Amiloride Analog Stimulation of Short-Circuit Current in Larval Frog Skin Epithelium

Recently, it has been established that pharmacologically distinct populations of amiloride-sensitive transport processes exist and are often preferentially expressed in different tissues (Table 1). They differ in ionic selectivity and in their relative affinities for amiloride analogs. Epithelial Na⁺ channels have been classified into two groups (Benos et al. 1992; Palmer, 1992). H-type channels are highly selective for Na⁺ and have a high affinity for amiloride, with an inhibition constant (K_i) in the nanomolar range. L-type channels are less selective for Na⁺, with K_i values for amiloride in the micromolar range. For the H channel, the affinity hierarchy is benzamil > amiloride > 5-(N,N-dimethyl)-amiloride (DMA), 5-(N-ethyl-N-isopropyl)-amiloride (EIPA). Although less well characterized, the profile for L channels seems to be EIPA > amiloride > phenamil. This may change as additional low-affinity channels are characterized. Amiloride analog sensitivity has also been used to characterize transducer, stretch-activated and dihydropyridine-sensitive L-type Ca²⁺ channels. In addition, distinct pharmacological profiles have been established for amiloride analog block of Na⁺/Ca²⁺ and Na⁺/H⁺ exchangers (Kaczorowski and Garcia, 1992; Kleyman and Cragoe, 1988). During metamorphosis, major changes occur in almost all organ systems of the amphibian. Previous studies have shown that ion uptake shifts from the gills in the larva to the skin in the adult (Dietz and Alvarado, 1974). Ion uptake across the ventral skin changes from an amiloride-stimulated non-selective monovalent cation transport process in the tadpole, which normally appears to be inactive, to an amiloride-inhibited, highly Na⁺-selective process in the adult (Cox and Alvarado, 1979; Alvarado and Cox, 1985; Hillyard and Van Driessche, 1989; Hillyard, et al. 1982). Whether the tadpole and adult channels are closely related or represent the expression of unrelated genes is at present unknown. However, antibodies to the adult Na⁺ channel cross-react with 65 and 150 kDa proteins in the tadpole skin (Hillyard and Benos, 1993), suggesting that there may be common epitopes on channel-related proteins in larval and adult skin.

Recent studies have identified several ligands that bind apical membrane receptors and regulate ion transport in various epithelia (Hwang et al. 1996; Wilson et al. 1996) including larval frog skin (Cox, 1992, 1993, 1997; Hillyard and Van Driessche, 1989). The purpose of the present study was to characterize further an apically regulated transport system in the bullfrog tadpole skin. I have used analogs with hydrophobic substitutions at either end of the amiloride molecule to help determine the nature of the ligand–receptor interaction. I show that amiloride analogs activate and apparently inhibit the I_sc in tadpole skin. In addition, the efficacy of the response appears to be related to the lipophilicity of the analog.

Rana catesbeiana tadpoles (stages XV–XVIII) were collected locally or purchased from commercial suppliers. Tadpoles were anesthetized in 0.1 % tricaine methane sulfonate buffered with 0.1 % sodium bicarbonate. The composite skin was dissected and mounted in modified Ussing chambers. Short-circuit current (I_sc) was measured, and solutions were changed as previously described (Cox, 1992). The surface area of the chamber was 0.7 cm². I_sc was sampled at 10 Hz using an IBM-compatible computer-based data-acquisition system and is reported as μA cm⁻². Baseline I_sc was typically less than 1 μA. The KCl Ringer contained (in mmol l⁻¹): 100 KCl, 2.0 CaCl₂ and 2.4 KHCO₃, pH 7.4. Ca²⁺-free KCl Ringer was made by omitting CaCl₂ and adding 0.5 mmol l⁻¹ EGTA. NaCl Ringer (same as above with 100 mmol l⁻¹ Na⁺ substituted for 100 mmol l⁻¹ K⁺) was maintained in the basolateral solution. K₂SO₄ Ringer was made with 56 mmol l⁻¹ K₂SO₄, 2.4 mmol l⁻¹ KHCO₃ and 1.2 mmol l⁻¹ CaSO₄. Na₂SO₄ Ringer was made by substituting Na₂SO₄ for K₂SO₄.

Ca²⁺-free KCl Ringer was periodically flushed into the apical solution just prior to and during drug addition. In some experiments, Ca²⁺-free NaCl or sulfate Ringer was used instead of KCl Ringer. Since prolonged incubation of the apical surface with Ca²⁺-free Ringer destroyed the integrity of the epithelium, KCl or NaCl Ringer was used for drug washout. Amiloride was dissolved directly in Ringer’s solution or in a distilled water stock solution for later addition. Nystatin and the analogs of amiloride were first dissolved in a methanol or dimethyl sulfoxide stock solution and added to the Ringer’s solution. The final solvent concentration never exceeded 0.5 %. The solvent itself at these concentrations had no effect on basal or amiloride-stimulated I_sc. All solutions were flushed directly into the apical chamber and removed from the top by vacuum with no interruption in I_sc (Cox, 1992).

Values are expressed as mean ± S.E.M. (N). Differences between means were established using Student’s t-test. Results were considered significant at the 95 % confidence level. Amiloride was a gift from Merck Sharp & Dohme, West Point, PA, USA. The amiloride analogs and N-(6 aminohexyl)-5-chloro-1-naphthalene sulfonamide (W-7) were purchased from Research Biochemicals, Natick, MA, USA. Nystatin and all other reagents were purchased from Sigma, St Louis, MO, USA.

It has previously been shown that amiloride stimulates short-circuit current across larval frog skin when it is added to the apical solution. The response is larger in a Ca²⁺-free KCl Ringer, suggesting that amiloride and Ca²⁺ may compete for a negatively charged binding site (Cox, 1979, 1992; Hillyard and Van Driessche, 1989). To determine which portion of the amiloride molecule might be involved in binding to the tadpole receptor, I tried several amiloride analogs that had substitutions on the pyrazine ring and at the guanidino end. Their names and structures are shown in Fig. 1. In normal NaCl Ringer, I_sc stimulation by these analogs was small at concentrations up to 100 μmol l⁻¹, so Ca²⁺-free KCl Ringer was used. As previously shown for amiloride (Cox, 1992; Hillyard and Van Driessche, 1989), the responses were larger in this solution, and therefore most experiments were carried out in Ca²⁺-free KCl Ringer.

Fig. 2 (top trace) shows typical responses over a range of amiloride concentrations in Ca²⁺-free KCl Ringer. Results were similar to, but larger than, those previously reported using KCl Ringer containing 2 mmol l⁻¹ Ca²⁺ (Cox, 1992; Hillyard and Van Driessche, 1989). The initial small rise in I_sc recorded just prior to amiloride addition was due to perfusion with Ca²⁺-free Ringer. This is more clearly shown in the inset of Fig. 2 and in Figs 5 and 7. An abrupt increase in I_sc occurred after the addition of various concentrations of amiloride. After 30–40 s, KCl Ringer (with Ca²⁺) was used for washout. The solid vertical bar indicates a 5–10 min recovery period (KCl Ringer outside) between amiloride additions. At 1000 μmol l⁻¹, there was a rapid increase in I_sc to 40 μA followed by a relaxation towards zero over several seconds. At 1 μmol l⁻¹, there was a considerably smaller increase with a much slower rate of decay. At 10, 50 and 100 μmol l⁻¹, the maximum peak response increased and the apparent rate of decay also increased. At 500 μmol l⁻¹, the peak was smaller than either value at 50 or 100 μmol l⁻¹. Data for several experiments are summarized in Fig. 3. The data in Fig. 3 were fitted to equation 1. The peak of the fitted curve occurred at 107 μmol l⁻¹. The constants for activation and block by amiloride were 12.8 and 918 μmol l⁻¹, respectively. The maximum response determined from the fit was 129 % of the value at 100 μmol l⁻¹ amiloride.

As mentioned above, the response decayed more rapidly as the concentration was increased. To quantify this, decay curves were fitted to equation 2 over a defined relaxation range (peak of response to washout). This gives a good relative index of the rate of decay (τ) of I_sc after stimulation. As before, the data were normalized to values obtained at 100 μmol l⁻¹ amiloride in each skin. Data for several experiments are summarized in Fig. 4: τ decreased as amiloride concentration increased.

Benzamil has a very high affinity for the Na⁺ channel in adult frog skin (Benos et al. 1992; Kleyman and Cragoe, 1988; Simchowitz et al. 1992). Block is almost complete at 1 μmol l⁻¹ and is only partially reversible. The enhanced blocking effect appears to result from the benzene ring at the guanidino end of the molecule (Fig. 1). From these results, I suspected that benzamil might have a relatively high affinity for the tadpole channel. As shown in Fig. 2, at 1–50 μmol l⁻¹, the benzamil response was similar in time course to that for amiloride. At 100 μmol l⁻¹, there was a smaller peak response and a more rapid decline in I_sc. Fig. 3 presents a summary of several experiments of this type. The maximum value of the fitted curve occurred at 5 μmol l⁻¹, which was an order of magnitude lower than the maximum for amiloride. The constants for activation and inhibition were 0.6 and 64 μmol l⁻¹, respectively. The decay constants were significantly smaller than those for amiloride at all concentrations (Fig. 4).

Next, I considered substitutions on the pyrazine ring. DMA has two methyl groups on the nitrogen at position 5 (Fig. 1) and a relatively low affinity for the adult channel (Benos et al. 1992; Kleyman and Cragoe, 1988; Simchowitz et al. 1992). As shown in Fig. 2 for the range 10–100 μmol l⁻¹, the time course of the increase and decline in I_sc was similar to that for amiloride. The response peaked at 52 μmol l⁻¹, approximately 60 % of that for the amiloride response (Fig. 3). At 500 μmol l⁻¹, only a small spike remained. At the lower concentrations (1 μmol l⁻¹), there was a variable response. In two skins, there was a slow decline after an initial peak. In three skins, there was a slow increase. An example of this is shown in Fig. 2. The constants for activation and inhibition were 12 and 230 μmol l⁻¹, respectively. The decay constant was significantly smaller than that for amiloride at all concentrations greater than 50 μmol l⁻¹. The decay time was particularly short at 500 μmol l⁻¹ (Fig. 4).

Larger hydrophobic substitutions at nitrogen at position 5 result in the ability to block the L-type Na⁺ channel, but reduce the affinity for the adult Na⁺ channel (Kleyman and Cragoe, 1988; Matalon et al. 1993; Moran et al. 1988; Simchowitz et al. 1992). I suspected that the affinity would also be low in the tadpole skin. As shown in Fig. 2, EIPA (10 μmol l⁻¹) gave a response with a time course very similar to that for amiloride but with a substantially lower amplitude. At 50 μmol l⁻¹ and above, the response turned on and off very quickly, resulting in an almost spike-like behavior and a low maximum response. On average, the maximum peak response occurred at 12 μmol l⁻¹ (Figs 2, 3). Note also that the I_sc decreases to a level below that occurring in Ca²⁺-free Ringer alone, which is consistent with inhibition (see the inset in Fig. 2). At lower concentrations (1 μmol l⁻¹), there was a relatively slow decline in I_sc after stimulation. The constants for activation and inhibition were 2.8 and 53 μmol l⁻¹, respectively. Decay data are summarized in Fig. 4.

MIBA is another analog of amiloride with a relatively large hydrophobic group at position 5. It also has a very low affinity for the adult channel (Kleyman and Cragoe, 1988; Simchowitz et al. 1992). In the tadpole, the I_sc response to MIBA was similar to that to EIPA (data not shown). The response was maximal at 12 μmol l⁻¹. The constants for activation and inhibition were 0.8 and 220 μmol l⁻¹, respectively. Data are summarized in Figs 3 and 4. As with EIPA, I_sc was rapidly reduced by MIBA to less than the value in Ca²⁺-free Ringer.

Previous studies of amiloride block of L-type Ca²⁺ channels and exchangers have shown that increased lipophilicity is correlated with more effective inhibition (Garcia et al. 1990; Kleyman and Cragoe, 1988; Simchowitz et al. 1992; Slaughter et al. 1988). The relationship between lipophilicity and decay was examined at 10 and 100 μmol l⁻¹ for analogs used in the present study. Lipophilicity was defined as the percentage distribution of drug between a lipid and aqueous phase (Kleyman and Cragoe, 1988). As the lipophilicity increased (Fig. 4, inset), τ decreased at both concentrations examined. The correlations were significant. This appeared to be independent of where on the amiloride molecule the lipophilic groups were placed.

Previous studies have shown that W-7 almost completely blocks the amiloride response in tadpole skin (Cox, 1992). I tested the ability of W-7 to block stimulation by amiloride analogs. In Fig. 5, an amiloride test pulse was followed by benzamil treatment. After washout of benzamil, W-7 was added. There was an immediate small increase in I_sc followed by a sharp decrease to levels below that observed in Ca²⁺-free Ringer. Subsequent addition of benzamil in the presence of W-7 had no effect. Upon washout, the benzamil and amiloride responses were restored. Values were summarized as the ratio (percentage) of I_sc in the presence of analog plus W-7 to the I_sc in the presence of the analog alone. Values were 19.1±9.1 % for benzamil, 5.8±1.9 % for EIPA, 1.7±0.8 % for DMA and 3.7±2.7 % for MIBA. The block was almost complete in all cases.

To test for interactions between the analogs and amiloride, a test pulse of amiloride (100 μmol l⁻¹) was given. The amiloride was washed out and followed by EIPA (100 μmol l⁻¹). Amiloride (100 μmol l⁻¹) was then added while maintaining the EIPA concentration. Pretreatment with EIPA reduced the amiloride response to 13.0±2.8 % of the response with amiloride alone. This is consistent with amiloride and EIPA interacting with the same process.

In high-resistance Na⁺-transporting epithelia, amiloride inhibits I_sc by blocking channels in the apical membrane (Palmer, 1992; Kleyman and Cragoe, 1988; Sariban-Sohraby and Benos, 1986). It has been argued in previous papers that, in tadpole skin, the primary effect of amiloride is to stimulate nonselective cation channels in the apical membrane (Cox, 1992; Hillyard and Van Driessche, 1989; Hillyard et al. 1982). Since amiloride and its analogs penetrate cell membranes, it is conceivable that the decrease in I_sc after analog stimulation is due to unknown effects that may increase the resistance of the basolateral membrane. One way to test for this is effectively to remove the apical membrane resistance using the polyene antibiotic nystatin (Cox and Alvarado, 1983). The effect of subsequent addition of amiloride analogs could reasonably be attributed to effects on the basolateral membrane or possibly the shunt pathway. A complication to this approach comes from the observation that polyene antibiotics may cause cell swelling which may, in turn, activate new conductances at the basolateral membrane (Germann et al. 1986; Hillyard and Van Driessche, 1992). These changes at the basolateral membrane can largely be eliminated by using non-permeant anions and hypertonic solutions.

Tadpole skins were mounted in chambers with Na₂SO₄ Ringer on the basolateral side and K₂SO₄ Ringer on the apical side. The effects of amiloride, benzamil and EIPA were tested before and after removing apical membrane resistance by adding 250 i.u. ml⁻¹ nystatin to the apical side. Experiments were performed with and without 100 mmol l⁻¹ sucrose on both sides of the skin. Typical experiments (with sucrose) are shown in Fig. 6. Amiloride, benzamil and EIPA had similar effects on I_sc when the skin was bathed in sulfate Ringer compared with chloride Ringer prior to nystatin treatment. Nystatin caused an increase in I_sc that stabilized after 30–60 s. Nystatin was perfused a second time to establish the nature of any perfusion artifact. A second perfusion of nystatin with the analog showed little or no effect on I_sc. Values were summarized as the ratio of the change in I_sc caused by the analog after nystatin treatment to the analog effect prior to nystatin treatment. For amiloride, the ratio was 0.06+0.04 using sulfate Ringer and 0.10±0.05 using sulfate Ringer with sucrose. For benzamil, values were 0.03±0.03 without and 0.11±0.03 with sucrose. For EIPA, values were 0.25±0.07 without and 0.10±0.06 with sucrose. For the experiments with sucrose, the transepithelial resistance was determined before and after adding the analog. Values are reported as the ratio of the resistance in the presence of the analog and nystatin to the resistance in the presence of nystatin immediately prior to analog addition. Values for amiloride, benzamil and EIPA were 0.96±0.08, 1.00±0.08 and 1.13±0.06, respectively, for five or more experiments each. None of these values was significantly different from 1.

To determine whether the amiloride analog response might be dependent on K⁺ in the apical solution, several experiments were performed with NaCl Ringer on both sides of the skin. Skins were perfused with Ca²⁺-free NaCl Ringer on the apical side followed by 100 μmol l⁻¹ amiloride. After amiloride washout, the process was repeated with either benzamil or EIPA (both 10 μmol l⁻¹). The protocol was repeated with KCl Ringer in the apical solution (see Fig. 7 for an example using benzamil). In this experiment, the relaxation in response to benzamil in NaCl Ringer was somewhat slower than that observed in KCl Ringer. On average, the time courses of the responses were similar in NaCl and KCl Ringer. The ratios of τ determined in KCl Ringer to τ in KCl Ringer for amiloride, EIPA and benzamil were 0.88±0.08 (N=16), 0.92±0.18 (N=6) and 1.06±0.11 (N=10), respectively. These values were not significantly different from 1.

Recent pharmacological and biochemical studies of amiloride-sensitive ion channels have shown that there is a wide variety of channel types found in many tissues (Benos et al. 1992; Canessa, 1996; Kleyman and Cragoe, 1988; Palmer, 1992; Simchowitz et al. 1992; Vigne et al. 1989). Many of these channels (including those found in the adult frog skin) were initially characterized according to their sensitivity to amiloride analogs. In contrast to the adult frog skin and most other amiloride-sensitive transport processes, the skin of the larval frog has an apical membrane channel that is activated by amiloride (Cox, 1992; Cox and Alvarado, 1979, 1983; Hillyard and Van Driessche, 1989; Hillyard et al. 1982). Little information is available as to how amiloride receptors in the larval frog skin compare with amiloride receptors in other transport systems. The purpose of the present experiments was to compare the effects of various analogs of amiloride on I_sc in frog tadpole skin. To summarize, the present studies have shown that amiloride analogs activate and inhibit I_sc in larval frog skin. Highly lipophilic analogs stimulated and inhibited I_sc more effectively than the more hydrophilic amiloride.

Amiloride is believed to bind to apical receptors and to promote the entry of monovalent cations into the cell across the apical membrane (Cox and Alvarado, 1979; Cox, 1992; Hillyard and Van Driessche, 1989; Hillyard et al. 1982). At present, I do not know whether activation and inhibition occur by direct effects on the channel or through separate, distinct processes. The initial rise in current is fast and may be too rapid and reversible to be due to activation of second-messenger systems or to binding of the drug to a basolateral membrane receptor (Cox, 1992, 1993). Elucidation of the precise role of second messengers in the amiloride response requires further study. However, as suggested by Hillyard and Van Driessche (1989), I will consider my results in terms of a receptor-operated channel at the apical membrane that is activated and desensitizes and/or is inhibited by amiloride analogs. The noise studies have established that amiloride stimulates a channel at the apical membrane.

I considered the possibility that amiloride analogs had significant effects on the basolateral membrane. Nystatin can be used to remove apical membrane resistance and, when nystatin is added using chloride Ringer, it has been shown to induce new transport pathways at the basolateral membrane, presumably because of cell swelling (Germann et al. 1986; Hillyard and Van Driessche, 1992). The new K⁺ conductance in tadpole skin can be blocked by quinidine and verapamil or eliminated if the skin is treated with 100 mmol l⁻¹ sucrose on both sides. Similar conditions were used in the present study to test for the effects of amiloride analogs on the basolateral membrane. Amiloride analogs had no measurable effect on either current or conductance after nystatin treatment (Fig. 6). Certainly, the nystatin-treated skin is not the same as the untreated skin, and the possibility of additional nystatin-induced changes that are not prevented by hyperosmotic solutions cannot be eliminated. However, the data are consistent with amiloride analogs having little or no effect at the basolateral membrane.

Prolonged incubation of epithelia in Ca²⁺-free solutions causes the breakdown of the shunt pathway (Contreras et al. 1992). Therefore, the responses observed in the present study (particularly in KCl Ringer) could be a result of increases in the conductance of the shunt pathway. While there is no way to measure the shunt resistance directly, shunt currents can be eliminated by using symmetrical NaCl solutions. The results shown in Fig. 7 demonstrate that the time courses in NaCl and KCl Ringer’s solutions for I_sc stimulation by benzamil were similar. Comparable results were obtained for EIPA. These results support the idea that transients in the shunt pathway resistance are not responsible for the time course of the I_sc response to amiloride analogs.

EIPA, MIBA and DMA have significant blocking effects on various ion transporters in other tissues (Garcia et al. 1990; Kaczorowski and Garcia, 1992; Kleyman and Cragoe, 1987; Lane et al. 1992). With regard to Na⁺/H⁺ exchange, at the time of analog addition in the KCl experiments, there was no Na⁺ in the apical solution. This eliminated an inward Na⁺ gradient across the apical membrane that would drive Na⁺/H⁺ exchange. When Na⁺ was present in the apical solution, the time courses of the responses were similar. In addition, benzamil gave a similar pattern of I_sc stimulation and inhibition to EIPA and MIBA. Benzamil has a very low affinity for the Na⁺/H⁺ exchanger (Kleyman and Cragoe, 1987; Lane et al. 1992). With regard to Na⁺/Ca²⁺ exchange, in the KCl experiments, Na⁺ and Ca²⁺ were eliminated from the apical solution at the time of analog addition, effectively eliminating Na⁺/Ca²⁺ exchange at the apical membrane. However, the possibility of amiloride analog effects on these exchangers (perhaps at the basolateral membrane) that result in modulation of the I_sc response cannot at present be totally ruled out.

The anticalmodulin drug W-7 specifically blocks voltage-dependent Ca²⁺ channels in Paramecium (Ehrlich et al. 1986; Hennessey and Kung, 1984) through a mechanism independent of calmodulin. W-7 has been shown to block amiloride stimulation of I_sc in tadpole skin (Cox, 1992). In the present study, W-7 at high concentrations blocked the stimulatory effects of all the analogs tested. The existence of a blocker common to all these analogs lends support to our implicit assumption they all work on the same transport path. Further support for this idea comes from the study where pretreatment with EIPA blocked the amiloride effect. While the simplest explanation for the action of W-7 is to assume that it has a direct effect on ion channels, as it does in Paramecium, I cannot at present eliminate the possibility that it has effects on calmodulin. It does block stimulation by all the ligands tested on tadpole skin to date (Cox, 1992, 1993, 1997).

The initial increase in I_sc upon exposure to amiloride was very rapid in all cases where the concentration was greater than 1 μmol l⁻¹, reaching a peak in 1–2 s. This is about as fast as solution can be perfused into the chamber (Cox, 1992). Similar responses were observed for the other analogs. This is consistent with a relatively high-affinity activation step. The rate of decay at 1 μmol l⁻¹ was slow or, in some cases, there was a small time-dependent increase in I_sc after the initial response. The simplest interpretation of these data is that there is a relatively high-affinity activation process and a lower-affinity inhibition. Presumably, the inhibiting effect is minimal at low concentrations. These results are consistent with the mechanism proposed by Hillyard and Van Driessche (1989).

Interpreting the maximum value of amiloride stimulation and the rate of decline of I_sc in tadpole skin is complex. The magnitude of I_sc is determined by the resistances of the apical and basolateral membranes as well as by the concentration gradients for Na⁺ and K⁺. With an increase in I_sc, there may be a build-up of K⁺ in the extracellular spaces on the basolateral side. This would lead to a decreased chemical gradient for K⁺ exit at the basolateral membrane and a decreased electrical gradient for cation entry at the apical membrane. Depolarization could also cause unknown effects on membrane resistances that might affect the current transient. However, nystatin also increases transport of Na⁺ or K⁺ without a transient in I_sc comparable with that induced by amiloride analogs, resulting in sustained currents that are much higher than the current observed after amiloride analog treatment (Cox and Alvarado, 1983).

With Na⁺ outside, an increase in I_sc may lead to a build-up of Na⁺ in the cell. This would decrease the gradient for Na⁺ entry, perhaps resulting in a decrease in I_sc. However, comparable transients are observed with K⁺ in the apical solution, arguing against this possibility. In addition, after the apical membrane resistance had been removed with nystatin, no effects of amiloride analogs were observed on either current or resistance. Therefore, changes in conductance at the apical membrane probably account for the majority of the transient in I_sc. Many ligand-activated channels desensitize by as yet poorly characterized mechanisms. Significant desensitization may occur with the tadpole channel. In addition, amiloride and its analogs are known to block most types of transporters. In the present study, the rate of decrease in I_sc was concentration-dependent, and in some cases the decreases were to below pretreatment values, which is consistent with inhibition of transport.

The peak response of I_sc may be severely attenuated if a significant block occurs prior to complete activation of transport. Amiloride is thought to interact directly with the tadpole channel to give the initial rise in I_sc (Cox, 1992; Cox and Alvarado, 1979; Hillyard and Van Driessche, 1989; Hillyard et al. 1982). If the effect is directly on the channels, allosteric binding would promote a conformational change, causing the channel to open. As previously suggested, the positive charge on the nitrogen at the guanidino end of amiloride may interact with a negatively charged receptor (Hillyard and Van Driessche, 1989). This is consistent with the larger responses observed in Ca²⁺-free Ringer. However, it is possible that Ca²⁺ also partially blocks the channel directly in addition to shielding a receptor site (Alvarado and Cox, 1985; Hillyard and Van Driessche, 1989). The data in Fig. 3 suggest that the activation sequence is benzamil = MIBA > EIPA > DMA = amiloride. Assuming that amiloride acts primarily on the apical membrane conductance, as suggested above, we can consider the decay after amiloride stimulation in terms of desensitization or amiloride inhibition of the transport process. If the channel desensitizes, the rate of decline of Isc would be independent of amiloride concentration. If amiloride inhibition is occurring, the rate of decline of I_sc would increase with increasing amiloride concentration. As shown in Fig. 2 and summarized in Fig. 4, the decay constant decreased slightly as amiloride concentration was increased. This is consistent with a relatively low-affinity block by amiloride (Hillyard and Van Driessche, 1989). The decay constants for the other analogs (Figs 2, 4) showed a greater dependence on concentration. Following this reasoning, a small decay constant would indicate a higher affinity of the amiloride analog for a binding site. From the data in Fig. 4, the relative inhibiting efficacy would be EIPA = MIBA > benzamil > DMA > amiloride. The fact that MIBA and EIPA at high concentrations rapidly decreased I_sc to values below that in Ca²⁺-free Ringer (Fig. 2) is consistent with a blocking effect. For MIBA and EIPA, the rate of decline at high concentrations was equal to, or more rapid than, the activation response. This could lead to block of a large number of channels before others have been activated. As mentioned above, a rapid block would significantly attenuate the maximum observed I_sc stimulation, decreasing apparent efficacy (Cox, 1992; Hillyard and Van Driessche, 1989).

Amiloride stimulation has a time course similar to the Na⁺ self-inhibition transient in I_sc seen in adult skin after Na⁺ restoration (Zeiske and Lindemann, 1974; Lindemann and Van Driessche, 1978; Van Driessche and Lindemann, 1979). An attractive hypothesis is that amiloride analogs also undergo a similar transient, with activation by amiloride at lower concentrations and inhibition at higher concentrations. If amiloride analogs partition into the membrane, they can perhaps be considered to float in the bilayer until they collide and inhibit an open channel.

Results from studies of amiloride analogs on other transporters may offer some insight into the present work. On the Na⁺/Ca²⁺ exchanger, the potency of amiloride analogs increases with increasing hydrophobicity of the guanidino nitrogen substituent. Benzamil is also an effective inhibitor of Na⁺/Ca²⁺ exchange. The analogs are thought to partition into the membrane and diffuse to the blocking site, resulting in a relatively high local concentration (Simchowitz et al. 1992; Slaughter et al. 1988). With regard to Na⁺/H⁺ exchange, the most potent inhibitors are analogs with hydrophobic substituents at the 5 position of the pyrazine ring (Kleyman and Cragoe, 1988; Simchowitz et al. 1992). Kaczorowski and Garcia (1992) suggest that amiloride analogs bind to L-type Ca²⁺ channels in an allosteric fashion. Binding of MIBA and EIPA significantly increases the rate of dissociation of nitrendipine, D-600 and diltiazem. The most effective blockers are lipophilic and are thought to partition into the membrane (Garcia et al. 1990). The results of the present study show a significant correlation between lipophilicity and the decay constant for I_sc (Fig. 4). As suggested above for some of the other transporters, hydrophobic substituents on amiloride analogs may partition into the membrane, promoting activation and/or inhibition.

Amiloride-sensitive Na⁺ transport has been divided into two categories according to the relative affinity of amiloride block (see also Table 1). The K_i for amiloride block of H channels is of the order of 50–100 nmol l⁻¹ (Benos et al. 1992; Palmer, 1992). In addition, high-affinity amiloride-sensitive Na⁺ channels have been cloned from several sources (Canessa, 1996; Canessa et al. 1994). The relationship between the cloned channel and the purified bovine renal Na⁺ channel has not yet been established (Benos et al. 1995; Canessa, 1996). The K_i for block of L-type Na⁺ channels is in the range 1–100 μmol l⁻¹ (Benos et al. 1992, 1995; Matalon et al. 1993; Moran et al. 1988; Vigne et al. 1989). Although there are diverse pharmacological profiles among different cell types, EIPA has generally been shown to be a more effective blocker than amiloride and is thought to block the channels directly. The amiloride receptor in tadpole skin would appear to have more in common with the L-type rather than the H-type channel. Although L-type Ca²⁺ channels can pass monovalent cations in the absence of Ca²⁺, the tadpole channel appears to be quite different in that it is activated by the Ca²⁺ channel blocker diltiazem (Cox, 1992) but it is not activated by dihydropyridines (T. Cox, unpublished observations).

Some analogs of amiloride stimulate I_sc in adult frog skin and toad urinary bladder. DMA and benzoylimidazole-2-guanidine have been shown to stimulate rather than inhibit transport in both adult (Li et al. 1987; Zeiske and Lindemann, 1974) and tadpole (Cox and Alvarado, 1979; Hillyard et al. 1982). In the adult, the response has a relatively slow onset and reaches a new steady state, unlike the transient stimulation observed in tadpole skin (present study; Cox, 1992; Hillyard and Van Driessche, 1989).

Amiloride analogs have been shown to block several other types of ion transporters (see Table 1). There is a wide range of pharmacological profiles, indicating diversity in receptor types. Amiloride analog interaction with the tadpole receptor does not exactly match that of any of the above transporters. Since amiloride analogs bind to several transporters with varying affinities, the amiloride analog profile does not provide a highly sensitive probe for channel characterization. Complete characterization of the relationships between the transporters will require a knowledge of their respective protein structures.

In conclusion, the results presented here describe a cation transport system in frog tadpole skin that is differentially stimulated by analogs of amiloride. Analogs with substitutions on the guanidino end, such as benzamil (very high affinity for the H form of the adult channel), activated at the lowest concentrations. The magnitude of the response was small, suggesting a moderate-affinity inhibition effect. Analogs with relatively large hydrophobic groups at the 5 site of the pyrazine ring, such as MIBA and EIPA, activated Isc at relatively low concentrations. At high concentrations (50–100 μmol l⁻¹), the maximum response was small and the rate of decay was fast, which is consistent with a relatively high-affinity inhibition effect.

A physiological role for the amiloride-activated transport system in tadpole has yet to be established. In tadpoles, Na⁺ uptake from pond water was not detectable across the skin. Under normal conditions, ion uptake is achieved by feeding and by transport across the gills (Dietz and Alvarado, 1974). Whether amiloride analogs would promote ion uptake from pond water (typically less than 1 mmol l⁻¹ Na⁺) has not been tested. Amiloride-sensitive Na⁺ channels are involved in the sense of taste (Ye et al. 1991), and von Seckendorff Hoff and Hillyard (1993) have suggested that adult toads can taste salt with their skin via amiloride-sensitive mechanisms. In addition, it has previously been shown that, in addition to amiloride, ligands such as acetylcholine and ATP also activate the tadpole channel (Cox, 1993, 1997). These responses are inhibited by a wide variety of blockers. Ligand binding at the apical membrane may turn out to be an important site for regulation of transepithelial transport.

Csaba (1992) has described primitive plasma membrane receptors present in other systems. He suggests that receptors that interface directly with a changing environment need to be fairly non-selective and to respond to a large variety of molecules. The tadpole amiloride receptor may have characteristics in common with these primitive plasma membrane receptors.

The excellent technical assistance of Dee Gates and Greg Schandelmeier is gratefully acknowledged. Supported by a grant from the American Heart Association, Illinois Affiliate.