Regulation of a TRPM7-like Current in Rat Brain Microglia*

Non-excitable cells use Ca2+ influx for essential functions but usually lack voltage-gated Ca2+ channels. The main routes of Ca2+ entry appear to be store-operated channels or Ca2+-permeable non-selective cation channels, of which the magnesium-inhibited cation (or magnesium-nucleotide-regulated metal cation) current has received considerable recent attention. This current appears to be produced by one of the recently cloned transient receptor potential (TRP) channels, TRPM7. In this study of rat microglia, we identified TRPM7 transcripts and a prevalent current with the hallmark biophysical and pharmacological features of TRPM7. This is the first identification of a TRPM7-like current in the brain. There is little known about how members of the TRPM sub-family normally become activated. Using whole-cell patch clamp recordings from rat microglia, we found that the TRPM7-like current activates spontaneously after break-in and that the current and its activation are inhibited by elevated intracellular Mg2+ but not affected by cell swelling or a wide range of intracellular Ca2+ concentrations. The TRPM7-like current in microglia appears to depend on tyrosine phosphorylation. It was inhibited by several tyrosine kinase inhibitors, including a peptide (Src 40–58) that was shown previously to inhibit Src actions, but not by inactive drugs or peptide analogues. The current did not depend on the cell activation state; i.e. it was the same in microglia recently removed from the brain or when cultured under a wide range of conditions that favor the resting or activated state. Because TRPM7 channels are permeable to Ca2+, this current may be important for microglia functions that depend on elevations in intracellular Ca2+.

Non-excitable cells use trans-membrane Ca 2ϩ influxes for essential cell functions, including proliferation, apoptosis, secretion, volume regulation, and ion homeostasis. Immune cells were among the first and remain among the most intensively studied, with focus on general and more specific roles of Ca 2ϩ signaling, such as mitogenic activation, secretion of lymphokines, cytokines, cytotoxic molecules, and antibodies, and on phagocytosis, respiratory burst, and migration. In parallel, there have been an increasing number of studies on the pathways of Ca 2ϩ entry into these cells. In almost all cases, the role of membrane potential in Ca 2ϩ influx in non-excitable cells is fundamentally different from excitable cells. Voltage-gated Ca 2ϩ channels are rare in immune cells, and many cell types appear to be devoid of them. In these cells, although Ca 2ϩ influx is electrogenic and tends to depolarize the cell, the main effect of depolarization is to reduce the driving force for Ca 2ϩ influx. Because hyperpolarization favors Ca 2ϩ entry, the focus has been on store-operated Ca 2ϩ channels, including the Ca 2ϩrelease-activated Ca 2ϩ (CRAC) 1 channel, and on Ca 2ϩ -permeable non-selective cation channels. The CRAC current is widely expressed in immune cells, where it is one of the best characterized store-operated channels (1,2). However, two Ca 2ϩpermeable channel types often co-exist. Recent studies on the Jurkat T cell line and rat basophilic leukemia cells were the first to distinguish between co-existing CRAC currents and a Mg 2ϩ -inhibited cation current (also called magnesium nucleotide-regulated metal cation current) (3)(4)(5). The latter current is nearly identical to one of the recently cloned transient receptor potential (TRP) channels, TRPM7.
The TRP family comprises Ͼ20 channels, all with six transmembrane domains and cytoplasmic N and C termini, and recently divided into three sub-families: TRPC, TRPV, and TRPM (for reviews see Refs. 6 -8). Many of these channels are widely distributed in mammalian tissues, and following their cloning and heterologous expression, properties of some have made them excellent candidates for the Ca 2ϩ entry pathways of non-excitable cells. Most members of the TRPC sub-family are store-operated and Ca 2ϩ permeable, one of the TRPV members (TRPV4) is activated by changes in cell volume, and some members of the TRPM sub-family (TRPM7, TRPM8) are permeable to divalent cations, including Ca 2ϩ (9,10). TRP channels are not voltage-gated and thus are open over a wide range of voltages once activated. Those that are Ca 2ϩ -permeable produce Ca 2ϩ influxes that increase with the driving force; i.e. with hyperpolarizing membrane potentials, the common feature of Ca 2ϩ entry into immune cells and many other nonexcitable cells. Indeed, the properties of TRPM7 (9, 11) make it the best candidate for the non-store-operated Ca 2ϩ pathway for Ca 2ϩ entry in many immune cells.
Expression of TRPM7 is widespread, with transcripts in brain, spleen, lung, kidney, heart, and liver (9,11). Within the brain, it is not known which cell types express this channel or what role it might play in normal or pathological brain function. We have been studying ion-channel expression and roles in the immune cell of the central nervous system, the microglia. These cells express several types of K ϩ channels, and some are involved in key microglia functions; e.g. proliferation and the respiratory burst (12,13), reviewed in Ref. 14. Although the precise roles of the K ϩ channels are not yet known, they may act by maintaining a hyperpolarized membrane potential and facilitating Ca 2ϩ entry, as in lymphocytes (for reviews see Refs. 14 and 15). The routes of Ca 2ϩ entry in microglia are poorly understood. Microglia have ionotropic purinergic receptors that are permeable to Ca 2ϩ and monovalent cations (16) and a current reported to be CRAC (17), both of which should produce Ca 2ϩ influxes that increase with hyperpolarization. The functional expression of voltage-gated Ca 2ϩ channels (18) is somewhat controversial, but if present they should produce a depolarization-activated Ca 2ϩ influx. The presence of other Ca 2ϩ -permeable channels in microglia has not been reported. In the present study, we have identified transcripts for TRPM7 and characterized a prevalent current in rat brain microglia that has the hallmark features of expressed TRPM7 channels. The same current was expressed in ex vivo microglia, studied within 2 days of removal from the brain, as well as in microglia cultured under a wide range of conditions that favor either the resting or activated cell state.
There is little known about how members of the TRPM sub-family normally become activated. For instance, TRPM7 contains an active protein kinase domain (atypical ␣-kinase), it can phosphorylate itself, and mutations in the ATP-binding motif reduce the current (9). But it appears that effects of changing ATP may be entirely because of inhibition of TRPM7 channel activation by intracellular free Mg 2ϩ (10,11), which is also a distinguishing feature of native Mg 2ϩ -inhibited cation or magnesium nucleotide-regulated metal cation currents (3-5; 19). We addressed this question in the present study, using whole-cell and perforated patch clamp recordings on rat microglia. We found that the TRPM7-like current activates spontaneously after break-in, but much less current develops in perforated-patch recordings. The current, like TRPM7, was inhibited by millimolar total intracellular Mg 2ϩ concentrations. It was not affected by cell swelling or a wide range of intracellular Ca 2ϩ concentrations. Activation of the current was inhibited by several tyrosine kinase inhibitors and by a peptide that interferes with actions of Src kinase but not by the appropriate inactive drug analogues or a scrambled peptide. Because TRPM7 channels are permeable to Ca 2ϩ , this current may be important for microglia functions that depend on elevations in intracellular Ca 2ϩ .

EXPERIMENTAL PROCEDURES
Microglia-Cultures of highly purified microglia were prepared from neopallia of 2-day-old Wistar rat pups (Charles River, Quebec) as described previously (20). In brief, finely minced neopallial tissue was pelleted and resuspended in minimal essential medium (MEM; University Health Network, Sera and Media Services, Toronto). The tissue was filtered through a 40-m cell strainer and re-suspended in MEM that was supplemented with 5% horse serum, 5% fetal bovine serum, and 0.05 mg/ml gentamycin (all from Invitrogen). The cells were then seeded into 75-cm 2 flasks and re-fed 2 days after isolation. After 10 -12 days in culture without feeding, the flasks were shaken (180 rpm, 8 -12 h), and the floating cells (Ͼ95% pure microglia) were harvested. For electrophysiology, microglia were plated on 15-mm sterile glass coverslips and cultured in either serum-free medium (i.e. MEM with 2% B27 supplement (Invitrogen), 2 mM L-glutamine, and 0.05 mg/ml gentamycin) or astrocyte-conditioned medium (ACM), collected as the supernatant from rat astrocyte cultures grown in MEM with 2% fetal bovine serum (21). ACM was collected twice a week, frozen at Ϫ70°C, and used within one month. Where indicated, microglia were treated with 100 ng/ml lipopolysaccharide or 100 nM phorbol myristate acetate overnight before patch clamping. For ex vivo cells, freshly prepared suspensions of neopallial tissue were plated onto 15-mm sterile glass coverslips, fed with ACM, and patch-clamped 2 days after isolation.
Reverse Transcriptase-PCR-Reverse transcriptase-PCR was performed as described previously (22). Briefly, total RNA was isolated using TRIZOL reagent (Invitrogen) from cultures that were Ͼ98% pure microglia. To eliminate genomic contamination, the cell suspension was digested with DNase I (Amersham Biosciences) at 0.1 units/ml, for 15 min at 37°C. First strand cDNA was synthesized using an oligo(dT) primer (Amersham Biosciences) and used as a template for PCR reactions with a gene-specific primer (from ACGT, Toronto, ON) for ChaK (GenBank TM accession number NM_053705) (forward primer CTGAA-GAGGAATGACTACAC; reverse primer ACAGGGAAAAAGAGAGG-GAG). As a control, oligo mouse actin (M12481) was amplified (forward primer CTACAATGAGCTGCGTGTGG; reverse primer TAGCTCT-TCTCCAGGGAGGA). The PCR reaction was conducted with 1.5 mM MgCl 2 , 0.5 M forward and reverse primers, 10% of the cDNA reaction mixture, and 1.25 M of Taq DNA polymerase, using a GeneAmp PCR 2400 system (PerkinElmer Life Sciences). The mixture was preheated to 95°C for 5 min and then subjected to 25 cycles of a denaturing phase at 94°C, a 30-s annealing phase at 55°C, and a 30-s extension phase at 72°C. The amplified products were resolved in 2% agarose gels containing 0.5 mg/ml ethidium bromide. The identities of products of the predicted sizes were confirmed by restriction endonuclease digestion.
Patch Clamp Recordings-We made whole-cell or perforated-patch configurations at room temperature with an Axopatch 200A or Axopatch 1B amplifier (Axon Instruments, Union City, CA) and 5-10megohm resistance pipettes pulled from thick-walled borosilicate glass capillaries (WPI, Sarasota, FL). Capacitance and series resistance (but not leak current) were compensated online, and recordings were filtered at 5 kHz before acquiring data. pCLAMP v6.0 or 8.0 (Axon Instruments) was used for generating voltage commands, recording current, and collecting data, which were analyzed offline with Origin v6.0 (Northampton, MA). Liquid junction potentials were measured between each bath and pipette solution with a 3 M KCl electrode, subtracted offline, and used to correct all voltages and current versus voltage relations.
Several combinations of bath and pipette solutions (indicated in each figure legend) were used to isolate the current and to characterize its selectivity and ion dependence. The standard bath solution contained (in mM) 125 NaCl, 5 KCl, 1 CaCl 2 , 1 MgCl 2 and 10 HEPES, and standard pipette solution contained 40 KCl, 100 potassium aspartate, 1 MgCl 2 , 1 CaCl 2 , 10 HEPES, 2 K 2 ATP, and 10 EGTA. To minimize contributions of the swelling-activated Cl Ϫ current (20) many experiments were performed with low Cl Ϫ solutions made by substituting NaCl and KCl with 125 mM NaMeSO 4 and 5 mM KMeSO 4 (bath) and 130 mM KMeSO 4 or potassium aspartate (pipette). Sucrose was added when necessary to adjust the osmolarity of all solutions to 300 milliosmolar, measured with a freezing point-depression osmometer (Advanced Instruments, Norwood, MA). For ion-selectivity studies, low Cl Ϫ solutions were used, with one of the following substitutions: 130 mM nMDG MeSO 4 (pipette), K ϩ -free bath, or pipette solutions were made by replacing KMeSO4 with CsMeSO 4 , or a divalent-free bath solution was prepared by replacing MgCl 2 and CaCl 2 with 4 mM NaCl and adding 10 mM HEDTA. To assess the dependence of the current on internal Ca 2ϩ , the pipette calcium solutions contained 1 mM K 4 1,2bis(2-aminophenoxy) ethane N,N,NЈ,NЈ-tetraacetic acid (K 4 BAPTA) instead of EGTA, with free Ca 2ϩ adjusted by adding CaCl 2 , according to the CaBuffer program (from J. Kleinschmidt, formerly of New York University). The effect of changing the internal Mg 2ϩ concentration was tested by adding MgCl 2 to the low Cl Ϫ pipette solution while keeping the ATP concentration fixed at 2 mM. For perforated-patch recordings, the pipette was back-filled with intracellular solution containing 20 g/ml amphotericin B (Sigma) diluted from a freshly prepared 1 mg/ml stock solution in Me 2 SO (23). All bath solutions were pH 7.4 (adjusted with NaOH), and pipette solutions were pH 7.2 (adjusted with KOH or CsOH).
To initially assess regulation of the current by tyrosine kinases, a broad-spectrum tyrosine kinase inhibitor, genistein (or its inactive analogue, daidzein, both at 50 M), or the more Src-selective inhibitor, herbimycin A (2 M) (24) (all from Sigma), was added to the bath. A specific role for Src tyrosine kinase was further examined using pipette solutions containing the Src peptide, Src40 -58, or the scrambled inactive peptide, Src40 -58s (0.1 mg/ml), as we described previously (25,26). The peptides were synthesized at the Hospital for Sick Children (Toronto). Spermine was purchased from Calbiochem, and unless indicated, the remaining chemicals were purchased from Sigma and were the highest purity available.
Statistics-Where appropriate, data are presented as mean Ϯ S.E., with n indicating the number of cells studied under each condition. Differences between means were analyzed by using either paired or unpaired Student's t tests as appropriate.

Biophysical Properties of the TRPM7-like Current in Rat Microglia
An Outward-rectifying Cation Current-Unless otherwise indicated, all recordings were from microglia that were cultured in serum-free medium or astrocyte-conditioned medium, conditions that reduce microglia activation (21). We patch-clamped microglia that were unipolar or bipolar with elongated cell bodies, as in one of our earlier studies (12). Throughout this study, the holding potential was Ϫ5 mV to inactivate the voltage-gated K ϩ currents that have been well characterized in microglia: Kv1.3-like (12,25,27) and Kv1.5-like currents (12). Voltage-clamp steps and ramps were applied between Ϫ85 and ϩ115 mV, avoiding more negative potentials that activate the well characterized inward rectifier (Kir2.1-like) current (20,27). Except where noted, we observed an outward rectifying current that lacked time-dependent activation or inactivation during voltage-clamp steps. Fig. 1, A and B shows examples of the current; similar currents were seen in Ͼ200 microglia and, as shown in subsequent figures, under a wide range of conditions. Responses to voltage steps and ramps illustrate some hallmarks of this current; that is, lack of detectable time-dependent gating and strong outward rectification with a reversal potential of about Ϫ15 mV with standard bath and pipette solutions (circles in Fig. 1C). These features are very similar to the recently cloned TRPM7 channel (9,11). Most reports on transfected TRPM7 channels (9,11) have not shown or mentioned the tail currents, and many have shown currents only in response to voltage-ramp protocols. We found very small tail currents upon return to the holding potential of Ϫ5 mV (Fig. 1, A and E), which was close to the reversal potential. When steps to more negative potentials followed strong depolarizing steps, the tail currents were immediately small (not shown), without the relaxations that occur when voltage-gated channels close with hyperpolarization. A similar lack of time dependence is expected for TRPM7 channels, because the inward current is rapidly blocked by external divalent cations.
The outward current is carried by efflux of monovalent cations. That is, replacing intracellular K ϩ with nMDG ϩ nearly eliminated the outward current ( Fig. 1C), demonstrating that the monovalent cation, K ϩ , is highly permeant, but the bulky cation, nMDG ϩ , is not. In contrast, replacing external Cl Ϫ with methanesulfonate (MeSO 4 ) ( Fig. 1D) did not affect the current amplitude or rectification, and similar results were seen with external aspartate (n ϭ 3 cells, not shown). Thus, the outward current is not because of anion influx. K ϩ appears to be more permeant than Cs ϩ , because the outward current was significantly reduced by replacing internal K ϩ with Cs ϩ (Fig. 1C). Although a selectivity sequence of K ϩ ϾCs ϩ ϾnMDG ϩ is characteristic of K ϩ channels, the reversal potential of Ϫ10 to Ϫ15 mV with internal K ϩ or Cs ϩ rules out a K ϩ -selective channel; E K is Ϫ85 mV with the normal K ϩ -containing bath and pipette solutions used in Fig. 1, A-C. However, these features are typical of the TRPM7 channel, which passes monovalent cations (not anions) with a permeability sequence K ϩ ϳCs ϩ ϳNa ϩ ϾCa 2ϩ (9). Possible contributions of ionotropic purinergic receptors, which produce a non-selective cation current in microglia (see "Discussion"), were ruled out by several observations. First, to eliminate the possibility that ATP leaking from the pipette during seal formation activated the cation current, some experiments were done without ATP; the current activated normally (n ϭ 4, not shown). Second, the outward-rectifying current took several minutes to fully activate (see Fig. 7) whereas purinergic currents activate rapidly in the presence of ATP. Third, purinergic-receptor-mediated currents in microglia have nearly linear I-V relations, with substantial inward current.
In the absence of external divalent cations, a large inward current was revealed during voltage-clamp steps (Fig. 1E) or ramps (Fig. 1F). As before, there was no time-dependent gating evident, thus the step-elicited currents were superimposed on the ramp currents. As shown here and in Fig. 8C (open circles), in divalent-free solution there was a dramatic decrease in current rectification; e.g. the ratio of current density at ϩ115 mV versus Ϫ85 mV decreased from 8.6 Ϯ 2.3 in standard bath solution to 1.8 Ϯ 0.3 (n ϭ 7) in divalent-free bath solution. In divalent-free bath solution, the current reversed at Ϫ11 Ϯ 1 mV (n ϭ 7), which expected for a non-selective cation channel through which Na ϩ (predominant in the bath) and K ϩ (predominant in the pipette) permeate well. That is, the K ϩ Nernst potential is about Ϫ85, and the Na ϩ Nernst potential is about ϩ60 mV; thus, if the TRPM7 channel is equally permeable to K ϩ and Na ϩ (9) the reversal potential should be ϳϪ12 mV.
Inhibition by Intracellular Mg 2ϩ but Not by Intracellular Ca 2ϩ -With the standard 1 mM Mg 2ϩ and 2 mM K 2 ATP in the pipette, the outward-rectifying current increased over the first several minutes of whole-cell recording ( Fig. 2A; for average time course see Fig. 7D). The current was not sensitive to internal Ca 2ϩ (Fig. 2B); i.e. there were no changes in current density, degree of outward-rectification, or reversal potential over a wide range of free Ca 2ϩ concentrations. Thus, the microglia channel is not Ca 2ϩ -activated, like some cation nonselective channels (28,29), or subject to Ca 2ϩ -dependent slow deactivation, as is the CRAC current (30). Consistent with our results, buffering internal Ca 2ϩ to ϳ100 nM had no effect on cloned TRPM7 channels (31). However, there was a dose-dependent inhibition of current by internal Mg 2ϩ . The current density at both early (Ͻ3 min; see Fig. 2C, part i) and later times (Ͼ20 min; see Fig. 2C, part ii) reached significantly larger values when internal Mg 2ϩ was omitted. In contrast, the normal time-dependent build up of current was essentially prevented by 6 mM internal Mg 2ϩ , a concentration found previously (11) to fully inhibit cloned TRPM7 channels. The free concentrations of Mg 2ϩ (calculated with the CaBuffer program and taking into account 2 mM ATP) were 73 M when total Mg 2ϩ was 1 mM and 2.8 mM when total Mg 2ϩ was 6 mM. Normal free Mg 2ϩ levels in mammalian cells are 0.25-1 mM (32,33). Although increasing total Mg 2ϩ reduces ATP availability, this cannot account for the changes in current, because we observed current run-up without ATP in the pipette solution (n ϭ 4, data not shown). The observed inhibition of this current by millimo-lar intracellular Mg 2ϩ is consistent with cloned TRPM7 and native TRPM7-like channels in other cells (3-5, 10, 11, 19). Although the reversal potentials (E rev ) were similar upon break-in, after 20 -23 min E rev values were about Ϫ10 mV with 1 and 6 mM Mg 2ϩ compared with Ϫ23 mV with 0 mM internal Mg 2ϩ . One possibility is that reduced block by internal Mg 2ϩ allowed more monovalent current to flow, seen as a larger inward current.

Pharmacological Properties of the TRPM7-like Current
We tested several ion-channel blockers (Fig. 3), focusing on a known inhibitor of TRPM7 channels, 2-aminoethyldiphenyl borate (2-APB), a rather non-selective cation channel blocker (quinidine), and known blockers of other ion channels in microglia: SKF-96365, clotrimazole, flufenamic acid, and 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB). In each case, the drug concentration chosen was higher than the K d or IC 50 for block of the relevant channel, and different cells were used for each test. To minimize Cl Ϫ currents, low Cl Ϫ bath and pipette solutions were used. Effects of the drugs could not be attributed to current run-down, because in time-matched control recordings there was no run-down by 30 min, the longest time used for drug treatments. Drug reversibility was not fully assessed, but all were at least partly reversed during a 2-3-min washing period. The TRPM7 blocker, 50 M 2-APB (3, 5), significantly inhibited the outward-rectifying current in microglia; i.e. by 42% at ϩ110 mV. The ability of the CRAC blocker, SKF-96365, to block TRPM7-like channels is somewhat controversial, with 20 M producing no block in one study (5) and irreversible block in another (4). We observed no block of the TRPM7-like current by 20 M SKF-96365 (n ϭ 4, data not shown). The trivalent cations, La 3ϩ and Gd 3ϩ , which block CRAC channels at micromolar concentrations (3,34), only reduce cloned or native TRPM7 currents at millimolar concentrations (6,9,10). We found about 50% block of the microglial current by 1-2 mM La 3ϩ (n ϭ 3, not shown). Quinidine (1 mM) reduced the outward-rectifying current by 67%. However, quinidine is not very selective at such high concentrations but also blocks some sodium channels (35), Kv (36), and non-selective cation channels (37). Clotrimazole blocks the Ca 2ϩ /calmodulin-activated K ϩ channel, SK4, in microglia (13,38) and other cells, with a K d of about 25 nM (22,39). It is intriguing that the TRPM7-like current in microglia was reduced by 47% by 1 M clotrimazole. This cannot be attributed to contamination by SK4 channels, because the current was not K ϩ -selective (Fig. 1C) and was completely insensitive to changes in internal Ca 2ϩ (Fig. 2B). Flufenamic acid is usually used as a Cl Ϫ channel blocker, and consistent with this finding, the swelling-activated Cl Ϫ current in rat microglia was fully blocked by 100 M flufenamic acid (20). The same concentration reduced the TRPM7-like current by 53% in the present study (Fig. 3B) and inhibited a nonselective cation current in epithelia (37). The lack of specificity of these high concentrations of flufenamic acid, quinidine, and clotrimazole mean that caution is needed in interpreting results of cell-functional studies using these drugs. In contrast, another Cl Ϫ channel blocker, NPPB, which fully blocks the swelling-activated Cl Ϫ current in rat microglia at 250 M (20), did not affect the TRPM7-like current in these cells (Fig. 3B). This result supports the use of NPPB for assessing the roles of Cl Ϫ channels in microglia and verifies that the TRPM7-like outward-rectifying current is not contaminated by Cl Ϫ current in our recordings.

Microglia Express Transcripts for the TRPM7 Channel
Essentially all properties described thus far for the outwardrectifying channel in rat microglia are similar to those of cloned TRPM7 channels (9,11), reviewed in Refs. 6, 7, and 40. TRPM7 is expressed in a wide range of tissues, including brain, skeletal muscle, kidney, heart, liver, and spleen. Thus, it was not surprising that rat microglia express transcripts for TRPM7 (Fig.  4). These microglia were cultured under the same conditions used for the patch clamp analyses in Figs. 1-7. Because there were no differences in current in response to several activating stimuli (see Fig. 9B), we did not assess whether the TRPM7 mRNA level changed with microglia activation.

Regulation of the TRPM7-like Current
Spontaneous Development of the TRPM7-like Current in Whole-cell Recordings-Currents from heterologously expressed TRPM7 channels (9,11) and native TRPM7-like currents (3)(4)(5) spontaneously activate in whole-cell recordings, as did the current in microglia. We showed (Fig. 2C) that the current amplitude is more than twice as large with 0 mM internal Mg 2ϩ than with 1 mM and much smaller with 6 mM Mg 2ϩ . Results in Fig. 5, A-C shows that the time course of development also depends on the internal Mg 2ϩ concentration. That is, with 1 mM internal Mg 2ϩ the current doubled in the first 6 -8 min (Fig. 5B, and see average for control cells in Fig.  7D). When internal Mg 2ϩ was omitted (Fig. 5A), the current activated more rapidly, taking about 70 s to reach the halfmaximal amplitude and attaining a plateau in about 2 min. Conversely, with elevated internal Mg 2ϩ (6 mM), the current progressively decreased to about 60% of the initial value by 3 min after establishing a whole-cell recording. Again, half of this decrease was achieved in about 70 s. By plotting both inward and outward currents as percent of their peak values, we demonstrate the simultaneous changes in inward and outward currents at each Mg 2ϩ concentration, further evidence that the same channel carries both currents and that the effects of internal Mg 2ϩ on the time course are not voltage-dependent.
In whole-cell recordings, significant current was present immediately after break-in (e.g. see Fig. 2C, Fig. 5, A-C, and Fig.  7, A and D). However, in perforated-patch recordings there was very little increase in current (Fig. 5D, closed circles); the average current density after 20 min of recording was 5.4 Ϯ 1.2 pA/pF at ϩ115 mV. Compare this with the 6-fold increase in current density during whole-cell recordings with 1 mM total Mg 2ϩ (open circles). Current development during whole-cell recordings produced a stable, activated state. That is, for two cells, the current was allowed to develop during a whole-cell recording and then the pipette was removed (not shown). When a second whole-cell recording was established on the same cells, the current was immediately large. This observation, and the spontaneous development of current in whole-cell but not perforated-patch recordings, suggests that the TRPM7-like current is activated by washout of an intracellular inhibitor. Cell swelling activates a pronounced Cl Ϫ current in microglia, which is blocked by NPPB (see Fig. 5E and Ref. 20) However, activation of the TRPM7-like current could not be attributed to changes in cell volume during whole-cell recording, because the current was not affected by a hypotonic bath solution (Fig. 5E).
Spermine Blocks the TRPM7-like Current in a Voltage-dependent Manner-Another characteristic of TRPM7 channels is their block by external polyvalent cations like spermine and spermidine. This block is voltage-dependent, with more block of inward currents at moderately negative potentials that help to drive the polyvalent cations into the external pore. A detailed study (41) of native TRPM7-like channels in rat basophilic leukemia cells showed some relief of block at strongly hyperpolarized potentials that drive the external polyvalent cations through the channel. In addition, they observed some block of outward current at positive membrane potentials when the polyvalent cation concentration was high. Thus, to examine the block by external spermine, we used 0 mM internal Mg 2ϩ to allow the current to develop and stabilize quickly (as in Fig. 5A) and increased the inward current by perfusing in a divalent FIG. 4. TRPM7 transcripts are present in rat microglia. mRNA was prepared from cultured microglia (see "Experimental Procedures"). TRPM7-specific forward and reverse primers amplified a product of the correct size (lane 2) but not in the negative controls when either reverse transcriptase or cDNA was omitted (lanes 3 and 4). Restriction enzyme digestion (lane 5) produced two products of the expected sizes (306, 354 bp), which appear as a broad band.
cation-free bath solution (as in Fig. 1E). Fig. 6 shows representative current versus voltage curves for one cell (part A) and the time course of outward current plotted at ϩ80 mV and inward current at Ϫ80 mV for the same cell (part B). The numbers in part B indicate times at which the currents in part A were recorded. After the current had stabilized in normal bath solution (trace 1), divalent-free solution was perfused into the bath, resulting in larger inward and outward currents with little rectification (trace 2; compare with Fig. 1F). Perfusing spermine (100 M) into the bath in divalent-free solution essentially abolished the inward current between 0 and Ϫ80 mV (leak current remained) and moderately reduced the outward current (trace 3). The block was reduced by washing out the spermine for 3-4 min (trace 4) before returning to the normal bath solution. Consistent with previous studies, the block was voltage-dependent with an average block of 94.4 Ϯ 5.7% at Ϫ80 mV and 41.1 Ϯ 8.2% at ϩ80 mV (n ϭ 4 cells).
Broad-spectrum Tyrosine Kinase Inhibitors Reduce the TRPM7-like Current-For all experiments using protein tyrosine kinase inhibitors, low Cl Ϫ bath and pipette solutions were used to eliminate Cl Ϫ currents. Some outward-rectifying current was present at ϩ115 mV upon break-in; i.e. ϳ30% of the maximal current in this cell. The TRPM7-like current was inhibited rapidly and reversibly by bath addition of the broadspectrum tyrosine kinase inhibitor, genistein (50 M) (Fig. 7A), with a maximal inhibition of 59 Ϯ 5% (Fig. 7C) within 10 min. After washing out genistein, the current recovered fully over the next ϳ10 min in this cell. The effects of genistein were rapid and reversible in all eight cells tested. Although there is precedent for direct block of some channels by genistein, this is unlikely to be the case for the TRPM7-like current. That is, the time needed to exchange the small bath volume (250 l) was Ͻ30 s, so the several minutes required for onset and recovery of genistein inhibition were not because of the perfusion time. The inactive analogue, daidzein, did not significantly reduce the current (Fig. 7, A and C). The genistein-sensitive current that was isolated by subtraction (Fig. 7B) displayed the typical outward-rectifying profile of TRPM7, with large currents above

FIG. 5. Controlled development of the TRPM7-like current.
A-C, intracellular Mg 2ϩ affects the time course of spontaneous development of the outward current (measured at ϩ115 mV) and inward current (measured at Ϫ85 mV). Whole-cell recordings were made with different Mg 2ϩ concentrations in low Cl Ϫ solutions (to eliminate Cl Ϫ currents). Before averaging, each point was calculated as a percent of the maximal inward or outward current for each cell, to standardize for different current amplitudes between cells and the much smaller inward than outward currents. Values are mean Ϯ S.E. for the number of cells indicated in parentheses. D, spontaneous development of the current was markedly reduced in perforated-patch recordings. First, perforated-patch recordings were alternated with whole-cell recordings on different cells, and the average current densities were compared at 20 -30 min of recording (by which time whole-cell currents reach steady-state; see Fig. 2A). Second, for two cells, perforated-patch recordings were followed by breaking into the whole-cell configuration and recording for a further 20 min (closed circles). Values are mean Ϯ S.E. for the number of cells indicated in parentheses. E, the TRPM7-like current is not sensitive to cell swelling. In control recordings (three left-hand bars) with standard bath and pipette solutions, perfusion of a hypotonic solution (75% of normal osmolarity) into the bath activated a swelling-sensitive Cl Ϫ current that was fully blocked by 250 M NPPB. In separate cells (right-hand bars), the TRPM7-like current was allowed to run up in isotonic solution and then 250 M NPPB was perfused in, followed by the hypotonic solution. There was no increase in the NPPB-insensitive TRPM7-like current. Values are current density at ϩ115 mV, mean Ϯ S.E. for the number of cells indicated above each set of bars.

FIG. 6. Spermine blocks the TRPM7-like current in a voltagedependent manner.
A, a representative cell showing current versus voltage curves from ramp voltage-clamp protocols. Dashed trace (labeled 1) shows control, outward-rectifying current in normal bath solution. Trace 2 shows larger, slightly inward-rectifying current after divalent cation-free solution was perfused into the bath. Trace 3 shows reduced current after perfusing 100 M spermine into the divalent-free bath; note greater reduction of inward current. Trace 4 shows partial restoration of the current after spermine was washed out of the bath for 4 min. B, the time course of the experiment in A plotted as outward current at ϩ80 mV and inward current at Ϫ80 mV. The numbers indicate the times at which the currents in A were recorded. After washing out spermine, the normal bath solution (containing divalent cations) was perfused in. ϩ40 mV.
Next, we used the more Src-selective inhibitor, herbimycin A (24). To focus on the time course, differences in current amplitude between cells were eliminated by normalizing the current as percent of maximal current at ϩ115 mV in the same cell (Fig. 7D). Again, ϳ40% of the maximal current was present upon break-in. The current spontaneously increased about 2.5fold over the next 20 -30 min and then slowly decreased to about 65% of the maximal value by 90 min of recording. In separate cells, when 2 M herbimycin A was added to the bath after 15 min of recording, the current significantly decreased, and only 25% remained by 90 min. Washout of herbimycin was too slow to complete during the lifetime of a recording. The inhibition by two chemically distinct tyrosine kinase inhibitors suggests that tyrosine kinases are involved in the activation or maintenance of the current.
Selectively Inhibiting Src Kinase Reduces the TRPM7-like Current-The inhibition by herbimycin A suggested involvement of an Src family kinase. Thus, we exploited a peptide, Src40 -58, made from the unique domain of Src. This peptide was originally used to produce a Src-inhibiting antibody, but the peptide itself was found to act like an inhibitor of Src kinase function when added to the patch pipette during whole-cell recordings (42). Although its precise mechanism of action is still unknown, the authors concluded that the Src40 -58 peptide interacts with an unknown protein that regulates Src function (42). Our earlier results are consistent with Src40 -58 acting as an Src inhibitor; that is, it produced effects opposite to those of a Src-activating peptide (26). In the present study (Fig.  8), the development of the current was compared with a normal pipette solution or one containing a scrambled version of the Src peptide as a control for the addition of an intracellular peptide. Under control conditions with normal levels of Ca 2ϩ and Mg 2ϩ in the bath, the outward current increased more than 2-fold within 20 min after break-in (p Ͻ 0.01, n ϭ 7). Perfusing in a divalent-free bath solution revealed the large inward monovalent current (as in Fig. 1, E and F) but had little FIG. 7. Tyrosine kinase inhibitors reduce the TRPM7-like current. All whole-cell recordings were made with low Cl Ϫ solutions to eliminate Cl Ϫ currents. A, genistein inhibits the current. A representative recording, plotted as current density at ϩ115 mV, shows the dramatic and reversible inhibition by genistein but not the inactive analogue, daidzein. After the current developed and stabilized (ϳ20 min after break-in), 50 M daidzein was perfused into the bath for about 10 min and then washed out. Then, 50 M genistein was perfused into the bath for about 8 min and then washed out. B, isolating the genistein-sensitive current. For each voltage-clamp ramp, the current remaining after adding 50 M genistein to the bath was subtracted from the control current to isolate the genistein-sensitive component. Then, the mean current density versus voltage relation for the eight cells was calculated (heavy trace). The two thin traces show the range in the current amplitudes at Ϯ 1 S.E. C, summary of inhibition by genistein. The current density at ϩ115 mV (mean Ϯ S.E., number of cells indicated on the bars) is shown before and ϳ5 min after adding 50 M genistein or daidzein to the bath. Student's paired t test, **, p Ͻ 0.01. D, the tyrosine kinase inhibitor, herbimycin A, also reduces the current. Whole-cell recordings were begun in control solutions (open symbols) and then half of the cells were exposed to 2 M herbimycin A at 15 min (closed squares). Each current was normalized to its maximal value at ϩ115 mV before calculating the mean values as a function of time. Values are mean Ϯ S.E. for the number of cells shown in parentheses. Student's unpaired t test, *, p Ͻ 0.05; **, p Ͻ 0.01.
FIG. 8. Inhibiting Src tyrosine kinase reduces the TRPM7-like current. All recordings were begun with standard low Cl Ϫ solutions containing 1 mM CaCl 2 and 1 mM MgCl 2 and then switched for a few minutes to a divalent-free bath solution. A, a representative recording of the outward (at ϩ115 mV) and inward (at Ϫ85 mV) currents. B and C, average current density was measured after 20 min of whole-cell recording with standard low Cl Ϫ solutions in the bath and pipette (control), with the Src peptide (Src40 -58, 0.1 mg/ml) or the inactive scrambled peptide (Src 40 -58s, 0.1 mg/ml) in the pipette solution. In B, the bath solution contained 1 mM CaCl 2 and 1 mM MgCl 2 . In C, divalent cations were omitted from the bath. Values are mean Ϯ S.E., with the number of cells in parentheses; *, p Ͻ 0.05; **, p Ͻ 0.01. effect on the outward current. The Src40 -58 peptide prevented the spontaneous run-up of the outward-rectifying current, whereas the inactive scrambled peptide had no effect (Fig. 8B). Next, a divalent-free bath solution was used to increase the inward current, making it easier to show that the Src peptide simultaneously inhibited both inward and outward currents. In the presence of Src40 -58, neither inward nor outward currents ran up even after 20 min, whereas the scrambled peptide had no effect (Fig. 8C).
The TRPM7-like Current Is Constitutively Active in Microglia, Regardless of Cell Activation State-The same current was present in microglia recently removed from the brain (see "Experimental Procedures") ( Fig. 9A) as in cells cultured for nearly 2 weeks (see Figs. 1-8 and Fig. 9B). Moreover, the current increased for several minutes after establishing wholecell recordings, as before, and reached the same average current density. Thus, the expression and activation of TRPM7like currents are constitutive, not a result of cell-culture conditions. This conclusion is further supported by the lack of difference in current density (Fig. 9B) between microglia cultured with serum-free or ACM media, conditions that favor the microglia resting state (21) 2 or cells treated with the activating agents, lipopolysaccharide or phorbol myristate acetate.

DISCUSSION
A TRPM7-like Current in Microglia-Numerous properties of the current we recorded in rat brain microglia are very similar to cloned TRPM7 channels. Heterologously expressed TRPM7 produces a strongly outward-rectifying current, where the outward current is carried by monovalent cations, and the inward current is normally carried by divalent cations, including Ca 2ϩ (10,11). In the absence of external divalent cations, TRPM7 channels are permeable to monovalent cations, and the outward rectification is largely abolished (11). TRPM7 is inhibited by millimolar concentrations of intracellular Mg 2ϩ or Mg 2 ATP (11). For native TRPM7-like currents, it is often difficult to see an inward Ca 2ϩ current, because the whole-cell current is much smaller than for heterologously expressed channels. Thus, Ca 2ϩ permeability has been convincingly demonstrated only for transfected TRPM7 (10,11), where expressed currents were Ͼ10 times larger (2-3 nA) than the native current we observed in microglia (ϳ200 pA at ϩ80 mV). The small amplitude of the native TRPM7-like current in mi-croglia also rules out a precise measurement of its reversal potential, especially in divalent-containing solutions (e.g. in Fig. 1, B-D and Fig. 2) where the very small inward current can include some leak current caused by the patch pipette. In every way we examined, the microglial current is similar to TRPM7. There was (i) a spontaneous increase in current over several minutes during whole-cell recordings, (ii) strong outward rectification above ϳϩ50 mV, (iii) low selectivity among small monovalent cations and a permeability sequence of K ϩ ϳCs ϩ ϾnMDG ϩ , (iv) no discernable permeability to Cl Ϫ , aspartate, or methanesulfonate, (v) a small inward current that was dramatically increased by omitting external divalent cations, (vi) concentration-dependent inhibition by intracellular Mg 2ϩ (at 1 and 6 mM total Mg 2ϩ ), and (vii) inhibition by external 2-APB, La 3ϩ , or spermine but not by SKF-96365. The lack of effect of changing intracellular Ca 2ϩ (between 16 nM and 40 M) rules out activation by depletion of Ca 2ϩ stores, which is a characteristic of CRAC and TRPC channels. From these properties, it is anticipated that the TRPM7-like current in microglia will mediate a Ca 2ϩ influx that increases with hyperpolarization. A role for the outward monovalent cation current is less obvious, but with a depolarized reversal potential (ϳϪ15 mV), this channel will also tend to depolarize the cell.
Several compounds commonly used as blockers of other ion channels also inhibited the microglial current. These were quinidine, which is often used as a K ϩ -channel blocker, clotrimazole, which is usually used to block the Ca 2ϩ /calmodulingated K ϩ channel, SK4 (22,43,44), and flufenamic acid, which is usually used to block Cl Ϫ channels (20,45). There is precedent for quinidine and flufenamic acid blocking a non-selective cation current in epithelia (37).
Channel Regulation-The perplexing question of what activates TRPM7 channels has been investigated using both cloned channels and native TRPM7-like currents. TRPM7 contains an active atypical ␣ kinase domain that was initially thought to be essential for channel activity, because the current was increased by intracellular ATP (9). However, this result has been re-interpreted by several studies showing that the channel is directly inhibited by internal Mg 2ϩ , even when it is added with ATP. In fact, high internal Mg 2ϩ inhibits in the absence of ATP, and poorly hydrolyzable analogs either do not inhibit (Na⅐ATP␥S) or inhibit in a reversible fashion (Mg⅐ATP␥S) under conditions where channel phosphorylation would have been essentially irreversible (11,46). Similarly, for native TRPM7like currents, direct inhibition by Mg 2ϩ , rather than ATP or Mg 2 ATP, has been clearly demonstrated (19). Some native current has been observed immediately upon establishing a whole-cell recording, and this current progressively declined when internal Mg 2ϩ was elevated to several millimolar (see Ref. 4 and present study). With standard or reduced levels of Mg 2ϩ in the pipette solution, the TRPM7-like current increased spontaneously for several minutes (see Refs. 9 and 11 and present study). It is notable that, in the current study, when perforated-patch recordings were made from microglia, the initial current was smaller and failed to increase during many minutes of recording, but run-up proceeded normally after the whole-cell configuration was attained in the same cells. Despite the Mg 2ϩ -sensitivity of the current, for several reasons we do not think it was activated in whole-cell recordings simply by washout of Mg 2ϩ . With 1 mM total Mg 2ϩ in the pipette (calculated free Mg 2ϩ , 73 M), the current required 6 -8 min for half-activation, much slower than predicted for diffusion of a small ion in a small cell. Moreover, the free Mg 2ϩ concentration in intact lymphocytes is 190 -290 M (33), whereas millimolar concentrations are needed to fully block TRPM7 channels (19). Thus, considerable current should have activated in perforat-2 R. Jagasia and L. C. Schlichter, unpublished results.
FIG. 9. The TRPM7-like current is present in microglia at differing activation states. A, whole-cell recordings were made from ex vivo microglia; i.e. in dissociated brain tissue recorded after only 1 day in culture. Low Cl Ϫ solutions were used in the bath and pipette. The current density (mean Ϯ S.E., n ϭ 4) is shown after at least 20 min of recording. B, the current density was not affected by culturing conditions or microglia activation. Microglia were grown in serum-free medium or astrocyte-conditioned medium or were treated overnight with the activating agents, 100 ng/ml lipopolysaccharide, or 100 nM phorbol myristate acetate. Values are mean Ϯ S.E. for the number of cells indicated in parentheses. Recordings were made with standard solutions (see "Experimental Procedures") wherein the bath contained 125 mM NaCl, and the pipette contained 40 mM KCl and 100 K aspartate. ed-patch recordings if Mg 2ϩ was the only cytoplasmic inhibitor. Nor could its activation be explained by the cell swelling that sometimes occurs spontaneously in whole-cell recordings, because the TRPM7-like current was not activated by hypotonic bath solutions. Finally, the microglial current was not activated by changing intracellular ATP, because it did not require ATP in the pipette solution to fully activate. This is in contrast to the swelling-activated Cl Ϫ current in these cells, which requires ATP (20).
The phosphatidylinositol pathway, which couples numerous receptors to phospholipase C (PLC), has been implicated in regulating TRPC3 and TRPC6, which are activated by the PLC product, diacylglycerol or its analogue, 1-deoyl-2-acetyl-snglycerol (47,48). It was initially suspected that TRPM7 might be regulated by the phosphatidylinositol pathway, because it physically interacts with the C2 domain of PLC-b 1 (9), an interaction that was reflected in its previous name, TRP-PLIK (''transient receptor potential-phospholipase interacting kinase"). Because TRPM7 interacts with the C2 domains of PLC-␤ 1 , PLC-␤ 2 , PLC-␤ 3 (but not PLC-␤ 4 ), and PLC-␥ 1 (but not PLC-␦ 1 ) (31), differences in PLC-dependent signaling can be expected between different cell types, depending on which PLC isoforms they express, and different receptors, depending on which downstream pathways they use. Receptor-mediated activation of PLC rapidly inhibited, rather than activated, TRPM7 (31). This inhibition appeared to be because of hydrolysis of the PLC substrate, phosphatidylinositol 4,5-bisphosphate, either by heterotrimeric G proteins linked to PLC-␤ or the receptor tyrosine kinase activation of PLC␥ by the epidermal growth factor receptor. It was not because of reduced diacylglycerol production, because inhibition was not prevented by adding the analogues, 1-deoyl-2-acetyl-sn-glycerol or diacylglycerol. Although the epidermal growth factor receptor is a receptor tyrosine kinase, TRPM7 activation was attributed to phosphatidylinositol 4,5-bisphosphate, thus no link has yet been made between TRPM7 and tyrosine phosphorylation.
We have now shown that several inhibitors of tyrosine kinases inhibit the activation of the TRPM7-like current during whole-cell recordings from rat microglia. The inhibitors used ranged from the broad-spectrum genistein to the more Srcselective herbimycin A to the peptide, Src40 -58, which prevents Src actions (42). Taken together, our results implicate Src itself in regulating the current but do not rule out the possibility that other protein tyrosine kinases also regulate TRPM7 channels. Further studies on the mechanism of regulation will be needed to address whether Src binds to and phosphorylates a specific tyrosine residue on TRPM7 and whether this residue is responsible for regulating the channel. The effect of inhibitors of protein tyrosine kinase(s) means that the spontaneous activation of current in whole-cell recordings might reflect spontaneous activation of a kinase or conversely, washout or inhibition of a tyrosine phosphatase.
Physiological Perspective-There is considerable interest in identifying Ca 2ϩ entry pathways in non-excitable cells, which usually lack voltage-gated Ca 2ϩ channels. Although it was initially thought that small-conductance CRAC channels provide this pathway, the more recently identified TRPM7 channels are very likely to be important (40). TRPM7-like currents have been identified in a number of non-excitable cells. Because of their Ca 2ϩ permeability and lack of voltage-dependent gating they are well designed to contribute to numerous Ca 2ϩdependent cell processes, which in immune cells include responses to mitogens, secretion of cytokines, lymphokines and antibodies, phagocytosis, and the concomitant respiratory burst. To accurately compare the contribution of TRPM7 with that of CRAC or other store-operated Ca 2ϩ channels, it will be important to develop potent and selective blockers. Although antisense knockdown is another potential approach, it may be problematic, because the level of TRPM7 activity appears to control cell survival. That is, targeted deletion in a chicken B-cell line induced growth arrest and cell death, and overexpression killed HEK cells within 24 h (11). If endogenous TRPM7 activity is tightly regulated, one possibility is that large changes in these currents cause cell death under pathological conditions. Although TRPM7 channels will likely respond to tissue ischemia, an expected and direct outcome should be a decrease in current, because the channels are inhibited by high internal free Mg 2ϩ (19). Intracellular ATP is a major buffer of Mg 2ϩ in mammalian cells (33), thus when it is depleted following severe metabolic disruption, as in ischemia, free Mg 2ϩ can rise substantially (33,49). This should inhibit TRPM7 channels, an effect that should be exacerbated by Mg 2ϩ permeation. TRPM7 channels are more permeable to Mg 2ϩ than Ca 2ϩ and have a permeability sequence of Zn 2ϩ ϳNi 2ϩ Ͼ ϾBa 2ϩ ϾCo 2ϩ ϾMg 2ϩ Mn 2ϩ Sr 2ϩ Cd 2ϩ Ca 2ϩ (10). This unusual permeability raises the possibility that these channels can regulate cellular levels of essential metal ions (e.g. Mg 2ϩ , Mn 2ϩ , Co 2ϩ ). However, the conjecture that TRPM7 channels allow toxic heavy metals to enter cells and kill them (10) is challenged by the observation that high intracellular levels of these divalent cations inhibit TRPM7-like current (19).
Our observation that inward and outward TRPM7-like currents are reduced by inhibiting the widely expressed Src tyrosine kinase could have profound implications for cell physiology, for instance of microglia. As the resident macrophages of the central nervous system, microglia play important roles in maintaining brain functions. In response to nearly every central nervous system insult (damage or disease), microglia transition from resting to activated states, which exhibit both beneficial and deleterious properties. Microglial activation is not an all-or-nothing response but appears to be a graded phenomenon. It includes some, but not necessarily all, of the following: proliferation, migration to the site of injury, phagocytosis and respiratory burst, synthesis of major histocompatibility complex molecules and antigen presentation, and morphological changes. Some of these changes are accompanied by release of potentially cytotoxic substances (e.g. reactive oxygen intermediates, nitric oxide, excitatory amino acids, proteases) or neurotrophic molecules, growth factors, and inflammatory cytokines (e.g. interleukin-1, interferon ␥, transforming growth factor-␤) (50 -53).
Studies of the roles of ion channels in microglia activation are at an early stage. Nevertheless, voltage-gated K ϩ (Kv) channels and swelling-sensitive Cl Ϫ channels appear to be important for proliferation (12,20), and the respiratory burst involves Kv1.3 and Ca 2ϩ /calmodulin-dependent K ϩ (SK3 and SK4) channels (13). The current model for the mechanism is that K ϩ channels hyperpolarize the microglia and enhance Ca 2ϩ entry though non-voltage-gated Ca 2ϩ channels. Moreover, activation of SK channels requires a rise in intracellular Ca 2ϩ and ATP-induced microglial activation, which includes release of TNF-␣, plasminogen and cytokines, and is Ca 2ϩ -dependent (54). Based on its expression and properties, the TRPM7 channel provides a likely pathway for this Ca 2ϩ entry. Numerous receptors expressed in microglia are functionally coupled to Src family tyrosine kinases (55)(56)(57); e.g. CSF-1R (linked to Src, Fyn, Yes), GM-CSF-R (linked to Lyn, Yes, Hck), Fc␥ receptors (Lyn), TNF-␣R (Fgr), IL-6R (Hck), and G proteincoupled receptors (e.g. for thrombin and PAF). Syk, which is downstream of Src family protein tyrosine kinases, tyrosinephosphorylates other effectors, including PLC␥. Our present results on tyrosine kinase-dependent regulation of the TRPM7like current in microglia means that any of these receptormediated pathways could, in principle, provide positive feedback wherein TRPM7 activation and concomitant Ca 2ϩ entry further enhances the cellular response.