Acta Ph.ysiol Scand 1992, 144, 191-198
C hlor ide -selective channels of large cond uctance in bovine aortic endothelial cells S.-P. O L E S E N and M. B U N D G A A R D Department of General Physiology & Biophysics, T h e Panum Institute, University of Copenhagen, Denmark OLESEN, S.-P. & BUNDGAARD, M. 1992. Chloride-selective channels of large conductance in bovine aortic endothelial cells. Acta Physiol Scand 144, 191-198. Received 7 July 1991, accepted 26 September 1991. ISSN 0001-6772. Department of General Physiology & Biophysics, The Panum Institute, University of Copenhagen, Denmark. Single-channel currents of an anionic channel in the plasma membrane of cultured bovine aortic endothelial cells have been recorded with the patch-clamp technique. The channel is selective for chloride over cations, and has an average single channel conductance of 382 picosiemens in symmetric 140 millimoles of chloride. In addition to the main conductance state it shows well-defined subconductance states of about 50, 100, 150 and 200 picosiemens. The channel is very active at membrane potentials close to 0 mV, but steps to either positive or negative membrane potentials above +20 millivolt lead to a rapid inactivation of the channel. Changes in the concentrations of free calcium or andenosine tri-phosphate on the cytosolic surface do not influence channel activity. The chloride channel rarely opens at resting membrane potential, but it may help repolarize endothelial cells following depolarizing stimuli. Key wurds : anion channel, electron microscopy, endothelial cell, large conductance voltage-dependent chloride channel, patch-clamp technique, Weibel-Palade body.
Activation of chloride channels is important for various cellular functions such as reduction of neuronal excitability (Hamill et al. 1983), transepithelial transport (Christensen et al. 1989, Liedtke 1989) and regulation of cell volume (Worrell et al. 1989). Chloride channels are present in the cell membranes of most cells, and there exists a large variety of different C1- channels (Frizzel 1987). However, chloride channels have not yet been studied in endothelial cells. Vascular endothelial cells lining the inner surface of arterioles and larger arteries release nitric oxide, prostacyclin and other vasoactive mediators (Moncada et al. 1988) and play an important role in the regulation of blood flow. Activation of endothelial cells with physical or chemical stimuli may lead to the opening of K+ channels (bradykinin, Colden-Stanfield et al. 1987; ATP, SauvC et al. 1988; acetylcholine, Correspondence : Dr. S.-P. Olesen, NeuroSearch A/S, 26 Smedeland, DK-26000 Glostrup, Denmark.
Olesen et al. 1988a; fluid shear-stress, Olesen et al. 1988b) or non-selective cation channels (stretch, Lansman et al. 1987; histamine, Bregestovski et al. 1988). I n addition, all endothelial cells contain an inward rectifier K+ channel having a conductance of a few pS in the depolarized range (Colden-Stanfield et a!. 1987, Sauve et al. 1988, Olesen et al. 1988a). We report here that bovine aortic endothelial cells also express a large conductance, chlorideselective channel. T h e channel is active at membrane potentials around 0 mV, and it does not require Ca2+ or ATP for activity.
M A T E R I A L S A N D METHODS Endothelial cells. Endothelial cells were mechanically isolated (scraped) from calf aortae and cultured, using a modification of the techniques outlined by Ryan et al. (1980). Subcultures, from passages 14, showing the characteristic cobblestone morphology were used for the present experiments. Samples were routinely immunostained with antibodies specific for von Willebrand factor (factor VIII). Other samples were
191
192
S.-P. Olesen and -41. Birndgirnrd
Fig. 1. (a) 1 . o magnification ~ electron micrograph ofboiine aortic endothelial cells in culture. Bar, 2 p i . (b) The overlapping zone of t i + o adjacent endothelial cells. The intercellular cleft appears obliterated by junctional contacts and appositions (arrow). Bar, 0.2 pm. (c) Segment of the nuclear zone of an endothelial cell. The arrow marks a \\'eibel-Palade body. Bar, 0 . j pm. selected for electron microscop!- and initially fixed in aldehydes. They %ere further processed in Petri dishes according to standard EhI-procedures. Thin sections of the cells were examined in a Zeiss 10B electron microscope (Zeiss hlP31, Oberkochen, Germany). Putc-/7-ckrmp t d i n i q u e . The cultured bo\ ine aortic endothelial cells were plated on glass cmerslips and studied for the following 3 days. Cell membrane currents were determined under Yoltage clamp conditions using the patch-clamp technique in cell-attached or inside-out recording modes. The signals were recorded nith a List EPC-7 amplifier, stored on a digital tape recorder and displayed on a chart recorder. The pipette solution always contained (in mxt): K - , 144; C1 , 140; Ca2-, 1; Mg", 2 ; HEPES. 10; pII \\-as 7.4. The bath contained extracellular solution in cellattached experiments and intracellular solution in inside-out experiments. The ion composition of the extracellular solution was as follow (in mzi): Na-,
140; K-, 4; CIF, 110; Ca2+, 1; Mg2+,2 ; HEPES, 10; pH was adjusted to 7.1 with HCI. The intracellular solution contained (in mM): K+, 146; CIF, 140; Ca2+, 1; \lg'-, 2 ; HEPES, 10; the EGTA concentration a-as 1.9 msi gi\ing a free Ca" concentration of 100 nZ1; p H \%asadjusted to 7.2 with HCI. .All agents were administered to the bath (10Opl volume) at a slow rate of 0.5 ml min-' in order to atoid stimulation of shear-stress-activated channels (Olesen er ul. 1988b). T = 2 2 + 2 "C.
R E S UI, T S Immunofluorescence microscopy showed the granular intracellular expression of factor VIII related antigen, typical for vascular endothelium in culture. I n the electron microscope the monolayer appeared a s flattened, overlapping cells (Fig. 1). Neighbouring cell membranes showed focal appositions or contacts, reminiscent
Endothelial chloride channels
193
60 mV
I: 20 mV
I + 10 mV c
Fig. 2. Voltage dependency of two endothelial chloride channels in an inside-out patch. The membrane potential ( V , ) was changed from 0 mV to 10, 20, or 60 mV as indicated on the upper tracing, and the outward currents were recorded for a period of 30 s. At a membrane potential of 10 mV the two channels stayed open for the whole period, whereas at V, = 60 mV both channels closed within 2 seconds. Arrows indicate the current level corresponding to the closed state of the
channels.
of tight junctions. The Weibel-Palade bodies, which are specific for endothelial cells and represent the factor VIII positive granula observed by immunofluorescence microscopy (Wagner et al. 1982), were occasionally encountered in the perinuclear cytoplasm.
Endothelial chloride channels The cultured bovine aortic endothelial cells express a large C1- selective ion channel. In the cell-attached recording mode it was seen only in 9 out of 64 patches, but it was often activated when the patch was excised and then observed in 30 out of 52 inside-out patches. The maximal number of channels seen in an inside-out patch was two. The results below are obtained on inside-out patches unless otherwise stated. The activity of the channel is voltagedependent. At a membrane potential of 0 mV or i l O m V the channel opened with an open probability close to 1, and it did not inactivate. Following steps in membrane potential from 0 mV to & 20 mV or numerically larger values,
the channel inactivated within seconds (Fig. 2). There was no significant difference in inactivation time after steps to positive or negative potentials, e.g. at a membrane potential of +60 mV the inactivation time was 4.0 s and at -60 mV it was 3.3 s. Although the channel rapidly closed at potentials of f60 mV, rare openings were observed at potentials as high as 90 mV. Furthermore, at the larger potentials the channel did not always stay closed following the initial inactivation, but often reopened briefly seconds later. In the standard solutions with a symmetric [Cl-] of 140 mM the single channel conductance was 382 pS (SD = 40pS, n = 18) and the reversal potential was 0 mV (Fig. 3 ) . The current-voltage relationship was linear in the voltage range studied from - 80 to +80 mV. Changing the solution on the cytosolic side to one containing 70 mM C1- by substituting 70 mM Cl- with gluconate reduced the single channel conductance to 268 pS (SD = 30 pS, n = 5 ) and the reversal potential was shifted to - 16.1 mV (SD = 1.4 mV, n = 5). This shift is
194
S.-P. Olesen (inn .+I. Bundgaard vm 20 rnV [CI-], = 70 m M
c
[CI 1, = 140 rnM -20 mV
ue c
-30 mV
V,
(mv)
c
270 pS
-40 mV
360 pS c
-1 0
-50 rnV
-60 mV
'
--
c
J
-20
4s
Fig, 3. Single chloride channel current as a function of membrane potential in an inside-out patch. The left panel shows the chloride currents at various membrane potentials obtained in standard solutions. The membrane potential was changed from 0 mV to the values indicated at the beginning of the traces. The right panel s h o w the current-voltage relationship of the chloride channel obtained with standard solutions on both sides (symmetric 140 mM [Cl-1) and after replacement of 70 r n l chloride with gluconate on the inside. Changing [Cl-1, to half concentration reduced the single channel conductance and shifted the reversal potential by 17.2 mV to the left, indicating a strong anionic sclecti\-it!-.
close to the change in chloride equilibrium potential calculated from the Nernst equation ( - 17.8 mi'), indicating a strong anion selectivity. I n -5 inside-out patch experiments, the bath solution was changed from the standard intracellular solution containing 116 m>r K- to one containing 146 mar Na'. T h i s substitution of cations did not significantly shift the currentvoltage relationship including single channel conductance and reversal potential. Several subconductance states of the channel were seen in addition to the main conductance state of 382 pS. Subconductances of about 50 pS, 100 pS, 150 pS, 200 pS in addition to several less well defined states were found (Fig. 4). .inalysis of the data depicted in Figure 4 shows that the subconductance states are less common than the main conductance state (Np, (50 pS) = 0.0451; Np,, (150 pS) = 0.0041; Np, (350 pS) = 0,1172) and that their mean open time (1,) is signiticantl!- shorter ( t o (50 pS) = 15.0 m s ; to (1.50 pS) = 3.1 m s ; t(,(350 pS) = 80.0 ms). T h e
reversal potentials of these subconductance states were the same as the main open level, and the conductances and reversal potentials of the subconductance states changed in the same proportion as the main conductance state when the intracellular [Cl-] was reduced to half concentration. T h e dependence of the chloride channel activity on the intracellular ligands Ca2+ and .4TP was studied by changing the concentration of the ligands in the bath in inside-out patch experiments. Free calcium concentrations of 1W6, lo-', and M were obtained by replacing KCI with K-EGTA at concentrations of 0, 1.07, 1.9, and 11 mM respectively, while keeping the total intracellular [Ca"] at 1 mM. T h e chloride channel activity recorded at membrane potentials of - 70- 60 mV was independent of the changes in free intracellular [Ca'+] ( n = 5 ; Fig. 5 ) . I n contrast to the C1channels the inward rectifying K+ channels also present in Figure 5 were sensitive to Ca2+
+
Endothelial chloride channels
195
(a)
,
350 pS
150 pS
50 pS
3s
25
0
1 8
8
12
16
Amplitude (PA)
Fig. 4. Subconductance states of endothelial chloride channel. (a). Tracing showing several subconductance states as well as the main conductance state of the chloride channel in inside-out patch. V, = -40 mV. (b). Amplitude histogram of the data in Figure 6a showing peaks around 2, 6 and 12-14 pA being equivalent to 50, 150 and 300-350 pS (V, = -40 mV). The ordinate is
logarithmic. although in low concentrations, i.e. the K+ channels opened normally in free [Ca2+], of 100 nM or above, but always closed at 10 nM free [Ca”],. Further, the C1- channel activity was not affected by changes in the ATP concentration in the bath from 0 to 0.1 and 1 mM (n = 3). Administration of the metabolic blockers cyanide (1 mM) plus iodo-acetate (1 mM) for a period up to 1; h was also without effect on C1- channel activity ( n = 9). In cellattached recordings performed with an extracellular solution in the bath and thus a physiological resting membrane potential of the endothelial cells the chloride current was inwardly directed at a pipette potential of 0 mV
and the open probability ranged from 0.01 to 0.06 (n = 9). The reversal potential of the chloride channel in this situation was 13 mV (SD = 5 mV, n = 4) depolarized in relation to the resting membrane potential. DISCUSSION The activation of single C1- channels has been recorded in membrane patches of aortic endothelial cells. The identity of the cells has been verified by localization of factor VIII in their cytoplasm and by the ultrastructural characteristics. The conductance of the main open level of the channel is 382 p s (symmetrical 140 mM
cl-),
[Ca2+l.(nM) I
1000
Po
20s Fig. .5. Inremiti\ it! of chloride channels to changes in calcium concentration on the cytosolic surtice of inside-out patch. T h c tracing shous the acti\-it!- of two innard rectifying K- channels e-ihihiting long regular openings as well as a chloride channel with brief openings and probably \arious substates. Changing the internal calcium concentration from 1000 nhf to 10 n v closed the
inward rectif! ing I(-channels, hut the chloride channels were unaffected. The tracing was filtered 100 H z by the strip chart recorder. I = - 60 m i - .
;II
and the channel expresses several subconduct- levels has been reported for a neuronal anion ance lelels. T h e channel is permeable to C1- channel (Geletyuk 8r Kazachenko 1983). T h e activitl- of the channel clearly depends on 15 ith no significant cation-permeabilit!. I n inside-out patches the channel opens spon- a factor in the cytosolic medium, since only 14(?0 taneously at transmembrane potentials b e t w e n of the patches showed channel actkity in cell-- 10 and 10 mi7, hut outside this range it rewrts attached recordings, in contrast to 58?& of the to the closed state with characteristic J-oltage- inside-out patches. Kolb & U b l (1987) studied dependent relaxation times. cell-attached patches of macrophages, which ere silent prior to administration of zymosan, This large-conductance anion-selective channel has not been observed in I ascular endothelial but expressed large numbers of the big C1cells earlier, but it has been reported in a wide channels minutes after the administration of this I-arietJ-of-othcr cell types. It mas discowxed by acti\-ator, and they proposed that a second lagleb! (1983) in rat muscle, but is also messenger such as Ca2' or a mediator of channel found in e.g. epithelial cells (Krouse e l a/. 1986, phosphorylation could be involved in the reguChristensen e l a / . 1989), I1-l!-mphoc!-tes (Bosma lation of the CI- channel activity. I n the 1989), macrophages (Kolb &- Ubl 1987) and endothelial cells we have studied two candidates neuroblastoma cells (Bolotina et o/. 198T). -411are for this intracellular ligand, Ca2+and ATP, and maximall!- open around 0 m i - and tend to close found that both were ineffective in changing the a ~ a yfrom that loitage. 'I'he voltage range in CI- channel activitj-. However, it is likely that which the endothelial channels are obseryed the channel is regulated by an intracellular ( - 80- + 80 m i - ) is about the same as for the B- mediator not yet identified. Iyniphoc!-te channel (Bosma 1989), but in some Single ion channels of vascular endothelial cell t!pes it n a r r o w down to i Z 5 m l - . T h e cells h a w been studied by several groups dependence of the single channel conductance (Lansman et a / . 1987, Rregestovski et al. 1988, on the chloride concentration seen in endothelial Sauve PI u/. 1988), but CI- channels have not cell5 has also been found in other cells (Blatz &- been identified before. T h e reason may be that Magleb!- 1983, Schn-arze & Kolb 1984). the C1- channel is not always expressed and that T h e equally spaced subconductance levels of certain conditions favour its activation, such as about 50. 100, 1.50 and 200 pS propose that the described for the macrophages (Kolb & Ubl endothelial CI- channel could be made u p of 7 1987). Thus, it is possible that the channel is identical subunits each with a pore. Krouse et a / . present in many endothelial cells in an inactive (1086) and Holotina r t 171. (1987) arrib-ed at the state. same number, but as man!- as 16 subconductance T h e physiological function of the large-
Endothelial chloride channels conductance C1- channel is not known. Schwarze & Kolb (1984) proposed that the channel could be a hemi-channel of a gap-junctional pore, and this idea was based on the high single channel conductance, its non-specific appearance in many cell types, its partial permeability to cations in some cells, as well as the voltage sensitivity and kinetic properties appearing somewhat similar to those of gap-junctional channels. Vascular endothelial cells express gap-junctional channels zn vivo and in vitro (Davies 1986), but the activity of purified hemi-channels has not been characterized, and so far the relationship between hemichannels and large anion channels remains speculation. I n a resting bovine aortic endothelial cell the membrane potential is about -77 mV (Olesen et al. 1988a), so due to the voltage sensitivity of the C1- channel its activity will be very low under resting conditions. However, if the endothelial cell is depolarized towards 0 mV, the channel will open and repolarize the cell towards the C1- equilibrium potential, which was determined to be 13 mV more positive than the resting membrane potential, i.e. about -64 mV (giving an intracellular C1- concentration of z 11.5 mM). T h u s following a depolarizing stimulus such as histamine or stretch (Lansman et al. 1987, Bregestovski et a/. 1988) the C1- channel may help repolarize an endothelial cell with a conductance which is many times higher than that of the inward rectifier Kfchannel present in the endothelial cells (Colden-Stanfield et al. 1987, SauvC et al. 1988).
197
conductance levels in a mouse B lymphocyte cell line. 3 Physiol 410, 67-90. BREGESTOVSKI, P., BAKHRAMOV, A., DANILOV,S., MOLDOBAEVA, A. & TAKEDA, K. 1988'. Histamineinduced inward currents in cultured endothelial cells from human umbilical vein. Br 3 Pharmacol 95, 429436. O., SIMON,M. & RANDLEV, T. 1989. CHRISTENSEN, Anion channels in a leaky epithelium. PfliiRers .4rc.h 415, 3746. COLDEN-STANFIELD, M., SCHILLING, W.P., RITCHIE, L.T. & KUNZE, D.L. A.K., ESKIN,S.G., NAVARRO, 1987. Bradykinin-induced increases in cytosolic calcium and ionic currents in cultured bovine aortic endothelial cells. Circ Res 61, 632-640. DAVIES,P.F. 1986. Biology of disease. Vascular cell interactions with special reference to the parhogenesis of atherosclerosis. Lab Inzest 55, 5-24. FRIZZELL, R.A. 1987. Cystic fibrosis: A disease of ion channels? Trends Neurosci 10, 19C193. V.N. 1985. Single GELETYUK, V.E. & KAZACHENKO, C1- channels in molluscan neurones : multiplicity of the conductance states. 3 Membrane Bid 86, 9-15. J. & SAKMANN, B. 1983. HAMILL,O.P., BORMANN, Activation of multiple-conductance state chloride channels in spinal neurones by glycine and GABA. Nature 305, 805-808. KOLB,H.-A. & UBL, J. 1987. Activation of anion channels by zymosan particles in membranes of peritoneal macrophages. Biochim Bioph,ys Actn 899, 239-246. KROUSE, M.E., SCHNEIDER, G.T. & GAGE,P.W. 1986. A large anion-selective channel has seven conductance levels. Nature 319, 58-60, LANSMAN, J.B., HALLAM, T.J. & RINK,T.J. 1985. Single stretch-activated ion channels in vascular endothelial cells as mechanotransducers ? Nuture 325, 811-813. The authors are grateful to the late Professor Christian LIEDTKE, C.M. 1989. Regulation of chloride transport Crone for his inspiration and support. Pia Hagman, in epithelia. Ann Rev Physiol 51, 143-160. Ellen Munch and Ingrid Kjzr provided excellent MONCADA, S., PALMER, R.M.J. & HIGGS,E.A. 1988. technical assistance. The work was supported by the The discovery of nitric oxide as the endogenous Johan and Hanne Weimann Foundation, the Carlsberg nitrovasodilator. €€$pertension 12, 365-372. Foundation (no. 1677/78) and the Danish Medical OLESEN, S.-P., DAVIES, P.F. & CLAPHAM, D.E. 1988a. Research Council (no. 12-7039/40, 12-8560). Muscarinic-activated K' current in bovine aortic endothelial cells. Circ Res 62, 1059-1064. D.E. & DAVIES, P.F. 1988b. OLESEN, S.-P., CLAPHAM, REFERENCES Haemodynamic shear stress activates a K+ current BLATZ,A.L. & MAGLEBY, K.L. 1983. Single voltagein vascular endothelial cells. Nature 331, 168-170. dependent chloride-selective channels of large RYAN,U.S., MORTARA, M. 8z WHITAKER, C. 1980. conductance in cultured rat muscle. Biophys J' 43, Methods for microcarrier culture of bovine pul237-241. monary artery endothelial cells avoiding the use of BOLOTINA, V., BORECKV, J., VLACHOVA, V., BAUDY- enzymes. Tissue & Cell 12, 619-635. L., SIMONEAU, C. & ROY,G. 1988. ~ , PARENT, sovtl, M. & VYSKOCIL, F. 1987. Voltage-dependent S A U VR., External ATP triggers a biphasic activation process chloride channels with several substates in excised of a calcium-dependent K' channel in cultured patches from mouse neuroblastoma cells. Neurosci bovine aortic endothelial cells. Pjugers A r c h 412, Lett 77, 298-302. 469-48 1. BOSMA.M.M. 1989. Anion channels with multiple
198
S.-P. Olesen and M . Bundgaard
SCHWARZE, W. & KOLB, H.-A 1984. voltagedependent kinetics of an anionic channel of large unit conductance in macrophages and myotube membranes. PJIUKCYS Arch 102, 281-291. WAGNER, D.D., OLMSTED, J.B. & MARDER, V.J. 1982. Immunolocalization of von Willebrand protein in
Weibel-Palade bodies of human endothelial cells. 3. Cell Biol 95, 355-360. WORRELL, R.T., BUTT,A.G., CLIFF,W.H. & FRIZZELL, R.A. 1989. .4 volume-sensitive chloride conductance in human colonic cell line T84. A m 3 Physiol 256, C1 111LC1119.
C hlor ide -selective channels of large cond uctance in bovine aortic endothelial cells S.-P. O L E S E N and M. B U N D G A A R D Department of General Physiology & Biophysics, T h e Panum Institute, University of Copenhagen, Denmark OLESEN, S.-P. & BUNDGAARD, M. 1992. Chloride-selective channels of large conductance in bovine aortic endothelial cells. Acta Physiol Scand 144, 191-198. Received 7 July 1991, accepted 26 September 1991. ISSN 0001-6772. Department of General Physiology & Biophysics, The Panum Institute, University of Copenhagen, Denmark. Single-channel currents of an anionic channel in the plasma membrane of cultured bovine aortic endothelial cells have been recorded with the patch-clamp technique. The channel is selective for chloride over cations, and has an average single channel conductance of 382 picosiemens in symmetric 140 millimoles of chloride. In addition to the main conductance state it shows well-defined subconductance states of about 50, 100, 150 and 200 picosiemens. The channel is very active at membrane potentials close to 0 mV, but steps to either positive or negative membrane potentials above +20 millivolt lead to a rapid inactivation of the channel. Changes in the concentrations of free calcium or andenosine tri-phosphate on the cytosolic surface do not influence channel activity. The chloride channel rarely opens at resting membrane potential, but it may help repolarize endothelial cells following depolarizing stimuli. Key wurds : anion channel, electron microscopy, endothelial cell, large conductance voltage-dependent chloride channel, patch-clamp technique, Weibel-Palade body.
Activation of chloride channels is important for various cellular functions such as reduction of neuronal excitability (Hamill et al. 1983), transepithelial transport (Christensen et al. 1989, Liedtke 1989) and regulation of cell volume (Worrell et al. 1989). Chloride channels are present in the cell membranes of most cells, and there exists a large variety of different C1- channels (Frizzel 1987). However, chloride channels have not yet been studied in endothelial cells. Vascular endothelial cells lining the inner surface of arterioles and larger arteries release nitric oxide, prostacyclin and other vasoactive mediators (Moncada et al. 1988) and play an important role in the regulation of blood flow. Activation of endothelial cells with physical or chemical stimuli may lead to the opening of K+ channels (bradykinin, Colden-Stanfield et al. 1987; ATP, SauvC et al. 1988; acetylcholine, Correspondence : Dr. S.-P. Olesen, NeuroSearch A/S, 26 Smedeland, DK-26000 Glostrup, Denmark.
Olesen et al. 1988a; fluid shear-stress, Olesen et al. 1988b) or non-selective cation channels (stretch, Lansman et al. 1987; histamine, Bregestovski et al. 1988). I n addition, all endothelial cells contain an inward rectifier K+ channel having a conductance of a few pS in the depolarized range (Colden-Stanfield et a!. 1987, Sauve et al. 1988, Olesen et al. 1988a). We report here that bovine aortic endothelial cells also express a large conductance, chlorideselective channel. T h e channel is active at membrane potentials around 0 mV, and it does not require Ca2+ or ATP for activity.
M A T E R I A L S A N D METHODS Endothelial cells. Endothelial cells were mechanically isolated (scraped) from calf aortae and cultured, using a modification of the techniques outlined by Ryan et al. (1980). Subcultures, from passages 14, showing the characteristic cobblestone morphology were used for the present experiments. Samples were routinely immunostained with antibodies specific for von Willebrand factor (factor VIII). Other samples were
191
192
S.-P. Olesen and -41. Birndgirnrd
Fig. 1. (a) 1 . o magnification ~ electron micrograph ofboiine aortic endothelial cells in culture. Bar, 2 p i . (b) The overlapping zone of t i + o adjacent endothelial cells. The intercellular cleft appears obliterated by junctional contacts and appositions (arrow). Bar, 0.2 pm. (c) Segment of the nuclear zone of an endothelial cell. The arrow marks a \\'eibel-Palade body. Bar, 0 . j pm. selected for electron microscop!- and initially fixed in aldehydes. They %ere further processed in Petri dishes according to standard EhI-procedures. Thin sections of the cells were examined in a Zeiss 10B electron microscope (Zeiss hlP31, Oberkochen, Germany). Putc-/7-ckrmp t d i n i q u e . The cultured bo\ ine aortic endothelial cells were plated on glass cmerslips and studied for the following 3 days. Cell membrane currents were determined under Yoltage clamp conditions using the patch-clamp technique in cell-attached or inside-out recording modes. The signals were recorded nith a List EPC-7 amplifier, stored on a digital tape recorder and displayed on a chart recorder. The pipette solution always contained (in mxt): K - , 144; C1 , 140; Ca2-, 1; Mg", 2 ; HEPES. 10; pII \\-as 7.4. The bath contained extracellular solution in cellattached experiments and intracellular solution in inside-out experiments. The ion composition of the extracellular solution was as follow (in mzi): Na-,
140; K-, 4; CIF, 110; Ca2+, 1; Mg2+,2 ; HEPES, 10; pH was adjusted to 7.1 with HCI. The intracellular solution contained (in mM): K+, 146; CIF, 140; Ca2+, 1; \lg'-, 2 ; HEPES, 10; the EGTA concentration a-as 1.9 msi gi\ing a free Ca" concentration of 100 nZ1; p H \%asadjusted to 7.2 with HCI. .All agents were administered to the bath (10Opl volume) at a slow rate of 0.5 ml min-' in order to atoid stimulation of shear-stress-activated channels (Olesen er ul. 1988b). T = 2 2 + 2 "C.
R E S UI, T S Immunofluorescence microscopy showed the granular intracellular expression of factor VIII related antigen, typical for vascular endothelium in culture. I n the electron microscope the monolayer appeared a s flattened, overlapping cells (Fig. 1). Neighbouring cell membranes showed focal appositions or contacts, reminiscent
Endothelial chloride channels
193
60 mV
I: 20 mV
I + 10 mV c
Fig. 2. Voltage dependency of two endothelial chloride channels in an inside-out patch. The membrane potential ( V , ) was changed from 0 mV to 10, 20, or 60 mV as indicated on the upper tracing, and the outward currents were recorded for a period of 30 s. At a membrane potential of 10 mV the two channels stayed open for the whole period, whereas at V, = 60 mV both channels closed within 2 seconds. Arrows indicate the current level corresponding to the closed state of the
channels.
of tight junctions. The Weibel-Palade bodies, which are specific for endothelial cells and represent the factor VIII positive granula observed by immunofluorescence microscopy (Wagner et al. 1982), were occasionally encountered in the perinuclear cytoplasm.
Endothelial chloride channels The cultured bovine aortic endothelial cells express a large C1- selective ion channel. In the cell-attached recording mode it was seen only in 9 out of 64 patches, but it was often activated when the patch was excised and then observed in 30 out of 52 inside-out patches. The maximal number of channels seen in an inside-out patch was two. The results below are obtained on inside-out patches unless otherwise stated. The activity of the channel is voltagedependent. At a membrane potential of 0 mV or i l O m V the channel opened with an open probability close to 1, and it did not inactivate. Following steps in membrane potential from 0 mV to & 20 mV or numerically larger values,
the channel inactivated within seconds (Fig. 2). There was no significant difference in inactivation time after steps to positive or negative potentials, e.g. at a membrane potential of +60 mV the inactivation time was 4.0 s and at -60 mV it was 3.3 s. Although the channel rapidly closed at potentials of f60 mV, rare openings were observed at potentials as high as 90 mV. Furthermore, at the larger potentials the channel did not always stay closed following the initial inactivation, but often reopened briefly seconds later. In the standard solutions with a symmetric [Cl-] of 140 mM the single channel conductance was 382 pS (SD = 40pS, n = 18) and the reversal potential was 0 mV (Fig. 3 ) . The current-voltage relationship was linear in the voltage range studied from - 80 to +80 mV. Changing the solution on the cytosolic side to one containing 70 mM C1- by substituting 70 mM Cl- with gluconate reduced the single channel conductance to 268 pS (SD = 30 pS, n = 5 ) and the reversal potential was shifted to - 16.1 mV (SD = 1.4 mV, n = 5). This shift is
194
S.-P. Olesen (inn .+I. Bundgaard vm 20 rnV [CI-], = 70 m M
c
[CI 1, = 140 rnM -20 mV
ue c
-30 mV
V,
(mv)
c
270 pS
-40 mV
360 pS c
-1 0
-50 rnV
-60 mV
'
--
c
J
-20
4s
Fig, 3. Single chloride channel current as a function of membrane potential in an inside-out patch. The left panel shows the chloride currents at various membrane potentials obtained in standard solutions. The membrane potential was changed from 0 mV to the values indicated at the beginning of the traces. The right panel s h o w the current-voltage relationship of the chloride channel obtained with standard solutions on both sides (symmetric 140 mM [Cl-1) and after replacement of 70 r n l chloride with gluconate on the inside. Changing [Cl-1, to half concentration reduced the single channel conductance and shifted the reversal potential by 17.2 mV to the left, indicating a strong anionic sclecti\-it!-.
close to the change in chloride equilibrium potential calculated from the Nernst equation ( - 17.8 mi'), indicating a strong anion selectivity. I n -5 inside-out patch experiments, the bath solution was changed from the standard intracellular solution containing 116 m>r K- to one containing 146 mar Na'. T h i s substitution of cations did not significantly shift the currentvoltage relationship including single channel conductance and reversal potential. Several subconductance states of the channel were seen in addition to the main conductance state of 382 pS. Subconductances of about 50 pS, 100 pS, 150 pS, 200 pS in addition to several less well defined states were found (Fig. 4). .inalysis of the data depicted in Figure 4 shows that the subconductance states are less common than the main conductance state (Np, (50 pS) = 0.0451; Np,, (150 pS) = 0.0041; Np, (350 pS) = 0,1172) and that their mean open time (1,) is signiticantl!- shorter ( t o (50 pS) = 15.0 m s ; to (1.50 pS) = 3.1 m s ; t(,(350 pS) = 80.0 ms). T h e
reversal potentials of these subconductance states were the same as the main open level, and the conductances and reversal potentials of the subconductance states changed in the same proportion as the main conductance state when the intracellular [Cl-] was reduced to half concentration. T h e dependence of the chloride channel activity on the intracellular ligands Ca2+ and .4TP was studied by changing the concentration of the ligands in the bath in inside-out patch experiments. Free calcium concentrations of 1W6, lo-', and M were obtained by replacing KCI with K-EGTA at concentrations of 0, 1.07, 1.9, and 11 mM respectively, while keeping the total intracellular [Ca"] at 1 mM. T h e chloride channel activity recorded at membrane potentials of - 70- 60 mV was independent of the changes in free intracellular [Ca'+] ( n = 5 ; Fig. 5 ) . I n contrast to the C1channels the inward rectifying K+ channels also present in Figure 5 were sensitive to Ca2+
+
Endothelial chloride channels
195
(a)
,
350 pS
150 pS
50 pS
3s
25
0
1 8
8
12
16
Amplitude (PA)
Fig. 4. Subconductance states of endothelial chloride channel. (a). Tracing showing several subconductance states as well as the main conductance state of the chloride channel in inside-out patch. V, = -40 mV. (b). Amplitude histogram of the data in Figure 6a showing peaks around 2, 6 and 12-14 pA being equivalent to 50, 150 and 300-350 pS (V, = -40 mV). The ordinate is
logarithmic. although in low concentrations, i.e. the K+ channels opened normally in free [Ca2+], of 100 nM or above, but always closed at 10 nM free [Ca”],. Further, the C1- channel activity was not affected by changes in the ATP concentration in the bath from 0 to 0.1 and 1 mM (n = 3). Administration of the metabolic blockers cyanide (1 mM) plus iodo-acetate (1 mM) for a period up to 1; h was also without effect on C1- channel activity ( n = 9). In cellattached recordings performed with an extracellular solution in the bath and thus a physiological resting membrane potential of the endothelial cells the chloride current was inwardly directed at a pipette potential of 0 mV
and the open probability ranged from 0.01 to 0.06 (n = 9). The reversal potential of the chloride channel in this situation was 13 mV (SD = 5 mV, n = 4) depolarized in relation to the resting membrane potential. DISCUSSION The activation of single C1- channels has been recorded in membrane patches of aortic endothelial cells. The identity of the cells has been verified by localization of factor VIII in their cytoplasm and by the ultrastructural characteristics. The conductance of the main open level of the channel is 382 p s (symmetrical 140 mM
cl-),
[Ca2+l.(nM) I
1000
Po
20s Fig. .5. Inremiti\ it! of chloride channels to changes in calcium concentration on the cytosolic surtice of inside-out patch. T h c tracing shous the acti\-it!- of two innard rectifying K- channels e-ihihiting long regular openings as well as a chloride channel with brief openings and probably \arious substates. Changing the internal calcium concentration from 1000 nhf to 10 n v closed the
inward rectif! ing I(-channels, hut the chloride channels were unaffected. The tracing was filtered 100 H z by the strip chart recorder. I = - 60 m i - .
;II
and the channel expresses several subconduct- levels has been reported for a neuronal anion ance lelels. T h e channel is permeable to C1- channel (Geletyuk 8r Kazachenko 1983). T h e activitl- of the channel clearly depends on 15 ith no significant cation-permeabilit!. I n inside-out patches the channel opens spon- a factor in the cytosolic medium, since only 14(?0 taneously at transmembrane potentials b e t w e n of the patches showed channel actkity in cell-- 10 and 10 mi7, hut outside this range it rewrts attached recordings, in contrast to 58?& of the to the closed state with characteristic J-oltage- inside-out patches. Kolb & U b l (1987) studied dependent relaxation times. cell-attached patches of macrophages, which ere silent prior to administration of zymosan, This large-conductance anion-selective channel has not been observed in I ascular endothelial but expressed large numbers of the big C1cells earlier, but it has been reported in a wide channels minutes after the administration of this I-arietJ-of-othcr cell types. It mas discowxed by acti\-ator, and they proposed that a second lagleb! (1983) in rat muscle, but is also messenger such as Ca2' or a mediator of channel found in e.g. epithelial cells (Krouse e l a/. 1986, phosphorylation could be involved in the reguChristensen e l a / . 1989), I1-l!-mphoc!-tes (Bosma lation of the CI- channel activity. I n the 1989), macrophages (Kolb &- Ubl 1987) and endothelial cells we have studied two candidates neuroblastoma cells (Bolotina et o/. 198T). -411are for this intracellular ligand, Ca2+and ATP, and maximall!- open around 0 m i - and tend to close found that both were ineffective in changing the a ~ a yfrom that loitage. 'I'he voltage range in CI- channel activitj-. However, it is likely that which the endothelial channels are obseryed the channel is regulated by an intracellular ( - 80- + 80 m i - ) is about the same as for the B- mediator not yet identified. Iyniphoc!-te channel (Bosma 1989), but in some Single ion channels of vascular endothelial cell t!pes it n a r r o w down to i Z 5 m l - . T h e cells h a w been studied by several groups dependence of the single channel conductance (Lansman et a / . 1987, Rregestovski et al. 1988, on the chloride concentration seen in endothelial Sauve PI u/. 1988), but CI- channels have not cell5 has also been found in other cells (Blatz &- been identified before. T h e reason may be that Magleb!- 1983, Schn-arze & Kolb 1984). the C1- channel is not always expressed and that T h e equally spaced subconductance levels of certain conditions favour its activation, such as about 50. 100, 1.50 and 200 pS propose that the described for the macrophages (Kolb & Ubl endothelial CI- channel could be made u p of 7 1987). Thus, it is possible that the channel is identical subunits each with a pore. Krouse et a / . present in many endothelial cells in an inactive (1086) and Holotina r t 171. (1987) arrib-ed at the state. same number, but as man!- as 16 subconductance T h e physiological function of the large-
Endothelial chloride channels conductance C1- channel is not known. Schwarze & Kolb (1984) proposed that the channel could be a hemi-channel of a gap-junctional pore, and this idea was based on the high single channel conductance, its non-specific appearance in many cell types, its partial permeability to cations in some cells, as well as the voltage sensitivity and kinetic properties appearing somewhat similar to those of gap-junctional channels. Vascular endothelial cells express gap-junctional channels zn vivo and in vitro (Davies 1986), but the activity of purified hemi-channels has not been characterized, and so far the relationship between hemichannels and large anion channels remains speculation. I n a resting bovine aortic endothelial cell the membrane potential is about -77 mV (Olesen et al. 1988a), so due to the voltage sensitivity of the C1- channel its activity will be very low under resting conditions. However, if the endothelial cell is depolarized towards 0 mV, the channel will open and repolarize the cell towards the C1- equilibrium potential, which was determined to be 13 mV more positive than the resting membrane potential, i.e. about -64 mV (giving an intracellular C1- concentration of z 11.5 mM). T h u s following a depolarizing stimulus such as histamine or stretch (Lansman et al. 1987, Bregestovski et a/. 1988) the C1- channel may help repolarize an endothelial cell with a conductance which is many times higher than that of the inward rectifier Kfchannel present in the endothelial cells (Colden-Stanfield et al. 1987, SauvC et al. 1988).
197
conductance levels in a mouse B lymphocyte cell line. 3 Physiol 410, 67-90. BREGESTOVSKI, P., BAKHRAMOV, A., DANILOV,S., MOLDOBAEVA, A. & TAKEDA, K. 1988'. Histamineinduced inward currents in cultured endothelial cells from human umbilical vein. Br 3 Pharmacol 95, 429436. O., SIMON,M. & RANDLEV, T. 1989. CHRISTENSEN, Anion channels in a leaky epithelium. PfliiRers .4rc.h 415, 3746. COLDEN-STANFIELD, M., SCHILLING, W.P., RITCHIE, L.T. & KUNZE, D.L. A.K., ESKIN,S.G., NAVARRO, 1987. Bradykinin-induced increases in cytosolic calcium and ionic currents in cultured bovine aortic endothelial cells. Circ Res 61, 632-640. DAVIES,P.F. 1986. Biology of disease. Vascular cell interactions with special reference to the parhogenesis of atherosclerosis. Lab Inzest 55, 5-24. FRIZZELL, R.A. 1987. Cystic fibrosis: A disease of ion channels? Trends Neurosci 10, 19C193. V.N. 1985. Single GELETYUK, V.E. & KAZACHENKO, C1- channels in molluscan neurones : multiplicity of the conductance states. 3 Membrane Bid 86, 9-15. J. & SAKMANN, B. 1983. HAMILL,O.P., BORMANN, Activation of multiple-conductance state chloride channels in spinal neurones by glycine and GABA. Nature 305, 805-808. KOLB,H.-A. & UBL, J. 1987. Activation of anion channels by zymosan particles in membranes of peritoneal macrophages. Biochim Bioph,ys Actn 899, 239-246. KROUSE, M.E., SCHNEIDER, G.T. & GAGE,P.W. 1986. A large anion-selective channel has seven conductance levels. Nature 319, 58-60, LANSMAN, J.B., HALLAM, T.J. & RINK,T.J. 1985. Single stretch-activated ion channels in vascular endothelial cells as mechanotransducers ? Nuture 325, 811-813. The authors are grateful to the late Professor Christian LIEDTKE, C.M. 1989. Regulation of chloride transport Crone for his inspiration and support. Pia Hagman, in epithelia. Ann Rev Physiol 51, 143-160. Ellen Munch and Ingrid Kjzr provided excellent MONCADA, S., PALMER, R.M.J. & HIGGS,E.A. 1988. technical assistance. The work was supported by the The discovery of nitric oxide as the endogenous Johan and Hanne Weimann Foundation, the Carlsberg nitrovasodilator. €€$pertension 12, 365-372. Foundation (no. 1677/78) and the Danish Medical OLESEN, S.-P., DAVIES, P.F. & CLAPHAM, D.E. 1988a. Research Council (no. 12-7039/40, 12-8560). Muscarinic-activated K' current in bovine aortic endothelial cells. Circ Res 62, 1059-1064. D.E. & DAVIES, P.F. 1988b. OLESEN, S.-P., CLAPHAM, REFERENCES Haemodynamic shear stress activates a K+ current BLATZ,A.L. & MAGLEBY, K.L. 1983. Single voltagein vascular endothelial cells. Nature 331, 168-170. dependent chloride-selective channels of large RYAN,U.S., MORTARA, M. 8z WHITAKER, C. 1980. conductance in cultured rat muscle. Biophys J' 43, Methods for microcarrier culture of bovine pul237-241. monary artery endothelial cells avoiding the use of BOLOTINA, V., BORECKV, J., VLACHOVA, V., BAUDY- enzymes. Tissue & Cell 12, 619-635. L., SIMONEAU, C. & ROY,G. 1988. ~ , PARENT, sovtl, M. & VYSKOCIL, F. 1987. Voltage-dependent S A U VR., External ATP triggers a biphasic activation process chloride channels with several substates in excised of a calcium-dependent K' channel in cultured patches from mouse neuroblastoma cells. Neurosci bovine aortic endothelial cells. Pjugers A r c h 412, Lett 77, 298-302. 469-48 1. BOSMA.M.M. 1989. Anion channels with multiple
198
S.-P. Olesen and M . Bundgaard
SCHWARZE, W. & KOLB, H.-A 1984. voltagedependent kinetics of an anionic channel of large unit conductance in macrophages and myotube membranes. PJIUKCYS Arch 102, 281-291. WAGNER, D.D., OLMSTED, J.B. & MARDER, V.J. 1982. Immunolocalization of von Willebrand protein in
Weibel-Palade bodies of human endothelial cells. 3. Cell Biol 95, 355-360. WORRELL, R.T., BUTT,A.G., CLIFF,W.H. & FRIZZELL, R.A. 1989. .4 volume-sensitive chloride conductance in human colonic cell line T84. A m 3 Physiol 256, C1 111LC1119.
