272

Brain Research. 584 (1992) 272-2,'%

Elsevier Science Publishers B.\;. BRES 17884

Mechanotransducing ion channels in astrocytes Charles

L. Bowman

a,

Jiu-Ping

Ding

b,

Frederick

Sachs

a

and Masahiro

Sokabe

c

a Department of Biophysical Sciences, School of Medicine, University of New York at Buffalo, Buffalo, N Y 14214 (USA), b Biology Department, Washington Unicersity, St. Louis, MO 63130 (USA) and c Department of Physiology, Nagoya University School of Medicine, Showa-Ku, Nagoya (Japan)

(Accepted 11 February 1992)

Key words: Astrocyte; Ion channel; Glial cell; Mechanotransduction

Ion channels present on the soma of neonatal rat astrocytes in primary cell culture were studied using the single channel recording technique. Ion channels were activated by changing the pressure in the back of the pipette. The morphological structure of the patch membrane was examined while recording channel activity. One class of channel was activated by increasing the pipette pressure (curvature-sensitive or CS channels). CS channels were observed in 150 mM KCI, 150 mM NaC1, or 150 mM sodium gluconate. At constant pressure the closed times decreased with depolarization. CS channels had a conductance of 50 pS in 150 mM NaC1, and displayed an inwardly rectifying current-voltage relationship. CS channel activity was found only in cell-attached patches, and were active only when the patch membrane curved towards the soma. The other class of channel was found to be activated by both suction and pressure (stretch-activated or SA channels). Four SA conductance levels were found: 360, 230, 144, and 70 pS in 150 mM KCI. Each conductance displayed a linear current-voltage relationship. At negative membrane potentials SA channels were inhibited by Cs ÷, Ba 2+ or Na ÷. The relationship between average mechanosensory current and pressure was biphasic for SA channels and monophasic for CS channels. Combinations of SA and CS channels could be observed in the same patch. We propose that CS channels are non-specific cation channels which sense membrane tension only when the patch membrane is in a specific, permissive curvature. SA channels appear to be K+-selective channels that sense membrane tension independent of the direction of curvature.

INTRODUCTION T h e r e a r e a growing n u m b e r o f r e p o r t s suggesting t h a t a v a r i e t y o f m e m b r a n e c o m p o n e n t s in astrocytes sense a n d t r a n s d u c e c h a n g e s in m e m b r a n e tension. F o r e x a m p l e o s m o t i c c h a n g e s e v o k e r e l e a s e of taurine 24'26'35, g l u t a m a t e , a n d a s p a r t a t e 24, a n d a d e p o l a r ization of t h e m e m b r a n e p o t e n t i a l 21-z3. Slightly changing t h e o s m o t i c p r e s s u r e strongly affects b o t h t h e isoproterenol-evoked and K+-stimulated release of t a u r i n e 26'27, t o u c h i n g t h e astrocyte s o m a with a glass m i c r o - n e e d l e evokes c h a n g e s in [Ca2+]in6'3°, a n d applying suction to the b a c k of t h e p i p e t t e causes t h e a p p e a r a n c e o f n o n - s e l e c t i v e c a t i o n c h a n n e l s in cella t t a c h e d p a t c h e s o n r e t i n a l m u l l e r cells 37. P r o m p t e d by t h e o s m o t i c studies a n d by the growing n u m b e r of r e p o r t s o f s t r e t c h - a c t i v a t e d ion c h a n n e l s 33'38, we d e c i d e d to d e t e r m i n e if a s t r o c y t e s have similar ion channels. This p a p e r d e s c r i b e s t h e m e c h a n o t r a n s d u c -

ing ion c h a n n e l s p r e s e n t on the s o m a o f n e o n a t a l rat astrocytes in p r i m a r y cell culture. M e c h a n i c a l l y sensitive ion c h a n n e l s have b e e n rep o r t e d in b o t h p l a n t a n d a n i m a l k i n g d o m s , at all levels of the p h y l o g e n e t i c l a d d e r , a n d in n e a r l y all types of cells 33. S e l e c t i o n m e c h a n i s m s strongly favor r e t e n t i o n o f m e c h a n o t r a n s d u c t i o n d u r i n g evolution, a n d it is likely t h a t such c h a n n e l s p l a y a f u n d a m e n t a l role in c e l l u l a r physiology. U s e s for such c h a n n e l s at the cellular level can b e envisioned, i n c l u d i n g t h e c o n t r o l o f cell size, cell s h a p e a n d cell division. M e c h a n o t r a n s d u c i n g ion c h a n n e l s m a y also p l a y a role in o s m o r e g u lation 7,39,4°, cell motility 32, a n d e m b r y o g e n e s i s 29. U n f o r t u n a t e l y , t h e r e a r e n o specific a n t a g o n i s t s for m e c h a n o s e n s o r y c h a n n e l s so t h a t t h e i r f u n c t i o n o u t s i d e of specific m e c h a n o r e c e p t o r s has n o t y e t b e e n d e m o n strated. T w o k i n d s o f m e c h a n o s e n s i t i v e c h a n n e l s have b e e n r e p o r t e d , t h o s e t h a t i n c r e a s e activity with m e m b r a n e

Correspondence: C.L. Bowman, 320 Cary Hall, Department of Biophysical Sciences, School of Medicine, State University of New York at Buffalo,

Buffalo, NY 14214, USA. Fax: (1) (716) 831-3395.

273 tension t7'a2 and those that decrease activity with membrane tension 32. The channels that activate with tension are called stretch-activated (SA) channels while those that inactivate with tension are called stretch-inactivated (SI) channels. The gating of such channels appears to be a function of membrane tension, which is related to the transmembrane pressure by Laplace's law. In the present study we show that astrocytes have SA channels and also a novel type of mechanically sensitive channel whose activation is not a simple function of membrane tension. The gating of this channel appears to be associated not only with tension, but with the direction of curvature of the patch membrane. We call this channel a curvature-sensitive, or CS channel. CS channels are non-selective for cations, voltage-dependent and have one conductance level (50 pS). There are four SA channels with conductance levels 360, 230, 144, and 70 pS. SA channels are K ÷ selective. Taking the two gating classes together there are five different mechanosensitive ion channels in neonatal rat astrocytes. Preliminary accounts of this work have been reported in abstract form 1°-12. MATERIALS AND METHODS

Cell cultures Astrocytes were grown in primary cell culture using the method of Frangakis and Kimeiberg ~6 with two modifications. Briefly, the cerebral cortices of six 1-day-old rats were removed and the cells dissociated using protease (type IX; EC 3.4.24.4; Sigma Chemical Co., St. Louis, MO). The cells were plated at a density of 2.0x104 cells/cm 2 on 22 mm z square glass coverslips, and grown in Eagle's MEM growth medium (Gibco, Grand Island, NY) containing Earles's salts, L-glutamine and fetal bovine serum (10%), and cultured in a 95%/air, 5%CO 2 environment at 37°C (Stericult incubator, model 3033, Forma Scientific, Marietta, OH). On average, 80-90% of the cells stain for glial fibrillary acidic protein (GFAP) 16, a marker considered specific for astrocytes1'13. For this study the age of the cells in culture ranged from 2 to 20 weeks. There appeared to be no relationship between the age of the cells in culture and the appearance of SA or CS channels. In growth medium the cells are flat and more difficult to patch clamp. To simplify patching the cells were rounded by replacing the growth medium with a Ringer's solution at 37°C containing (mM): 150 NaCI, 5 KC1, 1 CaC12, 0.5 MgCI2, 10 HEPES, and 1 dibutyryl-cAMP (N'-2'-O-dibutyryladenosine 3' :5'cyclic monophosphate; Sigma Chemical Co.) pH 7.4. The cells were incubated in this medium at 37°C for several hours before study. As a control for drug effects dibutyryl-cAMP was omitted in about half of the experiments. The appearance of SA or CS channels was not dependent on pretreatment of the cells with dibutyryl-cAMP. After incubation the coverslips containing the cells were placed in a chamber, washed with fresh Ringer's solution, and studied electrophysiologically. All experiments were carried out at room temperature (20-28°C). In using the method of Frangakis and Kimelberg 16 to make the cell cultures we found that an appropriate cellular yield, as controlled by the degree of stirring, was critical in obtaining cellular preparations having the characteristic properties of astrocytes in cell culture, and stretch channels. Use of a stirring bar (25.4 x 9.5 ram; 4-6 rotations per s) in a 30 ml beaker (30 mm diameter base)

resulted in the normal cellular yields of about 1 x 106 cells per brain. In such preparations the percentage of cells with GFAP-staining and the CS channels are both greater than 80% (six cell cultures). By contrast use of a smaller (12.6x3.3 mm) stirring bar resulted in cellular yields 10-fold less, a decrease in the percentage of GFAPstaining cells ( < 10 percent), and less than 10 percent of the cells showing CS or SA channels (four cell cultures). Our attention was drawn to the stirring bar because another laboratory, using identical lot numbers of reagents and a larger stirring bar did not have the problem of low cellular yields (H.K. Kimelberg, personal communication). It appears that the larger stirring bar more closely mimics the effects of tituration or vortexing procedures commonly used in starting cultures of brain cells. For the studies reported here the larger stirring bar was used during the dissociation of the cortices. Cells selected for patching were rounded-up and contained multiple process, some of which terminated in end-foot-like structures. These properties are characteristics of astrocytes2°.

Electrophysiology Standard patch-clamp methodsTM were used to record from gigaseals. An Axon Instruments (Burlingame, CA) Axopatch patchclamp was used to observe the channel currents. The channel currents were stored on a JVC video recorder using a PCM encoder (VR-10, Instrutech Corp., Long Island, NY). For analysis data was played back through an 8-pole low-pass Bessel filter (Frequency Devices, MA) and re-digitized at five-times the cut-off frequency with a PC/AT-based acquisition system (Data Translation Inc., Marlborough, MA; model 2801-A). The computer programs IPROC, LPROC (Axon Instruments), and PAT (version 6.0, a generous gift of John Dempster, Department of Physiology, University of Strathclyde, Glasgow, Scotland) were used to create and analyze amplitude histograms. The analyses were carried out on a Texas Instruments 286-based Business Pro and a Standard-286 computer (Standard Brand Products, Austin, TX). Micropipettes were made on a Sachs-Flaming micropipette puller (model PC-84, Brown-Flaming Instruments, CA) using borosilicate glass tubing (Drummond 100 lambda Microcaps, Drummond Scientific Co., Broomall, PA). The pipettes were coated with Sylgard 184 (DOW Coming Corp., Midland, MI) to within 100/~m of the tip, and annealed using a hair dryer. The tips were viewed at 500× and heat-polished using a home-built microforge. The polishing was stopped after the appearance of a change in tip morphology. Pipette resistances were 4-7 M r / w h e n measured with a series of 10 mV test pulses. Stress was applied to the patch by varying the pressure in the patch pipette. Pressure was monitored to an accuracy of + 1 mmHg using either a mercury column or a pneumatic pressure transducer (model DPM-1B, Bio-Tek Instruments, Winooski, VT). The pipette solutions were either (in mM): 150 KCI, 5 KEGTA, 10 HEPES, pH 7.4 (using KOH), or 120 KC1, 20 HEPES, pH 7.4. Positive pressure (2-4 mmHg) was kept on the pipette prior to sealing to keep the tip clear of particulate contamination. Gigaseals were readily formed by slowly increasing the suction from - 5 to 10 mmHg, and recordings were made from the soma of the cell. The reference electrode consisted of an agar bridge (2%) in normal Ringer's solution. To measure the membrane potential and input resistance after gigaseal formation we applied suction ( - 60 to - 130 mmHg) or, surprisingly, pressure (+50 to +150 mmHg) to the pipette. This procedure frequently resulted in a large increase in current noise, a decrease in resistance (generally from 5 x 109 to 24 + 10× 106 .Q, 30 cells), and new offset in the patch-clamp. Subtracting the pipette resistance left an input resistance of 10-20 x 106 12. Switching from voltage-clamp to current-clamp mode or zeroing out the current in voltage-clamp mode yielded a potential of - 59 + 14 mV (38 cells). Because the potential depolarized with increases in [K÷ ]out or glutamate (10 -4 M) (data not shown) we interpreted the zero current potential to be the membrane potential. After gigaseal formation we generally found a small current offset of about 1 pA. This current offset was zeroed out by adjusting the pipette potential to - 3 . 8 + 1 . 7 mV (12 cells). This potential appeared similar to that reported using KCl-filled pipettes 15. We did not correct the membrane potentials for this potential. The membrane potential and input resistance measured with -

274 patch pipettes are in general agreement with previous studies performed at higher temperatures using conventional microelectrodes 3-5.

could be carried out to a resolution of a few ( < 2) s. Pressures were maintained between 20 and 179 s so that the alignment error was small for the analysis.

l,Tdeo microscopy

Selection of patches for imaging

A Zeiss Axiovert 35M microscope (AZI Inc., Avon, MA) equipped for differential interference (Nomarski) optics was used to view the patch. The objective was a 63 x / 1 . 4 0 NA lens. A water immersion lens (40 X/0.75 NA) served as the condenser. A special holder was built to mount the water immersion lens. The microscope lenses and pipette were aligned such that the shear axis was parallel to the pipette axis. The image was magnified 6 x , and sent to a CCD camera for display on a TV monitor. The final image was displayed at a magnification of 58 nM/pixel. Images were archived on a S-VHS recorder, and analyzed off line. Photos were made using a video copy processor having 512x512 line resolution and a 6-bit contrast dynamic range (model P61U, Mitsubishi Electric Sales America Inc., Cypress, CA). Details concerning pipette construction, holders, chamber construction, and electronics have been published 4~. Channel data was digitized using a digital data recorder (VR-10, Instrutech Corp.), and stored on a separate VCR. Voice data concerning pressure changes was recorded on both VCRs. Using this audio information subsequent alignment of both tapes

In most patches, the patch membrane went up the pipette beyond the viewing range (about 25 ~m) during gigaseal formation so that these patches could not be analyzed visually. We obtained two patches (out of approximately 15 patches) which stayed within viewing range and had only CS channels. Similar results were obtained in these two patches.

A

-1 0 mm Hg

Currents and potentials Inward currents are shown as downward deflections. In the cell-attached configuration a downward deflection indicates that positive ions are moving from pipette to cell. For the cell-attached configuration, membrane potentials (MP) were calculated from the following equation: MP = - ( 5 9 mV + pipette potential).

Fitting dwell-time histograms We used John Dempster's PAT software for fitting exponential probability density functions to dwell-time histograms of the CS

B

-7 mm Hg

[~

-1 2 mm Hg

[

C

-4 mm Hg

Fig. 1. The effect of pipette pressure on the direction-sensitive (CS) channel. Records shown are from ,the same ceU-attached .patch recording. A: - 1 0 mmHg. B: - 7 mmHg. C: - 4 mmHg. D: - 1 2 mmHg. The large downward, unresolved deflections are from the 360 pS SA channel, described later. Vertical and horizontal scales (lower right-hand side) represent 5 pA and 50 ms, respectively.Memhrane potential was - 9 0 mV. Pipette contained (in mM): 150 KCI, 5 KEGTA, 10 HEPES, pH 7.4. Bandwidth of channel records was limited to 1,000 Hz and sampled at 5,000 Hz. Patch 572478.

275

T

II

D

C

B

A

-110mY

-U-

- 1 50 mV

.< O-

2

C

0

V

1+20 mV

0

30

-2 (3

"~ c c 0 tO

-4 -6

¢) - 8 O~ .c -10

80 -I 50 -I 20

-90

-60

-30

membrane potenti01 (mY) Fig. 2. CS channel current-voltage relationship. Examples of CS channel activity shown in IA-D and IIA-D. For IA-D, pipette contained 150 mM NaCI, 5 NaEGTA, 10 HEPES, pH 7.4. In example IIA-D, pipette contained 10 NaC1, 5 NaEGTA, 10 HEPES, pH 7.4. Horizontal and vertical calibrations represent 0.250 s and 1 pA, respectively. Patch membrane potentials are listed for each letter (e.g. - 130 mV applies to IA and IIA, - 1 1 0 mV applies to IB and IIB, etc.). The cells' resting membrane potential was - 7 0 mV, as determined by breaking into the cell following the experiment. Pipette potentials were +60 mV (IA, IIA), +40 mV (IB, liB), 0 mV (IC, IIC) and -90 mV (ID, IID). Records were filtered at 500 Hz and sampled at 100 Hz. In all cases the bathing media contained 150 mM NaCl, 5 mM KC1, 1 mM CaCI2, 0.5 mM MgCI2 and 10 mM HEPES, pH 7.4. The lower figure shows the current-voltage relationship. For each point on the curve, the amplitudes of 6-8 single channel events at each membrane potential were measured using PAT (see Materials and Methods), and averaged. The vertical lines through the points represents the standard deviation in the single channel amplitudes at each potential. Resting membrane potentials were determined by establishing the whole-cell configuration at the conclusion of the experiment, n , 150 mM K+-gluconate (patch 574893); A, 150 mM NaCI (patch 580822); e, 120 mM K+-gluconate plus 30 mM KCI (patch 573015); *, 150 mM KCI (patch 484116); II, 10 mM KCI (patch 584820); • 10 mM NaCI (patch 590000); A, 10 mM NaC1 (patch 591056). The lines through the points represent the best fit of the Goldman-Hodgkin-Katz current equation to the current voltage data2. For 150 mM salts the GHK current equation was fit to closed circles representing the data for 120 mM potassium gluconate and 30 mM KCI. For the fit, a K + was assumed to carry the current. Calculated PK = 1.21 +0.04X 10 -7 cm/s. The internal and .external [K + ] were assumed to be equal (120 raM). For 10 raM salts the GHK current equation was fit to the open triangles representing the data for 10 mM NaC1. The calculated permeability coefficient was 1.91+0.04x 10 -7 cm/s. The external and internal [Na + ] were both assumed to be 10 mM. No improvement of the fit was obtained if two ions were assumed to carry the current. A non-linear least squares estimation program (nonlin.prg) available in the Gauss programming language (version 1.49B; Aptech Systems Inc., Kent, WA) was used to fit the data.

channel. For the SA channel, we used IPROC (Axon Instruments) for detection of single channel events, creation of an events list and dwell-time histograms. NFITS software (a generous gift from Dr. Chris Lingle, Washington University, St. Louis, MO) was used to analyze the dwell-time histograms.

RESULTS

Curvature-sensitive channels (CS channels) I n the cell-attached c o n f i g u r a t i o n at the soma increasing the p r e s s u r e from - 1 0 to - 7 m m H g , a n d t h e n to - 4 m m H g , i n c r e a s e d the f r e q u e n c y of a p p e a r ance of c h a n n e l s , r e p r e s e n t e d by d o w n w a r d deflections in the c u r r e n t trace (Fig. 1). Such deflections indicate inward c u r r e n t s (flowing from the p i p e t t e to cell). C h a n g i n g the p r e s s u r e to - 1 2 m m H g (Fig. 1D) re-

duces the f r e q u e n c y of c h a n n e l o p e n i n g s . T h e r e f o r e , the effect of i n c r e a s i n g the p r e s s u r e is reversible. I n o t h e r cells the r a n g e of pressures u s e d to activate the c h a n n e l varied from - 15 to + 60 m m H g . W e have n o t observed such c h a n n e l s i n patches excised into NaCI R i n g e r ' s solution ( > 30 cells tested). T h e pressures used to activate the CS c h a n n e l are below that which is necessary to establish the whole cell c o n f i g u r a t i o n (generally m o r e negative t h a n - 1 0 0 m m H g , or m o r e positive t h a n + 100 m m H g ) , , a n d we. have n o t observed CS c h a n n e l s at pressures m o r e negative t h a n - 2 5 m m H g ( > 20 cells). W h i l e CS c h a n n e l s are activated by i n c r e a s i n g t h e pressure, c h a n n e l activity a p p e a r s to b e r e l a t e d to the direction of c u r v a t u r e of the patch

276 A

10-

24-

i2-

(.)

20(:::3 0 16-

d

12-

C)

co

8-

O O4 O O

6-

(~ O3

0

c4

10

8

0

c5

6

O (.)

2-

4-

co (.Z3 0 O

84-

0

0 I''''I

0.0

....

2.5

I''''I''''I

5.0

7.5

time (sec)

10.0

[

I

I

I

0.0

0.4

0.8

1.2

time (sec)

F'

0.0

I

I

I

I

0.4

0.8

1.2

1.6

time (sec)

Fig. 3. CS channel closed dwell-times (A,B) and open dwell-times (C) were fit by a sum of two exponential probability distributions. Long closed dwell-times (A). Short closed dwell-times (B). Long and short open dwell-times shown in C are displayable on the same plot. For 3A, the long closed dwell-time constant was 1.60 + 0.186 s (77 + 7% of the total area). For 3B, the short closed dwell-time constant was 0.029 + 5 s (17 + 2% of the total area). For 3C, long open = 0.579 5:0.322 s (46 + 30% of the total area) and short open = 0.086 + 0.015 s (53 5: 5% of the total area). The standard deviations reflect the errors in the fit. The ASCII output of P A T was reformed to give c o u n t s / b i n width vs. time. Cell-attached patch, patch 574543. M e m b r a n e potential - 8 6 mV. Pipette contained 120 m M K gluconate, 30 m M KCI, 5 m M K E G T A , 10 m M HEPES, p H 7.4. Bandwidth was 100 Hz and sampling rate was 500 Hz.

membrane (see below). We therefore call this channel the curvature-sensitive (CS) channel. We have observed the CS channel in more than 60 cells in eleven different primary cell culture preparations. In three cell cultures approximately 90% of the cells showed CS channel activity, while 50% of the cells showed activity in three other celt cultures. One celt culture had little CS channel activity ( < 5%). Cells from preparations of low cellular yield invariably had a low CS activity ( < 10%; four cell cultures, see Materials and Methods). Such differences in the appearance of CS channels suggest a regulatory influence of the conditions of culture. To determine the ionic selectivity of CS channels we studied the effect of ionic replacement on the amplitudes of single events at different membrane potentials. Examples of such events at different membrane potentials are shown in Fig. 2, labeled IA-ID and IIA-IID. Single channel current-voltage curves were obtained using pipettes filled with 150 mM KCI, NaC1, and 150 potassium gluconate (Fig. 2). Replacing 150 mM Na + with 150 K ÷, or reducing [CI-] from 150 to 30 mM, was without effect on the slope conductance (50 pS). Replacing 150 mM C1- with 150 mM gluconate was also without effect on the slope conductance. The CS channel appears not to be C1--selective since an increase in inward current amplitude at negative membrane potentials should have been observed when [CI-] was reduced. With such a replacement the driving force for the outward movement of C1- in-

creases, assuming a c o n s t a n t [ C 1 - ] i n of 31-43 mM m. The lack of effect when K + was replaced with Na + suggests that the CS channel is a non-selective cation channel. For such a channel reducing the concentration of inward-current (cation) carriers should reduce the single channel current amplitude and shift the reversal potential in the negative direction. Such a shift in the reversal potential should follow the change in equilibrium potential of the cation carrying the current. The amplitude of CS channel events was reduced when 150 mM NaCl was replaced by 20 mM NaC1 (compare I A - D with IIA-D, Fig. 2). The reduction in channel amplitude was accompanied by a shift, in the negative direction, of the reversal potential. A similar reduction in channel amplitude, and shift in reversal potential, was observed when 20 mM KC1 replaced 150 mM KCI (Fig. 2, IV curve). All of these observations suggest that the CS channel is a non-selective cation channel. In this regard, CS channels appear to be similar to the non-selective SA channels reported in Xenopus oocytes ~7 and ventricular myocytes9. The CS channel IV relationship cannot be explained by the Goldman-Hodgkin-Katz current equation. There are significant deviations between the fitted curve and IV data, particularly at depolarized membrane potentials (note solid lines in IV curve, Fig. 2). The deviations are more extreme with 20 mM salt solutions in the pipette. The rectification in the IV data is therefore non-Goldman. Such rectification may be due to a voltage-dependent energy barrier, or perhaps a block

277 of outward current by an internal molecule. Inwardly rectifying SA channels have been reported 8't4'44'45. In principle, application of pressure may force ions through the channel. However, in common with other mechanotransducing ion channels 38, CS single-channel amplitude is independent of pressure (not shown). The dependence of the single-channel amplitude on membrane potential indicates that the electro-chemical gradient, and not the pressure gradient, provides the energy for ion flow through CS channels. Changes in pipette pressure specifically alter the distribution of closed and open states (see below). Increasing the pipette pressure may create a leakage pathway between the wall of the pipette and the patch. Such a leakage pathway may lead to channel activity. Because the channel appeared to be non-selective for cations, we studied the pressure-evoked currents under ionic conditions that would favor the movement of Na ÷ or K + from bath to pipette, or favor no movement at all. For the former case we studied the effect of pipette pressure using a pipette solution of 10 mM NaCI, 5 mM NaEGTA and 10 mM HEPES, and a bath solution of 140 mM NaCI, 5 mM KCI, 1 mM CaCI 2, 0.5 mM MgC12 and 10 mM HEPES. Under such conditions, at 0 mV pipette potential, movement of Na +, K + and Ca 2÷ from bath to pipette through the leakage pathway is strongly favored and should result in an outward current. Using these ionic conditions we observed instead a 0.3 pA inward single channel current (Fig. 2; IIC) (4 patches). Such inward currents would be expected if Na ÷ was moving from pipette to cytoplasm through the CS channel, or CI- was moving from bath to pipette through the leakage pathway. To investigate the latter possibility we used the bath solution (see above) in the pipette and studied CS channel activity at 0 mV pipette potential. Under these conditions net movement of ions through the pipette-to-bath pathway is not favored. We observed only inward currents at 0 mV pipette potential (see Fig. 2; IC, and see Figs. 7 and 8). Such currents are expected if cations are moving from pipette to cytoplasm. These results demonstrate that the pressure-evoked non-selective ion current does not flow through a pathway between the glass-membrane interface linking pipette to bath. In common with SA channels 8'39, CS channel activity (NPo), at constant pipette pressure, showed considerable non-stationarity (not shown). The non-stationarity in CS activation was characteristic of most of the patches with such channels. The non-stationarity indicates that we cannot provide a constant input to the gating mechanism. There are morphological changes in the patch structure that accompany non-stationarity (see below).

i

C

A

(D tm4 (/3

-1 4 mm Hg

(.o

(5 3 CO

~2 Z

H,Hill HnNIIH

00 1

. . . .

0

30

i

i

i'

1

B

~24 6~8

- 6 mm Hg

~12 Z O 6 o o

n0J]~ 0

o

6

n

. . . . . . . . . 12

18

24

30

TIME (SEC) Fig. 4. CS channel closed dwell-time histogram. Closed dwell-times are sensitive to changes in pressure. Membrane potential was - 8 6 mV. Same patch as in Fig. 3. Inset: examples of channel records at each pressure. Calibration markers represent 0.250 s and 2 pA. Bandwidth was 100 Hz and sampling rate was 500 Hz.

In four patches the data was sufficiently stationary to permit a determination of the effect of pressure on dwell-times spent in the closed (Fig. 3A, B) and open states (Fig. 3C). In each of the four patches both open and closed dwell-times could be fit by two exponentials. An example of the effect of increasing the pressure from - 1 4 to - 6 mmHg on the long closed dwell-times is shown in Fig. 4. Increasing the pressure reduced the closed time. Both the long- and short closed dwell-times were sensitive to pipette pressure (Fig. 5A). By contrast, the open states were largely insensitive to changes in pressure (Fig. 5B). We conclude that the CS channel is only sensitive to pressure while in the closed conformation. The CS channel is voltage-dependent (Fig. 6). The long- and short closed dwell-times, and the long open dwell-time increased with depolarization (Fig. 6A,B). By contrast, the short open dwell-time was voltage independent (Fig. 6B).

Imaging patches with CS channels The effects of pipette pressure on the structure and movement of the patch, as well as CS channel activity, are shown in Figs. 7 and 8. In each figure the dark arrow points towards the patch dome and the open

278 arrow points to the cytoplasm below the patch dome. At a pipette pressure of + 7 mmHg, there was little CS channel activity while this pressure was maintained (25 s). The patch dome remained curved towards the shank of the pipette (i.e. towards the right in Fig. 7A). Increasing the pressure to +13 mmHg reversed the direction of dome curvature towards the pipette tip (towards the left in Fig. 7B). Much CS channel activity was evident while the pressure was maintained (147 s). At least 4 channels were present and are represented in the amplitude histogram. Further increases in pipette pressure evoked more single channel events and maintained the direction of curvature of the patch (not shown). Increasing the suction to - 1 0 mmHg caused the dome o f the patch to curve again towards the shank of the pipette. The channel activity was reduced considerably while this pressure was maintained (118 s; Fig. 7C). Therefore, the effect of the pressure increase on the curvature of the patch and CS channel activity is reversible. This effect of pressure on the curvature of the patch, and concomitant effect on CS channel activity, could be elicited repeatedly (see below). No CS

4 2

f

E

A

O

A

.

,:~......

1"t73

t

C

-

C)

~J

2-3 --4

•

-6

I 0

,

,

i

,

,

I

,

,

i

,

,

a

,

,

,

,

,

,

f

._..ld2 -2~-2°J s..B

,

~

,

04

c

03 -3

•

O

-4

•

,

-

60

i

-145

,

I

-130

membrone

i

-115

i

-1

O0

-85

p o t e n t i o l (mY)

Fig. 6. Voltage-dependence, at constant pressure, of the closed (A) and open (B) state dwell-times. A: ©, long closed dwell-times; o, short closed dwell times. B: ©, long open dwell-times; o, short open dwell-times. Same patch as in Fig. 5.

o

t)

CI

~-2

E -4 0 -1 tD -6

C2

•

l

l

l

l

J

l

,

,

i

,

,

i

i

,

F

B

o o o

r3

o o

01 0

o

02

c-4

_~

i

-20

i

I

-1 6

i

i

i

i

18

i

i

i

-12

-4

pressure (ram Hg) Fig. 5. Both closed dwell-time constants change with pressure. Summary of the closed dwell-time constants (C1,C2), and open dwell-time constants ( 0 3 , 0 4 ) . A: ©, long closed dwell-times; o, short closed dwell-times. B: ©, long open dwell-times; o, short open dwell-times. The closed dwell-times (A) are not corrected for the number of channels in the patch. From saturation of the pressure vs. average current relationship we estimated there were a m i n i m u m of 66 CS channels in this patch. Cell-attached patch. M e m b r a n e potential was - 86 mV, patch 573281. Bandwidth was 100 Hz and sampling rate was 500 Hz.

channel activity was observed during the time ( > 300 s total time) the patch was curved towards the shank (towards the right of the figure). Substantial CS channel activity was observed only when the dome was curved towards the left of the figure. CS channel activity (NPo), but not its directional nature, appears to be strongly affected by changes in structure underneath the patch dome. Figure 8A shows the patch dome (dark arrow, Fig. 8) at an earlier time, soon after gigaseal formation. The patch dome curved towards the shank of the pipette (towards the right), with no CS channel activity ( - 1 8 mmHg, 13 s). Increasing the pressure to + 9 mmHg evoked a change in the radius of curvature and CS channel activity (Fig. 8B, 113 s). There were no multiple openings of CS channels (Fig. 8B). Further increases in pipette pressure to +30 mmHg evoked channel openings and drove the patch dome towards the pipette tip. When the pressure was changed to - 5 mmHg the patch dome traveled up the pipette tip and there was a concomitant loss of the plug of cytoplasm (compare patch contrast in Figs. 7 and 8). The disappearance of the plug suggests dissolution of cytoplasmic material. Multiple openings were common following the loss of cytoplasmic material (cf. Figs. 7 and 8). In addition the activation of CS channels became more sensitive to

279 changes in pipette pressure following the loss of cytoplasm (cf. Figs. 9A and 9B). These results suggest that NPo, but not directional sensitivity, is affected by loss of cytoplasmic constituents underneath the patch dome. In other patches such increases in NPo, after several cycles of positive and negative pressure, were commonly observed ( > 10 patches). These results demonstrate a pressure-induced alteration in CS channel properties. The basis of such a change may reside in the ability to deliver mechanical energy to the channel (the plug of cytoplasm may act to prevent inward deformation of the dome). However, the directional character of CS channel activation does not appear to rely on cytoplasmic constituents underlying the patch dome.

Stretch-activated channels (SA channels) In the cell-attached configuration, increasing the suction from 0 to - 7 0 mmHg increased the frequency of appearance of inward currents (Fig. 10). We found a large range of conductance levels for SA channels: 360, 230 (Fig. 10), 144 and 70 pS (Fig. 11). The effect of increasing the suction was reversible (not shown). The SA channels appeared in all possible combinations. Taking all the SA conductance levels together we have observed SA channels in 60% of more than 100 patches, giving an overall SA channel density of a b o u t 0.015 channels per/~m 2 (assuming a 5/zm diameter patch). We have observed both SA and CS channels in the same patch (Fig. 1). By contrast to the CS channel, SA channels apiiiiii~ii!i!i!iil:!il

Fig. 7. Effect of pressure on CS channel activity and curvature of the patch. A: + 7 mmHg. B: + 13 mmHg. C: - 10 mmHg. Inset: image of the patch. The dark and open arrows point towards the patch dome and underlying cytoplasm, respectively. Cell-attached patch. Nomarski optics. Magnification x 1,400. Upper panels: channel records observed concomitantly with the image of the patch. Channel records were sampled at 2,500 Hz with the bandwidth limited to 500 Hz. Vertical and horizontal scales represent 2.1 pA and 190 ms, respectively. The ordinate of the amplitude histogram represents the percentage of total time spent at that current level: the abscissa is the current (pA). Total amplitude histograms are plotted semi-logarithmically. Such a plot increases the display range in NPo so that channel openings can be seen in long records dominated by baseline current. Movement of the patch can be observed relative to the two black fiducial marks in the right hand side of all photos. All photos and channel records from same patch. In A, the diameter of the pipette at the dome of the patch is about 4/~m.

280 peared in excised patches. Visually separating SA channels from CS channels in the same patch was easy because SA channels have the larger open channel noise, larger conductance, faster kinetics and different gating properties (see below). While the open channel noise helped to separate CS channels from SA channels, assigning a value to SA channels conductance was problematic. Although we measured the largest settled event for each SA channel the uncertainties in conductance levels associated with such a m e a s u r e m e n t are considerable. For the 360 pS channel, conductance levels of settled events within a burst differed at most by 78 pS (Fig. 10). Errors in conductance estimates for the other SA channels were 52 pS (230 pS channel), 70 pS (144 pS channel) and 14 pS (70 pS channel). Nevertheless, we were able to distinguish four different conductance levels among SA channels. The open channel noise characteristics and range of conductance levels of SA channels are similar to the

Ca 2+ activated maxi-K -~ channels 25. To determine if Ca z+ was necessary to activate these channels we studied the effect of eliminating Ca 2+ from the bath (no added Ca 2+) and pipette (using EGTA). We were able to observe all conductance levels of SA channels (not shown). These results appear to rule out the possibility that external Ca 2+ is a necessary ligand. The IV relationship for all SA channels, with KCI in the pipette, is shown in Fig. 12. The cell-attached IV relationship is linear for negative m e m b r a n e potentials, and all SA channels share a common reversal potential ( + 20 mV). The single channel current is independent of applied pressure in the pipette (not shown). The common reversal potential shared by all four SA channels suggested a common selectivity. Therefore, we decided to determine the effects of known ion channel blockers on SA channel activity. Both cesium and barium, well-known blockers of K + channels, reduced both the frequency and open time of all SA

Fig. 8. Same patch as in Fig. 7 but at an earlier time in the record. A: 0 mmHg. B: +9 mmHg. C: +15 mmHg. Note the low channel activity (NPo) in B and C (cf. Fig. 7B) and the plug of cytoplasmic material underneath the dome (open arrow), which is absent later in the recording (Fig. 7, open arrow). Magnificaton × 1,400.

281

,o

A

1

< 0

O.

o o 0

0.1

0.01 -30

;o

-20

-10

0

0

10

20

,30

40

lo Hg

20

30

4o

[3

1

< 0.1

o.ol -30

-2o

-1o

o mm

Fig. 9. CS channel activation curves associated with Figs. 7 and 8. Ordinate: averaged current determined by integrating the total amplitude histogram at each pressure step. Abscissa: pipette pressure. A: average current when the plug of cytoplasm was under the patch dome (solid line drawn by eye). B: average current following dissolution of the cytoplasmic plug (solid line drawn by eye). In this patch more pressure steps were done than are shown in Figs. 7 and 8. This diagram is a s u m m a r y of the average current determined at these pressure steps. Records were sampled at 2,500 Hz and limited to a bandwidth of 500 Hz.

channels (> 10 patches). An example of cesium block of the 144 and 70 pS SA channels in an outside-out patch is shown in Fig. 13. Addition of 5 mM Cs ÷ reduced the frequency of appearance of both SA channels (Fig. 13). The minimum concentration necessary to observe the block of Cs ÷ or Ba 2+ was 0.1 and 0.01 mM, respectively (not shown). Increasing or decreasing the pressure to the elastic limits of the patch did not relieve the block by either Cs + or Ba 2+, suggesting that membrane tension alone was not affected by barium (or cesium). We have observed similar results for the 360 and 230 pS channels (> 10 patches). The blocking effects of Ba 2÷ and Cs + suggest that SA channels are K ÷ selective. In the cell-attached configuration we also compared the effects of NaCI- and KC1-HEPES solution on SA channels. With 150 mM KCI in the pipette we easily observed all SA channels over a large range of membrane potentials ( - 120 to + 80 mV). By contrast, with 150 mM NaC1, we only observed SA channels at membrane potentials more positive than 0 mV (not shown). These results show that Na + is impermeable, and that SA channels cannot be non-selective cation channels. By contrast we observed the CS channels in cell-attached patches with either KC1 or NaC1 in the pipette

360 pS

230 pS

F-t+ 4 emJ~j

"•

-

--

-- :~

-

*" ~

'

*'l + + ' -

--

~- -T'--

4 oedee

e ml~

:1

OOm

:1

M m

Fig. 10. Stretch-activated channels, 360 and 230 pS, as a function of pressure. Cell-attached patches. Pipette contained 150 m M KC1, 20 m M HEPES, p H 7.4. M e m b r a n e potential was - 100 mV. Each conductance class was from a different patch.

282

144

pS

70 pS ....

:

: -

-

0

emile

..... r

3 cmHi

6 emil|

•

omHi

40 meee

40

ml,~

Fig. 11. Stretch-activated channels, 144 pS and 70 p5, as a function of pressure. Cell-attached patches. Pipette contained 150 mM KCL, 20 mM HEPES, pH 7.4. Membrane potential was - 60 mV. Each conductance class was from a different patch.

and over a large range of membrane potentials ( - 170 to +20 mV). We analyzed the dwell times for one patch containing the 70 pS SA channel. Figure 14 shows that the closed times can be described by three exponentials, and that the long closed dwell-time (C3) is most affected by suction.

Gating characteristics of CS and SA channels Figure 15 shows the effect of pipette pressure on the average current carried by CS channels (Fig. 15A) and

.(

O_

.i, £LUl, iJl IJJ, IJ

/ ~/]2 ,gJ

u

i

70 pS

L; c

SA channels (360, 144 and 70 pS; Fig. 15B). CS channel activity (average current) remained low at suctions between - 3 0 and 0 mmHg, but activity increased with positive pressures and reached saturation (Figs. 15A and 10). These observations demonstrate that CS channel NPo is not a symmetric function of pressure. By contrast, SA channel activity generally increased for either negative or positive pressures (70 and 144 pS; Fig. 15B). (In other patches, the 360 pS SA channel activity increased with positive pressures.) SA channel activity for the 70, 144 and 360 pS SA channels is a symmetric function of pressure. These results suggest that the gating ~a~chanisms of SA and CS channels are

£4 pS

50

230 p~

X~ -40q

-1

Z~ -5o ° c I ~ -601 -2@@

o/

56o

q

pS

i

- 150

"/00

membrone

00

50

potential

(mV)

Fig. 12. Single channel current-voltage relationship for all SA channels: data from different cell-attached patches. Pipette contained 150 mM KCI, 20 mM HEPES, pH 7.4. Individual points are amplitudes determined by fitting Gaussians to open and closed events. Lines through the points were determined by linear regression. The listed numerical values of the conductances were determined from amplitudes of individual settled events. Such amplitudes within bursts are different (see text). Due to the large amount of open channel noise the conductance values shown are approximately 10-20% higher than those conductances determined from the slopes of the fitted lines. (If there were no open channel noise the conductances determined from the fitted Gaussians and the conductances determined from amplitudes of settled events would be identical.)

A d d 5 mid 0 5 ÷

pA 5 sec

10

Fig. 13. Effect of Cs + on the 144 and 70 pS SA channels. Outside-out patch. Membrane potential was - 6 0 mV. Bath contained 150 mM KCI, 20 HEPES, pH 7.4. Pressure was - 30 mmHg.

283 DISCUSSION o

-2

03

J

tP

4

(D

C2

-6

C1

-8

- 1 0

i

-70

-56

i

I

,

i

i

-42

i

,

-28

pressure

I

,

(ram

J

D

- 1 4

Hg)

Fig. 14. Effect of pressure on closed (C1,C2,C3) and open (O) dwell-times. The long closed dwell-time is the most sensitive to pressure. 70 pS SA channel.

fundamentally different. The monophasic activation of the CS channel with pressure is consistent with the conclusion that CS channels sense the direction of curvature of the dome (see Figs. 7 and 8). By contrast, the symmetric gating characteristics of the SA channel suggest that these channels sense membrane tension independently of the sign of the curvature of the patch. Over a period of about 25 min the activity of the 144 pS SA channel, at 0 mmHg, increased irreversibly. Changing the pressure was without effect on the average current (not shown). By contrast, over the same time, the activity of the 70 and 360 pS channels remained low at 0 mmHg. The results shown in Fig. 15B are from the same patch and they suggest that the 144 pS SA channel is gated independently of the other two SA channels.

100.000-

A

The most striking aspect of the CS channel is the lack of symmetry in its activation. All other known stretch-sensitive channels appear to be sensitive to activation with either positive or negative pressures. The symmetry in activation of other stretch channels suggests that membrane tension controls gating. We have tried extreme suction to turn the CS channel on, as would be expected from a stretch-activated channel 17, and we have tried extreme positive pressures to turn the channel off, as would be expected for a stretch-inactivated channel 32. Neither stimulus was successful. It is possible that high positive pressures may turn it off (or high negative suction may turn it on), but loss of the seal occurs before we can show such changes in activity. To date all SA (and SI channels) show activity well within the range of pipette pressures used in our experiments. CS channels in astrocytes cannot be attributed to the pipette per se, as we have not observed such channels in either Xenopus oocytes or astrocytes derived from hippocampal lesions, using identical pipettes. Such oocytes 44'45 and astrocytes (Romanowski and Bowman, unpublished observations) have an abundance of SA channels. In principle the CS channel could be related to the seal breakdown phenomenon. Seal breakdown might be expected to occur when pressure is applied to the pipette, leading to non-selective channel-like events. However, a seal breakdown model does not easily predict the cationic selectivity (i.e. the exclusion of

1.000-

B

1 4 4 pS 7 0 pS 10.000

0.100-

1.000

0.010-

•

Z~

•l

o_ @J > 0

-

0. t 0 0

'

-80

'

'

I

'

-40 pressure

0

' ' ' ~ ' ' ' , ' ' • I 0.001 ' i ' ~ ' 40 80 120 -80-60-40-20 (mm

Hg)

~ .

pressure

. 0

o

n

. . . 2'0 4'0

(mm

360

pS

6~0 8~0

Hg)

Fig. 15. Gating characteristics of CS and SA channels. A: pressure-activated channel. Channel data filtered at 100 Hz and sampled at 500 Hz. Current was averaged by integrating the total amplitude histograms at each pressure. 13, patch 592353; zx, patch 591035. B: Stretch-activated channel. 13,360 pS SA channel; zx, 70 pS SA channel; o, 144 pS SA channel. 20 rain after the beginning of the patch, the 144 pS SA channel lost all sensitivity to changes in pipette pressure. All SA channels were in the same patch. Current was averaged by integrating the total amplitude histograms at each pressure. Data were filtered at 1,000 Hz and sampled at 5,000 Hz.

284 anions), the rectifying IV relationship of the amplitudes of single channels when pipette and bath NaCI concentrations are identical, CS channel activity when the electrochemical gradients for Na + and C1- across the leakage pathway are both zero, the voltage dependency of the dwell-times between openings, the selective disappearance of CS channel activity upon excision of the patch into Ringer's solution (with maintenance of the gigaseal and SA channel activity), the uniform amplitude of open events, or the dependency of CS channel appearance on conditions used to dissociate the neocortex (see Materials and Methods). Instead, all of these observations are commonly associated with ion channel proteins in biological membranes. These observations suggest that CS channel events are due to such proteins. In our initial studies the CS channel appeared to be identical to SI channels studied previously 32, and were so named in our preliminary reports u'12. On the basis of our studies of the relationship between pressure and activation and our imaging data we now call this channel the curvature-sensitive (CS) channel. It has been proposed that SA channels gather gating energy from membrane tension, which is controlled by either the cytoskeleton 17'42 or the lipid bilayer zS. In both models tension in the lipid bilayer or tension in the cytoskeletal elements beneath, and in parallel with, the plasmalemma may be calculated by Laplace's law, T = PD/4

where T is the tension, P is the pressure difference, and D is the diameter of curvature of the patch. Such a relationship predicts increases in tension for suction and pressure, for similar diameters, without regard for the direction of curvature. In the present study light microscopic examination of the patch dome showed that the activation of the CS channels correlates with a specific direction of curvature of the dome. With bending of the patch by positive pressure the CS channel protein, which spans the bilayer, may be squeezed in the external half of the bilayer and expanded on the internal half of the bilayer, opening the pore. The source of the gating energy of CS channels would be derived from lipids, in line with the proposals of Martinac and colleagues 28. Although gating energy might come from changes in bending energy it is more likely that the channels are activated by tension, but that proper curvature is necessary for the channel to open. With positive pressure the probability of being open is a steep function of pressure (Figs. 9 and 15A), but the radius of curvature changes by a small amount (Figs. 7 and 8). It is most likely that tension activated the

channel but that channel opening is inhibited when the membrane is curved towards the pipette shank, as if a gate sensing local geometrical configuration was in series with a gate sensing membrane tension. The channel can respond to membrane tension only when patch geometry is the permissive configuration. Physiologically, CS channels, located in a region of membrane in the permissive curvature, could detect the partitioning of hydrophobic compounds into the plasma membrane, or could detect when astrocytic processes are in specific geometrical configurations (such as surrounding synapses). Because pressure changes also affected the location of the patch in the pipette another model may be envisioned if CS channel gating depends on the direction of movement (i.e. velocity) of the patch structure, but this seems unlikely. It is difficult to envision a model coupling the kinetic energy of patch movement with the gating energy for channel opening. The mechanosensing channels we observe might be some form of ligand-gated channel. In such a model the ligand would come from within the astrocyte when the pipette pressure was changed. In fact the release of substances, some of which depolarize the membrane potential, is enhanced by hypotonicity, which would increase membrane tension. The released compounds include glutamate, aspartate, and taurine 24'26'2v'35. However, we have tested for the ability of glutamate and taurine to activate channels in cell-attached patches and found no effect. The finding of an asymmetrically gated mechanotransducing ion channel suggests a re-classification of mechanotransducing ion channels based on their gating properties. The SA (SI) channels, with symmetric activation (inactivation) curves, may be classified as scaler mechanosensors. These channels would transduce increases in cytoskeletal or bilayer tension independent of the direction of curvature (or movement) of the patch dome. By contrast the CS channels could be called vectorial mechanosensors. Such channels would be sensitive to the sign of the curvature of the patch. We predict that other vectorial mechanosensors will be found. The two time constants for the closed and open times of CS channels would appear to suggest a kinetic model with two closed and two open states. However, all patches contained multiple CS channels so that we were unable to determine the number of states of a single CS channel. The three-state kinetic model for the SA channels appears to be different from the model proposed for SA channels in Xenopus oocytes 17, but an analysis at a higher bandwidth would be necessary before such a conclusion could be drawn.

285 The CS channel closed dwell-times decrease with depolarization, at constant pressure. This result suggests the depolarization by K + and neurotransmitters, such as glutamate and norepinephrine, or GABA43 could increase the likelihood that the CS channel would open for an otherwise small change in membrane tension. Therefore, the ability to sense stresses applied to the membrane may be regulated by neurotransmitters that affect the membrane potential. In addition, since low levels of glutamate swell astrocytes43, glutamate may regulate SA/CS channels indirectly by increases in membrane tension. All these mechanical inputs into membrane tension may activate SA channels and cause a re-distribution of spatial K ÷ currents 34. Morris and Horn 31 have found that a lack of stress induced whole cell currents in snail neurons. Such currents (100's of pA) were expected because these cells have numerous SA and SI K ÷ channels, as judged from cell attached patches. Stresses were applied osmotically or mechanically, and little or no mechanosensory whole-cell current was recorded. These results raised the possibility that mechanosensory ion channels may be abnormally sensitive to stress in patch recordings. The increased sensitivity may be related to tearing of cytoskeletal elements. Such elements, in their normal state, would keep the channels from being mechanosensitive 31. Our patch imaging showed that CS channels can be activated with and without normal-appearing cytoplasm underlying the dome (open arrows, Figs. 7 and 8). Following the morphological change, there was an increase in channel activity (NPo) (cf. Figs. 7 and 8). These observations support Morris and Horn's supposition of an increased mechanosensitivity in cellattached patch recordings. Normally, cytoplasmic elements beneath the dome may dominate the compliance and act to prevent mechanical energy from being delivered to the channel. The dissolution of the cytoplasm would allow more stress to reach the channels. However, becaUse CS channels may be activated with normal appearing cytoplasm (open arrow, Fig. 8), the issue of appropriate physiological response is probably a matter of degree of activation, not one of all-or-none. As we do not know how the cell uses the information (ion flux) associated with channel opening, we cannot speculate about physiological levels of channel activity. The increase in channel activity (NPo) after the morphological change may also suggest that CS channel activity is regulated in the cell by local changes in cytoplasmic viscosity. The change in cytoplasmic morphology was without effect on the directional gating properties of CS channels (Fig. 9). Kimelberg and O'Connor 21 found that the astrocyte membrane potential depolarizes in response to hypo-

tonicity. The amount of depolarization scaled with the degree of hypotonicity and correlated in time with astrocyte swelling. In contrast to snail neurons 3x substantial osmotically evoked whole-cell currents have been observed in astrocytes23. For such cells under whole-cell voltage-clamp, hypotonic solutions always caused a conductance increase, sometimes preceded by a conductance decrease. The latter was attributed to the closing of K ÷ channels, while the conductance increase was attributed to openings of non-specific cation channels. By itself, the finding of both K ÷ and non-selective cation mechanosensory currents is in agreement with the studies reported here. However, if tension increases in the plasmalemma during swelling, we would have predicted SA channels to turn on, causing a conductance increase (selective for K ÷), while CS channels remained quiescent. On the basis of this there appears to be disagreement between the wholecell experiments and predictions from cell-attached studies. Such discrepancies may suggest the presence of other mechanosensory channels in the processes which dominate the whole-cell currents, or the distention of the plasmalemma is more complicated (perhaps local bebbing acts to turn on CS channels at the interface between the bleb and normal plasmalemma). These contradictions may be resolved when specific antagonists for mechanosensory ion channels are found. Acknowledgments. We thank Drs. A. Auerbach, H.K. Kimelberg, B. Pickard and X.C. Yang for critical discussions, and C.M. Severin for photographic help. We thank Robert Borschel for superb machining, and Ann Marie Bowman for maintaining the cell cultures. This work was supported by Grant R23 NS 24891 from NIH to C.L.B, and Grants NIH DK 37792, USARO 22560-LS, and the Muscular Dystrophy Association to F.S.

REFERENCES 1 Bignami, A., Dahi, D. and Rueger, D.C., Glial fibriUary acidic protein (GFA) in normal neural cells and in pathological conditions. In S. Fedoroff and L. Hertz (Eds.), Advances in Cellular Neurobiology, Academic Press, New York, 1980, pp. 285-310. 2 Bowman, C.L. and Baglioni, A., Application of the GoldmanHodgkin-Katz current equation to membrane current-voltage data, Z Theoret. Biol., 108 (1984) 1-29. 3 Bowman, C.L. and Kimelberg, H.K., Excitatory amino acids directly depolarize rat brain astrocytes in primary culture, Nature, 311 (1984) 656-659. 4 Bowman, C.L., Kimelberg, H.K., Frangakis, M.V., BerwaldNetter, Y. and Edwards, C., Astrocytes in primary culture have chemically activated sodium channels, J. Neurosci., 4 (1984) 1527-1534. 5 Bowman, C.L. and Kimelberg, H.K., Adrenergic receptor-mediated depolarization of astrocytes. In H.K. Kimelberg (Ed.), Glial Cell Receptors, Raven Press, New York, 1988, pp. 53-76. 6 Charles, A.C., Merrill, J.E., Dirksen, E.R. and Sanderson, M.J., Intercellular signaling in glial cells: calcium waves and oscillations in response to mechanical stimulation and glutamate, Neuron, 6 (1991) 983-992. 7 Christensen, O., Mediation of cell volume regulation by Ca 2+ influx through stretch-activated channels, Nature, 330 (1987) 6668.

286 8 Cooper, K.E., Tang, J.L., Rae, J.L. and Eisenberg, R.S., A cation channel in frog lens epithelia responsive to pressure and calcium, .L Membr. Biol., 93 (1986) 259-269. 9 Craelius, W.V., Che, P. and EI-Sherf, N., Stretch-activated ion channels in ventricular myocytes, Biosci. Rep., 8 (1988) 407-414. 10 Ding, J.-P., Yang, X.-C., Bowman, C.L and Sachs, F., A stretchactivated ion channel in astrocytes in primary culture, Soc. Neurosci. Abst., 14 (1988) 1056. 11 Ding, J.-P., Bowman, C.L., Sokabe, M. and Sachs, F., Mechanical transduction in glial cells: SICS and SACS, Biophys. J., 55 (1989) 244a. 12 Ding, J.-P., Bowman, C.L. and Sachs, F., Mechanotransduction in astrocytes in primary cell culture, OWNYICIG Meeting (1989), SUNY-Buffalo, Buffalo, NY, January 14, 1989, p. 6. 13 Eng, LF., The glial fibrillary acidic (GFA) protein. In R. Bradshaw and D. Schneider (Eds.), Proteins of the Nervous System, Raven Press, NY, 1980, pp. 85-117. 14 Falke, L.C., Edwards, K.L., Pickard, B.G. and Misler, S., A stretch-activated anion channel in tobacco protoplasts, FEBS Lett., 237 (1988) 141-144. 15 Fenwick, E.M., Marty, A. and Neher, E., A patch-clamp study of bovine chromaffin cells and of their sensitivity to acetycholine, J. Physiol., 331 (1982) 577-597. 16 Frangakis, M.V. and Kimelberg, H.K., Dissociation of neonatal rat brain by dispase for preparation of primary astrocyte cultures, Neurochem. Res., 9 (1984) 1689-1698. 17 Guharay, F. and Sachs, F. Stretch-activated single ion channel currents in tissue-cultured embryonic chick skeletal muscle, J. Physiol., 352 (1984) 685-701. 18 Hamill, O.P., Marty, A., Neher, E., Sakmann, B. and Sigworth, F., Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches, Pfluger's Arch., 391 (1981) 85-100. 19 Kimelberg, H.K., Active accumulation and exchange transport of chloride in astroglial cells in culture, Biochim. Biophys. Acta, 656 (1981) 179-184. 20 Kimelberg, H.K., Primary astrocyte cultures: a key to astrocyte function, Cell. Mol. Neurobiol., 3 (1983) 1-16. 21 Kimelberg, H.K. and O'Connor, E.R., Swelling-induced depolarization of astrocyte potentials, Glia, 1 (1988) 219-224. 22 Kimelberg, H.K. and Kettenman, H., Swelling-induced changes in electrophysiological properties of cultured astrocytes and oligodendrocytes. I. Effects on membrane potentials, input impedance and cell-cell coupling, Brain Res., 529 (1990) 255-261. 23 Kimelberg, H.K., Anderson, E. and Kettenmann, H., Swelling-induced changes in electrophysiological properties of cultured astrocytes and oligodendrocytes. II. Whole cell currents, Brain Res., 529 (1990) 262-268. 24 Kimelberg, H.K., Goderie, S.K., Higman, S., Pang, S. and Waniewski, R.A., Swelling-induced release of glutamate, aspartate and taurine from astrocyte cultures, J. Neurosci., 10 (1990) 1583-1591. 25 Latorre, R., The large calcium-activated potassium channel. In C. Miller (Ed.), Ion Channel Reconstitution, Plenum Press, New York, 1986, pp. 431-463. 26 Martin, D.L., Madelian, V. and Shain, W., Osmotic sensitivity of isoproterenol- and high [K +]o-stimulated taurine release by cul-

27

28

29

30

31

32

33 34

35 36

37 38 39 40 41

42

43 44

45

tured astroglia. In H. Pasantes-Morales, D.L. Martin, W. Shain. and R. Martin (Eds.), Functional Neurochemistry of Taurine, Liss, New York 1989, in press. Martin, D.L., Madelian, V., Seligmann, B. and Shain, W., The role of osmotic pressure and membrane potential in K+-stimu lated taurine release from astroglia, J. Neurosci., 1/) (1990) 571577. Martinac, B., Adler, J. and Kung, C., Mechanosensitive ion channels of E. coli activated by amphipaths, Nature, 348 (1990) 261-263. Medina, I.R. and Bregestovshi, P.D., Stretch-activated ion channels modulate the resting membrane potential during early embryogenesis, Proc. R. Soc. Lond. B, 235 (1988) 95-102. Merrill, C.J., Dirksen, E. and Sanderson. M., Elevation and oscillations of [Ca 2+ ]i and intracellular communication in glial cells in response to trauma and glutamate, Soc. Neurosci. Abst., 16 (1990) 190. Morris, C.E. and Horn, R., Failure to elicit neuronal macroscopic mechanosensitive currents anticipated by single channel studies, Science 251 (1991) 1246-1249. Morris, C.E. and Sigurdson, W.J., Stretch-inactivated ion channels coexist with stretch-activated ion channels, Science, 243 (1989) 807-809. Morris, C.E., Mechanosensitive ion channels, J. Memb. BioL, 113 (1990) 93-107. Orkand, R.K., Nicbolls, J.G. and Kuffier, S.W., The effect of nerve impulses on the membrane potential of glial cells in the central nervous system of amphibia, J. Neurophysiol., 29 (1966) 788-806. Pasantes-Morales, H. and Schousboe, A., Volume regulation in astrocytes: a role for taurine as an osmoeffector, J. NeuroscL Res., 20 (1988) 505-509. Pasantes-Morales, H., Moran, J. and Schousboe, A., Volume-sensitive release of taurine from cultured astrocytes properties and mechanism, Glia, 3 (1990) 427-432. Puro, D.G., Stretch-activated channels in human retinal Muller cells, Glia, 4 (1991) 456-460. Sachs, F., Mechanical transduction in biological systems, CRC Crit. Rev. Biomed. Eng., 16 (1988) 141-169. Sackin, H., Stretch-activated potassium channels in renal proximal tubule, Am. J. Physiol., 253 (1987) F1253-F1262. Sackin, H., A stretch-activated K ÷ channel sensitive to cell volume, Proc. Natl. Acad. Sci. USA, 86 (1989) 1731-1735. Sokabe, M. and Sachs, F., The structure and dynamics of patchclamped membranes: a study using differential interference light microscopy, J. Cell Biol., 111 (1990) 599-606. Sokabe, M., Sachs, F. and Jing, Z., Quantitative video microscopy of patch-clamped membranes' stress, strain, capacitance, and stretch-channel activation, Biophys. J., 59 (1991) 722-728. Walz, W., Role of glial cells in the regulation of the brain microenvironment, Prog. Neurobiol., 33 (1989) 309-333. Yang, X.C. and Sachs, F., Block of stretch-activated ion channels in Xenopus oocytes by gadolinium and calcium ions, Science, 243 (1989) 1068-1071. Yang, X.C. and Sachs, F., Characterization of stretch-activated ion channels in Xenopus oocytes, J. Physiol., 431 (1990) 103-122.