hyperpolarization-activated, cyclic nucleotide-gated, cationic nonselective (HCN) currents termed I_h, I_f, I_q, or inward pacemaker currents are found in a wide variety of excitable cells (1, 21, 32, 41). The I_h channels activate, with a slow time course, at hyperpolarizing voltages beyond −60 mV (near the resting potential of most cells) and are permeable to both Na⁺ and K⁺. They are regulated by cyclic nucleotides (cAMP and cGMP) and external K⁺ and blocked by external Cs⁺. The genes coding for I_h channels form the HCN family, with four members (HCN1 to -4) in mammals. Each HCN isoform is able to form homomeric conducting channels that differ in their kinetics, steady-state voltage dependence, and sensitivity to modulation by cAMP (1, 21, 41, 45). HCN isoforms may also coassemble to form heteromeric channel complexes (54, 27).

Although I_h and HCN gene expression is observed, almost exclusively, in excitable cells, a Northern blot study performed in mouse has shown the expression in liver and kidney of a splice variant of the HCN3 mRNA expressed in brain (44) while, more recently, the presence of mRNA encoding HCN2 and HCN3 in the kidney-derived transformed HEK293 cell was reported (57). If nonexcitable cells may express HCN genes, then the presence of the corresponding I_h current and HCN channels in such cells is a possibility deserving careful attention.

In a previous work (12), we have shown that inner medullary collecting duct cells in primary culture exhibit outward- and inward-rectifying cationic currents, and we identified the outward-rectifying current as a voltage-dependent K⁺ current, flowing through basolateral voltage-gated K⁺ channels. In the present work, we use the perforated-patch and conventional whole cell clamp to demonstrate that inward rectification is due to the presence of a hyperpolarization-activated, cyclic nucleotide-gated, nonselective cationic current that exhibits many similarities with I_h. We detect transcripts corresponding to the I_h channel genes (HCN1, -2, and -4) in both the cultured cells and kidney inner medulla. Western blot showed HCN2 immunoreactive proteins in cell cultures and inner medulla. Immunocytochemistry analysis showed the presence of HCN2 immunoreactivity in the cytoplasm and at the basolateral membrane of collecting duct cells. We propose that, in inner medullary collecting duct, the functional expression of HCN channels gives rise to this hyperpolarization-activated cationic current, and we discuss the possible participation of these cationic channels in cell-interstitium osmotic equilibrium and in the generation of the intracellular osmotic force required to drive vasopressin-stimulated water and urea reabsorption.

MATERIALS AND METHODS

Cell Culture

Primary cultures of rat inner medullary collecting duct (IMCD) cells were obtained using a modified hypotonic lysis method as described previously (12). Cells were plated on glass cover slips contained in 35-mm petri culture dishes and cultured in DMEM (GIBCO) supplemented with 10% FBS (GIBCO), antibiotics, and insulin at 37°C with an air-5% CO₂ atmosphere as described (7, 12). Cells were studied 6–11 days after plating. At this time, cells formed confluent cell monolayers exhibiting blister formation, an evidence of cell polarization and transepithelial transport. As described (12), electrophysiological recordings were performed in cells exhibiting (principal or IMCD cell) morphology (as evidenced by positive Dolichos biflorus lectin binding), the main cell population in our cultures.

RT-PCR

Total RNA was extracted from primary cultures and rat kidney inner medulla with the RNeasy Protect Mini Kit (Qiagen). After conventional reverse transcription with oligo(dT)₁₅ primer and SuperScript III RT (Invitrogen), HCN gene transcripts were detected by PCR using the following primers: for HCN1 (HCN1 forward 5′-CCAGCCCGGAGACTATATCA-3′ and HCN1 reverse 5′-GATTGGAGGGATCGCTTGTA-3′), for HCN2 (HCN2 forward 5′-GTGGAGCGAAGTCTATTCGT-3′ and HCN2 reverse 5′-GTCCTCGTCAAACATCTTCC-3′), for HCN3 (HCN3 forward 5′-TCGGACACTTTCTTCCTGCT-3′ and HCN3 reverse 5′-GGTTGAAGATGCGAACCACT-3′) and for HCN4 (HCN4 forward 5′-CGGCATGGTGAATAACTCCT-3′ and HCN4 reverse 5′-CCGCAACTTGTCAGCATAGA-3′). All primers were designed according to HCN sequences published in Genebank.

Amplification was initiated by denaturizing the sample for 5 min at 92°C, followed by 40 cycles of 1 min at 92°C, 1 min at 58°C, and 1 min at 72°C. Finally, samples were held at 72°C for 10 min to complete extension of all PCR products. The PCR products were purified with the PCR purification kit (Roche) and then were cloned into the pGEM-T Easy Vector System (Promega). The cloned DNAs were sequenced using the ABI PRISM 310 Genetic Analyzer to confirm HCN identity.

Immunocytochemistry

The distribution of HCN channel immunoreactivity in IMCD cultures and renal inner medulla was analyzed using an affinity-purified specific rabbit polyclonal antibody raised against human HCN2 (anti-HCN2; Alomone Laboratories, Jerusalem, Israel). In these experiments, glass cover slips previously coated with collagen (Type I; Sigma) were used to grow the cell cultures. Confluent IMCD cultures were fixed overnight with 4% paraformaldehyde in PBS (GIBCO). Renal inner medulla sections were prepared as previously described (12). Samples were preincubated 30 min in PBS with BSA (10 mg/ml) and then incubated 18–24 h at 4°C in PBS containing 0.25% Triton and the HCN2 antibody (4 μg/ml). After three washes with PBS (10 min each), samples were incubated with fluorescein isothiocyanate-conjugated secondary antibody (goat anti-rabbit IgG 1:100; Vector Laboratories) for 1 h. Additional experiments involved dual labeling of HCN channels with mouse anti-Na⁺-K⁺-ATPase (1:1,000; ab2867; Abcam). This labeling was visualized using CY3-conjugated goat anti-mouse secondary antibodies (1:100; Jackson Immunoresearch). Samples were mounted with Vectashield (Vector Laboratories) and viewed using standard epifluorescence and laser scanning confocal microscopy (Olympus FV-1000). Negative controls were IMCD cultures and renal inner medullary slices with either the omission of the primary antibody or with the primary antibody previously incubated with its control antigenic peptide (2 μg peptide/1 μg antibody; see Refs. 28 and 58).

Western Blot

HCN channel proteins also were detected in cell extracts from cerebral cortex, IMCD cultures, and renal inner medulla prepared as described briefly. Cells and tissues were homogenized in a lysis buffer (0.1 mM phenylmethylsulfonyl fluoride, 1 mM ethylenediaminetetraacetic acid, and 10 mM Tris·HCl; pH 7.6) by ultrasonic treatment. Next, samples were centrifuged at 10,000 g, and the supernatant was recovered. To enrich membrane fractions, extracts were centrifuged at 17,000 g for 20 min at 4°C; supernatant was discarded, and pellets were suspended in PBS. Protein concentration in samples was measured using the Coomassie method. From each sample, 50 μg of protein were size separated in a 10% SDS-PAGE and electroblotted to nitrocellulose membrane (Millipore). HCN2 were detected using standard methodology and the anti-HCN2 antibody (1:500; Alomone Laboratories). Protein bands were visualized with a chemoluminescent reaction system (Millipore). To confirm specificity of antibody, the same samples were incubated with antibody previously incubated with the control antigenic peptide (2 μg peptide/1 μg antibody).

Whole Cell Clamp Recordings

Membrane currents were routinely studied with the perforated-patch whole cell clamp technique, but a few experiments were performed with the conventional whole cell clamp technique, as previously described (12). Cover slips with a confluent cell monolayer were placed in a superfusion chamber and maintained in a standard (chloride-free) bath solution containing (in mM): 157 gluconic acid, 146 NaOH, 5 KOH, 2 Ca(OH)₂, 1 Mg(OH)₂, 10 HEPES, 10 glucose, and 30 μM amiloride, pH 7.4. Other bath solutions were used in some experiments: 1) a “no Ca²⁺ bath solution” containing no Ca(OH)₂; 2) a “K⁺-rich bath solution” containing 1 mM NaOH and 150 mM KOH; and 3) a “NMG-rich bath solution” containing 1 mM NaOH and 145 mM N-methyl-d-glucamine. Some experiments were performed in the chloride equivalents of the control (Cl⁻ control solution, see below) and NMG-rich bath (Cl⁻ NMG solution) solutions containing, in addition, 1 mM diphenyl-2-carboxylate (DPC) to avoid the contribution of Cl⁻ currents (39, 55). In conventional whole cell recordings, cells were maintained in a chloride bath solution (Cl⁻ control solution) containing (in mM): 146 NaCl, 5 KCl, 2 Ca(Cl)₂, 1 Mg(Cl)₂, 10 HEPES, 10 glucose, 1 DPC, and 30 μM amiloride, pH 7.4. Two other bath solutions were used in these experiments: a “45 mM K⁺ solution” containing 106 mM NaCl and 45 mM KCl; and a “146 mM K⁺ solution” containing 5 mM NaCl and 146 mM KCl. All experiments were performed at room temperature (20–25°C). Micropipettes, from Kimax-51 glass (Kimble), were filled from the tip with a pipette solution composed of (in mM): 156 gluconic acid, 141 KOH, 10 NaOH, 1.54 Ca(OH)₂, 1 Mg(OH)₂, 2.3 EGTA, and 10 HEPES, pH 7.4. Pipette filling was completed, from the back, with the same pipette solution containing, in addition, 200 μg/ml amphotericin B. In conventional whole cell clamp experiments, micropipettes were filled with a solution containing (in mM): 136.5 KCl, 4.5 KOH, 6 NaH₂PO₄, 0.1 Mg(Cl)₂, 0.1 EGTA, 2 Na₂ATP, 0.2 Na_x-GTP, 0.3 Na-cAMP, and 0.03 Na-cGMP, pH 7.3. Once filled, micropipettes had a resistance of 2–4 MΩ. Seals of at least 1 GΩ were obtained after pipettes had contacted the cell membrane and a gentle suction had been applied. Perforated-patch whole cell clamp configuration was obtained 4–8 min after the membrane contact, as monitored when a voltage square pulse (20 mV, 5 ms) evoked a capacitative current transient shorter than 4 ms, and a series resistance (R_s) smaller than 20 MΩ was measured. In conventional whole cell recordings, the membrane was ruptured by suction. Membrane potential was clamped at −50 mV. Membrane capacitance and R_s were compensated (80%) and measured using the Axopatch-1D compensation systems. The voltage-clamp protocols were generated, and the membrane currents were acquired with the Axopatch-1D under the control of the pClamp software (v.6; Axon Instruments) running in a Pentium 1 personal computer (Gateway 200) and using a Digidata 1200 A/D converter (Axon Instruments). Membrane currents were low-pass filtered (usually at 2 kHz), digitized (usually at a sampling rate of 416.7 Hz), and stored on the hard disk of the computer for subsequent analysis (12). Analysis was performed using the Clampfit module of pClamp, and curve fitting was performed using Sigmaplot (Jandel Scientific). As previously reported, the time course of the capacitative current (evoked by a pulse from −50 to −60 mV) exhibited a monoexponential decay, evidencing the absence of electrical coupling between cells, an indispensable condition for achieving space clamp. As reported, membrane capacitance was 24.3 ± 0.7 pF, and cell input resistance was 1.23 ± 0.12 GΩ (n = 121). A basic stimulation protocol was used in every cell: from a holding potential of −50 mV, a series of 720-ms voltage steps between −160 and 80 mV were applied in 20-mV increments and with 4-s intervals between the steps. The other protocols used are described below.

Current Kinetics Analysis

Hyperpolarization-activated current.

The hyperpolarization-activated current was obtained from current recorded at membrane potentials between −160 and −80 mV by subtracting instantaneous current (current values obtained within the first 0.4 ms of voltage pulses).

Instantaneous current.

Current recorded during the second half of voltages steps from −40 to 40 mV was plotted against voltage, and instantaneous linear slope conductance was calculated by linear regression. Cells exhibiting outward rectification (12) were excluded. Instantaneous linear current at voltages from −60 to −160 mV and at voltages of 60 and 80 mV was calculated by extrapolation. Inward-rectifying instantaneous current was calculated by subtracting linear current from the current measured at the onset of the voltage pulses.

Time course of the hyperpolarization-activated inward current.

The time course of the hyperpolarization-activated inward current was studied using instantaneous current-subtracted traces. The time course of current activation was fitted with a sum of two exponential functions

where I_t is the current measured at time = t, A is a constant related to the maximum value that can be reached by the hyperpolarization-activated and time-dependent current, and τ_f and τ_s are the time constants of activation, with τ_f being faster than τ_s. At membrane potentials more positive than −120 mV the time-dependent kinetics of the current could often be fitted with a single exponential function (Eq. 1 without the second exponential term).

Voltage dependence of the hyperpolarization-activated inward current activation.

Tail currents (at −50 mV) after each voltage step were linear current subtracted. The resultant tail currents (i_v) corresponding to each voltage step (V) were normalized as fractions of the tail current corresponding to the −160 mV step (i_-160). The normalized values were fitted with the following (Boltzmann type) equation

where c is a correction factor to account for a nonmaximal activation of the inward current at −160 mV, i_max is the estimated maximum value that can be reached by i_v, V_o is the voltage of half-maximal activation, and k is a constant that gives the steepness of the voltage dependence.

Tail currents and reversal potential.

From a holding potential of −50 mV, the current was activated with a prepulse to −160 mV during 1 s, and voltage then was returned to various test potentials ranging from −60 to 40 mV in 10-mV increments for 90 ms. Time-independent linear current, as measured from the recordings obtained with a similar voltage protocol in which the prepulse to −160 mV was omitted, was subtracted. The measured tail currents were plotted against voltage. The reversal potential of the tail currents was measured at the point where the curve that best fitted the plotted current points crossed the voltage axis.

Voltage ramps and reversal potential.

From a holding potential of −50 mV, the current was activated with a prepulse to −160 mV during 1.15 s. Voltage was returned to −60 mV for 3 ms, allowing occurrence of capacitative current, and then voltage was ramped to 60 mV over 30 ms. Time-independent linear current, as measured from the recordings obtained with a similar voltage protocol in which a prepulse to −40 mV was applied, was subtracted. The measured current response to the voltage ramps was plotted against voltage; the reversal potential was measured at the point where the current value crossed the voltage axis.

Statistical Analysis

All experimental results are expressed as means ± SE. Comparison among mean values was made by Student's t-test for paired or unpaired data. In some experiments, comparison among values was made by Wilcoxon signed test (Wst). Values of P < 0.05 were considered as an indication of a significant difference.

RESULTS

In the voltage range explored (from −160 to 80 mV), cells from IMCD cultures showed both inwardly and outwardly rectifying currents. Outward currents have been described previously (12). The present work focuses on voltage-activated inward currents.

A hyperpolarization-activated, time-dependent, inward current (I_vti) was observed in ∼50% of the studied cells (Fig. 1A). They activate at a potential close to −60 mV, require >720 ms to complete its activation at potentials between −80 and −160 mV, and appear to activate more quickly with larger hyperpolarizations (see below). Figure 1B shows the currents observed in I_vti-expressing cells: an instantaneous current (I_ins) exhibiting linear and inward-rectifying components and I_vti. I_ins has a rough voltage dependence, exhibiting inward rectification at potentials below −60 mV (Fig. 1C) and a time-dependent deactivation at potentials >20 mV (Fig. 1A). A current-voltage (I-V) relationship of the averaged total current amplitude of 40 I_vti-expressing cells is shown in Fig. 1C; note the presence of inward rectification and the zero current potential (E_0cur) close to 0 mV (−4.2 ± 1.3 mV) of the instantaneous linear current. This figure also shows the averaged amplitude of the inward rectification due to I_vti and I_ins; both currents exhibited inward rectification at potentials below −60 mV. An instantaneous current exhibiting inward rectification at potentials below −60 mV has been related to the cationic current I_h (4, 28, 31, 34, 42). To determine whether, in IMCD cells, I_ins was related to I_vti, we looked for a correlation between them. Figure 1D shows a plot of single cell I_ins linear slope conductance as a function of the corresponding I_vti chord conductance [determined at −140 mV, assuming a reversal potential (E_rev) = 0 mV, see below]. This plot revealed a significant correlation between the two currents (r = 0.79, slope = 0.47, P < 0.0001, n = 40). It suggests that both I_ins and I_vti may flow through the same channels, as suggested with respect to I_h and its related instantaneous current (25, 34, 42). Considering the ionic conditions in our experiments (virtual absence of any potentially permeating anion in the presence of 30 μM amiloride), it can be expected that I_vti is a K⁺ current or an amiloride-insensitive Na⁺ or a nonselective cation current. On the other hand, the E_0cur of I_ins suggests a basal activity of a nonselective channel exhibiting little or null discrimination between K⁺ and Na⁺.

To explore if I_vti is a Na⁺-carried current, we obtained recordings in the same cell before and after the Na⁺-rich standard bath solution was substituted for an NMG-rich solution. Figure 2, A and B, shows that I_vti was present when the cell was bathed in the standard solution but absent when bathed in an NMG-rich solution. Figure 2C illustrates subtraction (A − B) showing the Na⁺-carried currents. The effect of this ion substitution on the I-V relationship, averaged from seven cells, is shown in Fig. 2D. Note the absence of any inward-rectifying current when cells were studied in the NMG-rich bath; also note a change in linear conductance and in the E_0cur of I_ins. This ion substitution maneuver decreased the linear slope conductance (measured at ±40 mV, around E_0cur) from 2.3 ± 0.5 to 0.7 ± 0.1 nS (P < 0.02) and shifted I_insE_0cur from −7.2 ± 3.5 to −68.8 ± 17.3 mV (P < 0.02). This result suggests that, in our control condition, I_vti is mainly carried by Na⁺ and suggests that the I_vti channels have a constitutive basal activity that contributes to I_ins. On the other hand, the −61 mV shift of I_insE_0cur suggests the activity of a K⁺ conductance (as that previously described; see Refs. 20 and 48) contributing to some 30% of I_ins. Abolition of I_vti and its related I_ins induced by this ion substitution maneuver was not reversible, suggesting the intervention, in the observed response, of an unexpected inhibitory mechanism. Similar results were obtained in a series of similar experiments performed in chloride-containing solutions (Fig. 2, E-H), indicating that I_vti channels are not permeable to Cl⁻ and showing the absence of any inward-rectifying chloride current in the cultured cells studied in our chloride-containing control bath solution. We used these results as a control for our experiments with the conventional whole cell clamp (see below).

To gain more insight about the ionic permeability of I_vti channels, we performed, in the same cell, a tail current analysis at different potentials (Fig. 3A, top) at two ionic bath conditions: Na⁺-rich standard solution (records in middle) and K⁺-rich solution (records on bottom). As shown in Fig. 3B, the I-V relationship of these tail currents exhibited an E_rev close to 0 mV in both bath conditions, suggesting an Na⁺-to-K⁺ selectivity ratio close to unity. In the K⁺-rich condition, tail currents exhibited inward rectification and had higher amplitudes than in the control condition. This experiment was performed two times with similar results. To confirm the aforementioned Na⁺-K⁺ selectivity ratio, we employed the conventional whole cell clamp technique to assure a known intracellular Na⁺ and K⁺ concentration. These experiments were performed using chloride-containing solutions because of the lack of any time-dependent inward rectifying cationic current in the absence of intracellular chloride. Control perforated-patch whole cell clamp experiments were performed as described above (Fig. 2, E-H). In conventional whole cell clamp, I_vti was activated with a prepulse to −160 mV (Fig. 3C), and the current response to a voltage ramp (v-ramp response) from −60 to 60 mV was analyzed, in the same cells, at three ionic bath conditions (Fig. 3D): NaCl-rich standard solution (n = 4 cells), 45 mM K⁺ solution (n = 4), and 146 mM K⁺ solution (n = 3). The I_ins v-ramp response was measured and subtracted (see materials and methods). As shown in Fig. 3D, the I_vti v-ramp response exhibited an E_rev close to zero mV in every bath condition. Figure 3E shows the mean amplitude of the I_vti v-ramp response recorded in the three ionic conditions. Current exhibited inward rectification at any of these conditions, and current amplitude increased as external K⁺ concentration was increased. I_ins slope conductance was also increased (from 66.8 ± 12.6 nS in 5 mM K⁺ to 84.0 ± 13.5 nS in 45 mM K⁺, n = 22, P < 0.001; and to 85.4 ± 18.3 nS in 146 mM K⁺; n = 16, P < 0.01, Wst not shown). When external K⁺ concentration was decreased from 146 to 45 mM and further to 5 mM, both the I_vti v-ramp response amplitude and the I_ins slope conductance decreased as expected for a reversible response. From I_vti response amplitudes observed at −40 mV in 5 and 146 mM K⁺, a chord conductance ratio gK⁺-to-gNa⁺ close to 1.5 may be calculated, and from I_ins slope conductance a gK⁺-gNa⁺ ratio close to 1.3 may be inferred. The E_rev of I_vti response was only slightly modified (Fig. 3F). I_ins E_rev was also slightly modified. From these results, we conclude that I_vti, like I_h (11, 14, 40), is a nonselective cation current exhibiting a higher conductance (but, at membrane potentials close to E_rev, not a higher permeability) to K⁺ than to Na⁺.

The time-dependent kinetics of I_vti was studied using I_ins-subtracted traces. Figure 4A shows that the time course of current activation can be well described by a sum of two exponential functions (Eq. 1 in materials and methods). A sum of two exponential functions has been used to describe the time-dependent activation of I_h (4, 8, 40, 42, 58, 59). As reported for I_h, the time-dependent activation of I_vti shows clear voltage dependence, with τ_f and τ_s becoming smaller at more negative membrane potentials (Fig. 4B). From the time constants at −160 mV (mean: 71 and 342 ms), it can be calculated that, at the end of the 720-ms pulse, ∼93% of total time-dependent activation has occurred. Therefore, this pulse duration is adequate to study the I_vti time-dependent kinetics (8, 14, 62). On the other hand, the mean values of I_vti time constants at −120 mV (156 and 679 ms) are comparable to those reported for I_h at room temperature (4, 14, 51, 53, 58, 59, 62).

The voltage dependence of I_vti activation was well described by a Boltzmann-type equation (Eq. 2 in materials and methods; Fig. 4C). The mean values of the activation parameters were V_o = −102 ± 2 mV, k = 25 ± 1 mV, and c = 1.10 ± 0.01 (n = 23), so that within 720 ms at −160 mV the I_vti voltage-dependent activation was ∼ 90% of its full activation (note: when pulse duration was augmented to 4 or 6 s, n = 8, similar values for the activation parameters were obtained). As reported for I_h and I_f, the activation of I_vti starts at a potential close to −40 mV; however, I_vti does not become fully activated, even at −160 mV. Consequently, the k value of I_vti is larger than that of I_h and I_f, but its V_o is within the range reported for these currents (11, 14, 28, 31, 51, 58, 59, 62).

Despite the use of the perforated-patch whole cell clamp technique, I_vti characteristically suffers an apparently complete rundown 10–20 min after seal formation (Fig. 5, A–D). After complete rundown, inward rectification was almost undetectable, and I_ins linear slope conductance decreased from 3.42 ± 0.25 to 0.77 ± 0.09 nS (n = 5, P < 0.002). Figure 5, A and B, shows the initial (at t = 0 min) and final (at t = 20 min) recordings obtained in a representative cell. Figure 5C shows the subtraction (A − B) illustrating the currents abolished by rundown. Figure 5D illustrates the continuous decrement in total current observed in five cells during a period of 10 min and the final level reached by current after complete disappearance of I_vti (after 20 min). Figure 5E shows the continuous decrement observed (during the initial 10 min) in I_vti chord conductance (determined at −140 mV) and in I_ins linear slope conductance. During rundown development I_vti time-dependent activation became progressively slower; Fig. 5F shows the progressive increment observed in τ_f during the initial 10 min. I_vti voltage-dependent activation parameters were not affected by rundown during the initial 10 min of recording (data not shown). Rundown of I_vti made our inhibitor and stimulator studies more difficult.

Different types of amiloride-insensitive nonselective cation channels can be blocked, from the extracellular side, by any of the following cations: Ba²⁺, Mn²⁺, Cd²⁺, or La³⁺; on the other hand, I_h is blocked by Cs⁺ and ZD-7288 (4, 11, 14, 40). Therefore, we explored the sensitivity of I_vti to these blocking agents. Due to the occurrence of current rundown, it was important to be able to differentiate between the inhibitory effect of the agents and current rundown. Assuming a probable basolateral localization of I_vti channels and to assure a fast access of the agents to the basolateral membrane, these experiments were performed in a condition that favors the opening of tight junctions: cells were preincubated during 5 min in a no-Ca²⁺ bath solution plus 1 mM EGTA and maintained, thereafter, in the no-Ca²⁺ bath (without EGTA; see Ref. 26). Although, unexpectedly, in cells studied in this condition I_vti showed a slower rundown (after 1 h, current amplitude exhibited some 70% of its initial amplitude), its presence precluded the detection of a slowly developing or a weak inhibitory effect; 10 mM Cs⁺, 5 mM Ba²⁺, 5 mM Mn²⁺, 0.5 mM La³⁺, or 100 μM ZD-7288 had no appreciable inhibitory effect on either I_vti or the instantaneous current. On the contrary, 5 mM Cd²⁺ rapidly (within 2 min) blocked most of the I_vti and instantaneous currents, and this effect was, at least partially, reversible. Figure 6 illustrates this blocking effect of Cd²⁺. I_vti was virtually absent in the presence of Cd²⁺ (Fig. 6B), and linear current slope conductance diminished from 4.16 ± 0.96 to 0.42 ± 0.05 nS (n = 5, P < 0.005); thus, the Cd²⁺-sensitive I_ins conductance accounted for some 90% of total linear conductance. The E_0cur of the linear current was also shifted from −7.0 ± 3.6 to −60.0 ± 8.5 mV (P < 0.001), suggesting that, in the presence of Cd²⁺, instantaneous current is mainly due to the activity of a K⁺-selective channel. Thus I_vti seems to be a Cd²⁺-sensitive and Cs⁺-resistant I_h-like current.

One of the most outstanding properties of I_h is its regulation by cyclic nucleotides (1, 21, 32, 41). To explore if I_vti exhibits this property, the cAMP membrane-permeant analog 8-bromo-cAMP (8-Br-cAMP) was employed. Figure 7, A –F, illustrates the I_vti regulation by cAMP. We obtained I_vti recordings in the same cell before (Fig. 7A) and after (Fig. 7B) its exposure to 8-Br-cAMP (0.3 mM). Figure 7C shows that total inward current increased when cells were exposed to 8-Br-cAMP. I_ins linear slope conductance changed from 2.8 ± 0.4 to 3.3 ± 0.3 nS (n = 6; P < 0.05, Wst). I_vti amplitude increased in every cell studied (Fig. 7D). I_ins inward-rectifying component increased in five of the six cells studied. In the presence of 8-Br-cAMP, the fast component of I_vti time-dependent activation kinetics became faster in every cell studied. Figure 7E shows that τ_f became smaller in this condition. The slow component of this activation showed no consistent change. In the presence of 8-Br-cAMP, the I_vti voltage-dependent activation curve started at more positive values (Fig. 7F), without a change in the voltage of maximal activation, so that, while V_o shifted to the right (from −90.3 ± 2.5 to −79.4 ± 2.7 mV, n = 6; P < 0.03, Wst), k increased from 18.8 ± 1.3 to 21.5 ± 1.8 mV (P < 0.03, Wst). The presence of 8-Br-cAMP had no effect on the rundown of I_vti, as has been reported to occur in the rundown of I_h (4). These results indicate that I_vti channels can be regulated by cAMP. I_vti responded to cAMP stimulation in a similar way as I_h and I_f (14, 40, 42, 53, 56, 59, 62). Furthermore, the instantaneous conductance associated with I_vti, like that associated with I_h (34, 35, 42), increased also in response to cAMP. It supports our proposition that I_vti is a Cs⁺-resistant I_h-like current closely related to an instantaneous current.

I_f and I_h are potentiated when the extracellular K⁺ concentration is increased (11, 14, 40). Figure 8, A–D, shows that I_vti exhibits a similar behavior. We obtained I_vti recordings in the same cell before (Fig. 8A) and after (Fig. 8B) the standard bath solution was changed to a K⁺-rich solution. Figure 8C shows that total inward current increased when cells were exposed to a K⁺-rich solution. I_vti increased from −391 ± 106 to −627 ± 161 pA at −140 mV (n = 5; P < 0.05, Wst; data not shown). I_ins linear component was not affected, but its inward-rectifying component increased from −119 ± 33 to −185 ± 34 pA at −140 mV (P < 0.05, Wst; data not shown). In the K⁺-rich solution, I_vti time-dependent activation kinetics became faster. Figure 8D shows that τ_f became smaller. τ_s also changed from 530 ± 68 to 269 ± 53 ms at −140 mV (P < 0.02, Wst; data not shown). However, this ion substitution had no effect on the voltage-dependent activation curve of I_vti. These results indicate that the I_vti channels can be regulated by the extracellular K⁺ concentration, as has been reported to occur with the I_f and I_h channels (11, 10, 14, 40).

I_h and I_f flow through HCN channels (21, 41, 45); therefore, we looked for the expression of HCN genes in IMCD cells. An RT-PCR analysis was performed using specific primers for the four rat HCNs (see materials and methods). Transcripts corresponding to the four HCNs were observed both in the cultured cells and the kidney inner medulla (Fig. 9A); the fragments size was as expected, although, HCN4 fragments showed small differences in size between them. The sequence of the HCN1, HCN2, and HCN4 (but not that of the HCN3) fragments corroborated the identity of the genes. These results suggest that HCN channels mediate I_vti.

To explore the possible translation of HCN-mRNA in IMCD cells, we searched for the presence of HCN channel protein immunoreactivity. Based on I_vti activation kinetics, on its cAMP sensitivity, and on the presence of an I_vti-related I_ins (34, 40, 54), we chose to look for HCN2 immunoreactivity using a HCN2 specific antibody, Western blot, and confocal microscopy. Western blot analysis identified HCN2 immunoreactive products in protein samples from brain, inner medulla, and cell cultures (Fig. 9B). The main band detected in brain and primary culture corresponds to proteins with masses of ∼120 kDa (as expected for HCN2; see Refs. 30, 38, 61, and 62), and a weak ∼90-kDa band was, also, detected in cell cultures. However, in the inner medulla, an ∼90-kDa band was mainly observed, probably representing a modified or variant form of the HCN2 protein, whereas a weakly labeled ∼120-kDa band was also observed. Next, our Western blot results show that, in the renal inner medulla and in our cultured cells, the HCN genes are translated to proteins. They also suggest that, in renal medullary extract, the HCN2 polypeptide undergoes proteolytic or posttranslational modification, or possibly represents a transcriptional variant.

To further support the proposal that I_vti flows through HCN channels, we looked for the presence of HCN proteins at the membrane of the IMCD cells. Using confocal microscopy, HCN2 immunoreactivity was detected both at the membrane and in the cytoplasm of IMCD cells, as has been reported to occur in excitable cells and in nonexcitable transfected cells (28, 34, 38, 61, 62). Figures 10 and 11 show that HCN2 immunolabeling partially colocalized with that of the Na⁺-K⁺-ATPase both in the cultured cells (Fig. 10, A, B, and C) and the “in situ” IMCD (Fig. 11, A, B, and D). Interestingly, this partial colabeling was also present at the papillary epithelium (Fig. 11D), suggesting the presence of HCN channels (and expression of I_vti) in this epithelium. Colabeling of a protein with the Na⁺-K⁺-ATPase is usually interpreted as evidence of its localization at the basolateral membrane of epithelial cells (37, 60), and it seems to be the case in the in situ IMCD. However, because in our cultured cells the Na⁺-K⁺-ATPase labeling seems to be distributed mainly inside the cell, we cannot rigorously apply this interpretation to our cultures. Anyhow, our results are consistent with mediation of I_vti by HCN channel polypeptide(s) located at or mobilizable to the basolateral membrane of IMCD cells.

DISCUSSION

Our work demonstrates the presence of a hyperpolarization-activated, cyclic nucleotide-gated, inward cationic current in IMCD cells in primary culture, the expression of HCN RNAs in these cells, and in the renal inner medulla, the presence of ∼90- and ∼120-kDa HCN2 immunoreactive proteins in inner medulla and primary cultures, as well as the presence of HCN2 immunoreactivity in the IMCD cells both in situ and in culture. I_vti exhibits some characteristics resembling those of I_h and I_f, and some features that differentiate it from those currents, nonetheless, the presence of HCN transcripts and HCN immunoreactivity in the IMCD cells, lead us to propose that I_vti flows trough HCN channels. To the best of our knowledge, this is the first study reporting an I_h-like cationic current (probably mediated by HCN channels) in either kidney epithelial cells or any other nonexcitable mammalian cell.

Like I_h and I_f observed in brain and heart (11, 14, 58, 62), I_vti activates at hyperpolarizing voltages in a time-dependent manner, exhibits a slow time course, and may be carried by Na⁺ and K⁺, showing a higher conductance to K⁺. Like I_h and I_f, I_vti starts its activation at a potential close to −40 mV, but in the presence of a high cAMP concentration its activation starts close to 0 mV. This range may have a remarkable physiological role in the nonexcitable cells of the IMCD. According to the observed basal activation onset potential, it is possible that a fraction of the I_vti channels may be active at our basal holding potential of −50 mV, and this fraction may explain, at least partially (70–90%, as inferred from our NMG and Cd²⁺ results), the instantaneous current we observed. That I_vti and a cation-nonselective I_ins may flow through the same channels is suggested by: 1) the positive correlation between linear instantaneous conductance and hyperpolarization-activated, time-dependent conductance; 2) their parallel abolition when most extracelullar Na⁺ was replaced by NMG; 3) their similar E_rev, which exhibited a similar change when extracelullar K⁺ was increased; 4) their higher conductance for K⁺ than for Na⁺; 5) their parallel decrement during I_vti rundown development; 6) their increment in the presence of 8-Br-cAMP; and 7) their similar blockade by Cd²⁺. Regarding I_h and its related instantaneous current, they are produced by two functionally distinct populations of the same channel (25, 34, 35), with the instantaneous current being Cs⁺ insensitive. On the other hand, introduction of a cysteine residue in the HCN2 channel pore renders the instantaneous current sensitive to Cd²⁺ blockade (35). Our results do not support the hypothesis that, in our cultured cells, two distinct populations of the channel produce I_vti and its related cation-nonselective I_ins because all experimental maneuvers, and rundown, affecting I_vti had similar effects on I_ins. The proposal that a single population of I_vti channels produces both currents is also supported by the I_ins time-dependent deactivation observed at potentials more positive than 20 mV. This time-dependent deactivation seems similar to that observed in I_h when membrane potential is changed from a hyperpolarized one to a depolarized one (34, 25).

This study demonstrates the presence of mRNA corresponding to HCN1, -2, and -4 in our cultured cells and inner medulla. In a general way, this result agrees with previous observations of HCN-RNA in kidney cells (44, 57). We also show that IMCD cells express antigens compatible with the presence of HCN2 channel proteins at the basolateral membrane. HCN proteins may form heteromeric channels with a possible tetrameric structure (3, 27, 54); hence, it is conceivable that I_vti channel is a heteromultimer composed by α-subunits of HCN1, HCN2, and HCN4. An alternative hypothesis could be that I_vti is a mixed current flowing through HCN1, HCN2, and HCN4 homomeric channels, but this would imply that all of them are Cs⁺-resistant and Cd²⁺-sensitive channels. The block characteristics of I_vti may arise from the presence of splice variants of the HCN channels expressed in excitable cells or from posttranslational modification of channel proteins, as suggested by our Western blot results (45). In this regard, intracellular Cd²⁺ blocks the I_h current from HCN2 and spHCN, and addition of cysteine residues to the HCN pore increases its sensitivity to Cd²⁺ block (15, 16, 43).

To explore possible physiological roles for I_vti channels, we have to focus our attention on I_vti voltage-dependent activation onset potential. It is close to −40 mV in our control condition and close to 0 mV during cAMP stimulation. These values are within the range of basolateral membrane potential (−80 to 20 mV) exhibited by the IMCD cells (20, 48, 52). I_vti channels may thus activate in physiological conditions and play a role in determining the basolateral membrane potential. Due to the nonexcitable nature of IMCD cells, an I_vti pacemaker function seems unlikely. However, a cyclic contractile activity of pelvocalyceal smooth muscle has been related to inner medullary urine concentrating ability (23, 36, 46), and one may speculate that I_vti channels in IMCD (and, probably, papillary epithelium) exhibit a cyclic activity somehow related with that smooth muscle contraction pattern.

According to a “classic” point of view, if I_vti channels influence the basolateral membrane ionic permeability, they may participate in the IMCD cell-interstitium osmotic equilibration. IMCD basolateral membrane is immersed in a hyperosmotic interstitial fluid with a high Na⁺, Cl⁻, K⁺, and urea concentration (5, 9, 18, 22, 24). When the kidney acutely passes from a chronic diuresis state to an antidiuresis state, the inner medullary interstitial solute concentration is slowly raised (6, 18, 47). Based on previous observations (2, 13, 47), it may be proposed that the Na⁺ and Cl⁻ transport activity of the thin ascending limb of Henle's loop provides the basis for an early increment in inner medullary interstitial Na⁺ and Cl⁻ concentration. Next, during the early phase of acute antidiuresis, the IMCD cells become exposed to a slowly developing interstitial hypertonicity, and its membrane transport characteristics become influenced by vasopressin and its second messenger cAMP (6, 18, 47, 49). In this situation, an Na⁺ and Cl⁻ influx through the basolateral membrane of IMCD is thought to allow a rapid cell-interstitium osmotic equilibration, with only minimal and transient cell volume change (6, 17, 47, 50). Na⁺ and Cl⁻ influx may endow the IMCD cells with the necessary osmotic force to drive the vasopressin-stimulated apical water and urea reabsorption (13, 29). At least a small fraction of this Na⁺ influx may occur through the I_vti channels now stimulated by the increased cAMP levels. Alternatively, one may propose that an important fraction of this Na⁺ influx occurs through I_vti channels, accompanied by an anion influx through the basolateral HCO₃⁻ conductance previously described (20, 48). The resulting Na⁺ and HCO₃⁻ intracellular concentration increase could then stimulate the activity of both the Na⁺/H⁺ and the Cl⁻/HCO₃⁻ exchanger at the basolateral membrane, alleviating the induced cell alkalinization. This proposal would explain the previously observed participation of those mechanisms in IMCD cell volume regulation (17, 50), as well as the observed increment in intracellular Cl⁻ concentration (6, 17, 47). Intracellular Na⁺ increment, and a mild cell alkalinization, may also stimulate the basolateral Na⁺-K⁺-ATPase, conducting to replacement of Na⁺ by K⁺, and to an increment in intracellular K⁺ concentration, as previously observed (6, 47). The functional coupling of channel-mediated Na⁺ and HCO₃⁻ fluxes could explain the absence of intracellular K⁺ content increment observed, in isolated cells, in presence of no added HCO₃⁻ external solutions (17). Finally, this hypothesis is also in accordance with an IMCD basolateral membrane Na⁺-dependent HCO₃⁻ influx previously observed (19, 33).

GRANTS

This work was partially supported by Dirección General de Asuntos de Personal Académico (DGAPA), Universidad Nacional Autónoma de México, Grants IX211304 and IN224106 to J. J. Bolívar.

FOOTNOTES

• The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.