Introduction

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

Fenamates belong to the class of N-phenylanthranilic acids and are widely used as nonsteroidal anti-inflammatory drugs by their ability to inhibit cyclooxygenases (Brogden, 1986). In addition, fenamates modulate a diversity of ion channels. They have been identified as inhibitors of voltage-gated and ATP-sensitive potassium channels (Grover et al., 1994; Lee and Wang, 1999), voltage-gated calcium channels (Li et al., 1998), calcium-activated chloride channels (White and Aylwin, 1990), and nonselective cation channels (Gögelein et al., 1990). On the other hand, fenamates have been shown to activate calcium-activated and voltage-dependent potassium channels (Farrugia et al., 1993; Ottolia and Toro, 1994). Since we had found that flufenamic and niflumic acid inhibited the propagation of calcium action potentials in monolayers of normal rat kidney (NRK) fibroblasts (de Roos et al., 1997), we investigated whether fenamates could also block gap junctional channels.

Gap junctional intercellular communication (GJIC) is of paramount importance in the regulation of a variety of biological processes. Gap junctional channels allow intercellular diffusion of small (<1 kDa) hydrophilic molecules and ions. This intercellular diffusion of signal molecules regulates a variety of biological processes including embryogenesis, cell proliferation, cardiac function, and propagation of calcium waves (Bruzzone et al., 1996; Kumar and Gilula, 1996). Besides diffusion of small molecules, gap junctions also mediate electrical coupling between cells and allow clusters of cells to behave as an electrical syncytium. Electrical coupling underlies synchronous electrical activity between excitable cells and has been shown to be essential in the propagation of the cardiac action potential (Gros and Jongsma, 1996), the contraction of smooth muscle (De Mello, 1994) and the coordination of hormone secretion (Stauffer et al., 1993).

Gap junctions are present in the plasma membrane of cells as clusters of tightly packed particles, each of which represents a single channel (Bukauskas et al., 2000). Each channel is formed by the docking of two hemi-channels (connexons) located in apposing cell membranes of neighboring cells. Each hemi-channel consists of six polypeptides called connexins. Connexin43 (Cx43) is the major gap junctional protein identified in fibroblasts (Goldberg and Lau, 1993). GJIC can be regulated by intracellular calcium and pH (Loewenstein, 1981; Spray and Bennett, 1985). In addition, several processes that induce modifications of Cx43, including phosphorylation on serine and threonine residues following activation of PKC (Lampe et al., 2000) and on tyrosine residues upon growth factor receptor activation (Lau et al., 1996), have been shown to modulate GJIC. The contribution of each of these processes to the modulation of GJIC is dependent on the cell type.

We (de Roos et al., 1996) and others (Maldonado et al., 1988) have shown that confluent NRK fibroblasts are electrically well coupled and here we show that fenamates can completely block intercellular communication of not only NRK cells but also of SKHep1 cells overexpressing Cx43. The observed inhibition is reversible and not mediated by changes of intracellular calcium or pH, and was unrelated to PKC activity or an inhibition of cyclooxygenase activity.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

Cell Culturing. NRKs (clone 49F), SKHep1 human hepatoma cells, and SKHep1 stably transfected with rCx43, designated SKHep1-Cx43 cells (Kwak et al., 1995), were grown in bicarbonate-buffered Dulbecco's modified Eagle's medium (DMEM; Life Technologies, Paisly, UK) supplemented with 10% newborn calf serum (HyClone Laboratories, Logan, UT) under 5% CO₂ at 37°C. Confluent cultures were first made quiescent by subsequent incubation for 1 to 3 days in serum-free medium consisting of a 1:1 mixture of DMEM and Ham's F12 (Life Technologies) supplemented with 30 nM Na₂SeO₃ and 10 µg/ml human transferrin.

Electrophysiology. For patch-clamp experiments cells were perfused with serum-free medium consisting of a 1:1 mixture of bicarbonate-buffered DMEM and Ham's F12, equilibrated with 5% CO₂ to a pH of 7.4. Whole-cell patch-clamp experiments and voltage-clamp step response measurements were carried out as previously described (de Roos et al., 1996) using an EPC-7 patch-clamp amplifier (List Electronic, Darmstadt, Germany). Current and voltage-clamp protocols and data acquisition were performed using CED software in conjunction with a CED 1401 interface (Cambridge Electronic Design, Cambridge, UK). Data were filtered at 10 kHz and stored on hard disk for subsequent analysis. Briefly, recording of the membrane potential was shortly interrupted by switching to the voltage-clamp mode and the subsequent application of a voltage pulse of +10 mV (duration 180 ms) from a holding potential of 70 mV. The resulting capacitive current transient was analyzed (Bio-Patch software; BioLogic, Claix, France) to obtain the initial peak current Ipk, the final steady-state current Iss, and the time constant  of the initial component of the capacitive transient, representing the charging of the patched cell. This component was in particular recognizable in the later stages of uncoupling. Pipettes were made from borosilicate glass capillaries (GC150-15; Harvard Apparatus LTD, Edenbridge, UK) using a two-stage pipette puller (L/M-3P-A; List Electronic, Darmstadt, Germany). The intracellular pipette solution contained 25 mM NaCl, 120 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, 3.5 mM EGTA, pH 7.4, and pipettes had a resistance of 4 to 6 M.

Determining Electrical Coupling. We have used two simple equations allowing an estimation of the gap junctional conductance G01 between the patched cell (#0) and the cells in the surrounding cell ring (#1) (see Appendix). G01 was assessed by measuring parameters from the overshooting current response to voltage-clamp steps of 10 mV (Fig. 1B) and then filling in these parameters in one of the two equations under Appendix. The first equation, which may only be used for G01 < Gser (R01 > Rser), is as follows:

(1)

with Gser, the series conductance (Ipk/10 mV) between the patch pipette and the interior of cell #0; Ipk, the peak of the current transient after cancellation of the fast capacitive current charging the pipette capacitance or obtained by back extrapolation to t = 0 from the current transient after the fast transient; and Ish, the peak of the current component charging cell ring #1 and visible as a shoulder in the foot of the Cm0 charging current. The second equation is more general and practical and has the same shape:

(2)

with G01x, the (under)estimated value of G01, and Iss, the final steady-state current of the current transient, i.e., after completion of charging all the membrane capacitances of the successive cell rings around cell #0 (set at 180 ms in our recordings). Equation 2 is used in all cases, i.e., for measuring uncoupling and recoupling kinetics and for dose-response curves, while eq. 1 is only used for checking G01x values under conditions that R01 > Rser.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   A, membrane potential of a confluent monolayer of NRK cells was measured and addition of 100 µM MFA is indicated by the bar. The membrane potential recording shows short interruptions to the voltage-clamp mode. B, capacitive current transients, evoked by application of a short voltage-clamp step (10 mV), at the indicated time points (a-h). C, corresponding calculated gap junctional conductance (G01x) as a function of time. Addition of 100 µM MFA is indicated by the bar.

Fluorescence Measurements. Dye-coupling of cells was determined by microinjection of a 10% lucifer yellow (Molecular Probes, Eugene, OR) solution in 0.33 M LiCl into a single monolayer cell by means of a glass capillary tip using a vertical microinjection system (Olympus IMT-2; Olympus, Tokyo, Japan) as previously described (de Haan et al., 1994). About 20 different monolayer cells per dish were injected (one per 5 s) and the number of fluorescent cells was counted 10 min after injection. Only the five injections per dish that caused most substantial dye transfer were used for further analysis.

Chemicals. Meclofenamic acid, niflumic acid, flufenamic acid, tolfenamic acid, indomethacin, flurbiprofen, and nigericin were from Sigma (St. Louis, MO).

	Results

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

Previously, we showed that capacitance measurements can be used to study changes in GJIC in small clusters of NRK cells (de Roos et al., 1996). Here we added meclofenamic acid, flufenamic acid, niflumic acid, and tolfenamic acid to a confluent monolayer of NRK cells and investigated their effect on the membrane potential and capacitive current transient at different time points after their addition. Meclofenamic acid (100 µM) depolarized the membrane of monolayer NRK fibroblasts (Fig. 1A) and the average depolarization was 25.7 ± 2.8 mV (n = 33). The depolarization could be restored to the resting membrane potential by washout of the fenamate. Flufenamic acid, niflumic acid, and tolfenamic acid induced comparable reversible depolarizations in monolayers of NRK fibroblasts of 23.2 ± 4.6, 25.9 ± 2.6, and 24.8 ± 3.7 mV (all n = 10), respectively.

Application of a voltage-clamp step to a single NRK fibroblast evokes a characteristic current transient that is determined by the series resistance (Rser) and the membrane capacitance and conductance of the individual cell (de Roos et al., 1996). In confluent monolayers of NRK fibroblasts, however, membrane capacitance and conductance of the surrounding cells also contribute to this current transient, since these cells are electrically well coupled. This is reflected by a slow and multiexponential initial decay of the induced current and a subsequent large prolonged steady-state current (Fig. 1B, a). Application of 100 µM meclofenamic acid to a monolayer of NRK fibroblasts caused an acceleration of the decay of the induced current and reduced the level of the prolonged steady-state current. This effect of meclofenamic acid was already observed under conditions that the membrane potential was still unchanged (Fig. 1B, b). In this typical example, after 4 min the steady-state current reached a minimal value. The evoked current transient could be fitted by a single exponential, indicating a complete block of intercellular communication, since more than one exponential is needed to fit the transient during a partial block of intercellular communication (de Roos et al., 1996). The capacitance calculated from this transient was 23.7 pF (Fig. 1B, e). The capacitance of cells completely uncoupled by meclofenamic acid was 19.7 ± 1.0 pF (n = 11), which is of the same order of magnitude as previously measured in single isolated NRK cells (de Roos et al., 1996). Electrical coupling could be largely restored by washout of the fenamate (Fig. 1B, f-h). These results demonstrate that fenamates can reversibly block electrical coupling in NRK fibroblasts and that blocking is not mediated by changes in the membrane potential.

The time course of the uncoupling could be followed by calculating the estimated gap junctional conductance between the patched cell and the surrounding cell ring (G01x) from each capacitive current transient (Fig. 1C). G01x in the coupled monolayer was 140.7 ± 9.6 nS (n = 14) and was reduced after addition of 100 µM meclofenamic acid to 1.4 ± 0.4 nS (n = 11) within 6.0 ± 0.4 min (n = 11). Since the single cell conductance is close to 1 nS (de Roos et al., 1996), this indicates that the patched cell is completely uncoupled from the surrounding cells. Although G01x could not completely be restored, partial recovery to 51.8 ± 4.2 nS (n = 11) was reached within 7.2 ± 1.4 min (n = 11).

To compare the potency of different fenamates, the concentration dependence of their effects on electrical coupling has been determined. To obtain dose-response curves, G01x was calculated 15 to 20 min after addition of the fenamates when the capacitive current transient had reached a steady-state (Fig. 2). Although G01x reflects only an estimation of the gap junctional coupling, the dose-response curves clearly show that the potency of the fenamates in blocking electrical coupling decreases in the order meclofenamic acid > niflumic acid > flufenamic acid.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Dose-response curve of the inhibition of electrical coupling by fenamates. The calculated gap junctional conductance (G01x) as a function of the concentration FFA, niflumic acid (NFA), and MFA is shown. G01x was determined 15 to 20 min after addition of fenamates when the capacitive current transient had reached a steady state (mean ± S.E.M., n = 4; control, n = 14).

Fluorescent dye transfer of lucifer yellow was used to investigate whether fenamates, in addition to electrical coupling, can also affect dye coupling. Lucifer yellow was injected into a single cell of a confluent monolayer of NRK fibroblasts, and intercellular diffusion of the dye showed that these cells were extensively dye coupled (Fig. 3, top). Pretreatment for 5 min with 100 µM meclofenamic acid completely prevented this diffusion of the injected dye to the neighboring cells (Fig. 3, bottom). Dye transfer was also completely inhibited when flufenamic acid was used at a concentration of 250 µM. These results demonstrate that fenamates can also completely block dye coupling in NRK cells.

View larger version (117K): [in this window] [in a new window]   Fig. 3.   Dye coupling in NRK fibroblasts. Lucifer yellow was injected into one single cell of the monolayer and intercellular diffusion of the dye was monitored 10 to 15 min after injection. Phase contrast and lucifer yellow fluorescence in control cells (top) and after pretreatment with 100 µM MFA for 5 min (bottom). Arrows indicate the injected cell.

To quantify the reduction of intercellular communication by fenamates, the number of cells to which lucifer yellow could be transferred was determined 10 min after injection into a monolayer NRK cell. For meclofenamic acid (MFA), complete block of dye coupling was achieved at a concentration of 100 µM, while block was half-maximal at 25 µM (Fig. 4). Flufenamic acid (FFA) completely blocked dye coupling at 250 µM. The inhibition of dye coupling by flufenamic acid was half-maximal at 40 µM (Fig. 4). In agreement with their effects on electrical coupling, dye-coupling experiments show that meclofenamic acid is a more potent blocker of GJIC than flufenamic acid.

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Dose-response curve of the inhibition of intercellular communication by fenamates. Number of dye-coupled cells as a function of the concentration of MFA and FFA. Number of dye-coupled cells was determined 10 to 15 min after dye injection into a single cell (mean ± S.E.M., n = 4; control, n = 10).

Next, we investigated whether the block of intercellular communication by fenamates could result from changes in the levels of the intracellular calcium concentration or pH, which are physiological modulators of gap junctions (Spray and Bennett, 1985). However, exposure of the cells to 250 µM meclofenamic acid for 10 min did not change the intracellular calcium concentration significantly. Fluorescence (340 nm/380 nm) values of intracellular Fura-2 before and after exposure were 0.44 ± 0.01 and 0.43 ± 0.01 (n = 10), respectively. Furthermore, the intracellular pH was unaffected under these conditions. The basal intracellular pH was 7.06 ± 0.04 and after adding meclofenamic acid the pH was 7.01 ± 0.05 (n = 8). These data demonstrate that inhibition of GJIC by fenamates is mediated neither by changes in intracellular calcium nor pH.

PKC has been described as a regulator of GJIC (Lampe et al., 2000) and therefore we investigated whether block of GJIC by fenamates is mediated by activation of PKC. In a previous study, we investigated modulation of gap junctions by activation of PKC using the phorbol ester TPA (de Roos et al., 1996) and found that TPA caused a partial block of GJIC. In the present study we have found that after prolonged (24-h) pretreatment of NRK fibroblasts with TPA, which causes the complete down-regulation of PKC activity in these cells, electrical coupling was completely restored and could not be blocked anymore by 100 ng/ml TPA. However, in these PKC down-regulated NRK fibroblasts electrical coupling could still be blocked by fenamates. After addition of meclofenamic acid and niflumic acid (250 µM) to these cells for 10 min, the voltage step-induced current transients could be fitted by single exponentials and the calculated capacitance of the cells was 20.1 ± 1.2 and 18.7 ± 2.6 pF (n = 7), respectively. This excludes a role for PKC in the inhibition of GJIC by fenamates.

Whether the impairment of GJIC by fenamates could be directly related to an inhibition of cyclooxygenase activity in the NRK cells was investigated by application of two other potent cyclooxygenase inhibitors, indomethacin and flurbiprofen. Pretreatment of NRK fibroblasts for 3 h with 1 µM of either of these inhibitors has been shown to cause a complete inhibition of cellular cyclooxygenase activity (Lahaye et al., 1994). We have found in the present study, however, that pretreatment of the cells for 3 h with either indomethacin or flurbiprofen did not affect their electrical coupling at all. G01x before and after the pretreatment with these inhibitors was 155.3 ± 12.8 nS (n = 8) and 153.9 ± 16.4 nS (n = 8), respectively. These results indicate that the impairment of GJIC by fenamates is unrelated to an inhibition of cyclooxygenase activity.

Since GJIC in NRK fibroblasts is mainly mediated by Cx43 (Li et al., 1996) we investigated whether impairment of intercellular communication can be attributed to a direct effect of fenamates on Cx43 function. To address that issue we added meclofenamic acid, flufenamic acid, and niflumic acid to SKHep1 cells and SKHep1 cells that had been stably transfected with Cx43 (SKHep1-Cx43 cells). Monolayers of SKHep1 as well as SKHep1-Cx43 cells showed a membrane potential around 35 mV and application of 250 µM meclofenamic acid reversibly depolarized the membrane with 10.3 ± 2.5 mV (n = 7) in SKHep1 cells and 11.4 ± 2.4 mV (n = 7) in SKHep1-Cx43 cells. Voltage-clamp step response experiments showed that monolayers of wild-type SKHep1 cells, which do not express Cx43 and endogenously express only low levels of Cx45 (Moreno et al., 1995), were less well coupled than NRK fibroblasts (Fig. 5A, control). The capacitance derived from the initial fast current transient was 16.9 ± 0.3 pF (n = 7), which is in the range of that of single cells in isolation. Thus, this capacitance is that of the patched cell in the monolayer. The corresponding G01x was 10.3 ± 0.3 nS (n = 7) and was reduced to 2.9 ± 0.2 nS (n = 7) after application of 250 µM meclofenamic acid to these cells (Fig. 5A, 8 min). The capacitance of the patched cell was 17.1 ± 0.4 pF (n = 7). G01x could partially be recovered to 8.2 ± 0.3 nS (n = 7) by washout of meclofenamic acid, whereas the initial fast capacitive transient remained unchanged (Fig. 5A, washout). Thus, 250 µM meclofenamic acid at least partly blocked intercellular coupling mediated by Cx45. SKHep1-Cx43 cells showed larger capacitive current transients, indicating that these cells were better electrically coupled (Fig. 5B, control). G01x was 60.6 ± 4.3 nS (n = 7) for the coupled monolayer and was reduced to 2.7 ± 0.2 nS (n = 7) after adding 250 µM meclofenamic acid (Fig. 5B, 6 min). The capacitance calculated from the initial fast transient was 17.5 ± 0.4 pF (n = 7), also indicating that the cells were largely uncoupled under these conditions. Electrical coupling could be restored by washout of meclofenamic acid and G01x was partially recovered to 23.0 ± 1.9 nS (n = 7) (Fig. 5B, washout). Flufenamic acid and niflumic acid were also able to block electrical coupling in these cells when added at a concentration of 250 µM. After uncoupling with flufenamic acid G01x was 2.7 ± 0.4 nS (n = 3) and the calculated capacitance was 19.0 ± 0.8 pF (n = 3). Niflumic acid reduced G01x to 2.9 ± 0.2 nS (n = 3) and the capacitance to 17.9 ± 1.1 pF (n = 3). These results demonstrate that fenamates are reversible blockers of Cx43-mediated intercellular communication.

View larger version (9K): [in this window] [in a new window]   Fig. 5.   Effect of 250 µM MFA on the capacitive current transient, evoked by a short voltage-clamp step (10 mV), in confluent monolayers of SKHep1 cells (A) and SKHep1-Cx43 cells (B). Data shown are representative traces of at least four experiments.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

In the present study we show that fenamates represent a novel class of reversible blockers of Cx43-mediated GJIC. The effects of fenamates on GJIC were assessed by using an analysis of single electrode voltage-clamp step responses, which is based on a simplified electric circuit of the patched cell with the surrounding monolayer.

Gap junctions have been shown to be essential for the direct diffusion of, for example, calcium and small signal molecules such as inositol triphosphate between neighboring cells (Giaume and Venance, 1998) and for their electrical coupling. A variety of biological processes, including cellular growth, propagation of calcium waves, and cardiac function is regulated by GJIC (Yamasaki and Naus, 1996; Kumar and Gilula, 1996). Impaired GJIC has been reported in several diseases, including cardiovascular diseases (Jongsma and Wilders, 2000) and tumorigenesis (Zhang et al., 1998; Laird et al., 1999; Yamasaki et al., 1999). Although in some cells a decrease in GJIC correlates with tumor growth (Hotz-Wagenblatt and Shalloway, 1993), phenotypic transformation of NRK cells is accompanied by an increase in intercellular communication (van Zoelen and Tertoolen, 1991).

GJIC can be inhibited by physiological agents that initiate complex signaling pathways, resulting in activation of kinases, phosphatases, and interacting proteins (Hossain and Boynton, 2000) or changes in levels of intracellular calcium and pH (Bruzzone et al., 1996). Furthermore, different classes of chemicals have been shown to block GJIC. Some of them affect the conformation of the connexins by disturbing the bulk membrane fluidity or the membrane protein interface. Examples of chemicals that block GJIC by changing the lipid structure around connexins include alcohols such as heptanol and octanol (Johnston et al., 1980), halothane (Burt and Spray, 1989), and fatty acids such as oleamides (Lerner, 1997), arachidonamide (Boger et al., 1999), and anandamide (Venance et al., 1995). Phorbol esters block GJIC by phosphorylation of connexin residues (Lampe, 1994), whereas glycerrhetinic acid derivatives block by dephosphorylation of connexin43 (Guan et al., 1996). For selective inhibition of gap junctions, synthetic oligopeptides that interact with the external loop of connexins have been developed (Kwak and Jongsma, 1999).

Fenamates are widely used as inhibitors of cyclooxygenase activity (Brogden, 1986). Their inhibitory effect on gap junctions, however, is not related to their ability to inhibit cyclooxygenase activity, since indomethacin and flurbiprofen, two potent cyclooxygenase inhibitors, did not affect the electrical coupling of the cells at all.

Previously, we found that blocking GJIC with the phorbol ester TPA resulted in a depolarized membrane (de Roos et al., 1996). Furthermore, halothane, another gap junction blocker, has been reported to have effects on the resting membrane potential in skeletal muscle cells (Sauviat et al., 2000). Complete uncoupling by fenamates was also accompanied by depolarizations of the membrane of variable magnitude. Uncoupling of the NRK cells by fenamates, however, already started when the membrane potential was still unchanged (Fig. 1, A, b and B, b). Moreover, complete uncoupling was sometimes observed under conditions that the membrane potential was unaffected. These results exclude that membrane depolarization caused the uncoupling by fenamates.

Complete block of GJIC by fenamates was sometimes accompanied with increased fluctuations of the membrane potential (Fig. 1, A, e and B, e). Most likely, electrical coupling of the monolayer cells caused fluctuations in the membrane potential to be averaged by a mechanism of channel-sharing (Atwater et al., 1983; Sherman et al., 1988), resulting in a stable membrane potential. So, the increased fluctuations of the membrane potential after application of fenamates may reflect the loss of channel sharing by blocking GJIC. However, since fenamates have been reported to block anionic as well as cationic channels at various concentrations from 10 to 100 µM (Gögelein et al., 1990; White and Aylwin, 1990; Grover et al., 1994; Li et al., 1998; Lee and Wang, 1999), additional effects of fenamates on plasma membrane ion channels may also have contributed to the increased membrane potential fluctuations and the depolarization of the uncoupled cell.

Fenamates did not affect intracellular calcium and pH, two physiological modulators of gap junctions. In addition, fenamates could still block GJIC in PKC down-regulated cells, excluding a role for PKC. Although a role for other second messenger pathways cannot be excluded, it is more likely that the reversible block of GJIC by fenamates results from either a direct interaction with connexins or an indirect action through perturbations in the bulk membrane fluidity or the membrane protein interface that would affect the conformation of the connexins.

Anti-inflammatory therapy is often accompanied by unwanted side effects, including gastrointestinal toxicity. However, nonsteroidal anti-inflammatory drugs have also been described to prevent colon cancer and Alzheimer's disease (Rich et al., 1995; Sheng et al., 1997). Whether there is a link between these unwanted and/or beneficial effects of nonsteroidal anti-inflammatory drugs and the (partial) block of gap junctions by fenamates requires further investigation. It should, however, be mentioned that the concentration needed to inhibit cyclooxygenase is much lower than that required to block GJIC. For example, the IC₅₀ of meclofenamic acid is about 0.05 µM in inhibiting cyclooxygenase (Kalgutkar et al., 2000) and about 25 µM in blocking GJIC (Fig. 4). Nevertheless, the present finding that fenamates are able to block GJIC not only in NRK fibroblasts but also in Cx43-overexpressing SKHep-1 cells strongly indicates that fenamates can be used as a pharmacological tool to reversibly block Cx43-mediated intercellular communication. Whether fenamates can also be used as blockers of intercellular communication mediated by other connexins remains to be investigated.