INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

In experiments designed to study hormonal regulation of Na⁺ transport in epithelial tissues, the activation of chloride transport in addition to sodium transport imposes complexities in the design of experiments and interpretation of data unlike those encountered in studies of frog skin and toad urinary bladder where adenosine 3',5'-cyclic monophosphate (cAMP) selectively activates apical membrane amiloride-sensitive epithelial Na⁺ channels (ENaCs) (Els and Helman, 1981, 1997; Schlondorff and Satriano, 1985). To understand the complexities to be encountered in studies of cell-cultured A6 epithelia where cAMP activates both Na⁺ and Cl channels, we turned to noninvasive methods of impedance analysis to determine the time course and magnitude of change of apical membrane conductance to chloride and apical membrane capacitance in response to elevation of intracellular cAMP by forskolin and prostaglandin E₂ (PGE₂) that are known to elevate cAMP in target tissues (Chalfant et al., 1993; Hall et al., 1976; Sonnenburg and Smith, 1988; Yanase and Handler, 1986; Noland et al., 1992).

Although the measurement of transepithelial impedance is in principle straightforward, previous studies from our laboratory had indicated that the dielectric properties of epithelial plasma membranes may exhibit -dielectric dispersions at audio frequencies (Awayda et al., 1999; Liu and Helman, 1998; Punescu and Helman, 2000) that complicate measurement of plasma membrane capacitance. The theoretical principles and considerations relevant to the present studies have been discussed elsewhere (Punescu and Helman, 2001) where we examined the contributions of Maxwell-Wagner-like and Cole-Cole dielectric dispersions to the transepithelial impedance locus of the series arrangement of apical and basolateral membrane impedances that are paralleled electrically by paracellular shunt resistances.

Evaluation of the time-dependent changes of short-circuit current and transepithelial impedance has led us to conclude that PGE₂ maximally activates a rather large apical membrane chloride conductance in <1 min. Activation of chloride channels preceded activation of the apical membrane sodium conductance from basal states where chloride conductance was immeasurably small or absent in A6 epithelia grown on Transwell-clear inserts. Because the time-dependent increases of apical membrane capacitance paralleled the increases of sodium transport and were completely dissociated from those of activation of chloride conductance, we concluded that the slightly delayed and relatively slow increases of capacitance were associated with activation of apical membrane sodium conductance.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Tissues, solutions, and drugs

A6 cells at passages 109 to 114 were used in the present studies. After growth on 75 cm² plastic culture flasks at 28°C in a humidified incubator containing 1% CO₂, the cells were subcultured on Transwell-clear cluster inserts (Costar, Cambridge, MA) for at least 14 days to achieve confluence and development of their transepithelial transport characteristics (Helman and Liu, 1997). The tissues were fed twice weekly with a Cl-rich growth medium that was based on a mixture of equal parts of Ham's nutrient mixture F-12 with L-glutamine and without sodium bicarbonate (N-6760, Sigma Chemical Co., St. Louis, MO) and L-15 Leibovitz medium (L-4386, Sigma Chemical Co.). This mixture was supplemented with 10% fetal bovine serum (FBS) (SH0070, HyClone, Logan, UT), 2.57 mM sodium bicarbonate, 3.84 mM L-glutamine (Sigma Chemical Co.), 96 U/ml penicillin, and 96 µg/ml streptomycin (BioWhittaker, Walkersville, MD).

The tissues were transferred to edge damage-free chambers (Abramcheck et al., 1985; Awayda et al., 1999) and short-circuited for the duration of the experiments using a very low-noise, four-electrode (Ag/AgCl, 4.5 M NaCl, 3% agar) voltage clamp while being perfused continuously at flow rates of ~7 ml/min through chamber volumes of ~0.5 ml with the growth medium without FBS and glutamine. Na⁺, K⁺, Cl, Ca²⁺, and Mg²⁺ concentrations were 103.0, 3.34, 106.8, 0.59, and 0.91 mM, respectively. Short-circuit currents (I_sc) were allowed to stabilize for at least 2 h before onset of experimental periods. Characteristically, the I_sc increased transiently when tissues were initially short-circuited, returning to near steady-state values within 1 to 2 h (Punescu et al., 1997); 100 µM amiloride added to the apical perfusion solution was used to inhibit blocker-sensitive Na⁺ currents (I).

To elevate intracellular cAMP, forskolin (Sigma Chemical Co.) or prostaglandin E₂ (PGE₂) (Sigma Chemical Co.) was added to the basolateral perfusion solution at final concentrations of 25 µM or 1 µM, respectively. Stock solutions of forskolin and PGE₂ were dissolved in ethanol at 10² M and stored at 20°C. Furosemide (Sigma Chemical Co.) was added directly to the basolateral solution at 1 mM to inhibit electroneutral chloride transport at the basolateral membranes of the cells (Brazy and Gunn, 1976; Lambert and Lowe, 1980; Stoddard et al., 1985). A chloride-free, gluconate Ringer's solution containing 100 mM sodium gluconate, 2.4 mM KHCO₃, and 2.0 mM CaSO₄ was perfused through apical and basolateral chambers in those experiments where tissues were exposed for 1 h during control periods and chronically thereafter to chloride-free solution.

Impedance analysis

Transepithelial impedance was measured under voltage clamp conditions using three overlapping bands of frequencies (low, medium, and high) between 0.244 Hz and 10.45 kHz (Fig. 1) and using essentially the same approach described previously (Awayda et al., 1999), but with several modifications. The voltage command signals (V_cmd) consisted of the vectorial sum of 43 frequencies where the absolute amplitude of each sinusoid decreased with increasing frequency (Fig. 1 B) and with the phase angle indicated in Fig. 1 C. This design of the composite voltage command signal served to assure that the capacitive currents at each frequency would be closer in magnitude than would occur if the amplitudes of the voltage command sinusoids were of equal amplitude at all frequencies. With fundamental periods (T) of 4.098 s (low frequencies), 255.7 ms (medium frequencies), and 20 ms (high frequencies), corresponding to fundamental frequencies of 0.244, 3.91, and 50.0 Hz, the relative amplitudes of the voltage command sinusoids in all three frequency bands can be compared (Fig. 1 B), especially in the overlapping ranges of frequency between bands (3.91 to 51.02 Hz and 50.0 to 816.4 Hz). Consequently, impedance in the overlapping bands of frequency was measured with sinusoids of markedly differing voltage and current amplitudes at each time point of measurement, thereby providing a built-in check for testing and assuring that impedance was independent of the magnitude of V_cmd. The magnitudes of V_cmd (Fig. 1 A) were adjusted so that the peak-to-peak changes of transepithelial voltage were near 2 mV. Consequently, it could be assumed that impedance was measured for all practical purposes in linear ranges of the current-voltage relationships of the channels.

View larger version (18K): [in this window] [in a new window]   FIGURE 1   (A) Composite voltage command signal, Vcmd, consisted of the vectorial sum of 43 sinusoids with relative amplitudes (B) and phase angles (C) and with time periods, T, determined by their fundamental frequencies (0.244, 3.91, and 50.0 Hz).

An IBM-compatible computer containing a DSP2200 board (16 bit ADCs, 16 bit DACs, National Instruments, Austin, TX) was programmed using LabWindows for DOS to output the analog V_cmd signals and to simultaneously digitize the measured transepithelial voltage and current signals after amplification and filtration at the Nyquist frequencies. Low, medium, and high frequency bands were each output in sequence over four time periods and data collection for analysis was retained for only the last period of each band. Accordingly, the total time for data acquisition was near 5.2 s [4.098 + (4 × 0.2557) + (4 × 0.020)]. These signals were Fourier-transformed to yield the voltage and current vectors. Z_meas was calculated as the quotient of voltage and current vectors at the respective frequencies. Where appropriate, data were normalized to planar surface area and plotted as Nyquist (ReZ_meas vs. ImZ_meas) or Bode plots (|Z_meas| and phase angle versus frequency). See Results for calculations of apical and basolateral membrane resistances and capacitances.

Data are summarized as means ± SE. Statistical analyses were performed with SigmaStat (Jandel Scientific Software, San Rafael, CA) using paired or unpaired t-tests where appropriate. A p value < 0.05 was considered significant. All experiments were carried out at room temperature.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Activation of amiloride-insensitive chloride current by forskolin and PGE₂

In agreement with the observations of others (Yanase and Handler, 1986; Perkins and Handler, 1981; Chalfant et al., 1993; Niisato and Marunaka, 1997), increases of intracellular cAMP lead to activation of chloride currents in A6 epithelia. As indicated in the strip chart recording of Fig. 2 A, we have observed characteristically, following treatment with either forskolin or PGE₂, that the I_sc increases abruptly within tens of seconds to peak or quasi-plateau values (Punescu, 1999). After what appears to be a short delay, the I_sc increases markedly and relatively slowly to substantially elevated plateau values within 20 to 30 min that are sustained for several hours. Addition of amiloride to forskolin- or PGE₂-stimulated tissues at the ends of 2-h experiments (not shown) characteristically resulted in large inhibitions of the I_sc. Compared to untreated tissues where the steady-state amiloride-insensitive currents, I, normally average in the range of 0.1-0.3 µA/cm² (Punescu et al., 1997, 2000b) and that averaged 0.16 ± 0.03 µA/cm² (n = 6) in the present series of experiments (Fig. 2, B and C), the I of forskolin and PGE₂-treated tissues averaged 1.91 ± 0.18 µA/cm² (n = 12) and 1.66 ± 0.09 µA/cm² (n = 11), respectively. Because the responses to forskolin, PGE₂, exogenous cAMP, and theophylline caused similar increases of the amiloride-insensitive short-circuit currents (Punescu, 1999; Punescu and Helman, 1998, 2000), PGE₂ was used exclusively in the experiments reported below.

View larger version (23K): [in this window] [in a new window]   FIGURE 2   Typical strip chart recordings are shown of the short-circuit current responses to forskolin or PGE2 in control (A) and amiloride-pretreated tissues (B and C). The tissue in A was treated with 25 µM forskolin.

When amiloride-blocked tissues were challenged with PGE₂ as indicated in Fig. 2, B and C, the amiloride-insensitive short-circuit currents increased abruptly within 30 to 40 s to peak values, followed by relaxation of the currents in <10 min to stable but elevated values that were sustained for the duration of observation (1 to 2 h). In the absence of PGE₂ furosemide was without effect on the amiloride-insensitive currents (Fig. 2 C). In PGE₂-stimulated tissues furosemide addition to the basolateral solution inhibited reversibly, but not completely, the amiloride-insensitive current (Fig. 2, B and C). Peak current values averaged 3.32 ± 0.17 µA/cm² (amiloride, n = 3) and 2.26 ± 0.23 µA/cm² (amiloride + furosemide, n = 3) at 37.9 ± 2.7 s (n = 6) following exposure of the tissues to PGE₂. The currents decayed thereafter with a time constant of 2.48 ± 0.14 min (n = 6) to plateau values that averaged 1.20 ± 0.11 µA/cm² (amiloride, n = 3) and 0.51 ± 0.12 µA/cm² (amiloride + furosemide, n = 3). After additional treatment of amiloride-blocked tissues with furosemide, the currents decreased to 0.38 ± 0.04 µA/cm² (n = 3) (Fig. 2 B).

Shown in expanded form in Fig. 3 are representative changes of I_sc (I_sc) from basal levels caused by PGE₂ within 7 min in a control tissue, an amiloride-blocked tissue bathed in the chloride-rich perfusion solution, and a control tissue bathed in chloride-free solution. Whereas currents in tissues with functional ENaCs exhibit relatively large secondary increases of amiloride-sensitive current (not shown), the delayed increases of current are absent in amiloride-pretreated tissues but are observed in chloride-free media. Notably, the abrupt increases of current are completely absent in chloride-free media, thereby providing evidence that the initial response to PGE₂ and forskolin is due to activation of a chloride conductance that precedes full activation of Na⁺ transport. In these regards our observations are the same as those reported by Chalfant et al. (1993).

View larger version (14K): [in this window] [in a new window]   FIGURE 3   Time courses of change of the short-circuit current, Isc, caused by PGE2 within the first 7 min in a control tissue and a tissue pretreated with amiloride that were bathed with the chloride-rich solution. Also shown is the time course of change of the Isc of a control tissue bathed with the chloride-free solution. The delay in onset of increase of the sodium current was ~30 s, at which time stimulation of chloride current was near maximal.

The amiloride-insensitive currents following PGE₂ or forskolin represent the maximal increases of chloride current at their peak and during the subsequent steady states of transport. In this regard it is relevant to note that the amiloride-insensitive basal short-circuit currents prior to PGE₂/forskolin are due principally to amiloride-insensitive Na⁺ currents (Baxendale-Cox et al., 1997). It is unknown to what extent PGE₂/forskolin may activate amiloride-insensitive Na⁺ current together with the chloride current. Consequently, the specific current due to chloride would be less than that measured by I if amiloride-insensitive Na⁺ currents are increased by PGE₂/forskolin. Thus, for tissues continuously short-circuited, the steady-state chloride currents activated by PGE₂/forskolin are rather small in magnitude, averaging at most <2 µA/cm² in amiloride-blocked tissues and at most near 0.44 µA/cm² when tissues are additionally treated with furosemide. Neglecting the contribution of amiloride-insensitive Na⁺ currents to the I of furosemide-treated tissues, furosemide inhibited between ~70-100% of the chloride current activated by PGE₂. If the I of PGE₂-treated tissues bathed in chloride-free solution is taken as an estimate of the amiloride-insensitive Na⁺ current (0.47 ± 0.08 µA/cm², n = 5) then furosemide inhibited nearly 100% of the steady-state PGE₂-activated chloride current. In the absence of step changes of I in response to a high concentration of basolateral furosemide, it would be reasonable to conclude that furosemide acts to inhibit basolateral membrane chloride entry into the cells via an electroneutral mechanism(s) of transport because step changes of transcellular resistance, if they occurred, would cause step changes of short-circuit current that were never observed.

Impedance of control and amiloride-treated tissues

Typical Nyquist plots of the measured impedance (Z_meas) of control, unstimulated tissues before and after treatment with amiloride are illustrated in Fig. 4. I_sc averaged 7.41 ± 0.60 µA/cm² and was decreased to 0.46 ± 0.06 µA/cm² within 6 min after amiloride (Table 1). The impedance locus (Fig. 4 A) conformed to a depressed semicircle at frequencies <~100 Hz and could be fit to an equation in the form of Eq. 1, thereby permitting estimation of the dc resistance. Although not apparent when viewed this way, the data points at frequencies >100 Hz did not conform to this equation (see below) and so the apparent resistance of apical and basolateral solutions, R, calculated this way (156.6 ± 36.7  · cm², n = 6) overestimated the actual series resistance R_sol (47.5 ± 1.5  · cm², n = 6). The R_sol could be estimated by eye to within ~1-2  · cm² by extrapolation of the ImZ_meas to the real axis. In practice, R_sol was determined by nonlinear curve-fitting of the ReZ_meas to infinite frequency using TableCurve (SPSS Inc., Chicago, IL). The transepithelial resistance (R_T) was determined as the difference between the dc value of Z_meas (Z) and the R_sol.

(1)

In these and all subsequent experiments, R_sol was not changed by amiloride nor PGE₂. Control R_T averaged 6.11 ± 0.77 k · cm² and at 6 and 45 min was increased to means of 10.61 and 9.76 k · cm², respectively, at these time points (Table 1). With the values of transepithelial conductance (G_T = R) and short-circuit currents before and after amiloride, the shunt resistance R_p = G and the Thévenin emf of the cellular pathway for Na⁺ transport (E_Na) were calculated using the method of Yonath and Civan (1971) where G_T = I_sc/E_Na + G_p. R_p averaged 10.74 and 10.03 k · cm² at 6 and 45 min after amiloride (Table 1) and the E_Na averaged 112.5 ± 4.3 mV (n = 6), which is quite typical for A6 and other tight epithelial tissues that transport Na⁺ exclusively through apical membrane ENaCs (Helman and Liu, 1997; Koeppen et al., 1980; Helman and Thompson, 1982; Macchia and Helman, 1979; Yonath and Civan, 1971).

View larger version (22K): [in this window] [in a new window]   FIGURE 4   (A) Typical Nyquist plots of the measured impedance, Zmeas, are illustrated for a control tissue that was treated with amiloride. The frequencies of the impedance vectors are indicated at the apices of the semicircles and at 300 Hz. An expanded view of the impedance vectors between 300 Hz and 10,450 Hz is shown in B. The solid lines shown in A were determined by nonlinear curve-fitting of the impedance vectors to Eq. 1 (see text and Fig. 6).

From the differences between R_T and R_p, the series resistance of apical and basolateral plasma membranes of the cells, R_cell = R_a + R_b, was calculated to have been increased by amiloride from 15.83 k · cm² to >260 k · cm² (Table 1). To the extent that control tissues express amiloride-sensitive ENaCs and amiloride-insensitive Na⁺ currents that are small, the amiloride-insensitive resistance of the apical membrane is expected to be in the range of 250 k · cm² if apical membrane voltage is 100 mV and the blocker-insensitive Na⁺ current, I, is 0.4 µA/cm². Thus, in control tissues, apical membranes are predominantly permeable to Na⁺. If a finite apical membrane chloride conductance exists in control tissues, its value must be extremely low and immeasurably small.

Capacitance of control and amiloride-treated tissues

Frequency-dependent capacitances can arise from Maxwell-Wagner and/or Cole-Cole dielectric dispersions (Punescu and Helman, 2001). At sufficiently high frequencies where the apical and basolateral membrane capacitive reactances are considerably less in value than their respective membrane resistances, the equivalent cell capacitance, C_eq = C_aC_b/(C_a + C_b), would be constant if the C_a and C_b were frequency-independent. If, however, the capacitances exhibit audio frequency -dispersions, C_eq would appear as a complex frequency-dependent capacitance C^*_eq.

At all frequencies C^*_eq  [j(Z_a + Z_b)]¹ (Punescu and Helman, 2001). Accordingly, the transepithelial impedance is given by Eq. 2 and C^*_eq can be determined from the measured values of Z_T and R_p. Shown in Fig. 5 A are typical results for a tissue in its control and amiloride-treated states. With increasing frequency, |C^*_eq| decreased from values near 1.5 µF/cm² to near 0.9 µF/cm² at 10.45 kHz. Amiloride did not change the |C^*_eq|.

(2)

The mean |C^*_eq| of amiloride-treated tissues is summarized in Fig. 5 C. The solid line fitted manually defines an approximate Cole-Cole -dispersion with a characteristic frequency near 150 Hz. The deviations of |C^*_eq| and phase angle  of C^*_eq from pure Cole-Cole behavior shown by the manually fitted lines in Fig. 5, C and D at frequencies <~10 Hz are expected due to the Maxwell-Wagner-like dispersion (Punescu and Helman, 2001). It should be stressed that manual fitting of a single Cole-Cole relaxation process to the data is not exact, and so the solid lines shown in Fig. 5, B and C should be taken as reasonable but imperfect approximations.

View larger version (22K): [in this window] [in a new window]   FIGURE 5   Apical membrane capacitance is frequency-dependent. Shown in A are typical plots of the dependence on frequency of the absolute values of the equivalent cell capacitance, |C*eq|, in the control state of a tissue and after inhibition of Na+ transport by amiloride. When these data are plotted as Nyquist plots as indicated in B, the capacitance vectors appear to conform to a depressed semicircle (dielectric increment  0.58 µF/cm2; Cole-Cole power law coefficient  0.65; characteristic frequency  110 Hz) at the higher frequencies. Significant deviations from a pure depressed semicircle at the lower frequencies are expected and observed due to the Maxwell-Wagner effect at the lower frequencies (Punescu and Helman, 2001). A summary of the mean |C*eq| is shown in C with selected standard error bars at the lower frequencies and at several higher frequencies for the purpose of clarity. The corresponding mean phase angles of C*eq are summarized in D. To illustrate the approximate contribution and location of the Cole-Cole relaxation process in C and D (solid lines), the parameters of the process were adjusted manually (dielectric increment = 0.54 µF/cm2; power law coefficient = 0.625; characteristic frequency = 150 Hz; limiting static capacitance = 0.89 µF/cm2) according to Eq. 5 in Punescu and Helman (2001) to obtain an approximate fit of the lines to the mean data points. More robust methods were not attempted due to the less than adequate number and reliability of data points at the lowest frequencies.

View larger version (21K): [in this window] [in a new window]   FIGURE 6   Impedance data of control and amiloride-treated tissues were fit to Eq. 1 at frequencies <50 Hz as shown by the solid lines in Nyquist (A and B) and Bode plots (C and D). (A) The fitted solid lines were depressed from ideal semicircles (dashed line). (B) Expanded view of A at higher frequencies. The solid fitted line deviated from the data points at frequencies >50 Hz and intersected the real axis at values of R significantly less than those of Rsol, indicating that the impedance locus is skewed toward low frequencies.

Shown also in Fig. 5 B are typical Nyquist plots of C^*_eq for the control and amiloride-treated states of the tissues. Significant deviations from single ideal depressed semicircles were observed, as expected at the lower frequencies corresponding to those associated with the Maxwell-Wagner dispersions (Punescu and Helman, 2001). Nevertheless, it was apparent, at least to a first approximation, that an -relaxation process was responsible for the frequency dependence of C^*_eq. Because C_a C_b (see below), it follows that this dispersion arises predominantly from the apical membranes of the cells. Consequently, apical membrane capacitance could not be assumed to be constant in the audio range of frequencies.

Provided that the bandwidth of impedance vectors was limited to a maximum of ~50 Hz or less for the present set of experiments, it was possible to obtain satisfactory fits of the impedance vectors in this limited range of frequencies to Eq. 1, which is an equation of a depressed semicircle. Typically, as illustrated in Fig. 6 and viewed as Nyquist (Fig. 6, A and B) or Bode (Fig. 6, C and D) plots, the fitted lines deviated significantly from symmetrical depressed semicircles at frequencies >100 Hz. R significantly underestimated the values of R_sol as noted above (Fig. 6 B).

Capacitance calculated at the apex of the fitted semicircle yielded mean values near 1.36 and 1.40 µF/cm² for control and amiloride-treated tissues, respectively (Table 1) at mean frequencies near 22 and 14 Hz, respectively. The corresponding mean values of |C^*_eq| at these frequencies reported in Fig. 5 C are 1.33 ± 0.03 and 1.36 ± 0.03 µF/cm² at 22 and 14 Hz, respectively, and are quite similar to the fitted values expected at these frequencies (Punescu and Helman, 2001). It should be noted that the capacitances calculated at the apices of the fitted depressed semicircles are not unique, but will depend not only on the frequency dependence of C^*_eq but also on the values of R^fit  R_p because the frequency at the apices f^fit = (2R^fit|C^*_eq|)¹. Consequently, differences in the tightness or leakiness of the paracellular shunt pathways will result in differences in f^fit and thus differences in capacitance calculated this way, even though C^*_eq has not changed.

PGE₂ activates a large apical membrane conductance to chloride

PGE₂ causes not only a dramatic decrease in transepithelial impedance but also a marked change of the impedance locus as indicated for a typical experiment, illustrated in Fig. 7. The change of impedance was clearly maximal at 2 min (not shown) at frequencies >5 Hz. The impedance at the very low frequencies <~3-4 Hz could not be resolved during the transient relaxation of currents that violated the requirement of steady-state currents and voltages to measure impedance. It should be noted that these experiments were carried out in the presence of amiloride to block Na⁺ currents so that the responses could be attributed to an increase of Cl conductance. R_T of the amiloride-treated tissue in Fig. 7 was decreased from near 13 to 4 k · cm². From control values of 0.27 ± 0.01 µA/cm², the I averaged at 16 and 45 min after PGE₂ 2.01 and 1.93 µA/cm², respectively. The mean R_T was decreased from amiloride control values near 10 k · cm² (Table 1) to 3.52 and 3.39 k · cm² (Table 2), respectively, at these same time points. Within 2 min PGE₂ caused maximal decreases of impedance that were sustained. The changes of impedance were reversible after withdrawal of PGE₂ although reversal was considerably slower (Fig. 7 C) and not quite complete at 45 min. From the PGE₂-stimulated I of 1.82 ± 0.07 µA/cm² (n = 6), the I decreased after PGE₂ washout at 2, 6, and 45 min to 1.02 ± 0.07, 0.74 ± 0.06, and 0.39 ± 0.03 µA/cm², respectively.

View larger version (19K): [in this window] [in a new window]   FIGURE 7   PGE2 causes a dramatic change of impedance and its locus, as indicated in A. The data of A were normalized to the values of Z to better illustrate the change of impedance locus as shown in B. The impedance vectors at the lower frequencies <~200 Hz of PGE2-treated tissues required the fitting of at least two depressed semicircles (solid line) to Eq. 3 (see text). The effects of PGE2 were partially reversible within 45 min, as indicated in C. In comparison with the onset of PGE2 effects on impedance the washout was relatively slow, as indicated by the changes of impedance locus at 2, 6, and 45 min.

The impedance vectors normalized to the values of Z (Fig. 7 B) more clearly indicate that the impedance locus was changed by PGE₂ from a single depressed semicircle at frequencies <50 Hz in the amiloride control state to a locus consistent with a large decrease of apical membrane resistance. Under these latter circumstances where apical membrane resistance, R_a, is decreased by PGE₂ into a range of values comparable in magnitude to the basolateral membrane resistance, R_b, and where in addition the time constants of apical and basolateral membranes are sufficiently different from each other to result in observable interacting semicircles in Nyquist plots, the measured impedance is quite generally:

(3)

where apical (Z_a) and basolateral (Z_b) membrane impedances are:

(4)

(5)

It should be emphasized here, as above, that the values of C and C are those at the unique frequencies at the apices of the semicircles and where the semicircles would be depressed if the power law dependencies and were less than unity due to capacitance being frequency-dependent (Punescu and Helman, 2001).

To determine apical and basolateral membrane resistances and capacitances, impedance data were fit by nonlinear curve-fitting (Scientist for Windows, MicroMath, Inc., Salt Lake City, UT) to Eq. 3 at frequencies <50 Hz for the amiloride control data, as was indicated above, and at frequencies <200 Hz when tissues were treated with PGE₂ (Fig. 7). Starting values of R_p were determined independently before treatment of the tissues with PGE₂. Starting values of R_a and R_b were estimated with the values of R_cell and the fractional transcellular resistances (R_a/(R_a + R_b)) that were evident by inspection of the impedance plots. C_a and C_b were estimated at frequencies corresponding to the approximate apices of the impedance loci. We will refer to the apical and basolateral membrane resistances and capacitances below simply as R_a, R_b, C_a, and C_b, recognizing that these parameters were determined by fitting of the impedance loci to Eq. 3 at the very low audio frequencies.

During steady-state periods of transport and at time points of 16 and 45 min, R_a averaged 1.42 and 1.70 k · cm² and R_b averaged 4.02 and 3.90 k · cm², respectively (Table 2). R_p averaged near 10 k · cm² (Table 2) and was essentially unchanged by PGE₂. Although it was not possible to determine the values of R_b before PGE₂ in control or amiloride-blocked tissues, the PGE₂-related time-dependent changes of R_a and R_b could be assessed as indicated in Fig. 8 A. R_a was maximally decreased from greater than a few hundred k · cm² within 2 min by PGE₂ and remained essentially constant for the 45 min period of observation. In contrast, it became apparent following PGE₂ that R_b decreased from >5 k · cm² to steady-state values near 4.0 k · cm² within 6 min. R_b in control tissues determined by a different method was found to average >6.6 k · cm² (Punescu, 1999) and 9.0 k · cm² (Punescu et al., 2000b), which would be consistent with the time-dependent decrease of R_b measured here by impedance of tissues treated with PGE₂. Accordingly, whereas the PGE₂-related decrease of R_b was relatively small and relatively slow, the major and very rapid effect of PGE₂ was to decrease the apical membrane resistance to values considerably less than those of the basolateral membranes of the cells.

View larger version (12K): [in this window] [in a new window]   FIGURE 8   PGE2 causes decrease of both apical (Ra) and basolateral membrane (Rb) resistance of amiloride-pretreated tissues as shown in A. Whereas the decrease of Ra is maximal within 2 min, the decrease of Rb is relatively slow, but essentially complete within ~6-10 min. The corresponding changes of the transcellular Thévenin emf, Ecell, are shown in B (see text).

Transcellular emfs

The magnitudes of the transcellular current before and after PGE₂ will be determined not only by R_a and R_b, but also by changes of the transcellular driving force E_cell where E_cell = E_a + E_b and where E_a and E_b are, respectively, the Thévenin emfs of apical and basolateral membranes. Accordingly, I_sc = (E_a + E_b)/(R_a + R_b) = E_cell/(R_a + R_b). Before PGE₂, where apical membranes are populated principally if not solely by amiloride-sensitive and -insensitive Na⁺ channels and where E_a is at or very near zero (Punescu and Helman, 2001), I_sc = E_b/(R_a + R_b). If the R_b is determined principally by the conductance of basolateral membrane K⁺ channels, then at zero current flow through the basolateral membrane, E_b = |_K + I_pumpR_b| (Helman and Thompson, 1982), where _K is the Nernst equilibrium potential difference and I_pump is the current generated by the Na,K-ATPase. Accordingly, relatively fast decreases of R_b that occur before appreciable changes of _K or I_pump would be expected to cause increases of E_b, which together with decreases of R_b would lead to stimulation of I_sc. Although decreases of R_b will lead to increases of I_sc, E_cell would be expected to increase if the channels expressed at apical membranes were exclusively Na⁺ channels, as would be the case for tight epithelia like frog skin and toad urinary bladder, where generally E_cell = E_Na averages above 100 mV and as indicated above, 112.5 mV for the A6 epithelia of the present experiments.

For tissues like A6, where cAMP activates both Na⁺ and Cl channels and where chloride channels are relatively close to electrochemical equilibrium, the E_a after PGE₂ must move toward the Nernst equilibrium potential difference of Cl, _Cl. With amiloride-blocked tissues where apical membranes express blocker-insensitive Na⁺ channels (R), E_a = (E_ClR)/(R + R_Cl). Because R_Cl R, after PGE₂ E_a can practically be equated with _Cl. Thus, with reference to a grounded apical solution, E_cell = E_b  Cl = I(R_a + R_b) so that E_cell can be calculated from the I and transcellular slope resistance (R_a + R_b). As indicated in Fig. 8 B, E_cell decreased dramatically within 2 min from 112.5 mV to mean values of 20.4 mV, and then to 10.9 and 10.7 mV at 16 and 45 min, respectively, after PGE₂. Consequently, despite a rapid opening of a large apical membrane Cl conductance within 2 min, the initial peak magnitude of the chloride current (I_Cl = (V_a  Cl)/R_Cl) was compensated for by a correspondingly large decrease of E_cell. Because Cl current enters the cells initially, V_a > _Cl. With loss of Cl from the cells and increase of _Cl, E_cell declines toward steady-state values, as does the V_a  Cl and hence the Cl current component of I.

The apical membrane Na⁺ current components of I during these times are I = V_a/R, so that I = I + I_Cl. Because R R_Cl, the rapid initial increases of I must be due to I_Cl, with I_Cl relaxing to its steady-state value where chloride secretion is determined principally by the rate of basolateral membrane Cl entry (see Appendix). If the basolateral membrane chloride entry were zero, then chloride must come to electrochemical equilibrium at the apical membranes of the cells where at the steady state I_Cl = 0 or chloride secretion would be absent despite high values of apical membrane conductance to Cl.

Depressed semicircles of impedance loci

Because capacitance is frequency-dependent, it is expected that the impedance locus would consist of depressed semicircles (Punescu and Helman, 2001). Nonlinear curve-fitting of the impedance vectors as indicated above yielded mean power law dependencies _a and _b near 0.85 and 0.90, respectively (Table 2). The absolute values of capacitance at the apices of the depressed semicircles |C_a| and |C_b| averaged near 1.38 and 20.1 µF/cm², respectively, in PGE₂-treated tissues. The frequencies at the apices averaged ~70-80 Hz for |C_a| and ~2.0 Hz for |C_b| (Table 2).

PGE₂-related changes of impedance in amiloride-blocked tissues require Cl

To ensure that PGE₂ activated a chloride conductance in the tissues we studied, two experimental protocols were used to test for chloride dependence of the PGE₂-related changes of impedance in amiloride-pretreated tissues. In experiments illustrated in Fig. 9, A and B, tissues bathed in chloride-rich solution were treated first with PGE₂ and then exposed to a chloride-free gluconate solution. In the experiments illustrated in Fig. 9, C and D, tissues were bathed first with chloride-free solution, treated with PGE₂ in the absence of chloride, and then exposed to a chloride-rich solution. Regardless of the sequence, removal of chloride in the PGE₂-treated states of the tissues caused reversible decreases of the I to 0.47 ± 0.08 µA/cm² from 1.75 ± 0.08 µA/cm² (n = 5). The R_T and impedance locus were unchanged by PGE₂ when tissues were bathed in chloride-free solution (Fig. 9 C). Subsequent exposure to the chloride-rich solution resulted in a decrease of R_T and a change of the impedance locus (Fig. 9, C and D). Removal of chloride from tissues treated first with PGE₂ resulted in reversible increases of R_T and changes of the impedance locus (Fig. 9 A). These specific chloride-dependent changes of current, resistance, and impedance locus provide compelling evidence to support the view that PGE₂ acting through cAMP activates a large chloride conductance at the apical membranes of the cells that precedes significant activation of apical membrane ENaCs, and hence the amiloride-sensitive apical membrane Na⁺ conductance. Thus, the R_a of amiloride- and PGE₂-treated tissues can be equated with the apical membrane resistance to chloride, R_Cl. To the extent that in control tissues PGE₂ stimulates Na⁺ transport about threefold (Punescu and Helman, 1997; Punescu, 1999) to ~20 µA/cm², the amiloride-sensitive Na⁺ resistance, R, would be ~5.6 k · cm² (112.5 mV/20 µA/cm²) or about three to fourfold larger than R_Cl in PGE₂-treated A6 epithelia.

View larger version (23K): [in this window] [in a new window]   FIGURE 9   The changes of impedance and its locus caused by PGE2 are dependent on chloride. In A, gluconate replaced all chloride in the apical and basolateral solutions bathing a tissue that had already been treated with PGE2 and amiloride. The effect of chloride removal was completely reversible upon return to the chloride-rich solution. The impedance vectors in A normalized to the values of Z are shown in B. In C the tissue was treated with PGE2 while it was bathed with the chloride-free gluconate solution, causing a minuscule change of impedance. Exposure to the chloride-rich solution caused a typical change of impedance and its locus. The normalized impedance vectors in C are shown in D.

Time-dependent change of capacitance

In the face of a frequency-dependent apical membrane capacitance and changes of resistance, we could not rely on the estimates of capacitance at the apices of the semicircles to evaluate changes of capacitance caused by PGE₂ because it was not possible to compare the capacitances at the same frequencies in control and PGE₂-stimulated states of the tissues. Instead, we evaluated the changes of |C^*_eq| caused by PGE₂ where we could compare capacitance at all frequencies. We were particularly interested in evaluating the data at the higher frequencies, because at sufficiently high frequencies the apical membrane capacitive reactance, X_a = (jC^*_a)¹, decreases to values substantially less than those of the apical membrane resistance so that the Z_a can be equated with X_a. Because of the relatively high basolateral membrane capacitance and resistance, it is readily appreciated that at high frequencies the basolateral membrane impedance can be equated with its capacitive reactance. Although we do not know the frequency dependence of C^*_b, its contribution to C^*_eq at the higher frequencies would be relatively small if the ratio of C_b/C_a near 15:1 (Table 2) at the lower frequencies is indicative of its ratio at higher frequencies. Accordingly, |C^*_eq| approaches values near the apical membrane capacitance |C^*_a| when X_a R_a.

Shown in Fig. 10 A for a representative experiment are the |C^*_eq| at frequencies between 250 Hz and 10 kHz in the control state of the tissue and at various time points between 2 and 45 min after treating the tissue with PGE₂. It was evident that PGE₂ caused an increase of |C^*_eq| at every frequency and at every time point. It was also evident at frequencies >1000 Hz that capacitance increased relatively slowly with small increases discernible at 2 min but with the dominant increase occurring between 2 min and 16 min. Bearing in mind that PGE₂ decreases apical membrane resistance to values near 1.5 k · cm² and that this analysis requires that X_a R_a, we assumed as a criterion that this condition was met when at sufficiently high frequencies values of |C^*_eq| normalized to their control values were constant (independent of frequency). Typically, as determined empirically, this criterion was satisfied at frequencies >~1500 Hz, as indicated in Fig. 10 B for time points of 2, 6, and 45 min. At frequencies above 8 kHz the uncertainties in difference of values between Z_meas and those of R_sol limited reliable evaluation of the normalized |C^*_eq| as the difference values approached zero. To summarize, the normalized values of |C^*_eq| were averaged between 1.5 and 4.0 kHz at each time point after PGE₂.

View larger version (22K): [in this window] [in a new window]   FIGURE 10   Shown in A are the absolute values of the equivalent cell capacitance, |C*eq|, at frequencies between 250 Hz and 10 kHz for a typical tissue in its amiloride control state and at the time intervals indicated following treatment with PGE2. Shown in B are the values of |C*eq| at 2, 6, and 45 min after PGE2 normalized to their control values at the same frequencies between 1.5 and 8.0 kHz. Intermediate time points at 4 and 16 min have been omitted for clarity. Although the |C*eq| are dependent on frequency, the normalized changes of |C*eq| are constant (independent of frequency) in this range of frequencies. The mean of the 19 normalized values of |C*eq| are indicated by the solid lines. A summary (means ± SE, n = 6) of the time-dependent increases of the normalized |C*eq| is shown in C. Also shown in C for the purpose of direct comparison of time courses are typical normalized time-dependent changes of short-circuit current in a tissue pretreated with amiloride and bathed in chloride-rich solution (see Fig. 2 B), and a control tissue bathed with chloride-free gluconate Ringer's solution (dashed line) where basal and PGE2-stimulated currents are amiloride-inhibitable.

Shown in Fig. 10 C is a summary of the time-dependent changes of |C^*_eq| caused by PGE₂. After a short delay of <1 min capacitance was increased at 45 min by 9.8 ± 0.3% (n = 6). The best fit line to the data points was determined by nonlinear curve-fitting using the set of transition functions in TableCurve (SPSS Inc., Chicago, IL). The data were best fit to a Weibull cumulative function. This line and other asymmetric transition functions that fit nearly as well (pulse cumulative with power term, log-normal cumulative) all indicated that the onset of response of capacitance to PGE₂ was delayed by ~1 min. For comparison of time course, the I response to PGE₂ shown in Fig. 2 B is reproduced in Fig. 10 C. Remarkably, capacitance increased relatively slowly toward a sustained plateau within ~20-30 min, paralleling the time-dependent increases of amiloride-sensitive short-circuit current attributable to stimulation of Na⁺ transport and apical membrane functional ENaC densities (Helman and Punescu, 1998; Punescu and Helman, 1997; Punescu, 1999; Els and Helman, 1997). These findings are perhaps more remarkable given the rapidity (<1 min) with which PGE₂ maximally activates the apical membrane chloride conductance. Indeed, there is a virtually complete dissociation in time between activation of chloride and sodium conductances. In this regard the changes of capacitance caused by PGE₂ are correlated with activation of sodium conductance. If increases of capacitance are associated with activation of chloride conductance in A6 epithelia, they are undetectable by the methods used in our studies.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Chloride transport

Although PGE₂ is a potent activator of Na⁺ transport in A6 epithelia (Kokko et al., 1994; Matsumoto et al., 1997; Punescu and Helman, 1997), its effect on chloride transport is relatively minor. According to our own analysis, chloride transport is stimulated maximally by PGE₂ within 1 min and well before onset of appreciable increases of Na⁺ transport. Furosemide reversibly inhibits the steady-state PGE₂-activated chloride transport, does not block the PGE₂ activation of apical membrane chloride conductance, and thus acts most likely through inhibition of Cl entry into the cells through an electroneutral transporter at the basolateral membranes of the cells (Fan et al., 1992; see Appendix).

In the absence of PGE₂ the apical membranes are permeable only to Na⁺, principally through amiloride-sensitive ENaCs but also, to a relatively small degree, through amiloride-insensitive Na⁺ channels (Baxendale-Cox et al., 1997; Helman et al., 1998). In this regard the impedance locus of control tissues, and in particular amiloride-blocked control tissues, behaved as though the fractional transcellular resistances approached values of unity, indicating that the apical membrane resistance was far greater in value than the basolateral membrane resistance. Consequently, apical membrane chloride conductance in the absence of PGE₂ is for all practical purposes at or near zero, so that chloride channels are either in a permanently shut state or have open probabilities very near zero.

Clearly, PGE₂/cAMP activation of chloride conductance is remarkably a very rapid process that in the A6 epithelia we studied causes a large decrease of apical membrane resistance to values that averaged near 1500  · cm² in amiloride-blocked tissues. To achieve steady-state chloride currents at the apical membranes that averaged near 2 µA/cm², a small net electrochemical driving force of 3 mV displaced from equilibrium would be required. Such a rapid increase of chloride conductance requires direct activation of channels that are resident within the apical membranes and/or an equally rapid translocation of chloride channel-containing vesicles from the cytosol to the apical membranes of the cells (see below), where the channels are already functional or become functional within seconds once they have reached the apical membranes. Interestingly, Kokko et al. (1997) have noted that PGE₂ activates chloride channels in cell-attached patches, which is a procedure that completely prevents cAMP activation of ENaCs (Marunaka and Eaton, 1991). Because chloride channels are activated in cell-attached patches even though patch formation completely disrupts activation of ENaCs, the implication is that the activation process does not involve trafficking of new chloride channels to the apical membranes (Kokko et al., 1997) (see Capacitance below).

Functional channel densities

Patch clamp experiments have also revealed that PGE₂-activated apical membrane chloride channels in A6 epithelia are characterized by a single channel conductance of 7 pS and an open probability near 0.5 (Kokko et al., 1997). Hence, the apical membrane density of open chloride channels activated by PGE₂ is ~95 million channels/cm² (or 95 channels per cell = (10⁶ cells/cm² × 1500  · cm² × 7 pS/channel)¹) and the total number of open and closed channels is therefore ~190 channels/cell that are maximally activated by PGE₂. Such calculations serve to point out under maximal conditions of channel activation that the densities of channels involved in transport are remarkably low. Hence, it becomes particularly important in assessing location and trafficking of chloride channels to use methods of detection with sufficient sensitivity that take into account the relatively few functional channels actually involved in changes of apical membrane chloride conductance and transport.

Capacitance

Changes of capacitance have been used widely as an index of change of membrane area. However, a complicating factor is the existence of audio frequency dispersions regardless of whether they arise from Maxwell-Wagner and/or Cole-Cole dielectric dispersions. Our results here, as elsewhere (Liu et al., 1995; Liu and Helman, 1998), clearly indicate that apical membranes of A6 epithelia exhibit a major dielectric dispersion at low audio frequencies with absolute values of capacitance decreasing from near 1.5 µF/cm² to near 0.9 µF/cm² as frequency nears 10 kHz. It should be emphasized that currents, resistances, and capacitances reported here have been normalized to the planar surface area of the tissue. Actual membrane areas of apical and basolateral membranes are not known. Although there is considerable uncertainty in knowing the actual apical membrane area, it is known from recent confocal microscopic studies that apical membranes of living cells of A6 epithelia are dome-shaped (Butterworth et al., 2001). Thus, actual apical membrane area may, as a rough approximation, be about twice the planar area so that capacitance normalized to actual area is more likely in the vicinity of 0.45 (~10 kHz) to 0.75 µF/cm² (dc). To the extent that the capacitance of a vacuum with a dielectric thickness of 40-60 Å is in the range of 0.15-0.22 µF/cm² (Awayda et al., 1999), it is clear that dielectric increments that give rise to two to threefold increases of capacitance above vacuum must exist at radio and/or higher frequencies to account for the capacitance at 10 kHz. The major dielectric increment (0.9 to 1.5 µF/cm²), however, exists at audio frequencies, where changes not only of area, but also changes of the -dispersion dielectric increment and/or its relaxation frequency can give rise to changes of capacitance unrelated to changes of membrane area, and so confound interpretation of changes of capacitance. In this regard we chose, as have Van Driessche and his colleagues (Zeiske et al., 1998; Atia et al., 1999) to measure capacitance in the kHz range of frequencies where changes of capacitance, when they occurred, could be due to change of membrane area and/or change of dielectric increments at radio or higher frequencies. To the extent that it is impossible with intact epithelia to measure capacitances at these higher frequencies to verify constancy of the dielectric increments, it remains impossible to conclude unequivocally that changes of capacitance are due to changes of membrane area. Nevertheless, more recent observations have indicated parallel time-dependent changes of Na⁺ transport rates, apical membrane ENaC densities, and vesicle endocytosis upon withdrawal of forskolin from pretreated A6 epithelia, suggesting that cAMP-dependent changes of channel densities occur by trafficking of channels to the apical membranes of the cells (Butterworth et al., 2001). Hence, it would be plausible to believe that the changes of capacitance observed in our experiments are related to changes of membrane area. Indeed, the time course of change of capacitance, together with its relatively delayed and slow onset and increase toward sustained plateau values within ~20-30 min, suggests that the observed increases of capacitance could be due mainly if not solely to vesicle trafficking of Na⁺ channels to the apical membranes of the cells. In this regard it is especially noteworthy that capacitance was not changed within 1 min, at which time chloride conductance was increased maximally. On the one hand, this could be interpreted to indicate that PGE₂ activates chloride channels that are resident within the apical membranes. On the other hand, we do not know how many chloride channels can be packaged within a single vesicle. If the density of packaging of channels per vesicle is high, then it would remain possible that only a few chloride-containing vesicles are involved in exocytosis that could result in undetectable changes of membrane area/capacitance, thereby requiring that no firm conclusion be made regarding the origin of chloride channels activated by PGE₂/cAMP.

Our findings with regard to PGE₂-related increases of capacitance are quantitatively similar to those reported by Zeiske et al. (1998) and Atia et al. (1999) who reported that a variety of hormones/agents that increase intracellular cAMP cause about an 8% increase of capacitance measured at kHz frequencies in A6 epithelia bathed with an NaCl-free apical solution. Under these conditions the peak increases of capacitance were in some but not all cases delayed from the peak increases of the chloride short-circuit current and the transepithelial conductance.