Protein Kinase C-dependent Enhancement of Activity of Rat Brain NCKX2 Heterologously Expressed in HEK293 Cells*

Different members of the Na+/Ca2++K+ exchanger (NCKX) family are present in distinct brain regions, suggesting that they may have cell-specific functions. Many neuronal channels and transporters are regulated via phosphorylation. Regulation of the rat brain NCKXs by protein kinases, however, has not been described. Here, we report an increase in NCKX2 activity in response to protein kinase C (PKC) activation. Outward current of NCKX2 heterologously expressed in HEK293 cells was enhanced by β-phorbol dibutyrate (PDBu), whereas PDBu had little effect on activity of NCKX3 or NCKX4. The PDBu-induced enhancement (PIE) of NCKX2 activity was abolished by PKC inhibitors and significantly reduced when the dominant negative mutant of PKCϵ (K437R) was overexpressed. Moreover, PDBu accelerated the decay rate of the Ca2+ transient at the calyx of Held, where NCKX is the major Ca2+-clearance mechanism. Intracellular perfusion with alkaline phosphatase completely inhibited PIE. Consistently, β-phorbol myristate acetate (PMA), but not 4α-PMA, induced a 3-fold stimulation of 32P incorporation into NCKX2 expressed in HEK293 cells. To investigate the sites involved, PIE of wild-type NCKX2 was compared with mutant NCKX2 in which the three putative PKC consensus sites were replaced with alanine, either individually or in combination. Double-site mutation involving Thr-476 (T166A/T476A and T476A/S504A) disrupted PIE, whereas single mutation of Thr-166, Thr-476, or Ser-504 or the double mutant T166A/S504A failed to completely prevent PIE. These findings suggest that PKC-mediated activation of NCKX2 is sensitive to mutation of multiple PKC consensus sites via a mechanism that may involve several phosphorylation events.

Previously, we have shown that NCKX plays a major role in Ca 2ϩ extrusion at the fast glutamatergic presynaptic terminal, calyx of Held (11). It is well known that phorbol esters enhance synaptic transmission by a presynaptic mechanism, which includes an enhancement of Ca 2ϩ sensitivity of the molecular machinery that mediates synaptic vesicle fusion and an increase in the size of the readily releasable pool (23). The effect of phorbol esters on the presynaptic Ca 2ϩ dynamics, however, still remains to be elucidated. Given that Ca 2ϩ plays a key role in the activity-dependent changes of synaptic strength as well as in release of neurotransmitters, PKC modulation of NCKX would be expected to have substantial influence on presynaptic Ca 2ϩ dynamics, and in turn on short-term synaptic plasticity.
In the present study, we have analyzed the regulation of neuronal NCKX family members by PKC. We found that PKC activation phosphorylated and activated NCKX2 expressed in HEK293 cells, whereas the activity of NCKX3 and NCKX4 was not affected. By using pharmacological intervention and mutagenesis of potential PKC phosphorylation sites, we have shown that Thr-476 as well as Thr-166 and Ser-504 are important residues involved in PKC-mediated activation of NCKX2.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-Human embryonic kidney (HEK293) cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, and 1% penicillin/streptomycin in 5% CO 2 at 37°C. The cells were cotransfected with rat brain NCKX2 cDNA and DS-Red2 cDNA using a standard calcium-phosphate precipitation protocol with HBS buffer (140 mM NaCl, 0.75 mM Na 2 HPO 4 , 25 mM HEPES, pH 7.0). The cells were used for electrophysiological recordings 36 -48 h after the co-transfection.
Electrophysiology-Electrophysiological recordings were performed using the conventional whole cell configuration of the patch clamp recording technique. Patch pipettes with a resistance of 4 -5 M⍀ were prepared from borosilicate glass capillaries, which were filled with a pipette solution (see below). Voltage clamp experiments were conducted with an EPC-8 patch clamp amplifier (HEKA Elektronik, Lambrecht, Germany). The holding potential was set to 0 mV. The membrane current was filtered at 500 Hz by a built-in low-pass filter, sampled at 1 kHz, and stored on a PC. The raw current recordings were low-pass filtered off-line using a boxcar smoothing algorithm with a smoothing factor of 5. The capacitive component of the current was minimized by the cancellation circuit built in the voltage clamp amplifier. Voltage ramp pulses were applied before and after evoking NCKX2 current using perfusion solution changes (see below). The current-voltage (I-V) relationship was obtained from the current response to a ramp pulse of amplitude from Ϫ80 mV to ϩ80 mV over 200 ms. For clarity, in Fig. 7B and supplemental Fig. S2B, slow I NCKX2 traces are presented after removing artifacts caused by voltage ramp pulses. All experiments were performed at room temperature (24 Ϯ 1°C).
Site-directed Mutagenesis-Site-directed mutagenesis was carried out using the GeneTailer site-directed mutagenesis kit (Invitrogen) according to the manufacturer's protocol with appropriate pairs of complementary mutagenic primers. Mutants were constructed from the FLAG-tagged rat brain NCKX2 (fNCKX2) (4) in the pcDNA3.1 vector. Single or double point mutations were created to replace serine or threonine residues with alanine. All constructs were verified by DNA sequencing prior to use.
Dominant Negative Mutations of PKCs-The expression vector pHACE (24) was used to generate plasmids that encode the dominant negative mutant of PKC with a C-terminal hemagglutinin tag. pHACE-PKC-DN expression plasmids were generated by ligating full-length open reading frames of PKC isoforms with a dominant negative (DN) point mutation at the ATP binding site (K368R, K376R, K437R, and K384R for PKC␣, PKC␦, PKC⑀, and PKC, respectively) into pHACE digested with EcoRI (see Ref. 24 for details). The expression of each PKC-DN in HEK293 cells was confirmed by Western blot analysis using anti-HA antibody (1:5000, Santa Cruz Biotechnology, Santa Cruz, CA). An anti-tubulin blot was performed as a loading control (anti-tubulin antibody, Sigma). Relative expression of PKC-DN mutants was determined by comparative densitometry of Western blots (supplemental Fig. S1).
Photometry-Two days after transfection, HEK293 cells on poly-D-lysine-coated coverslips were loaded with 5 M fura-2 AM (Molecular Probes) and mounted in a perfusion chamber on a microscope stage. The fura-2 ratio was determined on a field of cells by measuring fluorescence from excitation at 340 nm and 380 nm with a D-104 microscope photometer using Felix version 1.42 software (Photon Technology International). The perfusion solutions used contained either 140 mM NaCl or LiCl together with 5 mM KCl, 0.1 mM CaCl 2 , 10 mM D-glucose, and 10 mM HEPES-tetramethylammonium (pH 7.4). The cells were perfused alternatively for 3 min with sodium buffer or 2 min with lithium buffer. Half way through the second perfusion with sodium buffer, the cells were incubated for 15 min with either 0.1 M, ␤-phorbol myristate acetate (PMA), or 4␣-PMA. Ca 2ϩ -transport rates for reverse-mode NCKX2 activity were determined by linear regression of the initial linear rate of change of the fura-2 ratio using GraphPad Prism version 4.0 software.
In Vivo Phosphorylation Assay-Two days after transfection, HEK293 cells in 100-mm dishes were washed twice with 5 ml of Dulbecco's modified Eagle's medium without sodium phosphate and then incubated with 100 Ci/ml 32 P-labeled orthophosphate in 2 ml of Dulbecco's modified Eagle's medium without sodium phosphate for 3-4 h at 37°C. The cells were then treated with or without 1 M ionomycin and/or 0.1 M PMA or 4␣-PMA for 15 min at room temperature. The cells were then placed on ice, washed twice with 5 ml of ice-cold phosphate-buffered saline and lysed in situ with 1 ml of IP buffer (10 mM Tris, pH 7.5, 1 mM EDTA, 1% Triton X-100, 120 mM NaCl, 10 mM NaF, 10 mM sodium pyrophosphate, 1 mM sodium vanadate, 100 units/ml aprotinin, 0.1 mM phenylmethylsulfonyl fluoride, and complete protease inhibitor tablets (Roche Applied Science)). All subsequent steps were performed at 4°C. The lysate was transferred to a microcentrifuge tube and incubated on ice for 20 min followed by centrifugation at 14,000 rpm for 5 min. 25 g of protein extract was frozen at Ϫ80°C for later SDS-PAGE and immunoblotting analysis. 5 mg of protein extract from the supernatant was adjusted to 1 ml with the IP buffer, precleared with protein A-Sepharose beads and transferred to a new tube. The supernatant was mixed with 5 g of M2 anti-FLAG monoclonal antibody by rotating for 2 h followed by addition of 80 l of 25% protein A beads for 30 min. The beads were washed three times in IP buffer by centrifuging at 3,000 rpm for 2 min. The sample was eluted by adding 40 l of 4ϫ SDS sample buffer containing 8% 2-mercaptoethanol and heating to 50°C for 5 min.
The immunoprecipitated samples and protein extracts were resolved on 9% SDS-PAGE gels and transferred to nitrocellulose membranes. The 32 P autoradiograph was obtained by incubating the membranes with x-ray film in a cassette containing one intensifying screen at Ϫ80°C for 3-48 h. No signal was observed after 15 min of incubation.
The membranes were subsequently incubated in phosphatebuffered saline containing 0.1% Tween 20, 5% skim milk powder and probed with the affinity-purified anti-NCKX2 polyclonal antibody F (21), followed by application of horseradish peroxidase-conjugated anti-rabbit IgG antibody. The membranes were developed using ECL reagents and typically required exposure times of 1 s to 2 min. The band intensities were quantified using Image J software (National Institutes of Health), and 32 P incorporation was determined by dividing the intensities of the 32 P autoradiograph signals by the intensities of the immunoblot signals using GraphPad Prism version 4.0 software.
Ca 2ϩ Imaging at the Calyx of Held-Transverse brainstem slices containing the medial nucleus of the trapezoid body were prepared from 8-to 9-day-old Sprague-Dawley rats as previously described (11), using a vibratome (VT 1000s, Leica, Germany) and ice-cold low calcium artificial cerebro-spinal fluid containing (in mM): 125 NaCl, 25 NaHCO 3 , 2.5 KCl, 1.25 NaH 2 PO 4 , 2.5 MgCl 2 , 0.5 CaCl 2 , 25 glucose, 0.4 sodium ascorbate, 3 myo-inositol, 2 sodium pyruvate, at pH 7.4 when saturated with carbogen (95% O 2 /5% CO 2 ) and with an osmolarity of ϳ320 mosM. Whole cell patch clamp recordings of calyces of Held were made under visual control using differential interference illumination in an upright microscope (BX50WI, Olym-pus, Japan). Ca 2ϩ concentrations were measured by fluorescence imaging using a 60ϫ water immersion objective (numerical aperture, 0.9; LUMPlanFl, Olympus), an air-cooled slow-scan charge-coupled device camera (SensiCam, PCO, Germany), and a monochromator (Polychrome II, Till Photonics, Germany), which were controlled by a PC, running a customized software programmed with MicroSoft Visual Cϩϩ (version 6.0). The procedures for cytosolic Ca 2ϩ measurement in slices have been previously described in detail (11). Calcium transients were evoked by stimulating presynaptic axons with a concentric bipolar stimulating electrode (World Precision Instruments, Sarasota, FL), placed between the medial border of the medial nucleus of the trapezoid body and the midline of the brainstem. The pipette solution contained (in mM): 120 potassium gluconate, 30 KCl, 20 HEPES, 4 Mg-ATP, 4 sodium ascorbate, and 0.3 Na-GTP at pH 7.3 (adjusted with KOH). Experiments were performed at 24 Ϯ 1°C.
Chemicals and Statistics-4-␣-PDBu, FK-506, fura-4F were purchased from Biomol, from A.G. Scientific, and from Molecular Probes, respectively. Exchanger inhibitory peptide (XIP, RRLLFYKYVYKRYRAGKQRG) was synthesized by AnyGen (Gwangju, Korea). All other chemicals, including PDBu, chelerythrine, GF109203X, alkaline phosphatase, and cyclosporin A, were purchased from Sigma. Values for the peak amplitude of outward NCKX2 currents in the presence of drug were normalized to those of control conditions obtained from the same cell. The normalized amplitudes were compared statistically using Student's t test. Statistical data are presented as mean Ϯ S.E. (n ϭ number of cells studied).

RESULTS
Na ϩ /Ca 2ϩ ϩK ϩ Exchange Currents (I NCKX2 ) in Rat NCKX2transfected HEK293 Cells-Using whole cell patch clamp technique, outward NCKX2 current was recorded at a holding potential of 0 mV from NCKX2-transfected HEK293 cells, which were internally dialyzed with a high concentration of NaCl (2). The reverse Na ϩ gradient favors Ca 2ϩ entry in the presence of extracellular Ca 2ϩ and K ϩ . To prevent the rapid increase in cytosolic [Ca 2ϩ ], which was expected to develop as a consequence of exchange in reverse mode, the pipette solution contained 10 mM BAPTA. The control bathing solution contained LiCl and EGTA, which provides no counterpart to be exchanged for Na ϩ . NCKX2-transfected cells generated outward current at 0 mV when they were exposed to bath solution containing both Ca 2ϩ and K ϩ together, but not with either alone (Fig. 1A, left). I NCKX2 was not significantly affected by 0.5 mM ouabain, indicating that the Na ϩ -pump current did not contaminate our data (not shown). When non-transfected HEK293 cells were analyzed under identical conditions, they produced no obvious current (Fig. 1A, right). To obtain current-voltage (I-V) relationships, the cells were subjected to voltage ramps ranging from Ϫ80 to ϩ80 mV during the bath perfusion of control or test solutions. The I-V curve observed from the NCKX2-transfected cell during the bath-perfusion of Ca/K solution showed no reversal, which was clearly distinct from the control I-V curve, indicating that the outward current corresponded to the electrogenic movement of Na ϩ in exchange for Ca 2ϩ and K ϩ catalyzed by the NCKX2 protein.
PDBu Enhanced I NCKX2 in a Dose-dependent Manner-When outward I NCKX2 was elicited repeatedly by alternating bath application of Ca/K solution (for 15-20 s, every 2 min) with Li solution, spontaneous enhancement of outward I NCKX2 was never observed in control conditions ( Fig. 2A, left). In the presence of 100 nM PDBu, however, outward I NCKX2 was significantly and progressively enhanced (Fig. 2B, left). The enhancement started as early as 2 min after the bath application of PDBu. The I-V relationship of outward currents in the control condition and in the presence of PDBu shows that PDBu enhanced the outward current uniformly over the entire voltage range tested (from Ϫ80 mV to ϩ80 mV), and thus no reversal was observed between the two I-V curves, indicating that PDBu did not activate endogenous nonselective cationic current (Fig. 2B, right). The time course for the change in the amplitude of I NCKX2 at different concentrations of PDBu is shown in Fig. 2C. The dose dependence of the PDBu effect on I NCKX2 was quantified as the peak amplitude of I NCKX2 measured at a given time following the application of PDBu, normalized to the averaged amplitude of two or three I NCKX2 peaks elicited before the drug application from the same cell (this estimate is referred to below as "relative amplitude of I NCKX2 "). PDBu enhanced NCKX2 activity in a concentration-dependent manner, resulting in relative amplitudes of I NCKX2 at 100 nM, 500 nM, and 1 M PDBu of 148 Ϯ 9.6% (n ϭ 9), 195 Ϯ 8.3% (n ϭ 5), and 306 Ϯ 41% (n ϭ 4), respectively (8 min treatment of PDBu). In contrast, the inactive PDBu analogue, 4␣-phorbol 12,13-dibutyrate (1 M) did not change NCKX2 activity (Fig.  2C, 95.1 Ϯ 6.8%, n ϭ 6), suggesting that the effect of PDBu is dependent on the activation of PKC.
We examined the effects of PDBu (100 nM) on outward NCKX currents (I NCKX ) recorded from NCKX2-, NCKX3-, and NCKX4-expressing cells (Fig. 3). The enhancement of NCKX2 activity by 100 nM PDBu was significantly higher than that of NCKX3 or NCKX4, whose activity was little affected by PDBu. The relative amplitude of I NCKX was estimated as 147 Ϯ 7.7% (n ϭ 5), 103 Ϯ 17.5% (n ϭ 5), and 105 Ϯ 11.7% (n ϭ 6) for NCKX2, NCKX3, and NCKX4, respectively (10-min treatment of PDBu). This result indicates that the PDBu-induced enhancement (PIE) is specific for NCKX2, and provides a basis for an intriguing possibility that such functional differences in members of the NCKX family may underlie their distinctive regulation.
The Enhancement of NCKX2 Activity by PDBu Is Dependent on the Activation of PKC-To confirm that the effect of PDBu on I NCKX2 was due to activation of PKC, we tested whether PKC inhibitors could prevent activation. When PDBu was washed out, the enhanced I NCKX2 slowly decreased to the control level over Ͼ10 min (data not shown). In contrast, when a PKC inhibitor was applied subsequent to PDBu, the enhancement of I NCKX2 was rapidly abolished within 2 min ( Fig. 4B for chelerythrine; Fig. 4C for GF109203X). Moreover, in the presence of the PKC inhibitor, reapplication of 100 nM PDBu no longer enhanced I NCKX2 . The effects of PKC inhibitors are summarized in Fig. 4D. Although PDBu alone enhanced I NCKX2 to 148 Ϯ 7% (n ϭ 6) of the control, the PIE was significantly inhibited by 1 M chelerythrine (94.7 Ϯ 2%, n ϭ 4, p Ͻ 0.05), a nonspecific PKC inhibitor, or by 100 nM GF109203X (82 Ϯ 17%, n ϭ 4, p Ͻ 0.05), an inhibitor for classic and novel type PKC (25,26). To confirm the involvement of PKC in PIE, we studied the effects of PDBu in HEK293 cells that overexpress one of dominant negative isoforms of PKC together with NCKX2. The relative amplitude of I NCKX2 was significantly reduced in the HEK293 cells expressing the dominant negative PKC⑀. Mean values for relative I NCKX2 measured at 12-min treatment of 100 nM PDBu are summarized in Fig. 4E. The magnitude of PIE was not significantly affected in the cells expressing dominant negative PKC␣, PKC␦, or PKC. Consistently, pretreatment of Gö6976 (200 nM), an inhibitor of classic PKC and protein kinase D (or PKC), had no significant effect on PIE (150 Ϯ 4%, n ϭ 6, Fig. 4D). These results suggest that PIE is primarily mediated by PKC⑀.
It has been reported that PKC can activate adenylate cyclase, protein kinase G, and protein kinase D (27)(28)(29). To investigate whether PKA or protein kinase G downstream of PKC is involved in PIE, we tested whether forskolin (10 M) or 8-Br-cGMP (100 M) can directly enhance I NCKX2 . Neither of these activators enhanced I NCKX2 , suggesting that neither PKA nor protein kinase G is involved in PIE (supplemental Fig. S1C). Recordings of whole cell current measured at a holding potential of 0 mV from an HEK293 cell expressing NCKX2 (NCKX2-expressing cell, left) or non-transfected HEK293 cell (control cell, right) using a pipette solution that contained 120 mM NaCl and 10 mM BAPTA. A, K ϩ dependence of NCKX2 current (I NCKX2 ) was tested by bath application of test solutions containing either K ϩ or Ca 2ϩ alone or both ions together. Application of each test solution is indicated by a bar below the current record. Bath application of each test solution was intervened by control bathing solution (white bars marked by "Li"), which contained (in mM) 120 LiCl, 20 tetraethylammonium-Cl, 0.5 EGTA, 10 HEPES, and 10 glucose. LiCl was replaced with equimolar KCl in the test solution marked with "K," and EGTA was replaced with 1 mM CaCl 2 in that marked with "Ca," and both were replaced in "Ca/K." Outward current was elicited in only in the NCKX2-expressing cell when the bath-applied test solution contained both K ϩ and Ca 2ϩ . B, I-V relationships were obtained from the current responses to voltage ramps from Ϫ80 mV to ϩ80 mV, which were applied at times labeled as a-g in A.
Moreover, Gö6976 (200 nM), a protein kinase D inhibitor, had no significant effect on PIE, suggesting that protein kinase D is not involved in PIE (Fig. 4D).
Effects of PDBu on Ca 2ϩ Decay Rate at the Calyx of Held-We tried to test whether PDBu can enhance the NCKX activity endogenously expressed at the calyx of Held, a large glu-tamatergic presynaptic terminal in the medial nucleus of the trapezoid body (Fig. 5A), where NCKX, NCX, and the plasma membrane Ca 2ϩ -ATPase account for ϳ40%, 25%, and 20% of the total calcium clearance mechanisms (11). Because calyx of Held expresses large Ca 2ϩ -activated K ϩ current, it is not practical to isolate NCKX current. Thus, we studied the effects of PDBu on the Ca 2ϩ -decay rate at the calyx of Held.
When [Ca 2ϩ ] i was measured using whole cell patch techniques, treatment of 200 nM PDBu typically increased resting [Ca 2ϩ ] level and slowed down the Ca 2ϩ -decay rate of calcium transients (CaT) evoked by a short depolarizing pulse. To prevent a possible artifact caused by dilution of cytosolic components necessary for normal signal transduction pathway during whole cell patch recordings, we made a whole cell patch on the calyx of Held during a brief time (90 s) with a pipette solution containing 200 M fura-4F, and gently withdrew the patch pipette. After the withdrawal, we could get fluorescence intensity comparable to 50 -100 M of fura-4F. We evoked CaTs by stimulating afferent axon fibers using a stimulation electrode located at the midline of the brain stem. About 15 or 20 action potentials at 200 Hz were required for obtaining a CaT whose amplitude was in the range of 1-1.5 M.
To inhibit the activity of NCX and plasma membrane Ca 2ϩ -ATPase, we included 200 M exchanger inhibitory peptide (XIP) and 50 M carboxyeosin in the pipette solution, which blocks NCX and plasma membrane Ca 2ϩ -ATPase, respectively (11,30). Because Ca 2ϩ clearance is steeply dependent on the level of [Ca 2ϩ ] i excursion from the resting value, we compared two CaTs only when the difference in the peak ⌬[Ca 2ϩ ] levels between the two CaTs is Ͻ100 nM. Mean values for peak Ca 2ϩ levels before and after PDBu application were 1.44 Ϯ 0.28 M and 1.42 Ϯ 0.64 M (n ϭ 5).
Each Ca 2ϩ transients recorded under these conditions were fitted with a bi-exponential function:  DECEMBER 22, 2006 • VOLUME 281 • NUMBER 51

PKC Enhances NCKX2 Activity
. We regarded the time derivative of the bi-exponential fit at t ϭ 0 normalized with the peak Ca 2ϩ amplitude, as a parameter that represents the activity of Ca 2ϩ clearance at the peak ⌬[Ca 2ϩ ] i level (see Ref. 11 for details). The t ϭ 0 value in control conditions with internal XIP and carboxyeosin was 7.99 Ϯ 0.54 s Ϫ1 (n ϭ 5), which was significantly slower than the mean t ϭ 0 value measured without XIP and CE included (14.0 Ϯ 0.90 s Ϫ1 , n ϭ 10). In the presence of XIP and carboxyeosin in the cytosol, PDBu clearly accelerated the Ca 2ϩ -decay rate of the Ca 2ϩ transient (Fig. 5B), and accordingly increased the t ϭ 0 to 12.88 Ϯ 1.27 s Ϫ1 (n ϭ 5, p Ͻ 0.02, paired t test, Fig. 5C). The total calcium clearance rate (d[Ca 2ϩ ] T /dt) was calculated from the time derivative of the decay phase of the Ca 2ϩ transient and calcium binding ratios of endogenous Ca 2ϩ buffers and Ca 2ϩ -indicator dye and plotted as a function of ⌬[Ca 2ϩ ] i . Fig. 5D shows that the difference in the total calcium decay rate between Ca 2ϩ transients before and after PDBu treatment is more pronounced at higher ⌬[Ca 2ϩ ] i levels, indicating the NCKX activity is enhanced by PDBu. Direct Phosphorylation of NCKX2 in Transfected HEK293 Cells-The above results show that PKC is a potent activator of the rat brain NCKX2. Next, we investigated whether PIE of I NCKX2 involves a phosphorylation process. When NCKX2-ex-pressing HEK293 cells were dialyzed intracellularly with pipette solution containing alkaline phosphatase (50 units/ml), neither 100 nM nor 1 M PDBu induced any significant enhancement of I NCKX2 (96 Ϯ 2% (n ϭ 5) and 88 Ϯ 4% (n ϭ 3), respectively, supplemental Fig. S2). No enhancement of the outward I NCKX2 in the presence of excess intracellular alkaline phosphatase indicates that the PDBu effect requires a phosphorylation process.
Phosphorylation of the NCKX2 protein molecule was examined by [ 32 P]orthophosphate labeling of HEK293 cells expressing FLAGtagged NCKX2, followed by various treatments to stimulate PKC. The upper panels of Fig. 6A demonstrate that the overall incorporation of 32 P into cellular proteins, and the level of expressed NCKX2, was equivalent under all conditions and was unaltered by PKC activation. The representative experiment shown in the lower panels of Fig. 6A illustrates that NCKX2 immunoprecipitated from untreated cells contained a constitutive level of phosphorylation, whereas no corresponding band was observed in immunoprecipitates from vector-transfected cells (lower left panel). Treatment with 0.1 M PMA, but not with the inactive analogue, 4␣-PMA, resulted in an increased level of NCKX2 phosphorylation (lower left panel), whereas the amount of NCKX2 present in the respective immunoprecipitates was similar (lower right panel). The presence of 1 M ionomycin had little effect on the enhanced phosphorylation. The averaged data from four independent experiments (Fig. 6B) indicated a significant 2-to 3-fold stimulation of 32 P incorporation when normalized to NCKX2 protein. These data suggest that the level of NCKX2 phosphorylation was enhanced upon specific activation of PKC, whereas the lack of any effect of ionomycin suggests that a Ca 2ϩ -independent isoform of PKC was responsible.
To ensure that PMA was effective in stimulating NCKX2 activity under these conditions, reverse-mode (i.e. Ca 2ϩ entry) operation of NCKX2 was assessed by photometry of transfected HEK293 cells grown on coverslips, loaded with the fluorescent Ca 2ϩ -indicator, fura-2, and mounted in a perfusion chamber on a microscope stage. A representative experiment is shown in Fig. 6C, and averaged data from three independent experiments are shown in Fig. 6D. NCKX2 activity was induced by a perfusion switch from Na ϩ -to Li ϩ -containing buffer, which caused an increase in Ca 2ϩ entry and a rise in the fura-2 ratio (vectortransfected cells do not display this change, data not shown). Note that the rate of rise of the fura-2 ratio, as well as the peak value, due to NCKX2 activity typically decreases monotonically with repeated perfusion switches (Fig. 6, A and B). Under these conditions, a 15-min treatment with 0.1 M PMA induced an increase of ϳ2-fold in the rate of NCKX2 operation not observed with the inactive analogue, 4␣-PMA (Fig. 6D). These data indicate that PMA treatment resulted in similar changes in NCKX2 activity as was observed upon PDBu treatment.
An increase in 32 P incorporation in response to PMA stimulation was also observed for all three NCKX2 double mutants, even those without significant PIE of I NCKX2 (data not shown). Due to the complexity of the phosphorylation experiment and the inherent variation in incorporation, our data were not sufficiently precise to determine statistically significant differences in the amount of phosphorylation between the wild-type and mutant NCKX2 molecules. It was thus unclear if the maximal level of phosphorylation observed in these mutants was less than that of the wild type.
Thr-476 Is Critically Involved in PIE-A scan of the rat NCKX2 amino acid sequence for PKC recognition motifs (31) revealed three potential sites, as illustrated in Fig. 7A. To identify which of these sites might be phosphorylated by PDBu treatment, we mutated each putative PKC phosphorylation acceptor serine or threonine to alanine, either individually or in combination. These mutations were introduced into the fNCKX2, and the PDBu response was analyzed in transiently transfected HEK293 cells expressing each mutant. To confirm that these mutations did not affect expression and basal function of NCKX2, we measured the outward NCKX current den-sity from cells expressing wild-type (WT) or each mutant (supplemental Table S1). No statistically significant difference in the current density between WT and mutants indicated that alanine substitution at the putative PKC phosphorylation sites, either individually or in combination, had negligible effects on the conformation and expression of NCKX2.
For each mutant, we examined the effect of treatment with 100 nM PDBu on the relative amplitude of I NCKX2 . Fig. 7B shows the time courses of PDBu effects on the relative amplitude of I NCKX2 (right), together with representative current traces (left) recorded from wild-type and three single-site mutants. None of the single-site mutants were significantly different from wild type in terms of the relative amplitude of I NCKX2 measured at 8 min of PDBu superfusion (148 Ϯ 9.6% for WT (n ϭ 9); 152 Ϯ 21% for T166A (n ϭ 7); 125 Ϯ 10% for T476A (n ϭ 9); 157 Ϯ 6.5% for S504A (n ϭ 13)). Although the I NCKX2 of the T476A mutant was initially enhanced by PDBu, PIE became smaller 8 -10 min after PDBu application, and the amplitude of I NCKX2 eventually returned to the control level following 14 min of treatment (Fig. 7B). The biphasic response observed in the T476A mutant suggests that phosphorylation targets (Thr-166 or Ser-504) other than Thr-476 might undergo dephosphorylation by endogenous protein phosphatases (PPs) during the late phase of PDBu treatment, subsequent to the initial PKC-dependent phosphorylation. To test this hypothesis, we examined the effect of various PP  DECEMBER 22, 2006 • VOLUME 281 • NUMBER 51 inhibitors on the decline in PIE of I NCKX2 observed in the T476A mutant (I NCKX2-T476A ). PP inhibitors were bath-applied 8 -10 min following PDBu application, when the decline of I NCKX2-T476A first becomes apparent. There are two kinds of PPs: Ca 2ϩ -independent and Ca 2ϩ -dependent PPs. PP-1 and PP-2A constitute the former group, and PP-2B (or calcineurin) belongs to the latter. We found that okadaic acid (1 M), a specific inhibitor of PP-1 and PP-2A, had no effect on the decline of I NCKX2-T476A (Fig. 7C). On the other hand, the specific inhibitors of calcineurin, cyclosporin A (1 M) or FK-506 (1 M), clearly prevented the decline phase normally observed in the presence of PDBu alone (Fig. 7C). The relative amplitudes of I NCKX2-T476A measured 6 min following application of 100 nM PDBu alone, PDBu plus okadaic acid (1 M), PDBu plus cyclosporin A (1 M), or PDBu plus FK-506 (1 M) were 96 Ϯ 14% (n ϭ 4), 115 Ϯ 18% (n ϭ 4), 133 Ϯ 6% (n ϭ 7, p Ͻ 0.05), or 139 Ϯ 5% (n ϭ 4, p Ͻ 0.05), respectively (p values indicate the statistical significance in the difference between PDBu alone and PDBu plus PP inhibitor). Unlike the T476A mutant, wild-type and other single mutants (T166A or S504A) exhibited no significant change in amplitude of I NCKX2 following treatment with calcineurin inhibitors (Fig. 8).

PKC Enhances NCKX2 Activity
It is well known that activation of calcineurin is dependent on an increase in cytosolic [Ca 2ϩ ] (half-maximal [Ca 2ϩ ] ϭ 600 nM (32)). Thus, it was surprising that calcineurin was activated in the above experiments, considering that the cells were dialyzed with pipette solution containing 10 mM BAPTA. To test if [Ca 2ϩ ] i rose high enough to activate calcineurin in our experimental conditions, we measured I NCKX2-T476A and [Ca 2ϩ ] i simultaneously by adding the Ca 2ϩ -indicator dye, fura-2 (500 M) to the same patch pipette solution used in the above experiments (containing 10 mM BAPTA). Indeed, outward I NCKX2 was accompanied by a Ca 2ϩ transient whose peak [Ca 2ϩ ] level increased higher than 2.5 M (supplemental Fig. S3A). Moreover, when I NCKX2-T476A was induced every 4 min instead of every 2 min, the decline in PIE was not observed over a similar time course (supplemental Fig. S3B). These results indicate that the increase in cytosolic [Ca 2ϩ ] induced by reverse-mode I NCKX2 under our experimental conditions was sufficient to activate calcineurin, which in turn dephosphorylated the T476A mutant, but not the other molecules. Thus, blocking the  Either 25 g of protein extract from lysed cells (lysate) or FLAG-immunoprecipitated NCKX2 (FLAG-IP) were resolved on 9% SDS-PAGE gels and transferred to nitrocellulose membranes. The 32 P autoradiograph (left panels) was recorded followed by immunoblotting with the anti-NCKX2 polyclonal F antibody (right panels). B, the specific phosphate incorporation into the immunoprecipitated samples was determined by dividing the intensities of the bands on the 32 P autoradiographs by those of the corresponding bands on the immunoblots. The mean values (ϮS.E.) from four separate experiments are shown. Statistically significant differences from the untreated sample are indicated with p Ͻ 0.01 (*) or p Ͻ 0.05 (**) using a paired ratio comparison t test. C, HEK293 cells grown on poly-D-lysine-coated coverslips were transiently transfected with FLAG-tagged NCKX2, loaded with 5 M fura-2 AM, and mounted in a perfusion chamber on a microscope stage. The cells were perfused alternatively with either Na ϩ or Li ϩ buffer containing 5 mM K ϩ . The fura-2 ratio from excitation at 340 nm and 380 nm was captured for the entire field of cells using a photometer. The perfusion was temporarily stopped, and the cells were incubated for 15 min with 0.1 M PMA or 4␣-PMA at the time indicated by the arrow. The reduction in both amplitude and rate of the rise in [Ca 2ϩ ] upon repetitive Li ϩ perfusion switches is typical also of untreated cells under these conditions. D, the mean initial rates (ϮS.E.) of increase in the fura-2 ratio for three pulses of NCKX2 activity were compared for PMA-or 4␣PMA-treated cells, respectively. Data were obtained from three separate experiments. ***, statistically different from control (PMA pulse 1) with p Ͻ 0.01 using a paired ratio comparison t test, and from 4␣-PMA pulse 2 with p Ͻ 0.05 using a t test to compare the rate data. decline in PIE by inhibiting calcineurin revealed that no singlesite mutation at the three putative PKC phosphorylation sites completely inhibited the stimulation of I NCKX2 induced by PDBu.
We hypothesized that multiple PKC consensus sites might be involved in the PKC-dependent stimulation of NCKX2 (33,34). To test this hypothesis, similar studies were carried out with three double mutants: T166A/T476A, T476A/S504A, and T166A/S504A. Fig. 8A shows representative current traces for each mutant before and following 6 min of treatment with 100 nM PDBu (I-V curves recorded for WT and T476A/S504A are shown in supplemental Fig. S4). The two double mutants involving Thr-476 showed significantly reduced PIE of I NCKX2 . The relative amplitudes of I NCKX2 6 min after application of PDBu were 137 Ϯ 10% for WT (n ϭ 12), 99 Ϯ 4.4% for T166A/ T476A (n ϭ 7, p Ͻ 0.05) and 104 Ϯ 8% for T476A/S504A (n ϭ 7, p Ͻ 0.05). On the other hand, the PIE of the T166A/S504A mutant was only slightly inhibited (131 Ϯ 7%, n ϭ 10). Finally, to rule out the possibility that activation of calcineurin might be responsible for the loss of PKC effects on the double mutants, we examined effects of 1 M FK-506 on PIE of I NCKX2 from the cells expressing each double mutant. When FK-506 was added 8 min following PDBu application, no significant effect on I NCKX2 was observed for any of the double mutants (Fig. 8B). The relative amplitudes of I NCKX2 in the presence of PDBu plus FK506 were 157 Ϯ 8% for WT (n ϭ 8), 105.6 Ϯ 6.6% for T166A/ T476A (n ϭ 7), 102 Ϯ 10% for T476A/S504A (n ϭ 7) and 134 Ϯ 14% for T166A/S504A (n ϭ 6). This lack of FK-506 effect indicates that activation of calcineurin was not related to the loss of PIE in the double mutants.
The above results suggest that multiple PKC consensus sites are involved in PIE of NCKX2. To evaluate the contribution of each PKC site to PIE, the mean values for relative amplitude of I NCKX2 of wild-type and various mutants measured at 14 min after PDBu treatment in the presence of 1 M FK506 were sorted and plotted on the bar graph in Fig. 9. The PIE was higher in the T166A/S504A (Thr-476 is the available target for PKC) mutant than in the T476A/S504A (Thr-166 is available) or T166A/T476A (Ser-504 is available) mutant, suggesting that these sites are not functionally equivalent, but each contributes to PIE with a different weighting factor (w). In addition, from the observation that PIE was higher in the T476A mutant (both of Thr-166 and Ser-504 are available) than T166A/T476A or T476A/S504A (either Thr-166 or Ser-504 is available), it could be inferred that either Thr-166 or Ser-504 is not sufficient to yield PIE, but both sites (Thr-166 and Ser-504) contribute cooperatively to induce PIE.

DISCUSSION
In the present study, we have investigated the effect of PDBu on neuronal NCKX2 expressed in HEK293 cells, and shown evidence for the stimulation of NCKX2 activity by PKC, which may be mediated via direct phosphorylation of the exchanger. Recently, ␤-phorbol esters have been shown to activate not only The values for the amplitude of I NCKX2(T476A) were normalized to their corresponding control amplitudes. Error bars represent Ϯ S.E. Asterisks denote statistical significance between PDBu alone and PDBu plus cyclosporin A or PDBu plus FK-506 (*, p Ͻ 0.05).
PKC but also Munc13, which enhances fusion of secretory vesicles to the plasma membrane (35). We think it unlikely that the PIE of NCKX2 function we have observed is caused by increased vesicular insertion to the plasma membrane by Munc13 (36) for the following reasons. 1) PIE was not observed in other NCKX isoforms (NCKX3 or NCKX4), as might be expected if PIE were mediated via a general vesicle insertion mechanism; 2) PIE of the NCKX2 current was blocked by PKC inhibitors GF109203X or chelerythrine or by inclusion of alkaline phosphatase in the recording pipette; 3) NCKX2 double mutants at PKC phosphorylation motifs (T166A/T476A and T476A/ S504A) abolished the PIE, suggesting a direct activation of NCKX2 by PKC; and 4) enhanced phosphorylation of the exchanger protein itself was observed in the presence of PMA. Thus, it appears that NCKX family members expressed in the brain are differentially regulated by PKC. To our knowledge, this is the first report on the modulation of an NCKX family member via protein kinases, although there are many studies on such regulation in the NCX family.
PKC Isoform Involved in PIE and Downstream Mechanisms-There are three categories of PKC isoforms according to the dependence of their activation on calcium and diacylglycerol. In the present study, two or three episodes of I NCKX2 were induced prior to the application of PDBu. During this period, [Ca 2ϩ ] varied typically in the range of 10 to 150 nM, because the cells were intracellularly perfused with 10 mM BAPTA (supplemental Fig. S3). The [Ca 2ϩ ] level prior to the induction of the very first enhanced I NCKX2 episode in the presence of PDBu was well below 100 nM, suggesting that Ca 2ϩ -independent novel PKCs might be involved in the PIE. Consistent with this observation, PMA enhanced the 32 P incorporation into NCKX2 without pre-treatment of ionomycin. Moreover, we showed that overexpression of the dominant negative mutant of PKC⑀ significantly reduced PIE. Interestingly, it was reported that a novel type PKC (PKC⑀) but not classic PKC is expressed at the presynaptic terminal, calyx of Held, where NCKX activity was demonstrated (11,37).
The present study suggests that PIE may involve multiple PKC sites on the molecule NCKX2. Previous studies on PKCdependent modulation of ionic channels indicate that PKC modulates channel activity not only by the modulation of the intrinsic channel properties but also by the regulation of receptor/channel trafficking. Modulation of intrinsic channel properties by PKC involves usually direct phosphorylation of the channel protein (38,39). On the other hand, PKC-dependent regulation of channel trafficking involves diverse mechanisms. PKC promotes insertion of N-methyl-D-aspartate channels or causes internalization of ATP-sensitive K ϩ channel through phosphorylation of channel-associated protein(s) involved in signaling and/or trafficking without phosphorylation of the channel protein itself (40,41). However, PKC-induced internalization of GluR2 and surface expression of Kir1.1 require phosphorylation of the channel itself (42,43). Although the inhibition of the PIE by point mutation favors the former scenario (change in intrinsic properties of the NCKX2 transporter), because the mechanism that leads  . Mean values for relative I NCKX2 of wild-type and various mutants measured at 14 min following PDBu treatment in the presence of 1 M FK-506. The relative contribution (w) of each putative PKC phosphorylation site was calculated from double mutants in which only a single site relevant to PIE is assumed to be available for PKC. The w values calculated in this way for Thr-166, Thr-476, and Ser-504 were 0.05, 0.6, and 0.05, respectively. The expected value for relative I NCKX2 in each mutant was calculated from simple addition of w values of available phosphorylation sites in each mutant (inverted triangles). To correct the discrepancy between the real and expected values, different cooperativity factors (␣) were multiplied to the summed w values, and the resultant expected value for each mutant is shown by triangles. The ␣ values between Thr-166 and Ser-504 and between Thr-476 and other sites were assumed to be 6.7 and 1.5, respectively. Error bars: Ϯ S.E., n ϭ number of cells studied. from phosphorylation of NCKX2 to enhanced activity, further experiments will be required to distinguish unambiguously between recruitment of new transport units to the plasma membrane or the enhancement in intrinsic activity of existing units.
Does Direct Phosphorylation of NCKX2 Induce PIE?-To address this issue, we carried out phosphorylation experiments similar to those of Fig. 6 with the three double mutants. An increase in 32 P incorporation in response to PMA was observed for all double mutants, even those that exhibited negligible PIE of I NCKX2 (data not shown). However, it was unclear if the maximal level of phosphorylation observed in these mutants was significantly less than that of the wild type. We believe two possible explanations may account for the discrepancy between the functional effects and direct phosphorylation of the NCKX2 mutants. First, NCKX2 may be phosphorylated in a PKC-dependent manner at sites in addition to Thr-166, Thr-476, and Ser-504. Although these other sites do not affect activity directly, the background incorporation prevents observation of only those phosphorylation events directly linked to activity changes. The activity-independent phosphorylation events may occur at PKC sites that do not match the typical consensus motif, or may be indirect, due to PKC activation of a kinase with different specificity. Second, it is possible that the mutations we have introduced do not alter PKC-dependent phosphorylation directly but, rather, have the effect of uncoupling phosphorylation from the increase in activity via a change in protein structure. Even if phosphorylation of Thr-476 is not a direct cause of PIE, our point-mutation studies suggest that Thr-476 plays a critically important role. Moreover, among the mutants that exhibited significant PIE, only the T476A mutant was susceptible to calcineurin, indicating that Thr-476 is necessary not only for the induction of PIE but also for maintenance of PIE by protecting other susceptible phosphorylated sites present on NCKX2 from subsequent dephosphorylation by calcineurin.
The PKC consensus sequences involving Thr-166, Thr-476 or Ser-504 found in NCKX2 are not conserved in NCKX3 or NCKX4. All three of these NCKX family members possess putative PKC-phosphorylation sites in regions of the proteins that are likely to be cytoplasmically exposed. The fact that only NCKX2 was overtly affected by PDBu indicates that only a subset of PKC consensus sites are available in the native proteins or that functional changes in exchanger properties result only from phosphorylation of specific sites.
In summary, the present study demonstrates that multiple amino acid residues are involved in PKC-mediated stimulation of NCKX2, with Thr-476 in the large intracellular loop serving as the critical residue for mediating enhancement of exchanger activity. In contrast, the activity of NCKX3 and NCKX4 was not regulated by a pathway involving PKC. This regulatory difference between family members, which is also found in other transporters, could provide an important physiological mechanism by which the multiplicity of protein functions can be generated.