ABSTRACT
The K+-secreting larval midgut of Manduca sexta in vitro was voltage-or current-clamped. In contrast to Tl+, NH4+and Na+, both Rb+and K+generated a short-circuit current, although with different saturation kinetics. The dependence of the short-circuit current on Rb+ /K+ mole fraction gave no evidence for multi-ion occupation of the basolateral K+ channels. After ‘functionally’ eliminating the apical membranes using the ionophore amphotericin B and the ‘apical K+pump’ blockers trimethyltin chloride or Tl+, the K+channels could be more closely investigated. By measuring zero-current potentials, permeability ratios PX/PK were estimated using an adapted version of the Goldman–Hodgkin–Katz voltage equation. Their sequence was K+ (1) = Tl+ > Rb+ (0.38) > NH4+ (≈0.3) > Cs+(0.03) > Na+(≈0). The K+channels could not be blocked by basally applied Cs+, Na+ or tetraethylammonium. Blockade of K+ current by Ba2+ was typically voltage-dependent, but only at moderate transbasal voltages. The relative electrical distance δ of the Ba2+ binding site from the basal channel opening was determined to be 0.2. At zero transbasal voltage, the apparent inhibition constant for barium KB* a was 1.7 mmol l−1.
INTRODUCTION
Active K+ secretion by vertebrate epithelia involves a basal Na+ /K+ -ATPase and apical K+ channels (Nielsen, 1984; Foster et al. 1984). In insect epithelial tissues, a novel mechanism has recently been found for the well-known (e.g. Harvey et al. 1968) electrogenic K+ secretion. The apical cooperation of a V-type H+ -ATPase and an H+ /cation antiporter mediates uphill K+ exit in Malpighian tubules (Bertram et al. 1991; Weltens et al. 1992) and in the midgut goblet cells of lepidopteran caterpillars (Wieczorek et al. 1991). In the tubules, basal K+ entry into brush-border cells may be driven by secondary active Na+ /K+ /2Cl− or KCl transporters, whereas the Na+ /K+ -ATPase (if present) usually plays a minor role (Maddrell and O’Donnell, 1992). In Formica polyctena tubules, however, K+ uptake from the haemolymph is passive (Weltens et al. 1992), just as it is in the larval lepidopteran midgut (Harvey and Zerahn, 1972; Zeiske et al. 1986).
In Manduca sexta midgut, the epithelium in which H+ -pump-energized K+ secretion was first discovered, the apical H+ /cation antiporter accepts Rb+, Li+ and Na+ besides K*(Wieczorek et al. 1989, 1991). Whether K+ -like Cs+, Tl+ and NH4+ are also transported is not known. Moreover, in the American silkworm Hyalophora cecropia, the entire midgut K+ -secreting system, including the basal transport step, handles Rb+ (Harvey et al. 1968), in some cases Cs+ (Zerahn, 1970) and, in the absence of haemolymph Ca2+ and Mg2+, even Na+ and Li+ (Harvey and Zerahn, 1971). If the Ba2+-blockable ion channels known to mediate basal K+ entry in the M. sexta midgut (Zeiske et al. 1986) were to have such a low selectivity, the term ‘K+ channels’ would surely be inadequate. In several vertebrate epithelia, there are relatively non-selective cation channels that are indeed permeable to K+, Rb+, Cs+, Na+ and Li+ and can be blocked by Ca2+ and Ba2+ (Aelvoet et al. 1988; Krattenmacher et al. 1991).
The present study attempts to determine the extent to which the basolateral K+ conductance in M. sexta midgut fits (at physiological haemolymph Ca2+ and Mg2+ levels) a typical K+ channel population with respect to permeability, ion occupation and block. We tested whether cations other than K+ could be secreted by the intact in vitro midgut and examined the selectivity of the basolateral ion channels after ‘electrical elimination’ of the apical membranes. The latter was performed by using the ionophore amphotericin B and an ‘apical K+ pump’ inhibitor. We also tested the efficacy of some putative K+ channel blockers and investigated the voltage-dependent block by Ba2+.
MATERIALS AND METHODS
Tobacco hornworm (Manduca sexta L.) larvae were raised at 26°C (12 h:12 h light:dark) from eggs kindly provided by Professor H. Wieczorek (Munich) and fed a synthetic diet (Carolina Biological Supply Co., Burlington, NC, USA). All larvae were in their fifth instar and weighed between 3 and 5 g when used. The preparation and experimental apparatus were as described previously (Schirmanns and Zeiske, 1994). The chamber was like that used by De Wolf and Van Driessche (1986) except that the exposed tissue area was only 0.26 cm2. Both chamber compartments were continuously superfused by a gravity feed system.
Solutions
The bathing solutions used in this study were either standard saline [32K(Cl); Nedergaard and Harvey (1968)] or modifications thereof (Table 1). In all Cl−-free salines, Mg2+ and Ca2+ were added as gluconate instead of chloride; in all K+ -free salines, KHCO3 was replaced by NaHCO3; all other cations were introduced as chloride, nitrate or gluconate salts instead of KCl. Tl+ was added as the nitrate salt, since TlCl is insoluble. Osmotic strength was balanced with sucrose. The K+ chemical activity, aK+, of the solutions was kindly measured by Professor H. Wieczorek. aK+ of 90K salines was 61 mmol l−1, whereas the activity coefficient for K+, and presumably also for other test ions (X+ ), in 32X salines was approximately 1. Amphotericin B (Sigma) was added from a dimethylsulphoxide stock solution (final concentration 5 μmol l−1) and trimethyltin chloride (TMT; Merck-Schuchardt, Hohenbrunn, Germany) was added from an aqueous stock solution (final concentration 50 μmol l−1).
Electrical recordings
The tissues were clamped to zero current during the first 30 min and thereafter short-circuited. Current pulses in the former and voltage pulses in the latter state allowed the determination of the tissue resistance (Rt) and of the spontaneous transepithelial potential difference (PDt) or short-circuit current (Isc), respectively. The experiments were started after a total equilibration period of 45 min. In most equilibrated tissues, Isc declined slightly with time (less than 20% h−1) as expected (Wolfersberger and Giangiacomo, 1983; Crawford and Harvey, 1988). The data were corrected for the extrapolated Isc decay if necessary (i.e. in dose–response or ion-substitution experiments on intact tissue).
For current–voltage (I–V) measurements, voltage pulses of 400 ms duration with alternating polarity and increasing magnitude were applied to the otherwise short-circuited tissue. The sequence was thus positive pulse, negative pulse, short-circuit (1 s); increase pulse amplitude, positive pulse, negative pulse, short-circuit (1 s); etc. The current was monitored on a chart recorder (Linseis, Selb, Germany). Very small (less than 10%) current transients (about 200 ms) at the onset of a voltage pulse have been described for the intact epithelium (Moffett, 1980) and were also observed in this amphotericin-treated preparation. Because of the limited response frequency of our chart recorder, we were unable to register ‘instantaneous’ I–V curves. As a compromise, we chose a voltage pulse length of 400 ms, close to the 300 ms used by Moffett (1980).
All I–V curves were recorded after perforating the apical membranes with amphotericin B, so tissue resistances were small and the maximal voltage amplitude was limited to ±50 to ±100 mV by the maximal compensation current of the clamp circuit (±4.4 mA cm−2).
Statistics
All values are presented as means ± standard errors. Differences between paired groups were evaluated with the Student’s t-test. P<0.05 was considered to represent a significant difference.
RESULTS
The magnitude of K+ transport
With 32K(Cl) standard saline on both sides of the in vitro tobacco hornworm midgut, active K+ secretion drops rapidly during the first 30 min after mounting, thereafter entering a slow-decline phase or ‘pseudo steady state’ (Cioffi and Harvey, 1981; Wolfersberger and Giangiacomo, 1983). The average changes in the electrical variables immediately after mounting the preparations in the chamber until the end of a 45 min equilibration period were as follows: the lumen-positive PDt fell from 76±3 to 41±2 mV, the luminally directed Isc fell from 1450±100 to 520±30 μA cm−2 and Rt rose from 59±3 to 86±5 D cm2 (N=36 preparations).
Selectivity of the active transport pathway
A series of experiments was performed to investigate the extent to which the epithelial secretory system could distinguish between K+ and monovalent cations of similar size. The short-circuited tissues were perfused with symmetrical salines containing the appropriate test ion. When ions are being actively transported, one can check, by using the specific blocker Ba2+, whether the basal passage occurs through K+ channels. Fig. 1A demonstrates that substitution of K+ by Rb+ hardly affected Isc, which was more sensitive to basal Ba2+. Replacing K+ with Tl+ at the end of the experiment caused Isc to drop to zero after 2 min. This is expected because Tl+ is not only a ‘K+ -like’ cation but also inhibits the apical transport step (Zerahn, 1982; Schirmanns and Zeiske, 1994). Two further typical features may be seen. The first is the slow decay of Isc for the above-mentioned in vitro midgut, the second is the Isc transient that occurs during K+ /Rb+ substitutions in the M. sexta midgut (as well as in the H. cecropia midgut; Nedergaard and Harvey, 1968). From 16 experiments similar to that shown in Fig. 1A, we obtained a ratio of Rb+ -carried to K+ -carried current (IRb/IK) of 0.80±0.02. Fig. 1B shows the Isc time course when K+ was replaced by NH4+. As with Tl+, Isc was quickly abolished (after 4 min) in this and four analogous experiments. Only for brief exposure times (2–15 min) was there a slow but full recovery of K+ current after treatment with Tl+ (see also Fig. 3, in the presence of amphotericin) or NH4+ (see Fig. 4B). In the intact tissue, therefore, Rb+ is virtually able to replace K+ in the secretory process, whereas Tl+ and NH4+ are not.
Moffett and Koch (1988) reported that a rise in basal K+ concentration ([K+ ]b) results in a saturation of Isc but leaves cytoplasmic [K+ ] virtually unaffected. This finding suggests that K+ transport across the basolateral membranes is the rate-limiting step of K+ secretion. Consequently, the observed behaviour of Isc would be a property of the K+ channels. We therefore conducted six experiments to compare the substrate-dependence of IK and IRb. The appropriate current-carrying cation was substituted by impermeant (Zeiske et al. 1986) Na+ and its concentration was then stepped up by gradual Na+ replacement. We did not use test ion concentrations below 8 mmol l−1because, at least for K+, there is no passive inward driving force across the basolateral membranes under such conditions (Chao et al. 1990). For Isc carried by K+ or Rb+ the Michaelis–Menten relationship (equation 6 in Appendix 1) was fitted (Fig. 2A) to pooled data (N=6) whose Hanes plot ([K+ ]/Iscversus [K+ ]) is linear (not shown), indicating saturation kinetics. KRb was significantly higher than KK, whereas the extrapolated maximal Isc ratio was Imax,Rb/Imax,K=1.1. Therefore, and despite the lower binding affinity of Rb+ compared with that of K+, the basolateral K+ channels seem to have about equal maximal capacity for both ions. In order to detect possible interactions between Rb+ and K+ within the channel, we tested the Isc for anomalous, i.e. non-linear, behaviour when mixtures of both ions were varied from 100% K+ to 100% Rb+. Fig. 2B summarizes the data from three preparations and demonstrates that there were no anomalies: a linear relationship roughly fits the mole-fraction-dependence of Isc in K+ –Rb+ mixtures, as does a fit (dashed line) based on equation 7, which describes a one-site model (Appendix 1).
Basolateral permeability to small monovalent cations
In a previous study on the M. sexta midgut (Schirmanns and Zeiske, 1994), we showed that apical use of amphotericin B virtually eliminates the resistance of the apical membranes without significantly affecting the basolateral or (supposedly paracellular) ‘shunt’ conductances. This treatment effectively short-circuits the apical K+ driving force. After additional inhibition of remaining active transport activity (e.g. by Tl+ or TMT), the preparation may be modelled as a parallel arrangement of basolateral membranes and septate junctions. When tissues held under these conditions are bathed with bilaterally identical salines, there is no transepithelial driving force, so net ionic movements can only be generated by electrical or concentration gradients. The following experiment investigates whether Tl+, being transepithelially ‘impermeant’ (Fig. 1A), is able to pass through the basolateral K+ channels. As can be seen from Fig. 3, Tl+ and K+ generated currents of about equal magnitude when a luminally directed concentration gradient was established. The slow wash-out of Tl+ is also typical (see also Fig. 4A). The Isc could be blocked by basal Ba2+, with a seemingly higher blocking efficiency when K+ rather than Tl+ was the permeant cation. Thus, Tl+ appeared to permeate the cells through basolateral Ba2+-blockable K+ channels. The average current ratio ITl/IK of five similar experiments was 0.92±0.04. We conducted an analogous attempt to force a NH4+ current across the preparation. Unfortunately, under these conditions Isc hardly ever approached a steady state (see also Fig. 4B).
Rb+ and Tl+ could easily pass through the basolateral K+ channels, whereas the situation for NH4+ remained unclear. Next, we inspected more closely the ‘true’ basolateral permeability to K+ -like cations by adopting the Goldman–Hodgkin–Katz (GHK) approach. In these ‘single-membrane’ experiments, the apical salines were 90K(Cl) or 90K(Glu), whose [K+ ] and pH were at cytoplasmic levels (see Chao et al. 1991). The GHK current equation for absolute permeabilities was not applicable to current–voltage curves (N=4, not shown) recorded after electrical elimination of the apical membranes. Despite a 10:1 [K+ ] asymmetry, the corresponding I–V relationship was almost linear, both before and after tentative correction for ‘shunt’ current contributions (see Appendix 2).
The GHK current equation does not adequately describe ion permeation through channels that exhibit, among other things, saturation (Hille, 1992; see Fig. 2). Therefore, so-called reversal potentials (Erev) across the ‘single-membrane preparation’ were measured under zero-current conditions, and permeability ratios were determined by using the GHK voltage equation, which is not susceptible to the above restrictions (Hille, 1992; Lester, 1991). Fig. 4A illustrates an experiment in which Cs+, Na+, K+, Rb+ and Tl+ were successively substituted on the basal side. It can be seen that the permeability to Cs+ was very small (large Erev), whereas that to Rb+ was considerable. Tl+ was just as permeant as K+ (small Erev). To check the effects of Tl+, we had to work with NO3− solutions: Erev was more negative at a given [K+ ] in NO3− solutions than in Cl− solutions, probably as a consequence of the higher ‘shunt’ permeability to NO3− (Chao et al. 1989). The minute change in Erev when basal [Na+ ] was decreased from 90 to 32 mmol l−1was also typical (N=9) and indicated the relative impermeability (of the shunt as well as of the K+ channels) to this cation. Fig. 4B shows that basal introduction of NH4+ resulted in a rapid potential change followed by a slow depolarisation (with concomitant resistance increase) that did not reach a steady value. We interpreted this result to indicate that NH4+ was permeant, and used the starting point of the slow phase to estimate the permeability ratio. The tissue resistance, Rt, increased with time in the presence of basal Tl+ and NH4+ and was only slowly reversible after basal reintroduction of K+, again demonstrating the long-term effects of these two cations.
An algorithm developed in Appendix 2 (equation 13) describes Erev as a function of the product [X+ ]bPX/PK, where X+ is the test ion and PX/PK is its permeability ratio with K+ as reference. However, this approach is based on the assumption that the ‘shunt’ is equally and/or hardly permeable to small monovalent cations. As demonstrated in Fig. 5, equation 13 fitted the measured dependence of Erev on the basal K+ concentration ([K+ ]b=[X+ ]bPX/PK) very well. Once the function Erev=f([K+ ]b) was known, the permeability ratios of other test cations could be easily evaluated. One way to do this is to determine the value of [K+ ]b that generated the same Erev as 32 mmol l−1of the basal cation in question (as indicated by the dashed lines in Fig. 5) and then to calculate PX/PK=[K+ ]b/(32 mmol l−1). For this purpose, we used nomograms like that shown in Fig. 5, from which we obtained the relative permeability to Cs+ and Rb+ from one curve (in the presence of Cl−) and that to Tl+ from the other curve (in the presence of NO3−). Pooling of such experiments (N=11 for Rb+, N=6 for Cs+, N=4 for Tl+, N=3 for NH4+ ) yielded the permeability ratio sequence: K+ (1) = Tl+ > Rb+ (0.38) > NH4+ (≈0.3) > Cs+ (0.03) > Na+ (≈0).
Blockade of K+ channels
To characterize the basolateral K+ channels further, we tested the effects of the putative channel blockers Cs+, Na+ and tetraethylammonium (TEA+ ) on K+ -carried Isc across the intact tissue. Cs+ did not significantly inhibit Isc or alter Rt when applied basally at 32 mmol l−1[as 32K32Cs(Cl) saline; N=7]. The same was true for symmetrical application of 32 mmol l−1Na+ [32K32Na(Cl) saline; N=4]. TEA+, when added basally at 10 mmol l−1, did not affect Isc (N=3). As shown above, Ba2+ is a suitable tool for confirming cationic currents across the basolateral K+ channels. In a final series of experiments, we inspected the mode of action of this blocker. We added Ba2+ to the basal saline and approximately balanced ionic and osmotic strength on both sides with an inert cation (Tris at pH 7) and with sucrose, as described previously (Schirmanns and Zeiske, 1994). Fig. 6A depicts the dose-dependent effect of the blocker on the K+ -carried Isc of a typical intact tissue: 12 mmol l−1basal Ba2+ did not even halve Isc and tissue conductance (Gt). As shown in Fig. 6B, the inhibition of Isc by Ba2+ appeared to conform to Michaelis–Menten kinetics. According to this cumulative analysis of six experiments, only about 60% of Isc was susceptible to Ba2+ with a mean half-maximal effect (apparent inhibition constant, ) at 4.6 mmol l−1.
Voltage-dependence of Ba2+ block
In order to record current–voltage (I–V) curves of the basolateral membranes at different [Ba2+]b, we again electrically eliminated the apical membranes using amphotericin B and trimethyltin chloride (TMT). Fig. 7A shows that the I–V relationship for one such preparation was, as expected for equal [K+ ] on both sides, basically linear (ohmic) over most of the measured voltage range in the absence of Ba2+, but was transformed by the blocker into a curve; i.e. the current block was voltage-dependent. A somewhat clearer picture emerges from the graph in Fig. 7B, in which the slope between adjacent data points of the I–V curve (i.e. the conductance, Gt) is plotted as a function of Vt. Under control conditions, Gt remained roughly constant between −75 and +25 mV, but it started dropping at more (apical) positive voltages, apparently stabilizing at about half its original level. The Ba2+ block of the transepithelial conductance between +25 and −25 mV was dose-dependent and increased with increasing apical negativity. At more positive and more negative voltages, Gt increased again and tended to become larger than values measured in the absence of Ba2+.
To prove the voltage-dependence of the blocker’s binding affinity, we adopted the Woodhull model (see Appendix 1). This assumes a single binding site for the blocker within the channel and predicts an exponential dependence of the apparent inhibition constant on the transmembrane voltage (equation 8). Therefore, we evaluated at each Vt from plots similar to that shown in Fig. 6B for the intact tissue. The current was not precisely zero at Vt=0 (Fig. 7A), as would be expected for symmetrical salines, so we assumed, for the sake of simplicity, that this deviation was voltage-independent (possibly caused by minute residues of active transport) and corrected the tissue current data for this amount. Fig. 8A shows a plot of lnversus Vt for the data based on the I–V relationship depicted in Fig. 7A. It indeed displays the expected linearity of the lnversus Vt function, but only within the range from −20 to +20 mV; at more extreme voltages, was almost constant. Although these features appeared in all four such experiments, the Vt range in which the Ba2+ block was voltage-dependent varied considerably: Fig. 8B shows data from an experiment in which changed exponentially with voltage between −60 and +60 mV. Applying equation 8 to the linear part of the lnversus Vt plots permitted an evaluation of the relative electrical distance δ of the blocker-binding site from the outer channel opening and the apparent inhibition constant at zero voltage (; see Appendix 1). The mean values from four tissues were δ=0.2±0.1 and =1.7±0.3 mmol l−1.
DISCUSSION
Transepithelial secretion of cations
Cooperation of the apical H+ pump with the apical H+ /cation antiport and the basal K+ channel has been reviewed for insect midgut and Malpighian tubules (Zeiske, 1992; and Introduction). The antiport accepts Rb+ and Na+ besides K+ (Wieczorek et al. 1991), whereas Tl+ inhibits this transporter (Schirmanns and Zeiske, 1994).
The basal side contains K+ channels that are part of the transcellular pathway (Zeiske et al. 1986; Zeiske, 1992). They are sensitive to Ba2+ (Moffett and Koch, 1985; Zeiske et al. 1986, 1992), lidocaine (Moffett and Koch, 1991) and quinidine (Alpert, 1989). The basal K+ permeability was stimulated by Cl−, but inhibited by agents that usually impair Cl− movements across membranes (Zeiske et al. 1992). These effects were not seen in the presence of basal Ba2+ (Zeiske and Marin, 1992), suggesting that the K+ channel shares some similarity with K+ /Cl− cotransporting molecules.
The pharmacological profile was ambiguous so we wanted to extend it and, in addition, to elucidate the selectivity profile of the K+ channel. The active transport of Rb+ by the in vitro midgut of M. sexta (Fig. 1A) is not surprising, since it is already known for Hyalophora cecropia (Harvey et al. 1968; Nedergaard and Harvey, 1968; Zerahn, 1980). At first glance, Rb+ behaves as a fairly good substitute for K+ in the M. sexta midgut, where it carries a Ba2+-sensitive current whose magnitude, at 32 mmol l−1Rb+, is 80% of that of the K+ current. However, differences in handling of both ions appear from the concentration-dependence of Isc (Fig. 2A), which saturates with a significantly higher Michaelis constant in the presence of Rb+ (KRb=26 mmol l−1) than in the presence of K+ (KK=10 mmol l−1). We return to this point later in the Discussion. The KK values published hitherto vary considerably: Zeiske et al. (1992) measured a value of 64 mmol l−1; Moffett and Koch (1988) one of 14 mmol l−1(extracted from their Fig. 5) and our estimate is even lower. These workers used the same 32K(Cl) saline but larger larvae than we used. Hence, we are left with the explanation that larval development results in a change in the midgut transport properties, as suggested by Chamberlin (1990). Tl+ is not secreted by the midgut (Fig. 1A), since it blocks the apical K+ pump. But even after ‘elimination’ of the apical membranes, wash-out of Tl+ is slow (Figs 3, 4A) and long-term exposure (for longer than 30 min) irreversibly increases the tissue resistance (see Schirmanns and Zeiske, 1994). Such non-specific actions of Tl+ as well as multiple cytoplasmic effects have also been described for other epithelia (Zeiske and Van Driessche, 1983, 1986; Benos et al. 1980), explaining why this heavy metal ion is called toxic.
Like Tl+, NH4+ is not capable of carrying an active current (Fig. 1B), and its reversibility is slow and often incomplete. A plausible explanation for this finding is that the well-known shunting of the H+ gradient by hydrolysis of NH4+ leads to a breakdown of active cation pumping, as has been shown in vesicle studies on midgut goblet cell apical membranes (Wieczorek et al. 1991).
Are there K+ channel subtypes?
In addition to Ba2+ blockade, Moffett and Koch (1991) observed K+ current inhibition in Manduca sexta midgut by lidocaine, and Alpert (1989) described a marked quinidine sensitivity. These organic blockers typically act on volume-activated K+ channels (Hille, 1992; Germann et al. 1986). On the basis of additive blocker effects, Moffett and Koch (1991) proposed the existence of two channel populations: one sensitive to Ba2+ and the other to lidocaine. However, because Ba2+ was used at concentrations far below those at which it has its maximal effect (cf. Figs 3 and 6 in the present paper and Fig. 1 in Zeiske et al. 1986), their conclusion no longer stands (see below). In our hands, the efficacy of 20 mmol l−1Ba2+ varied, for intact tissue, between 50% (Fig. 6) and 90% (Alpert, 1989; Zeiske et al. 1986). The present study showed almost complete K+ current block after ionophore treatment (Fig. 3). However, volume-dependent K+ channels are also usually blocked by Ba2+ (Hille, 1992; Germann et al. 1986). Indeed, manipulations that swell cells lead to a clearly Ba2+-blockable K+ current increase in Manduca midgut (Zeiske et al. 1990). Moffett and Lewis (1990) briefly reported patch-clamp data for midgut goblet cells indicating the existence of four basolateral K+ channel types, two of which were activated by Ba2+. Unfortunately, these data have not yet been published in detail. It is conceivable that in the midgut, as in many other tissues, a given transport state, and thus volume state, leads to unpredictable activity of different K+ channel populations. This could modify our view of the sometimes incomplete current inhibition by Ba2+ (but see below), whereas an evaluation of channel selectivity may not be affected, because the permeability ratios for different cations do not vary much among K+ channel subtypes (Hille, 1992; and see below).
Selectivity of basolateral K+ channels
Our model for the evaluation of PX/PK ratios is based on a single K+ channel population (at least with respect to selectivity and Ba2+ sensitivity). Indeed, taking into account our assumption of a poorly selective shunt (equation 12, Appendix 2), the data and theoretical expectations are in agreement (Fig. 5). Nevertheless, it cannot yet be ruled out that we have described a ‘lumped’ population of K+ channels that have similar selectivity characteristics.
The experiments using Tl+ showed that the absence of transepithelial transport of a cation by the midgut does not necessarily mean that the cation is basally impermeant (Fig. 3): after ionophore-induced perforation of the apical membranes, a concentration gradient can drive Tl+ through the basolateral channels at a rate similar to that for K+ (ITl/IK=0.92). This finding encouraged us to estimate ionic permeability ratios for the K+ channels. The apical membranes become electrically invisible after treatment with amphotericin B and TMT, allowing measurement of the reversal potentials of the basolateral membranes. However, this method requires circumvention of the problem of unknown parallel ‘shunt’ permeability to cations (Appendix 2). We assumed the shunt cation conductance to be small and of equal magnitude for all monovalent cations on the basis of several observations. (1) Basal addition of 32 mmol l−1Cs+ and symmetrical addition of 32 mmol l−1Na+ to 32K salines did not cause a significant change in the conductance of the intact tissue. The same was true for an apical increase of [K+ ] (from 32K(Cl) to 90K(Cl); N=4. (2) After electrical ‘elimination’ of the apical membranes, a change in [Na+ ]b hardly altered Erev and Rt (Fig. 4A); basal substitution of Na+ by the larger choline or N-methylglucammonium ions did not significantly affect Erev and Rt (N=2).
Although the Goldman current equation was linear, as would be expected for symmetrical [K+ ] after amphotericin treatment (Fig. 7A, control), it was not curved in the presence of a K+ gradient. The I–V linearity could be due to an erroneous correction for the shunt contribution, but the accurate description (Fig. 5) of the data given by the GHK voltage equation, as modified by our model, suggests that this is not the case. Ionic redistribution caused by lengthy voltage pulses (see Materials and methods) could explain the I–V linearity (Moffett, 1980). In addition, the decrease in conductance that was observed at apically positive potentials (see Fig. 7B and below) could contribute to the observed deviation from the expected Goldman behaviour.
Equation 13 fits the experimental relationship between Erev and [K+ ]b very well (Fig. 5). Consequently, the PX/PK ratios for the tested ions become calculable. Among these, only basal NH4+ did not generate a constant Erev; instead, it was biphasic (Fig. 4B). Although NH4+ seemed to permeate the K+ channels fairly well, the tissue resistance climbed steadily during basal exposure to NH4+. It seems likely, therefore, that the second, slow phase is due to secondary inhibitory effects. It is well known that cell acidification (e.g. caused by cytosolic NH4+ dissociation) blocks basal epithelial K+ channels (Harvey et al. 1988).
Gap-junctional coupling of goblet and columnar cells (Moffett and Koch, 1988) prevented us from allocating the K+ channels to one specific cell type or to both. The estimated permeability ratios yielded a sequence of: K+ (1) = Tl+ > Rb+ (0.38) > NH4+ (≈0.3) > Cs+ (0.03) > Na+ (≈0). The above sequence for the alkali metal ions K+, Rb+, Cs+, Na+ [Eisenman (1962) sequence IV] appears in almost all K+ channels so far investigated, including the basal channels of frog skin (Lindley and Hoshiko, 1964). Equally remarkable is the similarity to many K+ channels of excitable membranes, which display the same sequence, except that Tl+ is more permeant than K+ (reviewed by Hille, 1992). Consequently, with regard to the many different cations transported by the H. cecropia midgut (see Introduction), there is no evidence for the presence of poorly selective basolateral channels in the M. sexta midgut. Nevertheless, it is still possible that such channels become apparent when the basal medium is free of Mg2+ and Ca2+(Aelvoet et al. 1988), which would explain why the K+ current increases under the latter conditions (Moffett and Koch, 1983, 1985).
K+ channel block
Extracellular Cs+ (32 mmol l−1), Na+ (32 mmol l−1) or TEA+ (10 mmol l−1) do not inhibit IK, at least in the short-circuited midgut, which develops an intracellularly negative Vb of ‘only’ 32 mV (Chao et al. 1991). The insensitivity to Cs+ and TEA+ is rare compared with that of K+ channels in excitable (Hille, 1992) and epithelial (Van Driessche and Zeiske, 1985) tissues. Basal Ba2+, which seems to be a universal blocker of various types of K+ channels (Hille, 1992; Germann et al. 1986; Van Driessche and Zeiske, 1985), causes a marked hyperpolarization of Vb (Moffett and Koch, 1988) and clearly reduces the current through basolateral K+ channels, irrespective of whether K+, Rb+ or Tl+ is the permeant ion. Compared with K+, however, the blockade is more efficient when Rb+, and less efficient when Tl+, is present (Figs 1A, 3). Basolateral K+ channel block by Ba2+ is competitive (Zeiske et al. 1986), so the former effect probably results from the higher KRb, i.e. the lower binding affinity of Rb+ (Fig. 2A). The weaker block of Tl+ current in turn suggests a high binding affinity of Tl+ which, because of the ion’s strongly polarizable outer electron shell, seems to be characteristic (Zeiske and Van Driessche, 1983; Eisenman and Horn, 1983).
The incomplete inhibition of K+ current by Ba2+ displayed a Michaelis–Menten-like dose-dependence, both for the intact tissue (Fig. 6) and after electrical ‘elimination’ of the apical membranes (Fig. 7). I–V recordings conducted under the latter conditions revealed that Ba2+ block also depended on voltage (Fig. 7). Two features of the voltage-dependent change in tissue conductance are surprising. First, when Vt was apically (i.e. intracellularly) moderately positive, Gt (and thus the conductance of the ‘shunt’ and/or K+ channels) dropped in the absence of Ba2+, suggesting a voltage-dependent permeability. Second, in the presence of basal Ba2+, Gt rose at high positive and negative voltages and finally exceeded the control value. Since the effect is symmetrical, Ba2+ probably induces an extra current, which must be carried by symmetrically present ions other than Ba2+. We cannot yet confirm that anion or cation movements across shunt pathways are responsible for this effect. However, one intriguing possible explanation for this finding is that high voltages open Ba2+-activated K+ channels known to exist in the M. sexta midgut (Moffett and Lewis, 1990). The unusual behaviour during Ba2+ treatment may then also apply to the short-circuited intact tissue, where the blocker hyperpolarizes Vb to −70 mV (Moffett and Koch, 1988).
The, at first glance ‘simple’, dose–response relationship up to 12 mmol l−1Ba2+(Fig. 6B) would then be more complicated: the apparently blocker-insensitive 40% of active current (Fig. 6B) would be ‘induced’ in the presence of Ba2+. In contrast, at Vb=0 mV (in apically perforated tissue), the blockade of K+ current was much more complete (Fig. 3).
Blockers of volume-activated K+ channels were effective in the intact tissue, suggesting the existence of volume-dependent channels. Cell volume probably increases in the presence (Fig. 3) of nitrate, to which amphotericin channels are permeable (Schirmanns and Zeiske, 1994). The volume increase may then open (additional) lidocaine-and quinidine-sensitive pores (Hille, 1992; Germann et al. 1986). However, Schirmanns (1993) found that a complete Ba2+ inhibition of current is only rarely obtained when cell volume changes are putatively prevented by using gluconate as the major anion in the presence of apical amphotericin. Rather, it should be noted that the inhibition by Ba2+ of the K+ channel (as investigated with current-noise analysis; Zeiske et al. 1986) is reduced at [Ba2+] greater than 5 mmol l−1, which may explain some of the above observations. Consequently, the incomplete block of Isc by Ba2+ (this paper; Moffett and Koch, 1985; Zeiske et al. 1986) seems not to be due to the presence of a Ba2+-insensitive, and perhaps volume-activated, K+ channel population but instead may reflect an intrinsic property of a unique K+ channel type (Zeiske et al. 1986, 1992; Zeiske and Marin, 1992; Zeiske, 1992).
Woodhull’s model for voltage-dependent block (Appendix 1) is no more applicable at higher voltages: the apparent inhibition constant loses its voltage-dependence (Fig. 8) simultaneously with the Ba2+-induced increase in Gt. At moderate voltages, however, the model fits the measured relationship between Vt and . It yields a relative electrical distance δ (the fraction of Vb that drops between the binding site and the extracellular side) of 0.2 and a of 1.7 mmol l−1. The relative electrical distance may not exactly match the corresponding physical distance within a heterogeneous channel with funnels and narrow passages (Hille, 1992). Nevertheless, for a typical K+ channel, Ba2+ is, in our case, not a very ‘deep’ blocker, and its binding affinity lies in the medium or lower range. Two other epithelia may serve as examples. Since Ba2+ block is competitive, one has to take into account that in both cases applies to the ‘harder’ conditions of a K+ -rich medium. For the apical K+ channels of frog skin, δ is 0.34 whereas is 1.1 mmol l−1(at [K+ ]a=115 mmol l−1; Van Driessche and De Wolf, 1991). Basolateral K+ channel block in the locust rectum displays a very high δ of 0.76 and a K*Ba of 2–3 mmol l−1(at −8 to −26 mV and [K+ ]b=80 mmol l−1; Hanrahan et al. 1986).
Ion occupation and energy profile of the K+ channel
If a channel contains several ions at a time, these ions will interact while passing in single file. One way to detect presumed mutual ionic repulsions on the macroscopic level is to analyse the current carried simultaneously by two ions of different binding site affinity (Zeiske and Van Driessche, 1983; De Wolf, 1989). In multi-ion pores, Isc goes through a minimum or maximum as a function of the ratio of ionic concentrations. Another characteristic of multi-occupancy is an unusually high voltage-dependence of blocking ions (Eisenman and Horn, 1983). In our case, however, the observed dependence of Isc on varied K+ /Rb+ mixtures is not anomalous, but is in reasonable agreement with the prediction of a one-site channel model (Fig. 2B). Furthermore, the voltage-dependence of Ba2+ block is, as stated above, quite low. Consequently, there is no evidence for multiple ion occupation in the basolateral midgut K+ channels, at least up to a total cation concentration of 32 mmol l−1. This is surprising, as most K+ channels are multi-ion pores (usual [K+ ] is approximately 100 mmol l−1) and many of them can hold more than one Ba2+ at a time (Latorre and Miller, 1983; Hille, 1992). One of the other rare exceptions, in which no multi-ion characteristics have yet been found, is the normal (not the swelling-induced) basolateral K+ channel of the turtle colon (Germann et al. 1986).
This approach only applies to the lowest and the highest energy levels inside a one-ion channel, but it remains valid in the presence of other, flatter extrema (Hille, 1992). Although the above model is a first working hypothesis, it fits the results of this study and those of a previous finding: Zeiske et al. (1992) reported that Cl− stimulates active K+ current in the M. sexta midgut through a proportional rise in KK and Imax,K. A possible explanation for this effect is that Cl− increases the energy level of the external binding site for K+, but does not change Wpeak at the selectivity filter and thus PK remains the same.
ACKNOWLEDGEMENTS
The authors wish to thank Professor H. Wieczorek for providing lepidopteran eggs and for measuring K+ chemical activities of the solutions and Dr H. Onken for reading the manuscript critically.