Functional and molecular characterization of a volume-sensitive chloride current in rat brain endothelial cells


Corresponding author S. F. von Weikersthal: Department of Pharmacology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QJ, UK. Email:


  • 1Volume-activated chloride currents in cultured rat brain endothelial cells were investigated on a functional level using the whole-cell voltage-clamp technique and on a molecular level using the reverse transcriptase-polymerase chain reaction (RT-PCR).
  • 2Exposure to a hypotonic solution caused the activation of a large, outward rectifying current, which exhibited a slight time-dependent decrease at strong depolarizing potentials. The anion permeability of the induced current was I (1.7) > Br (1.2) > Cl (1.0) > F (0.7) > gluconate (0.18).
  • 3The chloride channel blocker 5-nitro-2-(3-phenylpropylamino)-benzoate (NPPB, 100 μM) rapidly and reversibly inhibited both inward and outward currents. The chloride transport blocker 4,4′-diisothiocyanatostilbene-2,2′-disulphonic acid (DIDS, 100 μM) also blocked the hypotonicity-induced current in a reversible manner. In this case, the outward current was more effectively suppressed than the inward current. The volume-activated current was also inhibited by the antioestrogen tamoxifen (10 μM).
  • 4The current was dependent on intracellular ATP and independent of intracellular Ca2+.
  • 5Activation of protein kinase C by phorbol 12,13-dibutyrate (PDBu, 100 nM) inhibited the increase in current normally observed following hypotonic challenge.
  • 5Extracellular ATP (10 mM) inhibited the current with a more pronounced effect on the outward than the inward current.
  • 6Verapamil (100 μM) decreased both the inward and the outward hypotonicity-activated chloride current.
  • 7RT-PCR analysis was used to determine possible molecular candidates for the volume-sensitive current. Expression of the ClC-2, ClC-3 and ClC-5 chloride channels, as well as pICln, could be shown at the mRNA level.
  • 8We conclude that rat brain endothelial cells express chloride channels which are activated by osmotic swelling. The biophysical and pharmacological properties of the current show strong similarities to those of ClC-3 channel currents as described in other cell types.

Endothelial cells from brain microvessels form the structural and functional component of the blood-brain barrier (BBB) which isolates the brain parenchyma from normal variations in body fluid composition. Their properties are very different from those found in other parts of the vasculature. In particular they possess highly resistant, tight intercellular junctions, have minimal pinocytotic transport and there are no fenestrations. These characteristics ensure a stable environment in the brain protected from peripheral fluctuations and allowing normal neuronal activity. Various conditions such as ischaemia, hypoxia, brain injury and inflammation are accompanied by brain oedema which can be caused or aggravated by the swelling of endothelial cells which then can result in increased permeability and disruption of the BBB. The ability of endothelial cells to regulate volume efficiently is therefore a prerequisite for the preservation of the structural integrity of the BBB.

Studying the ionic mechanisms underlying endothelial cell volume regulation is important in understanding the processes and regulation of fluid movement across the BBB. Under normal steady state conditions, the constancy of the cell volume is maintained by the Na+-K+-ATPase. In addition, most cells possess transport systems that become activated after changes in cell volume. These include the Na+-H+ exchangers, the Na+-K+-Cl cotransporter, chloride channels, potassium channels and water channels. Cell swelling leads to activation of these transport systems and this results in a loss of osmolytes and a concomitant loss of water. This has been termed a regulatory volume decrease (RVD) (for a review, see Hoffman & Dunham, 1995).

Chloride channels involved in RVD have been described in a number of different cell types, such as endothelial cells, epithelial cells, various cell lines, tumour cells and even excitable cells (for reviews, see Nilius et al. 1996; Strange et al. 1996; Okada, 1997). However, for brain endothelial cells, the identity and possible role of chloride channels had not been described in any depth prior to our work (von Weikersthal et al. 1997).

In recent years, several proteins have been described as possible molecular candidates for volume-sensitive chloride currents. To date, the ClC chloride channel family offers two candidates: firstly, ClC-2 has been cloned and identified as a volume-sensitive chloride channel (Thiemann et al. 1992; Gründer et al. 1992) and, secondly, the ClC-3 channel has been sequenced by Kawasaki et al. (1994) and recently described to be swelling activated (Duan et al. 1997). The protein pICln has also been linked with volume-sensitive chloride currents. Paulmichl et al. (1992) sequenced pICln and suggested that it might be a possible candidate for a chloride channel. However, it could be shown that pICln is localized mainly in the cytosol and that its protein structure is not that of a typical channel. Instead it was suggested that pICln is a regulator for a volume-sensitive chloride channel (Krapivinsky et al. 1994). P-glycoprotein (P-gp), the product of the multidrug resistance MDR1 gene, is another protein which has been suggested to be involved in cell volume regulation. Valverde et al. (1992) reported that P-gp is a volume-sensitive chloride channel, but this specific role was subsequently questioned (see, for example, McEwan et al. 1992). Recent work still suggests a possible link between P-gp and volume-sensitive currents, but one in which P-gp exerts a regulatory role (Valverde et al. 1996; Wu et al. 1996).

In the present study, we have endeavoured to clarify the situation for brain endothelial cells, following up previous observations of increased anion permeability after hypotonicity-induced swelling (von Weikersthal et al. 1997) and some patch-clamp results reported previously in abstract form (von Weikersthal et al. 1996). In order to identify a possible candidate channel, we have pursued a two-pronged approach. Firstly, at a functional level, we have characterized in detail the volume-sensitive chloride current in these cells using the whole-cell patch-clamp technique. Secondly, at a molecular level, employing the RT-PCR technique, we have demonstrated which of the possible candidate chloride channels are expressed in the cells. Putting these data together and comparing with the literature, we suggest that the volume-sensitive chloride channel we have identified functionally in rat brain endothelial cells is most probably the ClC-3 channel.


Cell preparation

Brain microvessels were isolated from the cortical grey matter of rats using a method previously described (Abbott et al. 1992). Briefly, male Wistar rats (180-200 g in weight) were killed by exposure to a rising concentration of CO2. Cortical grey matter was dissected free of meninges and digested in collagenase-dispase (Boehringer, Mannheim) for 1 h; subsequently myelin was removed by centrifugation through 22 % BSA. This was followed by a more prolonged digestion (3 h) and the separation of individual microvessels from contaminating cells by centrifugation on a Percoll gradient. Cells were grown from these microvessels in collagen-coated culture flasks or dishes in Ham's F10 medium containing 20 % plasma-derived serum, 75 μg ml−1 endothelial cell growth supplement (First Link, Brierley Hill, West Midlands, UK), 80 μg ml−1 heparin, 0.5 μg ml−1 vitamin C, 2 mM L-glutamine, 100 U ml−1 penicillin and 100 μg ml−1 streptomycin. Cultures were maintained at 37°C in a humidified 5 % CO2-air atmosphere. Cells of passages 2-4 were used.

Electrophysiological recordings

Experiments were performed at room temperature (20-24°C) using the whole-cell configuration of the patch-clamp technique. One hour before the experiment, the cells were harvested using a brief exposure to trypsin and seeded onto a coverslip mounted in a perfusion chamber containing an isotonic bath solution. The chamber was continuously perfused throughout the experiment. Changes in tonicity and additions of drugs were brought about by changing the solution perfused through the recording chamber. Patch-clamp pipettes were manufactured from borosilicate glass (GC 150F-15, Clark Electromedical, Pangbourne, UK) using a two-stage puller (Narashige, Tokyo, Japan), and fire-polished to give final resistances of 4-6 MΩ when filled with pipette solution. The reference electrode, a Ag-AgCl pellet immersed in pipette solution, was connected to the bath by a 3 M KCl salt bridge. Whole-cell currents were recorded with an Axopatch 200A amplifier and pCLAMP 6 software (Axon Instruments). For the majority of experiments, voltage pulses of 100 ms duration were applied in 20 mV steps to levels between -100 and +100 mV, from a holding potential of -50 mV. Current-voltage relationships for the average current in the last 40 ms of 100 ms pulses were calculated using Origin software (Microcal Software Inc., Northampton, MA, USA). The percentage effect of a drug for the indicated voltage step was determined as follows:

display math

Reversal potentials were determined from I-V relationships measured using 1 s voltage ramps from -100 to +100 mV. The potentials have not been corrected for the junction potentials at the tip of the microelectrode and the ends of the salt bridges. Data are expressed as means ±s.e.m.; n stands for number of observations. Statistical analysis was by Student's paired t test; a value of P < 0.05 was considered to be statistically significant.

Solutions and chemicals

The standard isotonic bath solution contained (mM): NaCl, 150; KCl, 2.5; CaCl2, 2.5; MgCl2, 0.5; Hepes, 10; NaH2PO4, 1; glucose, 10; buffered to pH 7.4 with NaOH. The osmolarity was 290 mosmol l−1. The standard hypotonic bath solution was obtained by reducing the NaCl concentration to 120 mM, which gave an osmolarity of 230 mosmol l−1. Further hypotonic bath solutions were obtained by reducing NaCl to 135 mM (260 mosmol l−1), or 110 mM (190 mosmol l−1). In experiments to test the anion selectivity, 120 mM NaCl of the hypotonic bath solution was replaced by the same concentration (120 mM) of NaI, NaBr, NaF or sodium gluconate. The osmolarity was adjusted to 230 mosmol l−1 using mannitol. The standard pipette solution contained (mM): caesium gluconate, 110; CsCl, 20; Hepes, 10; EGTA-Na4, 1; EDTA-Na2, 0.01; MgSO4, 8; ATP-Na2, 4; GTP-Na3, 0.5; and buffered at pH 7.2 with CsOH. The osmolarity was 270 mosmol l−1. In experiments to investigate the type of current rectification, the pipette solution was changed by increasing CsCl to 120 mM and reducing caesium gluconate to 10 mM. The osmolarity of all solutions was measured with the Advanced Micro Osmometer (Advanced Instruments, Inc., Needham Heights, MA, USA). The test drugs, 4,4′-diisothiocyanostilbene-2,2′-disulphonic acid (DIDS), 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB), tamoxifen, phorbol 12,13-dibutyrate (PDBu), ATP and verapamil were obtained from Sigma. Stock solutions used for addition to the bath perfusate were 10 mM DIDS in hypotonic bath solution, 10 mM NPPB in ethanol, 10 mM tamoxifen in ethanol, 10 mM PDBu in DMSO, 10 mM verapamil in distilled water and 1 M ATP in distilled water.

Isolation of total RNA and RT-PCR

Total cellular RNA was prepared from a cell pellet ((2-5) × 106 cells) by guanidine hydrochloride lysis, phenol-chloroform extraction and ethanol precipitation using glycogen to assist precipitation, as described by Sambrook et al. (1989). Total RNA was reverse transcribed into first strand cDNA using 0.5 μg random hexamer pd(N)6 (HT Biotechnology, Cambridge, UK) with 10 mM dithiothreitol, 6 mM MgCl2, 40 mM KCl, 25 mM Tris buffer, pH 8.4, 2.5 μg DNase/RNase-free BSA, 1 mM of each deoxytriphosphate nucleotide (dATP, dGTP, dCTP and dTTP). Ten units of super reverse transcriptase (HT Biotechnology) were added to the reaction mix and incubated for 1 h at 42°C. The cDNA produced by the RT step was used for simultaneous amplification of sequences specific to the different genes. Amplification was performed using 4 μl of the cDNA mixture. The PCR reaction mixture included 2 mM MgCl2, 50 mM KCl, 20 mM Tris buffer pH 8.4, 5 μg BSA, 200 μM of each deoxytriphosphate nucleotide (dATP, dGTP, dCTP and dTTP), 12.5 pmol forward primer, 12.5 pmol reverse primer, 1 unit SuperTaq (HT Biotechnology) and water to give a total reaction volume of 24 μl. PCR was performed in a Techne PHC-3 Thermal Cycler (Cambridge, UK) using the following conditions: 30 cycles of 94°C for 1 min, 55°C for 2 min and 72°C for 2 min, followed by a final step of 72°C for 5 min. Oligonucleotide primers used in the described experiments are listed below. The PCR products were separated by electrophoresis in a 2 % agarose gel and visualized by UV in the presence of ethidium bromide.

Primer design

Primers were designed with the aid of the designer program ‘Primer3’ at using the corresponding rat mRNA sequences. The specificity of the primers for the target gene was checked against the databases using ‘Fasta3′ at and primers were checked for hairpin loops and palindromes using the Cybergene Utility at The oligonucleotides were synthesized by Genosys, Cambridge, UK. The primers designed and used are listed in Table 1.

Table 1. Primers designed and used in the RT-PCR
Target genePrimers Product size (bp)Accession no


Biophysical characterization of the volume-activated chloride current

Rat brain endothelial cells were voltage clamped in the whole-cell configuration and held at -50 mV. Under isotonic conditions, these cells possessed a small current of 13 ± 2.2 pA at an applied potential of +80 mV and -4.2 ± 0.8 pA at -80 mV (n= 5). After change to a hypotonic solution, the current increased significantly; the increase in current was dependent on the level of hypotonicity. The reduction of osmolarity by 30 mosmol l−1 from 290 to 260 mosmol l−1 increased the current at an applied potential of +80 mV to 45 ± 6.2 pA and at -80 mV to -4.7 ± 0.9 pA (n= 5). A further reduction of osmolarity to a 60 mosmol l−1 difference raised the current at +80 mV to 163.3 ± 23.7 pA and at -80 mV to -15.8 ± 3.2 pA (n= 5). A strong hyposmotic solution (90 mosmol l−1 difference) evoked an increase in current at ± 80 mV to 451.5 ± 50.7 and -43.3 ± 6.9 pA (n= 5), respectively. The current was outward rectifying; in experiments (n= 3) with nearly symmetrical Cl concentrations inside and outside the cell (hypotonic bath solution containing 126 mM Cl and pipette solution 120 mM Cl), the current at an applied potential of +80 mV was 137 ± 6.7 pA and, at -80 mV, it was -36.1 ± 2.8 pA. The reversal potential of the I-V curve was at 2.3 ± 0.6 mV. The increase in current was reversible as it decreased on return to isotonic conditions. A second exposure to a hypotonic solution reactivated the current again. Figure 1A shows current traces from a typical cell exposed to different hypotonic solutions. Recordings shown here were made after a steady state was reached under each hypotonic condition. Figure 1B shows the corresponding current-voltage relationships in response to an isotonic solution and different hypotonic solutions of five experiments. The current showed a time-dependent inactivation at potentials more positive than +80 mV.

Figure 1.

Effect of different levels of hypotonicity on anion current in rat brain endothelial cells obtained from whole-cell voltage-clamp recordings

A, whole-cell currents in response to voltage pulses, to levels from -100 to +100 mV in 20 mV steps. Recordings were made once steady state had been reached. Thereafter solutions were changed. B, current-voltage relationships; recordings were made when the response reached steady state after the change to the corresponding bath solution. ○, isotonic, 290 mosmol l−1; ▴, hypotonic, 260 mosmol l−1; •, hypotonic, 230 mosmol l−1; ♦, hypotonic, 190 mosmol l−1. Data are shown as means ±s.e.m., n= 5.

To examine the chloride dependence and the anion selectivity of the channel, Cl in the hypotonic solution was replaced by Br, I, F or gluconate (each 120 mM). Voltage ramps between -100 and +100 mV over 1 s were applied in the presence of the different extracellular anions. I was the most permeable whereas gluconate was the least permeable. The sequence of permeabilities was I > Br > Cl > F > gluconate. The reversal potentials (Vrev) under the different conditions are given in Table 2, as well as the relative permeability for each anion X to Cl (PX/PCl) which were calculated using the Goldman-Hodgkin-Katz equation. The shifts of reversal potential and the corresponding permeability ratios indicate the chloride dependence of the current.

Table 2. Effects of Cl ion replacement on reversal potentials of volume-sensitive anion current
AnionVrev (mV) P X/PCl n
  1. V rev values were obtained from I–V plots as indicated in Methods. Relative anion permeabilities (PX/PCl) were calculated using the Goldman–Hodgkin–Katz equation.

I−52.6 ± 1.61.704
Br−44.4 ± 1.61.213
Cl−38.1 ± 0.61.04
F−31.0 ± 0.80.74
Gluconate−2.2 ± 0.60.183

During each recording, the size of the patched cell was observed. The cells patched were rounded. Within 10 min after the change to a hypotonic solution (230 mosmol l−1), the cell diameters had increased by up to typically 50 %. Increase in cell diameter, as viewed by the microscope, always preceded the increase in current; increase in current was apparent about 2 min after cell swelling was observed. A time course of the response can be seen in Fig. 3. This may indicate that, under the given condition of whole-cell patch-clamp recording of trypsinized cells, cells have to swell visibly before showing an increase in current. After changing back to an isotonic solution, the cells shrank to their original size within 3 min.

Figure 3.

ATP dependence of volume-activated chloride currents in rat brain endothelial cells

Whole-cell currents at +80 mV and -80 mV in response to a hypotonic solution when ATP was present in the pipette solution (circles) and when ATP was omitted (squares). Change from isotonic to hypotonic solution is indicated by an arrow. Data are shown as means of 5 experiments.

Pharmacological characterization of the volume-activated chloride current

To characterize the type of chloride current present in the rat brain endothelial cells, we investigated the effects of the chloride transport blockers NPPB and DIDS. At 100 μM, NPPB blocked the hypotonicity-induced outward and inward chloride currents at an applied potential of -80 mV by 92.4 ± 5.2 % (P < 0.05,n= 5) and at +80 mV by 90.8 ± 8.3 % (P < 0.05,n= 5). Figure 2A shows current traces of a typical cell exposed to hypotonicity and the effect of the subsequent addition of NPPB and the corresponding I-V relationships of five experiments. By contrast DIDS (100 μM) blocked only the outward current evoked by hypotonicity, at +80 mV, inhibiting it by 87.1 ± 7.3 % (P < 0.05,n= 5). The inward current was not sensitive to DIDS, showing a current reduction at -80 mV of only 0.2 ± 14.6 % (n.s., n= 5). This finding is consistent with an earlier observation that hypotonicity-induced iodide efflux, which is a measure of anion permeability and which is equivalent to the inward current, was not inhibited by DIDS (von Weikersthal et al. 1997). A similar response to DIDS was described by Gosling et al. (1995). Figure 2B shows the current traces and the I-V relationships which illustrate the influence of DIDS. The effects of NPPB and DIDS were completely reversible. Following wash-out of the drugs, the hypotonic solution brought back the normal increase in current.

Figure 2.

Effect of NPPB (100 μM) (A), DIDS (100 μM) (B) and tamoxifen (10 μM) (C) on volume-sensitive chloride currents in rat brain endothelial cells

A, left, whole-cell currents in response to voltage pulses, applied from -100 to +100 mV in 20 mV steps. Right, current-voltage relationships obtained in isotonic (○), hypotonic (•) and hypotonic + NPPB (▪) solutions. Data are shown as means ±s.e.m., n= 5. B, left, whole-cell currents in response to voltage pulses, applied from -100 to +100 mV in 20 mV steps. Right, current-voltage relationships obtained in isotonic (○), hypotonic (•) and hypotonic + DIDS (▪) solutions. Data are shown as means ±s.e.m., n= 5. C, left, whole-cell currents in response to voltage pulses, applied from -100 to +100 mV in 20 mV steps. Right, current-voltage relationships obtained in isotonic (○), hypotonic (•) and hypotonic + tamoxifen (▪) solutions. Data are shown as means ±s.e.m., n= 5.

Tamoxifen, an antioestrogen and a known chloride channel blocker, at 10 μM inhibited reversibly both inward and outward currents, at -80 mV by 97.7 ± 3.1 % (P < 0.05,n= 5) and at +80 mV by 99.7 ± 6.3 % (P < 0.05,n= 5) of control values (Fig. 2A).

Activation of the chloride current was dependent upon intracellular ATP. When ATP (4 mM) was omitted from the pipette solution (n= 5), no increase in chloride conductance was observed after a hypotonic stimulus (Fig. 3). The activation was not dependent on intracellular Ca2+, as all recordings were made with 1 mM EGTA and 0.01 mM EDTA present in the pipette solution to chelate intracellular Ca2+. These results are in agreement with previous observations from iodide efflux experiments (von Weikersthal et al. 1997).

We investigated the possible modulatory effect of protein kinase C (PKC) by external application of the PKC activating phorbol ester PDBu. Application of PDBu (100 nM) caused no reduction in outward or inward currents if the volume-sensitive chloride current was already elicited. However, if cells were exposed to PDBu before or concomitantly with hypotonic challenge, then no increase in current was observed. At an applied potential of +80 mV the current was 19.3 ± 3.2 pA, and at -80 mV it was -2.9 ± 1.1 pA (n= 5), which is not significantly different from the current under isotonic conditions. To ensure that each cell under investigation was capable of responding to hypotonic challenge, we adopted the procedure of testing each cell both before drug treatment and after drug wash-out. In every case, hypotonicity-induced responses were obtained in the absence of drug. Figure 4 shows current traces which illustrate the effect of PDBu according to this protocol.

Figure 4.

Effect of PDBu (100 nM) on volume-sensitive chloride currents in rat brain endothelial cells

Whole-cell currents in response to voltage pulses, applied from -100 to +100 mV in 20 mV steps. Recordings were made once steady state had been reached. Thereafter solutions were changed.

To characterize the channel further we determined the effect of extracellularly applied ATP. At 10 mM, ATP significantly reduced, but did not abolish, the hypotonicity-activated current, lowering the current at +80 mV by 74.6 ± 7.8 % (P < 0.05,n= 5). At -80 mV an even smaller inhibition of the current of only 29.3 ± 6.1 % (n= 5) was observed. Figure 5A shows the I-V relationships of the currents before and after inhibiton by ATP.

Figure 5.

Effect of ATP (10 mM) (A) and verapamil (100 μM) (B) on volume-sensitive chloride currents in rat brain endothelial cells

A, effect of ATP on current-voltage relationships. Recordings were made in isotonic (○), hypotonic (•) and hypotonic + ATP (▪) solutions. Data are shown as means ±s.e.m., n= 5. B, effect of verapamil on current-voltage relationships. Recordings were made in isotonic (○), hypotonic (•) and hypotonic + verapamil (▪) solutions. Data are shown as means ±s.e.m., n= 5.

The effect of verapamil was also determined. At a concentration of 100 μM, verapamil decreased the hypotonicity-activated current at +80 mV by 67.2 ± 9.3 % (P < 0.05,n= 5) and at -80 mV by 46.6 ± 5.8 % (P < 0.05,n= 5). Figure 5B shows the I-V relationships of the currents before and after inhibition by verapamil.

Molecular characterization of volume-activated chloride current

The RT-PCR technique was used to investigate the molecular expression of mRNA of the chloride channels ClC-2, ClC-3 and ClC-5, and of pICln. With primers specific to the rat cDNA sequences of ClC-2 and ClC-3, bands corresponding to the expected fragment size, i.e. 259 bp and 277 bp, were obtained from RNA prepared from cultured rat brain endothelial cells. We also detected a good band with a fragment size of 246 bp with primers for pICln. A further member of the ClC family, the ClC-5 channel, could also be identified (fragment size 284 bp). As an internal control, we used β-actin, which gave a band of the expected size of 296 bp. Figure 6A shows a representative gel of one sample out of four analysed.

Figure 6.

RT-PCR analysis of ClC-2, ClC-3, ClC-5, pICln and β-actin mRNA expression in rat brain endothelial cells

A, RT-PCR using total RNA. The primers used are as follows: lane 1, ClC-2 (259 bp); lane 2, ClC-3 (277 bp); lane 3, ClC-5 (284 bp); lane 4, pICln (246 bp); lane 5, β-actin (296 bp); and lane 6, molecular weight marker (100 bp ladder from Gibco). B, PCR using total RNA, leaving out the RT step. The primers used are as follows: lane 1, ClC-2 (259 bp); lane 2, ClC-3 (277 bp); lane 3, ClC-5 (284 bp); lane 4, pICln (246 bp); lane 5, β-actin (296 bp); and lane 6, molecular weight marker.

To exclude possible cDNA contamination of our samples, we designed the primers for β-actin in such a way as to amplify a cDNA product extending over two neighbouring exon regions. The intervening intron between these regions is so small that the presence of any genomic DNA would produce a fragment of 421 bp. As we did not detect a band of this size (Fig. 6A, lane 5), we can exclude DNA contamination of our samples. Further confirmation of this was obtained by omitting the RT step that would normally generate cDNA for subsequent amplification. No fragments were obtained in the absence of reverse transcription with any of the primers used (Fig. 6A).


Regulatory volume change represents a vital adaptive response of cells to osmotic challenge. In particular, brain endothelial cells, which form the main structural component of the BBB, must avoid significant changes in shape or size that could compromise their barrier function. Volume regulation in these cells is continuously challenged by large fluxes of solutes such as glucose and amino acids, which must pass through them from the blood to the brain. With alterations in intracellular osmolarity, controlled efflux of chloride, potassium and organic osmolytes occurs to compensate for changes in cell volume and these processes provide important volume regulatory mechanisms. This study presents evidence of functionally active volume-sensitive chloride channels in rat brain endothelial cells. Exposure to hypotonic solution leading to cell swelling is accompanied by the activation of a current that has the characteristics of that of a chloride current. To establish the possible identity of the channel or channels responsible for this, we compare here our functional and molecular data with properties and characteristics of volume-sensitive chloride currents already described by others.

The first candidate to consider is the ClC-2 channel, which is known to have widespread tissue distribution, with high mRNA levels in brain, kidney, heart and liver. This channel has been expressed in Xenopus oocytes where it gives rise to a functional chloride channel activated by strong hyperpolarization and by extracellular hypotonicity. It shows inward rectification and an anion selectivity of Cl > Br > I (Thiemann et al. 1992; Gründer et al. 1992). The current is only moderately inhibited by classical chloride channel blockers, and DIDS is almost ineffective. Such properties are very different from those we describe here in brain endothelial cells suggesting that the volume-sensitive currents in brain endothelial cells are not mediated by a ClC-2 channel even though mRNA encoding ClC-2 is detectable in these cells.

The ClC-3 channel was also found in brain endothelial cells on a mRNA level. This channel was first characterized by Kawasaki et al. (1994) as an ubiquitously expressed protein, with high levels in brain, kidney and lung. Expression of ClC-3 in Xenopus oocytes produced a large time-independent chloride current which was slightly outward rectifying, with an anion selectivity of I > Cl= Br > acetate > gluconate, and was inhibited by DIDS and the PKC activators TPA and PDBu. Subsequently, Duan et al. (1997) have shown that functional expression of a cardiac clone of ClC-3 from guinea-pig in NIH/3T3 cells resulted in a volume-sensitive chloride conductance with similar characteristics, which are outward rectification, time- and voltage-dependent inactivation at potentials higher than +80 mV, an anion selectivity of I > Cl as well as inhibition by DIDS, tamoxifen, extracellular application of ATP, and by the PKC activator PDBu. These characteristics are identical to the electrophysiological data we obtained in brain endothelial cells and as our molecular characterization showed a high level of ClC-3 mRNA, we propose that ClC-3 is likely to be the channel that mediates volume-sensitive currents in brain endothelial cells, or at least participates in the volume response. Recent findings by Yamazaki et al. (1998) support this link as they have shown volume-regulated chloride currents with similar functional characteristics in canine vascular smooth muscle cells, which they relate to a high level of mRNA ClC-3 expression in these cells.

ClC-3 contains two putative phosphorylation sites and phosphorylation by PKC is known to modulate the channel properties (Kawasaki et al. 1994). The ClC-3 mediated chloride currents observed by Duan et al. (1997) were indeed inhibitable by the PKC activator PDBu after the channel was activated by hypotonic stress. In brain endothelial cells prior exposure to PDBu can prevent channel opening, but once the channel is activated by hypotonicity, it can no longer be inhibited by PDBu. Presumably inhibition of channel activation by phosphorylation is only possible when the channel is in a non-activated state.

As pICln mRNA could be detected in the brain endothelial cells, we also consider whether this protein might be involved in the volume regulatory processes observed. Opinions about the identity and function of pICln differ; some believe it to be a volume-sensitive chloride channel, others suggest it is a regulator of a channel and still others consider that it has no specific link at all. Paulmichl et al. (1992) observed that, when expressed in Xenopus oocytes, pICln elicited a chloride conductance under isotonic conditions. The current, named ICln, exhibited outward rectification, inactivation at strong positive potentials, sensitivity to block by DIDS and NPPB and inhibition by extracellular ATP. A swelling-induced current with apparently the same characteristics has been described in NIH/3T3 cells. This could be reduced by treatment with antisense oligonucleotides against pICln, suggesting pICln to be a volume-sensitive chloride channel (Gschwentner et al. 1995). However, an investigation by Ackerman et al. (1994) revealed that Xenopus oocytes already possess an endogenous volume-sensitive chloride current, named ICl.swell, with properties similar to those of the current described by Paulmichl et al. (1992). Monoclonal antibodies against pICln could inhibit this endogenous ICl.swell (Krapivinski et al. 1994). It was suggested that pICln, identified as a soluble cytosolic protein which forms specific complexes with actin and other cytosolic proteins, might be involved as a sensor or mediator in the change of cell volume, rearrangement of cytoskeletal components, and activation of chloride channels (Krapivinsky et al. 1994, 1998). A model introduced by Coca-Prados et al. (1996) integrates the function of ClC-3 and pICln. Their hypothesis was that ClC-3 provides a channel for chloride movement directly regulated by PKC, and pICln is a volume-sensitive regulator which controls the movement of chloride through the ClC-3 channel by a mechanism yet to be identified. The biophysical and functional properties of the chloride current we report here found in brain endothelial cells are very similar to those of ICl,swell described by Ackerman et al. (1994) but whether pICln is responsible or at least involved in volume responses in our cells remains to be determined.

It has also been suggested that P-glycoprotein (P-gp) may have a function in volume regulatory mechanisms. In the original experiments of Valverde et al. (1992), overexpression of P-gp in NIH/3T3 fibroblasts resulted in the appearance of a swelling-activated outward rectifying chloride current dependent on intracellular ATP and blockable by DIDS and NPPB. A similar current detected in certain P-gp-expressing cell lines was shown to be blockable by various inhibitors of P-gp and this led to the idea that P-gp might have a dual role as a drug efflux pump and a chloride channel (Gill et al. 1992; Mintenig et al. 1993). These suggestions have since been questioned as there appears to be no direct correlation between P-gp expression and the presence of swelling-activated currents (McEwan et al. 1992; Dong et al. 1994; Altenberg et al. 1994; Tominaga et al. 1995). More recently it has been suggested that P-gp may function as a channel regulator (Han et al. 1996; Wu et al. 1996; Valverde et al. 1996), possibly overexpression of P-gp altering cytoskeleton-membrane interactions, which then modify the activity of volume-sensitive channels (Strange et al. 1996). Though we have shown already that brain endothelial cells express P-gp (Barrand et al. 1995), in studies in which we used iodide efflux as a measure of anion permeability (von Weikersthal et al. 1997), we were unable to see any effective block with verapamil and other substrates of P-gp. In the present study, we see that even a high concentration of verapamil (100 μM), well above that required to completely block P-gp activity, produces only a 60 % reduction in the hypotonicity-induced current. Though tamoxifen, another P-gp substrate, completely inhibits the volume-activated current, this agent is known to block currents whether or not P-gp is expressed (Nilius et al. 1994) and is probably acting as a direct chloride channel blocker (Zhang et al. 1994) rather than a specific substrate for P-gp-associated chloride currents. We have thus no clear evidence in brain endothelial cells for involvement of P-gp.

As rat brain endothelial cells have shown ClC-5 mRNA expression we also consider whether this channel might mediate volume-sensitive chloride currents. The ClC-5 chloride channel is predominantely expressed in the kidney, with lower amounts in the brain, the liver and other tissues (Steinmeyer et al. 1995). Two studies have shown that ClC-5 expression in Xenopus oocytes elicits an outward rectifying chloride current; however, the current characteristics differ in the two studies. Steinmeyer et al. (1995) reported an anion selectivity of Cl > Br > I, and almost no effect on the current by the chloride channel blockers DIDS and NPPB, whereas Lindenthal et al. (1997) observed an anion selectivity of I > Cl > gluconate and sensitivity to block by DIDS. In a recent study Schmieder et al. (1998) expressed in Xenopus oocytes a ClC-5 construct in which they replaced native non-coding regions of ClC-5 with a sequence which facilitates protein expression. With a high ClC-5 protein expression it is more likely that this chloride channel is actually responsible for the observed current. The obtained electrophysiological data, which show similarities to those described by Steinmeyer et al. (1995), are different from the current characteristics we observed in brain endothelial cells, therefore suggesting that the volume-sensitive currents are not mediated by a ClC-5 channel.

In this study we have demonstrated that brain endothelial cells possess volume-activated chloride currents, which share many characteristics of volume-sensitive chloride currents described in other cells. Our data so far suggest that ClC-3 forms the actual channel which mediates the volume-activated current, and leave open the possibility that pICln may have a regulatory role. As we found mRNA of several ClC genes, it is likely that brain endothelial cells express several members of the ClC family in parallel. Their localization, e.g. on cell membrane or intracellular compartments, and therefore their function may differ. There is also the possibility that coexpression of different ClC members in the same cell may lead to the formation of heteromultimeric ClC channels (Lorenz et al. 1996).


We wish to thank The Wellcome Trust for their generous support for this work and for the salary of S. F. v. W. S. F. v. W. is also very grateful to Atticus Hainsworth and Rafael Rosales for helpful advice on patch-clamp techniques.