=6) than in the presence of Ba²⁺alone (P<0.01, paired t-test).

In this series of experiments, the tubules were less sensitive to Ba²⁺. Ba²⁺still caused some hyperpolarization of V_bl (compare V_bl with the control values in Table 1 and in Weltens et al. 1992), but the effect of [K⁺]_bl on V_bl was only partly abolished by Ba²⁺(compare the effect in Tables 1 and 4).

In the present study, we have shown that in the presence of 6 mmol l⁻¹Ba²⁺in response to a rise in bath [K⁺], [K⁺]_i increased to the same extent and with a similar time course as in the absence of Ba²⁺(Fig. 1 and Table 1). This result and previous findings in Formica polyctena (see Introduction for references) strongly argue for the role of other K⁺uptake systems, besides entry through conductive channels, in maintaining fluid secretion and in adapting the cytosolic K⁺concentration to the surrounding K⁺concentration.

Ba²⁺itself seemed to have contradictory effects on [K⁺]_i: in some experiments, [K⁺]_i increased in the presence of Ba²⁺; in others, it decreased. As Ba²⁺slows down both K⁺entry across the basal membrane and K⁺exit across the apical membrane (see Leyssens et al. 1993b), a difference in balance between these two effects might explain the rise in [K⁺]_i in some tubules and the decrease in others.

A Na⁺/K⁺-ATPase has been found in insect Malpighian tubules, but only in a few species does its activity affect fluid secretion significantly (for a review, see Nicolson, 1993).

In Malpighian tubules of Formica polyctena fluid secretion continued at a very high rate in a high-K⁺(113 mmol l⁻¹) but completely Na⁺-free solution (Van Kerkhove et al. 1989), and omission of Na⁺from a 51 mmol l⁻¹K⁺solution had no effect (Table 2A). The latter findings do not support a significant role for any Na⁺-dependent K⁺uptake mechanism in fluid secretion by Malpighian tubules of Formica polyctena, at least in the presence of relatively high K⁺concentrations.

In low K⁺concentrations, omission of Na⁺from the bath decreased fluid secretion (Table 2A), but the Na⁺/K⁺-ATPase inhibitor ouabain was without effect in the presence of 5 mmol l⁻¹K⁺on either fluid secretion (Table 2B) or V_bl in the presence or absence of Ba²⁺(Fig. 2A,B, Table 3A). Thus, although the insensitivity of Na⁺/K⁺-ATPase to ouabain in Formica polyctena tubules, or even excretion and sequestration of ouabain (for a review, see Vaughan and Jungreis, 1977; Anstee and Bowler, 1979), cannot be excluded, our results suggest that, even in a low-K⁺medium, an active Na⁺/K⁺pump is of little importance for transepithelial K⁺transport in Malpighian tubules of Formica polyctena. The small Na⁺-dependence of the fluid secretion rate in low [K⁺]_bl can probably be explained by the presence of another Na⁺-dependent transport system, i.e. a Na⁺/K⁺/2Cl⁻cotransport system (see below).

Another candidate for active K⁺uptake, the K⁺/H⁺-ATPase, if present, does not seem to play any role in fluid secretion: no effect of its blocker Sch28080 could be detected.

Since vanadate interacts with both Na⁺/K⁺- and K⁺/H⁺-ATPases, no significant effect on fluid secretion was expected on adding vanadate to the bath solution. Surprisingly, when applied at 10⁻³ mol l⁻¹, vanadate strongly inhibited the fluid secretion rate, the inhibitory effect becoming statistically significant after 20 min (Fig. 3A). In both 5 and 51 mmol l⁻¹[K⁺]_bl, V_te significantly depolarized to a new steady-state level within 5 min of the addition of vanadate. The observation that V_bl, [K⁺]_i and [K⁺]_l remained unaffected (Fig. 3D and Table 3A,B) seems to indicate that K⁺(Cl⁻?) and water transport were affected to the same extent: the speed of passage of K⁺and water was slowed down without changing the actual K⁺concentrations. This, in itself, seems to exclude an effect of vanadate on a Na⁺/K⁺-ATPase, as this would have altered the K⁺and Na⁺concentrations in the cell and, consequently, V_bl. Furthermore, since the inhibitory effects of vanadate on fluid secretion rate in Na⁺-containing (i.e. 51 mmol l⁻¹[K⁺]_bl) and Na⁺-free (i.e. 113 mmol l⁻¹[K⁺]_bl) conditions were comparable (Fig. 3A,B, Table 2B), its strong inhibitory effect was probably not due to an interaction with a hypothetical Na⁺/K⁺-ATPase. These results are different from the response of Malpighian tubules of Locusta migratoria, where both vanadate and ouabain caused a slow depolarization of both V_bl and V_ap (Baldrick et al. 1988).

Because in Formica polyctena there was no evidence for the presence of a K⁺/H⁺-ATPase either (see above), we had to consider other possible sites of action. A complex indirect effect of vanadate was suggested for Malpighian tubules of Drosophila melanogaster (Bertram et al. 1991). A multitude of stimulatory and inhibitory intracellular mechanisms for vanadate have been reviewed by Chasteen (1983) and Nechay et al. (1986). The effects on the fluid secretion rate and electrophysiological variables of Formica polyctena Malpighian tubules were comparable to, and followed a similar time course to, the effects of the V-type ATPase inhibitors bafilomycin A₁ and N-ethylmaleimide (NEM) reported by Weltens et al. (1992) on the same preparation. In the latter study, the reduction of the Ba²⁺-induced hyperpolarization was attributed to an electrogenic current generated by the apical H⁺-ATPase and blocked by bafilomycin A₁ and NEM. Unlike ouabain (in 5 mmol l⁻¹[K⁺]_bl), vanadate (10⁻³ mol l⁻¹in 51 mmol l⁻¹[K⁺]_bl) did significantly reduce the Ba²⁺-induced hyperpolarization of V_bl (Fig. 3C, Table 3A).

Reports of a direct effect of vanadate on the apical H⁺pump are controversial. According to Forgac (1989), vacuolar-type H⁺-ATPases are insensitive up to at least 1 mmol l⁻¹. Chatterjee et al. (1992), however, described the presence of a novel V-type H⁺-ATPase in osteoclast plasma membrane vesicles with a unique pharmacology and with specific isoforms of two subunits in the catalytic portion of the enzyme; the H⁺-ATPase activity was completely inhibited by 10⁻³ mol l⁻¹vanadate. 5×10⁻⁴ mol l⁻¹vanadate caused 50% inhibition of the ATP-dependent proton uptake of isolated chromaffin granules and yeast vacuoles (Beltrán and Nelson, 1992). This is in line with a small effect (10–15% inhibition only) of a lower concentration of vanadate (10⁻⁴ mol l⁻¹) on the cation-stimulated ATPase activity in purified goblet cell apical membranes of Manduca sexta (Wieczorek et al. 1986), later identified as a V-type H⁺-ATPase (Wieczorek et al. 1989). The effect was not significant and was considered to be negligible by the authors at that time, but might represent the threshold for vanadate-sensitivity of this ATPase.

In summary, no conclusive evidence was found for the existence of basal active K⁺uptake mechanisms in exchange for either Na⁺or H⁺. Their role in maintaining an asymmetric ion distribution across the basal membrane and a particular rate of fluid secretion is apparently minimal. The strong inhibitory effect of vanadate on fluid secretion can probably be explained by an effect on the apical H⁺pump.

The existence of coupled entry mechanisms for K⁺, Na⁺and Cl⁻has been investigated in Malpighian tubules of other insects, but only in a low bath K⁺concentration (for a review, see Nicolson, 1993). In Malpighian tubules of Formica polyctena a more extensive study was carried out to find evidence for an electroneutral coupled transport of K⁺in different bath K⁺(and Na⁺) concentrations.

In high-K⁺(113 mmol l⁻¹) Na⁺-free conditions, K⁺entry via a Na⁺/K⁺/2Cl⁻cotransporter can be excluded. K⁺uptake via a K⁺/Cl⁻cotransporter is not very likely either, as bumetanide had no effect at a concentration as high as 10⁻⁴ mol l⁻¹(Fig. 4C, Table 2B).

In 51 mmol l⁻¹[K⁺]_bl, the fluid secretion rate was not affected by omitting Na⁺from the bath solution (Table 2A). Although bumetanide binding to a Na⁺/K⁺/2Cl⁻cotransporter should be favoured in the 51 mmol l⁻¹ K⁺, 62 mmol l⁻¹ Na⁺bath solution (see Haas and Forbush, 1986), no change in fluid secretion rate was observed after the addition of 10⁻⁵ mol l⁻¹bumetanide (Fig. 4A, Table 2B). These findings do not favour K⁺entry via a Na⁺/K⁺/2Cl⁻cotransporter. However, the significant reduction of the fluid secretion rate after the addition of 10⁻⁴ mol l⁻¹bumetanide suggests that a large portion of the basal K⁺uptake may occur via a K⁺/Cl⁻cotransporter (Fig. 4A, Table 2B). Since, at this concentration, no effect of bumetanide was observed in 113 mmol l⁻¹[K⁺]_bl, specific effects are unlikely. The observation that Br⁻has a stimulatory effect on fluid secretion in 51 mmol l⁻¹[K⁺]_bl corroborates the hypothesis of a K⁺/Cl⁻cotransport system (Table 2A; see Ellory and Hall, 1988). At 10⁻³ mol l⁻¹, the inhibitory effect of bumetanide was much more pronounced. However, at this high concentration, interaction with other Cl⁻-dependent transport mechanisms cannot be excluded; in whole-cell recordings of the rat lacrimal gland, it was shown that 4X10⁻⁴ mol l⁻¹bumetanide inhibited Cl⁻channels, for instance (Evans et al. 1986).

In a low-K⁺, high-Na⁺medium (i.e. 10 or 5 mmol l⁻¹[K⁺]_bl), both Na⁺omission and Cl⁻substitution by Br⁻caused a reversible reduction of the secretion rate (Table 2A), and bumetanide clearly inhibited fluid secretion at 10⁻⁵ mol l⁻¹(see Fig. 4B, Table 2B). These findings argue for the presence of a Na⁺/K⁺/2Cl⁻cotransporter (Ellory and Hall, 1988). Evidence for a Na⁺/K⁺/2Cl⁻cotransporter (Maddrell, 1969; O’Donnell and Maddrell, 1984; Hegarty et al. 1991) and/or Cl⁻-dependent K⁺uptake (Wessing et al. 1987; Baldrick et al. 1988) has also been found for Malpighian tubules of other species.

That bumetanide (and/or another loop diuretic, furosemide, for other species) affects electroneutral transport mechanisms is confirmed by the lack of a significant effect on electrical potential differences (Table 3A,B) (cf. V_bl in Locusta migratoria, Baldrick et al. 1988, and Onymacris, Nicolson and Isaacson, 1987; V_ap in Aedes aegypti, Hegarty et al. 1991). Long-term application, however (several minutes), might change the intracellular K⁺, Cl⁻and/or Na⁺concentrations and hence V_bl (cfr. Rhodnius prolixus, O’Donnell and Maddrell, 1984; Aedes aegypti, Hegarty et al. 1991). In Malpighian tubules of Formica polyctena, [K⁺]_i did not change much, as V_bl did not alter.

An important observation was that bumetanide did not hamper the response of [K⁺]_i to a change in [K⁺]_bl from 5 to 51 mmol l⁻¹(Fig. 4E, Table 3B). This means that other pathways, probably K⁺channels, can ensure basal K⁺uptake and adapt the intracellular K⁺content to the surrounding K⁺concentration. This finding indirectly points to the existence of an inwardly directed basal electrochemical K⁺gradient in the latter conditions.

To summarize, in a high-K⁺(113 mmol l⁻¹) Na⁺-free solution, coupled entry of K⁺and Cl⁻seems to be of little importance. In 51 mmol l⁻¹[K⁺]_bl, a substantial portion of K⁺entry can occur via a K⁺/Cl⁻cotransporter. In low [K⁺]_bl, a Na⁺/K⁺/2Cl⁻cotransporter seems to become important.

When both Ba²⁺and bumetanide were present, fluid secretion was completely blocked. It was possible, however, that this was due solely to the continued presence of Ba²⁺. Also, the effect of [K⁺]_bl on [K⁺]_i was reduced, but not abolished. This could mean either that the proposed model (see below) is still incomplete and that other K⁺ uptake mechanisms are present in the basal membrane or that Ba²⁺does slow down K⁺ entry, but does not completely block it (see, for instance, the difference in sensitivity to Ba²⁺of the series of tubules in Tables 1 and 4). Dijkstra et al. (1994) found that Ba²⁺always increased the basal membrane resistance, but that the hyperpolarization of V_bl was smaller in the ant tubules in their study than in those investigated by Weltens et al. (1992). It would be interesting to study the Ba²⁺-sensitivity of single K⁺ channels using the patch-clamp technique. The observation that Ba²⁺can completely block fluid secretion if applied for long enough can be explained by its concomitant hyperpolarizing effect on the apical membrane. This will severely reduce the turnover of the electrogenic H⁺ pump and the build-up of the proton concentration gradient that is needed to drive K⁺ into the lumen (see Leyssens et al. 1993b).

Several hypothetical transport mechanisms (Fig. 10 in Leyssens et al. 1992) have now been investigated. Previous results have shown that, at the apical side, K⁺ extrusion into the lumen is realized via the combination of an electrogenic H⁺ pump and an electroneutral K⁺/H⁺ antiporter (see Weltens et al. 1992; Zhang et al. 1994; Leyssens et al. 1993a). The present study shows that the pathways for basal K⁺ entry seem to be conductive channels and secondary active cotransporters and that their relative importance varies depending on the bath K⁺ (and Na⁺) concentration. Primary active K⁺ uptake systems, i.e. a Na⁺/K⁺-or K⁺/H⁺-ATPase, if present, do not seem to contribute significantly to transepithelial transport or to the maintenance of intracellular K⁺ (and Na⁺) levels. Fig. 5 shows the proposed transport model for Malpighian tubules of Formica polyctena.

In high-K⁺, Na⁺-free conditions, a coupled entry of K⁺ and Cl⁻ seems to be of little importance. K⁺ uptake probably occurs via basal K⁺ channels, driven by a favourable electrochemical gradient (Fig. 5C) (see also Leyssens et al. 1993a). In more physiological conditions, i.e. 51 mmol l⁻¹ [K⁺]_bl, when the basal electrochemical K⁺ gradient is smaller (see Leyssens et al. 1993a), K⁺ entry via channels may become less important and a K⁺/Cl⁻ cotransport system probably accounts for a large portion of K⁺ uptake (Fig. 5B). In 5 or 10 mmol l⁻¹ [K⁺]_bl (and high-Na⁺ conditions), little K⁺ is available for transepithelial transport; the basal electrochemical K⁺ gradient is even smaller than in 51 mmol l⁻¹ [K⁺]_bl and possibly outwardly directed (see Leyssens et al. 1993a). The fluid secretion rate is low but still present. In this condition, a Na⁺/K⁺/2Cl⁻ cotransporter may take over a large part of the basal K⁺ uptake (Fig. 5A).

Furthermore, the different basal transport systems for K⁺ entry may function at different rates depending on the basal K⁺ (and Na⁺) concentration and may thereby determine the amount of K⁺ available for transport across the apical membrane via the K⁺/H⁺ antiporter. The rise in the intracellular [K⁺] in response to an increased basal K⁺ concentration could occur either via K⁺ channels or via a coupled K⁺/Cl⁻ entry mechanism.

In a low-K⁺, high-Na⁺ solution, the intracellular K⁺ content is markedly reduced (see Leyssens et al. 1993a). Consequently, intracellular K⁺ has to be replaced by another cation, probably Na⁺. A Na⁺ conductance could not be detected (see Leyssens et al. 1992), but the uptake of Na⁺ might occur via a Na⁺/K⁺/2Cl⁻ cotransporter. The luminal K⁺ concentration also decreased when [K⁺]_bl was lowered, but not to the same extent as [K⁺]_i (see Leyssens et al. 1993a). This implies that K⁺ is still the major cation secreted. Consequently, although the apical K⁺/H⁺ antiporter may have some affinity for Na⁺ when intracellular Na⁺ levels rise, secretion of K⁺ still predominates. This means that net basal K⁺ uptake must exceed net Na⁺ uptake. Additional uptake via a K⁺/Cl⁻ cotransporter seems unlikely; it would need a very low intracellular Cl⁻ concentration. Preliminary results show that this is not the case (unpublished observations). As no evidence was found for a Na⁺/K⁺-ATPase, we have no indication of how intracellular [Na⁺] is regulated in our preparation. Theoretically, in view of the high intracellular pH found in Malpighian tubules of Formica polyctena (Zhang et al. 1994), a Na⁺/H⁺ antiporter in the basal membrane, working in reverse mode and unexpectedly functioning at an alkaline pH, might extrude Na⁺ in exchange for H⁺.

The transepithelial Cl⁻ pathway awaits further investigation. In high [K⁺]_bl (113 mmol l⁻¹), the electroneutral K⁺/Cl⁻ cotransporter did not seem to be functional and no relative Cl⁻ conductance was found. This suggests that Cl⁻ transport is primarily paracellular. The lack of a relative Cl⁻ conductance in the basal membrane (Leyssens et al. 1992) points to a paracellular pathway for Cl⁻ down an electrical gradient in 113 mmol l⁻¹ [K⁺]_bl. In 51 and 5 mmol l⁻¹ [K⁺]_bl, K⁺/Cl⁻ and Na⁺/K⁺/2Cl⁻ cotransporters, respectively, seem to be functional and at least part of the Cl⁻ transport may occur transcellularly; hence, Cl⁻ channels would be required to ensure Cl⁻ exit across the apical membrane. In a few experiments in 5 mmol l⁻¹ [K⁺]_bl, a small hyperpolarization of V_ap was observed when bumetanide was applied (Fig. 4D) (cf. Rhodnius prolixus, O’Donnell and Maddrell, 1984; Locusta migratoria, Baldrick et al. 1988). If a Cl⁻ conductance were present in the apical membrane, this might be explained by a drop in intracellular Cl⁻ concentration due to inhibition of basal Cl⁻ influx. The existence of a transcellular Cl⁻ pathway was also suggested by a study on symmetrically perfused Formica polyctena tubules (Dijkstra et al. 1994). Cl⁻ channels in the apical membrane have been detected by a patch-clamp study in Aedes aegypti (Wright and Beyenbach, 1987) and in Drosophila melanogaster tubules (J. Dow and B. Harvey, personal communication). Apical Cl⁻ channels (as in rectal gland; Greger et al. 1989) and/or bumetanide-sensitive basal cotransporters (as in Aedes tubules, Hegarty et al. 1991) could be a site of regulation by hormones and second messengers to accelerate fluid secretion rate even at lower bath K⁺ concentrations.

Preliminary results obtained from intracellular and luminal ion-selective measurements in different bath K⁺ concentrations have demonstrated that the calculated gradients for Cl⁻ and K⁺ are consistent with the transport mechanisms for K⁺ and Cl⁻ proposed in the model.

The authors wish to thank Mr P. Pirotte for making the electrodes and performing part of the fluid secretion measurements, Mrs J. Vanderhallen for preparing the solutions, Mr R. Van Werde for help with the electronics, Mr W. Leyssens for administrative tasks, Ms K. Ungricht for typing the tables and Mr and Mrs Withofs for artwork. This work was supported by a grant from NFWO (Nationaal Fonds voor Wetenschappelijk Onderzoek, Belgium) and by a grant from the EC (European Community: SC1-CT90-0480).