Arachidonic Acid Activates an OpenRectifier Potassium Channel in Cultured RatCortical Astrocytes

Stefano Ferroni,1* Pierluigi Valente,1 Marco Caprini,1 Mario Nobile,2 Peter Schubert,3

and Carmela Rapisarda1

1Department of Human and General Physiology, University of Bologna, Bologna, Italy2Institute of Biophysics, CNR, Genoa, Italy3Department of Neuromorphology, Max Planck Institute of Neurobiology, Martinsried, Germany

A pathophysiological increase in free arachidonic acid(AA) is thought to regulate the channel-mediated astro-cytic swelling occurring in several brain injuries. We re-port that in cultured rat type-1 cortical astrocytes, expo-sure to 10 �M AA activates an open rectifier K� channel,which exhibits many similarities with TREK/TRAAK mem-bers of the two-pore-domain K� channel family KCNK.Patch-clamp experiments showed that the current devel-oped with a long latency and was preceded by a depres-sion of the previously described outward rectifier K�

conductance. Pharmacologic studies indicate that theK� open rectifier was differentially sensitive to classicK�-channel blockers (quinine, quinidine, tetraethylam-monium, and barium) and was inhibited potently by gad-olinium ions. The activation of this K� current occurredindependently of the AA metabolism as pharmacologicinhibition of the lipoxygenase, cyclooxygenase, and cy-tochrome P450 epoxygenase signaling cascades did notalter the AA effect. Moreover, neither the neutralization ofthe NADPH-oxidase pathway nor scavenging intracellu-lar free radicals modified the AA response. Finally, theAA-induced K� current was unaffected by protein kinaseC inhibitors. The activation mechanism of the K� openrectifier was through an extracellular interaction of AAwith the plasma membrane. RT-PCR analysis revealedthat the AA-induced K� conductance was mediatedlikely by TREK-2 channels. Collectively, the results dem-onstrate that in cultured cortical astrocytes, pathologicallevels of AA directly activate an open rectifier K� channel,which may play a role in the control of K� homeostasisunder pathophysiological conditions.© 2003 Wiley-Liss, Inc.

Key words: astroglia; polyunsaturated fatty acid; KCNKchannel; ischemia; potassium homeostasis

Arachidonic acid (AA) is a cis-polyunsaturated fattyacid present in the plasma membrane and is linked co-valently to other molecules to form phospholipids. Inresponse to stimulation of metabotropic receptors, AA canbe released by activation of cellular phospholipases and can

act as biochemical messenger, directly or through its me-tabolites, to regulate the activity of various cellular sub-strates (Katsuki and Okuda, 1995). In the brain, the phys-iological level of free AA is lower than 0.01 mmol/kg, butunder pathophysiological conditions such as ischemia, ac-tivation of phospholipases A2 and C promotes the accu-mulation of AA to up to 0.5 mmol/kg (Rehncrona et al.,1982). An increase in AA was shown to induce brainedema both in situ and in vitro (Chan and Fishman, 1978;Unterberg et al., 1987), an effect that was postulated to bedue to astrocytic swelling triggered by an augmentedproduction of eicosanoids (e.g., prostaglandins and leuko-trienes) and superoxide radicals (Chan et al., 1988; Staubet al., 1994). Limited information is available, however,concerning the ionic mechanisms underlying the AA-mediated astrocytic swelling. This is noteworthy becauseion channels are important targets for free AA in variouscell types (Meves, 1994).

The regulation of ion channel activity by AA occursthrough both indirect and direct mechanisms. AA metab-olites (cyclooxygenase and lipoxygenase products), AA-produced oxygen free radicals, and protein kinase C areimportant mediators of the indirect mechanisms. In thenervous system, they have been demonstrated to mediateAA modulation of some neuronal channels (Piomelli et al.,1987; Keyser and Alger, 1990). Direct effects imply thatAA exerts its modulatory action by interacting with thechannel or with the phospholipid environment (Ordwayet al., 1991). In non-neuronal brain cells, such a directaction of AA has been demonstrated in oligodendrocytes,in which AA negatively modulated both outward- andinward-rectifying potassium (K�) channels (Soliven and

Contract grant sponsor: MIUR, Italy.

*Correspondence to: Stefano Ferroni, Department of Human and GeneralPhysiology, University of Bologna, Via San Donato 19/2, 40127 Bologna,Italy. E-mail: ferroni@biocfarm.unibo.it

Received 7 October 2002; Revised 18 December 2002; Accepted 30December 2002

Journal of Neuroscience Research 72:363–372 (2003)

© 2003 Wiley-Liss, Inc.

Wang, 1995). In activated microglial cells, a direct mech-anism was shown to underlie the AA-mediated inhibitionof the outward rectifier K� current (Visentin and Levi,1998) whereas in retinal glial cells, AA was reported todirectly depress the electrogenic uptake of glutamate (Bar-bour et al., 1989).

Because of the importance of K� channels in cellvolume regulation in astroglial cells (Olson and Li, 1997),we have addressed the question whether pathophysiolog-ical concentrations of AA were able to modulate K�

channel activity in cultured rat cortical astrocytes. Weshow that AA causes an initial partial inhibition of thedelayed rectifier K� current, which is followed by a large,sustained increase of the whole-cell K� conductance me-diated by the activation of an open rectifier K� channel.This K� channel displays functional and pharmacologicproperties similar to those of TREK/TRAAK members ofthe two-pore-domain K� channels family (KCNK)(Goldstein et al., 2001). Finally, we demonstrate the pres-ence of TREK-2 transcript in cultured cortical astrocytes.A preliminary account of this work has appeared else-where (Ferroni et al., 2001).

MATERIALS AND METHODS

Cell Culture

Primary cultures of pure cortical rat astrocytes were pre-pared as described previously (Ferroni et al., 1995). Briefly,cerebral cortices devoid of meninges were triturated and placedin cell culture flasks containing DME-glutamax medium(Gibco-Invitrogen, Carlsbad, CA) with 15% fetal calf serum(FCS) and penicillin/streptomycin (100 U/ml and 100 �g/ml,respectively; Gibco-Invitrogen). Culture flasks were maintainedin a humidified incubator with 5% CO2 for 2–5 weeks. Atconfluence, astroglial cells were dispersed enzymatically(trypsin-EDTA) in petri dishes (33-mm diameter) at a density of1–2 � 104 per dish. Electrophysiological experiments werecarried out 3–7 days later. Immunostaining for glial fibrillaryacidic protein (GFAP) and the flat, polygonal morphologicalphenotype of the cultured cells indicated that more than 95%were type-1 cortical astrocytes (Ferroni et al., 1995).

Electrophysiological Recording

Current recordings were obtained with the whole-cellconfiguration of the patch-clamp technique (Hamill et al.,1981). Patch pipettes were prepared from thin-walled borosili-cate glass capillaries to have a tip resistance of 2–4 M� whenfilled with the standard internal solution. Membrane currentswere amplified (EPC-7, List-Electronic, Darmstadt, Germany),filtered at 2 kHz and stored on a microcomputer for off-lineanalysis (pClamp 5.5.1, Axon Instruments, Foster City, CA, andOrigin, MicroCal, Northampton, MA). Because of the largeamplitude of the currents measured, series resistance (�10 M�)was corrected for to 60–80%. Experiments were carried out atroom temperature (20–24°C).

Solutions and Chemicals

Salts and other chemicals were of the highest purity grade(Sigma, St. Louis, MO). For the electrophysiological experi-

ments, the standard bath saline was (mM): 140 NaCl, 4 KCl,2 MgCl2, 2 CaCl2, 10 HEPES, 5 glucose, pH 7.4 with NaOHand osmolarity adjusted to �315 mOsm with mannitol. Theintracellular (pipette) solution was composed of (mM): 144 KCl,2 MgCl2, 5 EGTA, 10 HEPES, pH 7.2 with KOH and osmo-larity �300 mOsm. When using external solutions with differ-ent ionic composition, salts were replaced equimolarly. Thedifferent saline containing the pharmacologic agents were ap-plied with a gravity-driven, local perfusion system at a flow rateof �200 �l/min positioned within �100 �m of the recordedcell. Aliquots (10 mM) of AA were prepared in 70% ethanol andkept at �80°C for no longer than 1 month. On the day ofexperiments, perfusing solutions containing AA were preparedby diluting AA in the appropriate saline and sonicating for 5 minbefore use. Ethanol concentration in the perfusing solution was�0.1% and control experiments with ethanol were negative.The putative channel blockers were administered by co-application with AA except gadolinium (Gd3�), which tendedto form small, insoluble aggregates when applied in conjunctionwith AA. For this agent the effect was evaluated by comparingthe rate of AA washout with that relative to the current inhi-bition upon Gd3� exposure. For experiments aiming at inves-tigating the role of AA metabolic byproducts in the observedeffects, nordihydroguaiaretic acid, indomethacin, diphenyleneiodonium, and 17-octadecynoic acid (all from Sigma) werediluted into the extracellular bathing saline 30 min before theexperiments and added into the recording pipette. This para-digm was also used for the experiments in which the proteinkinase C inhibitors were utilized (H7 and Ro38-8220; Sigma).Because superoxide dismutase (SOD; Sigma) was appliedthrough the pipette only, measurements were started 15–20 minafter accessing the cells to permit SOD to dialyze completely theastrocyte cytoplasm (Pusch and Neher, 1988).

RT-PCR Analysis

Total RNA was extracted from rat brain or astroglial cellcultures with TRI-reagent (Sigma) following the manufacturer’sinstructions. First strand cDNA was synthesized from total RNAusing M-MLV Reverse Transcriptase (Gibco-Invitrogen) accord-ing to the manufacturer’s instructions. First strand cDNA was usedas template for PCR amplification (Taq DNA Polymerase; Roche)using TRAAK specific primers (5�-CCGAGACCAGTT-TCTGAG-3� and 5�-CTACAGTGGTGAGAGTCAC-3�) (Finket al., 1998) to amplify a fragment of 517 base pairs (bp), TREK-1specific primers (5�- GAACTCATCCAGCAAATAGTGGC-3�and 5�-TCTTCCACTTTGGCAATTCC-3�) (Meadows et al.,2000) to amplify a 281-bp fragment, TREK-2 specific primers(5�-GGCTAATGTCACTGCTGAGTTCC-3� and 5�-AAGC-CACACTTTAGTCCAGCTCC-3�) (Bang et al., 2000) to am-plify a 622-bp fragment, and using -actin as a control (Clontech)by amplifying a 330-bp fragment. Moreover, as positive internalcontrol, two primers for the mouse Kv1.1 channel (5�-CC-GCATCGACAACACCACAG-3� and 5�-CTCCTGCTCAGC-TATCTCCGTGCCCAGGG-3�) were used to amplify a frag-ment of 223 bp.

The reaction was carried out in a final volume of 50 �lcontaining 1� Taq Polymerase PCR buffer, 1.5 mM MgCl2,0.5 �M primers, 0.2 mM dNTPs, and 2.5 U Taq Polymerase(5 U/�l). The following PCR cycles were then carried out:

364 Ferroni et al.

94°C for 5 min, one cycle (94°C , 45 sec; 45°C for1 min; rampto 72°C for 2 min) followed by 35 cycles (94°C, 45 sec; 55°C,for 1 min; ramp to 68°C for 2 min) and finally 8 min at 72°C.Assays for K� channels in which the expected products were notdetected initially were also carried out under different PCRconditions. On a 1.5% agarose gel, 10 �l of the amplificationreaction was run in parallel with a known molecular weightmarker (Roche) and stained with ethidium bromide.

Statistics

Data are expressed as mean standard deviation (SD) ofseveral cells (n) in the different conditions. The statistical eval-uation was carried out with two-tailed Student’s t-test orANOVA followed by Bonferroni’s posthoc test as appropriate.A P-value � 0.05 was taken as statistically significant.

RESULTS

Dual Effects of Arachidonic Acid on AstrocyteMembrane Conductance

In agreement with our previous observations (Fer-roni et al., 1995), in cultured cortical astrocytes a slowlydepolarizing voltage ramp from a holding potential (Vh) of�60 mV elicited only outwardly rectifying K� currents(Fig. 1A). Ramp currents activated at potentials positive to

�40 mV and were partially (�70%) inhibited by 10 mMTEA (data not shown, n � 14). Exposure to 10 �M AAresulted in dual, sequential effects (n � 300): a reductionof the K� current followed by a robust and sustaineddevelopment of an outwardly rectifying conductance (Fig.1A). The inhibitory action of AA varied greatly from cellto cell, with maximal effects ranging from 10–70% (41 11%, mean SD, n � 112). The effect was rapid, withonset within 10 sec and maximal blockade after exposureof 30–60 sec (Fig. 1B). By contrast, the increase in ramp-evoked outward current occurred with a delay (1–3 min)after starting the AA application and developed slowly,needing 3–7 min after onset to reach the maximum value(Fig. 1B). Moreover, whereas the current inhibition didnot affect the astrocyte resting membrane potential (con-trol: �33 7 mV, n � 79; AA: �34 8 mV; n � 65,P � 0.05) the AA-induced current increase was accom-panied by a negative shift in zero-current potential(�70 5 mV; n � 86, P � 0.001). Upon AA washout, thecurrent slowly (�2 min) returned to a steady level, whichwas always smaller than the control, even after prolongedwashing (�10 min), thereby indicating that whereas theincrease in outward conductance was fully reversible, theAA-mediated current blockade was largely irreversible.

Fig. 1. Dual modulation of whole-cellplasma membrane conductance in corti-cal astrocytes. A: Representative currenttraces activated from a holding potential(Vh) of �60 mV after stepping to�120 mV for 500 msec and then slowlydepolarizing (180 mV/600 msec) to�60 mV (right inset); pulse interval 10sec. Upon exposure to 10 �M AA, theramp (*) outward potassium current (1)showed a partial inhibition (2), whichwas followed by a large, outward recti-fying increase in membrane conductancein the whole range of ramp potentials(3). Note that upon washout of AA (4),the current remained smaller that thecontrol trace. The traces obtained bydigital subtraction of the currents depict-ing the net decrease (1–4) and increase(3–4) in ramp currents produced by AAindicate that whereas AA blocked thebasal voltage-gated outward K� current,the AA-mediated current augment wasin the whole range of ramp potentials.B: The time course of the AA effects onthe currents measured at �60 mV illus-trated in A depicts the differences in theonset and time course of the dual actionsof AA. Numbers correspond to thetraces in A. The dashed line in A is thezero-current level, whereas in B it indi-cates the current level at �60 mV rela-tive to the control trace.

Arachidonate Activates Astroglial K� Channel 365

To characterize the gating mechanisms underlyingthe dual modulation of the whole-cell conductance, avoltage step protocol (Fig. 2A, inset) was used to activatevoltage- and time-dependent membrane currents. As re-ported previously (Ferroni et al. 1995), delayed rectifierK� currents activated at potentials above �40 mV and didnot inactivate during the voltage pulses. This K� conduc-tance was depressed rapidly upon challenge with 10 �MAA, an effect that was followed by the development oflarge fast-activating, non-inactivating membrane currents(fold-increase at �60 mV compared to washout: 6.8 3.5, n � 55; Fig. 2A–C). The steady-state current-voltagerelationship (I-V) of the AA-induced currents shows thatthis conductance was outwardly rectifying and generatedsubstantial currents at potentials more positive than�60 mV (Fig. 2D). The specificity of these effects wasdemonstrated by the result that administration of the sat-urated AA analog, arachidic acid (10 �M), did not modifythe profile of the control ramp currents (n � 12; data notshown).

AA Positively Modulates an Open RectifierPotassium Channel

Because the AA-induced current increase was paral-leled by a shift of the zero-current potential toward theNernst potential for K� under our experimental condi-tions (EK � �91 mV), it was postulated that such aconductance was a K� current. To address this question indetail, experiments were carried out under various extra-cellular K� concentrations ([K�]ext); by raising [K�]ext theAA-generated ramp currents changed polarity at more

positive potentials, and this effect was accompanied by areduction in current rectification (Fig. 3A). The plot ofthe reversal potentials as function of [K�]ext is indicativethat the AA-activated K� current was selectively perme-able to K� (Fig. 3B). Taken together, these findingsdemonstrate that the positive AA modulation of the K�

conductance was mediated by an open rectifier channel,which has remarkable similarities to AA-sensitive mem-bers of the two-pore-domain K� channel family (Gold-stein et al., 2001).

Pharmacologic Modulation of the AA-ActivatedPotassium Channel

Open rectifier K� channels regulated by polyunsat-urated fatty acids display a partial sensitivity to the classicK� channel blockers (Patel and Honore, 2001). More-over, because these two-pore-domain K� channels areactivated also by membrane stretch, they are blockedpotently by gadolinium ion (Gd3�) (Lesage and Lazdun-ski, 2000), which inhibits several stretch-activated ionchannels (Hamill and McBride, 1996). Therefore, we nextaddressed the pharmacologic sensitivity of the AA-inducedK� current by exposing astrocytes to various K� channelblockers and to Gd3�. AA-evoked K� currents were inhib-ited strongly by quinine (200 �M), and quinidine (200 �M),but less by Ba2� (2 mM) and TEA (10 mM). All theseagents had a rapid effect, which was reversible promptlyupon washout. Extracellularly applied Gd3� (50 �M) alsodepressed potently the K� open rectifier. Interestingly, theGd3� effect became irreversible by increasing the time ofexposure (data not shown). Because the Gd3� sensitivity

Fig. 2. Arachidonic acid positivelymodulates the whole-cell astrocytemembrane conductance through activa-tion of a rapidly activating, non-inactivating outward current. A: Repre-sentative family of current traces evokedfrom Vh �60 mV with voltage stepsfrom �120 to �60 mV in �20 mVincrements delivered every 10 sec (in-set). Voltage- and time-dependent cur-rents, consistent with the activation ofthe delayed rectifier K� conductance,were elicited at potentials positive to�40 mV. B: In the same cell, prolongedapplication (�2 min) of 10 �M AA gen-erated larger outward currents, whichwere fast-activating and did not inacti-vate during the 250-msec voltage pulses.C: Smaller currents, compared to con-trol conditions, were evoked after re-moval of AA. D: The I-V curve of thecurrent traces (inset) obtained by digitalsubtraction of families in B and C illus-trates the outward rectifying profile ofthe AA-mediated conductance, which isidentical to that observed with thevoltage-ramp protocol.

366 Ferroni et al.

might be indicative that the AA-induced K� current wasmediated by a stretch-sensitive channel, we next exploredits osmosensitivity. We did not measure AA-mediatedcurrents in astrocytes exposed to hypotonic conditionbecause this would also lead to activation of swelling-induced Cl� currents (Olson and Li, 1997). Upon mildastrocyte shrinkage obtained by raising the extracellularosmolarity by �50 mOsm, however, the AA-induced K�

current was depressed significantly compared to isotonic

conditions. Collectively, this pharmacologic profile (Fig.4) evidenced further the similarities of the AA-inducedK� conductance expressed by cultured cortical astrocyteswith the AA-activated, stretch-sensitive two-pore-domainK� channels displayed by neuronal cells (Lesage and Laz-dunski, 2000; Goldstein et al., 2001).

Cellular Mechanisms Underlying AA-MediatedActivation of the K� Open Rectifier

Several lines of evidence indicate that AA can mod-ulate ion channel activity through various intracellularsignaling cascades, some of which are dependent on AAmetabolism (for a review, see Meves, 1994). Therefore,we next explored whether the byproducts of AA metab-olism were involved in the AA-mediated increase in K�

conductance. The data demonstrate that lipoxygenase,cyclooxygenase, and cytochrome P450 epoxygenase sig-naling cascades were not involved as the magnitudes of themaximally AA-activated currents did not differ signifi-cantly in the presence of various metabolic inhibitors (Fig.5A). Likewise, the time course of the current increase alsowas not altered. Protein kinase C (PKC), which has beenreported to mediate some of the modulatory effects of AAon several ion channels (see Meves, 1994), was also notinvolved as its pharmacologic inhibition did not modifythe positive modulation of the K� channel (Fig. 5B).Finally, because there is substantial evidence that AA gen-erates large amounts of oxygen free radicals, we addressedthe question of whether the K� current increase entailedan increase in free radicals; however, neither the intracel-lular addition of a superoxide free radical scavenger nor theinhibition of NADPH-oxidase had significant effects onthe AA-stimulated channel activity (Fig. 5C). Thus, AAactivates the K� open rectifier through a mechanism thatis not dependent on classic AA-activated signal transduc-tion pathways.

Because AA can readily cross the plasma membrane(Glatz et al., 1997), we next explored whether AA actedextracellularly or from the cytoplasmic site of the plasmamembrane. To this end, initial experiments were carriedout by adding AA into the recording pipette and analyzingthe effect of extracellular administration of AA. Underthese experimental conditions, extracellular AA evoked anincrease in K� conductance that did not differ significantlyfrom that produced without AA in the pipette (Fig. 6A).Furthermore, intracellular AA alone did not produce anincrease in K� conductance, even when a higher intra-cellular AA concentration was used (30 �M dialyzed for20 min; n � 7). Next, we evaluated the AA effect in thepresence of bovine serum albumin (BSA) in the pipette(2.5 mg/ml). Because of the large capacity of the BSA tobind free AA and act as fatty acid scavenger (Bojesen andBojesen, 1994), its addition into the pipette would preventthe intracellular accumulation of AA. The data indicatethat with intracellular BSA, external AA remained able togenerate large K� currents that did not differ significantlyfrom AA-evoked control traces (Fig. 6A). Finally, weanalyzed the rate of recovery of the AA-mediated con-ductance increase in the presence of BSA applied extra-

Fig. 3. The AA-mediated conductance is due to the activation of a K�

open rectifier. A: AA-induced ramp currents were evoked in various[K�]ext concentrations by replacing sodium. Elevation of [K�]ext in-duced a positive shift in reversal potential (Erev) and a reduction inrectifying profile, to indicate the activation of an open rectifier K�

channel whose behavior approximates that predicted by the constantfield theory for simple K� electrodiffusion. The traces relative to thetwo different [K�]ext were obtained by digital subtraction of the controlcurrents and those relative to the maximal effect of AA. B: The shiftsin reversal potential measured in various [K�]ext were fitted by aleast-squares linear regression, which depicts a 53-mV shift in reversalpotential per 10-fold change in [K�]ext, close to the theoretical Nernstpotential of 58 mV. Numeric values in parentheses are the number ofcells tested in each condition.

Arachidonate Activates Astroglial K� Channel 367

cellularly and intracellularly. Compared to control condi-tions, extracellular administration of BSA accelerated therate of washout, as expected for BSA stripping the AAinteracting with the extracellular side of the plasma mem-brane (Fig. 6B; n � 9). This result was not modifiedsignificantly by addition of BSA into the recording pipette(n � 8, data not shown). Taken together, these findingssupport the notion that to activate the K� open rectifierAA must interact extracellularly with some components ofthe plasmalemma.

RT-PCR Analysis of Two-Pore-Domain K�

ChannelsBecause both the biophysical and pharmacologic

properties of the astrocytic K� open rectifier resembledthose of the AA-activated members of the two-pore-domain K� channel family, we examined the mRNAexpression of TRAAK, TREK-1 and TREK-2 channels(Fig. 7). RT-PCR amplifications with primer pairs spe-cific for TREK/TRAAK channel cDNA sequences werecarried out with RNA from pure cultured cortical astro-cytes and from total brain extracts used as positive control.The expression of Kv1.1, a voltage-gated K� channel alsopresent in cultured astroglia (Smart et al., 1997), was takenas further internal control. The analysis shows that PCRproducts of the appropriate size corresponding toTREK-1 and TRAAK transcripts could be detected intotal brain, but were lacking in cultured astrocytes. Bycontrast, the product of TREK-2 was present in bothbrain extracts and astrocytic cultures. Taken together,

these data suggest that TREK-2 may be the molecularentity underlying the AA-mediated increase in astrocyteK� conductance.

DISCUSSIONThe main finding of this study is the demonstration

that the biologically active polyunsaturated fatty acid AAincreases the whole-cell K� conductance of cultured ratcortical astrocytes by activating a novel open rectifier K�

channel. The opening of this channel follows the rapid,partial inhibition of the delayed rectifier K� current iden-tified previously in cultured astrocytes (Ferroni et al.,1995). Altogether, these observations confirm the com-plexity of AA effects on ion channel activities (Meves,1994).

The result that exposure to Gd3�, which inhibitedstrongly the AA-induced K� current, did not affect sig-nificantly the basal K� current (unpublished results) indi-cates that the AA-mediated increase in K� conductancedid not entail a positive modulation of voltage-gated K�

channels. By contrast, the overall biophysical and pharma-cologic properties of the positively-modulated K� channelresemble those of the TREK/TRAAK members of thelarge family (KCNK) of two-pore-domain K� channelsthat recently have been cloned (Goldstein et al., 2001).This family comprises open rectifier channels that areregulated differently by several biological parameters, in-cluding intracellular and extracellular pH, oxygen tension,changes in osmolarity, membrane stretch, and unsaturatedfatty acids (Patel and Honore, 2001). Concerning the

Fig. 4. Pharmacology of the AA-activated open rectifier K� channel incortical astrocytes. Histograms depicting the inhibition of the AA-induced K� conductance measured at �60 mV upon extracellularapplication of several classic K� channel blockers (A), and underconditions that negatively modulate stretch-activated channels (B).

Normalized currents were obtained by subtracting the residual currentafter AA washout to the current value measured after exposure to10 �M AA alone (ctrl) or co-application of AA with the different testagents. Numeric values in parentheses are the number of determina-tions in each condition. **P � 0.01 vs. ctrl.

368 Ferroni et al.

members that are activated by AA and other polyunsatu-rated fatty acids, their properties have been characterizedbest upon gene expression in heterologous systems (Finket al., 1996, 1998; Bang et al., 2000); so far there has beenno information about their functional presence in non-neuronal cells of the central nervous system (CNS). In fact,immunohistochemistry and in situ hybridization studieshave shown that they are expressed in neurons of differentregions of the CNS, but not in areas with high density ofglial cells (Talley et al., 2001). Recently, it was reportedthat TREK-1, a KCNK member activated both by AAand membrane stretch, is the molecular counterpart of theS-channel in Aplysia neurons (Patel et al., 1998). Anothermember, named TRAAK, was found distributed widelyin neurons of the CNS (Fink et al., 1998). Other studiesdemonstrated that the transcript of TREK-2, a secondmember of the TREK family, is expressed mainly in thecerebellum, and also suggested that it might underlie partof the AA-activated K� current in cultured neurons (Kimet al., 1995; Bang et al., 2000). Recently, two newTREK-2 isoforms have been identified, one of which,called TREK-2c, has a strong expression pattern in severalbrain regions (Gu et al., 2002). Our data indicate thatamong the different members of the two-pore-domain K�

channels modulated positively by AA, cultured corticalastrocytes express TREK-2 transcript. While preparingthis article, a study reporting the identification and single-channel characterization of TREK-2 in cultured astrocyteswas published (Gnatenco et al., 2002). We did not inves-tigate the single-channel behavior, which both in astroglialcells and under heterologous expression showed an in-wardly rectifying profile at negative membrane potentials(Bang et al., 2000; Gnatenco et al., 2002). A notabledifference between our data and those reported in mixedastroglial culture composed of protoplasmic (type-1) andfibrous astrocytes is that whereas in mixed cultures, astro-cytes responded to AA exposure with either an increase ora decrease in K� conductance (Gnatenco et al., 2002), inour pure type-1 astrocytic cultures, AA administrationinvariably produced an initial K� current depression fol-lowed by a delayed activation of the K� open rectifier.

The latency in the activation of TREK-2 channelafter AA application is unlikely to be mediated by intra-cellular signaling cascades regulated by AA metabolism,because the AA response was not altered by astrocytetreatment with pharmacologic agents inhibiting lipoxy-genase, cyclooxygenase, or cytochrome P450 epoxygenasepathways, or by removing extracellular calcium (unpub-lished results). Likewise, blocking PKC activation andinterfering with oxygen free radical formation, also in-volved in AA-mediated channel modulation in other celltypes (Meves, 1994), did not prevent the development ofthe K� current. Finally, the increase in K� conductancecould be elicited several times and without rundown dur-ing each experiment, which lasted up to 45 min. Theseresults mirror those of TREK/TRAAK channels ex-pressed in heterologous systems (Fink et al., 1996, 1998;Bang et al., 2000), and support the notion that the slow

Fig. 5. The positive modulation of the K� open rectifier is not me-diated by classic intracellular signal transduction pathways activated byarachidonic acid. A: Effects of the blockage of AA metabolism on thedevelopment of the K� current. Astrocytes were treated for 30 minwith various test agents before electrophysiological recordings weremade. The agents were also added into the pipette solution. Theblockade of cyclooxygenase (1 �M indomethacin), lipoxygenase (2 �Mnordihydroguaiaretic acid, NDGA) and cytochrome P450 epoxygenase(10 �M 17-octadecynoic acid, 17-ODYA) cascades did not alter theAA-elicited currents at �60 mV when compared to control traces(ctrl). B: Inhibition of protein kinase C with different agents (50 �MH7 or 1 �M Ro38-8220) did not prevent the AA-mediated positivemodulation of the K� open rectifier. C: Scavenging oxygen freeradicals by dialyzing the astrocytes with superoxide dismutase (SOD;100 U/ml) or inhibiting the NADPH-oxidase (1 �M diphenylene-iodonium, DPI) did not affect the current amplitudes evoked by AA.Numeric values in parentheses are the number of determinations ineach condition.

Arachidonate Activates Astroglial K� Channel 369

onset of the current increase is likely due to a slowlydeveloping interaction of AA directly with the channelproteins or the phospholipid environment. In this context,it was shown recently that deletion of a charged amino

acid cluster in the C-terminus of TREK-2 abolished thechannel sensitivity to intracellular pH, mechanical stimuli,and fatty acids, suggesting that an intracellular domain ofthe channel protein is essential for AA action (Kim et al.,2001). It has been postulated also that fatty acid sensitivityof TREK-1 depends on the insertion of negatively-charged amphiphiles into the external leaflet of the plasmamembrane, which, by producing a change in the mem-brane structure, activate the channel through a chargedsensor region in the K� channel (Patel et al., 2001). Ourresults are consistent with this hypothesis, because whereasAA effect was not modified by intracellular BSA, it wasreversed rapidly by its extracellular administration. Fur-thermore, inclusion of AA into the pipette did not pro-duce any change, both in the basal current and in theAA-mediated current increase. Additional evidence sug-gesting that the AA-induced opening of TREK-2 was dueto an AA-mediated membrane crenation effect is the findingthat another membrane phospholipid crenator, lysophos-phatidylcholine (10 �M) (Patel at al., 2001), caused a rapidincrease in K� conductance, which was even larger than thatgenerated by AA (unpublished results). Altogether, these dataoffer an interpretation of the sidedness of the AA action thatis different from that postulated from heterologous expres-sion of TREK-2, which suggested that AA binds intracel-lularly (Kim et al., 2001).

In neurons, the positive modulation of TREK/TRAAK channels by AA was postulated to be beneficialby decreasing the cell excitability through activation of abackground K� current (Lesage and Lazdunski, 2000).This interpretation was supported further by the findingthat these channels were activated also by neuroprotectiveagents and anesthetics (Patel at al., 1999; Duprat et al.,

Fig. 6. Arachidonic acid interacts with the extracellular site of theplasma membrane to open the K� open rectifier. A: The K� currentswere elicited upon exposure to AA in control conditions (ctrl) and byadding either bovine serum albumin (BSA; 2.5 mg/ml) or 10 �M AAinto the recording pipette. Neither the intracellular addition of AA (AApip) nor the scavenging of intracellular AA with BSA (BSA pip)modified the current responses to extracellular AA. Currents at�60 mV were normalized with respect to the traces generated byexposure to AA under control conditions in astrocytes recorded the

same day of experiments. Numeric values in parentheses are the num-ber of determinations in each condition. B: Time course of the currentrecovery upon AA washout in control condition (ctrl) and in thepresence of extracellular BSA (10 mg/ml). Time zero was set withrespect to the normalized amplitude of the AA response at �60 mVbefore the initial current decline. Note the increase in the rate ofrecovery upon BSA exposure. The data points for each condition werebest fitted by a single exponential decay with time constants of 56 19sec (ctrl) and 5 2 sec (BSA).

Fig. 7. RT-PCR analysis of AA-activated two-pore-domain K� chan-nels in astroglial culture and total brain extracts. A: RNA from totalbrain (TB) and pure type-1 astrocytes in primary culture (CA) wereused to prepare first strand cDNAs. Specific primers were used andfragments of expected sizes were obtained for TREK-1 (281 bp),TRAAK (517 bp), -actin (330 bp) and Kv1.1 (223 bp) in TB extracts,whereas in CA, TREK-1 and TRAAK were not detected. Twonegative controls were included: the RT(�) without the reverse tran-scriptase enzyme, and the PCR(�) without the cDNA. B: TREK-2specific primers were used to generate the expected PCR product of622 bp in TB and CA. DNA ladders are also shown.

370 Ferroni et al.

2000). In astrocytes, which are nonexcitable cells, this K�

channel must have a different role. Because the AA con-centration that activates this current is in the pathophysi-ological range, and because at concentration below 2 �M,no AA-mediated current could be detected (unpublishedresults), it is unlikely that in physiological situations thischannel has a housekeeping role. Instead, it may havesome relevance under pathophysiological conditions suchas hypoxia/ischemia, when free AA is increased (Rehnc-rona et al., 1982). Astrocytes in vivo are involved in thecontrol of extracellular K� homeostasis through plasmamembrane uptake mechanisms (Walz, 2000). It is assumedgenerally that such a regulation is mediated mainly byinward rectifier K� channels (Sontheimer, 1994). It maybe, however, that under hypoxic/ischemic conditions orduring spreading depression when extracellular K� risesup to 30–80 mM (Kraig and Nicholson, 1978; Hansen,1985), and there is a concomitant increase in free AA andastrocytic swelling (Chan and Fishman, 1978; Unterberget al., 1987; Lauritzen et al., 1990; Jing et al., 1994), theK� open rectifier may contribute to the uptake of extra-cellular K�. It is noteworthy that under our culture con-ditions, cortical astrocytes did not express inward rectifierK� channels (Ferroni et al., 1995). This finding, togetherwith the fact that cultured astrocytes resemble reactiveastrocytes because of their high proliferative rate, suggeststhat this AA-modulated channel may have an alternativefunctional significance. In this context, it was reportedrecently that activation of two-pore-domain K� channelstriggers cell death by favoring apoptotic volume decrease(Trimarchi et al., 2002). Because astrocytes can die by anapoptotic mechanism in response to hypoxic/ischemicconditions in vitro (Yu et al., 2001), it is tempting tospeculate that under these conditions the increase in freeAA may generate astrocytic apoptosis, at least in part, byactivating this novel K� open rectifier. Further studies willprovide a definitive answer to this crucial question.

ACKNOWLEDGMENTSWe thank A. Minardi for excellent technical assis-

tance in cell culturing. This study was supported by grantsfrom MIUR (Italy) to S.F., M.N., and C.R.

REFERENCESBang H, Kim Y, Kim D. 2000. TREK-2, a new member of the mech-

anosensitive tandem-pore K� channel family. J Biol Chem 275:17412–17419.

Barbour B, Szatkowski M, Ingledew N, Attwell D. 1989. Arachidonic acidinduces a prolonged inhibition of glutamate uptake into glial cells. Nature342:918–920.

Bojesen IN, Bojesen E 1994. Binding of arachidonate and oleate to bovineserum albumin. J Lipid Res 35:770–778.

Chan PH, Chen SF, Yu AC. 1988. Induction of intracellular superoxideradical formation by arachidonic acid and by polyunsaturated fatty acids inprimary astrocytic cultures. J Neurochem 50:1185–1193.

Chan PH, Fishman RA. 1978. Brain edema: induction in cortical slices bypolyunsaturated fatty acids. Science 201:358–360.

Duprat F, Lesage F, Patel AJ, Fink M, Romey G, Lazdunski M. 2000. Theneuroprotective agent riluzole activates the two P domain K� channelsTREK-1 and TRAAK. Mol Pharmacol 57:906–912.

Ferroni S, Marchini C, Schubert P, Rapisarda C. 1995. Two distinctinwardly rectifying conductances are expressed in long term dibutyryl-cyclic-AMP treated rat cultured cortical astrocytes. FEBS Lett 367:319–325.

Ferroni S, Valente P, Nobile M, Caprini M, Rapisarda C. 2001. Dualmodulation of whole cell potassium conductance by arachidonic acid incultured rat cortical astrocytes. Abstr Soc Neurosci 868:10.

Fink M, Duprat F, Lesage F, Reyes R, Romey G, Heurteaux C, LazdunskiM. 1996. Cloning, functional expression and brain localization of a novelunconventional outward rectifier K� channel. EMBO J 15:6854–6862.

Fink M, Lesage F, Duprat F, Heurteaux C, Reyes R, Fosset M, LazdunskiM. 1998. A neuronal two P domain K� channel stimulated by arachi-donic acid and polyunsaturated fatty acids. EMBO J 17:3297–3308.

Glatz JF, Luiken JJ, van Nieuwenhoven FA, Van der Vusse GJ.1997. Molecular mechanism of cellular uptake and intracellular translo-cation of fatty acids. Prostaglandins Leukot Essent Fatty Acids 57:3–9.

Gnatenco C, Han J, Snyder AK, Kim D. 2002. Functional expression ofTREK-2 K� channel in cultured rat brain astrocytes. Brain Res 931:56–57.

Goldstein SA, Bockenhauer D, O’Kelly I, Zilberberg N. 2001. Potassiumleak channels and the KCNK family of two-P-domain subunits. Nat RevNeurosci 2:175–184.

Gu W, Schlichthorl G, Hirsch JR, Engels H, Karschin C, Karschin A, DerstC, Steinlein OK, Daut J. 2002. Expression pattern and functional char-acteristics of two novel splice variants of the two-pore-domain potassiumchannel TREK-2. J Physiol 539:657–668.

Hamill OP, Marty A, Neher E, Sakmann B, Sigworth FJ. 1981. Improvedpatch-clamp techniques for high-resolution current recording from cellsand cell-free membrane patches. Pflugers Arch 381:85–100.

Hamill OP, McBride DW. 1996. The pharmacology of mechanogatedmembrane ion channels. Pharmacol Rev 48:231–252

Hansen AJ. 1985. Effect of anoxia on ion distribution in the brain. PhysiolRev 65:101–148.

Jing J, Aitken PG, Somjen GC. 1994. Interstitial volume changes duringspreading depression (SD) and SD-like hypoxic depolarization in hip-pocampal tissue slices. J Neurophysiol 71:2548–2551.

Katsuki H, Okuda S. 1995. Arachidonic acid as a neurotoxic and neuro-trophic substance. Prog Neurobiol 46:607–636.

Keyser DO, Alger BE. 1990. Arachidonic acid modulates hippocampalcalcium current via protein kinase C and oxygen radicals. Neuron 5:545–553.

Kim D, Sladek CD, Aguado-Velasco C, Mathiasen JR. 1995. Arachidonicacid activation of a new family of K� channels in cultured rat neuronalcells. J Physiol 484:643–660.

Kim Y, Gnatenco C, Bang H, Kim D. 2001. Localization of TREK-2 K�;channel domains that regulate channel kinetics and sensitivity to pressure,fatty acids and pHi. Pflugers Arch 442:952–960.

Kraig RP, Nicholson C. 1978. Extracellular ionic variations during spread-ing depression. Neuroscience 3:1045–1059.

Lauritzen M, Hansen AJ, Kronborg D, Wieloch T. 1990. Cortical spread-ing depression is associated with arachidonic acid accumulation andpreservation of energy charge. J Cereb Blood Flow Metab 10:115–122.

Lesage F, Lazdunski M. 2000. Molecular and functional properties oftwo-pore-domain potassium channels. Am J Physiol Renal Physiol 279:793–801.

Meadows HJ, Benham CD, Cairns W, Gloger I, Jennings C, Medhurst AD,Murdock P, Chapmann CG. 2000. Cloning, localisation and functionalexpression of the human orthologue of the TREK-1 potassium channel.Pflugers Arch 439:714–722.

Meves H. 1994. Modulation of ion channels by arachidonic acid. ProgNeurobiol 43:175–186.

Olson JE, Li GZ. 1997. Increased potassium, chloride, and taurine con-ductances in astrocytes during hypoosmotic swelling. Glia 20:254–261.

Arachidonate Activates Astroglial K� Channel 371

Ordway RW, Singer JJ, Walsh JV. 1991. Direct regulation of ion channelsby fatty acids. Trends Neurosci 14:96–100.

Patel A, Honore E. 2001. Properties and modulation of mammalian 2Pdomain K� channels. Trends Neurosci 24:339–346.

Patel AJ, Honore E, Lesage F, Fink M, Romey G, Lazdunski M. 1999.Inhalation anesthetics activate two-pore-domain background K� chan-nels. Nat Neurosci 2:422–426.

Patel AJ, Honore E, Maingret F, Lesage F, Fink M, Duprat F, Lazdunski M.1998. A mammalian two-pore-domain mechano-gated S-like K� chan-nel. EMBO J 17:4283–4290.

Piomelli D, Volterra A, Dale N, Siegelbaum SA, Kandel ER, Schwartz JH,Belardetti F. 1987. Lipoxygenase metabolites of arachidonic acid as sec-ond messengers for presynaptic inhibition of Aplysia sensory cells. Nature328:38–43.

Pusch M, Neher E. 1988. Rates of diffusional exchange between small cellsand a measuring patch pipette. Pflugers Arch 411:204–211.

Rehncrona S, Westerberg E, Akesson B, Siesjo B. 1982. Brain cortical fattyacids and phospholipids during and following complete and severe in-complete ischemia. J Neurochem 38:84–93.

Smart SL, Bosma MM, Tempel BL. 1997. Identification of delayed rectifierpotassium channel, Kv1.6, in cultured astrocytes. Glia 20:127–134.

Soliven B, Wang N. 1995. Arachidonic acid inhibits potassium conduc-tances in cultured rat oligodendrocytes. Am J Physiol 38:341–348.

Sontheimer H. 1994. Voltage-dependent ion channels in glial cells. Glia11:156–172.

Staub F, Winkler A, Peters J, Kempski O, Kachel V, Baethmann A. 1994.Swelling, acidosis, and irreversible damage of glial cells from exposure toarachidonic acid in vitro. J Cereb Blood Flow Metab 14:1030–1039.

Talley EM, Solorzano G, Lei Q, Kim D, Bayliss DA. 2001. CNS distri-bution of members of the two-pore-domain (KCNK) potassium channelfamily. J Neurosci 21:7491–7505.

Trimarchi JR, Liu L, Smith PJ, Keefe DL. 2002. Apoptosis recruits two-pore domain potassium channels used for homeostatic volume regulation.Am J Physiol Cell Physiol 282:C588–C594.

Unterberg A, Wahl M, Hammersen F, Baethmann A. 1987. Permeabilityand vasomotor response of cerebral vessels during exposure to arachidonicacid. Acta Neuropathol 73:209–219.

Visentin S, Levi G. 1998. Arachidonic acid-induced inhibition of microglialoutward-rectifying K� current. Glia 22:1–10.

Walz W. 2000. Role of astrocytes in the clearance of excess extracellularpotassium. Neurochem Int 36:291–300.

Yu AC, Wong HK, Yung HW, Lau LT. 2001. Ischemia-induced apoptosisin primary cultures of astrocytes. Glia 35:121–130.

372 Ferroni et al.