The recombinant human TRPV6 channel functions as Ca2+ sensor in human embryonic kidney and rat basophilic leukemia cells.

The activation mechanism of the recently cloned human transient receptor potential vanilloid type 6 (TRPV6) channel, originally termed Ca(2+) transporter-like protein and Ca(2+) transporter type 1, was investigated in whole-cell patch-clamp experiments using transiently transfected human embryonic kidney and rat basophilic leukemia cells. The TRPV6-mediated currents are highly Ca(2+)-selective, show a strong inward rectification, and reverse at positive potentials, which is similar to store-operated Ca(2+) entry in electrically nonexcitable cells. The gating of TRPV6 channels is strongly dependent on the cytosolic free Ca(2+) concentration; lowering the intracellular free Ca(2+) concentration results in Ca(2+) influx, and current amplitude correlates with the intracellular EGTA or BAPTA concentration. This is also the case for TRPV6-mediated currents in the absence of extracellular divalent cations; compared with endogenous currents in nontransfected rat basophilic leukemia cells, these TRPV6-mediated monovalent currents reveal differences in reversal potential, inward rectification, and slope at very negative potentials. Release of stored Ca(2+) by inositol 1,4,5-trisphosphate and/or the sarco/endoplasmic reticulum Ca(2+)-ATPase inhibitor thapsigargin appears not to be involved in TRPV6 channel gating in both cell lines but, in rat basophilic leukemia cells, readily activates the endogenous Ca(2+) release-activated Ca(2+) current. In conclusion, TRPV6, expressed in human embryonic kidney cells and in rat basophilic leukemia cells, functions as a Ca(2+)-sensing Ca(2+) channel independently of procedures known to deplete Ca(2+) stores.

The activation mechanism of the recently cloned human transient receptor potential vanilloid type 6 (TRPV6) channel, originally termed Ca 2؉ transporterlike protein and Ca 2؉ transporter type 1, was investigated in whole-cell patch-clamp experiments using transiently transfected human embryonic kidney and rat basophilic leukemia cells. The TRPV6-mediated currents are highly Ca 2؉ -selective, show a strong inward rectification, and reverse at positive potentials, which is similar to store-operated Ca 2؉ entry in electrically nonexcitable cells. The gating of TRPV6 channels is strongly dependent on the cytosolic free Ca 2؉ concentration; lowering the intracellular free Ca 2؉ concentration results in Ca 2؉ influx, and current amplitude correlates with the intracellular EGTA or BAPTA concentration. This is also the case for TRPV6-mediated currents in the absence of extracellular divalent cations; compared with endogenous currents in nontransfected rat basophilic leukemia cells, these TRPV6-mediated monovalent currents reveal differences in reversal potential, inward rectification, and slope at very negative potentials. Release of stored Ca 2؉ by inositol 1,4,5-trisphosphate and/or the sarco/endoplasmic reticulum Ca 2؉ -ATPase inhibitor thapsigargin appears not to be involved in TRPV6 channel gating in both cell lines but, in rat basophilic leukemia cells, readily activates the endogenous Ca 2؉ release-activated Ca 2؉ current. In conclusion, TRPV6, expressed in human embryonic kidney cells and in rat basophilic leukemia cells, functions as a Ca 2؉sensing Ca 2؉ channel independently of procedures known to deplete Ca 2؉ stores.
Calcium is involved in a multitude of intracellular signal transduction mechanisms ranging from contraction to secretion. To achieve a high bandwidth of signal transmission with sometimes even opposing effects, cells need to control the spatio-temporal resolution of their Ca 2ϩ signals (1). In nonexcitable cells the rise in [Ca 2ϩ ] i 1 during stimulation of G-protein-coupled receptors or receptor tyrosine kinases is regulated in a complex fashion by Ca 2ϩ release from endogenous IP 3 -sensitive stores followed by store-operated Ca 2ϩ influx across the plasma membrane. One of the best characterized store-operated Ca 2ϩ entry pathways is the Ca 2ϩ release-activated Ca 2ϩ current, termed I CRAC (2,3). The molecular identity of store-operated channels is a matter of debate. Several members of the TRP family, a group of Ca 2ϩ permeable channels related to the Drosophila melanogaster TRP gene product, have been implicated in store-dependent cation influx (4,5). However, so far, none of the trp genes has been shown to encode channels with the ion-permeation properties and the sensitivity to store depleting agents of CRAC channels.
The family of the trp genes in vertebrates can be divided into three subfamilies, based on similarities in the structures of the encoded proteins (6), the TRPC, the TRPM, and the TRPV subfamilies. Members of the TRPV subfamily appear to be regulated by physical or chemical stimuli such as heat, osmotic, or mechanical stress. The recently cloned human TRPV6 gene product (GenBank TM accession no. CAC20417), formerly called CaT-L or TRP8 (7,8) or CaT1 (9), belongs into this group. Human TRPV6 channels (7), like rat TRPV6 (formerly called rat CaT1; Ref. 10) and rabbit TRPV5 channels (former ECaC for epithelial Ca 2ϩ channel; Ref. 11), show a high Ca 2ϩ selectivity, Ca 2ϩ and Na ϩ permeation properties indicative of an anomalous mole fraction behavior, and current-voltage relationships similar to CRAC channels.
Recently these common features between TRPV6 and CRAC channels were demonstrated by two groups (12,13). In addition Voets et al. (13) also revealed several differences between I CRAC in RBL cells and TRPV6-mediated Ca 2ϩ currents when expressed in HEK cells, including insensitivity of TRPV6 channels to store depletion and the effect of intracellular Mg 2ϩ , which causes voltage-dependent block of TRPV6, but not of CRAC channels. In contrast, Yue et al. (12) demonstrated activation of TRPV6 channels in CHO cells by depletion of calcium stores with IP 3 and thapsigargin. This gating mechanism could be accomplished when recordings were performed 8 -12 h but not 24 h after transfection (12). The authors assumed that, after 24 h, TRPV6 overexpression is not matched by the signal transduction apparatus presumed to be endogenously present in CHO cells and responsible for sensing store depletion. Subsequently, Yue et al. (12) suggested that the TRPV6 protein comprises all or a part of the CRAC pore.
In the present study, two different cell lines, RBL and HEK cells, were used to study the gating of human TRPV6 channels. RBL cells are a common model system to study I CRAC , whereas in HEK cells ionic conductances associated with store depletion * This work was supported in part by the Wilhelm Sander-Stiftung and the Fonds der Chemischen Industrie. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed.  (14 -16). Accordingly, one could assume a higher or different expression level of the signal transduction machinery that leads to I CRAC activation, in RBL than in HEK cells. We therefore expressed the TRPV6 cDNA in both cells and studied whether there is a cell-specific modulation of TRPV6 channel function. Under conditions of high intracellular Ca 2ϩ buffering, TRPV6-mediated Ca 2ϩ currents develop to such an extent even in RBL cells that it swamps the current through CRAC channels. However, at low intracellular Ca 2ϩ buffering, store depletion activates I CRAC but not TRPV6 channels, suggesting that the TRPV6 protein is not sensitive to store depletion under these conditions. Instead, TRPV6 channels function as Ca 2ϩ -sensing Ca 2ϩ pores and their current amplitudes are inversely correlated with the [Ca 2ϩ ] i .

EXPERIMENTAL PROCEDURES
Cell Culture, Transfected cDNA, and Transfection-HEK-293 (ATCC, 1573-CRL) and RBL-1 cells (ATTC, 1378-CRL) were from the American Type Culture Collection (Manassas, VA). Minimum essential medium with Earle's salts and L-glutamine was used for HEK cells and Dulbecco's modified Eagle's medium including L-glutamine, sodium pyruvate, and 1000 mg/liter D-glucose for RBL cells. Both culture media were supplemented with 10% fetal calf serum (Invitrogen, Paisley, UK). Cell lines were grown in a 5% CO 2 humidified incubator at 37°C. Cells were plated onto glass coverslips in 35-mm diameter Petri dishes 24 h prior to transient transfection with 4 g of DNA in 5 ml of the PolyFect ® reagents (Qiagen, Hilden, Germany). The bicistronic expression plasmid pdiCaT-L was constructed as described (7) and contained the entire protein-coding regions of the b-variant of human TRPV6 (formerly CaT-Lb, DDBJ/EMBL/GenBank TM accession no. CAC20417) followed by an internal ribosomal entry side and the green fluorescence protein DNA. The a-variant of human TRPV6 (accession no. CAC20416) differs in three amino acid residues to that of the b-variant (R157C, V378M, and T681M) and is identical with the human TRPV6 variant (accession no. AAL40230) reported by Peng et al. (17).
Electrophysiological Recordings and Solutions-For experiments, coverslips with cells were transferred 24 -32 h after transfection to the recording chamber and kept in a modified Ringer's solution containing (in mM): 145 NaCl, 10 CaCl 2 , 10 CsCl, 2.8 KCl, 2 MgCl 2 , 11 glucose, 10 HEPES, adjusted to pH 7.2 with NaOH. The divalent-free solution contained (in mM): 145 NaCl, 2.8 KCl, 10 CsCl, 11 glucose, 10 EGTA, 10 HEPES, adjusted to pH 7.2 with NaOH. Patch-clamp experiments were conducted in the tight-seal whole-cell configuration (18) using an EPC-9 amplifier (HEKA Elektronik, Lambrecht, Germany). Patch pipettes were pulled from borosilicate glass (Kimax ® ) and had resistances between 2 and 3 megohms when filled with the standard internal solution containing (in mM): 145 cesium glutamate, 10 HEPES, 8 NaCl, 1 MgCl 2 , 2 Mg-ATP adjusted to pH 7.2 with CsOH. When using high intracellular chelator concentrations (30 and 60 mM), the osmolarity was kept constant by appropriate reduction of cesium glutamate. The intracellular Mg-ATP concentration was increased to 6 mM for the experiments in the absence of external divalent cations. This was done to minimize contamination of monovalent currents presumably mediated by endogenously expressed TRPM7 channels in HEK and RBL cells (19). The Mg 2ϩ -free pipette solution contained 10 mM EGTA, Na 2 -ATP instead of the Mg 2ϩ salt and no addition of MgCl 2 to study the time-dependent removal of the intracellular Mg 2ϩ block. The series resistances were typically in the range of 5-10 megohms and not electronically compensated. Currents were filtered using an 8-pole Bessel filter at 8 kHz and digitized at 100 s. Voltage-clamp recordings were performed using ramps (Ϫ110 mV to ϩ90 mV in 50 ms) applied every 2 s using PULSE software (HEKA Electronics) on a personal computer. Cells were held at Ϫ10 mV between ramps. Several parameters (capacitance, series resistance, holding current) were displayed simultaneously at a slower rate (2 Hz) using the X-Chart display (HEKA Electronics). The membrane potential values were corrected for 10-mV liquid junction potential. No additional voltage correction was performed for the experiments under divalent-free conditions. Effects of changes in surface charge screening were ignored. All experiments were carried out at room temperature (20 -23°C); internal solutions were kept on ice to minimize hydrolysis of the nucleotides.
Northern Blot Analysis-For Northern blot analysis, 2 g of poly(A) ϩ RNA from rat duodenum and from RBL cells were separated by electrophoresis on 0.8% agarose gels and thereafter transferred to Hybond N nylon membranes (Amersham Biosciences Europe, Freiburg, Germany) as described (7). The membranes were hybridized in the presence of 50% formamide at 42°C overnight. No additional signals were detected if the exposure time was 1 week. The probe was the 1539-bp ApaI cDNA fragment of rat TRPV6 (DDBJ/EMBL/GenBank TM , accession no. AF160798) encoding amino acid residues 208 -721 and the 345-bp EcoRI/BamHI fragment of human TRPV6 (amino acid residues 528 -643), both labeled by random priming with [␣-32 P]dCTP.
Data Analysis-Analysis was performed with PulseFit and programs written in the IGOR macro language (Wave Metrics, Lake Oswego, OR). Ca 2ϩ currents elicited by voltage ramps were leak-subtracted by subtracting either the first ramp in TRPV6-transfected cells or by averaging the first two to four ramps after establishing the whole-cell mode in untransfected cells (depending on how fast I CRAC developed) and then subtracting the mean from all subsequent traces. This type of analysis is well established for I CRAC measurements in RBL cells and was also used for TRPV6-expressing cells to allow a better comparison of the data in Figs. 2 and 3. No background current subtraction was performed for the data from TRPV6-expressing cells shown in Figs. 1A, 4A, and 5-8. The peak current amplitude was measured after leak subtraction at Ϫ80 mV for the data presented in the dose-response curve in Fig ] i at 60, 30, 10, 3, 1.5, 1, 0.5, and 0.1 mM EGTA, respectively). Throughout, average data are given as means Ϯ S.E. for n cells. Student's paired t test was used for comparison of means.
Drugs-All chemicals were purchased from Sigma, except BAPTA (Molecular Probes).

Current-Voltage Relationship from TRPV6-expressing HEK
Cells and Nontransfected RBL Cells-In TRPV6-transfected HEK cells, a large current was elicited by voltage ramps when [Ca 2ϩ ] i was buffered by 10 mM EGTA in the patch pipette ( Fig.  1). The inwardly rectifying current reversed at positive potentials (Ն30 mV, n ϭ 5). At potentials more negative to Ϫ40 mV the current showed a fast inactivation (time constant of 7.6 Ϯ 1.3 ms at Ϫ100 mV, n ϭ 5). No such current was recorded in the absence of external Ca 2ϩ from TRPV6-transfected HEK cells (data not shown, n ϭ 5). Mock and nontransfected cells exhibited a small background current with a linear I-V relationship (data not shown, n ϭ 5 each).
The biophysical properties of heterologously expressed TRPV6 channels are similar to those of endogenous storeoperated channels in RBL cells (Fig. 1B). Both currents show a strong inward rectification with a positive reversal potential, indicating a high Ca 2ϩ selectivity.
Store Depletion at High Intracellular Ca 2ϩ Buffering in TRPV6-expressing HEK and RBL Cells in Comparison to Untransfected Cells-It has been argued for the TRPV6 protein from rat that this channel is only store-operated when its expression level stands in a special relation to the native signal transduction machinery of CHO cells (12). This idea was tested by expressing the human clone in RBL cells, which show a much more pronounced I CRAC than CHO and HEK cells. Subsequently, RBL cells might offer a more native environment for store-operated channels than HEK cells. Whether this affects the gating of TRPV6 channels was tested in the following experiments. I CRAC is gated by the filling state of IP 3 -sensitive Ca 2ϩ stores. It can be activated by store depletion obtained by intracellular perfusion of IP 3 , the sarco/endoplasmic reticulum Ca 2ϩ -ATPase inhibitor thapsigargin, and/or EGTA. Using IP 3 (30 M) and EGTA (10 mM) in the internal solution resulted in rapid activation of I CRAC (Fig. 2). TRPV6-expressing HEK cells showed a rapid current activation followed by a slow inactivation until a steady-state level was reached (Fig. 2). This was also the case for TRPV6-expressing RBL cells; immediately after obtaining the whole-cell configuration, the Ca 2ϩ current developed, reached a maximum, and subsequently decayed to a lower level (Fig. 2).
Very similar results were obtained with thapsigargin (2 M) instead of IP 3 or both substances (Table I); I CRAC activated rapidly in RBL cells, no prominent current was detected in HEK cells, and a large inward current was measured in TRPV6-expressing HEK and RBL cells such as shown in Figs. 1 and 2. No significant differences between both cell types were detected with respect to the peak Ca 2ϩ current, activation, and inactivation kinetics. The activation of I CRAC by high concentrations of the Ca 2ϩ buffer EGTA (10 mM) in the intracellular solution is much slower than in the recordings with IP 3 and/or thapsigargin (Fig. 2). Strikingly, the activation kinetic of TRPV6-mediated Ca 2ϩ currents was not changed in both expression systems tested. Similar results were obtained with BAPTA instead of EGTA (data not shown). Taken together, these data suggest that the presence of store-depleting agents such as thapsigargin and IP 3 does not influence TRPV6 channel activation.
Store Depletion at Low Intracellular Ca 2ϩ Buffering in TRPV6-expressing HEK and RBL Cells in Comparison to Untransfected Cells-It is possible to record I CRAC under conditions of moderate cytoplasmic Ca 2ϩ buffering (20). With low concentrations of EGTA (1.5 mM) and the addition of IP 3 (30 M) in the pipette solution, peak current amplitude in RBL wild-type cells was similar to the recordings presented before (Fig. 2). However, current densities were dramatically decreased in TRPV6-transfected cells (Fig. 3) when compared with the recordings with higher EGTA concentrations (10 mM, Fig. 2).
This difference became more obvious when I CRAC was activated with thapsigargin (2 M) and IP 3 (30 M) under more physiological conditions of low intracellular Ca 2ϩ buffering (0.1 mM EGTA). Prominent I CRAC activation was recorded in RBL wild-type cells (Fig. 3), which was very similar to the recordings with higher EGTA concentrations (10 mM) under otherwise identical conditions (data not shown, n ϭ 5). Although store depletion was achieved as proved by the activation of I CRAC in RBL wild-type cells, no macroscopic current was detected in TRPV6-expressing HEK cells (Fig. 3). However, using 10 mM EGTA with thapsigargin and IP 3 in the intracellular solution induced a large Ca 2ϩ influx, as described earlier (Table I). Therefore, it is likely that the Ca 2ϩ current seen in TRPV6transfected RBL cells was I CRAC and that TRPV6 channels are not activated by thapsigargin and IP 3 .
TRPV6 Channel Activity Depends on the Intracellular Ca 2ϩ Chelator Concentration-Because TRPV6 channels activated independently to store-depleting agents, it is possible that the intracellular EGTA concentration is crucial for current activation. Various intracellular EGTA concentrations were tested on their effect to change TRPV6-mediated Ca 2ϩ entry (Fig. 4A). At low intracellular Ca 2ϩ buffering with 0.1 mM EGTA in the pipette solution, no current was activated. Increasing the EGTA concentration slightly up to 0.5 mM resulted in Ca 2ϩ current activation with a small current density. Higher concentrations of the chelator caused a further increase in current size. The relationship between Ca 2ϩ current amplitude and EGTA concentration in the intracellular solution is plotted as a dose-response curve in Fig. 4B. Interestingly, no saturation of the current amplitude was detected with 60 mM EGTA in comparison to the current densities recorded with 30 mM EGTA. No such behavior was detected in nontransfected HEK cells.
Immediately after establishing the whole-cell configuration, the current, measured from the first ramp at Ϫ80 mV, was not significantly different between TRPV6-expressing and nontransfected HEK cells, arguing against a maximally activated channel under resting conditions (Table II).
TRPV6 Channels Function as a Ca 2ϩ Sensor-The chelator EGTA is characterized by selectivity more than 5 orders of magnitude higher in chelating Ca 2ϩ over Mg 2ϩ (21). Therefore, mainly the [Ca 2ϩ ] i and to a lesser extent the [Mg 2ϩ ] i are decreased by EGTA. Thus, it is necessary to find out which ion is responsible for the EGTA-induced augmentation of TRPV6mediated Ca 2ϩ currents. In the previous experiments, the standard internal solution contained 1 mM MgCl 2 and 2 mM Mg-ATP, which in the presence of 0.1 mM EGTA translates to 1.05 mM Mg 2ϩ and is reduced to 0.64 mM in the presence of 60 mM EGTA (see "Experimental Procedures"). With the latter solution a prominent Ca 2ϩ current was detected, as shown in Figs. 4 and 5. Using the same pipette solution but elevating the [Mg 2ϩ ] i to 1.05 mM resulted in very similar Ca 2ϩ entry with respect to the activation kinetic and peak current amplitude (Fig. 5). However, no current activation was detected with an internal solution containing 1.05 mM [Mg 2ϩ ] i and 100 nM [Ca 2ϩ ] i (Fig. 5). Because chelators can induce effects unrelated to their Ca 2ϩ buffering properties (22), it was necessary to use a pipette solution with the same [Mg 2ϩ ] i and [Ca 2ϩ ] i but 60 mM EGTA instead of 10 mM. Again, no current developed, which makes it unlikely that EGTA activates TRPV6 channels by pharmacological means (data not shown, n ϭ 3). Furthermore, no inhibition of peak current amplitude was detected in the presence of 4.

TRPV6-mediated Ca 2ϩ Currents in the Presence of High
Intracellular Concentrations of Either EGTA or BAPTA-Prominent TRPV6-mediated Ca 2ϩ entry was detected at a calculated global [Ca 2ϩ ] i in the picomolar range. Because the physiological [Ca 2ϩ ] i is not that low, it is tempting to speculate that higher levels were achieved within a microdomain of elevated Ca 2ϩ near the channel. This idea was tested by using intracellular BAPTA instead of EGTA. BAPTA has a faster Ca 2ϩ binding kinetic than EGTA (20), resulting in more effective competition with the slow endogenous Ca 2ϩ buffers. High chelator concentrations (30 or 60 mM) were dialyzed into the cells to avoid buffer saturation on channel opening in the presumed microdomain. Under these conditions Ca 2ϩ currents were larger and showed less slow inactivation with BAPTA in comparison to EGTA, despite the fact that Ca 2ϩ entry was elicited by short voltage-ramps of 50-ms duration, which were applied only once every 2 s (Fig. 6). These differences in the time courses of TRPV6-mediated Ca 2ϩ influx was also seen with chelator concentrations of 60 mM instead of 30 mM (data not shown; n ϭ 7 for 60 mM BAPTA and n ϭ 10 for 60 mM EGTA). Hence, it is likely that local Ca 2ϩ gradients control TRPV6 channels rather than the overall [Ca 2ϩ ] i .

TRPV6-mediated Monovalent Ca 2ϩ Currents When Divalent Cations Are Removed from the Extracellular Solution-
The presumable built-up of subplasmalemmal Ca 2ϩ gradients can be abolished by recording monovalent currents in the absence of extracellular divalent cations. Fig. 7 illustrates the currentvoltage relationship under these conditions where TRPV6 channels were activated with 10 mM intracellular EGTA. The I-V curves showed a pronounced inward rectification of a large TRPV6 Channels Are Ca 2ϩ Sensors current with a slightly positive reversal potential. During hyperpolarizing steps the current activated slowly, which results in a negative slope at potentials roughly below Ϫ80 mV during the ramp protocol, apparently because of a time-dependent removal of an intracellular Mg 2ϩ block (data not shown, n ϭ 8; Ref. 13). However, no such behavior was observed for monovalent currents, which reversed at more positive potentials in nontransfected RBL cells after I CRAC activation by IP 3 (30 M) and EGTA (10 mM; Fig. 7B). Dialysis of 10 mM EGTA into a TRPV6-expressing HEK cell resulted in rapid activation of Ca 2ϩ currents followed by inactivation (Fig. 8). When the external Ringer's solution was substituted by a divalent-free saline, inward currents were ini-tially blocked in an anomalous fashion before large monovalent currents gradually developed and remained quite stable with time. Thus, the time course of monovalent current activation and inactivation was relatively slow in comparison to the fast development and run-down of Ca 2ϩ currents. When switching back to the normal bath solution, monovalent currents rapidly decayed in opposition to their slow development under divalent-free conditions. The negative correlation of [Ca 2ϩ ] i and TRPV6 channel activity was shown for Ca 2ϩ currents and also remained valid for monovalent currents; under divalent-free conditions, currents increased in parallel to an increase of intracellular EGTA concentrations (Ϫ97 Ϯ 27 pA/pF, n ϭ 6 for 1 mM EGTA; Ϫ195 Ϯ 84 pA/pF, n ϭ 6 for 10 mM EGTA; and Ϫ353 Ϯ 96 pA/pF, n ϭ 4 for 60 mM EGTA). Similar currents were not detected in HEK wild-type cells (data not shown), whereas monovalent currents in nontransfected RBL cells reached their maximum amplitude immediately after application of divalent-free solution and subsequently decayed (data not shown). Apparently these currents comprise a completely different kinetic compared with TRPV6-mediated currents (13) and are supposed to occur through CRAC channels endogenously present in RBL cells. In summary the experiments under divalent-free condition with various intracellular EGTA concentrations suggest that TRPV6 channels are blocked by extracellular divalent cations and subplasmalemmal Ca 2ϩ .
Northern Blot Analysis for HEK and RBL Cells-TRPV6mediated Ca 2ϩ currents behaved very similarly if expressed either in HEK or RBL cells, although store-operated Ca 2ϩ influx is much more prominent in RBL than in HEK cells. Northern blot experiments were, therefore, performed to reveal TRPV6 transcript expression. Using a human TRPV6 cDNA as a probe, no transcripts could be detected in nontransfected HEK cells (data not shown). With the rat cDNA as probe, the 3.0-kb transcript of rat TRPV6 (10) was readily detected in poly(A) ϩ from rat duodenum but not in poly(A) ϩ from RBL cells (Fig. 9), indicating no expression of TRPV6 in these cells. DISCUSSION Ca 2ϩ currents mediated by recombinant TRPV6 and endogenous CRAC channels had similar biophysical properties and could not clearly be discriminated by their current-voltage relationship. The data presented here allow one to distinguish both currents by their current-voltage relationship in divalentfree saline and by their activation mechanism; TRPV6-mediated Ca 2ϩ currents were strongly dependent on [Ca 2ϩ ] i and were significantly augmented by decreasing [Ca 2ϩ ] i . However, I CRAC activated by store depletion, which was without effect on the gating of TRPV6 channels. Interestingly, current amplitudes of TRPV6-expressing cells did not saturated but continuously increased if very high concentrations of Ca 2ϩ chelators were used in the intracellular solution (30 and 60 mM EGTA or BAPTA). The calculated global [Ca 2ϩ ] i was several magnitudes below physiological values (see "Experimental Procedures"). Therefore, it is likely that TRPV6 channels sense the local intracellular Ca 2ϩ concentration in the close proximity of the pore.
Voltage-activated EAG K ϩ channels undergo Ca 2ϩ -mediated inhibition in situations where K Ca channels start to become activated (23). The ubiquitous Ca 2ϩ -binding protein CaM was identified as the Ca 2ϩ sensor protein that activates K Ca channels in response to rising [Ca 2ϩ ] i but closes EAG channels. This reverse action is regulated by CaM binding to the C terminus of hEAG1 (24) and to the ␣-subunit of small conductance K Ca channels (25). Interestingly the TRPV6 protein also binds CaM in a Ca 2ϩ -dependent manner (8), which makes it an attractive candidate for the Ca 2ϩ sensor.
It has been postulated that a decrease in [Ca 2ϩ ] i near the plasma membrane in response to store depletion leads to activation of I CRAC (26). Indeed it could be shown for RBL cells that low [Ca 2ϩ ] i in the range of 30 -50 nM activates I CRAC spontaneously and independently of global Ca 2ϩ store depletion, whereas elevation of [Ca 2ϩ ] i up to 100 nM resulted in store-dependent gating of I CRAC (27). Cell dialysis with low [Ca 2ϩ ] i might passively deplete subtypes of Ca 2ϩ stores, which are presumably responsible for I CRAC activation. Unfortunately it was not possible to resolve these CRAC stores with mag fura-2. The activation properties of recombinant TRPV6 channels presented here were similar to those reported by Krause et al. (27) for I CRAC with respect to the spontaneous activation at very low [Ca 2ϩ ] i . However, no Ca 2ϩ current increase was detected for TRPV6 channels if stores were depleted with thapsigargin and/or IP 3 at 100 nM free Ca 2ϩ in the pipette solution (data not shown).
The recombinant TRPV6 channel from rat (12) shows electrophysiological properties similar to those reported here and previously for the human channel (7) e.g. the high Ca 2ϩ selectivity, anomalous mole fraction effect, block by inorganic cations, and loss of selectivity in the absence of divalents. The activation kinetics of rat TRPV6 channels were unaltered if store-depleting agents such as IP 3 and thapsigargin were added to the EGTA (10 mM) containing pipette solution. Furthermore, no current activation was detected under conditions of weak intracellular Ca 2ϩ buffering (0.05 and 0.5 mM EGTA). Very similar results were recorded for human TRPV6 channels in  Ϫ3.0 Ϯ 0.9 (5) Ϫ2.5 Ϯ 0.3 (6) 1.5 mM EGTA Ϫ3.0 Ϯ 0.6 (5) Ϫ2.9 Ϯ 0.8 (5) TRPV6 Channels Are Ca 2ϩ Sensors the present study (Table II). On the other hand, it has been suggested that TRPV6 and the closely related TRPV5 protein form constitutively active Ca 2ϩ channels (7,11). This conclusion is based on recordings where intracellular Ca 2ϩ buffering capacities were relatively high (11) or only considered in part (7). When IP 3 and thapsigargin were included under conditions of low Ca 2ϩ buffering, rat TRPV6 channels were activated in CHO cells transfected 8 -12 h previously. One possible explanation is a mismatch between the protein expression levels and the still unknown signal transduction apparatus that senses store depletion (12). This possibility was tested for human TRPV6 channels by using different expression systems. RBL cells have a pronounced store-operated Ca 2ϩ entry and are therefore extensively used to study I CRAC . HEK cells represent a common model system to express foreign cDNAs, and it has recently been suggested that store depletion also activates I CRAC in these cells (14 -16). However, current densities were extremely small (Ϫ0.2 pA/pF at 0 mV with 20 mM external Ca 2ϩ ) and only 20% of that measured in Jurkat T-lymphocytes, where I CRAC reaches peak amplitudes comparable with those in RBL cells. Accordingly, one can assume that the signal transduction machinery that leads to I CRAC activation is expressed to a higher degree in RBL than HEK cells. However, no cellspecific modulation of TRPV6 channel function was found 24 -32 h after transfection in both HEK and RBL cells.
The kinetic of human TRPV6 channel activation remained unaltered if Ca 2ϩ stores were either actively or passively depleted (Fig. 2). Peak current densities were slightly larger in RBL than HEK cells when store depletion was achieved by IP 3 and/or thapsigargin (Table I). These values were measured ϳ1 min after establishing the whole-cell configuration and reflect Ca 2ϩ currents from TRPV6 and CRAC channels, which activate on a similar time scale. The small difference in current density might, therefore, be the result of larger I CRAC in RBL cells. Using only EGTA in the pipette solution resulted in rapid activation of TRPV6-mediated Ca 2ϩ currents and slow activation of I CRAC (Fig. 2). The maximum Ca 2ϩ entry was determined by TRPV6 channels so that the peak current was measured early in the experiments and without contamination of I CRAC . Subsequently the difference in current size was smaller between RBL and HEK cells in comparison with the data with thapsigargin and/or IP 3 (Table I). However, no significant differences were detected as a result of the broad fluctuations of these Ca 2ϩ currents, which is at least partially a consequence of variations in the expression levels.
Yue et al. (12) concluded that TRPV6 comprises all or part of the CRAC pore, whereas Voets et al. (13) reported that the TRPV6 protein and CRAC channels are not identical. In the latter study, no differences were detected between TRPV6 channel activation in dependence to Ca 2ϩ current densities. In these experiments, currents were measured within 12 h after transfection and selected for HEK cells with small current densities (below 100 pA/pF at Ϫ100 mV with 10 mM external Ca 2ϩ and 10 mM EGTA in the internal solution). Interestingly, these authors recorded a large inward current with 100 nM free Ca 2ϩ in the intracellular solution. The amplitude was largest immediately after establishing the whole-cell mode and rapidly decreased until a steady-state level was reached. No such activation kinetic was seen for human TRPV6 channels in HEK cells at the same Ca 2ϩ concentration, despite the fact that voltage ramps were applied immediately after establishing the whole-cell mode. It remains to be determined whether structural differences among TRPV6 channels from different species are responsible for the heterogeneous electrophysiological find-ings observed. The human TRPV6 protein has 90% identity to mouse TRPV6 (BC016101) used by Voets et al. (13) and 89% identity to rat TRPV6 (NM_053686) studied by Yue et al. (12). TRPV6 proteins from mouse and rat share 98% identical amino acid residues.
The rabbit TRPV5 protein, which shares 73% identical amino acid residues with human TRPV6, has been implicated to play a role in Ca 2ϩ reabsorption by the kidney and intestinal epithelial cells. It appears to be colocalized with 1,25-dihydroxyvitamin D 3 -dependent calbindin-D 9k and/or calbindin-D 28k (28,29), which might enhance Ca 2ϩ currents as a result of a relief of Ca 2ϩ -induced inhibition (30). Although it is not clear whether TRPV6 is expressed in human kidney and small intestine (7,9), it will be interesting to find out whether differences in the endogenous Ca 2ϩ buffer capacity affects TRPV6 channel gating under in vivo conditions.
In summary, TRPV6-mediated Ca 2ϩ currents and I CRAC have several features in common such as the high Ca 2ϩ selectivity and the similar I-V relationship. However, although I CRAC is store-operated, no such activation mechanism was found in TRPV6-expressing HEK and RBL cells where TRPV6 channels function as Ca 2ϩ sensors. Differential suppression of TRPV6-mediated Ca 2ϩ currents was achieved under conditions of weak intracellular Ca 2ϩ buffering. With only 0.1 mM EGTA in the intracellular solution, I CRAC was activated with IP 3 and thapsigargin and showed very similar properties in TRPV6expressing and nontransfected RBL cells. Therefore, the contribution of recombinant TRPV6 channels to store-operated Ca 2ϩ entry in HEK and RBL cells appeared to be minor. TRP proteins are believed to form homo-and/or heterooligomeric channels in vivo, and it has been suggested that the gating of TRP channels depends on the level of channel protein expression (12,31). Assuming that the TRPV6 protein is, or is a part of, the CRAC pore (12) and is endogenously present in RBL cells at only very low levels (12), it is interesting that overexpression of TRPV6 in RBL cells, as shown here, appears not to interfere with the activity of endogenous CRAC channels because it was possible to discriminate between Ca 2ϩ currents mediated by either CRAC or TRPV6 proteins. These findings might indicate a close stoichiometric coupling between subunits that form the CRAC channel complex and its activation machinery.