Activation of TRPM7 Channels by Phospholipase C-coupled Receptor Agonists*

TRPM7 is a ubiquitously expressed nonspecific cation channel that has been implicated in cellular Mg2+ homeostasis. We have recently shown that moderate overexpression of TRPM7 in neuroblastoma N1E-115 cells elevates cytosolic Ca2+ levels and enhances cell-matrix adhesion. Furthermore, activation of TRPM7 by phospholipase C (PLC)-coupled receptor agonists caused a further increase in intracellular Ca2+ levels and augmented cell adhesion and spreading in a Ca2+-dependent manner (1). Regulation of the TRPM7 channel is not well understood, although it has been reported that PIP2 hydrolysis closes the channel. Here we have examined the regulation of TRPM7 by PLC-coupled receptor agonists such as bradykinin, lysophosphatidic acid, and thrombin. Using FRET assays for second messengers, we have shown that the TRPM7-dependent Ca2+ increase closely correlates with activation of PLC. Under non-invasive “perforated patch clamp” conditions, we have found similar activation of TRPM7 by PLC-coupled receptor agonists. Although we could confirm that, under whole-cell conditions, the TRPM7 currents were significantly inhibited following PLC activation, this PLC-dependent inhibition was only observed when [Mg2+]i was reduced below physiological levels. Thus, under physiological ionic conditions, TRPM7 currents were activated rather than inhibited by PLC-activating receptor agonists.

TRPM7 is a ubiquitously expressed nonspecific cation channel that, intriguingly, contains a C-terminal serine-threonine kinase domain. It belongs to the transient receptor potential melastatin-related (TRPM) 2 subfamily of TRP channels that transduce sensory signals (2). TRPM7 functioning appears essential for life in that both knock-out and overexpression of the channels cause growth arrest, loss of cell adhesion, and rapid cell death (3)(4)(5). We recently discovered that low overexpression of TRPM7 induces cell spreading and adhesion and that its activation by PIP 2 -hydrolyzing receptor agonists leads to the formation of adhesion complexes in a kinase-dependent manner (1).
Currents carried by TRPM7 channels exogenously expressed in mammalian cells have been analyzed by several groups. In physiological solutions, the channel conducts mainly Ca 2ϩ and Mg 2ϩ (6), but in the absence of these divalent cations, K ϩ and Na ϩ (3, 7) permeate efficiently. A characteristic feature is the inhibition of TRPM7 currents by physiological (1-2 mM) intracellular Mg 2ϩ levels; in whole-cell patch clamp experiments, large outwardly rectifying TRPM7 currents (3,7,8) are evoked by perfusion with Mg 2ϩ -free pipette solutions. Furthermore, MgATP and MgGTP also inhibit the channels (3,9), although some controversy was raised on this issue (10). TRPM7 currents have been termed MagNuM (for Mg 2ϩ nucleotide-regulated metal ion (3)) or MIC (for Mg 2ϩ -inhibited cation (10)) currents. These terms will here be used interchangeably to reflect whole-cell currents evoked by internal Mg 2ϩ depletion. MIC/MagNuM currents revert at about 0 mV and lack voltageand time-dependent activation (3,7). Inward currents are predominantly carried by divalent cations, whereas outward currents consist mainly of monovalent cations (at low [Mg 2ϩ ] i ). Outward rectification is most likely due to divalent permeation block of inward currents at negative potentials (3), because perfusion with divalent-free extracellular solutions augments inward currents and linearizes the I/V relationship. The activation of TRPM7 by internal perfusion with Mg 2ϩ -free solutions does not reflect the release of the permeation block, but the precise mechanism has not yet been solved (10). Several reports document that the set point for [Mg 2ϩ ] i sensitivity is governed by the kinase domain; however, TRPM7 channels lacking the kinase domain can still be activated by internal Mg 2ϩ depletion (4,11,12). Thus, the interactions of TRPM7 with Mg 2ϩ are complex; Mg 2ϩ is conducted through the channel pore, causes voltage-dependent permeation block, and influences gating at the cytosolic surface.
The exact mechanisms by which receptor agonists regulate TRPM7 are less well characterized, and the published data are, at least partly, conflicting. An initially claimed indispensable role for the kinase domain (7) was challenged in subsequent studies (4,(11)(12)(13). We have recently shown (1) that the TRPM7 ␣-kinase specifically phosphorylates the heavy chain of myosin-II, thereby strongly influencing cell adhesion. Importantly, association with and subsequent phosphorylation of myosin-II depend on prior activation of the channel by phospholipase C (PLC)-coupled receptors, and influx of extracellular Ca 2ϩ constitutes an essential step in this process (1). In accordance with this, TRPM7 binds directly to several PLC isoforms, including PLC␥ and PLC␤ (14). The stimulatory effect of PLC on TRPM7 observed in intact cells by biochemical, cell-biological, and live cell-imaging studies contrasts with a report that, in human embryonic kidney 293 (HEK293) cells, whole-cell TRPM7 currents are inhibited by PIP 2 hydrolysis (14). PIP 2 -dependent gating also occurs in other TRP family members, including TRPV1 (15), TRPM4 (16,17) TRPM5 (18), and TRPM8 (19,20). Finally, Takezawa and colleagues (13) recently reported that, in HEK-293 cells (expressing only endogenous muscarinic receptors), carbachol attenuated TRPM7 currents via the G s -cAMP signaling pathway, whereas PLC activation was not involved.
Given the discrepancies between the documented whole-cell patch clamp results and our cell biological observations, we re-examined how PLC-activating receptor agonists affect TRPM7 currents using non-invasive techniques in cells that moderately overexpress TRPM7. To this end, we combined fluorescence resonance energy transfer (FRET) assays for second messengers with perforated patch experiments and Ca 2ϩ fluorometry to show that opening of TRPM7 channels closely correlates with PLC activation but not with cAMP/cGMP signaling. We also showed that, in perforated patches, TRPM7 currents are evoked by treatment of intact cells with a membrane-permeable Mg 2ϩ chelator (EDTA-AM) and that these currents are inhibited rather than augmented by PLC-activating receptor agonists. We conclude that PLC-coupled agonists activate rather than inhibit TRPM7 in intact cells.
Constructs-TRPM7 constructs were as described previously (1). The PIP 2 FRET sensor (eCFP-PH␦1 and eYFP-PH␦1) was previously generated in our laboratory (21). Sensors for cAMP and cGMP were kind gifts from Dr. M. Zaccolo (University of Padova, Padova, Italy) and Dr. W. Dostmann (University of Vermont, Burlington, VT), respectively, and Yellow Cameleon 2.1 was a gift from Dr. R. Tsien (University of California-San Diego, La Jolla, CA).
Immunostaining and Kinase Reaction-A TRPM7 antibody was raised to amino acids 1748 -1862 in rabbits as detailed in Ref. 1. Preimmune serum was used as a control in all experiments. For immunolabeling, cells were fixed with 4% paraformaldehyde in phosphate buffer, permeabilized with 0.1% Triton X-100, and incubated with rabbit anti-TRPM7 sera (1:200) fol-lowed by horseradish peroxidase-conjugated anti-rabbit Ig (1:1000). Amplification was by tyramide-conjugated fluorescein isothiocyanate (PerkinElmer Life Sciences). For the in vitro kinase assay, TRPM7 was precipitated from lysed cells with anti-TRPM7 antibodies and assayed for kinase activity as published recently (1).
Cell Culture, Fluorimetric Experiments-Culture of mouse N1E-115 and Phoenix packaging cells was as described previously (1). Ratiometric and pseudoratiometric Ca 2ϩ recordings on cells on glass coverslips were carried out essentially as published in Refs. 21-23 in HEPES-buffered saline, pH 7.2, at 37°C. All traces were calibrated with ionomycin and BAPTA as published previously (23). Dynamic FRET essays were carried out as described previously (21,24). Excitation was at 425 nm using an ND3 filter, and cyan and yellow fluorescent protein emissions were collected simultaneously at 470 Ϯ 20 and 530 Ϯ 25 nm, respectively. Data were acquired at 4 Hz, and FRET was expressed as the ratio of cyan to yellow fluorescent protein signals. This ratio was set to 1.0 at the onset of the experiments, and changes are expressed as the percent of deviation from this initial value.
Patch Clamp Experiments-Electrophysiological recordings were collected using the HEKA EPC9 system. Current recordings were digitized at 100 kHz (ramp and block pulse protocols) or 10 Hz (steady-state whole-cell currents). Borosilicate glass pipettes were fire-polished to 2-4 M⍀. After establishment of the G⍀ seal, the patched membrane was ruptured by gentle suction to obtain whole-cell configuration, or amphotericin B (240 g/ml) was used to obtain the perforated patch configuration with the typical access resistance of 3-10 M⍀.
Solutions were (in mM) 120 whole-cell pipette potassiumglutamate, 30 KCl, 1 MgCl 2 , 0.2 CaCl 2 , 1 EGTA, 10 HEPES, pH 7.2, and 1 MgATP; 140 external solution NaCl, 5 KCl, 0 -1 MgCl 2 , 0 -10 CaCl 2 , 10 HEPES, and 10 glucose adjusted to pH 7.3 with NaOH; for perforated patch recordings, the pipette solution was complemented with 240 g/ml amphotericin B, and MgATP was omitted. (3), we found that transient overexpression of TRPM7 channels was lethal to N1E-115 cells and several other cell lines tested within a few days (data not shown). Furthermore, at early time points after transfection, much of the TRPM7 protein was found localized to biosynthetic compartments. To circumvent these problems, we expressed TRPM7 in N1E-115 cells at low levels by retroviral transduction (1). These cells, termed N1E-115/TRPM7 cells, were viable, divided normally, and could be routinely kept in culture for several months (1). Using a TRPM7-specific antibody (1), the expression level in N1E-115/ TRPM7 cells was found to be 2-3 times higher than that in the parental cells (Fig. 1a). Under these conditions, the TRPM7 mainly localized to the plasma membrane, as shown in Fig. 1b. Moreover, depletion of intracellular Mg 2ϩ in whole-cell patch clamp recordings showed that N1E-115/TRPM7 has ϳ2.5 times higher current density of MIC/MagNuM currents than wild type (wt) cells (supplemental Fig. S1).
Because cell biological and biochemical data (1) indicate that TRPM7 channels are activated by bradykinin (BK), we monitored the effect of the addition of BK on TRPM7-mediated Ca 2ϩ influx by comparing N1E-115/TRPM7 to parental cells. In parental cells, 1 M BK induced a cytosolic Ca 2ϩ increase (765.9 Ϯ 9.7 nM, n ϭ 80) that peaked within seconds and subsequently returned to values close to resting levels (92.9 Ϯ 2.9 nM) within 60 s (Fig. 1d, left panel). This corresponds to a mean increase of 7.6 nM from basal levels, which was not statistically significant. In contrast, in N1E-115/TRPM7 cells, the BK-induced transient Ca 2ϩ increase (peak 843.3 Ϯ 12.7 nM) was followed by a prominent sustained Ca 2ϩ elevation (141.0 Ϯ 7.0 nM, n ϭ 120) that lasted for several minutes before returning to resting levels rather abruptly (Fig. 1d, right panel; quantification in Fig. 1c). This corresponds to a mean increase in sustained Ca 2ϩ levels of 32.1 nM (p Ͻ 0.001, paired t test) in accordance with the higher TRPM7 expression levels in these cells. Thus, in N1E-115/TRPM7 cells, the addition of BK elicits a sustained phase in the Ca 2ϩ response, in good agreement with our cell biological data (1). In summary, the observed increase in Ca 2ϩ influx is mediated by TRPM7 channels and is not due to the retroviral transduction procedure (supplemental Fig. S2).
TRPM7 Activation Is Downstream of PLC in Intact N1E-115 Cells-Endogenous B2 bradykinin receptors in N1E-115 cells signal predominantly via the G q -linked PLC pathway (27), but some reports suggest that, depending on the cell type, the B2 receptors may occasionally either inhibit (28) or stimulate (29) the production of cAMP. We set out to identify the signaling events responsible for BK-induced activation of TRPM7 in N1E-115/TRPM7 cells by correlating Ca 2ϩ fluorometry with the activation of intracellular signaling pathways, as detected by various FRET assays.
BK caused rapid breakdown of a significant fraction (60 -80%) of the plasma membrane PIP 2 pool, as detected by a FRET assay (21) that reports membrane PIP 2 content (Fig. 2a, upper  panel, first trace). The G q /PLC-coupled receptor agonist lysophosphatidic acid and thrombin receptor-activating peptide also activate PLC, although to a lesser extent (ϳ20 -30% of BK values; n Ͼ 200) (Fig. 2a, upper panel, second and third trace (21)). Similar to BK, the initial Ca 2ϩ peak induced by these agonists was followed by sustained Ca 2ϩ influx in N1E-115/ TRPM7 cells (Fig. 2a, lower panel). In contrast, sustained Ca 2ϩ influx was not seen when cells were stimulated with agonists of receptors that do not couple to PLC, such as prostaglandin E1 (Fig. 2a, 4th trace, and data not shown). Thus, TRPM7 activation correlates well with PLC activation/Ca 2ϩ signaling.
To address the possible involvement of cAMP in the BKinduced opening of TRPM7, we monitored cAMP levels in intact N1E-115/TRPM7 cells using a genetically encoded cAMP sensor (30,31). The addition of BK to N1E-115/TRPM7 cells had no effect on cAMP levels (n ϭ 8), whereas forskolin (25 M) readily raised cAMP levels (Fig. 2b, left panel). Furthermore, pretreatment of cells with pertussis toxin to specifically inhibit G i and thereby block receptor-induced decreases in cAMP levels did not affect the BK-induced Ca 2ϩ influx (data not shown). Therefore, in N1E-115/TRPM7 cells, changes in cAMP levels do not mediate the BK-induced opening of TRPM7. In addition, neither prostaglandin E1, which activates G s to cause a rapid and sustained increase in [cAMP] i (Fig. 2b, middle panel), nor sphingosine-1-phosphate (S1P, data not shown), which couples to G i and G 13 but not to PLC in N1E-115 cells (32), had any effect on Ca 2ϩ levels. We conclude that there is no evidence for a role of cAMP in BK-mediated Ca 2ϩ influx in N1E-115/TRPM7 cells.
We also investigated the effects of the nitric oxide donor nitroprusside, because nitric oxide was reported to activate TRPM7-mediated Ca 2ϩ influx in cultured cortical neurons (33). In intact parental and N1E-115/TRPM7 cells, nitroprusside triggered the production of cGMP (Fig. 2c, left panel (24)) without affecting [Ca 2ϩ ] i (Fig. 2c, right panel). In conclusion, TRPM7 opening closely correlates with PLC activation but not with other G protein-linked signals.
Activation of PLC Inhibits Whole-cell TRPM7 Currents in N1E-115/TRPM7 Cells-In whole-cell patch clamp experiments using HEK-293 cells overexpressing M1 muscarinic receptors, TRPM7 channels were shown to be inhibited by carbachol-induced PIP 2 breakdown (14). This inhibition was reverted to by intracellular perfusion with a water-soluble PIP 2 analogue. However, our cell biological observations (1) and Ca 2ϩ data (see above) show that PLC activation causes opening rather than closure of TRPM7 channels. We therefore tested whether voltage ramp-induced TRPM7 currents, recorded in whole cells with Mg 2ϩ -free pipette solution, were similarly sup-pressed by PLC activation in our cells. Indeed, BK rapidly suppressed outward-rectifying whole-cell TRPM7 currents both in native N1E-115 cells (data not shown) and N1E-115/TRPM7 cells (Fig. 3a). Although inward currents were completely abolished (99.7 Ϯ 1.0% of control values, n ϭ 9), small outward currents were still observed at high depolarizing potentials (Fig. 3b, trace 3 and see Fig. 6b). Other PLC-coupled agonists, including TRP and lysophosphatidic acid, also inhibited the currents, although to a lesser extent. Therefore, it appears that the original observations on PIP 2 dependence of whole-cell TRPM7 currents hold true for N1E-115/TRPM7 cells.
PLC-coupled Receptor Agonists Activate TRPM7 Currents in Intact Cells-The above observations leave us with a paradox; PLC activation inhibits TRPM7 currents when using whole-cell electrophysiology, whereas it opens the channels when assayed by Ca 2ϩ fluorometry. What causes this difference? In wholecell patch clamp experiments, diffusible signaling molecules will be rapidly diluted, which could influence channel gating properties. We therefore analyzed the electrophysiological effects of PLC activation using the perforated patch configuration (34,35), which provides electrical access to the cell without disturbing the cytosolic composition. Voltage-clamped perforated patches were held at Ϫ70 mV, and membrane conductance was monitored from currents evoked by a biphasic block pulse protocol (ϩ10 and Ϫ10 mV from resting potential (Fig.  4). In unstimulated cells, currents were small (0.46 Ϯ 0.16 pA/pF, n ϭ 10), corresponding to a membrane conductance of 23 pS/pF. Note that because Mg 2ϩ levels remain intact in this configuration, the magnitude of the currents should match those of whole-cell experiments when Mg 2ϩ is included in the pipette. Indeed, as shown in supplemental Fig. S1a, lower panel, the currents induced by 120-mV voltage steps in unstimulated whole cells measured 3.1 Ϯ 0.4 pA/pF (n ϭ 5), which corresponds to ϳ26 pS/pF (in close agreement with the perforated patch data). As expected, spontaneously developing currents were never observed in perforated patches (n ϭ 12) (see Fig. 5a).
Strikingly, stimulation with bradykinin significantly increased the membrane conductance (250 Ϯ 53 pS, n ϭ 5) The addition of the adenylyl cyclase activator forskolin together with the phosphodiesterase inhibitor isobutylmethylxanthine serves as a positive control. Conversely, prostaglandin E1 strongly activates the G s /cAMP pathway through its cognate G protein-coupled receptor (middle panel) but fails to elicit the sustained Ca 2ϩ elevation (right panel) that is typically seen after G q /PLC activation. c, in N1E-115/TRPM7 cells, BK fails to alter cGMP levels, whereas the nitric oxide donor nitroprusside readily activates this pathway, as assessed using Cygnet, a FRET sensor for cGMP (41). The addition of nitroprusside had no effect on [Ca 2ϩ ] i in these cells (right panel). ( Fig. 4b) in N1E-115/TRPM7 cells. In all cases, currents were transient, reverting to baseline with somewhat variable kinetics (range, 2-8 min). Similarly, the addition of BK slightly but significantly augmented the endogenous TRPM7 current in parental cells (Fig. 4a). Perforated patch I/V plots of BK-induced currents in wt and N1E-115/TRPM7 cells revealed almost linear currents (Fig. 4, a and b, right panels), probably because of the divalent permeation block of outward currents. The magnitude of these currents correlates with the TRPM7 expression levels, and they are inhibited by the same panel of inhibitors (see Table S1). We found that the distinction between PLC-induced inhibition (in whole cells) versus stimulation (in perforated patches) holds true for TRPM7-transfected HEK-293 cells as well (see supplemental Fig. S3). We conclude that a prototypic PLC-activating receptor agonist (bradykinin) activates TRPM7 channels under perforated patch clamp conditions.
Chelation of Intracellular Mg 2ϩ Mimics Whole-cell TRPM7 Currents in Perforated Patches-To further investigate the differences in TRPM7 current behavior under different recording conditions, we sought to lower [Mg 2ϩ ] i in intact cells. Cells were pretreated with EDTA-AM, a membrane-permeable precursor of the Mg 2ϩ /Ca 2ϩ chelator EDTA. Upon removal of its lipophilic AM moiety by cytosolic esterases, the compound becomes trapped in the cytosol and attains its chelating properties. Although in the perforated patch no spontaneous run-up of TRPM7 currents was observed (Fig. 5a), exposure of N1E-115/TRPM7 cells to 10 M EDTA-AM consistently evoked large currents with outward rectifying I/V relationships, identical to those obtained in whole cells (Fig.  5b). The slow onset of this current likely reflects the slow development of Mg 2ϩ buffering during buildup of the intracellular EDTA concentration. As a control, treatment with BAPTA-AM, a highly Ca 2ϩ -selective membrane-permeable chelator, had no effect (Fig. 5d). Physiological and pharmacological properties of EDTA-AM-evoked perforated patch currents were identical to whole-cell TRPM7 currents (see supplemental Fig. S4 and Table S1).
To date, most studies on TRPM7 have been performed on cells transiently overexpressing the channels, commonly HEK-293 or Chinese hamster ovary cells. We therefore tested whether EDTA-AM also activates TRPM7 channels in perforated patch experiments on HEK-293 cells. Indeed, heterologously expressed as well as endogenous TRPM7 channels were strongly activated by EDTA-AM treatment ( Fig. 5c; for statistics, see the legend). Importantly, application of EDTA-AM in Ca 2ϩ -fluorimetric experiments also opened the channels, as evidenced by increased [Ca 2ϩ ] i (detected using a plasma membrane-targeted version of the Ca 2ϩ FRET sensor Cameleon), whereas again BAPTA-AM had no effect (Fig. 5d). Thus, the use of EDTA-AM provides a novel and reliable method to evoke TRPM7/MIC currents in intact cells.

Effects of Agonist-induced PIP 2 Breakdown on EDTA-AMinduced and Whole-cell MIC/MagNuM Currents-How does
PLC activation affect EDTA-AM-induced TRPM7 currents in perforated patches? Strikingly, under these conditions, stimulation with BK caused a large fraction of TRPM7 channels to close rapidly (Fig. 6, a and b). This effect was rapid and transient rather than sustained in all cells tested. Rapid inhibition was also observed in parental N1E-115 cells, and it was not specific for BK, in that lysophosphatidic acid and thrombin receptoractivating peptide had the same effect (data not shown).
In contrast, in the whole-cell experiments of Runnels et al. (14), PIP 2 hydrolysis caused irreversible closure of whole-cell TRPM7 currents. We therefore compared the effects of the addition of G protein-coupled receptor agonists on PIP 2 levels in whole-cell and perforated patch experiments in N1E-115/ TRPM7 cells. Automated analysis of confocal time lapse images (21) show that the PIP 2 indicator GFP-PH(PLC␦1) was retained at the plasma membrane during intracellular accumulation of EDTA-AM (data not shown). Subsequent stimulation with BK resulted in fast translocation of GFP-PH to the cytosol (Fig. 6c,  right panel) followed by its relocation to the plasma membrane within a few minutes. The quantification in Fig. 6d shows that 5 min after BK stimulation recovery is almost complete. In contrast, in whole-cell experiments with N1E-115/TRPM7 cells, the BK-induced GFP-PH translocation was sustained (Fig. 6c,  left panel), in good agreement with published data for HEK-293 cells (14).
Thus, the time course of the inhibition of Mg 2ϩ depletioninduced TRPM7 currents correlates well with that of the loss of PIP 2 following receptor activation. The data also strongly suggest that lack of recovery of BK-induced TRPM7 currents in whole-cells is due to impaired PIP 2 resynthesis in this configuration.

DISCUSSION
In this study, we have addressed the regulation of TRPM7 channels in intact cells. Transient overexpression of TRPM7 was avoided because this caused mislocalization of TRPM7 to endomembranes and cell death because of the constitutive activity of TRPM7. Furthermore, transient overexpression is also more likely to titrate out essential binding partners and thereby may influence signaling events. Indeed, Takezawa et al. (13) note that TRPM7 overexpression abolishes the activation of PLC by carbachol receptors. Recently, Kim et al. (36) have also noted that regulation of native TRPM7 channels by PLC differs from the results obtained with transfected HEK-293 cells (14). We therefore used perforated patch clamping and Ca 2ϩ imaging to avoid the disturbances in cell signaling that likely occur in the whole-cell configuration.
Physiological Properties-Our results reveal that retroviral transduction results in the appearance of a conductance with all the expected properties of TRPM7 (see supplemental "Discussion"). At first sight, it may seem disturbing that the observed perforated patch currents are relatively small and lack strong outward rectification, a feature that has been termed "TRPM7 signature." However, small amplitude basal and evoked currents were to be expected, because TRPM7 was expressed at low levels, and the currents were recorded at non-depolarized voltages to allow comparison with the Ca 2ϩ data. Moreover, in HEK-293 overexpression studies, the TRPM7 inward currents (mainly Ca 2ϩ and Mg 2ϩ ) were also minor (3,7).
Strong outward rectification is seen under conditions of low [Mg 2ϩ ] i (whole cells; EDTA-AM-treated perforated patches). Interestingly, similar to the perforated patch currents, the whole-cell currents evoked by PLC activation in TRPM7-transfected HEK-293 cells with 1 mM Mg 2ϩ in the pipette showed only weak outward rectification (supplemental Fig. S3c). Most likely, this reflects an anomalous mole fraction effect, whereby intracellular Mg 2ϩ attenuates outward monovalent currents. However, separation of this effect from the inhibitory effect of intracellular Mg 2ϩ on channel gating would require systematic analysis (preferably in inside-out patches), which to our knowledge has not been performed thus far. In addition, intracellular Mg 2ϩ levels may act through the TRPM7 protein itself, in view of the reported function of the kinase domain in determining the set point for Mg 2ϩ inhibition (4,9) as well as through additional mechanisms (37). The use of EDTA-AM to activate the channels in intact cells should present an important new experimental paradigm for studies in this direction.
Signaling Pathways-Takezawa et al. (13) recently reported that activation of PLC through endogenous M1-muscarinic receptors had little effect on MIC/MagNuM currents in TRPM7-overexpressing HEK-293 cells. Perhaps the discrepancy between the Runnels et al. (14) and our study on the one hand and that of Takezawa et al. (13) on the other hand is a difference in potency of the PLC-activating receptors involved, as in their (and our) hands, endogenous M1 receptors cause only minor PIP 2 breakdown. Furthermore, unlike in our low level retroviral overexpression studies, in HEK-293 cells, TRPM7 overexpression inhibited carbachol-induced PLC signaling (13). Their study also relied on the use of the "PLC inhib-itor" U-73122, a compound with many known side effects (38,39). Rather, Takezawa et al. (13) suggest that the modulatory effects of carbachol are mediated by cAMP, as lowering [cAMP] i attenuated whole-cell TRPM7 currents in HEK-293 cells. However, in N1E-115 cells, bradykinin had no effect on cAMP levels, and conversely, potent cAMP-raising agonists did not trigger sustained Ca 2ϩ influx or TRPM7 currents detected in the perforated patch. Rather, our analysis indicates a key role for PLC-derived signals.
Dual Regulation by Phosphoinositide Signals-At odds with a previous report (14), we observed that stimulation of PLC-activating receptors caused TRPM7 channel opening rather than closure. This discrepancy is not cell type-dependent nor does it depend on TRPM7 expression levels. How can the paradoxical effects of PLC activation on TRPM7 currents in whole-cell and perforated patch experiments be reconciled? We speculate that fundamentally different modes of regulation mediate stimulation and inhibition.
On the one hand, in unperturbed cells, currents increase upon PLC activation, and therefore we propose that this must be the more physiological mode of action. It is reassuring that this concurs with our recent biochemical and cell biological data, which unequivocally demonstrate the activation of TRPM7 by PLC-activating agonists in intact cells (1), and with evidence from a very recent study by Kim et al. (36). However, the precise mechanism of this activation pathway remains elusive, although we can exclude the involvement of the PIP 2 -derived messengers diacylglycerol (because the addition of membrane-permeable analogues did not activate TRPM7, data not shown) and cytosolic Ca 2ϩ (supplemental Fig. S2a).
On the other hand, PIP 2 undoubtedly is an important cofactor for normal TRPM7 functioning (14,37,40). In analogy to other TRP family channels (17,20), C-terminal stretches of positive amino acids may bind PIP 2 to impose a proper tertiary structure. In TRPM7, positive stretches are present in the TRP consensus domain and at amino acids 1147-1154 and 1196 -1218 in the C terminus. Loss of PIP 2 could inhibit normal opening of TRPM7, but this would only be revealed following prior full activation of the channel by Mg 2ϩ depletion. In line with this notion, spontaneous rundown of MIC/MagNuM currents in whole-cell and inside-out experiments is accompanied by loss of PIP 2 (37,40). Our data show that, in whole cells but not in perforated patches, PLC-activating agonists cause a sustained PIP 2 drop due to impaired resynthesis. Thus, the effects of PLC activation on whole-cell TRPM7 currents could be viewed as "accelerated rundown." Alternatively, the opposing effect of PIP 2 depletion on PLC-mediated activation may reflect a subtle feedback mechanism whereby ongoing loss of PIP2 counteracts or limits TRPM7 activation. A similar mechanism was proposed recently for TRPM8 (20). In strong support of this dual regulatory model, kinetic analysis of BK-mediated stimulatory and inhibitory effects shows that activation proceeds distinctly faster than inactivation.
In summary, our experiments reveal a second mode of activation for TRPM7; not only can the channel be activated by depletion of intracellular Mg 2ϩ and Mg 2ϩ nucleotides but also by stimulation of endogenous PLC-activating receptors.