Acid-sensing Ion Channels in Malignant Gliomas*

High grade glioma cells derived from patient biopsies express an amiloride-sensitive sodium conductance that has properties attributed to the human brain sodium channel family, also known as acid-sensing ion channels (ASICs). This amiloride-sensitive conductance was not detected in cells obtained from normal brain tissue or low grade or benign tumors. Differential gene profiling data showed that ASIC1 and ASIC2 mRNA were present in normal and low grade tumor cells. Although ASIC1 was present in all of the high grade glial cells examined, ASIC2 mRNA was detected in less than half. The main purpose of our work was to examine the molecular mechanisms that may underlie the constitutively activated sodium currents present in high grade glioma cells. Our results show that 1) gain-of-function mutations of ASIC1 were not present in a number of freshly resected and cultured high grade gliomas, 2) syntaxin 1A inhibited ASIC currents only when ASIC1 and ASIC2 were co-expressed, and 3) the inhibition of ASIC currents by syntaxin 1A had an absolute requirement for either (cid:1) - or (cid:2) -hENaC. Transfection of cultured cells originally derived from high grade gliomas serum-free Eco 100 Optimem plus lipid Optimem coverslips incubated for in the DNA/lipid Optimem At the end of this and with Dulbecco’s Eagle’s plus fetal bovine The were allowed to overnight and used for patch-clamp experiments over the course the next 48 control peptide. Translated proteins were immunopurified, reconstituted into liposomes in different combinations, and solubilized before co-precipitation as described un- der “Materials and Methods.” To test for co-precipitation, antibodies directed against nonlabeled protein were used, and the presence of co-precipitated radioactively labeled ASIC was detected. the first of the two middle lines , protein from solubilized proteoliposomes contain- ing [ 35 S]Met ASIC1 plus unlabeled ASIC2 was immunoprecipitated using anti-ASIC2 antibodies; the next that ASIC2 is in the immunoprecipitate of a mixture of 35 S]Met ASIC2 plus unlabeled ASIC1 using anti-ASIC1 antibodies. The two demon- strate that neither ASIC1 nor ASIC2 can be co-precipitated with CLC5 demonstrating specificity of interaction between ASIC1 and ASIC2. three different rENaC subunits ( (cid:5) , (cid:1) , and (cid:2) ) were transcribed and translated in vitro in the presence of radioactive methionine. An aster- isk was added to the name of the channel subunits to mark radioactively labeled components. Transcription products were analyzed using SDS-PAGE. subunit with in vitro translated nonlabeled syntaxin 1A, in proteoliposomes, and solubilized before precipitation as described under “Materials and Syntaxin precipitated using monoclonal anti-syntaxin 1A antibod- ies, precipitates radioactively labeled subunits and autoradiography. (cid:1) and (cid:2) co-precipitate

Primary intracranial neoplasms remain a significant cause of mortality and morbidity in both children and adults. In patients with malignant gliomas of World Health Organization Grades III and IV, disease progression is uniformly rapid despite aggressive surgical and adjunctive therapies. Median survival of optimally treated individuals with the most aggressive of these tumors, glioblastoma multiforme, is 12 months, and this statistic has not varied for more than 30 years. The ability to treat these tumors successfully has been hampered by a fundamental lack of understanding of the control of the growth and differentiation of glial cells, how the transformed phenotype evolves, and how glioma cells modify their environment to support their increased energy demands. We have identified a novel amiloride-sensitive inward Na ϩ current that appears to be constitutively activated in malignant gliomas but not in low grade or normal astrocytes (1). In addition, glioma cells display up-regulation of Cl Ϫ and K ϩ channels not found, at least functionally, in normal glia (2,3). Thus, it is reasonable to hypothesize that ion transport systems specifically expressed by glioma cells are intimately related to and indeed may define the unique growth and migratory ability of these cells.
The main objective of the present study was to explore the molecular mechanisms that underlie the constitutively activated Na ϩ currents present in high grade glioma cells. We assume, based on our previous electrophysiological, pharmacological, and molecular biological studies, that the brain Na ϩ channels, also known as acid-sensing ion channels (ASICs), 1 may comprise the core conduction element of these channels. To date, six members of the ASIC family have been cloned in mammals (4).
These channels share the common property of generating excitatory currents in response to acidic pH when studied in heterologous expression systems, except for ASIC2b that, at least in its homomeric form, does not appear to respond to low pH (5). Moreover, ASIC4 is inactive by itself and hence is not thought to encode a proton-gated ion channel (6,7). Although the subunit composition of brain Na ϩ channels in native tissues is unknown, evidence for heteromultimeric channel formation with distinctive functional characteristics has been obtained (7)(8)(9)(10). These brain Na ϩ channels, like their epithelial counterparts, can be inhibited by amiloride and its analogs, although with a much lower affinity (11). A role for chemical pain sensation has been proposed for these channels in sensory neurons (12,13), but their role in the brain is obscure. Proton-activated neuronal currents have been identified in different brain regions (13,14). Hence, these channels may function as acid pH sensors in normal brain and in pathophysiological states such as ischemia or epilepsy where tissue acidification occurs (15,16). Gain-of-function mutations of ASIC1 and ASIC2 have also been detected and have been proposed to participate in neurodegenerative disease (7,17,18). The function of these channels in glia remains a mystery, yet functional amiloride-sensitive Na ϩ current expression selectively characterizes high grade gliomas.
There are a number of potential mechanisms that may produce a constitutive activation of this class of channel. We tested two hypotheses. First, as a consequence of oncogenic transformation, gain-of-function mutations of ASIC1 may occur. Second, we tested the hypothesis that syntaxin 1A may downregulate ASIC1 activity by analogy with the effects of syntaxin 1A on other ion channel activities, such as voltage-dependent Ca 2ϩ channels, CFTR, and ENaC. Our results show that gainof-function mutations of ASIC1 are not present in a multitude of freshly resected and cultured high grade gliomas (glioblastoma multiforme or GBM, World Health Organization classification Grade IV). However, syntaxin 1A inhibits ASIC currents, but only when both ASIC1 and ASIC2 are co-expressed. Moreover, the inhibition of ASIC currents by syntaxin 1A requires either ␥or ␦-hENaC. Furthermore, message for ASIC2 could be detected in 30 -40% of high grade gliomas. We suggest that it is the failure of syntaxin 1A to inhibit ASIC1 activity in the absence of plasma membrane-localized ASIC2 that accounts for the functional expression of amiloride-sensitive Na ϩ currents in astrocytoma cells.

MATERIALS AND METHODS
DNA Constructs-The cDNAs encoding full-length ␥-ENaC subunits, ASIC1, and ASIC2 are described elsewhere (19 -21). The human ASIC 1a (ASIC1) and ASIC2a (ASIC2) cDNAs were a gift from Drs. David Corey and Jaime Garcia-Añ overos of the Harvard Medical School. The human ENaC subunits were given to us by Dr. Michael J. Welsh of the University of Iowa.
In Vitro Transcription and Translation-cDNAs were transcribed and translated in vitro using the TNT transcription/translation system (Promega) as previously described (22). To test for protein-protein interaction between different ENaC subunits and syntaxin 1A or between ASIC1 and ASIC2, we translated these constructs either with radioactive or nonradioactive methionine, immunopurified them, and reconstituted them in different combinations in proteoliposomes as previously described (22). Proteoliposomes were solubilized in precipitation buffer that contained 50 mM Tris, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, and 0.1% SDS. All of the precipitation reactions were carried out in this buffer. As a rule, antibodies directed against nonlabeled proteins were used, and the presence of co-precipitated radioactively labeled proteins were detected using SDS-PAGE and autoradiography.
ASIC Expression in Xenopus Oocytes-DNA samples were in vitro transcribed using the SP6 or T3mMessage Machine kits (Ambion, Austin, TX). The integrity of the cRNA was assessed by running the samples under denaturing conditions on a formaldehyde-agarose gel. RNA concentration and purity were determined by UV spectrophotometry at 260 nm. The oocytes were removed from appropriately anesthetized adult female Xenopus laevis (Xenopus Express, Beverly Hills, FL) using standard techniques (23). Follicular cells were removed by the addition of collagenase to calcium-free medium as described (23). Defolliculated oocytes were washed in OR-2 medium (82.5 mM NaCl, 2.4 mM KCl, 1.0 mM MgCl 2 , 1.8 mM CaCl 2 , and 5.0 mM HEPES, pH 7.4) and allowed to recover overnight in half-strength Liebovitz medium at 18°C. Groups of stage VI oocytes were injected with cRNA in a 50-nl volume containing 12.5 ng of ASIC and/or ENaC subunits depending upon the experiments. Standard two-electrode voltage clamp procedures were performed at room temperature on the oocytes 24 -72 h post-injection.
The oocytes were impaled with two 3 M KCl-filled electrodes, each having resistances of 0.5-2 M⍀. A Dagan TEV-200 voltage clamp amplifier was used. Two Ag-AgCl reference electrodes were connected to the bath by 3 M KCl, 3% agar bridges. The regular perfused solution was ND96 (96 mM NaCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 2 mM KCl, and 5 mM HEPES, pH 7.4). To prepare ND96 having pH values below 6.0, MES replaced HEPES. To permit ASIC to recover completely from desensitization following acidification, at least 45 s elapsed prior to a subsequent challenge of the oocyte with another acidic pH solution. The experiments were controlled by pCLAMP 8.0 software (Axon Instruments, Burlingame, CA), and the proton-sensitive currents measured at Ϫ60 mV were digitized. The current-voltage (I-V) relationships were determined by changing the clamp potential in 20-mV increments from Ϫ100 to ϩ100 mV from a holding potential of 0 mV. All of the animal care and experimental protocols were approved by the University of Alabama at Birmingham Institutional Animal Care and Use Committee (protocol number APN 020506241).
Planar Lipid Bilayer Experiments-Planar lipid bilayers were formed from a solution of a 2:1 diphytanoyl-phosphatidyl-ethanolamine:diphytanoyl-phosphatidylserine dissolved in n-octanol (total concentration, 25 mg/ml). The membranes were spread onto a 200-m diameter hole drilled in a polystyrene cup. Membrane capacitance averaged 250 -350 picofarads. The standard bathing solution was 100 mM NaCl plus 10 mM MOPS buffer, pH 7.4. The lipids were obtained from Avanti Polar Lipids (Alabaster, AL). All of the solutions were filtered sterilized using 0.22-m Sterivex-GS filters (Millipore, Bedford, MA). Electrical connections and current measurements were made as previously described (24). The voltage was applied to the cis chamber, and the trans chamber was held at virtual ground. Oocyte membrane vesicles containing the channels of interest were applied to a preformed bilayer with a glass rod from the trans compartment with the potential held at Ϫ40 mV. Only membranes containing a single ion channel were used for experiments. The data were analyzed as before (24).
Whole Cell Patch Clamp-Micropipettes were constructed using a Narashigi pp-83 two-stage micropipette puller. The tips of these pipettes had internal diameters of ϳ0.3-0.5 m and outer diameters of 0.7-0.9 m. When filled with an electrolyte solution containing 100 mM potassium gluconate, 30 mM KCl, 10 mM NaCl, 20 mM HEPES, 0.5 mM EGTA, Ͻ10 nM free Ca 2ϩ , 4 mM ATP at a pH of 7.2, the electrical resistance of the tip was 1-3 M⍀. The bath solution was serum-free RPMI 1640 cell culture medium. The solutions approximate the ionic gradients across the cell membrane in vivo. The pipettes were mounted in a holder and connected to the head stage of an Axon 200A patch clamp amplifier affixed to a three-dimensional micromanipulator system attached to the microscope. The cells were viewed using a Nikon model TE200 inverted microscope fitted with an ultraviolet light source. Under fluorescence, transfected cells expressing jellyfish green fluorescence protein were identified for whole cell patch clamp analysis.
The micropipettes were abutted to the cells, and slight suction was applied. Seal resistance was continuously monitored using pCLAMP 8. After the formation of seals with resistances in excess of 1 G⍀, another suction pulse was applied to form the whole cell configuration by rupturing the membrane within the seal but leaving the seal intact. Successful completion of this procedure produced a sudden increase in capacitance with no change in seal resistance, indicative of the whole cell configuration. The cells were then held at a membrane potential of 0 mV and clamped sequentially for 1600 ms each to membrane potentials of Ϫ100 mV to ϩ100 mV in 20-mV increments, returning to the holding potential of 0 mV for 1 s between each test voltage. The currents were recorded digitally and filed in real time. The entire procedure was performed using a DOS Pentium computer modified for A/D signals with pCLAMP 8 software and with an A/D interface controlled by pCLAMP (Axon Instruments, Sunnyvale, CA).
Antibodies, Immunoprecipitation, Gel Electrophoresis, and Western Blots-The anti-ASIC2a antibodies (Alamone, Jerusalem, Israel), anti-ASIC1 antibodies (Chemicon International, Temecula, CA), anti-␥ENaC antibodies (raised against peptide; CNTLRLDRAFSSQLT-DTQLTNEL), and anti-syntaxin 1A antibodies (25) were used for immunoprecipitation and Western blot detection. Crude membrane fractions from the glioblastoma cell line SK-MG1 were prepared as previously described (26). Immunoprecipitation and co-precipitation from SK-MG1 cell lysate were performed using the size X protein A immunoprecipitation kit from Pierce according to the manufacturer's instructions. Briefly, affinity-purified ASIC2 antibody (280 g) was bound and cross-linked to protein A beads. After extensive washing, the beads were added to the SK-MG1 cell lysates (1.5 ml) and incubated overnight at 4°C. The cells were lysed with 1.5 ml of 1% Triton X-100, 150 mM NaCl, 5 mM EDTA, 50 mM Tris, pH 7.5, containing protease inhibitors. Immunoprecipitated proteins were eluted and analyzed by electrophoresis and Western blotting. Standard electrophoresis and blotting protocols were followed. Briefly, the protein was run on 8% SDS-PAGE mini gels with 4% stacking gels for about 1 h at 200 V in a Bio-Rad Minisubcell apparatus. The gels were transferred onto polyvinylidene fluoride, treated with 5% nonfat dry milk/Tris-buffered saline/ Tween, and probed with the appropriate antibodies. Secondary antibody was conjugated to horseradish peroxidase, and visualization was performed with chemiluminescent reagents (Amersham Biosciences). Controls included substitution of nonimmune rabbit IgG for primary antibodies.
ASIC2 Transfections-To determine whether expression of hASIC2 could restore the normal astrocyte Na ϩ current phenotype to astrocy-toma cells, we transfected hASIC2 contained in a eukaryotic expression vector (GW1-CMV, the kind gift of Dr. David Corey, Massachusetts General Hospital) into the SK-MG1 (gift of Dr. Gregory Cairncross, University of Calgary) or U87-MG (purchased from ATCC) astrocytoma cell lines using LipofectAMINE (Invitrogen) for whole cell patch-clamp analysis. Briefly, 5-10 g of hASIC2 was mixed with 100 l of serumfree Optimem (Invitrogen). In most experiments, hASIC2 was co-transfected with 3 g of a green fluorescence protein reporter vector. Control cells were either untransfected or transfected with GW1-CMV vector from which the hASIC2 insert had been removed by cutting with EcoRI and religating the cut ends. The DNA/Optimem mixture was mixed with 100 l of Optimem plus 7.5 l of lipid and incubated at room temperature for 45 min, after which time a further 800 l of Optimem was added. Tumor cells plated onto glass coverslips at 60 -80% confluency were rinsed twice in Optimem and then incubated for 6 h in the DNA/lipid Optimem mixture. At the end of this incubation period, the medium was aspirated and replaced with fresh Dulbecco's modified Eagle's medium plus 10% fetal bovine serum. The cells were allowed to recover overnight and were used for patch-clamp experiments over the course of the next 48 h.
Preparation of Syntaxin 1A and Syntaxin 3 Proteins-Defined regions of syntaxin 1A and syntaxin 3 (amino acids 1-266; referred to as ⌬C as they lack the C-terminal membrane anchor) were generated by polymerase chain reaction. Restriction sites (EcoRI in forward and XhoI in reverse), 5Јoverhangs (TATA), and a stop codon were introduced into the primers. The PCR product was cloned into pGEX5X-1 vector (Pharmacia Corp.) and transformed in a protease deficient Escherichia coli strain (BL21-DE3). The GST fusion protein (GST-Syn1A⌬C and GST-Syn3⌬C) was purified on glutathione-Sepharose beads (Pharmacia Corp.). The protein was eluted using 20 mM reduced glutathione in phosphate-buffered saline, pH 7.4, and dialyzed extensively at 4°C with at least four changes of cold phosphate-buffered saline. The protein was concentrated using a centrifugal filter device (Centriprep; 10,000 Dalton cut-off; Millipore). The protein was estimated using the BCA kit (Pierce) and stored in small aliquots at Ϫ80°C.
RNA Extraction-Freshly excised human brain tissue or primary cultured brain tumor cells (GBM or normal brain from temporal lobe) were frozen and stored in liquid nitrogen by the University of Alabama at Birmingham Neurosurgery Brain Tissue Bank under Institutional Review Board approval X9804090300. The frozen tissue was ground into a fine powder using a mortar and pestle, under liquid nitrogen, after which 1 ml of Trizol (Invitrogen) containing 250 g of glycogen was added. The Trizol/powder mixture was transferred to a chilled glass/Teflon homogenizer and ground for 10 strokes while on ice. The homogenate was sequentially passed through 25-and 26-gauge needles to further reduce cellular debris and then transferred to a 1.5-ml microcentrifuge tube. 200 l of chloroform was added to the homogenate, vortexed for 30 s to mix, and centrifuged at maximum speed (14,000 rpm) for 5 min using a tabletop centrifuge at room temperature. The aqueous phase was transferred to a fresh 1.5-ml tube, 500 l of ice-cold isopropanol were added, and the RNA was allowed to precipitate overnight at Ϫ20°C. The precipitated RNA was pelleted by centrifugation at maximal speed for 15 min at room temperature, washed with 1 ml of 70% ethanol, and centrifuged for 5 min. The pellet was dissolved in 100 l of RNase-free H 2 O, and 100 l of phenol:chloroform (1:1) was added and mixed by vortexing. After centrifugation, the aqueous phase was transferred to a fresh 1.5-ml tube, and 100 l of chloroform was added, mixed by vortexing, and centrifuged. After transferring the aqueous phase to a fresh 1.5-ml tube, 10 l of 7.5 mg/ml ammonium acetate and 250 l of 100% ethanol were added and mixed well, and the RNA was allowed to precipitate overnight at Ϫ20°C. The precipitated RNA was pelleted by centrifugation, washed twice with 70% ethanol, and resuspended in RNase-free H 2 O. The integrity of the RNA was verified following electrophoresis through 1% agarose-formaldehyde gels. All of the equipment (e.g. homogenizers, mortar, pestle, etc.) were pretreated with RNase-Zap (Ambion) and rinsed with diethyl pyrocarbonatetreated H 2 O prior to use. All of the human tissue acquisition and protocols were reviewed and approved by the University of Alabama at Birmingham Institutional Review Board (protocol numbers X021015002 and X980409003).
RT-PCR-RT-PCR was performed using a OneStep RT-PCR kit (Qiagen) according to the manufacturer's instructions, using 0.1-1 g of total RNA as template. Custom primers specific to ASIC1 were synthesized by Invitrogen and used at a final concentration of 0.6 M. For example, the forward and reverse primers 5Ј-CCCGCATGGCAAA-GAG-3Ј and 5Ј-GGCTCAGCAGGTAAAGTCC-3Ј corresponded to bases 1109 -1125 and 1572-1587 (plus 3 bases of the 3Ј-untranslated region) of ASIC1, respectively. Reverse transcription was performed using a single cycle of 50°C for 30 min. This was followed by a single cycle of 95°C for 15 min, which inactivates the reverse transcriptase while activating the HotStart Taq DNA polymerase, followed by 40 cycles of 94°C for 1 min, 56°C for 1 min, and 72°C for 1 min, and finally a single 10-min cycle at 72°C. Aliquots of each reaction mixture were electrophoresed on a 2% Nu-Sieve (FMC Corp.) agarose gel using PCR markers (Promega) to determine molecular size. Products of the correct molecular size were isolated from the gel using the QIAquick gel extraction kit (Qiagen) and subcloned into the pCR-2.1 vector using the TOPO-TA cloning kit (Invitrogen) following the manufacturer's instructions. Recombinants were selected by blue/white screening and restricted with EcoRI (Promega) to verify incorporation of insert of correct size. ASIC1 sequences were verified by further restriction enzyme digest analysis and automated DNA sequencing (DNA Sequencing Facility, Iowa State University).

RESULTS
We previously reported that there was an amiloride-sensitive component to whole cell inward currents in high grade, highly invasive brain tumors (1). These currents were seen in primary cultures of tumors as well as in established glioma cell lines. This amiloride-sensitive current was not present in low grade brain tumors or normal brain tissue. To verify these initial findings and to extend these results to freshly resected GBM tumors, we performed whole cell patch-clamp experiments on freshly excised normal astrocytes and freshly resected and primary cultured brain tumor tissue samples (Fig.  1). In the basal state, the current records for both freshly resected and primary cultures of World Health Organization Grade III and IV tumor cells were characterized by large inward currents (Fig. 1A), and these results were completely FIG. 1. Representative whole cell patch clamp recordings from freshly isolated normal human astrocytes and GBM (World Health Organization Grade IV) and primary cultures of different World Health Organization Grade astrocytomas. The cells were voltage-clamped between Ϫ100 and ϩ100 mV in 20-mV increments from a holding potential of 0 mV, and the resulting currents were recorded. The cells were superfused with RPMI 1640 medium, and the pipette contained 100 mM potassium gluconate, 30 mM KCl, 20 mM HEPES, 0.5 mM EGTA, 4 mM ATP, and Ͻ10 nM free calcium at a pH of 7.2. Top traces, basal currents. Middle traces, currents following superfusion with 100 M amiloride. Bottom traces, difference currents (amiloride-sensitive). This experiment was repeated at least four separate times for each cell type.
inhibited following superfusion with 100 M amiloride (Fig.  1B). Fig. 1C shows the difference current (i.e. the amiloridesensitive component). These results should be contrasted to the lack of effect of 100 M amiloride in normal astrocytes and Grades I and II astrocytoma cells. The absolute magnitudes of the outward currents at ϩ40 mV ( Fig. 2A) and inward currents at Ϫ60 mV (Fig. 2B) in the absence and presence of amiloride for normal astrocytes, different grade gliomas, medulloblastoma, and two continuous GBM cell lines are summarized in Fig. 2. Although there was no discernible pattern to the magnitudes of either the outward or inward currents, amiloride only blocked inward currents in the more aggressive, higher grade tumors (Grades III and IV and medulloblastoma). Amiloride likewise blocked inward currents in SK-MG and U87-MG cells, both originally derived from GBM. A summary of the current-voltage (I-V) characteristics of both freshly resected normal astrocytes and GBM cells is presented in Fig. 3. It is apparent that the GBM cells are depolarized by an average of 31 mV compared with normal astrocytes under these recording conditions. The depolarized zero current membrane potential is due to the presence of an enhanced Na ϩ conductance. These results confirm our previous findings and establish that this amiloride-inhibitable component to whole cell currents is present in freshly excised GBM cells.
Because these currents were inhibited by amiloride and because the apparent amiloride inhibitory constant (K i ) was high (i.e. ϳ10 M), we hypothesized that the channels underlying these currents were of the ASIC variety. Thus, to begin to test this hypothesis, we performed RT-PCR on total RNA extracted from human brain tissue samples obtained during craniotomy for epilepsy (normal tissue, lanes 1-3) or primary GBM resection (lanes 4 -6) and from continuous cell lines derived from anaplastic astrocytoma (lane 7), a gliosarcoma (lane 8), and 7 different GBMs (lanes 9 -15) (Fig. 4A). Specific primers were designed to amplify a 482-bp product for ASIC1a and a 447-bp product for ASIC2a. We also designed primers to identify ASIC 3, ASIC4, and ␣-, ␤-, ␥-, and ␦-hENaC subunits. The ASIC1a product was detected in all of the samples (both normal and tumor), including the pancreatic carcinoma cell line BxPC-3 (lane 16). In contrast, the ASIC2a message was found in the three normal samples and in only one of the freshly resected GBMs (lane 5) and in the D54MG, SK-MG1, and U373MG cell lines (lanes 9, 11, and 15). Direct sequencing of the PCR products confirmed their identity. ASIC 3 was present in many of the glioma cell lines and in both normal tissue and GBM tumors, whereas ASIC 4 was present in most of the fresh tissues samples but in none of the cultured cell lines. ␣-hENaC was found in the majority of the tumor cell lines, in one freshly excised GBM tissue but not in any of the normal samples. ␤-hENaC was found in one of the freshly resected GBMs (lane 4) and in the pancreatic carcinoma cell line but not in any other samples. Also, either ␦and/or ␥-hENaC were observed in most of the freshly excised GBM tissues as well as in the tumor cell lines. The ␥-hENaC was also present in the normal tissue as compared with ␦-hENaC.
Another independent set of experiments was performed to examine specifically the distribution of ASIC1 and ASIC2 message between different tumor samples (Fig. 4B). Total RNA was extracted from tissue obtained from resected GBM tumors (lanes labeled G), primary cultures of GBM cells (lanes labeled P), cultured normal human astrocytes (Astrocyte), and continuous GBM cell lines. Examining in toto the data in Fig. 4, These results support the hypothesis that ASIC may be a component of the channels that underlie the amiloride-sensitive currents seen in GBM. ASIC1 is present in both normal astrocytes and GBM tissues, but ASIC2 and ASIC4 are present in normal astrocytes and not in the majority of high grade gliomas. Therefore, although the actual molecular constitution of the observed amiloride-sensitive component found in high grade brain tumor cells is not known, it may be comprised of a variety of subunits from the degenerin/ENaC superfamily.
Electrophysiology of ASIC Expressed in Oocytes and Reconstituted into Planar Bilayers-We used Xenopus oocytes to express ASIC1 and ASIC2 individually and in combination. Whole cell acid-activated sodium currents, after expression of wild type ASIC1, ASIC2, and ASIC1/2 in oocytes, are shown in Fig. 5A. The upper panels show the actual currents measured during the voltage clamp protocol. There was no measurable current when the extracellular pH (pH O ) was 7.5. However, upon reduction of pH O to 4.0, there was a transient current response. Two currents are routinely measured: the peak inward current (I p ) and the steady state current (I s ), which was measured 8 s after I p . I p was greatest for ASIC1 and least for ASIC2. The combination of ASIC1/2 resulted in a transient current signature intermediate between ASIC1 and ASIC2. The reversal potentials for I p and I s were identical within all three groups of oocytes, but the reversal potential for ASIC2 and ASIC1/2 expressing oocytes was ϳ20 mV lower than for ASIC1 expressing oocytes (Fig. 5A, bottom panels). These results suggest that ASIC induces cation selective currents with P Na /P K ranging from 6 -20 depending upon whether the oocytes were injected with ASIC1, ASIC2, or ASIC1/2 cRNA. Moreover, these functional results suggest that ASIC1 and ASIC2 interact in such a way that the conductance characteristics are altered, consistent with what has been suggested in earlier studies (8).
We have presented evidence that both wild type and gainof-function mutants of ASIC1 and ASIC2 can be incorporated and studied in planar lipid bilayers (27). Well defined pHactivated (for the wild type) single channels with a unitary conductance of ϳ18 -20 picosiemens were recorded. However, little work has been done with gain-of-function ASIC mutants expressed in oocytes. Therefore, we expressed the G430F mutation of ASIC2 in Xenopus oocytes and subsequently assessed sodium currents (Fig. 5B). Voltage clamp traces in the absence and presence of amiloride are shown in the upper half of the figure. In contrast to the wild type channel, the currents do not inactivate, they are constitutively active (even at neutral pH), and the inward currents are sensitive to amiloride. The bottom half of the figure shows the inhibition in response to amiloride. These currents were recorded at Ϫ100 mV. Almost all of the inward current was inhibited by amiloride. The reversal potential for this gain-of-function mutation was ϩ3 mV, indicating a somewhat reduced P Na /P K as compared with the wild type channel as in bilayers. This gain-of-function mutation was less sensitive to reductions of external pH. No amiloride-sensitive currents could be elicited in oocytes (n ϭ 18, 3 frogs) injected with the gain-of-function ASIC1a mutant cDNA (i.e. G433F). However, this mutant channel could be recorded in planar bilayers and displayed the same biophysical and pharmacological characteristics as the G430F ASIC2a channel (not shown). We have previously shown that the protein actin, when added to the presumptive cytoplasmic surface of ENaC incorporated into planar lipid bilayers, can functionally interact with the channel reducing single channel conductance as well as altering cation selectivity (28). Addition of actin to the gain-offunction ASIC1 or ASIC2 channel had no effect on the single channel signatures. However, when vesicles were prepared from channels containing a mixture of gain-of-function ASIC1

FIG. 5. Acid-activated ASIC currents in Xenopus oocytes.
A, inward sodium currents versus time were measured under voltage clamp conditions. The oocytes were voltage-clamped from Ϫ100 mV to ϩ100 mV in 20-mV increments. The currents were activated by reduction in pH to 4.0. Between voltage clamp levels, the extracellular pH was returned to pH 7.5 for 45 s prior to the next voltage clamp level and pH 4.0 exposure. Bottom panels, current-voltage relationships for ASIC1, ASIC2, and ASIC1/2 expressing oocytes. f, I p ; q, I s (8 s after I p ). Each experiment was repeated four times with similar results. B, representative current traces of gain-of-function mutation of ASIC2 G430F expressed in Xenopus oocytes. The experimental conditions were the same except that experiments were all conducted at pH 7.5. Bottom left panel, representative current trace obtained at a voltage clamp level of Ϫ100 mV in an oocyte expressing ASIC2 G430F. At pH 7.5 (time 0), a very large inward current could be measured. 100 M amiloride obliterates the current, and it is reversible. Bottom right panel, summary current-voltage curve for the amiloride sensitive component of the gain-of-function mutant (n ϭ 4). The reversal potential was ϩ3 mV. plus ASIC2, actin reduced single channel conductance to ϳ9 picosiemens, and the channel was almost always open (not shown). Therefore, both our bilayer and oocyte expression studies show that ASIC1 and ASIC2 can be functionally studied, and more importantly, co-expression of ASIC1 plus ASIC2 results in channels that are functionally distinct from their homomeric counterparts.
Gain-of-function Point Mutations in ASIC1 Isolated from Brain Tumor Samples-To test the hypothesis that gain-offunction mutations may underlie the appearance of an amiloride-sensitive conductance in malignant tumor cells, we isolated total RNA from malignant gliomas following resection or in established cell lines. The results of these experiments are presented in Table I. Using RT-PCR and the appropriately designed primers, we sequenced the entire coding region of ASIC1 from these cells. From the sequence analysis, we determined that there was no gain-of-function in positions that have previously been described, namely, Gly 433 (ASIC1) or Gly 430 (ASIC2) (7,17,18). The results of these experiments demonstrate that obvious gain-of-function mutations do not occur in ASIC1. The ASIC1 primary sequence, for the most part, was identical to the wild type channel. However, during these analyses we did note some other random mutations in ASIC1, namely, F197S, D212G, E219K, and N395T. We constructed ASIC1 mutants containing these specific amino acid changes and expressed them in oocytes. None of these mutant constructs displayed gain-of-function activity (not shown). Thus, from these results we conclude that gain-of-function mutations in ASIC1 do not contribute to the constitutive amiloride-sensitive sodium conductance measured in high grade glioma cells.
SNARE Protein Interactions with ASIC- Fig. 6 presents data obtained from our previous differential gene profiling studies on human brains showing that syntaxin 1A is present in normal and high grade glioma tumor cells. A more extensive survey by RT-PCR demonstrates the presence of syntaxin 1A in normal human astrocytes, freshly resected GBMs, and GBM cell lines (Fig. 7). Moreover, ␥-hENaC appears to be present in all these samples as well. We tested the hypothesis that syntaxin 1A may down-regulate ASIC activity by analogy with the effects of syntaxin 1A on other ion channel activities such as voltage-dependent calcium channels, CFTR, and ENaC (25, 29 -32). The results of these experiments are shown in Fig. 8 and summarized in Table II. GST-Syn1A⌬C (0.175 M) had no effect on the activity of channels comprised of ASIC1 plus ASIC2 (either alone or in combination) when the ␥ subunit of ENaC was not present (Fig. 8A). However, if ␥-hENaC was simultaneously reconstituted with ASIC1 plus ASIC2, the same concentration of GST-Syn1A⌬C significantly inhibited channel open probability (Fig. 8A). The concentration dependence of this inhibitory effect is shown in Fig. 8B (K D ϭ 207 Ϯ 31 nM). If the syntaxin 1A-binding protein Munc-18 is added to the bath, subsequent addition of GST-Syn1A⌬C had no effect (Fig.  8C). GST by itself did not have any effect (Fig. 8D). Truncation of either the N or C termini of ␥-ENaC rendered GST-Syn1A⌬C ineffective (Table II). Interestingly, ␦-hENaC can substitute for ␥-hENaC. We conclude that syntaxin 1A can specifically downregulate ASIC activity only if ASIC1 plus ASIC2 and ␥ (or ␦)-hENaC are present. Essentially the same results were found in oocyte expression studies, i.e. inhibition of AISC1 plus ASIC2 by syntaxin required ␥-hENaC (Fig. 8E). This effect was specific for syntaxin 1A, because syntaxin 3 was without effect (not shown). Based on our molecular biological data (Figs. 4  and 6), it appears that in normal cells ASIC1, ASIC2, and ␥ (or ␦)-hENaC are present, but in most tumor cells ASIC2 is lacking or attenuated. Thus, inhibition of channel activity by syntaxin 1A would be relieved in tumor cells because of the absence or reduced levels in the plasma membrane of one necessary component of the complex, namely, ASIC2.
Biochemical Evidence of ASIC1a, and ASIC2a, and ␥-ENaC Interactions-The results of our preliminary functional studies in oocytes and bilayers imply an interaction between ASIC1, ASIC2, and ␥-ENaC. To provide biochemical evidence of such an interaction, we did a co-precipitation study in a glioma cell line (SK-MG1). The strategy was to immunoprecipitate ASIC2 and test for the co-precipitation of ASIC1 and ␥-ENaC by Western blot. The results are shown in Fig. 9. ASIC2 can be detected by Western blot after precipitation with anti-ASIC2 antibodies, but not if nonimmune IgG was used for precipitation (Fig. 9A), demonstrating the specificity of the anti-ASIC2 antibodies. Fig. 9 (B and C) shows co-precipitation of ASIC1 and ␥-ENaC with ASIC2 and lack of co-precipitation if nonimmune IgG was used for precipitation. These biochemical results demonstrate that ASIC1, ASIC2, and ␥-ENaC can associate to form heteromeric complexes in an intact glioma cell line.
To provide additional evidence for these potential interactions, co-immunoprecipitation experiments have been performed with in vitro translated and proteoliposome-reconstituted ASIC1, ASIC2, and the voltage gated chloride channel 5 (ClC5) as a control peptide (Fig. 10A). Transcripts of ASIC1 and ASIC2 were radioactively labeled (two left lanes). These transcripts were immunopurified and reconstituted either with the unlabeled conjugated ASIC partner or with unlabeled ClC5. The strategy was to immunoprecipitate one partner to detect the presence of the other by autoradiography. Our results dem-  6. Cluster diagram of selected ENaC and syntaxin 1A expressed in GBM samples compared with normal brain tissue. Colors represent the relative levels of gene expression with the brightest red indicating the highest level of expression and the green depicting the lowest level or absence of expression (these data were taken from supplemental information associated with a previously published study (39) with permission). onstrated that ASIC1 and ASIC2 can indeed associate to form heteromeric complexes, and co-precipitation between ASIC1 and ASIC2 is specific because of the lack of co-precipitation of ASIC1 and ASIC2 with ClC5. We used the same strategy to test for protein-protein interactions between ␥-hENaC and syntaxin 1A, between ␦-hENaC and syntaxin 1A, and between ␤-hENaC and syntaxin 1A (Fig. 10B). We used in vitro translated proteins for these co-precipitation experiments because of a lack of good anti-␦-hENaC antibodies. In vitro translation allowed us to test co-precipitation between two proteins, even if only one antibody is available, because the co-precipitated protein is radioactively labeled and can be detected without antibodies. In Fig. 10B, the results show that immunoprecipitation with anti-syntaxin 1A antibodies resulted in co-precipitation of both ␥-hENaC and ␦-hENaC but not ␤-hENaC.
Transfection of U87-MG and SK-MG-1 Tumor Cells-If the hypothesis is correct, namely, that the absence of an amiloridesensitive current in normal cells is due to the presence of ASIC1, ASIC2, ␥ (or ␦)-hENaC, and syntaxin 1A in the mem-brane and that the basally activated amiloride-sensitive conductances seen in high grade glioma cells is due to a relief of inhibition by syntaxin 1A because of the absence of ASIC2, then transfecting ASIC2 back into a tumor cell (displaying amiloride-sensitive currents) should re-establish the nontumor phenotype (i.e. no inward amiloride-sensitive current). The results of such experiments are shown in Fig. 11A. These experiments were performed in two different cultured glioma cell lines (U87MG and SK-MG1) that were originally derived from GBMs. The cells were transfected with an empty expression vector and green fluorescence protein (left panels). Superfusion of 100 M amiloride abolished the inward current in these GBM-derived tumor cells, as reported earlier ( Ref. 1 and Fig.  2). In contrast, in the right panels, tumor cells that were transfected with both green fluorescence protein and the vector containing ASIC2 resulted in a phenotype in which there was no amiloride-sensitive inward current. Likewise, inclusion of Munc-18, a syntaxin 1A-binding protein that blocks syntaxin interaction with other proteins (25), in the patch pipette of FIG. 8. The effects of GST-Syn 1A⌬C on ASIC1 plus ASIC2 incorporated into planer lipid bilayers. A, bilayers were bathed with symmetrical 100 mM NaCl and 10 mM MOPS, pH 7.4; the holding potential was 100 mV. These traces are representative of at least four individual experiments. B, concentration dependence of the inhibition effects of GST-Syn 1A⌬C on P o of ASIC1 ϩ ASIC2 ϩ ␥-ENaC channel in planar lipid bilayers. C, Munc-18 (0.5 M) addition prevents inhibitory effects of GST-Syn 1A⌬C on ASIC1 ϩ ASIC2 ϩ ␥-hENaC in planar lipid bilayers (n ϭ 4). D, GST by itself has no effect on ASIC1 ϩ ASIC2 ϩ ␥-hENaC (n ϭ 4). E, ASIC1, ASIC2, and heteromeric ASIC1/2 Ϯ ␥-hENaC following expression in oocytes. The currents (I p ) were normalized to the values measured at Ϫ60 mV in the absence of syntaxin 1A. The currents were evoked by a step decrease in pH o to 4.0. Co-expression of syntaxin 1A with ASIC1 ϩ ASIC2 ϩ ␥-hENaC resulted in significantly (p Ͻ 0.01) lower mean currents ASIC2-transfected U87MG cells rapidly restored the inward amiloride-sensitive Na ϩ currents (Fig. 11B). These results are consistent with the hypothesis that SNARE protein interactions (specifically syntaxin 1A) with ASIC may underlie the lack of channel activity seen in normal astrocytes and the appearance of such activity in the malignant state.
Induction of Amiloride-sensitive Currents in Normal Astrocytes-In normal astrocytes, no amiloride-sensitive inward currents are present. We hypothesize a tonic inhibition of these currents by two independent, convergent mechanisms: inhibition by PKC-␤ (33) and inhibition by syntaxin 1A. To test this hypothesis, we performed whole cell patch clamp experiments on primary cultured normal human astrocytes treated with chelerythrine chloride, a generalized PKC inhibitor, and Munc-18 (to disrupt syntaxin 1A-protein interactions) (Fig. 12). Chelerythrine treatment alone or inclusion of Munc-18 in the patch pipette had no influence on macroscopic currents. However, if astrocytes were pretreated with chelerythrine prior to entering the whole cell mode with Munc-18 included in the pipette, activated, amiloride-sensitive inward currents resulted. These results support the hypothesis that both PKC and syntaxin 1A inhibit ASIC currents in astrocytes. DISCUSSION Our previous results indicated that an inward sodium conductance is present in high grade gliomas but not in low grade tumors or normal astrocytes (1). Moreover, messenger RNA analysis (either by gene microarray or RT-PCR) indicated that ASIC1 message was present in normal and GBM tissues and that ASIC2 mRNA was only present in normal and not in the majority of glial cells examined thus far. In the present work, we addressed several potential molecular mechanisms that may account for the constitutive presence of an inward Na ϩ conductance in high grade glioma cells that is absent in normal astrocytes and lower grade astrocytic tumors. There are three main conclusions that can be drawn from our experiments: 1) gain-of-function mutations of ASIC1 do not account for the constitutive activation of sodium currents seen in high grade glioma tumor cells; 2) interactions between an ASIC-based channel complex and a SNARE protein (syntaxin 1A) result in channel inhibition; and 3) the molecular composition of functional brain amiloride-sensitive sodium channels is complex and may include both brain and degenerin/ENaC subunits.
Gain-of-function mutations in both ASIC1 and ASIC2 have been described. Using RT-PCR, we sequenced the entire coding region of ASIC1 from malignant tumor cells obtained either from freshly resected tissue or from established cell lines. From the sequence analysis, we determined that there were no gainof-function mutations in positions that have been previously described (8,17,34). Although we detected polymorphisms of ASIC sequence in positions other than those previously described for gain-of-function mutations, these ASIC mutants were not basally activated when heterologously expressed in Xenopus oocytes. Thus, we postulate that gain-of-function mutations of ASIC1 do not underlie the activated current seen in high grade gliomas. However, if the amiloride-sensitive currents are derived from a combination of subunits of ASIC and ENaC, as we suggest, our results do not exclude the possibility of gain-of-function mutations occurring in other channel com-  9. Analysis of the interaction between ASIC2, ASIC1, and ␥-ENaC. Crude membrane fractions from SK-MG cells were used for precipitation and co-precipitation. Anti-ASIC2 antibodies or nonimmune IgG cross-linked with protein A beads were used for precipitation. Precipitated and co-precipitated proteins were probed with anti-ASIC2 (A), anti-ASIC1 (B), or anti-␥-ENaC antibodies (C). ASIC2 can be detected by Western blot after precipitation with anti-ASIC2 antibodies but not if nonimmune IgG was used for precipitation (A). ASIC1 and ␥-ENaC co-precipitated with ASIC2 but not if nonimmune IgG was used for precipitation (B and C).
ponents. Moreover, we posit but have not yet proved that this complex of ASIC/ENaC components is by itself constitutively activated.
Observations from several different laboratories indicate that syntaxin 1A is expressed not only in the brain but also in epithelial cells such as the gut and airway where it can interact with ion channels such as CFTR (25) and ENaC (31,32). The inhibition of CFTR by syntaxin 1A requires a C-terminal membrane anchor of this SNARE (30). In addition, the modulation of CFTR channel function involves a direct binding interaction between the N-terminal tail of CFTR and the membrane-proximal helical domain of syntaxin 1A (30). There is also a large body of evidence indicating that syntaxin 1A can regulate neuronal calcium channels (29,(35)(36)(37). Syntaxin 1A has been shown to bind directly to the N-type calcium channel between loops 2 and 3 (also referred to as the synprint peptide) (38). Functionally, syntaxin regulates calcium channel activity at the level of gating rather than by altering membrane trafficking. Interestingly, epithelial sodium channels have also been demonstrated to interact both physically and functionally with syntaxin 1A (31,32). Co-expression with ENaC and syntaxin 1A in Xenopus oocytes has a regulatory effect on the ability of this channel to conduct sodium. The mechanism of this inhibition is not well understood, but it is known that the ␥-ENaC subunit is required (32). There is disagreement as to whether the inhibitory effect is due to altering the number of channels at the cell surface or to an alteration of the gating properties of the channel (31,32). Our data demonstrate that syntaxin 1A can also affect ASIC. Based upon our gene profiling RT-PCR and functional data, we hypothesized that syntaxin 1A inhibits ASIC, that the C-terminal anchor is not required and that ASIC1, ASIC2, and ␥-ENaC must be present within the channel complex for this inhibition to occur. These results were confirmed in our single channel bilayer experiments, indicating a direct interaction between syntaxin and the channel. Moreover, our RT-PCR results indicate that ␥-ENaC is not present in all tumors (Fig. 2). However, ␦-ENaC has a widespread distribution. Our functional studies do, in fact, indicate that ␦-ENaC can substitute for ␥-ENaC in this regard. Biochemically, we show that syntaxin 1A can immunoprecipitate both ␥and ␦-ENaC but not ␤-ENaC (Fig. 10). Our results show that ASIC1 ϩ ASIC2 ϩ ␥ (or ␦)-ENaC exists in normal brain tissue or low grade tumors and that ASIC2 is not present in more than half of the higher grade glioma cells that we have examined. In the other subset of tumors where ASIC2 message is present, we hypothesize that ASIC2 protein is not present in the plasma membrane. Thus, in the absence of plasma membrane ASIC2, syntaxin 1A, which is present in all cells of astrocytic origin (Fig. 7), cannot inhibit channels comprised primarily of ASIC1. In the absence of inhibitory inputs, the conductance would be constitutively active because of the heterogeneous combination of ENaC/degenerin components and/or low prevailing [Ca 2ϩ ] (27). Our biochemical and electrophysiological experiments are consistent with this hypothesis.
Our working paradigm for amiloride-sensitive Na ϩ current regulation in normal astrocytes and glial tumors is as follows. Two sodium channels, ASIC1 and ASIC2, are co-expressed in normal cells; however, in cells derived from high grade astrocytic tumors, ASIC2 is either not expressed at all or is retained within the cell such that it is absent from the plasma membrane. ASIC1 is present in the plasma membrane of both normal and tumor cells. The consequence of this difference in expression and cellular localization between ASIC1 and ASIC2 is that astrocytic tumor cells can be characterized by a large inward Na ϩ current that is sensitive to the Na ϩ channel blocker, amiloride. In contrast, in normal cells, this current is not expressed; thus, the role of ASIC2 in this context is to suppress the otherwise rampant Na ϩ conductance. Moreover, in normal cells, two independent inhibitory pathways are operative: inhibition by syntaxin 1A and inhibition by PKC-␤ (33). Addition of a broad spectrum PKC inhibitor (chelerythrine chloride) to normal astrocytes, in combination with a disruption of syntaxin 1A/Na ϩ channel interaction with Munc-18, results in the appearance of a large, amiloride-sensitive Na ϩ conductance.
Although our results provide a plausible explanation for the constitutive activation of sodium conductance seen in malignant glioma cells, there is yet another possibility to consider, namely, that ASIC2 is not expressed in normal astrocytes. Our gene profiling studies were performed on tissue specimens obtained in surgery for intractable epilepsy (39). We recognize that there are limitations in the use of any source of intact human tissue. Even though we obtained tissue samples from white matter and not cortex and the number of astrocytes exceeds the number of neurons by at least 1 order of magnitude in these samples, these tissue specimens, nonetheless, contain other cell types such as neurons. Even though RT-PCR results indicate the presence of ASIC2 in primary cultures of human astrocytes (Fig. 4B) 10. Analysis of interactions between ASICs and different hENaC subunits with syntaxin 1A using co-precipitation. A, in vitro (IN V) transcription and translation of ASIC1 and ASIC2 were performed using either radioactive (two left lanes) or nonradioactive methionine. Nonlabeled ClC5 was used as a control peptide. Translated proteins were immunopurified, reconstituted into liposomes in different combinations, and solubilized before co-precipitation as described under "Materials and Methods." To test for co-precipitation, antibodies directed against nonlabeled protein were used, and the presence of co-precipitated radioactively labeled ASIC was detected. In the first of the two middle lines, protein from solubilized proteoliposomes containing [ 35 S]Met ASIC1 plus unlabeled ASIC2 was immunoprecipitated (IP) using anti-ASIC2 antibodies; the next lane shows that ASIC2 is present in the immunoprecipitate of a mixture of [ 35 S]Met ASIC2 plus unlabeled ASIC1 using anti-ASIC1 antibodies. The last two lanes demonstrate that neither ASIC1 nor ASIC2 can be co-precipitated with CLC5 demonstrating specificity of interaction between ASIC1 and ASIC2. B, three different rENaC subunits (␤, ␥, and ␦) were transcribed and translated in vitro in the presence of radioactive methionine. An asterisk was added to the name of the channel subunits to mark radioactively labeled components. Transcription products were analyzed using SDS-PAGE. Each subunit was mixed with in vitro translated nonlabeled syntaxin 1A, reconstituted in proteoliposomes, and solubilized before precipitation as described under "Materials and Methods." Syntaxin 1A was precipitated using monoclonal anti-syntaxin 1A antibodies, and the precipitates were analyzed for the presence of radioactively labeled rENaC subunits using SDS-PAGE and autoradiography. Both ␥-hENaC and ␦-hENaC, but not ␤-hENaC, co-precipitate with syntaxin 1A.
munofluorescence suggest that ASIC2 is not expressed in normal astrocytes. But even in the absence of ASIC2, and thus the elimination of the syntaxin 1A inhibitory arm, ASIC1-based amiloride-sensitive currents would still be blocked because of the presence of PKC␤ I and ␤ II in normal astrocytes, PKC␤ I and ␤ II are not present in GBM (40). The experiments presented in   12 are consistent with this postulate. Our findings of the presence of constitutively activated, amiloride-sensitive sodium channels in glial-derived tumor cells run counter to the suggestions of Horimoto et al. (40), who show that transferring a gain-of-function ASIC mutation into cancer cells using tumorspecific promoters can produce cell death and suggest that this may be a new strategy for cancer gene therapy. Although this may be the case for certain somatic cell cancers, it cannot be so for brain tumors because a functionally activated sodium conductance is operative in the glial tumors that we have examined.
Thus, our work demonstrates that constitutive amiloridesensitive currents are a specific feature of the more aggressive brain tumors. At present, little is known about the function of ASIC in the brain and less about how the composition and activity of these channels are regulated. Our results provide new insights into the cellular mechanisms that control ASIC function in the brain and how these controlling mechanisms integrate channel function with the other physiologic demands of both normal and malignantly transformed glial cells. Moreover, this work demonstrates that amiloride-sensitive sodium channels cannot easily be classified based on simple biophysical parameters, such as single channel conductance and/or sensitivity to amiloride. This class of ion channel, both in the brain and in epithelial tissues, appear to have a variable composition, and hence tissue-specific differences in biophysical parameters may result from different channel compositions in different tissues. In any case, the novel findings concerning the regulation and biophysical properties of ASICs that have been presented in this work are important in defining the cell biology of brain tumor cells. Understanding the molecular targeting or regulatory sites of ASIC will undoubtedly promote the development of new strategies to circumvent the unwanted activation of these channels and thus have important therapeutic ramifications.