A Constitutive, Transient Receptor Potential-like Ca2+ Influx Pathway in Presynaptic Nerve Endings Independent of Voltage-gated Ca2+ Channels and Na+/Ca2+ Exchange*

Calcium levels in the presynaptic nerve terminal are altered by several pathways, including voltage-gated Ca2+ channels, the Na+/Ca2+ exchanger, Ca2+-ATPase, and the mitochondria. The influx pathway for homeostatic control of [Ca2+]i in the nerve terminal has been unclear. One approach to detecting the pathway that maintains internal Ca2+ is to test for activation of Ca2+ influx following Ca2+ depletion. Here, we demonstrate that a constitutive influx pathway for Ca2+ exists in presynaptic terminals to maintain internal Ca2+ independent of voltage-gated Ca2+ channels and Na+/Ca2+ exchange, as measured in intact isolated nerve endings from mouse cortex and in intact varicosities in a neuronal cell line using fluorescence spectroscopy and confocal imaging. The Mg2+ and lanthanide sensitivity of the influx pathway, in addition to its pharmacological and short hairpin RNA sensitivity, and the results of immunostaining for transient receptor potential (TRP) channels indicate the involvement of TRPC channels, possibly TRPC5 and TRPC1. This constitutive Ca2+ influx pathway likely serves to maintain synaptic function under widely varying levels of synaptic activity.

larization of an invading action potential. In a variety of preparations, it has also been shown that Ca 2ϩ release from internal stores in the presynaptic terminal, mainly through a ryanodinesensitive pathway (via Ca 2ϩ -induced Ca 2ϩ release; CICR), contributes to regulation of synaptic neurotransmitter release (reviewed in Ref. 2), mainly as a function of action potential frequency or, in some cases, presynaptic receptor action (3)(4)(5). Following bursts of activity, partial store depletion may thus occur, necessitating the replenishment of the internal Ca 2ϩ stores to restore resting homeostasis in the terminal. In addition, changes in external [Ca 2ϩ ] as a result of synaptic activity (6 -10), typically dependent on firing frequency (11), often coupled with postsynaptic ionotropic receptor activation (12,13) or postsynaptic depolarization (14), may alter presynaptic [Ca 2ϩ ] i homeostasis, as will various pathological conditions (e.g. hypoparathyroidism, secondary to hypomagnesia).
In non-neuronal cells, various store-operated channels (SOCs) appear to function in the maintenance of internal Ca 2ϩ stores, although they also contribute to regulated changes in cytosolic [Ca 2ϩ ] i under some circumstances (15,16). In the presynaptic terminal, the calcium-sensing receptor (CaSR) (17) and various TRPC channel subunits (18) have been detected, as has the neuronal calcium sensor-1 (NCS-1) (19), although the case for the latter has been controversial (20); however, a clear case for SOCs has not been forthcoming. Rather, the Na ϩ /Ca 2ϩ exchanger was originally implicated as a possible pathway for Ca 2ϩ entry following nerve activity, but despite an internal "Na ϩ load" with sustained activity, the inward-directed Na ϩ gradient still provides sufficient drive on the exchanger to favor Ca 2ϩ efflux, even during nerve activity (Refs. 21-23; however, see Ref. 24). Thus, the Na ϩ /Ca 2ϩ exchanger largely contributes to clearance of [Ca 2ϩ ] i , and not Ca 2ϩ influx. In addition, the mitochondria (when present) and plasma membrane Ca 2ϩ -ATPase appear to play dominant roles in resetting the final resting [Ca 2ϩ ] i (25,26). Here, we have identified a presynaptic calcium influx pathway for restoring homeostasis, revealed upon depletion of external calcium using isolated adult mature nerve endings from mouse cortex.

EXPERIMENTAL PROCEDURES
Purification of Intact Isolated Presynaptic Nerve Endings (Synaptosomes)-The protocol used in this study was approved by the Drexel University College of Medicine Institutional Animal Care and Use Committee. The intact isolated presynaptic nerve endings were purified as described previously (27). In brief, cortices from adult C57/Bl6J mice (Jackson Laboratory, Bar Harbor, ME) were removed and placed into ice-cold 0.32 M sucrose. The cortices were homogenized in 0.32 M sucrose with a glass-Teflon tissue grinder. Synaptosomes were isolated using the Percoll step gradient method (28). The purified synaptosomes were washed with oxygenated HEPES-buffered saline (HBS, pH 7.4) containing 142 mM NaCl, 2.4 mM KCl, 1.2 mM K 2 HPO 4 , 1 mM MgCl 2 , 5 mM D-glucose, and 10 mM HEPES, plus 1 mM Ca 2ϩ . Under these conditions, the terminals are intact, closed structures, as they respire, maintain ionic gradients and functional ion channels, transport and secrete neurotransmitter in a Ca 2ϩ -dependent fashion (29). That they are fully intact functional structures is also evidenced by retention of fluorescent dyes (27,30) and patch clamp recording (31).
Cell Culture of NG108-15 Cells-The neuronal cell line NG108-15 was cultured under differentiation conditions (32). In brief, the cells were plated onto glass coverslips coated with Cell-Tak (BD Biosciences) in Dulbecco's modified Eagle's medium containing 1% fetal bovine serum and 1 mM dibutyryl adenosine 3Ј:5Ј-cyclic monophosphate for 3-5 days. Under these conditions, the cells extend long neurites, each with multiple, large presynaptic-like axonal varicosities, capable of Ca 2ϩ -dependent secretion of acetylcholine (33) and synaptic innervation of rodent myotubes (32). The presynapticlike varicosities contain voltage-gated ion channels typical of presynaptic sites (34,35), synaptic vesicles and mitochondria, and Ca 2ϩ stores (35).
Expression of shRNA Constructions in Differentiated NG108-15 Cells-Transfection of mouse shRNA constructs into differentiated NG108-15 cells was accomplished using FuGene (Roche Diagnostics), with which we typically obtain 80 -90% transfection efficiency as gauged by expression of green fluorescent protein. Effects of mouse TRPC1, TRPC5, and TRPC3 (as control) shRNAs were assessed after 3-4 days, using Ca 2ϩ imaging and immunoblot analysis.
Measurement of Relative Ca 2ϩ Levels-The purified synaptosomes or differentiated NG108-15 cells were loaded with 5 M of various fluorescent Ca 2ϩ indicator dyes (fluorimetric studies: fura-2, fluo-4, mag-fluo-4, mag-fura-2, X-rhod-1, fura-2FF; confocal imaging: fluo-4) in HBS containing 1 mM Ca 2ϩ using the acetoxymethyl ester derivatives of the dyes for 45-60 min at 37°C as previously described (27,30). In the case of mag-fluo-4 or X-rhod-1, the synaptosomes were further incubated at room temperature for 30 min, to favor localization of the dyes into compartmentalized or cytosolic sites, respectively (36). The dye-loaded synaptosomes were then washed by centrifugation and resuspended in HBS made without Ca 2ϩ (nominal Ca 2ϩ ; nCa 2ϩ ), unless otherwise indicated, to better control the Ca 2ϩ concentration with or without EGTA. For fluorimetric studies, aliquots of the dye-loaded synaptosomes were suspended in HBS in a stirred quartz cuvette held at 37°C in a temperature-controlled PerkinElmer LS50B luminescence spectrometer. For most experiments, fura-2 was used to detect changes in Ca 2ϩ levels under ratiometric conditions (emission at 510 nm in response to alternating excitation at 340 and 380 nm). Although maximum and minimum ratio values were determined at the end of control runs with fura-2-loaded synaptosomes by adding 5 M ionomycin followed by 4 mM EGTA, Ca 2ϩ depletion during a run led to spurious calculations of Ca 2ϩ levels using the Grynkiewicz equation (37). Hence, the data using fura-2 are presented simply as raw ratios. In addition, caffeine, which has been reported to quench certain fluorescent dyes (38), had no discernible quenching effect on fura-2 in synaptosome preparations. In experiments where recording at high time resolution was necessary, fluo-4 was used with 0.1-s sampling intervals. For the low affinity dye mag-fluo-4, emission was recorded at 515 nm in response to 490 nm. For the intermediate affinity dye X-rhod-1, fluorescent emission was recorded at 605 nm in response to 505 nm. For the high affinity dye fluo-4, fluorescence was recorded at 506 nm in response to 488 nm. For confocal imaging, the synaptosomes or differentiated NG108-15 cells were plated onto coverslips coated with Cell-Tak and then inserted into a rapid-exchange Warner perfusion system mounted on a Nikon Diaphot microscope attached to a Nikon PCM 2000 laser-scanning confocal imaging system. Fluorescent images were recorded in response to excitation at 488 nm. During the confocal imaging, the preparations were under constant perfusion at ϳ5 ml/min with HBS without or with Ca 2ϩ , as appropriate. Images were typically collected at 2-or 4-s intervals, with the first 5 consecutive images collected as a baseline. Each experiment corresponds to sequential images collected from a single preparation. The quantification of fluorescence intensities associated with individual synaptosomes or varicosities recorded in digitized images was calculated using MetaMorph (Molecular Devices, Downingtown, PA) and corrected for photobleaching based on the baseline images (typically Ͻ3%). Analysis was performed by an observer blind to the experimental conditions. Response to depolarization evoked by elevated extracellular K ϩ was used as a criterion for synaptosomal or varicosity viability. Data are presented as normalized responses (F/F 0 , where F 0 is the fluorescence intensity associated with a given structure at t 0 ). All reagents were used at or above maximal concentrations based on literature values (e.g. Ref. 39). Depletion was initially accomplished using 0.1 mM EGTA, final concentration. Note that the preparations were partially depleted prior to addition of EGTA, because of a final wash with nominally Ca 2ϩ -free HBS during preparation. For sequential depletion, EGTA was initially added at 0.1 mM and then following Ca 2ϩ re-addition to 0.5 mM net concentration, EGTA was re-added to 0.8 mM concentration (see Fig. 1A). For experiments involving lanthanides (Gd 3ϩ and La 3ϩ ) at Ͻ10 M, preparations were pre-depleted and then washed free of EGTA, due to the very high stability constants of the lanthanides for EGTA (40).
Immunocytochemistry-Immunostaining was performed as described (27,41,42). In brief, the synaptosomes or differentiated NG108-15 cells on Cell-Tak-coated coverslips were fixed with 4% paraformaldehyde in HBS for 30 -45 min. The preparations were then permeabilized by incubation with Tris-buffered saline (50 mM Tris pH 7.2, 0.9% NaCl) containing 0.1% Triton X-100 for 30 min, followed by extensively washing with Tris-buffered saline. For double immunolabeling, the preparations were first incubated for 60 -120 min at room temperature with affinity purified primary rabbit antibodies for specific TRP channels (1: 100 -200) and, in the case of the synaptosomes, a mouse monoclonal antibody for synaptophysin (clone SVP38; 1:500), which serves as a highly specific nerve terminal marker (43). After extensive washing with Tris-buffered saline, the synaptosomes were incubated for 60 -120 min at room temperature with fluorescein-conjugated goat anti-rabbit IgG (1:500) and, in the case of the synaptosomes, rhodamine-conjugated donkey anti-mouse IgG (1:500) secondary antibodies in the presence of 10% goat serum in Tris-buffered saline. Control preparations were incubated without primary antibodies but with secondary antibodies. After extensive washing, stained preparations were mounted and imaging of the synaptosomes was performed using the confocal microscope, recording the fluorescence emitted in response to excitation at 488 (fluorescein) or 568 nm (rhodamine) as described (27,41,42). In particular, the black level was set using the control preparations incubated with secondary antibodies only, eliminating the low level of background fluorescence. This black level was used when imaging all subsequent samples stained with primary and secondary antibodies. For immunostaining of synaptosomes, merged images are shown; for immunostaining of the NG108-15 cells, images of fluorescence in the fluorescein channel are shown in grayscale. Positive immunostaining for TRP subunits in the synaptosomes was denoted by fluorescence contained within the margins of the staining for synaptophysin (see Ref. 44 for discussion). The specificity of the anti-TRPC antibodies was verified via immunoblotting of SDS extracts of synaptosomal preparations (18), visualized using the Odyssey Infrared Imaging System (LI-COR Biosciences).
-Conotoxin MVIIC, -conotoxin GVIA, agatoxin TK, and thapsigargin were from Alomone (Jerusalem, Israel). Anti-synaptophysin monoclonal antibody (clone SVP-38), flufenamic acid, ruthenium red, ryanodine, (Ϫ)-xestospongin C, cyclopiazonic acid, 2,5-di-(t-butyl)-1,4-hydroquinone, pertussin toxin, and caffeine were purchased from Sigma. All anti-TRPC antibodies were from Alomone Labs. Fluorescein-conjugated goat followed by re-addition of (mM) Ca 2ϩ (ϩCa 2ϩ ) and finally K ϩ depolarization (K ϩ : ϩ30 mM) were recorded fluorimetrically (A-E and G) or via confocal imaging (F and H). In F and H, synaptosomes in the confocal microscopic field (see inset in F for example sequence for an individual synaptosome) were identified by their size (0.5-2 m) and their characteristic changes in Ca 2ϩ levels obtained on K ϩ -induced depolarization. In G, the effect of variable depletion on subsequent Ca 2ϩ influx was tested. In testing various concentrations of Ca 2ϩ , confocal imaging was used to minimize effects of external dye and to control the [Ca 2ϩ ] o in the absence of EGTA. As shown in H, an ED 50 of ϳ0.1 mM net free Ca 2ϩ was observed for Ca 2ϩ following re-addition, whereas the ED 50 for K ϩ depolarization was ϳ0.5 mM net free Ca 2ϩ (n ϭ 3; see also, Ref. 99). F/F 0 , fractional change in fluorescence reflecting the relative change in [Ca 2ϩ ] i ; 340/380 nm: ratio of fluorescence at alternating wavelengths.
anti-rabbit IgG antibody and rhodamine-conjugated donkey anti-mouse IgG antibody came from Jackson ImmunoResearch Laboratories (West Grove, PA). The mouse shRNA containing pLKO.1 constructs were obtained from the Drexel University RNAi Resource Center (under management of Dr. Aleister Saunders in the Department of Bioscience), having originally been obtained from the Open Biosystems Mouse Arrest TM shRNAmir Library (Huntsville, AL). Each construct was optimized for selective RNA knock-down. Separate constructs were engineered to co-express green fluorescent protein to assess transfection efficiency. All other chemicals were of the highest reagent grade.
Statistics-All experiments were independently replicated at least 3 times. Where indicated, the significance of the difference between mean values was determined by one-way analysis of variance followed by Scheffé F test or t test. Differences were considered significant when p was minimally Ͻ0.05.

Ca 2ϩ Depletion Reveals a Presynaptic Ca 2ϩ Influx Pathway-
Using a variety of fluorescent dyes with varying relative affinities for Ca 2ϩ , an increase in [Ca 2ϩ ] i occurred on re-addition of Ca 2ϩ to isolated mouse cortical nerve terminals following depletion of Ca 2ϩ by incubation in Ca 2ϩ -free HBS containing EGTA (Fig. 1), with responses evident at the level of individual terminals (Fig. 1F). This increase in [Ca 2ϩ ] i following re-addition of Ca 2ϩ to depleted synaptosomal preparations has been previously noted in numerous studies of synaptosomal Ca 2ϩ regulation (e.g. Refs. [45][46][47]. The time course for the increase in [Ca 2ϩ ] i was fairly rapid, as seen using non-ratiometric Ca 2ϩsensitive dyes (e.g. Fluo-4) at high time resolution (Fig. 1, B, F, and G; 0.1-s sampling interval). With direct addition of fluorescent dye as a measure of the time limitation for mixing (0.7 Ϯ 0.17 s, mean Ϯ S.D., n ϭ 3), the time course for the initial increase in [Ca 2ϩ ] i measured for synaptosomes loaded with Fluo-4 was 1.55 Ϯ 0.5 s (mean Ϯ S.D., n ϭ 18; binding of Ca 2ϩ to Fluo dyes is diffusion-limited (48)). This time course was similar to that found for K ϩ depolarization-induced increases in [Ca 2ϩ ] i (1.61 Ϯ 0.13 s, mean Ϯ S.D., n ϭ 9), measured after re-equilibration of the preparation with Ca 2ϩ . It was also similar to the increase observed in preparations twice-rewashed with HBS to remove any residual extracellular dye (2.0 Ϯ 0.12 s, mean Ϯ S.D., n ϭ 3; Fig. S1A, inset). The time course for the initial depletion was slower (e.g. fluo-4, 3.1 Ϯ 0.87, mean Ϯ S.D.; n ϭ 3; Fig. S1A, inset), but this response was highly variable. For the Ca 2ϩ dyes having relative lower affinities, fura-2FF and the less specific mag-fluo-4, the Ca 2ϩ changes with depletion and re-addition will reflect a mixture of cytosolic and compartmentalized Ca 2ϩ (Fig. 1, D and E). Depletion in the presence of these dyes displayed a second, slower phase, as compared with the high affinity dyes. In contrast, X-rhod-1, a higher affinity Ca 2ϩ dye, was loaded under conditions favoring localization to the cytosol (see Ref. 36) (Fig. 1C). Thus, there is an indication of Ca 2ϩ present in a store/mitochondrial compartment(s) in addition to the cytosol.
Influx was confirmed using Mn 2ϩ as the divalent cation in a typical fura-2 quench assay ( Fig. 2; see also, Ref. 49). The initial rate of Mn 2ϩ influx was not significantly affected following addition of Ca 2ϩ (Fig. 2, inset). In addition, Ca 2ϩ influx via this pathway could be detected within 1 s of depletion (partial) of the preparation, and had kinetics that did not change with longer periods of depletion (Fig. 1G) or with repetitive depletion (Fig. 1A). The increase in [Ca 2ϩ ] i on re-addition of Ca 2ϩ to depleted preparations was insensitive to prior treatment with pertussis toxin (see Ref. 50), nor was it affected by pretreatment with the permeant diacylglycerol analog 1-oleoyl-2-acetyl-sn-glycerol (supplemental Fig. S1). Thus, the apparent influx pathway for Ca 2ϩ into the depleted nerve terminal has kinetic features of a channel activity independent of a pertussin toxin-sensitive or DAG-coupled pathway; however, receptor regulation of this pathway is not excluded by these results.
Ca 2ϩ Influx Revealed by Ca 2ϩ Depletion Is Independent of VGCCs, Na ϩ /Ca 2ϩ Exchange, and Ca 2ϩ Stores-Although K ϩ depolarization-induced increases in [Ca 2ϩ ] i were invariably elicited following re-equilibration of the depleted preparations with Ca 2ϩ , the extent to which VGCCs are involved in the influx pathway revealed by Ca 2ϩ depletion was assessed using VGCC blockers. In the presence of a mixture of peptide toxin blockers (agatoxin and conotoxins) that completely suppressed K ϩ -induced Ca 2ϩ increases (27,51), there was no effect on the Ca 2ϩ influx pathway (Fig. 3A). In addition, depolarization in the presence of VGCC blockers prior to addition of Ca 2ϩ back to depleted synaptosomes had no effect on the Ca 2ϩ influx (Fig.  3B). Moreover, Sr 2ϩ , which is conducted poorly by classical SOCs but is carried by TRPC channels (52), could substitute for Ca 2ϩ in the influx pathway (Fig. 3C), whereas Gd 3ϩ had little if any effect, either as a SOC blocker (53) or cation carrier at Յ10 M (Fig. 3D). Ba 2ϩ , which also appears to be poorly conducted FIGURE 2. Divalent cation influx following Ca 2؉ depletion as measured using Mn 2؉ quench. Mn 2ϩ was added at various final concentrations, where indicated, before (blue traces) and after (red traces) addition of Ca 2ϩ , to synaptosomes loaded with fura-2, as described in the legend to Fig. 1. Inset, comparison of initial rates of Mn 2ϩ influx before (dotted blue trace) versus after (red traces) addition of Ca 2ϩ . Millimolar Mn 2ϩ is known to block VGCCs (49, 100 -102). Ratios were performed at 340 and 380 nm to gauge effects of both Mn 2ϩ and Ca 2ϩ in the same run.
To address other potential sources of Ca 2ϩ , blockade of the Na ϩ /Ca 2ϩ exchanger was performed using the inhibitor KB-R7943 (58) and had no detectable effect when used at 5 M. Block by KB-R7943 at 50 M did result in an enhanced second phase of the Ca 2ϩ influx after Ca 2ϩ re-addition (Fig.  4A), indicating no direct involvement of the exchangers in the depletion-coupled influx per se but an involvement in the equilibration of [Ca 2ϩ ] i , as expected (59), wherein under normal external [Na ϩ ], efflux of Ca 2ϩ is favored. Furthermore, removal of external Na ϩ (substituted with N-methylglucamine), which should also facilitate Ca 2ϩ entry via all of the Na ϩ /Ca 2ϩ exchangers, had no significant effect on the depletion-coupled Ca 2ϩ influx (Fig. 4B), although this result does not preclude conductance of Na ϩ . Taken together, these results further support a voltage-insensitive channel being the mediator of this Ca 2ϩ influx pathway, but not a SOC.
There is abundant evidence for the release of Ca 2ϩ from ryanodinesensitive stores in brain nerve terminals (for review, see Ref. 2), presumably from an endoplasmic reticulum compartment (60), having a significant impact on synaptic dynamics (61)(62)(63). However, we observed no detectable increase in [Ca 2ϩ ] i after application of thapsigargin, caffeine, or ryanodine to synaptosomes equilibrated in normal [Ca 2ϩ ] o (not shown), consistent with previous studies (64). It may have been that release of Ca 2ϩ from the stores was highly localized, resulting in a very small signal to noise ratio under conditions of normal basal levels of [Ca 2ϩ ] i , and/or the released Ca 2ϩ was very rapidly buffered, perhaps by the mitochondria (65). Caffeine had no effect on the depleted terminals or the Ca 2ϩ influx on Ca 2ϩ readdition, but it did strongly inhibit the K ϩ depolarization-induced Ca 2ϩ increase following re-addition ( Fig. 5A; 53 Ϯ 11% of control, . Similar effects of La 3ϩ and Gd 3ϩ were also observed at 100 M (not shown). Responses carried by 1 mM Sr 2ϩ (C) or 1 mM Ba 2ϩ in the presence of VGCC toxin blockers (E) were also measured. Note that K ϩ depolarization did not stimulate a significant increase in Sr 2ϩ entry. (The K d values for fura-2 for both Sr 2ϩ and Ba 2ϩ are higher than that for Ca 2ϩ .) mean Ϯ S.E., n ϭ 3), consistent with the presence of a ryanodine receptor-linked Ca 2ϩ store involved at least in part, in CICR evoked via VGCCs (4, 66 -68). During depletion or upon subsequent readdition of Ca 2ϩ , sarco/endoplasmic Ca 2ϩ -ATPase inhibitors (thapsigargin, cyclopiazonic acid or 2,5-di-(t-butyl)-1,4-hydroquinone) had no effect at concentrations typically maximal (100 nM TO 1 M) for triggering Ca 2ϩ release from stores. On the other hand, relatively high concentrations of thapsigargin and cyclopiazonic acid induced an apparent small increase in basal calcium, more clearly evident using the cytosolic dye X-rhod-1 (supplemental Fig. S2), followed by an enhanced rise in [Ca 2ϩ ] i with Ca 2ϩ re-addition. The rise in [Ca 2ϩ ] i with readdition and basal Ca 2ϩ during depletion were also enhanced after inhibiting the mitochondria with micromolar carbonyl cyanide p-trifluoromethoxyphenylhydrazone (protonophore) (Fig. 5B), or after inhibiting the mitochondrial Na/Ca antiporter with 10 M CGP37157 (Fig. 5C). (High micromolar concentrations of thapsigargin have been reported to directly release Ca 2ϩ from isolated mitochondria (69), here causing an overshoot of the second phase of the Ca 2ϩ influx.) In addition, pretreatment with ruthenium red, an inhibitor of the mitochondrial Ca 2ϩ uniporter (70), among other activities (39), led to a decrease (75 Ϯ 7% of control; mean Ϯ S.E.; n ϭ 3; p Ͻ 0.05) in the depletion-coupled Ca 2ϩ influx (Fig. 5D). Moreover, no effect of xestospongin C, an inositol 1,4,5-trisphosphate receptor antagonist, was evident on re-addition of Ca 2ϩ to depleted synaptosomes (Fig. 5E). Together, these results confirm that the major site for Ca 2ϩ storage essential for presynaptic Ca 2ϩ homeostasis is the mitochondria, with a contribution by a caffeine-sensitive (ryanodine receptor-linked) store. It is proposed that Ca 2ϩ released from such a store is rapidly taken up by the mitochondria, most likely due to very close proximity of the endoplasmic reticulum Ca 2ϩ release sites to the mitochondria (71)(72)(73).
Pharmacological and Immunocytochemical Evidence for TRPC Channels at Presynaptic Sites-TRPC channel components have been identified in synaptosomal preparations (18), although definitive localization to presynaptic terminals was not confirmed. To consider whether functional TRPC channels may be involved in the Ca 2ϩ entry following Ca 2ϩ depletion, several inhibitors of TRP channels were used. The Ca 2ϩ influx revealed on depletion was significantly reduced (64 Ϯ 6% of control; mean Ϯ S.E.; n ϭ 3; p Ͻ 0.01) following pretreatment with the non-selective cation channel inhibitor flufenamic acid (74), as was the K ϩ -induced Ca 2ϩ influx (Fig. 6A). In addition, SKF96365, another modulator of TRPC channels (39), strongly blocked the influx in response either to depletion (19 Ϯ 6% of control; mean Ϯ S.E.; n ϭ 3; p Ͻ 0.01) or K ϩ depolarization (Fig. 6B). Neither  All preparations were preloaded with fura-2, except C for which the preparation was preloaded with mag-fura-2 (furaptra), to avoid a small quenching effect of CGP37157 on fura-2 (not shown). Inhibition of K ϩ -induced increases in [Ca 2ϩ ] i have been previously noted for CGP37157 (103) and ruthenium red (39). spermidine, 2-aminoethoxy phenylborane, nor 3,5-bis(trifluoromethyl) pyrazole had any effect (supplemental Fig. S3), each of these inhibitors typically having significant inhibitory activity toward SOCs, but displaying variable inhibition of certain TRP channels (16,(75)(76)(77). As noted previously, influx was also not affected by prior treatment with 1-oleoyl-2-acetyl-sn-glycerol (Fig. S1). These results, together with the previous data, are consistent with a store-independent pathway possibly involving TRP channels.
To assess more directly the presence of TRP channels, synaptosomal preparations were immunostained for individual subunits. Positive immunostaining for TRPC1 and TRPC5 but not TRPC3 or TRPV5 (control) in synaptophysin-positive terminals was observed (Fig. 6C). There was also no detectable staining for TRPC4 and TRPC6 (data not shown). Staining for TRPC1 and, to a lesser degree, TRPC5 was also observed (Fig.  7A) in presynaptic-like axonal varicosities of the differentiated neuronal cell line NG108-15 (32), which, in turn, displayed Ca 2ϩ influx following Ca 2ϩ depletion (Fig. 7, B and C). The Ca 2ϩ influx into the presynaptic-like varicosities following depletion displayed much slower kinetics as compared with that observed for synaptosomal preparations, likely due to equilibration with Ca 2ϩ stores previously shown to be the dominant regulatory element within the varicosities (35), the latter also having a substantially larger volume (ϳ1000 times) as compared with the synaptosome volume. The Ca 2ϩ influx revealed on depletion was attenuated following expression of RNAi (shRNAmir) against mouse TRPC5, and to a lesser extent TRPC1, but not against TRPC3 (Fig. 8).

DISCUSSION
The most important intraterminal element responsible for presynaptic Ca 2ϩ homeostasis is the mitochondrion (23), initially identified in a series of in-depth biochemical studies (78). Nearly all mature, active presynaptic terminals contain mitochondria, although their localization is highly dynamic (79), leading to their (transient) absence from some presynaptic sites in the brain (80). Nonetheless, they likely reside in close association with endoplasmic reticulum that is continuous with the endoplasmic reticulum in the axoplasm (71,81). In addition, at the presynaptic membrane both the Na ϩ /Ca 2ϩ exchanger and the Ca 2ϩ -ATPase function in removing elevated Ca 2ϩ after an impulse (Fig. 9), and hence are also essential to presynaptic Ca 2ϩ homeostasis (78), as are VGCCs. However, substantial fluctuations in [Ca 2ϩ ] in the synaptic cleft may occur, particularly with high impulse frequency and postsynaptic depolarization (13), necessitating a mechanism to rapidly re-equilibrate intraterminal Ca 2ϩ (82).
Here, we have identified a presynaptic Ca 2ϩ influx pathway independent of Ca 2ϩ stores, Na ϩ /Ca 2ϩ exchange, and VGCCs that we propose operates in a constitutive fashion to maintain intraterminal Ca 2ϩ . Its putative constitutive behavior is supported by results of Mn 2ϩ influx assay, wherein the Mn 2ϩ influx was unaffected whether external Ca 2ϩ was present or not (Fig.  2). On the other hand, the impact of this influx pathway on presynaptic homeostasis will be affected by the level of cytosolic [Ca 2ϩ ] achieved via VGCCs following presynaptic stimulation. Re-equilibration of basal presynaptic Ca 2ϩ will thus involve all Ca 2ϩ sources and pathways.
This Ca 2ϩ influx pathway has properties distinct from a previously identified, outward rectifying non-selective cation channel activated in isolated terminals following reduction of [Ca 2ϩ ] o via a putative G-protein-coupled pathway (31). The previously identified cation channel was strongly inhibited by FIGURE 6. TRPC channels in presynaptic terminals. The effects of the TRP channel inhibitors flufenamic acid at 100 M (A) and SKF96365 at 1 mM (B) were tested. C, localization of TRP channels to presynaptic terminals was determined using double immunostaining for the presynaptic marker synaptophysin (red) and individual TRP subunits (green), and was displayed as merged images. The percentages of synaptosomes with positive immunostaining for TRPC1 and TRPC5 over background were 50 -80%. Also included is a panel (inset with TRPC1) of staining with secondary antibodies only, as background control (44). There was no significant immunostaining for TRPC4 or TRPC6 on isolated cortical terminals (not shown), although a prominent localization of TRPC4 to cortical neurons has been noted (94). There was also no positive staining for TRPC3 on synaptophysin-positive terminals, but there was significant non-terminal staining in the synaptosomal preparation (not shown). Scales bars, 1 m. D, TRPC1, TRPC3, and TRPC5 in synaptosomal extracts detected via immunoblot analysis using rabbit anti-TRPC-specific antibodies (red channel), co-stained for actin (control) using a monoclonal anti-␤actin (green channel). Immunodetection for TRPC1, which displays multiple bands, and TRPC3 was similar to previous results (18); however, TRPC5 was rather weak.
high micromolar to millimolar Mg 2ϩ or Ca 2ϩ , and it was also blocked by Gd 3ϩ and spermidine. These properties, together with outward rectifying behavior of the cation current, are more consistent with the TRPM7 channel (83,84), which is strongly inhibited by physiological concentrations (1-2 mM) of intracellular Mg 2ϩ , perhaps in concert with Mg 2ϩ -ATP via its kinase domain (85)(86)(87), and by activation of phospholipase C-coupled receptors (88,89), and that appears to function in cellular Mg 2ϩ homeostasis (90). In contrast, the presynaptic Ca 2ϩ influx pathway identified here was strongly enhanced by lanthanides, enhanced in high micromolar Mg 2ϩ , and was completely insensitive to spermidine or other typical SOC inhibitors. The Mg 2ϩ and lanthanide sensitivity of the influx pathway, in addition to its ionic, pharmacological, and shRNA sensitivity, and the results of immunostaining for TRP channels indicate the involvement of TRPC channels, specifically TRPC5 and TRPC1 (55)(56)(57). The lack of effect of 1-oleoyl-2-acetyl-snglycerol is also consistent with this possibility (16,91). Thus, although TRPM7 might appear to be an attractive candidate for the presynaptic Ca 2ϩ influx pathway, because of its constitutive Ca 2ϩ (and Mg 2ϩ ) permeability, the properties of TRPM7 would indicate otherwise, particularly the strong suppression of cation inward currents by millimolar Ca 2ϩ or Mg 2ϩ , as noted previously. In contrast, the properties of TRPC5 containing channels fit better with the pharmacological and ionic sensitivity of the presynaptic Ca 2ϩ influx pathway revealed following external Ca 2ϩ depletion.
Previous studies have identified TRPC1, TRPC4, and TRPC5 channels in cortical synaptosome preparations (18), TRPC1 in axonal processes of neonatal hippocampal neurons (55), and TRPC5 in growth cones (92). TRPC1 has been found throughout the brain, including cortex (55). TRPC5 has also been noted in cortical regions (93), although it appears to be predominantly FIGURE 7. TRPC channels in presynaptic-like axonal varicosities. Localization of specific TRPC channels (A) and functional responses (B) in presynapticlike axonal varicosities (arrows) of differentiated NG108-15 cells, a model rodent neuronal cell line, are shown, to demonstrate the presence of the Ca 2ϩ influx pathway revealed by Ca 2ϩ depletion (C) in a presynaptic-like structure (104) having well characterized Ca 2ϩ stores (35), which display CICR on K ϩ depolarization. The slower apparent rate of influx, as compared with depletion-coupled Ca 2ϩ influx into the synaptosomes, is likely due to the ratelimiting CICR within the relatively large volume of the varicosity (5-10-m diameter) as compared with the isolated brain nerve terminal (1 m average diameter; ϳ1/125-1/1000 relative volume). Positive immunostaining (gray) of varicosities (arrows) for TRPC1 and TRPC5 was evident, although the latter was just above background. There was no staining of varicosities for TRPC3 above background, but there was significant staining in cell bodies. Scale bars, 10 m. FIGURE 8. TRPC1 and TRPC5 RNAi reduce the Ca 2؉ influx revealed by Ca 2؉ depletion in presynaptic-like axonal varicosities. Ca 2ϩ influx was measured following Ca 2ϩ depletion, as shown in Fig. 7B, in the presynaptic varicosities of differentiated NG108-15 cells, 3 days after transfection with: control vector (black; n ϭ 6); shRNA against mouse TRPC1 (blue; n ϭ 10); shRNA against mouse TRPC5 (red; n ϭ 6); shRNA against mouse TRPC3 (green; n ϭ 20), used as additional control. In immunoblot analysis, TRPC levels were reduced by ϳ60 -90%.  DECEMBER 7, 2007 • VOLUME 282 • NUMBER 49 localized to hippocampus. (TRPC4 appears to be expressed in many brain regions, including cortex (94), and although no significant immunostaining in cortical terminals was found in the present study, it cannot be completely excluded.) Although the precise roles for TRP channels at presynaptic sites are largely unknown, it is proposed that TRPC1 and TRPC5 combine (55) to form a constitutively active Ca 2ϩ influx pathway in brain presynaptic nerve terminals, revealed upon depletion of internal Ca 2ϩ as a consequence of lower [Ca 2ϩ ] o in the synaptic cleft and functioning in nerve terminal Ca 2ϩ homeostasis following return of [Ca 2ϩ ] o to normal values.

Constitutive Presynaptic Ca 2؉ Influx
It remains to be determined the extent to which this presynaptic Ca 2ϩ influx pathway is regulated and also the degree to which it is Ca 2ϩ selective, with some significant Na ϩ permeability likely present, despite the lack of a pronounced effect of Na ϩ -free conditions (Fig. 4). As for specific regulators, NCS-1, present in presynaptic terminals (19), has been suggested to interact and functionally enhance TRPC5 containing channels in growth cones (95). Thus, the influx pathway in mature presynaptic terminals may exist in an activated state. It is also possible that several other processes regulate this pathway (56), including phosphorylation of TRPC (96), and activation of TRPC via lysophospholipids (97) and/or STIM1 (98). It is also possible, in view of their presence and proposed role in growth cone function (92), that TRPC5 containing channels function in presynaptic terminal dynamics.