Excessive Na+/H+ Exchange in Disruption of Dendritic Na+ and Ca2+ Homeostasis and Mitochondrial Dysfunction following in Vitro Ischemia*

Neuronal dendrites are vulnerable to injury under diverse pathological conditions. However, the underlying mechanisms for dendritic Na+ overload and the selective dendritic injury remain poorly understood. Our current study demonstrates that activation of NHE-1 (Na+/H+ exchanger isoform 1) in dendrites presents a major pathway for Na+ overload. Neuronal dendrites exhibited higher pHi regulation rates than soma as a result of a larger surface area/volume ratio. Following a 2-h oxygen glucose deprivation and a 1-h reoxygenation, NHE-1 activity was increased by ∼70–200% in dendrites. This elevation depended on activation of p90 ribosomal S6 kinase. Moreover, stimulation of NHE-1 caused dendritic Na+i accumulation, swelling, and a concurrent loss of Ca2+i homeostasis. The Ca2+i overload in dendrites preceded the changes in soma. Inhibition of NHE-1 or the reverse mode of Na+/Ca2+ exchange prevented these changes. Mitochondrial membrane potential in dendrites depolarized 40 min earlier than soma following oxygen glucose deprivation/reoxygenation. Blocking NHE-1 activity not only attenuated loss of dendritic mitochondrial membrane potential and mitochondrial Ca2+ homeostasis but also preserved dendritic membrane integrity. Taken together, our study demonstrates that NHE-1-mediated Na+ entry and subsequent Na+/Ca2+ exchange activation contribute to the selective dendritic vulnerability to in vitro ischemia.

Neuronal dendrites are vulnerable to injury under diverse pathological conditions, including cerebral ischemia, epilepsy, and Alzheimer disease (1,2). The hallmark of dendritic injury is the formation of focal swelling or beads along the length of the dendritic arbor (3). However, the underlying mechanisms for this selective dendritic injury remain poorly understood. The initial NMDA or kainite-mediated swelling in dendrites of cultured neurons depends on intracellular accumulation of Na ϩ and Cl Ϫ but not Ca 2ϩ (4). On the other hand, excessive Ca 2ϩ entry plays a role in the long lasting structural damage and delayed recovery in hippocampal slices in response to NMDA (4,5). A correlation between dendritic bead formation and ATP reduction/mitochondrial dysfunction has been demonstrated in cultured hippocampal neurons following glutamate exposure (6). However, the relationship between selective dendritic damage, loss of Na ϩ and Ca 2ϩ homeostasis, and mitochondrial dysfunction following ischemia remains to be defined.
NHE-1 (Na ϩ /H ϩ exchanger isoform 1) is a plasma membrane protein present in virtually all mammalian cells and plays a central role in intracellular pH (pH i ) and cell volume regulation (7). NHE-1 activity is directly activated by intracellular acidification and/or by protein phosphorylation mediated by ERK-p90 ribosomal S6 kinase (p90 RSK ) 2 in ischemic neurons (8). Excessive NHE-1 activation results in intracellular Na ϩ accumulation, which subsequently promotes Ca 2ϩ entry via reversal of Na ϩ /Ca 2ϩ exchange (NCX rev ) and plays an important role in myocardium ischemia/reperfusion injury (9). We recently reported that NHE-1 activity in the soma of neurons and astrocytes is stimulated following ischemia, and inhibition of NHE-1 activity is neuroprotective (8,10). In addition, inhibition of NHE-1 either pharmacologically or by genetic knockdown reduces infarction at 24 h following in vivo focal ischemia (11). However, it remains unexplored whether concurrent activation of NHE-1 and NCX rev contributes to the selective vulnerability of postsynaptic neuronal dendrites to ischemic damage.
In the current study, we demonstrated that neurons exhibited robust NHE-1-dependent pH i regulation in their dendrites as a result of their large surface area/volume ratio. Further, in vitro ischemia (oxygen glucose deprivation and reoxygenation, OGD/REOX) stimulated NHE-1 activity in large dendrites (Lgdendrites). NHE-1-mediated Na ϩ entry and subsequent stimulation of NCX rev activity contributed to selective ischemic damage of dendrites. The underlying mechanisms involved the loss of mitochondrial Ca 2ϩ homeostasis and mitochondrial membrane dysfunction.
Pure Cortical Neuron Cultures-Pure cortical neurons from embryonic day 14 -16 mouse fetuses (SV129/Black Swiss) were prepared as described previously (8). The cortices were removed from E14 -16 fetuses and treated with 0.5 mg/ml trypsin at 37°C for 25 min. The cells were centrifuged at 300 ϫ g for 4 min. The cell pellet was diluted in B-27 supplemented neurobasal medium (2%) containing 0.5 mM L-glutamine and penicillin/streptomycin (100 units/ml and 0.1 mg/ml, respectively). The cells were seeded at a density of 1 ϫ 10 5 cells/cm 2 on glass coverslips in 6-well plastic plates coated with poly-D-lysine. The cultures were maintained in an incubator (model 3130, Thermo Forma, Waltham, MA) with 5% CO 2 and atmospheric air at 37°C. Half of the medium was replaced twice a week. 10 -15day cultures were used in the study.
OGD Treatment-10 -15-day neuronal cultures grown on coverslips in 6-well plates were rinsed with an isotonic OGD solution (pH 7.4) containing 0 mM glucose, 21 mM NaHCO 3 , 120 mM NaCl, 5.36 mM KCl, 0.33 mM Na 2 HPO 4 , 0.44 mM KH 2 PO 4 , 1.27 mM CaCl 2 , and 0.81 mM MgSO 4 . This solution has a K ϩ concentration (ϳ5.8 mM) that is similar to that of the neurobasal medium (5.6 mM) used for cell cultures. The cells were incubated in 1 ml of OGD solution for 2 h in a hypoxic incubator (model 3130, Thermo Forma) containing 94% N 2 , 1% O 2 , and 5% CO 2 . Normoxic control cells were incubated for 2 h in 5% CO 2 and atmospheric air in a buffer identical to the OGD solution except for the addition of 5.5 mM glucose. REOX was achieved by the addition of glucose (5.5 mM) and incubation at 37°C in 5% CO 2 and atmospheric air. Alternately, REOX was performed on the microscope stage by superfusion with HCO 3 Ϫ -EMEM at 37°C, equilibrated with 5% CO 2 and ϳ18% O 2 (monitored by an in-line oxygen electrode, model 16-730; Microelectrodes, Bedford, NH). pH i Measurement-pH i measurement and prepulse treatment were performed as described previously with some modifications (8). Briefly, pure neuronal cultures grown on coverslips were incubated with 2.5-5 M BCECF/AM for 30 min during normoxia or during the last 30 min of REOX at 37°C.
The coverslips were washed with HCO 3 Ϫ -free HEPES-EMEM and placed in a temperature-controlled (37°C) open bath imaging chamber (model RC24, Warner Instruments, Hamden, CT). The chamber was mounted on the stage of the TE 300 inverted epifluorescence microscope, and 1-3 neurons were visualized with a ϫ100 oil immersion objective. The cells were excited every 10 -30 s at 440 and 490 nm, and the emission fluorescence at 535 nm was recorded. Images were collected using a Princeton Instruments MicroMax CCD camera and analyzed with MetaFluor image-processing software. Fluorescence changes in regions of interest in soma, Lg-dendrites, and small dendrites (Sm-dendrites) were determined. Lg-dendrites were defined as dendritic segments with a width of 5.3 Ϯ 1.2 m, whereas Sm-dendrites were ones with a width of 1.8 Ϯ 0.4 m. The ratio of the background-corrected fluorescence emissions (F490/F440) for each region was calibrated using the high K ϩ /nigericin technique (8). pH i values were calculated for soma, Lg-dendrites, and Sm-dendrites using the respective BCECF calibration values collected from each region.
For the prepulse treatment, cells were subjected to an acid load by a transient application (1.5 min) of a 30 mM NH 4 ϩ /NH 3 solution. NH 4 ϩ /NH 3 solutions were prepared by replacing 30 mM NaCl in the HEPES-buffered solution with an equimolar concentration of NH 4 Cl. pH i recovery rates were determined from the slope of a fitted linear regression within the first minute after NH 4 ϩ /NH 3 prepulse (8). To minimize differential allosteric effects of H ϩ on NHE-1 activity, pH i recovery rates were measured at pH i ϳ6.2 throughout the study. In the Na ϩfree experiments, NaCl in the HEPES-buffered solution was replaced with an equimolar concentration of NMDG. NMDGsubstituted Na ϩ -free solutions (ϳ5 min) do not cause cell swelling in acutely isolated CA1 neurons (12).
Determination of Intrinsic Buffer Power (␤ i )-␤ i was determined in somata, Lg-and Sm-dendrites over a range of pH i by subjecting the cells to progressively decreasing concentrations of NH 4 ϩ in Na ϩ -free HEPES-EMEM as described previously (8). The total H ϩ net efflux rate (J H ϩ, mM H ϩ /min) was determined in three neuronal regions by multiplying ␤ i by ⌬pH i /⌬t at pH i ϳ6.2. In some experiments, J H ϩ was also calculated in the presence of HCO 3 Ϫ . The buffering by CO 2 /HCO 3 Ϫ was determined as Ϫ ] i ϭ S ϫ PCO 2 ϫ 10 (pHi Ϫ pK) , where S ϭ 0.0314, PCO 2 ϭ 40 mm Hg, and pK ϭ 6.12. At pH 6.2, the contribution of ␤ HCO 3 Ϫ to total buffering in normoxic cells was ϳ6%.
Intracellular Na ϩ Measurement-Intracellular Na ϩ concentration ([Na ϩ ] i ) was measured with the fluorescent dye SBFI/AM as described previously with some modifications (14). Cultured neurons grown on coverslips were loaded with 30 M SBFI/AM plus 0.02% pluronic acid during a 45-min REOX following a 2-h OGD. The coverslips were placed in the open bath imaging chamber and superfused (1 ml/min) with HCO 3 Ϫ -EMEM at 37°C. Using the Nikon TE 300 inverted epifluorescence microscope and a ϫ100 oil immersion lens, neurons were excited at 345 and 385 nm, and the emission fluorescence at 510 nm was recorded. Regions of interest (1-3 cells/ area) were drawn to determine SBFI fluorescence changes in soma, Lg-dendrites, and Sm-dendrites. The 345/385 ratios were analyzed with the MetaFluor image-processing software.
Absolute [Na ϩ ] i was determined for each cell by performing an in situ calibration as described previously (14). Multiple time point data acquisition induced phototoxicity in neurons. Therefore, [Na ϩ ] i was only determined in normoxic controls or 45-min REOX-treated neurons.
Intracellular Ca 2ϩ Measurement-Neurons grown on coverslips were incubated with 5 M fura-2 AM during a 2-h OGD. Following OGD, the cells were placed in the open bath imaging chamber and superfused (1 ml/min) with HCO 3 Ϫ -EMEM at 37°C. Using the Nikon TE 300 inverted epifluorescence microscope and a ϫ100 oil immersion objective lens, neurons were excited every 5 min at 345 and 385 nm, and the emission fluorescence at 510 nm was recorded. Images were collected and analyzed with the MetaFluor image-processing software. At the end of each experiment, the cells were exposed to 1 mM MnCl 2 in Ca 2ϩ -free HCO 3 Ϫ -EMEM and 5 M 4-bromo-A-23187. The Ca 2ϩ -insensitive fluorescence was subtracted, and the MnCl 2corrected 345/385 emission ratios were converted to [Ca 2ϩ ] as described previously (14).
Measurement of Mitochondrial Ca 2ϩ -Neurons on coverslips were incubated at 37°C for 60 min with 200 nM Mito-Tracker Green and 9 M rhod-2/AM, which was reduced with a minimum of sodium borohydride in HCO 3 Ϫ -EMEM containing 3 mM sodium succinate (14). Coverslips were then incubated for 2 h under either OGD or normoxia conditions. For REOX, coverslips were placed in the perfusion chamber on the stage of the Leica DMIRE2 confocal microscope and superfused (1 ml/min) with HCO 3 Ϫ -EMEM at 37°C. Cells (1-3 in the field) were visualized with a ϫ100 oil immersion objective and scanned sequentially for MitoTracker Green (excitation 488 nm (argon laser line), emission 500 -545 nm) and rhod-2 (excitation 543 nm (HeNe laser), emission 544 -677 nm). The Mito-Tracker Green signal was used to maintain focus prior to each sequential scan. Sequential scans were analyzed using the Leica confocal software. Average grayscale values were collected from regions of interest around mitochondrial clusters exhibiting colocalization of MitoTracker Green and rhod-2. Ca 2ϩ m levels were expressed as relative change of rhod-2 signals from the base-line values, and summarized data represent the average of the calculated values from 2-3 cells as described previously (14).

Measurement of Mitochondrial Membrane Potential (⌿ m )-
The fluorescent probe JC-1 was used to monitor ⌿ m as described previously (14). Neurons on coverslips were loaded with 9 M JC-1 during 2 h of OGD at 37°C. Following OGD, the cells were placed in the temperature-controlled open bath imaging chamber and superfused (1 ml/min) with HCO 3 Ϫ -EMEM at 37°C. Cells were visualized using the Nikon TE 300 inverted epifluorescence microscope and a ϫ60 oil immersion objective. Cells were excited at 480 nm, and emission fluorescence images were recorded at 535 nm (the monomer) and 640 nm (JC-1 aggregates). The ratio of the aggregate to monomer fluorescence was measured in regions of interest in soma, Lgdendrites, and Sm-dendrites (2-5 cells/area). In this study, we applied 1.0 M FCCP for 1 min to determine the maximal loss of JC-1 signals. ⌿ m was expressed as the percentage of the maximal FCCP-induced change under normoxic controls (14). We believe that the FCCP-sensitive loss of JC-1 signals largely reflects changes in ⌿ m and is not affected by plasma membrane potentials. This is based on a study where 2.5 M FCCP caused the immediate collapse of ⌿ m and complete depolarization of plasma membrane potential. However, a low concentration of FCCP (0.25 M) had no effect on plasma membrane potential (15).
To further confirm that the changes in JC-1 reflect changes of ⌿ m , we also conducted some parallel experiments using the cationic membrane-permeant fluorescence probe TMRE. Neurons were loaded with 5 nM TMRE and 200 nM MitoTracker Green in a buffer supplemented with 1 M tetraphenylboron for 30 min 37°C. The coverslips were then placed in the perfusion chamber on the stage of the Leica DMIRE2 confocal microscope and superfused (1 ml/min) with HCO 3 Ϫ -EMEM at 37°C supplemented with 5 nM TMRE. Cells (1-3 in the field) were visualized with a ϫ100 oil immersion objective and scanned sequentially for MitoTracker Green (excitation 488 nm (argon laser line), emission 500 -545 nm) and TMRE (excitation 543 nm (HeNe laser), emission 544 -677 nm). Sequential scans were analyzed using the Leica confocal software. The MitoTracker Green signal was used to maintain focus prior to each sequential scan and to identify mitochondrial clusters exhibiting colocalization of MitoTracker Green and TMRE. Maximal ⌿ m dissipation was induced by FCCP (1.0 M) at the end of each experiment. At a concentration of 5 nM, TMRE behaves in the non-quench mode and decreases its fluorescence intensity when ⌿ m is reduced. Data are expressed as relative percent change in FCCP-sensitive TMRE signals.
Determination of Surface Area/Volume Ratio in Soma and Dendrites-To determine differences in the ratio of surface area to volume in soma and dendrites, neurons grown on coverslips were loaded with 0.5 M calcein/AM (cytosol dye) and 5 M SYTO 60 (nucleus dye) for 30 min at 37°C. The coverslips were then placed in the perfusion chamber on the stage of a Leica DMIRE2 confocal microscope and visualized with a ϫ100 oil immersion objective. A 110-m-thick image stack (300 slices at 512 ϫ 512 pixels) was collected sequentially (excitation 488 nm (argon laser line), emission 500 -545 nm; excitation 543 nm (HeNe laser), emission 544 -677) and imported into ImageJ (version 1.41, National Institutes of Health). A cellular region (soma, nucleus, and Lg-or Sm-dendrites) was defined, and the surface area was calculated by summing the product of the region perimeter with the distance between each image section (0.38 m). The volume of the region was calculated with the region area and section distance. The soma volume was corrected by subtracting the calculated volume for the nucleus. No attempt was made to correct for the intracellular volumes of endoplasmic reticulum or mitochondria.
Detection of Dendritic Beading Formation (Varicosities)-To monitor dendritic beading formation, neurons grown on coverslips were loaded with the plasma membrane dye Vybrant DiO as per the manufacturer's instructions. Following OGD, the coverslips were placed in the open bath imaging chamber and superfused (1 ml/min) with HCO 3 Ϫ -EMEM at 37°C on the stage of a Leica DMIRE2 confocal microscope. A single neuron was visualized with a ϫ100 oil immersion objective and scanned (512 ϫ 512, 200 Hz) with an argon laser (excitation 488 nm, emission 500 -545 nm). The images were analyzed for den-drite beading with ImageJ analysis software. Beads with a diameter ϳ4 times larger than the width of the corresponding dendrite were counted in a 90 ϫ 90-m area. Data represent the average of the calculated values from three or four experiments.
Immunofluorescence Staining-Cells grown on coverslips were fixed in 4% paraformaldehyde in PBS for 15 min. After rinsing, cells were incubated with a blocking solution for 20 min followed by application of a primary polyclonal antibody for NHE-1 (1:50; Abcam Inc., Cambridge, MA). After rinsing in PBS, cells were incubated with Alexa Fluor TM 488 goat antirabbit IgG (1:200; Invitrogen) for 1 h. The coverslips were then covered with Vectashield mounting medium (Vector Laboratories, Burlingame, CA). Fluorescence images were captured by the Nikon TE 300 inverted epifluorescence microscope (ϫ40) using a Princeton Instruments MicroMax CCD camera and MetaMorph image-processing software.
Statistics-Statistical significance was determined by Student's t test or an analysis of variance (Bonferroni post hoc test) in the case of multiple comparisons. A p value smaller than 0.05 was considered statistically significant. n values represent the number of cultures in each experiment.

Surface Area/Volume (A/V) Ratio in Soma and Dendrites-
In order to accurately calculate ionic flux rates in soma and dendrites, we first estimated surface area to volume (A/V) ratios in these cellular regions. Fig. 1A shows a single slice twodimensional image of cultured neurons from a confocal stack image (300 slices, 110 m thick). The arrows in Fig. 1A illustrate the areas in the somata, Lg-dendrite, and Sm-dendrite where the A/V ratios and ionic changes were determined. Fig. 1B is a three-dimensional reconstruction of the stack of images with Metamorph software, highlighting neuronal morphology with the distinctly higher A/V ratios in Lg-and Sm-dendrites. The A/V ratio in Lg-dendrites was 3.8 times larger than somata (Fig.  1C). An ϳ7 times larger A/V ratio was estimated for Sm-dendrites. Interestingly, 2 h of OGD and 1 h of REOX did not significantly change the A/V ratio either in soma or in dendrites.
We then determined ␤ i in three regions as shown in Fig. 1D. ␤ i in each region was plotted against pH i and fit with a linear regression. The slopes of the lines in the three regions were not significantly different under normoxic control or OGD/REOX conditions. These findings imply that the changes of pH i regulation may result from altered function of H ϩ transporters, such as NHEs.
Changes of pH i in Soma and Dendrites following OGD/REOX-Lg-or Sm-dendrites exhibited more alkaline resting pH i values than soma under normoxic conditions ( Fig. 2A). Inhibition of NHE-1 with its potent inhibitor HOE 642 (1 M) or the newly developed NHE-1 kinase p90 RSK inhibitor fluoromethylketone (FMK; IC 50 of 15 nM, (17)) acidified pH i and significantly decreased pH i recovery rates (Fig. 2, A and B). OGD/REOX caused an alkalization of pH i in soma (a shift from 6.96 Ϯ 0.03 to 7.19 Ϯ 0.05; p Ͻ 0.05). Inhibition of NHE-1 reversed the OGD/REOX-mediated increase in pH i (Fig. 2C). Moreover, inhibition of the NHE-1 kinase p90 RSK with FMK prevented the post-OGD alkalization. OGD/REOX did not trigger additional changes in pH i in dendrites. However, either HOE 642 or FMK significantly acidified dendrites following OGD/REOX. These data suggest that NHE-1 activation plays a role in resting pH i maintenance and contributes to the intracellular post-OGD alkalization.
Increased H ϩ Efflux in Soma and Dendrites following OGD/ REOX-We further determined NHE-1 activity in soma and dendrites by measuring the pH i recovery rate following the NH 3 /NH 4 ϩ prepulse-induced acidification. As shown in Fig. 3A, when neurons were exposed to 30 mM NH 3 /NH 4 ϩ , pH i in Lgdendrites rose rapidly as NH 3 diffused into the cell and combined with H ϩ to form NH 4 ϩ (a and b) and then declined slowly (b and c). Returning cells to the standard HCO 3 Ϫ -free HEPES-EMEM solution caused pH i to decrease due to the rapid diffusion of NH 3 , which was dissociated from the newly formed NH 4 ϩ , and trapping H ϩ inside the cells (c and d). Both normoxic control and OGD/REOX-treated cells were able to restore pH i to their basal levels (Fig. 3A). However, the pH i recovery rate increased by ϳ2-fold in Lg-dendrites following OGD/REOX (1.23 Ϯ 0.24 unit/min versus 0.57 Ϯ 0.05 unit/min in normoxic neurons, p Ͻ 0.05).
pH i recovery rates were significantly higher in the Lg-dendrites (90%) and in the Sm-dendrites (330%) than the soma under normoxic conditions (Fig. 3B). The apparent higher pH i recovery rates in the dendrites could result from the larger A/V ratios in the dendrites. Thus, we corrected the pH i recovery rate for A/V ratio in the three different regions. After the correction, the rates were similar in all three regions under normoxic control conditions (Fig. 3C). 2-h OGD/1-h REOX triggered a further increase in the H ϩ efflux in the soma (264%), the Lg-dendrites (218%), and the Sm-dendrites (69%; Fig. 3B). After the correction for the A/V ratio, the OGD/REOX-induced elevation in pH i recovery rates remained significant in soma and Lg-dendrites (Fig. 3C).
This finding was further validated by calculating J H ϩ (Fig.  3D). Dendrites exhibited smaller J H ϩ than soma under normoxia and OGD/REOX conditions. The OGD/REOX-mediated selective stimulation of J H ϩ persisted in soma and Lg-dendrites but not in Sm-dendrites. Similar changes of J H ϩ were observed in the presence of HCO 3 Ϫ (21 mM; Fig. 3D). The lack of changes in the A/V ratios and ␤ i following OGD/REOX suggest that the OGD/REOX-induced stimulation of pH i recovery rates mainly reflect J H ϩ.
Differential NHE-1 Activity in Soma and Dendrites-We directly evaluated NHE-1-dependent pH i regulation activity in the soma and the dendrites using the NHE-1 inhibitor HOE 642 (1 M, IC 50 of 0.08 M) at a concentration that inhibits only the NHE-1 isoform (18). As shown in Fig. 4A, the OGD/REOXmediated elevation of the H ϩ extrusion rate in the somata was nearly abolished in the presence of HOE 642. However, HOE 642 only partially blocked the elevated pH i recovery rate in the dendrites (ϳ50 -60%). To determine the possible role of other isoforms of NHE in neuronal processes, we examined the effects of removing extracellular Na ϩ , which inhibits the function of all NHE isoforms by abolishing the inward Na ϩ driving force. In the absence of extracellular Na ϩ , H ϩ extrusion was absent in soma, similar to the NHE-1 inhibition via HOE 642 (Fig. 4A). In the dendrites, the pH i recovery rate was eliminated by ϳ76 -86%. Inhibiting all NHE isoforms with a general NHE inhibitor EIPA (100 M) had a similar effect as removing extracellular Na ϩ . Moreover, the residual Na ϩ -independent H ϩ extrusion in the Sm-dendrites could be mediated by vacuolar    (Fig. 4A). These data imply that NHE-1 is the dominant isoform in soma. However, in the dendrites, pH i regulation is governed by NHE-1 as well as other NHE isoforms and H ϩ -ATPases.
The HOE 642-sensitive portion of the H ϩ extrusion rate was obtained under normoxic control and OGD/REOX conditions (Fig. 4B). Consistently, NHE-1 activity (HOE 642-sensitive portion) in the soma and the Lg-dendrites was significantly elevated following OGD/REOX. Sm-dendrites exhibited a significantly higher basal activity of NHE-1, and OGD/REOX did not cause additional activation. Fig. 4C demonstrates that NHE-1 is the dominant form in cortical neurons, whereas the NHE-3 isoform is restricted to the cerebellum (19). Localization of NHE-1 protein in soma and Lg-and Sm-dendrites was also shown in Fig. 4D.
NHE-1-mediated Na ϩ Entry in Soma and Dendrites following OGD/REOX-The robust NHE-1 activation in the soma and the dendrites following REOX led us to speculate that NHE-1 plays a role in the dendritic Na ϩ i dysregulation following in vitro ischemia. [Na ϩ ] i in the soma and the dendrites was monitored under normoxic controls and at 45 min REOX. There were no significant differences in the base-line [Na ϩ ] i between the soma and the dendrites (Fig. 5A, arrowhead). A 45-min REOX following 2 h OGD led to an increase in [Na ϩ ] i through-out the neuron. Localized increases in [Na ϩ ] i were detected in the dendrites (Fig. 5A, arrow). Summary data show that [Na ϩ ] i increased from a resting level of 12.5 Ϯ 0.3 to 44.4 Ϯ 2.6 mM in the soma (p Ͻ 0.05; Fig. 5B). The Sm-dendrites exhibited the largest Na ϩ accumulation (56.5 Ϯ 4.1 mM, p Ͻ 0.05). Inhibition of NHE-1 with HOE 642 during REOX abolished the OGD/REOXinduced Na ϩ overload in both the soma and the dendrites. The rise in HOE 642-sensitive changes in [Na ϩ ] i was shown in Fig. 5C. These data imply that OGD/REOX-mediated accumulation of [Na ϩ ] i is largely mediated via NHE-1 activation.
We recently reported that stimulation of NHE-1 depends on activation of the ERK-p90 RSK signal transduction pathways and phosphorylation of NHE-1 (8). In the current study, we examined whether direct inhibition of the NHE-1 kinase p90 RSK with its potent inhibitor BI-D1870 (BI-D, IC 50 of 10 -30 nM (20)) could reduce OGD/REOX-mediated NHE-1 activation. The OGD/REOX-induced Na ϩ i loading in Lg-and Sm-dendrites was abolished by BI-D1870 (Fig. 5B). This effect is similar to the one mediated by the NHE-1 inhibitor HOE 642. Thus, p90 RSK function is largely responsible for NHE-1 activation and Na ϩ entry. These data also imply that the initial increased intracellular acidification associated with OGD (8) is not sufficient to drive NHE-1 activity but requires altered phosphorylation of the transporter.
We further compared the effects of BI-D along with FMK on inhibition of NHE-1 activation. Both BI-D and FMK abolished the OGD/REOX-mediated stimulation of pH i recovery in the soma and the dendrites (Fig. 5D). Especially in the Sm-dendrites, BI-D or FMK profoundly suppressed the pH i recovery rate, which was only ϳ25% of the normoxic basal levels (p Ͻ 0.05; Fig. 5D). These data suggest that p90 RSK pathways play a more dominant role in pH i regulation in the dendrites than in the soma.
Last, the pH i recovery rates in all three regions were determined in the presence of 21 mM bicarbonate under normoxia and OGD/REOX conditions (Fig. 5E). The results are similar to those in the absence of physiological bicarbonate. These data suggest that the role for bicarbonate-dependent ion transporters in regulation of pH i is negligible under these conditions.
To further test NHE-1-mediated Na ϩ overload, we examined whether there was a differential Na ϩ overload in soma and Sm-dendrites when Na ϩ /K ϩ -ATPase was blocked by ouabain (0.1 mM) during REOX. As shown in supplemental Fig. 1A,  NOVEMBER 5, 2010 • VOLUME 285 • NUMBER 45

JOURNAL OF BIOLOGICAL CHEMISTRY 35161
Na ϩ i clearly loaded faster in Sm-dendrites than in soma under these conditions. Moreover, at 30 min of REOX, blocking NHE-1 activity decreased the Na ϩ influx in Sm-dendrites (supplemental Fig. 1B). A similar trend was also observed in Lgdendrites (supplemental Table 1).
Delayed Dysregulation of Ca 2ϩ Depends on a Concurrent Activation of NHE-1 and NCX rev -One consequence of Na ϩ i overload is to trigger NCX rev and Ca 2ϩ entry. To investigate the possible concerted activation of NHE-1 and NCX rev , we first monitored changes of local [Ca 2ϩ ] i following REOX. The somata [Ca 2ϩ ] i was 68 Ϯ 9 nM under normoxic control conditions and increased modestly to 104 Ϯ 8 nM after 2 h of OGD (p Ͻ 0.05), which remained unchanged over the initial 35-min REOX (Fig. 6, A and B). In contrast, 2 h of OGD caused a slightly higher elevation in the dendritic [Ca 2ϩ ] i (157 Ϯ 21 nM, p Ͻ 0.05). Twenty min of REOX triggered a secondary rise in the dendritic [Ca 2ϩ ] i that initiated from local "hot spots" and then spread toward the soma over time (Fig. 6, A and C). The amplitude of the dendritic Ca 2ϩ dysregulation was significantly higher than the soma. By 45 min of REOX, the dendritic [Ca 2ϩ ] i rose dramatically to 1206 Ϯ 440 nM and spread to the soma, which exhibited slightly lower levels (766 Ϯ 300 nM). Sustained elevation in [Ca 2ϩ ] i during REOX (60 -100 min) resulted in cell death, as reflected by a sudden loss of the dye (data not shown).
We speculated that NCX rev contributes to this Ca 2ϩ dysregulation as a result of the robust NHE-1 activation and Na ϩ overload. First, we investigated whether inhibition of NHE-1 activity with HOE 642 could block the delayed rise in [Ca 2ϩ ] i . As shown in Fig. 6 (B and C), when HOE 642 was present only during 60 min of REOX, no secondary rise in [Ca 2ϩ ] i occurred in the soma and the dendrites following REOX. These data imply that NHE-1 activation is involved in the secondary loss of Ca 2ϩ i homeostasis during REOX. To establish whether this Ca 2ϩ rise results from activation of NCX rev , we conducted the experiments in the presence of SEA0400 (1 M), a potent inhibitor of NCX rev . As shown in Fig. 6, B and C, REOX failed to elicit the secondary elevation in [Ca 2ϩ ] i in both the soma and the dendrites. The results were similar to those of the HOE 642treated cells. This led us to conclude that concerted activation of NHE-1 and NCX rev contributed to the delayed Ca 2ϩ dysregulation in soma and dendrites following in vitro ischemia. However, these inhibitors did not affect basal levels of [Na ϩ ] i and [Ca 2ϩ ] i under normoxic control conditions (supplemental Table 2).
Changes of Ca 2ϩ m and ⌿ m in Soma and Dendrites following OGD/REOX-Excessive dendritic Na ϩ i and Ca 2ϩ i overload will affect mitochondrial Ca 2ϩ homeostasis and mitochondrial function. To investigate this, we first monitored changes of Ca 2ϩ m in the soma and the dendrites during 0 -60 min REOX. There was a slow and progressive elevation in the somata Ca 2ϩ m starting by 10 min REOX and reaching a plateau value (ϳ2.5-fold control) by 40 min REOX (Fig. 7A). Interestingly, the somata ⌿ m did not decrease significantly, whereas Ca 2ϩ m was increasing during early REOX. However, ⌿ m depolarized significantly after Ca 2ϩ m reached its plateau levels, and it was reduced to 47 Ϯ 6% of control at 60 min of REOX (Fig. 7A).
By 10 min of REOX, Ca 2ϩ m levels had increased significantly in the Sm-dendrites but not in the Lg-dendrites (Fig. 7, B and   C). The rate of Ca 2ϩ m increase in the Sm-dendrites during early REOX was significantly faster than in the soma (0.018 versus 0.003 relative change/min, p Ͻ 0.05). At 60 min REOX, Ca 2ϩ m was increased by ϳ4.5-fold in the Sm-dendrites and ϳ3.2-fold in the Lg-dendrites. Ca 2ϩ accumulation in the mitochondria remained elevated throughout the neuron until ϳ100 min, when a sudden loss in the rhod-2 dye signal occurred, probably due to a collapse of mitochondrial function (data not shown).
The earlier rise in Ca 2ϩ m in the dendrites was accompanied by a faster decrease of ⌿ m (Fig. 7, B and C). ⌿ m in the dendrites (particularly in the Sm-dendrites) depolarized at a rate twice that of the somata (1.2 versus 0.7%/min). ⌿ m in the Sm-dendrites dropped to the lowest level (17 Ϯ 4% of control) by 40 min of REOX. The kinetics of the ⌿ m collapse in the dendrites exhibited a significant negative correlation with the dendritic Ca 2ϩ m accumulation (Pearson product moment correlation coefficient ϭ Ϫ0.964, p Ͻ 0.001). Thus, compared with the soma, the dendrites show two characteristics: earlier onset time and larger magnitude in the loss of mitochondrial Ca 2ϩ homeostasis and ⌿ m . This demonstrates that the dendritic mitochondria are more sensitive to OGD/REOX damage than the soma. These changes are consistent with the earlier loss of Na ϩ and Ca 2ϩ homeostasis in the dendrites. Interestingly, inhibition of NHE-1 activity with 1 M HOE 642 prevented the REOX-mediated changes of Ca 2ϩ m and ⌿ m in soma. In the presence of 1 M HOE 642, there were no significant increases in Ca 2ϩ m in the Lg-dendrites and the somata. A slow accumulation in Ca 2ϩ m was detected in the Sm-dendrites, which was not statistically significant from 0 min REOX. Moreover, the delayed depolarization of the somata ⌿ m during 50 -60 min of REOX was absent with the HOE 642 treatment (Fig. 7A). Strong protective effects of HOE 642 on ⌿ m were also found in the Lg-and the Sm-dendrites. The small dendritic ⌿ m depolarized to ϳ44% of control (instead of 17% of control) at 60 min of REOX when NHE-1 activity was inhibited (p Ͻ 0.05). Thus, preservation of ⌿ m by NHE-1 inhibition may result from decreased mitochondrial Ca 2ϩ loading.
To determine the role of NHE-1 in mitochondrial dysfunction following OGD/REOX, we examined whether FMK (3 M) would prevent mitochondrial damage in the soma and the dendrites. As shown in Fig. 8, A and B, inhibition of the NHE kinase p90 RSK during REOX attenuated loss of ⌿ m following OGD/ REOX in the soma and the Lg-dendrites. In the Sm-dendrites, FMK effectively prevented depolarization of ⌿ m as early as 10 min of REOX (Fig. 8C). Interestingly, these effects were similar to the direct inhibition of NHE-1 mediated by HOE 642. Taken together, we can firmly conclude that blocking either NHE-1 or p90 RSK significantly preserves mitochondrial function in ischemic neurons.
In a parallel study, in order to confirm the reliability of the ⌿ m determination with JC-1, we also used the cationic dye TMRE to monitor changes of ⌿ m at 0 and 60 min of REOX (supplemental Fig. 2, A and B). Both JC-1 and TMRE measurements indicated that the REOX-induced decrease in ⌿ m was more profound in the Lg-and the Sm-dendrites. Additionally, inhibition of NHE-1 with HOE 642 during REOX significantly attenuated the OGD/REOX-induced decrease in ⌿ m in all areas of the cells (p Ͻ 0.05; supplemental Fig. 2B).
Role of Mitochondrial Uniporter in Mitochondrial Ca 2ϩ Accumulation-To investigate the role of the uniporter in mitochondrial Ca 2ϩ loading following OGD/REOX, we determined changes of Ca 2ϩ m when the mitochondrial uniporter was inhibited with 10 M RU360 during 60 min of REOX. Inhibition of the uniporter prevented accumulation of Ca 2ϩ m in the mitochondria in all regions of the neuron (Fig. 9).
Dendritic Damage Is Reduced by Inhibition of NHE-1 following OGD/REOX-We further investigated dendritic damage by monitoring the dendritic beading (varicosities) and membrane integrity changes following OGD/REOX. As shown in Fig. 10, A and B, 2 h of OGD caused some swelling in the dendrites without formations of varicosities (arrow). Varicosities developed in the dendrites over 30 -60 min REOX (Fig. 10C, arrowhead). At 60 min of REOX, the bead density increased by Ͼ13 times compared with the normoxic neurons (Fig. 10, C and E). In contrast, in the presence of 1 M HOE 642 during REOX, the bead formation increased only by ϳ3 times (Fig. 10, D and E, arrow; p Ͻ 0.05). These data suggest that inhibition of NHE-1 activity with HOE 642 not only reduces the dendritic ionic dysregulation but also decreases dendritic swelling following OGD/REOX.

DISCUSSION
Robust NHE-1 Activity in Dendrites-In the current study, we characterized NHE-1 activity in the somata and the Lg-and the Sm-dendrites under normoxic and OGD/REOX conditions. We observed that pH i regulation in the Lg-and the Smdendrites was ϳ90 -300% faster than in the somata under normoxic conditions. The differential pH i regulation rates between the dendrites and the soma were abolished when they were corrected by the differences in the A/V ratios. Therefore, the data illustrate that the dendrites can change pH i more rapidly than the soma due to the small cytosolic volume compared with its surface area. Robust NHE activity has previously been detected in the hippocampal nerve terminals following intracellular acidification under normoxic conditions (21). In our  study, following OGD/REOX, the Na ϩ -dependent H ϩ extrusion activity was further elevated in the soma (264%) and the Lg-dendrites (218%), whereas the A/V ratios remained unchanged.
The H ϩ extrusion mechanisms in the soma and the dendrites have not been well defined. In this study, we concluded that the somata pH i regulation under HCO 3 Ϫ -free conditions is exclusively mediated by NHE-1 activity, which is consistent with our previous findings using both HOE 642-mediated pharmacological inhibition and NHE-1 genetic knockdown approaches (11). Moreover, the significance of NHE-1 in neuronal ionic regulation is further highlighted by the abundant expression of NHE-1 compared with NHE-2, NHE-3, or NHE-5 in the neurons. This is consistent with the earlier reports on the preeminence of both NHE-1 mRNA and protein expression in brains over the isoforms NHE-2 to -4 and the abundance of NHE-3 in the cerebellum (19,22).
We report here that NHE-1 plays a dominant role in the regulation of dendritic pH i (60%). The remaining pH i regulation in the dendrites depended on the functions of the less abundant NHE isoforms (NHE-2 and NHE-5) and vacuolar H ϩ -ATPases. The H ϩ pumps are highly expressed in the vesi-cles of synaptic terminals and responsible for acid loading and accumulation of neurotransmitters (23). Although the H ϩ pumps are typically expressed in membranes of organelles, they have been detected in the plasma membrane of hippocampal astrocytes and are active in pH i regulation under Na ϩ -and HCO 3 Ϫ -free conditions (24,25). Our findings of H ϩ pump activity in the dendritic plasma membrane suggest that dendrites are equipped with multiple H ϩ extrusion mechanisms to counteract the robust Ca 2ϩ -dependent intracellular acidification during synchronous neural activity (26).
Recently, spatial nonuniformity in pH i has been reported in the proximal and distal dendrites of oligodendrocytes (27). The alkaline microdomains in the perikaryon and proximal dendrites of the oligodendrocytes were attributed to localized increases in NHE activity, whereas the acidic pH i in the distal dendrites may be the result of Na ϩ /HCO 3 Ϫ cotransporter-mediated HCO 3 Ϫ extrusion (27). Thus, the pH i microdomain and regulatory mechanisms in the oligodendrocyte distal dendrites appear to be different from those in the Sm-dendrites of neurons. These findings suggest that different cell types express different pH i -regulating mechanisms in regulating microdomain pH i .
Lack of NHE-1 Activation in Sm-dendrites following OGD/ REOX-The basal level of NHE-1-mediated H ϩ extrusion was high in the Sm-dendrites compared with soma. OGD/REOX did not further stimulate it. This suggests that, given their large A/V ratio and their higher basal J H ϩ , Sm-dendrites are able to maintain pH i without significant further elevation of NHE activity. This may also be a reflection of the sophisticated pH i regulatory mechanisms expressed in Sm-dendrites, preventing overstimulation of H ϩ extrusion. On the other hand, this Smdendritic J H ϩ phenomenon may be unique under the circumstances of culture models in plastic plates but not characteristic of dendrites under in vivo conditions. Indeed, we have observed much higher A/V ratios and faster pH i recovery rates in neurites grown in microfluidic devices, a model that mimics the slow diffusion and convection of the in vivo microenvironment (28). It remains to be investigated whether OGD/REOX affects NHE-1 function differently in the Sm-dendrites using the microfluidic device model. p90 RSK -mediated Stimulation of NHE-1 Activity following OGD/REOX-In the current study, OGD/REOX-mediated stimulation of NHE-1 activity was abolished in both the soma and the dendrites when NHE-1 kinase p90 RSK was inhibited by its potent inhibitors BI-D1870 and FMK. Activation of p90 RSK and NHE-1 phosphorylation is downstream of the ERK1/2 signaling pathways (8). FMK is a novel, specific inhibitor for p90 RSK isoforms 1 and 2 (17). FMK blocks the ␣ 1 -adrenoceptermediated NHE-1 phosphorylation and stimulation in rat ventricular myocytes (29). On the other hand, BI-D1870 has been shown to be a potent ATP-competitive inhibitor of all p90 RSK isoforms (20). In this study, we found that FMK and BI-D1870 were equally effective in inhibiting OGD/REOX-mediated NHE-1 activation, implying a role for p90 RSK isoforms 1 and 2. Moreover, the p90 RSK inhibitors reduced the NHE-1 activity to below the base-line levels. This suggests that p90 RSK is also involved in the regulation of basal NHE-1 activity in all three regions.  NOVEMBER 5, 2010 • VOLUME 285 • NUMBER 45

JOURNAL OF BIOLOGICAL CHEMISTRY 35165
NHE-1-mediated Na ϩ Entry following OGD/REOX-Disruption of dendritic ionic homeostasis occurs during early cerebral ischemia and may play a role in irreversible dendritic damage. Excessive Na ϩ influx via ionotropic glutamate receptors or tetrodotoxin-sensitive Na ϩ channels leads to neuronal death under excitotoxic or hypoxic conditions (4,30,31). However, subsequent studies have suggested that hypoxia-induced Na ϩ influx could be through pathways other than ionotropic glutamate receptors or tetrodotoxin-sensitive channels (32). In cultured hippocampal neurons, Na ϩ entry immediately after anoxia results from activation of NHE and a Gd 3ϩ -sensitive pathway (33). Recent reports demonstrate that dendritic damage following brief in vivo ischemia (34) or axonal morphological changes following in vitro hypoxia (35) are independent of ionotropic glutamate receptor activation. In the present study, we observed that OGD/REOX triggered an ϳ3-fold increase in [Na ϩ ] i (ϳ50 mM). The Na ϩ accumulation was eliminated when NHE-1 activity was inhibited by its inhibitor HOE 642 or the p90 RSK kinase inhibitor BI-D1870. Thus, the elevated NHE activity in the dendrites not only accelerates pH i recovery after OGD/REOX but also intensifies disruption of Na ϩ ionic homeostasis and causes dendritic vulnerability to ischemic damage. These findings also suggest that Na ϩ /K ϩ -ATPase function is not sufficient to maintain Na ϩ i homeostasis following OGD/ REOX when there is an increase in Na ϩ influx. Blocking NHE-1 activity would decrease the need for Na ϩ extrusion via Na ϩ / K ϩ -ATPase and preserve cellular ATP levels (Fig. 11). This imbalance between Na ϩ extrusion via Na ϩ /K ϩ -ATPase and NHE-1-mediated Na ϩ influx can also have a significant impact on [Na ϩ ] i in Sm-dendrites, particularly because of their large A/V ratio and high basal J H ϩ, even without further elevation of NHE-1 activity following OGD/REOX. Taken together, we conclude that the OGD/REOX-mediated stimulation of NHE-1 plays a dominant role in dendritic Na ϩ overload.
We failed to detect elevation of NHE-1-mediated H ϩ extrusion in the Sm-dendrites after OGD/REOX. The causes for the discrepancy between the NHE-1-mediated H ϩ extrusion and Na ϩ overload in the Sm-dendrites are not apparent. One possible explanation is that a subtle increase in NHE-1 activity may not be detected with the instantaneous measurement of H ϩ extrusion (dpH i /dt), whereas its impact on Na ϩ overload over time (at a steady-state level) can be revealed. This speculation is also supported by HOE 642-sensitive effects on mitochondrial dysfunction and Ca 2ϩ dysregulation. Future study is needed to further address this issue.
NHE-1-dependent Changes in Dendritic Ca 2ϩ following OGD/REOX-Activation of NHE activity in hippocampal nerve terminals following intracellular acidification is accompanied with an elevation in [Na ϩ ] i and [Ca 2ϩ ] i as well as increased postsynaptic currents (21). The authors attribute these changes to the concurrent activation of NHE and NCX rev in the nerve terminals (21). These findings suggest that NHE and NCX rev could play a coordinating role in the regulation of dendritic Na ϩ and Ca 2ϩ homeostasis and affect Ca 2ϩ -dependent release of neurotransmitters. In dendrites and dendritic spines close to postsynaptic localities, all three isoforms of NCX (NCX-1 to -3) are preferentially expressed, suggesting a role for NCX in Ca 2ϩ signaling at the excitatory postsynaptic sites (36).
In this study, 2-h OGD triggered a moderate elevation in dendritic Ca 2ϩ i , but, during 60 min of REOX, a delayed accelerated Ca 2ϩ i rise occurred. The OGD/REOX-induced Ca 2ϩ deregulation was initiated in the dendrites and then propagated to the soma. Interestingly, the secondary Ca 2ϩ i deregulation can be prevented when either NHE-1 activity was inhibited by HOE 642 or NCX rev was blocked by SEA0400. These findings imply that a coupled NHE-1 and NCX rev function is a major contributor to Ca 2ϩ i deregulation in the dendrites of cultured cortical neurons. Moreover, we believe that NCX-1 is the dominant NCX isoform in this study because SEA0400 has a high affinity against the reverse mode function of NCX-1 (IC 50 ϳ50 nM) as compared with NCX-2 (IC 50 ϳ1 M) and is ineffective against either NCX-3 or NCKX-2 (37).
The role of NCX rev in dendritic Ca 2ϩ i deregulation has also been examined during sustained NMDA exposure. Delayed Ca 2ϩ deregulation in CA1 neurons of acute hippocampal slices depended on Na ϩ loading but was not prevented by the nonspecific NCX inhibitor KB-R7943 (38). However, when Na ϩ loading is potentiated with low levels of ouabain (30 M), NMDA can trigger secondary Ca 2ϩ deregulation in dendrites, which is completely blocked by KB-R7943 (39). These findings further suggest that activation of NCX rev requires excessive Na ϩ loading. Our previous thermodynamic analysis predicts that NCX rev occurs when [Na ϩ ] i is elevated to ϳ20 mM in cortical neurons at a resting membrane potential of Ϫ60 mV (40).
The vulnerability of neuronal dendrites is characterized by the initial membrane depolarization, mitochondrial structure collapse, and dendritic beading in the dendrites and the subsequent propagation toward the soma during hypoxia and activation of NMDA receptors (1,6). Our current study illustrates that blocking of NHE-1 activity attenuated many similar changes in the dendrites following OGD/REOX. Therefore, concerted activation of NHE-1 and NCX rev may also play a role FIGURE 11. Illustration of dendritic ionic disruption and mitochondrial dysregulation in ischemic neurons. Following ischemia, activation of NHE-1 causes an increase in dendritic [Na ϩ ] i , which triggers NCX rev and leads to increases in [Ca 2ϩ ] i . The [Na ϩ ] i overload also causes increased consumption of ATP by Na ϩ /K ϩ -ATPase to maintain dendritic ionic homeostasis. On the other hand, the [Ca 2ϩ ] i overload stimulates Ca 2ϩ m uptake by the uniporter (UP) and formation of the mitochondrial permeability transition pore (PTP). Blocking NHE-1 and NCX rev would reduce disruption of Na ϩ and Ca 2ϩ homeostasis and preserve mitochondrial bioenergetics and dendritic membrane integrity.
in dendritic injury in conditions such as glutamate-mediated neurotoxicity, epilepsy, etc.
Changes of Dendritic ⌿ m and Ca 2ϩ m following OGD/REOX-In the current study, depolarization of ⌿ m in the Sm-dendrites occurred 40 min earlier than the soma following OGD/REOX. The loss of ⌿ m in the dendrites closely correlated with the Ca 2ϩ m accumulation. Mitochondria are capable of sequestering large amounts of Ca 2ϩ under various pathological conditions (41). Increases in free mitochondrial Ca 2ϩ would occur when Ca 2ϩ entry into the mitochondria exceeds the capacity of mitochondrial Ca 2ϩ extrusion and the mitochondrial robust phosphate buffering system (42). The Ca 2ϩ m accumulation reported in this study probably reflects Ca 2ϩ entry via a voltagedependent Ca 2ϩ uniporter before the collapse of ⌿ m (41). Small increases in Ca 2ϩ m stimulate Ca 2ϩ -dependent dehydrogenases and mitochondrial metabolism, but massive Ca 2ϩ loading of mitochondria leads to depolarization of ⌿ m (6,42). Sustained loss of ⌿ m will eventually trigger the opening of the permeability transition pore and release of Ca 2ϩ m and collapse of mitochondria bioenergetics (6). In the current study, most of the Ca 2ϩ m accumulation occurred before ⌿ m decreased below 50% and remained at a sustained level when ⌿ m was ϳ45% of control in the soma and ϳ20% of control in the Sm-dendrites. This implies that a residual level of ⌿ m (20%) for a short period (ϳ40 min) is sufficient to maintain high mitochondrial Ca 2ϩ levels. However, when low ⌿ m and high Ca 2ϩ m were extended past 60 min, there was a sudden loss of residual ⌿ m that coincided with the loss of the rhod-2 signal, suggesting a release of Ca 2ϩ m from the permeability transition pore under these conditions. Interestingly, when either NHE-1, p90 RSK , or NCX rev was inhibited during 60-min REOX, loss of ⌿ m and Ca 2ϩ m accumulation in dendrites was significantly reduced. This finding is consistent with the earlier reports on NHE-1 inhibition-induced attenuation of the mitochondrial Ca 2ϩ overload and mitochondrial permeability transition pore opening in cardiomyocytes and in ischemic/reperfused rat hearts (43)(44)(45). Taken together, these studies demonstrate a conserved role of the NHE-1 signaling mechanism in ischemic reperfusion injury among multiple cell types. Moreover, it has been suggested that NHE-1 inhibitors, including HOE 642, have a direct effect on reactive oxygen species production and mitochondrial permeability transition pore formation (13). In the current study, it is unknown whether any protective effects mediated by HOE 642 result from its direct actions on mitochondria. We speculate that such a possibility is low in light of the similar protective effects offered by inhibition of p90 RSK or NCX rev in this study as well as the protection observed in NHE-1 transgenic knock-out neurons (11). Moreover, in general, mild acidosis can inhibit neurotransmission, whereas alkaline pH i stimulates excitability. Thus, we cannot rule out that HOE 642 may protect neurons in part via directly correcting NHE-1mediated alkalinization.
In summary (Fig. 11), the current study reports that NHE-1mediated Na ϩ entry and subsequent stimulation of NCX rev activity contribute to the selective vulnerability of dendrites following in vitro ischemia. A newly emerging hypothesis speculates that dendritic Na ϩ overload and the subsequent activation of Na ϩ /K ϩ -ATPase would consume more ATP and fur-ther collapse mitochondrial biogenesis (6). However, to date, the mechanisms underlying the excessive Na ϩ influx and mitochondrial dysfunction are not well defined. Our current study demonstrates that activation of NHE-1 in dendrites presents a major pathway for Na ϩ overload. NHE-1 inhibition prevents Na ϩ accumulation, which is required for dendritic beading. Blocking NHE-1 function also attenuates loss of the dendritic ⌿ m and Ca 2ϩ m homeostasis and preserves mitochondrial bioenergetics and dendritic membrane integrity.