Heat Shock Protein 90β1 Is Essential for Polyunsaturated Fatty Acid-induced Mitochondrial Ca2+ Efflux*

Nonesterified fatty acids may influence mitochondrial function by alterations in gene expression, metabolism, and/or mitochondrial Ca2+ ([Ca2+]m) homeostasis. We have previously reported that polyunsaturated fatty acids induce Ca2+ efflux from mitochondria, an action that may deplete [Ca2+]m and thus contribute to nonesterified fatty acid-responsive mitochondrial dysfunction. Here we show that the chaperone protein heat shock protein 90 β1 (hsp90β1) is required for polyunsaturated fatty acid-induced mitochondrial Ca2+ efflux (PIMCE). Retinoic acid induced differentiation of human teratocarcinoma NT2 cells in association with attenuation of PIMCE. Proteomic analysis of mitochondrial proteins revealed that hsp90β1, among other proteins, was reduced in retinoic acid-differentiated cells. Blockade of PIMCE in NT2 cells by 17-(dimethylaminoethylamino)-17-demethoxygeldanamycin, a known inhibitor of the chaperone activity of hsp90, and hsp90β1 RNA interference demonstrated that hsp90β1 is essential for PIMCE. We also show localization of hsp90β1 in mitochondria by Western blot and immunofluorescence. Distinctive effects of inhibitors binding to the N or C terminus of hsp90 on PIMCE in isolated mitochondria suggested that the C terminus of hsp90β1 plays a critical role in PIMCE.

Cell Culture-The NT2 cell line was cultured in 100-mm dishes in Dulbecco's modified Eagle's medium, supplemented with 10% fetal calf serum (FBS) and penicillin (50 units/ml)/ streptomycin (50 g/ml) at 37°C in a humidified 5% CO 2 atmosphere incubator. In some experiments, cells were treated with retinoic acid or other chemicals as indicated for various periods. Cells grown to near confluence (90%) were harvested with trypsin (0.05%)/EDTA (0.02%), and suspensions of NT2 cells were used for measurement of intracellular Ca 2ϩ ([Ca 2ϩ ] i ) and preparation of mitochondria.
Induction of NT2 Cell Differentiation-NT2 cells were incubated with 10 M RA for 8 weeks to induce differentiation of neuron-like cells as previously described (14).
Measurement of [Ca 2ϩ ] i -NT2 cells or RA-differentiated cells suspended in HEPES/sodium/glucose (HNG) buffer containing 140 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 1.2 mM KH 2 PO 4 , 10 mM glucose, 10 mM HEPES, pH 7.4, were labeled with fura-2 AM (1 M) at 37°C for 30 min with agitation at 50 rpm. Loaded cells were washed and resuspended in HNG, and [Ca 2ϩ ] i mobilization in populations of NT2 cells or RAdifferentiated cells was measured in a fluorometer (QM-6; Photon Technology International), using a cuvette with the temperature stabilized at 37°C through a connected water bath. The ratio of fluorescence excited at 340 and 380 nm with emission of 510 nm was recorded and used to index the change of [Ca 2ϩ ] i , as previously reported (15).
Measurement of [Ca 2ϩ ] i mobilization in single NT2 cells was conducted using a Bio-Rad laser scan confocal imaging analysis system. The cells were grown in 12-well plates at a density of 10 4 cells/well for 48 h and treated with 17-DMAG or hsp90␤1 RNAi as indicated for an additional 48 -72 h. Fluorescent images of individual cells were then acquired, and [Ca 2ϩ ] i was measured as previously described (16).
Preparation of Mitochondria and Measurement of Mitochondrial Ca 2ϩ Efflux-The preparation of mitochondria from NT2 and RA-differentiated cells and rat tissues (liver, brain, spleen, heart, kidney, and skeletal muscle from 3-4-month-old male Sprague-Dawley rats, provided by Dr. Kong Zhang, Department of Medicine, University of Texas Health Science Center, San Antonio, TX) and the measurement of PIMCE were performed as described previously with modification (8). Rat tissues were cleaned and minced in 2 ml of ice-cold mitochondrial preparation buffer (MB1) containing 250 mM mannitol, 75 mM succinic acid, 0.1 mM EDTA, 0.5 mM EGTA, 10 mM HEPES, pH 7.4. Cleaned tissue samples were homogenized in 15-20 ml of MB1 on ice, using a Teflon-glass homogenizer (Glas-Col, Terre Haute, IN) at 50 rpm for 40 strokes. The homogenates were centrifuged at 130 ϫ g at 4°C for 15 min, and the supernatant was carefully removed and centrifuged at 10,000 ϫ g to precipitate the mitochondria. The samples were resuspended and washed in MB1, and protein concentrations were determined as described (17) using bovine serum albumin as the standard and following the manufacturer's instructions (Pierce). Mitochondrial preparations containing equivalent protein concentrations were then resuspended in 200 l of MB1, diluted with 1.8 ml of HEPES/potassium/glucose buffer (HKG) containing 20 mM NaCl, 100 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 10 mM glucose, 20 mM HEPES, pH 7.4, and loaded with 10 M X-rhod-1 AM for 30 min at 37°C with agitation at 20 rpm. The loaded mitochondria were resuspended in 100 -150 l of HKG and used to measure PIMCE as described in Ref. 8. Briefly, X-rhod-1 and Ca 2ϩ -loaded mitochondria in HKG were diluted to 2 ml of PBS0Ca (PBS supplemented with 100 M EGTA and 1 mM MgCl 2 ) in a cuvette with constant magnetic stirring. The fluorescence excited at 578 nm and emitted at 602 nm was measured in a fluorometer manufactured by Photon Technology International (Lawrenceville, NJ). The base line was established by incubation of mitochondria in the cuvette without the addition of PUFA. In experiments measuring PIMCE in rat tissue mitochondria, 40 g/ml mitochondrial protein was used for all tissues; 15-50 g/ml mitochondrial proteins was used for measuring PIMCE in mitochondria from cultured cells. Equivalent results were obtained when linoleic acid (LA)-induced Ca 2ϩ efflux from mitochondria of NT2 cells was measured in PBS0Ca or Ca 2ϩ -free HKG solution (containing 100 M EGTA instead of 1 mM CaCl 2 ).
Two-dimensional Gel Electrophoresis-Isoelectrofocusing of mitochondrial proteins from NT2 and RA-differentiated neural cells was performed in a Bio-Rad PROTEAN IEF Cell with 7-cm Immobilized pH gradient strips (pH 3-10) according to the manufacturer's instructions. The mitochondrial samples were suspended and solubilized in a buffer containing 9 M urea, 2.5 M thiourea, 2% CHAPS, 0.8% Bio-Lyte (pH 3-10), and 15 mM dithiothreitol at room temperature for 60 min. The dissolved samples were centrifuged at 150,000 ϫ g for 60 min at 4 -6°C. A 200-l aliquot of each supernatant was loaded, and the ReadyStrip IPG strip (pH 4 -7; Bio-Rad) was drawn gel side down through the solution without air bubbles. Each Ready-Strip Immobilized pH gradient strip was then covered with mineral oil to prevent precipitation of the urea. Following a 12-h rehydration, the strips were electrofocused for 4 h at 10,000 V. When completed, the strips were equilibrated in 5 ml of buffer containing 6 M urea, 2% SDS, 30% glycerol, 1% dithiothreitol, 50 mM Tris, pH 8.8, for 20 min. A subsequent equilibration was performed in 5 ml of the same buffer with 2.5% iodoacetamide substituted for the dithiothreitol. Protein samples in the equilibrated strips were separated on 8% SDS-PAGE. The gels were stained with Coomassie Brilliant Blue R-250 and scanned with a high resolution scanner (GS-800 Calibrated Imaging Densitometer; Bio-Rad). The protein levels in the gel images were analyzed with the PD-Quest program (Bio-Rad). The mitochondrial proteins with a molecular mass of Ն100 kDa and expression levels largely reduced in RA-differentiated cells were subjected to further analysis with mass spectrometry.
Protein Identification by Mass Spectrometry-Spots of interest in the SDS-PAGE were robotically excised by means of a Proteome Works spot cutter (Bio-Rad) and digested in situ with trypsin according to standard protocols based on the initial work of Mann and co-workers (18). Briefly, protein spots were excised from the gel and destained in 50% acetonitrile, 40 mM ammonium bicarbonate, pH 7.4, prior to digestion. Gel plugs were then dehydrated in 100% acetonitrile and rehydrated with 5-10 ng/l trypsin (Promega; modified) in 40 mM ammonium bicarbonate and incubated overnight at 37°C. The resulting digests were analyzed by capillary HPLC-electrospray tandem mass spectrometry on a Thermo Fisher LTQ ion trap mass spectrometer coupled to an Eksigent NanoLC micro-HPLC by means of a PicoView (New Objective) nanospray interface. Capillary on-line HPLC separation of tryptic peptides was conducted using the following conditions: column, New Objective PicoFrit, 75 m inner diameter, packed to 11 cm with C18 adsorbent (Vydac 218MSB5); mobile phase A, 0.5% acetic acid, 0.005% trifluoroacetic acid in water; mobile phase B, 90% acetonitrile, 0.5% acetic acid, 0.005% trifluoroacetic acid in water; gradient, 2% B to 42% B in 30 min; flow rate, 0.4 l/min. A data-dependent acquisition protocol was employed consisting of one survey scan followed by seven collision-induced dissociation spectra. The uninterpreted collision-induced dissociation spectra were searched against the NCBInr data base using Mascot (Matrix Science; 10-processor in-house license). Variable modifications considered include methionine oxidation and cysteine carbamidomethylation. A 95% confidence level threshold was used for Mascot peptide scores.
Immunohistochemistry-NT2 cells grown in 12-well plates were treated with vehicle (Lipofectamine only), 17-DMAG (1 nM), or RNAi of hsp90␤1 (20 M) in Opti-MEM with reduced FBS (5%) for 48 -72 h. The cells were labeled with MitoTracker (100 nM) at 37°C for 30 min in Dulbecco's modified Eagle's medium with 10% FBS. The cells were then fixed with 4% paraformaldehyde and permeabilized with 40 g/ml digitonin in PBS for 30 min at room temperature. The permeabilized cells were incubated with 10% FBS in PBS for 60 min and subsequently hybridized with or without monoclonal antibody to GRP94 (1:500 dilution in PBS containing 2% FBS, 0.01% Triton X-100; Santa Cruz Biotechnology) at 4°C overnight. The cells were washed with PBS containing 0.1% Tween 20 and incubated with Alexa 488-labeled goat anti-mouse IgG (1:1000 dilution; Invitrogen) for 1 h at room temperature. Labeled cells were washed, and images were acquired using a Bio-Rad confocal laser scan imaging system. The excitation laser beams used were 488 and 554 nm for Alexa and Mitotracker, respectively. Fluorescent images were sequentially recorded from the same field with 515 nm band pass and 570 nm long pass filters. The levels of hsp90␤1 in mitochondria were determined by the fluorescence ratio of Alexa 488 and MitoTracker from eight or nine randomly selected areas in each experiment.
Statistics-Data are presented as means Ϯ S.E. Comparisons were performed using a two-tailed Student's t test or analysis of variance. A significant difference was defined at p Ͻ 0.05.

RESULTS
Following our previous observations of PIMCE in NT2 cells, we investigated PIMCE of various rat tissues using isolated mitochondria. Mitochondria from a number of tissues tested The dark traces indicate the response to PUFA (arachidonic acid, LA, and docosahexaenoic acid), and the light traces represent basal Ca 2ϩ efflux in the absence of PUFA. The responses of brain and liver mitochondria are compared in B. V 0 and V LA are the mean rates Ϯ S.E. of Ca 2ϩ efflux at basal and following LA application, respectively. *, p Ͻ 0.01 liver versus brain, n ϭ 6.

Hsp90␤1 and Fatty Acid-induced Mitochondrial Ca 2؉ Efflux
(liver, spleen, and heart) showed PIMCE ( Fig. 1A; data from kidney and skeletal muscle not shown). However, mitochondria from rat brain showed diminished PUFA responses compared with mitochondria of other tissues (Fig. 1A). For example, the initial rate of LA-induced mitochondrial Ca 2ϩ efflux in the brain was less than one-tenth of that in the liver (Fig. 1B). Based on this finding, we designed experiments to identify the protein pathway(s) responsible for PIMCE. It has been demonstrated previously that following prolonged treatment with RA, NT2 cells are induced to differentiate into mature neurons (14). It is possible that the mitochondria of neurons differentiated from NT2 cells behave similar to those of neuronal brain cells. Accordingly, we hypothesized that as in the mitochondria of brain tissue, PIMCE would be attenuated in neurons differentiated from NT2 cells. As shown in Fig. 2, in control NT2 cells, LA (3 ϫ 10 Ϫ5 M) induced increases of intracellular Ca 2ϩ ([Ca 2ϩ ] i ; Fig.  2B), and PIMCE was observed in isolated mitochondria (Fig. 2C). After treatment of NT2 cells with RA according to the procedures described (14), we observed the differentiation of neuron-like cells (cf. differentiated cells in Fig. 2D with control cells in Fig. 2A), consistent with previous findings (14). Interestingly, LA-induced [Ca 2ϩ ] i responses in RA-differentiated cells were abolished (Fig.  2E), as was LA-responsive PIMCE in mitochondria isolated from these RA-treated cells (Fig. 2F).
Analysis of mitochondrial proteins from control NT2 and RA-differentiated cells with two-dimensional gel electrophoresis revealed alterations in the expression levels of multiple proteins (cf. gel images in Fig. 3, A and B). To identify the proteins   MARCH 21, 2008 • VOLUME 283 • NUMBER 12 that might be Ca 2ϩ channels or transporters involved in PIMCE, we selected three protein spots for further analysis by mass spectroscopy based on the following two criteria: 1) the protein levels were undetectable or largely reduced after RA treatment; 2) the molecular mass of the proteins should be Ն100 kDa (marked by arrowheads in Fig. 3A). A fourth protein spot with a molecular mass of ϳ30 kDa, which was also undetectable following RA treatment, was selected randomly as a control for mitochondrial localization. Corresponding positions of the spots in RA-treated cells are marked by the numbers and arrowheads in Fig. 3B. The three spots of interest were analyzed with mass spectroscopy and identified with Mascot (20) as follows: hsp90␤1, a homologue of tumor rejection antigen 1, gp96, and GRP94 (glucose-regulated protein 94) (Fig. 3A, spot 1) (cf. the NCBI site on the World Wide Web); the 150-kDa oxygen-regulated protein variant 1 (spot 2); and leucine-rich PPR motif-containing protein and nodal modulator 2 isoform 2 (spot 3). The randomly selected small protein (spot 4) was identified as pre-mRNA splicing factor SF2p32, the C terminus of which contained the complete sequence of MAM33 mitochondrial matrix protein. The MAM33 protein is thought to be involved in mitochondrial oxidative phosphorylation and in nucleus-mitochondrion interactions (21). The parameters used to identify these proteins are presented in Tables 1 and 2.

Hsp90␤1 and Fatty Acid-induced Mitochondrial Ca 2؉ Efflux
A literature search for information on the functions of the above identified proteins indicated that none of the proteins of interest was known to be a mitochondrial Ca 2ϩ channel or transporter. It has been reported that hsp90 and other chaperone proteins are involved not only in protein folding but also protein importation to the mitochondria (22,23) and/or interactions between mitochondrial and endoplasmic reticulum Ca 2ϩ stores (24). We thus focused on the protein hsp90␤1 and tested whether PIMCE in NT2 cells was affected by 17-DMAG, which binds to the N-terminal ATPase domain of hsp90 proteins and inhibits their chaperone activity. Prolonged treatment with 17-DMAG may also cause degradation of hsp90␤1 (37). Additionally, we also tested the effect of hsp90␤1 down-regulation with hsp90␤1 RNAi.
Treatment of cells with 17-DMAG for 48 h during culture caused a concentration-dependent inhibition of the LA-induced [Ca 2ϩ ] i signal measured in suspensions of cells, with maximal effect at 10 Ϫ9 to 10 Ϫ8 M 17-DMAG (Fig. 4A); over the same range of concentrations, 17-DMAG had no effect on the [Ca 2ϩ ] i response to the muscarinic agonist carbachol (Fig. 4B). 17-DMAG blocked the carbachol-induced [Ca 2ϩ ] i signal only at high concentrations (5 ϫ 10 Ϫ8 to 2 ϫ 10 Ϫ7 M) that significantly reduced cell counts (Fig. 4, B and C).
Image analysis of individual NT2 cells treated with 17-DMAG (10 Ϫ9 M) or hsp90␤1 RNAi for 48 h revealed that both treatments effectively blocked LA-induced [Ca 2ϩ ] i mobi-

TABLE 1 Potential candidates of the pathway(s) responsible for PUFA-induced mitochondrial Ca 2؉ efflux identified by capillary-HPLC-electrospray tandem mass spectrometry
Candidate proteins were identified in spots 1-3 from two-dimensional gel electrophoresis of mitochondrial proteins (Fig. 1G). Spot 4 was randomly selected and used as a control for localization in mitochondria.   (Fig. 5, A-F). Measurement of LA-induced Ca 2ϩ efflux in isolated mitochondria indicated that treatment with 17-DMAG and RNAi largely reduced the rate and amplitude of the LA response (Fig. 5, G-I). The rate of LAinduced mitochondrial Ca 2ϩ efflux was lowered to 13 and 22% (p Ͻ 0.01 versus control, n ϭ 3), and the amplitude was lowered to 28 and 46%, of control values by 17-DMAG and RNAi, respectively. In control experiments, 17-DMAG did not alter Ca 2ϩ loading to isolated mitochondria from NT2 cells (supplemental Fig. 1). Images demonstrating the effect of 17-DMAG (10 Ϫ9 M) and RNAi from a typical experiment using intact cells are presented in Fig. 6. In vehicle-treated (control) NT2 cells, LA (3 ϫ 10 Ϫ5 M) or carbachol (10 Ϫ4 M) increased [Ca 2ϩ ] i , whereas 17-DMAG (10 Ϫ9 M) or RNAi treatment blocked the response to LA but not carbachol (Fig. 6). These data demonstrate that hsp90␤1 is required for PUFA-induced mitochondrial Ca 2ϩ efflux.

Spot
In previous studies, hsp90␤1 has been detected primarily in the endoplasmic reticulum (25). However, it has been shown recently that during ontogeny of sperm cells, hsp90␤1 is dynamically redistributed between the cell surface and cellular organelles, including mitochondria (26). Immunofluorescence analysis of NT2 cells double-labeled with MitoTracker for mitochondria and a monoclonal antibody against GRP94 for hsp90␤1 was performed. As shown in Fig. 7A, nonspecific staining in the absence of anti-GRP94 was not observed. Positive staining by anti-GRP94 (indicated by the green fluorescence) co-localized with MitoTracker (indicated in the overlay images), demonstrating the presence of hsp90␤1 in mitochondria. In NT2 cells treated for 48 h with 17-DMAG (10 Ϫ9 M) or hsp90␤1 RNAi, measurement of green fluorescence co-localizing with MitoTracker staining demonstrated significantly reduced levels of hsp90␤1 in mitochondria (Fig. 7, A  and B). Localization of hsp90␤1 to mitochondria of NT2 cells was also demonstrated by immunoblot analysis using anti-GRP94 (supplemental Fig. 2).
It has been reported that the N and C termini of hsp90␤1 play independent roles in substrate interaction and protein importation (22,27). We thus compared the acute effects of 17-DMAG and novobiocin, two inhibitors of the chaperone activity of hsp90␤1 that bind to the N and C terminus of the protein, respectively, on PIMCE in isolated mitochondria. 17-DMAG inhibits hsp90 chaperone activity with an I 50 of 5.1 ϫ 10 Ϫ8 M (28), whereas novobiocin typically inhibits hsp90 activity at concentrations of 2 ϫ 10 Ϫ4 to 10 Ϫ3 M (22). As shown in Fig. 8, the addition of 17-DMAG (3 ϫ 10 Ϫ6 M) to mitochondria immediately prior to LA had no effect on LA-induced mitochondrial Ca 2ϩ efflux (Fig. 8, A  and B). In contrast, the addition of novobiocin (2 ϫ 10 Ϫ3 M) to mitochondria inhibited the LA response (Fig. 8, compare C with  A). The rate of LA-induced mitochondrial Ca 2ϩ efflux was reduced 73.4% by novobiocin (p Ͻ 0.02, novobiocin versus control, n ϭ 4) The effect of novobiocin was concentration-dependent,  MARCH 21, 2008 • VOLUME 283 • NUMBER 12 with maximal inhibition observed at 2 ϫ 10 Ϫ3 M. These data suggest a critical role for the C terminus of hsp90␤1 in PIMCE.

DISCUSSION
In the present study, we have identified the essential role of hsp90␤1 in LA-induced mitochondrial Ca 2ϩ efflux. The data presented in this study demonstrate an important novel function for the well known chaperone protein hsp90␤1 (i.e. its involvement in PIMCE, a process by which elevated NEFA may lead to altered mitochondrial Ca 2ϩ homeostasis and mitochondrial dysfunction).
NEFA in animal and human plasma and tissues are composed of saturated FA, monounsaturated FA, and PUFA (29), and there is evidence that the deleterious effects of these different types of FA may be mediated by distinct cellular pathways. For example, a recent study indicates that saturated and monounsaturated FA, but not PUFA, are effective in increasing cellular levels of diacylglycerol and thus causing stimulation of protein kinase C (30). In our study, PIMCE was observed only with PUFA and not with saturated or monounsaturated FA (8). However, the effect of PUFA on mitochondrial Ca 2ϩ could potentially alter the metabolism of saturated and monounsaturated FA through changes in substrate oxidation.
Although the majority of studies examining the deleterious effects of FA have been performed using saturated FA, LA is a major component of NEFA, and its concentration is significantly elevated in diabetes (29). The PIMCE response to LA and other PUFA may account for a portion of the adverse effects of NEFA, especially at the mitochondrial level, whereas saturated and monounsaturated FA may exert their effects via other pathways.
Multiple Ca 2ϩ transport pathways responsible for mitochondrial Ca 2ϩ uptake and Ca 2ϩ efflux have been characterized functionally. The pathways linked to mitochon-  drial Ca 2ϩ uptake include the mitochondrial Ca 2ϩ uniporter (31)(32)(33) and the rapid Ca 2ϩ uptake mechanism detected in liver and heart mitochondria (34). The currently known pathways responsible for mitochondrial Ca 2ϩ efflux under physiological conditions include the mitochondrial Na ϩ /Ca 2ϩ exchanger, which is predominantly expressed in excitable tissues, such as neuron and muscle, and the H ϩ /Ca 2ϩ exchanger, which is mainly detected in nonexcitable tissues, such as liver and kidney (31,35). In addition to the specific Ca 2ϩ transport pathways described above, opening of the mitochondrial permeability transition pore during apoptosis leads to efflux of substances with molecular mass of Ͻ1.5 kDa, including Mg 2ϩ , Ca 2ϩ , cytochrome c, adenine nucleotides, and mitochondrial matrix proteins, with the resulting collapse of ⌬ and mitochondria swelling (36). Currently, the molecular identities of the mitochondrial Ca 2ϩ transport pathways have not been determined. Using two-dimensional gel electrophoresis and proteomic analysis, we show here that the molecular chaperone hsp90␤1 is reduced parallel to the loss of PIMCE during RA-induced NT2 cell differentiation (Figs. 2 and 3 and  Tables 1 and 2). The inhibition of PIMCE by 17-DMAG and hsp90␤1 RNAi further confirms that hsp90␤1 is required for PIMCE (Figs. 4 -6). The inhibitory effect of 17-DMAG, like that of hsp90␤1 RNAi, may result from down-regulation of hsp90␤1 protein levels, since prolonged treatment with other geldanamycin derivatives has been shown to cause degradation of hsp90␤1 proteins (37). The results of immunofluorescence experiments in the current work demonstrated that treatment with 17-DMAG and hsp90␤1 RNAi reduced protein levels of hsp90␤1 in NT2 cells (Fig. 7). Previous studies have revealed differential roles for the N or C terminus of hsp90␤ in interaction with client proteins (22). We thus compared the acute effect of 17-DMAG (which binds to the N terminus of hsp90␤) and novobiocin (which binds to the C terminus) on PIMCE and found that only novobiocin effectively inhibited LA-induced PIMCE (Fig. 8). The distinctive effects of inhibitors binding to the N or C terminus of hsp90␤ on PIMCE in isolated mitochon-dria suggest that the C terminus of hsp90␤1 plays a critical role in PIMCE. How hsp90␤1 functions as a molecular constituent of the mitochondrial Ca 2ϩ transporters activated by PUFA is currently unknown and is the subject of ongoing study in our laboratory.
The lack of PIMCE in brain mitochondria was unexpected, although the rate and amplitude of PIMCE in mitochondria were found to differ among other tissues tested (Fig. 1). Comparisons of component proteins and oxidative phosphorylation capacities indicate substantial diversities among the mitochondria of different rat tissues (muscle, heart, liver, kidney, and brain) (38,39). The lack of PIMCE may be a unique feature of brain mitochondria. If hsp90␤1 is a component of the pathway responsible for PUFA-induced mitochondrial Ca 2ϩ transport, we would expect brain mitochondria to lack or have low levels of hsp90␤1 proteins. Although the expression level of hsp90␤1 in brain mitochondria was not directly investigated in the present work, diminished hsp90␤1 levels detected upon RA-induced differentiation of NT2 neurons (Fig. 3) provided suggestive evidence that mitochondria of neural systems might contain less hsp90␤1 compared with other tissues, such as liver and spleen.
We found that the inhibitory effect of 17-DMAG added to NT2 cells for 48 h was concentration-dependent but biphasic (i.e. at concentrations of 17-DMAG Ն10 Ϫ8 M, the inhibitory effect on the LA-induced [Ca 2ϩ ] i signal was partially lost) (Fig.  4). The reduced inhibitory effect of 17-DMAG at high concentrations was associated with decreased NT2 cell numbers, which may occur through induction of cell death or arrested cell growth. The larger LA-induced [Ca 2ϩ ] i signal in cells treated with high concentrations (Ն10 Ϫ8 M) of 17-DMAG may reflect elevated [Ca 2ϩ ] m during apoptosis. We have previously shown that during apoptosis, G protein-coupled receptor-mediated [Ca 2ϩ ] i signaling is inhibited (40). Thus, the reduction of carbachol-induced [Ca 2ϩ ] i signal observed at high concentrations (Ն10 Ϫ8 M) of 17-DMAG may indicate that NT2 cells undergo apoptosis in the presence of high concentrations of 17-DMAG.
It has been shown that the hsp90 family is composed of multiple isoforms and is involved in diverse cellular processes, such as signal transduction, protein folding, and mitochondrial protein import (22,41). Our data demonstrate that hsp90␤1 plays an essential role in PIMCE.