Irritant contact dermatitis is an innate inflammatory cutaneous overreaction to corrosive chemicals applied to the epidermis. Normal human epidermal keratinocytes (NHEKs) are used for toxicity testing of industrial products without using living animals and give insights into the mechanisms of irritant dermatitis. In vitro tests with NHEKs and with murine and human keratinocyte cell lines demonstrated that irritant and corrosive compounds induce secretion of the inflammatory mediators cytokines and ATP (Mizumoto et al., 2002, 2003; Coquette et al., 2003). Depending on the chemical nature of the irritant, the release of ATP by keratinocytes may or may not be mediated by plasma membrane disruption (Mizumoto et al., 2003). Many of the chemicals causing ATP leakage through cell lysis contain phenol residues (e.g., butylphenol, carvacrol) or long-chain hydrophobic hydrocarbons (e.g., octanoic acid, cetyltrimethylammonium bromide). Irritant molecules causing ATP release without cell lysis include amine-containing weak organic bases (e.g., aminoethyl-piperazine, heptylamine, diamine propane) and inorganic bases (e.g., sodium and potassium hydroxide) (Mizumoto et al., 2003). How they act on NHEKs and induce inflammatory mediators release remains poorly understood.

In a variety of cell types, volume-regulated anion channels (VRACs) are permeable to metabolites, some of which are proinflammatory mediators. These metabolites include amino acids (e.g., glycine, glutamate, and aspartate), polyols, lactate, bicarbonate, and ATP (Grygorczyk and Guyot, 2001; Hisadome et al., 2002; Darby et al., 2003; Nilius and Droogmans, 2003; Ullrich et al., 2006). VRACs also participate in the regulation of fundamental cell functions, particularly the regulatory volume decrease during cell swelling (Sardini et al., 2003) and the control of proliferation and apoptosis. VRAC inhibition suppresses cell proliferation of T lymphocytes, microglia, and endothelial cells, and its activation regulates normotonic cell shrinkage during apoptosis (Maeno et al., 2000; Okada et al., 2001, Eggermont et al., 2001). Moreover, VRAC can be partially activated under isotonic conditions contributing to the setting of membrane potential (Nilius and Droogmans, 2003; Suzuki et al., 2006). VRAC has been described in the keratinocyte cell line HaCaT (Zholos et al., 2005) and in NHEKs (Rugolo et al., 1992). Thus, modulation of VRAC by irritants may interfere with cellular processes of keratinocytes and contribute to ATP release and inflammatory cutaneous reaction. Therefore, we investigated whether VRACs are molecular targets for irritant chemicals.

In the present article, we report that VRACs in NHEKs are modulated by heptylamine, one of the referenced irritants causing ATP release without cell lysis. We developed experimental conditions to study VRAC current (I_VRAC) in relative isolation from other conductances in NHEKs in primary cultures. We observed that heptylamine reversibly inhibits VRAC activity and causes Ca²⁺-store depletion. This inhibition by heptylamine decreased during VRAC up-regulation due to hypotonic stress, suggesting that heptylamine interferes with the volumetric gate of VRAC. These results provide evidence that the referenced chemical irritant heptylamine acts on NHEKs through intracellular Ca²⁺ mobilization and inhibition of VRAC activity.

Materials and Methods

NHEK Primary Cultures. NHEK cultures were prepared from healthy human prepuce biopsies. Skin explants were conserved at 4°C for up to 72 h in Hanks' balanced salt solution (HBSS; Invitrogen, Carlsbad, CA) supplemented with penicillin/streptomycin (200 U/ml; Invitrogen). The mesenchyme under the dermis was removed, and the explants were decontaminated by washes in HBSS supplemented with ethanol (30%, v/v) and in HBSS supplemented with penicillin/streptomycin (200 U/ml). Explants were then cut in ∼0.5-cm² pieces and incubated for 16 to 20 h at 4°C with dispase grade II in phosphate-buffered saline (2.4 U/ml; Roche, Indianapolis, IN) supplemented with penicillin/streptomycin (100 U/ml). Epidermal layers were lifted from the dermis, mechanically dissociated, and incubated in HBSS supplemented with trypsin-EDTA (0.5 mg/ml; Invitrogen) for 20 min at 37°C. The resulting epidermal cell suspension was filtered through a cellular sieve (pore size, 70 μm) and centrifuged at 1200 rpm (200g) for 6 min. Epidermal cells were resuspended and cultivated in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with Ham's F-12 (30%; Invitrogen), fetal calf serum (10%; Invitrogen), l-glutamine (2 mM; Invitrogen), penicillin/streptomycin (100 U/ml), insulin (5 μg/ml; Sigma, St. Louis, MO), hydrocortisone (0.5 μg/ml; Sigma), and epidermal growth factor (10 ng/ml; Sigma). To obtain NHEK proliferation, cells were plated at 80 × 10³ cells/cm² and cocultured with feeder mitomycin C-treated 3T3 fibroblasts (30 × 10³ cells/cm²) in Falcon flasks (37°C, 95% O₂). The culture medium was renewed every 2 to 3 days.

After 8 to 15 days in vitro, subconfluenced NHEKs were isolated for cryopreservation using a two-step trypsinization procedure: cocultures were treated with trypsin-EDTA in HBSS (0.5 mg/ml) for ∼1 min to dislodge inactivated 3T3 fibroblasts, and then NHEKs were dislodged by a longer trypsin-EDTA treatment (∼3 min). NHEKs were centrifuged at 1200 rpm (200g) for 6 min and resuspended in Dulbecco's modified Eagle's medium supplemented with fetal calf serum (20%) and dimethyl sulfoxide (10%). NHEKs were finally cryopreserved at 3 × 10⁶/ml in cryovials.

For experiments, cryopreserved NHEKs were thawed, plated at 13 × 10³/cm² in Falcon flasks, and cocultured with feeder mitomycin C-treated 3T3 fibroblasts (13 × 10³/cm², 37°C, 95% O₂) for 5 to 20 days. Keratinocytes grown to a confluent monolayer become electrically coupled via gap junctions, and this precludes voltage-clamp analysis (Kam et al., 1987). Therefore, NHEKs were isolated by the two-step trypsinization procedure and plated at 10 × 10³/cm² in Nunclon dishes (Thermo Fisher Scientific, Waltham, MA) 3 to 30 h before patch-clamp experiments.

Whole-Cell Patch-Clamp Recording. Whole-cell NHEK currents were measured at room temperature (20–22°C) with an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA). Recordings were sampled at 20 kHz and filtered at 1 kHz. Patch pipettes pulled from borosilicate glass had resistance of 8 to 11 MΩ. The access resistance was 12 to 25 MΩ. Voltage errors were minimized using ∼80% series resistance compensation. Currents were typically recorded using two voltage-clamp protocols: a depolarizing voltage ramp from –120 to +40 mV (or +100 mV, when specified) with a rising rate of 0.45 V/s; and voltage steps (200 ms) from –120 to +100 or +130 mV with a 10-mV increment. Steps were preceded by a 125-ms prepulse at –120 mV to remove inactivation. Depolarizing voltage ramps were repeated every 4 s. The holding potential was –60 mV.

Pipettes were filled with an internal solution consisting of 130 mM KCl, 10 mM HEPES, 1 mM MgCl₂, 4 mM MgATP, 0.4 mM Na₂GTP, 0.5 mM CaCl₂, and 10 mM EGTA. The free cytosolic Ca²⁺ concentration ([Ca²⁺]_i) estimated with the free software WEBMAXCLITE version 1.15 (http://www.stanford.edu/~cpatton/webmaxc/webmaxclite115.htm) was 10 nM. Most experiments were carried out using this low [Ca²⁺]_i to avoid activation of Ca²⁺-dependent Cl^– and K⁺ channels of keratinocytes (Koegel and Alzheimer, 2001; Zholos et al., 2005). In some experiments, [Ca²⁺]_i was set at 100 nM using 2.8 mM CaCl₂ and 5 mM EGTA or to 2 or nominally 0 nM using 20 mM EGTA and 0.5 or 0 mM CaCl₂, respectively (the KCl concentration was adjusted to avoid osmolar variations). Internal anionic substitutions were obtained by replacing 130 mM KCl with equimolar KF or potassium gluconate (KGlu). In some experiments, 0.1% (v/v) heptylamine was added to the internal solution. The pH value of internal solutions was adjusted to 7.35 with KOH or HCl, and the osmolarity was 308 ± 5 mOsM. A slightly hypotonic internal solution (290 ± 5 mOsM) was obtained by reducing the KCl concentration.

The external isotonic solution (ITS, 296 ± 5 mOsM) consisted of 128.5 mM NaCl (or choline chloride, when specified), 3 mM KCl, 1 mM MgCl₂, 10 mM HEPES, 2.5 mM CaCl₂, and 10 mM glucose. Increase in the external KCl concentration from 3 to 27 mM was obtained with an equimolar reduction in NaCl. The external hypotonic solution (HTS; 236 ± 5 mOsM) consisted of 98.5 mM NaCl, 3 mM KCl, 1 mM MgCl₂, 10 mM HEPES, 2.5 mM CaCl₂, and 10 mM glucose. Adding 60 mM d-mannitol to HTS allowed to obtain the corresponding ITS (296 ± 5 mOsM) without changing ionic concentrations. The pH values for both ITS and HTS were adjusted to 7.35 with NaOH.

All chemicals were purchased from Sigma. Heptylamine is a straight-chain primary alkyl amine (Scheme 1). Heptylamine solutions were prepared in ITS or HTS, with pH being adjusted to 7.35 with HCl. 4,4′-Diisothiocyanatostilbene-2,2′-disulfonic acid disodium salt hydrate (DIDS, 200–400 μM) and histamine (1 mM) were solubilized in water.

Cell-Attached Patch-Clamp Recording. For cell-attached patch experiments, pipettes had resistances of 6 to 10 MΩ when filled with a solution containing 128.5 mM NaCl, 3 mM KCl, 1 mM MgCl₂, 10 mM HEPES, 2.5 mM CaCl₂, and 10 mM glucose. The pH value was adjusted to 7.35 with NaOH, and the osmolarity was 296 ± 5 mOsM. External solutions were the same as those for whole-cell patch-clamp experiments. Recordings were sampled at 20 kHz and filtered at 2 kHz. In cell-attached patch experiments, voltages given correspond to –V_pipette and are relative to the resting potential (V_rest). Currents were recorded by using a depolarizing voltage ramp from –100 to +100 mV with a rising rate of 2 V/s repeated every 5 s, with the holding potential set at –40 mV. Some experiments were carried out in NHEKs loaded with EGTA AM (100 μM; Tebu Bio, Le Perray en Yvelines, France) for 30 min (37°C, 95% O₂).

Calcium Imaging. For calcium imaging experiments, NHEKs were plated at 10 × 10³/cm² in 96-well Falcon dishes and loaded with Fluo-4 AM (5 μM; Invitrogen) for 30 min (37°C, 95% O₂). Experiments were performed in ITS containing either 2.5 or 0 mM CaCl₂. Images were acquired using an inverted microscope (Nikon Eclipse TE 300) and processed using Nikon Lucia analysis software (Nikon, Tokyo, Japan). The Fluo-4 fluorescence was recorded by exciting the probe at 485 nm. Heptylamine was applied to wells at a final concentration of 0.1%. In control experiments, the same volume of heptylamine-free ITS was applied. The fluorescence intensity (F) normalized to basal fluorescence (F0), expressed as F/F0, was used to monitor [Ca²⁺]_i variations. F/F0 variations were measured in 10 to 12 cells per wells.

Data Analysis. NHEK currents were analyzed using pCLAMP 9 software (Axon Instruments) and PRISM 4.0 software (GraphPad Software Inc., San Diego, CA). Junction potentials calculated with pCLAMP 9 software were 5.3, 17.4, and 9.6 mV at 22°C for KCl, KGlu, and KF internal solutions, respectively. To compare null-current voltages and reversal potentials of current-voltage (I-V) relationships under different conditions, voltages were corrected for junction potentials in Figs. 1, c to e, and 3d. In other experiments with KCl internal solution, voltages were not corrected. We refer to reversal potential in the case of a pure (i.e., pharmacologically isolated) current and to null-current voltage in the case of a composite macroscopic current.

The modified Goldman-Hodgkin-Katz equation was used to calculate the permeability ratio P_Glu/P_Cl of gluconate anions (Glu^–) relative to that of Cl^–: P_Glu/P_Cl = {[Cl^–]_n exp(–ΔE_RevF/RT) – [Cl^–]_s}/[Glu^–]_s, where [Cl^–]_n and [Cl^–]_s are the Cl^– concentration in the normal and the substituted internal solutions, respectively, [Glu^–]_s is the concentration of Glu^– in the substituted internal solution, ΔE_Rev is the difference of the reversal potentials for Cl^– and Glu^–, F is the Faraday constant, R is the gas constant, and T is the absolute temperature.

Dose-response curves were fitted by the Hill equation Y = Y_max + (Y_max – Y_min)/(1 + 10^{((logIC₅₀–log[heptylamine])×H)}), where Y is the percentage of current blocked (I_blocked/I_ctr.), IC₅₀ is the dose of heptylamine necessary to obtain 50% of Y_max, and H is the Hill slope. Conductances were calculated between +30 and +40 mV. All results are presented as mean ± S.E.M., with error bars and gray areas in figures representing S.E.M.

Results

Characterization ofI_VRAC in NHEKs. Because I_VRAC has been described in many cell types but poorly investigated in keratinocytes, we first sought to isolate and characterize this current in NHEKs. In all experiments, cells were dialyzed with an internal patch pipette solution containing 4 mM ATP, which is required for I_VRAC activation (Lewis et al., 1993; Bond et al., 1999; Bryan-Sisneros et al., 2000). To minimize Ca²⁺-activated Cl^– and K⁺ channels that are expressed in keratinocytes (Koegel and Alzheimer, 2001; Zholos et al., 2005), [Ca²⁺]_i was set to 10 nM. To differentiate Cl^– conductances from others, internal and external Cl^– were symmetrical, giving a Cl^– reversal potential (E_Cl) of ∼0 mV. Immediately after patch rupture, the whole NHEK current was dominated by a very small voltage-independent leak current (–4.86 ± 0.53 pA/pF at –120 mV and 3.20 ± 0.26 pA/pF at +40 mV, conductance 0.050 ± 0.005 nS/pF, n = 16) (Fig. 1a). In all cells, the current began to increase after 1 min of establishing the whole-cell mode and reached a plateau after 15 min at 0.56 ± 0.05 nS/pF (n = 16). The I-V relationship for currents stable for at least 5 min exhibited outward rectification and, unlike the leak current, showed clear inactivation at potentials greater than +40 mV (bottom inset, Fig. 1a). The whole-cell current had a null-current voltage near E_Cl at 2.18 ± 0.55 mV, and the current density was –28.66 ± 2.34 pA/pF at –120 mV and 18.02 ± 1.53 pA/pF at +40 mV (n = 16). This run-up current was also observed when the osmolarity of the internal solution (308 ± 5 mOsM) was decreased to slightly hypotonic values (290 ± 5 mOsM) (data not shown). This run-up current was inhibited by 200 μM DIDS, a specific blocker of Cl^– channels, in a voltage-dependent manner: the current was blocked to a greater extent at positive (74.22 ± 2.63% at +40 mV, n = 7) than at negative potentials (29.99 ± 8.06% at –120 mV, n = 7) (Fig. 1b). Ca²⁺-activated K⁺ currents have been described in keratinocyte cell lines (Koegel and Alzheimer, 2001), so we tested whether, under our conditions (i.e., 10 nM [Ca²⁺]_i), recordings were contaminated by such conductances. No K⁺ currents could be detected, because increasing the external K⁺ concentration from 3 to 27 mM in the presence of DIDS had little effect either on the whole-cell conductance or on the null-current voltage (Fig. 1b). In addition, nonselective cation conductance was absent because replacing 128.5 mM external NaCl with equimolar choline chloride had no effect on the whole-cell current (n = 4; data not shown).

The ionic composition of the intracellular solution was then changed to determine the dependence of the current magnitude and null-current voltage on the intracellular Cl^– concentration. For these experiments, we routinely used a voltage ramp protocol running from –120 to +40 mV in 400 ms, and voltages were corrected for junction potentials. Equimolar exchange of 130 mM intracellular Cl^– with fluoride (F^–) and Glu^– negatively shifted the null-current voltage of the I-V curve by 18 and 53 mV, respectively [null-current voltages, –3.12 ± 0.55 mV for KCl (n = 16), –21.17 ± 1.42 mV for KF (n = 3), and –56.17 ± 3.29 mV for KGlu (n = 8)] (Fig. 1c). These findings indicate a substantial anion permeability of the channel with the following permeability sequence: Cl^– > F^– > Glu^–. This was confirmed by estimating the reversal potential of the DIDS-sensitive current recorded with intracellular Cl^– or Glu^– (130 mM). Indeed, the reversal potential of the DIDS-sensitive current was shifted from –4.64 ± 0.96 mV (n = 6) to –47.96 ± 1.36 mV (n = 5) when internal Cl^– was replaced with Glu^– (Fig. 1d). This change in the reversal potential corresponds to a P_Glu/P_Cl of 0.4. Overall, these findings indicate that the whole-cell current records made with a high Ca²⁺-buffering internal solution and 4 mM ATP were dominated by a Cl^– current with very little, if any, contribution from other K⁺ or cation currents.

We also examined the contribution of the Cl^– current to the whole-cell current when the [Ca²⁺]_i was 100 nM. Under symmetrical Cl^– conditions, the null-current voltage of the whole-cell I-V curve was slightly shifted toward more negative values: –3.12 ± 0.55 mV (n = 16) to –17.00 ± 4.58 mV (n = 9) (Fig. 1e). The amplitude of the total current recorded at +30 mV was increased by 59% (from 15.07 ± 1.27 pA/pF, n = 16, to 24.02 ± 3.88 pA/pF, n = 9), which indicates that additional components contribute to the whole-cell current. In the presence of DIDS, increasing the external K⁺ concentration from 3 to 27 mM slightly depolarized the null-current voltage of the I-V relationship (from –10.25 to –3.00 mV for the example depicted in Fig. 1f). However, DIDS still blocked 65% of the outward current recorded at +40 mV, indicating that this K⁺ current component contributed, albeit moderately, to the whole-cell current (Fig. 1f). These results demonstrate that recordings with 100 nM [Ca²⁺]_i were contaminated by Ca²⁺-dependent K⁺ current components. Therefore, all subsequent experiments investigating I_VRAC in NHEKs were performed using 10 nM [Ca²⁺]_i. The properties of this Cl^– current (i.e., rectification in the outward direction, inactivation at positive potentials, voltage-dependent block by DIDS, and permeability sequence) are in some ways comparable with those found for VRACs in other cell types (Voets et al., 1997; Shuba and McDonald, 2000; Jentsch et al., 2002; Nilius and Droogmans, 2003).

We confirmed that VRAC carried this Cl^– current by showing that its conductance was increased by exposure to an HTS (236 ± 5 versus 296 ± 5 mOsM). Figure 2a illustrates the changes induced by HTS on the whole-cell current of a representative cell during successive osmotic challenges. The swelling-sensitive current typically started to increase ∼1 to 2 min after the change of solution and reached a plateau within 463 ± 36 s (n = 8) of the beginning of the hypotonic stress. Once the full effect of hypotonicity had developed, the conductance was 0.87 ± 0.11 nS/pF, and the current was 2.65 ± 0.31 and 2.75 ± 0.39 times greater in magnitude, respectively, at –120 and +40 mV (n = 12). The current increase was entirely reversed on return to ITS (Fig. 2a). The Cl^– nature of the swelling-induced current was apparent from the close proximity of its reversal potential (–0.33 ± 1.30 mV, n = 12) to the theoretical E_Cl and its typical outward rectification. In addition, using voltage steps revealed that the swelling-induced current was inactivated at potentials more positive than +40 mV (Fig. 2b). Like the Cl^– current described in ITS, the swelling-induced Cl^– current was reversibly inhibited by DIDS (Fig. 2d). Likewise, the block was voltage-dependent with 79.81 ± 5.85% of inhibition at +40 mV and 15.03 ± 1.45% at –120 mV (n = 3). Figure 2c illustrates pooled ramp-derived I-V curves from 12 cells recorded successively in ITS and HTS. Note that HTS caused a large increase in conductance over the entire voltage range without any noticeable change in the null-current voltage (inset, Fig. 2c).

Altogether, these results indicate that hypotonicity up-regulates a Cl^– current in NHEKs. This osmosensitive current shares properties with VRAC currents and is hereafter referred to as I_VRAC.

Heptylamine InhibitsI_VRAC in NHEKs. Whether I_VRAC was sensitive to heptylamine was first examined in isotonic conditions. Figure 3, a to c, illustrates the response of a cell to the application of 0.01% (i.e., 674 μM) heptylamine. Heptylamine reversibly reduced I_VRAC amplitude by 46% in a voltage-independent manner (inset, Fig. 3b; mean values, 52.07 ± 3.11% of inhibition at –120 mV and 56.08 ± 4.10% at +40 mV, n = 7). The current blocked by heptylamine was outwardly rectifying and reversed near E_Cl in symmetric Cl^– conditions (reversal potential, –4.51 ± 0.57 mV, n = 12) (Fig. 3d). Its reversal potential was negatively shifted by 49 mV when internal Cl^– was substituted with equimolar Glu^– (reversal potential, –53.46 ± 1.69 mV in KGlu, n = 9). This anionic substitution preserved the maximum conductance and the outward rectification of the current. Note that the effect of heptylamine was 15- to 20-fold slower in onset than the block caused by DIDS (62.7 ± 4.8 s, n = 3), suggesting that heptylamine and DIDS inhibit VRAC via different mechanisms.

Incremental increase of the heptylamine concentration (0.0001–0.1%) produced a progressive decrease in I_VRAC. The degree of block was voltage-independent over a wide range of concentrations and reached a maximum value of 75.66 ± 3.71% (measured at +40 mV, n = 11) at the highest concentration of heptylamine tested (0.1%, i.e., 6.74 mM) (Fig. 3e). Dose-response curves, constructed by plotting the percentage of inhibition against heptylamine concentration, gave an IC₅₀ value of 262 μM (i.e., 0.0039%) at –120 mV and 256 μM (i.e., 0.0038%) at +40 mV (Fig. 3f). Note that the dose-dependence of the heptylamine block did not deviate from a one-to-one binding curve with a Hill coefficient of 0.98.

Hypotonicity Reduces the Inhibiting Effect of Heptylamine on VRAC. Inhibition of I_VRAC by heptylamine was investigated in hypotonic conditions (Fig. 4) by the following procedure: after a steady I_VRAC was reached, the external solution was changed to HTS; once the current settled at a new plateau, the cells were exposed to the HTS containing the desired concentration of heptylamine. An example is shown in Fig. 4, a and b, in which heptylamine was tested under both isotonic and hypotonic conditions. The inhibition of I_VRAC by 0.01% heptylamine was 52.07 ± 3.11% at –120 mV and 56.08 ± 4.10% at +40 mV in ITS (n = 7) but only 17.34 ± 3.10% at –120 mV and 18.19 ± 3.52% at +40 mV in HTS (n = 7). The inhibitory effect of heptylamine under hypotonic versus isotonic conditions was quantified by measuring the amplitude of I_VRAC at –120 and +40 mV in the presence of a series of concentrations of heptylamine (Fig. 4, c and d). Dose-response curves derived from these experiments gave IC₅₀ values of 2.9 mM (0.0424%) at –120 mV and 2.1 mM (0.0313%) at +40 mV. These values are an order of magnitude greater than the IC₅₀ determined in ITS (Fig. 4, e and f).

Inhibition of VRAC in Cell-Attached Patches. Heptylamine is an amphiphilic molecule able to incorporate into cellular lipid bilayers and to interact with membranous proteins (Hrnjez et al., 1999; Keller et al., 2004). One way to test whether heptylamine acts directly on the external side of the channel or after its insertion into the lipid bilayer is to prevent the access of heptylamine to the mouth of the channel. We therefore conducted experiments in which we combined recordings of VRACs in cell-attached patches with the application of heptylamine in the bath. The left portion of Fig. 5a shows single-channel activities in a cell-attached patch when the bath ITS was replaced with HTS. No single-channel activity was observed with ITS. In most patches, however, the replacement of ITS with HTS activated single or multichannel activities. Using a voltage ramp protocol, single-channel activity was characterized by a unitary amplitude of 7 pA at +100 mV (relative to V_rest), relatively long openings, and rectification in the outward direction. Because osmotic swelling could mobilize intracellular Ca²⁺ (Boudreault and Grygorczyk, 2004), we tested whether activation of these channels by swelling could be secondary to [Ca²⁺]_i increase. This was not the case, because similar swelling-induced single-channel activities were obtained in NHEKs loaded with 100 μM EGTA AM, a condition that abolished [Ca²⁺]_i increase induced by bath application of histamine (1 mM) (Fig. 5c).

The activity of osmosensitive channels was inhibited by bath application of heptylamine at 0.1%, a concentration that inhibited ∼70% of whole-cell I_VRAC in HTS (Fig. 5b, left). This swell-activated, heptylamine-inhibitable current is extracted as shown in the Fig. 5, a and b, right. Its I-V relationship with outward rectification resembles that of the whole-cell I_VRAC. These data indicate that heptylamine kept its ability to inhibit VRACs trapped in cell-attached patches. It is interesting that dialyzing heptylamine (0.1%) via the patch pipette solution abolished whole-cell I_VRAC in both ITS and HTS (data not shown), which suggests that direct access of heptylamine to the external side of VRAC is not required for the blockade.

Heptylamine Mobilizes Intracellular Ca²⁺, which Is Not Responsible for the Inhibition of VRAC. Thapsigargin-induced store-operated Ca²⁺ entry inhibits VRAC in HaCaT cells (Zholos et al., 2005). Therefore, we tested whether inhibition of VRAC in NHEKs by heptylamine was mediated by a [Ca²⁺]_i increase. Calcium imaging performed in ITS containing 2.5 mM Ca²⁺ showed that heptylamine (0.1%) induced an increase in [Ca²⁺]_i 5 s after its application. The maximal [Ca²⁺]_i was reached within 35 s with an F/F0 ratio of 1.64 ± 0.11 (n = 8) (Fig. 6a). Then, F/F0 slowly decreased to basal values in ∼5 min (data not shown). This effect on [Ca²⁺]_i was reproducible in a Ca²⁺-free ITS (F/F0 = 1.66 ± 0.17 after 35 s, n = 7), suggesting that heptylamine induces the release of Ca²⁺ from the intracellular store. To test whether this [Ca²⁺]_i increase contributes to VRAC inhibition by heptylamine, whole-cell recordings were performed in NHEKs maintained in ITS and dialyzed with an intrapipette solution containing 20 mM EGTA and no added Ca²⁺ (nominally 0 nM [Ca²⁺]_i). This internal solution efficiently clamped intracellular Ca²⁺ as tested by bath application of the Ca²⁺ ionophore ionomycin (1 μM). These drastic Ca²⁺-chelating conditions did not impair the I_VRAC block by heptylamine (Fig. 6, b and c): the percentage inhibition of I_VRAC by 0.01% heptylamine with this pipette solution was 54.13 ± 5.68% (n = 5), which was indistinguishable from the inhibition observed using internal solutions with lower Ca²⁺-buffering capabilities (Fig. 6d).

Discussion

Properties ofI_VRAC in NHEKs under Isotonic andHypotonic Conditions. We show here that VRAC is expressed by subconfluent proliferating NHEKs cocultured with feeder 3T3 fibroblasts. We demonstrate that intracellular solutions containing 10 nM [Ca²⁺]_i allow isolation of I_VRAC from the Ca²⁺-activated Cl^– and K⁺ channels expressed in keratinocytes (Koegel and Alzheimer, 2001). We show, however, that recordings with 100 nM [Ca²⁺]_i are contaminated with small Ca²⁺-dependent K⁺ currents, although I_VRAC is the main contributor to the macroscopic current.

In isotonic conditions, I_VRAC is absent immediately after patch-rupturing but develops within ∼15 min. It is characterized by an outward rectification, by inactivation at potentials greater than +40 mV, and by the permeability sequence Cl^– > F^– > Glu^–. As described previously in other cell systems, it is inhibited in a voltage-dependent manner by DIDS (Nilius and Droogmans, 2003). Application of HTS increases the I_VRAC amplitude ∼3-fold. Under hypotonicity, I_VRAC retains its outward rectification, inactivation at positive potentials, and voltage-dependent block by DIDS. Therefore, the Cl^– current recorded under isotonic and hypotonic conditions is carried by the same channel, described as VRAC in other cells (Voets et al., 1997; Shuba and McDonald, 2000; Jentsch et al., 2002; Nilius and Droogmans, 2003).

VRAC has been described in the epidermal cell line HaCaT (Zholos et al., 2005) and in monolayers of human keratinocytes cocultured with 3T3 fibroblasts (Rugolo et al., 1992). However, this is the first time that VRAC activity is observed in isotonic conditions in the absence of G protein-coupled receptor stimulation. The slow development of I_VRAC in ITS after patch rupturing together with the lack of single-channel activity in the cell-attached patch mode in ITS indicate that tonic activity of VRAC under isotonic conditions resulted from the pipette solution dialysis. It might not be caused by a change in the resting cell volume because we observed a similar development of I_VRAC using a slightly hypotonic pipette solution. The internal solutions we used did not differ in [Ca²⁺]_i (10 nM) from those of Rugolo et al. (1992). However, we used a higher intracellular ATP concentration (4 mM) than Zholos et al. and Rugolo et al. (1 mM). Intracellular ATP has been implicated in the development of I_VRAC in other cells (Lewis et al., 1993; Bond et al., 1999; Bryan-Sisneros et al., 2000). However, the role of ATP in VRAC activation remains uncertain, because nonhydrolyzable ATP analogs also increase I_VRAC (Nilius and Droogmans, 2003).

The Mechanism of VRAC Inhibition by Heptylamine. The pharmacological effects of heptylamine as a blocker of the cardiac Na⁺/Ca²⁺ exchanger (IC₅₀ = 2.6 mM; Keller et al., 2004) and of intestinal epithelial basolateral K⁺ channels (IC₅₀ = 0.42 mM; Hrnjez et al., 1999) have been reported. Here, we describe that heptylamine induces a slow, reversible, voltage-independent and dose-dependent inhibition of VRAC in NHEKs. In addition, heptylamine could act on keratinocytes through the inhibition of other channels, and notably K⁺ channels as suggested previously by Hrnjez et al. (1999). Moreover, our data show that heptylamine increases [Ca²⁺]_i via Ca²⁺-store depletion but not by external Ca²⁺ entry. Store-operated Ca²⁺ entry induced by thapsigargin inhibits VRAC in HaCaT cells (Zholos et al., 2005); therefore, we tested whether inhibition by heptylamine in NHEKs was mediated by the increase in [Ca²⁺]_i. It was not the case, because drastic intracellular Ca²⁺-chelating conditions did not prevent I_VRAC inhibition by heptylamine.

Heptylamine, an amphiphilic molecule, may have inhibited VRAC via the incorporation into the plasma membrane. Indeed, the logarithm of its partition coefficient is positive (+2.02), indicating a larger positive molar ratio of solubility in octanol than in water (Hrnjez et al., 1999). A mechanism requiring incorporation of heptylamine into the NHEK membrane could explain the slower time course of the VRAC-inhibiting effect of heptylamine than DIDS. In addition, the blockade is obtained when heptylamine is applied either in extracellular or intracellular solution, suggesting that direct access to the external side of VRAC is not required for the blockade. A strong evidence that heptylamine reaches its molecular target by being inserted into the plasma membrane is that heptylamine applied outside the cell blocks VRACs trapped inside the pipette. We conclude that the slow block of VRAC by heptylamine requires incorporation of this amphiphilic compound into the plasma membrane. A similar mechanism of blockade by heptylamine and other primary alkyl amines has been described for the basolateral K⁺ channels in T84 epithelial cells and the Na⁺/Ca²⁺ exchanger in cardiac myocytes (Hrnjez et al., 1999; Keller et al., 2004). Indeed, the blocking ability of lipid-soluble amines increased with increasing length of the aliphatic chain and lipophilicity. Confocal laser scanning microscopy with fluorescent lipid analogs suggested that these amphiphilic molecules needed to be incorporated into the membrane and that the Na⁺/Ca²⁺ exchanger block required transbilayer movement of these chemicals to the inner leaflet (Keller et al., 2004).

It is interesting that the inhibiting effect of heptylamine on VRAC is reduced by hypotonicity, indicating that the signaling pathway activated by cell-swelling reduces the effect of heptylamine. The mechanism by which swelling antagonizes the effect of heptylamine was not determined. The insertion of heptylamine into the plasma membrane may disturb the lipid environment of VRAC or its molecular partners. VRACs are compartmentalized in caveolae in endothelial cells and are known to be sensitive to changes in the lipidic composition of the plasma membrane (Trouet et al., 1999, 2001). In addition, depletion of membrane cholesterol enhances the effects of low ionic strength on VRACs (Romanenko et al., 2004). Thus, once inserted into the plasma membrane, heptylamine may inhibit VRACs by affecting the lipid/protein interactions and/or membrane deformability. VRACs in endothelial cells are insensitive to changes in sterol/protein interactions but are regulated by membrane deformability (Byfield et al., 2004; Romanenko et al., 2004). Changes in membrane deformability caused by cholesterol depletion are also associated with increased stiffness of the “deep cytoskeleton.” The disruption of actin filaments abolishes both the cell stiffening and the up-regulation of VRAC caused by cholesterol depletion (Byfield et al., 2006). It is possible that heptylamine inserts into the membrane, decreases the lipid bilayer deformability, alters the cytoskeleton organization, and thereby inhibits VRACs. Furthermore, there is evidence that hypotonicity modulates VRAC through the Rho-kinase signaling pathway and the subsequent F-actin stress fiber formation (Nilius et al., 1999, 2000; Amano et al., 2000; Koyama et al., 2001; Ridley, 2001; Carton et al., 2002). Therefore, we suggest that, under hypotonicity, cell-swelling induces changes in the cytoskeleton that antagonize the inhibitory effect of heptylamine on VRACs.

ATP Release, Ca²⁺Mobilization, and Irritant Contact Dermatitis. VRAC can be blocked by a variety of Cl^– channel blockers, but this is the first time that it is shown to be inhibited by a cutaneous irritant. Heptylamine is a primary amine classified as a corrosive chemical (R34) by the European Union. It gives positive results in the Draize test within 4 h of application (erythema and edema). According to the United Nations classification, heptylamine belongs to group II/III, causing skin damage after a minimal exposure period of 1 to 4 h.

Like many amines, heptylamine belongs to the category of cutaneous irritants that induce release of the inflammatory mediator ATP from keratinocytes without plasma membrane disruption (Mizumoto et al., 2003). Previous evidence suggests that VRAC may serve as a pathway for ATP release. Indeed, osmotic swelling induces ATP release from various cells that is inhibited by VRAC blockers (Grygorczyk and Guyot, 2001; Hisadome et al., 2002; Darby et al., 2003; Ullrich et al., 2006). Heptylamine inhibits VRAC, implying the existence of a VRAC-independent pathway for the heptylamine-induced release of ATP in NHEKs. This result confirms the observation of Boudreault and Grygorczyk (2004) in epithelial cells and fibroblasts that the cell-swelling ATP release does not correlate with cell volume expansion and activation of VRAC but is tightly synchronized with cytosolic Ca²⁺ elevations. The present study shows that the irritant compound heptylamine releases Ca²⁺ from intracellular stores. It remains to test whether heptylamine-induced [Ca²⁺]_i increase in NHEKs is directly responsible for ATP release and linked to irritant contact dermatitis. Alternatively, Ca²⁺ mobilization by the autocrine and paracrine action of the released ATP on metabotropic P2Y receptors is also likely because it can occur in endothelial and epithelial cells such as NHEKs during osmotic swelling or mechanical stimulation (Grygorczyk and Guyot, 2001; Koyama et al., 2001; Koizumi et al., 2004).